Reoperations and Specific Local and Perioperative Complications
This chapter addresses the frequency of local and perioperative complications associated with breast implants. Later chapters of this report deal with long-term safety, particularly in terms of cancer and connective tissue disease. Local and perioperative complications are important outcomes in their own right, and to the extent that they lead to significant further medical interventions or impair the achievement of expected and desirable results, they are also relevant to implant safety. Five-year reoperative or secondary surgery rates or average number of implants placed per breast or per woman provide approximations of the sum of these complications. They are important to safety because, even though breast surgery is of low systemic morbidity, every operation and the attendant anesthesia carry risk. Many of the same complications that occur when implants are placed may occur when they are removed, revised, or replaced, e.g., infection, hematoma or seroma, pneumothorax, tissue necrosis (Rohrich et al., 1998a). Furthermore, other interventions such as closed capsulotomies, extra manipulations for mammographic screening or diagnosis, and medical care for rash, pain, infection and the like are often not included in reoperation and multiple replacement data and can be contributors to, or comorbidities with, the need for surgery. These other interventions can be very frequent; for example, as many as 192 closed capsulotomies in 140 patients (254 implants) have been reported (Brandt et al., 1984). Patient satisfaction, or the lack of it, is another indica-
tor that can generate further interventions, as noted in Chapter 1. It should be kept in mind with respect to the following discussion that implants, surgical experience, surgical techniques, and perhaps other factors have evolved since the studies reported here were undertaken, so current experience may differ. This argues for careful prospective studies as the committee concludes at the end of this section.
This chapter addresses the following topics because they have significant effects on implant safety: reoperation or secondary procedures as indicators of overall frequency of local and perioperative complications; aggregate complications in breast reconstruction; aggregate complications in breast augmentation; rupture and deflation; factors contributing to loss of implant shell integrity; detection of gel implant rupture; strength and durability of implant shells; frequency of implant rupture and deflation; description of implant fibrous tissue capsules and contractures; capsular, local breast, and distant tissue exposures to silicone and their complications; frequency of saline implant capsular contracture; barrier implants and contractures; effect of implant surface and contracture; effect of local adrenal steroids and contracture; presence of bacteria around implants, antimicrobial treatment and contracture or other complications; hematomas, their frequency and relationship to contractures; the effect of implant placement on contracture; and other relevant complications including pain.
Many other local and perioperative complications in addition to those noted above require explantation or other secondary surgical or medical interventions. A reasonably complete list (see Table 5-1) would include fibrous contracture of the implant capsule; gel implant rupture (with or without migration of silicone gel outside the capsule) or saline implant deflation; filler port or implant valve malfunction; shell folds or wrinkling; infection of the surgical wound; infection around or within the implant; infection associated with toxic shock syndrome; hemorrhage and hematoma; seroma; swelling of the breast; various skin rashes and other skin manifestations such as localized morphea; epidermal proliferative reactions (Spiers et al., 1994); middermal elastolysis; edema; blistering; cysts (Copeland et al., 1993); ulceration; necrosis of the skin, nipple or mastectomy or reconstruction flap; exudation of silicone through the skin or from the nipple (Erdmann et al., 1992; Leibman et al., 1992; McKinney et al., 1987); implant extrusion, misplacement, or displacement; silicone granuloma; axillary adenopathy; sensory loss and paresthesia; pain; abnormal lactation (Hartley and Schatten, 1971; Mason, 1991) and/or galactocele (DeLoach et al., 1994; Johnson and Hanson, 1996); thoracic skeletal asymmetries (Dickson and Sharpe, 1987; Peters and McEwan, 1993); pneumothorax (Brandt et al., 1984); and calcification. "Bleed" or diffusion of small quantities of mostly lower molecular weight linear (and cyclic) silicone gel fluid compounds through the silicone elastomer shell (and to a
Implant fibrous capsular contracture
Gel implant rupture (intra- and extracapsular)
Skin blistering, cysts, and necrosis
Swelling of the breast
Nipple or flap necrosis
Silicone exudation through skin or nipple
Implant shifting or displacement
Acute and chronic breast and chest wall pain
Saline implant deflation
Implant filler port or valve leakage
Loss or change in sensation of the breast or nipple
Operative wound infection
Chest wall skeletal changes
Infection with toxic shock syndrome
Hemorrhage at the operative site
Lactation and galactocele
Peri-implant hematoma or seroma
lesser extent outside the fibrous capsule) is also reported as a complication, but gel fluid diffusion is intrinsic to the design and physical characteristics of gel-containing implants (see Chapter 3 of this report).
Many of these complications have been cited in the approximately 100,000 adverse event reports to the Food and Drug Administration (FDA) summarized by Brown et al. (1998). This chapter does not rely on that reporting system, however. The FDA system is sensitive to national publicity and includes voluntary reports, which frequently consist of undocumented assertions (Brown et al., 1998), and is therefore subject to distortions of the frequency and nature of implant adverse effects.
Overall Frequency of Local Complications
Several studies with representative cohorts of 583 to 7,008 women address the frequency of secondary interventions in saline- and gel-filled implants for both augmentation and reconstruction (Gabriel et al., 1997; McGhan Medical Corporation, 1998; Mentor Corporation, (1992). Gabriel et al. (1997) reported that 178 (23.8%) of all 749 Olmstead County women of the usual age distribution, noted in Chapter 1, who were implanted at the Mayo Clinic (95% with gel-filled implants) had clinical indications requiring reoperation ranging from explantation to drainage of a hematoma over an average 7.8-year follow-up after implantation. This amounted to 18.8% of the 1,454 breasts implanted. Multiple complications occurred in 61% of these. Although the incidence of complications requiring surgery after augmentation or reconstruction did not differ at two
months, by the end of the fifth year when 83% of all first complications had occurred, the percentage of patients with complications after reconstruction (30-34%) was almost threefold that after augmentation (12%). Only surgical complications were analyzed, and some that may be important, such as silent ruptures, may have been missed. The frequency of complications reported in this study is consistent with the frequencies reported in other studies cited later, especially given the long average follow-up in this series.
The McGhan AR90 preliminary report (McGhan Medical Corporation, 1998) describes the 583 women of the usual age distribution who agreed to participate in this study and received McGhan 1990s, mostly textured, single- and multilumen gel-filled implants, 549 for augmentation and 34 for reconstruction, with a five-year follow-up. In this cohort, 23% of augmented women and 42.4% of reconstructed women required secondary surgery ranging from explantation to evacuation of hematoma or seroma, to correction of implant placement or contracture, to biopsy during the five-year study period. These are underestimates because implant rupture was diagnosed by physician evaluation; and therefore a number of silent ruptures were likely missed. Explantation is overestimated since about one-third (or roughly 6 and 14% of augmentation and reconstruction secondary surgeries, respectively) were at patient request because of safety concerns prevalent during the entry period (1990-1992) of this study not because of clinical indication.
The McGhan large simple trial (LST) (McGhan Medical Corporation, undated) was a one-year prospective observational study of all 2,855 women of the usual age distribution who agreed to participate in this study and received McGhan 1990s, predominantly textured, room temperature vulcanized (RTV) saline-filled implants, 81.1% for augmentation. Women entered this trial in 1995 and 1996. The cumulative results after a year for four complications (infection, deflation, explant, and severe [Baker Class III or IV] capsular contracture) were 18.9 and 35.9% of women following augmentation and reconstruction (includes revisions), respectively. These figures are more reflective of actual clinical conditions since saline implant deflations are likely to be observed, and women are much less likely to have requested explantation of saline implants on nonclinical grounds. While secondary surgical frequencies were not identified as such, they are likely quite similar although probably slightly higher than the cumulative percentage for the four complications, since these complications are highly predictive of surgical intervention. Additional complications in this cohort will occur in the one- to five-year interval. Comparison of these results with overall complications of the previously cited AR90, long follow-up gel-filled implants is inappropriate.
Women receiving Mentor gel-filled implants (and some receiving ex-
panders) were enrolled in the company's adjunct study. At the three-year follow-up, there were 3,559 women with reconstructions and 3,449 women with augmentation, almost all with low-bleed, gel-filled implants. The overall frequency of infection was 2.8-4.3% in reconstructions and 1.3% in augmentations. The overall Class III-IV contracture prevalence was 12-13% in reconstructions and 11% in augmentations. The overall frequency of rupture was 1.3-1.7% in reconstructions and 0.7% in augmentationsbut these ruptures were determined by physical examination only (Purkait, Mentor Corporation., IOM Scientific Workshop, 1998).
In addition to these studies, Gutowski et al. (1997) reported the outcomes of 504 patients with 995 predominantly Heyer-Schulte-Mentor saline implants in an 11-center retrospective cohort study. These patients represented the 41.5% of those identified by the plastic surgery centers that were successfully interviewed, but it is not clear how patients were identified or what proportions of the total number of women with implants at the centers were identified. This uncontrolled accession of patients raises significant concerns about the interpretation of any results. Implants were placed almost entirely (93.8%) for augmentation, evenly divided between submuscular and submammary positioning in 1980-1989, and were followed for an average of six years. Of these women, 20.8% underwent secondary surgery, primarily for replacement, removal, or capsulotomy. The complications of infection (0.2%), hematoma (1.6%), and seroma (0.1%) were infrequent. Deflation occurred in 55 implants (5.5%) and 51 women (10.1%). Deflation (and rupture) frequency differed by implant model. Only about 4.2% of the predominant late model RTV implants deflated. These probably represent minimum figures (Gutowski et al., 1997).
Fiala et al., (1993) reported the results of a survey of 106 women representing 62.9% of a cohort of 167 women who could be located from the original 304 women who had undergone breast augmentation from 1973-1991. Their implants were primarily smooth silicone gel (70.8%) but also included some polyurethane-coated (27.1%) and a few textured implants. In this survey, 73.9% of women reported being ''highly satisfied," 19.8% of women underwent secondary surgery, and the complications were mostly contractures. Contractures occurred more often as time progressed and significantly more in submammary than submuscular implants. There were fewer contractures, though not statistically significantly, around the polyurethane compared to the smooth implants (Fiala et al., 1993). Edworthy et al. (1998), surveying the experience of a population of 1,112 unselected women with silicone gel-filled implants, found that 214 women (19.25%) had undergone secondary surgical procedures and 38.5% of breasts implanted (average of frequencies reported for left and right breasts) had Class III-IV contractures. The odds of a woman
needing more than one implant per breast over time are high, and placement of as many as 16 implants per woman has been reported (Roberts et al., 1997). In one small cohort of 52 mostly (67%) augmented women who agreed to participate in the study, out of 138 consecutive women with breast implant problems, the average was 3.19 gel implants per woman over an average of 11.9 years (Wells et al., 1995). In another small cohort (N = 60) of consecutive women undergoing immediate reconstruction with expanders, 2.78 operations were required on average for each woman, 0.78 for complications and 2.00 for original expander insertion and the following permanent implant replacement (Slavin and Colen, 1990). Shanklin and Smalley (1998a) reported 3.45 implants per women in a small experience (N = 130) with patients self-selected for problems and a 49.3% frequency of procedures in addition to implant replacement including 15.4% closed capsulotomies. Although the data were reported in a way that made it difficult to aggregate them, each woman appeared to have undergone 1.8 to 1.9 operations, many of which were on the normal breast for correction of asymmetry in the series of 109 women with delayed postmastectomy reconstructions reported by Houpt et al. (1988). Worseg et al. (1995) reported 83 secondary operations in a cohort of 77 women implanted with inflatable (saline or dextran) Heyer-Schulte-Mentor implants with a mean follow-up of nine years. There was an average of 1.08 secondary operations per woman, predominantly for deflation (23.9%) or severe contracture (37.6%). Similarly, Middleton (1998b) reported a series of 1,251 women seen since 1992 at the University of California at San Diego with a diverse array of implants who were referred for magnetic resonance scans because of suspected implant problems. He found that 15.35, 30.56 and 45.19% of these women required replacement implants within five years following augmentation, cancer mastectomy, and prophylactic mastectomy, respectively. This population, which was referred for problems, averaged 1.54 implants per breast (Middleton, 1998b).
These studies covered different follow-up periods, different kinds of implants, and different indications for implantation. Some were surveys, some record reviews, and some prospective observational trials. Each group was of unknown relation to the total group from which it was selected with respect to the events being studied. The results, therefore, cannot be compared scientifically. Although a quantative estimate is not possible, it appears that a significant number of women can expect additional procedures in the first five years after implantation. Women with implants for reconstruction and with gel-filled implants appear more likely to be at the upper end of the range of frequency.
Breast Reconstruction After Mastectomy
Perioperative and local complications are significant medical and patient events. Some complications are procedure related, that is, they would occur independent of the presence of an implant, and some are implant dependent and may vary with the characteristics of the implant, as noted in Chapter 3. This is particularly clear in reports that compare perioperative complications in reconstructions after mastectomy with matched mastectomy patients without reconstruction, in a sense "operative controls." O'Brien et al. (1993) in a short follow-up study, reported a similar complication frequency of 28% (N = 82) in 289 mastectomized women who were not reconstructed, compared with 31% (N = 35) in 113 women who were reconstructed, primarily with subpectoral expanders, after mastectomy. Most perioperative complications were the same, but seromas requiring one or more aspirations were present in 19% (N = 55) of those without reconstruction and only 3% (N = 3) of those with reconstruction. Implant-related problems occurred in 14% (N = 16) of the reconstructed women, including eight who required explantation (O'Brien et al., 1993). Vinton et al. (1990) in a study primarily about immediate, surgical complications, reported a similar total complication rate of 48% in 305 women undergoing modified radical mastectomy without reconstruction and 37% in 90 women with mastectomy and immediate reconstruction, primarily with expanders. Again, seromas were more frequent in the nonreconstructed group (30% versus 13%), and the reconstructed group had a 6% prosthesis complication rate with 4% requiring explantation (Vinton et al., 1990).
With respect to total short-term complications in reconstruction, these reports suggest that the implant may prevent seromas. Other complications such as infection, hematoma, and epidermolysis or skin necrosis occur with about equal frequency in women undergoing mastectomy who have implants and those who do not. The frequencies of early complications in implanted and nonimplanted women after mastectomy are roughly equal in these reports, but implant related complications are underestimated because of the short follow-up. Comparisons such as these are not possible in augmented patients because there can be no operative controls. In reconstruction after mastectomy, surgery is a precondition and the avoidable risk is only the implant-dependent fraction; in augmentation, surgery is not a precondition, and the risks of silicone implants may not be as separable from the operative risks. Implant technology designed to minimize risk is important in both instances.
Similar results are reported from case series of implant patients. Noone et al. (1985) reported on 85 women undergoing immediate reconstruction after mastectomy with saline and double-lumen implants with
short follow-up and the usual complications of skin necrosis (15%), seroma or hematoma (12%), extrusion (4%), infection (2%), and severe contracture (11%)or 44% overall15% of which (contracture, extrusion) were clearly implant dependent. In addition, secondary surgery later was required for open capsulotomy and explantation in 14.4%. Francel et al. (1993) reported 57% revision surgeries with permanent saline implants or expanders in immediate reconstruction after mastectomy and 30% in delayed reconstruction, with minor complication frequencies of 8.1 and 14% respectively, and implantation failures of 3.5% in both groups. Eberlein et al. (1993) reported 19 (27%) secondary surgical procedures (replacement or capsulotomy) and 8% prosthetic loss in 71 women with submuscular double-lumen implants after mastectomy, and Bailey et al. (1989) reported 18% implant loss in 165 women reconstructed with submuscular expanders or gel implants. Crespo et al. (1994) reported 115 consecutive implant reconstructions at the time of mastectomy using McGhan double-lumen smooth implants. Secondary surgery was performed in 20% of these women, and there were 5% infections, 8% explantations, 11% seroma or hematomas and 3% tissue flap necroses (Crespo et al., 1994).
Gylbert et al. (1990a) reported breast reconstruction in 65 women with randomly selected gel or saline implants followed for an average of six years: 6 of 37 (16%) patients with saline implants required replacements because of deflations, and three other operations were needed for misplacement, severe contracture and extrusion (8%). This study was primarily of contracture and is reviewed again later in this chapter. Using expanders and gel implants for immediate and delayed reconstruction, Slavin and Colen (1990) had an overall complication rate of 60% (among them, 15% seromas, 13.3% skin necrosis, 8.8% extrusion, 6.7% infection) in 60 consecutive immediate reconstructions involving expanders. Kroll and Baldwin (1992) had 23% "failures" (poor aesthetics or failure to complete reconstruction) at 22 months' follow-up in 87 women reconstructed immediately with expanders, followed by permanent replacement with polyurethane or other gel-filled implants.
Schlenker et al. (1978) studied 89 women with immediate or delayed reconstruction after simple mastectomy for fibrocystic disease over 6 months to 12 years. They removed implants in 28% of these patients for infection, extrusion or necrosis. Using primarily the Mentor 1600 inflatable saline implant, Schuster and Lavine (1988) reported 98 women undergoing immediate submuscular reconstruction after subcutaneous prophylactic mastectomy over a nine-year period. Eighteen patients suffered tissue loss, and there were three extrusions, among a number of less troublesome complications (Schuster and Lavine, 1988). In a study of wound complications in implant or expander immediate breast recon-
structions in 112 women, Furey et al. (1994) observed complications in 25 patients (22.3%) and removed 8 implants; other complications were not reported. Camilleri et al. (1996) reported 111 consecutive women reconstructed using the Becker (reverse double-lumen, gel outside) permanent expander with an average follow-up of only one year. Complications more typical of expanders, such as wound dehiscence (8%) and filling port failure (6%), occurred in addition to contracture (9%), expander infection and removal (5%), skin flap necrosis and expander exposure (5%), and other sequelae such as pain on expansion (20%). Despite these complications, 89% of women expressed satisfaction to the plastic surgeon on follow-up (Camilleri et al., 1996).
Gibney (1987) using CUI or Heyer-Schulte expanders for reconstruction in 65 women with three to seven years of patient follow-up, reported 5.8% of breasts with contractures, 2.5% with infections, and 4.5% with deflations, resulting in loss of the implant in 4.6%. Mandrekas et al. (1995) compared 19 women with immediate to 25 women with delayed reconstruction using subpectoral smooth tissue expanders after cancer mastectomy. The longest follow-up was seven years, but most complications were assessed by one year. These included one seroma, one infection, one skin necrosis, one valve deflation, ten Class II-IV contractures and two malpositions16 complications in 15 (34%) women, more frequent in delayed reconstruction. There were no rheumatic complaints (Mandrekas et al., 1995). Mahdi et al. (1998) carried out a prospective trial using the McGhan reverse double-lumen expander in 16 immediate and 4 delayed subpectoral reconstructions followed for an average of 10.1 months. There were seven reoperations for correction of placement, one hematoma, and two filler port problems. Follow-up was insufficient to evaluate rupture or contracture (Mahdi et al., 1998).
Spear et al. (1991) reported 76 women with 89 double-lumen implants for immediate reconstruction, randomized to 16 mg methylprednisolone in the outer saline lumen or to controls. Except for a lower frequency of contracture in the steroid group, the two groups were comparable three years after implantation. There were a total of 38 operative revisions, 2 significant infections, 3 extrusions, 16 fluid collections, 6 instances of skin necrosis and 26 Class II-IV contractures. Spear and Majidian (1998) subsequently reported 171 immediate postcancer mastectomy reconstructions with textured McGhan expanders in 142 women. Expanders were mostly replaced by textured saline implants, and follow-up after completion of reconstruction averaged 19 months. There were 14 (8.1%) skin necroses, 2 (1%) hematomas, 6 (3.5%) infections, 8 expanders and 11 implants (6.4%) requiring replacement and 5 (3%) significant capsular contractures. Equal or better results using textured expanders were reported by Fisher et al. (1991; see also Maxwell and Falcone, 1992; Russel et al., 1990; Beasley,
1992 who reported infrequent loss of expanders and rare contractures). In a review of a large experience with saline expanders (currently the most widely used technology for reconstruction according to the 1997 American Society of Plastic and Reconstructive Surgeons [ASPRS] survey), Woods and Mangan (1992) implied that results at the low end of reported complications could be achieved through experience and care.
As earlier, these studies include a number of variables. However, based on these reports, it appears that women, historically, could expect early (postoperative) complications, up to 30%-40% after reconstruction with implants, because reconstruction typically involves a significant surgical procedure to begin with (i.e., mastectomy for breast cancer). In addition to the surgical complications from mastectomy and implantation, there are the usual complications that depend on the presence of the implant.
Using Mentor implants for submammary augmentation, with a follow-up of four or more years for 87%, Capozzi (1986) reported 3.4% of breasts with contractures, 3.4% with deflations at intervals from nine months to seven years, and 100% satisfaction in 100 women between 1976 and 1985. Cocke (1994), using Heyer-Schulte-Mentor saline implants in 75 women for augmentation, mostly submammary, followed for 1.5 to 13 years, reported 29% (N = 22) secondary surgeries and 52% (N = 39) complications (23 contractures requiring treatment). McKinney and Tresley (1983) reported a series of 58 Women using Heyer-Schulte saline implants in the submammary position, with a number of complications including deflation (N = 9), infection (N = 4), capsules (N = 14), and hematomas (N = 9). In a letter report, Bell reported 10 deflations on average at 32 months of the same implant model in a series of 193 women (Bell, 1983).
Others have reported very infrequent failures. Mladick (1993) summarized results from his experience with saline augmentation over 17 years in 1,327 women with 9.1% secondary surgery. Most of these implants were modem RTV saline inflatables, but high termperature vulcanized (HTV) saline implants had been used earlier. Although the average follow-up was short, only 28 months, as is often the case, the deflations of the older implant models were 37.7% compared to the 1.33% for the more recent model, and complications were infrequent, mainly contractures in 1.1% of breasts, and no infections (Mladick, 1993). Frequent deflations (5-8%) were reported with the early saline models by others (Grossman, 1973; Regnault et al., 1972). Lavine (1993) reported placing 2,018 saline implants, with 4.2% of patients needing revisions, 1.1% Class III-IV contractures, and 2.3% deflations of all implants, but only 0.56% deflations of recent model Heyer-Schulte inflatables. The follow-up of these implants
ranged from 6 months to 13 years (Lavine, 1993). These reports also include a number of different variables, but they generally find a lower complication frequency, more consistent with the overall reoperation frequencies cited earlier, which were based mostly on results of women with implants for augmentation.
The important events for the safety of breast implantation are those that require significant interventions and seriously detract from the desired cosmetic objective. These include gel implant rupture (especially extracapsular) or saline implant deflation, severe contracture, infection, significant hematoma, severe and continuing pain, granuloma and axillary adenopathy, and implant displacement and extrusion. These are events that may require surgical revision, extensive medical or surgical attention or explantation, or leave the patient with deformity and discomfort. The occurrence of individual local and perioperative complications varies enormously among reports. Differences in complication frequency result from multiple factors: (1) individual unexplained variability in biological reaction to the device (e.g., fibrous capsular contractures); (2) differences in women's ages, physical conditions, habits, comorbidities and indications for implantation; (3) differences in types of implants and their physical and chemical characteristics, as described below and in Chapter 3; (4) differences in the design, adequacy, and reporting of clinical and basic research that may distort the true biological and medical picture; and (5) variable techniques and skills of surgeons and other medical personnel (e.g., operative techniques and skill and/or coincident medical interventions such as antibiotics, antiseptics, closed capsulotomies, steroids, treatment for cancer, and others).
In the discussions that follow, evidence for the contributions of these factors to a particular complication or the frequency of complications is reviewed. In some instances, a role is likely but speculative. In other instances, there are data to support objective statements, at least of limited or suggestive evidence of an association. Although a great deal has been learned, much more work on biologic variation is needed to fully understand the influences of this factor. As noted earlier, there are differences in the frequency of complications in women who receive implants for augmentation and for reconstructions and in those with immediate and delayed reconstructions. Although age may not influence most complications, the amount of body tissue and fat available for implant coverage, habits such as smoking or alcohol abuse (which could affect tissue viability), and significant medical illness (diabetes) are reported to make a difference (Cohen et al., 1992). Implant types and characteristics are im-
portant in a number of ways, as noted here and Chapter 3. Deficiencies in design and reporting of research may result in confusing or even misleading or incorrect information concerning complications. These problems are mentioned elsewhere in this report. Differences in surgical skill or technique may also play an important role. It seems intuitively reasonable that this should be so, given the great variation in reported results, which often appears to have no other obvious explanation. Although few studies explore these factors, techniques, and skills, some reports claim that they are critical in explaining frequencies of hematoma, infection, contracture, or other complications that differ greatly from the average values cited in the medical literature (e.g., Freeman, 1967; Mladick, 1993). The roles of some of the specific medical interventions are detailed further below.
In addition, several major operative approaches to implant placement are described in the medical literature (Salomon and Barton, 1997). The operative incision can be made in the axillary fold (and the implant placed under direct vision or endoscopically), in the circum- or periareolar position, or in the inframammary fold. The implant can be placed subcutaneously, under the mammary gland on the chest muscles (submammary, subglandular), or under the chest muscles (submuscular, subpectoral), depending on the status of the breast and a number of other factors including surgeon-patient preferences. The axillary incision is generally not visible unless the arm is raised; the inframammary incision lies under the breast and is covered by normal clothing. Circum- or periareolar incisions are usually around the inferior pole of the nipple-areolar complex from 9 o'clock to 3 o'clock. Some of these approaches may give better operative exposure for control of bleeding or may allow greater ease of insertion of bulky implants. They may also make operative scars, which can be unsightly in 2 to 5% of patients, less visible (Baker, 1992). Some concern has been expressed also that periareoloar incisions may be more likely to interfere with subsequent lactation and breast feeding (see Chapter 11). The periareolar approach may interrupt some lactiferous ducts, but inferior pedicle mammaplasty interrupts ducts substantially, and many women are said to breast-feed successfully after this procedure. Some have speculated that any interference with breast feeding may be due to compression by the implant. In any event, no conclusive evidence was found that these different surgical approaches have significant influences on complications related to the safety of breast implants.
Implant Rupture and Deflation
All silicone gel implants are subject to the bleed or diffusion of gel fluid composed of relatively low molecular weight linear and cyclic sili-
cone compounds through the implant silicone elastomer shell. The compounds range in molecular weight from 5,200 to 400,000 but are primarily less than 25,000 or 10,000 in implants without or with barrier shells. The higher molecular weight compounds may be from uncross-linked shell silicones (Varaprath, 1991, 1992). The large, highly cross-linked poly- (dimethlysiloxane) (PDMS) molecule of the gel cannot diffuse through the shell, and gel does not appear outside the implant unless there is a physical passage caused by a breach in the integrity of the shell. Diffusion is potentially important if silicone fluid or other substances inside the implant are toxic, if low molecular weight compounds can permeate the capsule and get into the circulation or into local lymph nodes with adverse effects, or if the fluid contributes to local reactions such as capsule formation, infection, or local effects that have systemic consequences. As the implant ages, silicone penetrates intact capsules and appears in breast tissue outside the capsule more often. This does not necessarily relate to flank ruptures or result in granulomas or adenopathy, however (Beekman et al., 1997a). These questions are taken up elsewhere in this report.
Factors Contributing to Loss of Shell Integrity
Silicone gel fluid is regularly found on and outside the shells of gel-filled implants. Implant rupture, a loss of integrity of the implant shell of varying severity, is diagnosed only when silicone gel itself is present outside the implant. This may occur, at one end of the spectrum of severity, through tiny flaws or pinholes in the shell, such as these caused by inadvertent needle sticks during suturing at implantation (Goldwyn, 1969), open capsulotomy, or other surgery, or through injection, needle biopsy or aspiration of seromas and hematomas. Wrinkles or folds are observed in 15-67% of measurable gel or saline implant shells by methods such as palpation, mammography, or observation at explantation (Frankel et al., 1995; Ganott et al., 1992; Gylbert et al., 1990a; Rolland, 1989a). Ex-plant cases may not be representative and are probably high-end estimates. Tears can occur because a shell distorted by folds abrades itself through the continuous motion of the breast and implant on the chest wall or through muscular contraction over a submuscular implant (Schmidt, 1980). Such abraded areas can be detected by scanning electron microscopy (Young et al., 1996a). Older implants and those with tight capsules have been noted to have significant distortion, folding, and often calcification which suggests ample opportunity for wear (De Camara et al., 1993). Some of these defects may not be visible to the eye, but they have been characterized by scanning electron microscopy that can define small fold flaws and suture needle holes (Brandon et al., 1997a,b).
The breast may be subjected to considerable compressive force dur-
ing squeezing maneuvers to break fibrous capsules in closed capsulotomies (Gruber and Friedman, 1978) and when greater-than-necessary compression is used during routine film mammography or mammography modified for better visualization of breasts with implants (Eklund et al., 1988). Both of these maneuvers have been associated in some reports with implant rupture or deflation and rarely with other complications such as infection, gel migration, silicone granulomas and exudation of gel from the skin and nipple, or conversion of intra- to extra-capsular rupture. During mammography, this is very unusual and rarely, if ever, has been proved conclusively and should not discourage mammographic screening for breast cancer (Addington and Mallin, 1978; Andersen et al., 1989; Apesos and Pope, 1985; Argenta, 1983; Bassett and Brenner, 1992; Beraka, 1995; Brandt et al., 1984; Cocke, 1978; Cohen et al., 1997; De Camara et al., 1993; Edmond and Versaci, 1980; Eisenberg and Bartels, 1977; Eklund, 1990; Feliberti et al., 1977, Goin, 1978; Gruber and Jones, 1981; Hawes, 1990; Huang et al., 1978; Hueston and Hare, 1979; Laughlin et al., 1977; Pay and Kenealy, 1997; Pickford and Webster, 1994; Renfrew et al., 1992; Robinson et al., 1995; Scott et al., 1988; Wilflingseder et al., 1983; Williams, 1991; Zide, 1981).
Recurrence of contracture after either open or closed capsulotomy is substantial when patients are followed long enough and has been reported to range from 23 to 89% (Baker et al., 1976; Brandt et al., 1984; Burkhardt et al., 1981; Hetter, 1979; Hipps et al., 1978; Little and Baker; 1980; Moufarrege et al., 1987; Vecchione, 1977). Complications after closed capsulotomy are reported to reach 10%-12%. and primarily involve distortion, displacement, rupture (with migration of gel), infection, and bleeding. Rarely, severe persistent pain and extrusion, among other complications, may also occur (Gruber and Jones, 1981; Laughlin et al., 1977; Nelson, 1980a). Such data support those who question the wisdom of this procedure. In a follow-up to his survey of closed procedures, Nelson (1981) reported a complication prevalence in 5,579 open capsulotomies of 6.24%, primarily hematoma, displacement, and infection.
Open (operative) or closed capsulotomies have significant safety implications because they have historically been performed, often repeatedly, on women with contracture of the implant fibrous capsule. These procedures may be performed in a third to almost all cases, depending on surgeons' and women's tolerance of this complication and confidence in the effectiveness of the procedure (see above references and those in the discussion of contractures). Peters et al. (1997) reported that in their series of 100 consecutive explants, 36% had had at least one closed and 54% at least one open capsulotomy. Baker (1975) commented that he performed open capsulotomies in all Class IV and 50% of Class III contractures and had 40% recurrences. He performed closed capsulotomies in cases with
any firmness, repeated without limitation, and had 31% recurrences (Baker, 1975). Moufarrege et al. (1987) reported doing multiple closed and open capsulotomies on 82% (134 out of 164) of women with contractures, 28% of their series of 482 women who received silicone gel implants between 1973 and 1978. They noted that 18% of women with contractures were satisfied despite contracture and declined further intervention. Recurrences after closed capsulotomies occurred in 67% on the first attempt, 80% after the second attempt, and 90% after a third attempt. Similarly, recurrences after open capsulotomies occurred in 54% after one, 56% after two, and 71% after three attempts in the same breast. More than three attempts to relieve contracture by capsulotomy were fruitless (Moufarrege et al., 1987). In view of women's tolerance for contracture without seeking medical attention, the subjective nature of contracture assessment by surgeons, the loss to follow-up of women with implants, and the varying periods of follow-up for a problem that continues to accumulate over time (all noted elsewhere in this chapter), reports of the frequency of contracture recurrence after capsulotomy are probably underestimates.
Gruber et al. (1981) wrote that they performed open capsulotomies in 72-100% of breasts with Class III and Class IV contractures. Nelson's (1980a) survey of closed capsulotomies found 30,001 of these procedures in 114,617 augmentations. Capsular contracture-related procedures made up 27.5% of secondary procedures in augmentation and 14.3% in reconstruction in the McGhan (1998) AR90 study. Other sources of trauma, ranging from gunshot wounds to automobile accidents to tight hugs, may also burst the shell (Cohen et al., 1997; De Camara et al., 1993; Dellon et al., 1980; Wise, 1994a,b). Some shells are broken or abraded during insertion. Some shells that are diagnosed as ruptured are actually broken on explantation (Slavin and Goldwyn, 1995).
At the most severe end of the spectrum of loss of integrity, the implant shell can become completely disrupted and float in a pool of silicone gel and fluid. In other instances, only small quantities of gel may escape through small holes such as those described earlier. If the fibrous capsule surrounding the ruptured implant is intact, it can contain the silicone in more or less the same shape and to the same effect as the implant shell. This occurs in what is called an intracapsular rupture. Such a rupture is often asymptomatic, without cosmetic effect and imperceptible on physical or radiological examination. If the fibrous capsule is weakened and bulges, a ruptured (or intact) implant can adopt, or conform to, this new shape. If the capsule has lost its integrity at some point, varying amounts of silicone gel can escape the capsule entirely. This gel can either rein-capsulate close by in the breast tissue (Argenta, 1983; Mason and Apisarnthanarax, 1981) or migrate along tissue planes to other parts of the body, usually the axilla and chest or upper abdominal wall (Aim and
Shaw, 1994; Goin, 1978; Hueston and Hare, 1979). Occasionally, gel migrates to even more distant sites, including inside the thoracic cage, down the arm, or into the groin. Rarely, this causes damage to structures such as neurovascular bundles, the formation of granulomas and cysts, and/or the breakdown of overlying tissues (Capozzi et al., 1978; Edmond and Versaci, 1980; Foster et al., 1983; Hirmand et al., 1994; Kao et al., 1997; Masson et al., 1982; Persellin et al., 1992; Sanger et al., 1992; Teuber et al., 1995b, 1999; Walsh et al., 1989). Extracapsular rupture can be present with implants that are completely ruptured and collapsed, or it can be associated with implants that are ruptured but not collapsed (Middleton, IOM Scientific Workshop, 1998).
Rupture of saline implants is referred to as deflation. Because saline, unlike silicone, is a physiologic solution and is rapidly absorbed in tissue, it does not remain or accumulate where deposited. The collapse and loss of aesthetics of the implant generally occur over one to two days (Rheingold et al., 1994) but can be quite gradual with small perforations or valve leaks. These leaks may cause deflations (as few as two to three drops over 12 hours in vitro), which may become apparent slowly over as long a period two years and are observed as partial deflations (Dickson and Sharpe, 1987; Nordström, 1988; Peters, 1997; Rapaport et al., 1997). Deflation, even partial, usually should not pose a diagnostic problem except in multilumen implants where the deflation of a small outer saline-filled lumen may, as was the designer's original intent, escape notice. According to Rees et al. (1973b) water (but not salt) may pass through the silicone elastomer shell in response to an osmotic gradient despite the insolubility of silicone in water, but usually there is no reason for disequilibration because at physiologic salt concentrations the osmotic pressure is the same on both sides of the shell. If the shell and valves remain intact, only small increases or decreases in volume over time are reported (Cederna, 1996). For unknown reasons, larger volume changes have been reported to develop over the years in rare cases (Botti et al., 1994; Robinson and Benos, 1997). Volume changes in intact saline implants might reflect correction of an osmotic imbalance due to variation in human body fluid and saline filler osmolality by movement of water into the implant lumen (Frisch, 1997, 1998). In vitro experiments found that volume changes pursuant to osmotic gradients occurred slowly, over six months to a year. Subjecting the implant to pressure, which might occur especially in the submuscular position or with severe contracture, accelerated the movement of fluid out of implants. This might explain some in vivo changes (Rubin, 1983), or in some cases with inflatable implants the valve leakage could be so slight that it was not apparent and partial deflation was mistakenly ascribed instead to movement of fluid through the shell (possibly what happened in Cederna, 1996, and other studies).
Needle sticks at implant placement surgery are a significant cause of early deflation although, as noted, this may not be apparent for several months. In one study, about 1% of implants sold were returned because of perioperative deflation. Small needle holes were discovered in about 7% of these (Rapaport et al., 1997). Small holes were found in 75% of saline implants involved in one series of deflations (Rheingold et al., 1994). Abrasion and compression cause deflation of saline implants in ways similar to their effect on gel implants. However, saline implants lack the solidity and supportiveness of the gel interior and the lubrication of silicone fluid on the shell surface. They are therefore more prone to wrinkling, fold flaws, surface abrasion, and consequent deflation. Fold flaws have been associated with underfilling of saline implants, which causes wrinkling, especially of the upper part of the implant. These fold flaws have been cited as an important cause of deflation (Gylbert et al., 1990a; Lantieri et al., 1997; Worton et al., 1980). Even with overfilling or attention to proper filling of newer implants, however, a smooth saline implant or a non-adherent textured implant may still sag toward the bottom of a large tissue pocket in the breast, producing some collapse and vertical, so-called traction wrinkling of the superior part (Tebbetts, 1996). As described earlier, these implant problems are common, and submammary or subcutaneous placement of saline implants often allows these wrinkles to become easily visible through the skin. Skin wrinkling over a wrinkled implant can be quite pronounced and require corrective interventions. It has been reported with a frequency of 3.3-5% in sub glandular, smooth saline implants for augmentation (Cocke, 1994; McKinney and Tresley, 1983; Mladick, 1993). In reconstructed patients, skin wrinkling with smooth saline implants has been reported to occur in as many as 26% of cases (Gylbert et al., 1990a). In two additional studies, skin wrinkling occurred with 14% of saline compared to 3% of gel smooth implants or 7.3% of saline compared to 2.1% of gel textured implants (Asplund, 1984; Handel et al., 1995).
The valves and valve-shell interfaces of saline implants and expanders can also leak or become incompetent under increased pressure in the implant. Some have reported average pressures of 4.5 mm of mercury (Hg) in implants in soft breasts and 11.5 mm Hg in hard breasts and pressures of as much as 320 mm Hg during closed capsulotomies (Jenny, 1980). In other instances, such pressures are known to reach levels on the order of 93 mm Hg or 126 cm of water [H2O] or more, which is substantially greater than historic test standards (Caffee, 1993a; described in FDA, 1992b). Peters (1997) also demonstrated in vitro that valves that appear competent may leak saline slowly under pressure.
Detection of Rupture of Gel-Filled Implants
As noted above, intracapsular rupture of gel implants can go unrecognized. There may be no patient complaints and no physical diagnostic findings. The sensitivity of commonly used imaging technologies, such as film mammography, is reported to be around 50%, and some report a sensitivity as low as 15-20% (Ahn et al., 1994a; Bassett and Brenner, 1992; Robinson et al., 1995; Samuels et al., 1995), although there are some who claim it is much higher (Cohen et al., 1997; also see Chapter 12). Ultrasound may detect more ruptures. Magnetic resonance imaging is generally reported to have a sensitivity of around 90% in experienced hands, but this technology is currently too costly and time-consuming for routine screening. Physical diagnosis of extracapsular rupture, when the shape or feel of the gel mass has changed, is much easier, although occasionally such changes can escape detection (Edmond and Versaci, 1980). The changed contour of the gel is also observed on film mammography, although contour irregularities may be secondary to accommodation of an unruptured implant to, or herniation through, a capsulotomy defect. Presence of gel is reported outside the capsule in about 12-26% of ruptures (Ahn et al., 1994a; Berg, et al. 1995; Frankel et al., 1995; Gorczyca et al., 1994a; Middleton, 1998b). It has been as high as 35% in selected series (Andersen et al., 1989). These figures generally refer to more fragile second generation implants and come from case series of patients who present because of problems and, therefore, may overstate complications. There are also reports of series of 19 or 30 instances in which there has been implant rupture without any extracapsular movement of gel (Beekman et al., 1997a; Malata et al., 1994a) or a low proportion of extra-capsular movement, e.g., 5.7% of ruptured implants (Peters et al., 1997). Extracapsular rupture is also more easily detected on ultrasound and MRI.
The diagnosis of rupture of a gel implant is important because the release of silicone gel and fluid into the tissues may result in local complications. An intracapsular rupture may become extracapsular, and both are generally, but not always (Hardt, IOM Scientific Workshop, 1998), agreed to be an indication for explantation (Young., 1998). Moreover, rupture should be anticipated at some point since implants have a finite life span, although what this might be with current models is not certain. Careful explantation and direct visual examination are the standard for diagnosis of silicone gel-filled implant rupture, both unsuspected or silent, and for confirmation of rupture. Explantation allows only a retrospective or confirmatory diagnosis. It is not a prospective means of resolving the question of presence or timing of rupture in an individual patient.
Capsulectomy along with explantation is recommended if the implant is ruptured (Rohrich et al., 1998b; Young, 1998). Although Thomson (1973) reported that capsules left at explantation were reabsorbed, this is not necessarily the case. Retained capsules can interfere with adequate compression for mammography, can confuse mammographic diagnosis of breast cancer if calcifications are present (Hardt et al., 1995a; Peters et al., 1995d) and may lead to further complications necessitating additional surgery (Copeland, 1996; Hardt et al., 1995a; Hayes et al., 1993; Peters et al., 1995d; Rockwell et al., 1998; Stewart et al., 1992).
Implant Shell Strength
A number of additional factors could contribute to implant rupture. As discussed earlier, some shells were thicker and stronger than others. Different implants had valve or valve attachments that were more or less secure. Manufacturers developed elastomers with various performance characteristics, that also affected implant integrity. The manufacturing process itself has been cited as a possible contributor to deflation by Rubin (1983) who noted changes by scanning electron microscopy such as accentuation of defects in the surfaces of saline implants subjected to 3/4-pound tension.
Although not universally agreed upon, it appears that the elastomer shell is relatively stable in vivo once the effects of gel fluid permeation, that can decrease the tensile strength of shells not protected by barriers to gel fluid diffusion by around 30%, are taken into consideration (FDA, 1992a, p. 75). The shell appears to maintain strength, rupture resistance, and bulk material stability as determined by measurements of mechanical and chemical properties and cross-link density over many years of implantation. In Silastic II explants, elongation values were greater than those recommended by the American Society for Testing and Materials (ASTM) for 12 years (ASTM, 1983a; Brandon et al., 1998a,b). Analyses of elastomer by nuclear magnetic resonance (NMR) have reported or suggested some chemical changes, but the quantitative effects of these on shell strength were not assessed (Pfleiderer et al., 1995; Picard et al., 1997). A number of reports have cited decreases of modest proportions in silicone elastomer strength characteristics either in materials testing or in non-breast implants in mid-length, one-to-two year studies (Langley and Swanson, 1976; Leininger et al., 1964; Roggendorf, 1976; Swanson and Lebeau, 1974). Some more marked decreases were found in one long-term study of pacemaker leads (Dolezel et al., 1989), but a shorter-term in vitro investigation of the material found physicochemical and strength properties were maintained (Kennan et al., 1997).
The resistance to rupture and deflation observed in some studies is
quite good, but there is a considerable experience that describes rupture and deflation prevalence as increasing, sometimes markedly, with time after implantation. This increased prevalence may relate to repeated stressing, folding, and abrading of an implant, especially with tight capsules and incidental trauma of various kinds (Cohen et al., 1997; De Camara et al., 1993; Gutowski et al., 1997; Malata et al., 1994a; Peters et al., 1994a, 1996; Phillips et al., 1996; Young et al., 1996a, 1998). Shell strength and rupture resistance can be measured by tensile strength, elongation, tear resistance, and modulus, which are measurements (and standards) of the ability of an elastomer to stretch and withstand measured stresses and tearing forces. Greenwald et al. (1996) measured the shell tensile strength (resistance to stress and strain) and modulus of 25 smooth gel implants explanted after 2 to 18 years and found consistently less strength over time in older implants with considerable scatter of results. As the authors point out however, the explants were not compared to control devices, and the makes and models were not described; the implants were those removed by one of the authors from consecutive women between 1991 and 1994 (Greenwald et al., 1996). Phillips et al. (1996) also tested a series of 29 explants, 4 months to 20 years of age, and found considerable variation. They concluded that shell strength diminished with age, although they also noted considerable differences over time among implants of various types and manufacturers. They tested an unused Silastic II breast implant to validate their methods, but otherwise had no controls. Van Rappard et al. (1988) testing implants of the same make, noted a decrease in bursting strength with implant age. Marrota et al. (1998) compared explants with unused implants made by the same manufacturer; unfortunately, implant types were not specified in this short report, but explants in general had 25% less strength.
Addressing these problems, Wolf et al. (1996, 1997) and Brandon et al. (1997a,d, 1998a-f, 1999) described measurements of a large series of ex-planted and unused (some same lot) control gel implants of ages up to 28 years, manufactured primarily by Dow Corning, both as is and after gel fluid extraction. They also reported that gel fluid permeation in vivo decreases various parameters of shell strength. In the aggregate, physical properties of shells weaken but are not dramatically changed by length of implantation when measured after extraction of gel fluid, nor does scanning electron microscopy detect significant, visible surface deterioration in regions of nonwear even after 28 years in vivo. Different types and manufacture of implants have quite different physical parameters. Shell thickness varied twofold when measurements were made in a number of different places. Most striking was the lot-to-lot variation in implants of the same type from the same manufacturer. For example, tensile strengths of Silastic I and Silastic II models from different lots varied 3.5-fold and
from minimum to maximum between the two types by fivefold. Strength parameters of two 28-year-implanted Silastic0 implants' were decreased 20-40% by gel fluid permeation, but both extracted and nonextracted values fell within the range of control Silastic I (same elastomer) shells and were higher than ASTM standards (Brandon et al., 1997a-f, 1999; Wolf et al., 1996 and 1997; and see also Frisch and Langley, 1985; Swanson, 1973; Swanson and Lebeau, 1974).
Chawla and Hinberg (1996) also found that shells implanted for up to seven years maintained surface integrity, and neither they nor Ratner et al. (1994) were able to demonstrate free silica at the shell surface by any of a number of techniques. Peters (1981) tested a small number of gel and inflatable prostheses, some explanted, some unused, with a compression (burst) test and found no pattern, extreme variability among implant types, and lot-to-lot variability. Rupture occurred at pressures ranging from 0.62-10.8 pounds per square inch (psi) considerably lower than the pressures measured in various kinds of closed capsulotomy, 10-15 psi, (Gruber and Friedman, 1978) which was the objective. Different makes and models of implants clearly will have different tolerances for closed capsulotomy (Lemperle and Exner, 1993). Curtis and Hoshaw (1998) plotted Dow Corning mechanical test data against duration of implantation from approximately 5-20 years and observed no degradation with increasing in vivo duration, although there was considerable scatter of re-suits.
These results reflect the nature of silicone chemistry and manufacture. They indicate that rupture will depend on the manufacturer, type and model and even the lot of saline and gel implants, as well as on underlying physical parameters such as designed thickness and chemical formulation. Any analyses of rupture resistance and shell strength or of rupture prevalence in cohorts of women should try to control for these confounding variables. As indicated earlier, rupture also will depend on distortion, wear, and stress in the body, untoward events, and perhaps other unknown factors.
Biomedical polymers, including silicone, may absorb or adsorb certain lipids and lipoproteins (Carmen and Mutha, 1972; Chin et al., 1971; Dong et al., 1987). One investigator has proposed that the absorption of lipids by an elastomer may weaken it over time (Caffee, 1993b). Although lipids were reported in Silastic sheet implants and in explant silicone elastomer shells at low levels, <0.2% (Pfleiderer et al., 1995; Picard et al., 1997), and at less than 1% by weight in slabs of silicone elastomer implanted in dogs for two years (Swanson and Lebeau, 1974), others have not found them (Chawla and Hinberg, 1996). In discussing this issue, Frisch claimed that lipid adsorption accounted for at best an early but stable weight increase of about 1.5-2% and very little, if any, loss of elas-
tomer strength. The effect of lipids, if present, on shell rupture remains hypothetical; only small decreases in physical strength of the elastomer were found in the dogs of Swanson and Lebeau (1974), and Swanson (1973) observed lipids generally below 1% with no relationship to duration of implantation or fracture of the elastomer in the elastomer of finger joints that had been implanted for up to five years.
The Frequency of Rupture and Deflation
Rupture reports have been based on findings at explantation, detection by various mammographic technologies (often confirmed, at least in part, by explantation) and descriptions or surveys of results in cohorts of patients. None of these is absolutely reliable. Explant series are the standard for diagnosing rupture and for confirming rupture suspected by other means, but they undoubtedly include some ruptures that occur during implantation and explantation procedures. Further, women undergoing explantation are a special subset of the population of women with implants, and even of the population of women with ruptured implants. Mammography series may be drawn from explant series, may involve groups of women with suspected problems, or may depend on a less-than-certain diagnosis, although MRI has an excellent record. Case series reported from practitioners or facilities have the advantage that they can be (unless selected in some way) more representative of the population of women with implants, but the detection of rupture depends on patient complaint, routine screening film mammography, or physical examination, all of which are subject to considerable error and lack of sensitivity (see, for example, data on mammography sensitivity and Malata et al., 1994a, who suspected clinically only 6 of 19 ruptures seen on explantation, including 10 intracapsular ruptures with complete shell disintegrations). The accuracy of very low rupture rates reported by manufacturers based on returns, complaints, legal proceedings and other sources probably suffers from these same problems of ruptures' going undetected and unreported.
The medical literature on rupture of gel implants includes reports of prevalence ranging from 0.3 to 77% (Beekman et al., 1997b; Berg et al., 1993, 1995; Chung et al., 1996; Cohen et al., 1997; Davis et al., 1995; De Camara et al., 1993; Destouet et al., 1992; Dowden, 1993; Duffy and Woods, 1994; Gabriel et al., 1997; Gorczyca et al., 1992; Harris et al., 1993; Ko et al., 1996; Malata et al., 1994a; Middleton, 1998b; Nelson, 1981; Netscher et al., 1995a; Park et al., 1996b; Peters et al., 1996; Phillips et al., 1996; Robinson et al., 1995; Rohrich et al., 1998a; Rolland et al., 1989a; Slavin and Goldwyn, 1995; Weizer et al., 1995; Yeoh et al., 1996; Young et al., 1996a, 1998). These are reports of either the percentage of women who have one or both
implants ruptured or the percentage of all the implants placed that have ruptured. Since patients may have two implants, only one of which is ruptured, rupture frequency by number of implants is usually lower. The list of references here omits some mammography series that could have been cited because they report low numbers, patients are often referred for problems and ruptures are not always confirmed by explantation (see Chapter 12). The very low value, 0.3%, was from a review of 307 women examined by ultrasound and thus subject to the level of sensitivity of this technology. Three confirmatory explantations and six confirmatory MRIs were performed in the 307 women. One explant was clearly ruptured. The other, which was discounted, was covered with sticky silicone and was probably a small leak. This implant probably should not be counted as intact. This would raise the prevalence to at least 0.6%, with another upward correction of unknown magnitude for the ruptures missed due to the lack of sensitivity of ultrasound.
Deflation does not pose the same problems in definition, or generate the same research interest or attention in the literature because its detection is not prone to the uncertainties associated with gel rupture, and release of implant contents is of little consequence beyond the loss of aesthetic effect. Because of the reliability of detection, saline deflation estimates are much firmer than those for gel-filled implants. Women do not always seek medical attention from their operating surgeon for deflation, however, so it should not be assumed that these complications are always discovered and reported. Gutowski et al. (1997) noted that women had not notified the responsible plastic surgeon of 10 of 55 deflations, and no intervention was taken for some of these; in this large survey, deflations occurred in 10.1% of women and 5.5% of implants. Lantieri et al. (1997) reported their experience with 709 implants in 407 women who responded to a questionnaire mailed to 454 women who had received saline inflatable implants in their two facilities in France and the United States between 1981 and 1995 (489 smooth Mentor and 220 textured Mentor Siltex breast implants). The indication for implantation was reconstruction, usually delayed, in 40% of breasts; augmentation in 41%; and replacement in 19%. There was no significant difference in deflation between smooth or textured implants, between submammary or submuscular placement, or by indication for implantation. The overall deflation prevalence was 6.6%, with an average follow-up of 7.1 years, and under-filling was highly significantly associated with deflation; fold flaws were observed in 81% of deflations. In this report of implants filled on average to 103.1% of design capacity, the Kaplan-Meier survival curve showed a deflation rate of less than 1% at 12 months (Lantieri et al., 1997).
Presence of deflation at the end of one year in the McGhan LST was 3.9% of patients receiving implants for augmentation and reconstruction
combined (McGhan Medical Corporation). Deflations varied between 1.33 and 37.7% of implants, depending on type, in the experience of Mladick (Mladick, 1993) and similarly between 0.56 and 20% of implants in Lavine's (1993) series comparing the 1600 with the 1800 (defective valve) Heyer-Schulte-Mentor model. In the patient cohorts cited earlier in the discussion of aggregate complications, deflation usually was not the focus of the report and varied between 2.6 and 16% over different periods of follow-up for a number of different inflatable saline implants and expanders. In a discussion of explantation of gel-filled implants, Robinson (1992) noted 0.6% deflations of 1,600 mostly recently placed saline implants of unknown model and manufacture. Rheingold et al. (1994) reported a survival analysis of 67.34% at slightly less than ten years. They commented that their ''study confirms the obvious: Inflatable breast implants deflate with time."
Rubin (1983) reported his own experience with saline inflatable implants. He found 29 deflations in 478 implants from three different manufacturers, or 6% over five years (not counting those [N = 14] noted to be defective at the time of implant surgery) and commented that he expected to see more deflations with time. In a survey of plastic surgeons with 528 responses, he found 1,181 leaking implants of 13,200 reported, or 8.95% over an unknown period of time (Rubin, 1983). Gutowski et al. (1997) reported an actuarial survival estimate at ten years of 90.2-95.2%. Some underreporting is probable since, as noted earlier, not all women seek attention from their surgeons for deflation (Rheingold et al., 1994; Gutowski et al., 1997). Other reports on saline implant and expander deflation are very variable. Some frequencies are quite high (e.g., 76%), but since they include early implants that are known to have had problems, they have little to teach about what can be expected in terms of deflation of modem saline implants and expanders (Bell, 1983; Williams, 1972; Worseg et al., 1995; Worton et al., 1980). Taking implant type and model into account and reviewing relatively recent representative case series, the committee concludes that modern first-year deflations might be of the order of 1-3% of implants and that this percentage would rise slowly with time (see Capozzi, 1986; Francel et al., 1993; Gutowski et al., 1997; Lantieri et al., 1997; McGhan (undated); Mladick, 1993; O'Brien et al., 1993; Schuster and Lavine, 1998; Vinton et al., 1990).
Estimates of gel implant rupture can start with measures from cohorts of women examined in inclusive follow-up reviews or from routine screening of populations. Gabriel et al. (1997) reported 5.7% ruptures in a 7.8-year mean follow-up of 749 women, 3.9% of 1,454 implants, based on chart review. An unknown number of ruptures undoubtedly went undetected (Gabriel et al., 1997). At 10.7 years, Harris et al. (1993) detected 22 ruptures (6.5%) in 336 gel-implanted women routinely screened by film
mammography, with 42% followed up with ultrasound. In view of the low sensitivity of film particularly, but also of ultrasound mammography (Ahn et al., 1994a; see also Chapter 12), a significant number of ruptures must have escaped detection (Harris et al., 1993). In the McGhan (1998) AR90 clinical study, the five-year cumulative risk of rupture suspected by symptoms or physical examination was 4% of augmented patients (2.6% of implants in augmented women) and 9.8% of reconstruction patients (6.3% of implants in women with implants for reconstruction). Similarly, at two years, a 1990 Dow Corning observational study reported no ruptures in 69% of an original cohort of 360 augmented women returning for evaluation (Bowlin et al., 1988). The prevalence of rupture detected on physical examination in the Mentor adjunct study (1992) was 1.3% at the three-year follow-up of about 8,000 women. The low sensitivity of these detection methods (and, especially for Dow Corning, the short follow-up) inevitably means an underestimate of implant rupture (Bowlin et al., 1998; McGhan Medical Corporation, 1998; Purkait, Mentor Corp, IOM Scientific Workshop, 1998). Destouet et al. (1992), Dowden (1993), and Peters and Pugash (1993) screened consecutive routine patients by film mammography, physical examination confirmed later for a subset by explantation, and ultrasound. Ruptures were detected, respectively, in 4.6% of women at 10 years, 4.5% at an unknown interval, and in 8% at 8.5 years.
Clearly in all of these instances, detection was attempted with low-sensitivity technology and significant numbers of ruptures were missed (Destouet et al., 1992; Dowden, 1993; Peters and Pugash, 1993). The ex-plant or selected series, on the other hand, presumably overestimates prevalence (Young et al., 1998). The survey of Nelson (1981) represents an interesting hybrid between these two study types, explant and routine follow-up or screening series. He reported rupture in 15.9% of 5,579 breasts undergoing open capsulotomy. The selection bias of this series is unknown. Of these, 57% had previously been subjected to closed capsulotomy. The duration of implantation was not reported (although it may have been a relatively few years given the usual time of onset of contracture), and the implants in question were most likely of the thin-shelled, second generation variety. Nevertheless, this represents a considerable number of ruptures identified by direct observation in a large group of women with presumably a cross section of implant models who were gathered because of only one problem and thus not subject to the more obvious bias of explant cohorts (Nelson, 1981).
Unless the increasing prevalence of implant rupture with age can be shown to correlate with specific kinds of implants having specific fragility (Peters et al., 1996), it seems reasonable that the relationship between implant age and rupture, noted earlier, reflects a real trend (Goldberg et al., 1997). Capsular contracture and implant distortion tend to continue
over time, although no studies were found that provided conclusive evidence of an association between contracture and rupture or deflation (Lantieri et al., 1997). Stress and incidental trauma also accumulate over time. Shells are nonuniform in thickness and strength even within the same lot, and they can be expected to weaken after implantation due to permeation by gel fluid. They may then gradually lose strength over more prolonged times. Thus, deflation and rupture can be expected, as the data show, to continue over time.
Authors cited above reported that 96% of gel implants ruptured at or beyond 10 years (De Camara et al., 1993); up to 63% ruptured at 12 years or more (Cohen et al., 1997); 62.5% of those in situ for 10 years ruptured (Malata et al., 1994a); 71.2% ruptured at 14 years, and 95.4% ruptured at 20 years according to a survival analysis (Robinson et al., 1995); increasing failure was reported between 10 and 15 years, and 50% ruptured at about 15 years (Beekman et al., 1997b), 69-70% ruptured at 6 years or more (Peters, 1994a), 49% ruptured (includes leaking) at 10 years (Rohrich et al., 1998); and frequency of pinholes and frank ruptures increased from 8% at 0-5 years to 61% at 10 years to 73% at 20 or more years (Young et al., 1996a, 1998). Different rupture percentages have been found according to implant vintage. First-generation implants (1963-1972) had no ruptures, second-generation implants (1972 to mid-1980s) were 95% ruptured at 12 years, and third-generation implants (late 1980s to date) had only 3.5% ruptures by 1992 (Peters et al., 1996). Other investigators have reported similar findings. In fact, they found no ruptures in third-generation implants (Francel et al., 1998). On the other hand, the large series (N = 161) described by Yeoh et al. (1996) of 4- to 27-year-old, primarily Dow Corning gel-filled prostheses, which had 25% ruptures, did not show a correlation of implant age with rupture. The analysis by Gutowski et al. (1997) indicated 5-10% deflation at ten years for saline implants of recent vintage. A very high, in fact unacceptable, frequency of deflation is reported for older models, as noted earlier.
It is not possible to predict current rupture frequencies or rates. Rupture depends on implant type, model (and lot), and manufacturer; implant physical characteristics; silicone gel fluid permeation, which weakens implants; and the many other factors that stress, compress, and abrade implants. The reported high rupture prevalences cited above reflect experience with a great many thin-shell, compliant gel models according to the dates of explantation from these reports. Clearly it is possible to build an implant of sufficient strength to endure a long time, as the experience with first-generation 1960s thick-shell, thick-gel implants shows (Brandon et al., 1999; Peters et al., 1996). It is unknown where the balance of sturdiness, durability, and cosmetic softness versus wear and stress in the body lies, and whether implants in current use with short histories will
also reach the end of their lives at 10 to 15 years as data seem to imply for older models. Silicone gel fluid permeation of the shell seems to have a deleterious effect on durability, more so at high percentages of extractable fluid. Contracture, which is more common in gel implants, may be a risk factor for rupture (Feng, IOM Scientific Workshop, 1998), although there is insufficient evidence for this (Lantieri et al., 1997). Gel-filled implants have a softer HTV shell than saline implants. These factors, and indeed some of the studies summarized in this section, suggest that rupture may be more common in current gel-filled implants than current saline-filled implants and expanders, other things being equal, although properly conducted new studies will be needed to resolve this point as suggested below. Neither the submuscular or submammary site of implantation nor the augmentation or reconstruction indications appear to be important variables for rupture.
Estimates of rupture frequency of gel-filled implants must consider the bias problems of explant series, which are likely to seriously inflate rupture rates while providing an accurate diagnosis through direct examination, against the advantages of studying a more random group of women with the associated problems of underreporting of rupture that are inherent in insensitive diagnostic tools. Experience relying more heavily on random case series or explant series that identified recent implants (see Bowlin et al., 1998; Destouet et al., 1992; Dowden, 1993; Francel et al., 1993; Gabriel et al., 1997; Harris et al., 1993; McGhan, 1998; Peters and Pugash, 1993) suggests a relatively modest number of ruptures. The size, sensitivity of detection, duration of implantation, and other problems of these studies also suggest the need for an upward adjustment or considerable caution in relying on them to predict current rupture experience.
On the other hand, the variation, selection bias, and inclusion of implants no longer in use in most explant series do not provide a good basis for an estimate of modern or future rupture frequencies. The ruptures of gel-filled implants with certain characteristics such as soft, thin shells and compliant gels with high levels of extractable gel fluid, referred to as second generation, are substantial but are not really at issue. Experience with these implants clearly implies that women who still have them will need them removed or replaced, but it cannot be used to predict modem implant rupture frequency. (The very high deflation rates of early HTV saline implants are also generally accepted.) Rupture rates or frequencies of modern implants will not be known until long-term, prospective observational studies of sufficiently large and random cohorts of women are completed, using sophisticated and reliable detection methods such as MRI by skilled and experienced investigators. The recent experience of
Brown et al. demonstrates that good agreement on ruptures defined by MRI can be achieved among experienced experts (Brown et al., 1999).
Preliminary and admittedly anecdotal reports from such experts suggest that the frequency of rupture in third-generation implants is much lower than the prevalence reported for second-generation implants (Middleton, IOM Scientific Workshop, 1998). The committee stresses that a determination of rupture (and deflation) prevalence or rate in such a way is of value, should include a determination of other complications and characteristics of implantation in a disciplined way, and should be carried out because of its implications for the safety of silicone breast implants. Such a study or studies will be valuable only for implants of the same types and with the same physical-chemical characteristics. If manufacturers change silicone implant formulations or shell thicknesses or modify gel-filled implants in other ways such as by changing gels, studies of these defined formulations would have to be carried out. Until such time, only a guess at the order of magnitude of gel-filled implant rupture can be hazarded.
Keeping in mind the results Of explant series of modem implants and other recent observations as noted above, the committee is of the opinion that, with a conservative guess at upward adjustment to account for underdiagnosis, a modest number (perhaps less than 10%) of modem gel implants will have ruptured by five years and that ruptures will continue to accumulate and prevalence will increase in ensuing years. Whatever the actual rupture prevalence or incidence, the safety implications of rupture and deflation noted here include the risks of additional surgery and anesthesia when explantation or replacement of an implant (and capsulectomy) is required to remove silicone or to address the loss of aesthetics in saline- and gel-filled implants. Operative interventions to treat granulomas and significant axillary adenopathy may also be required, though infrequently. There will also be the rare complications of gel migration and tissue damage at more distant sites and whatever health conditions may follow tissue exposure to silicone gel.
Implant Capsules and Contracture
Placement of an implant in the body causes a reaction in the tissues that varies depending on size, shape (Matlaga et al., 1976), and surface texture and porosity (James et al., 1997; Salthouse, 1984; Taylor and Gibbons, 1983). Other surface physical characteristics, such as charge and energy, chemical characteristics of the implant, location of the implantation site, and animal species studied also make a difference (Bakker et al., 1988; Ksander et al., 1981). These differences affect the ability of some materials to adsorb body proteins of various kinds, to activate certain
cells, to cause the production of cellular factors and cytokines that may influence cellular proliferation and other activities both locally and perhaps systemically, and to excite inflammatory reactions of varying extent, intensity, and type of cell participation. All implants, including all breast implants, provoke some form of this general foreign body tissue reaction.
The production of an enclosing capsule of fibrous scar tissue around the implant is an important component of the foreign body reaction and significantly affects the effectiveness of breast augmentation or reconstruction with implants. Contracture of this capsule, if severe, causes often painful and disfiguring squeezing and distortion of the implant (and overlying tissue). At the extreme, it is compressed into its smallest volume, a sphere, rather than the shape designed to achieve normal breast contour. The contributing causes and management of this variably occurring complication are important and incompletely resolved issues in breast surgery. It is the adverse event most frequently reported to the FDA (Brown et al., 1998).
The reaction to a foreign body is a normal part of the body's intrinsic defense mechanisms. Silicone elastomer, gel, fluid, or some other component of a breast implant might also provoke an immune response as part of the body's reaction to implantation. In this case, certain cells would recognize specific foreign molecules associated with the implant and would initiate immune cellular proliferation and the production of specific protein antibodies to these molecules, and/or proliferation of, and the specific targeting by immune cells to attack these molecules. Changes in the local tissue reaction and possible systemic effects would likely follow an adaptive immunological response in addition to a foreign body or innate response. The committee did not find scientific evidence for this, but such questions are discussed further in the following chapter.
Whatever the contributing causes, the breast implant capsule can be described by its numbers, orientation, and types of cells and cell products. Different types of breast implants will produce differing capsules. Some have speculated that damage to the breast during implant placement, which could lead to fat necrosis, mammary gland degeneration, and muscle atrophy, may also contribute to capsule thickness and contracture (Smahel, 1978b), although there is insufficient evidence to support or refute this. Of course, the healing of an operative wound is a part of the response to implantation. In elementary terms, the early response to tissue injury and to the presence of an implant consists of the migration and activation of many different cells including polymorphonuclear leukocytes (PMNs) and mononuclear cells. PMNs disappear within a few days, but monocytes, lymphocytes, and fibroblasts remain around the implant. Monocytes differentiate into macrophages, and some of these cells coalesce to become foreign body giant cells. Fibroblasts are stimulated to
proliferate and produce collagen, and new capillaries develop. An array of different kinds of lymphocytes may be present (Katzin et al., 1996; Miller, IOM Scientific Workshop, 1998).
Contracted capsules, like hypertrophic burn scar, contain greater amounts of glycosaminoglycans and proportionately more chondroitin 4-sulfate, characteristics of immature scar (Vistnes et al., 1981). The types of collagen (I, III, and v) in capsules surrounding different types of implants and expanders and of different degrees of severity and age are similar to the types in breast dermis and commonly observed cutaneous scar. Contracted capsules have more collagen than soft capsules and normal dermis (Marshall et al., 1989; McCoy et al., 1984). Gradually the process and capsule become mature and take on more of the appearance of normal scar tissue with few cells around smooth implants and a thicker, more cellular, chronic inflammatory appearance with less regularly organized fibrous tissue around textured or polyurethane foam-coated implants. The fibrous tissue capsule surrounding breast implants is described in a number of reports (e.g., Copeland et al., 1994; Emery et al., 1994; Hameed et al., 1995; Kasper, 1994; Lesesne, 1997; Luke et al., 1997; Raso and Greene, 1997; Smahel, 1977; Wickman et al., 1993; Wyatt et al., 1998; Yeoh et al., 1996).
On direct visual observation, the usual implant capsule is a variably thin, grayish, glistening membrane. Microscopically, smooth-surfaced gel or saline implant capsules may have a flattened unicellular lining or a layer of pseudoepithelial cells next to the implant, overlaid with a regularly and linearly oriented dense collagen network that progresses to looser, better-vascularized connective tissue merging with the surrounding breast tissue. Synovial linings are also seen around smooth implants. Macrophages, multinucleated giant cells, T lymphocytes, fibroblasts, and occasionally other cells such as eosinophiles and plasma cells, are seen in varying numbers, although the older capsules generally have fewer cells set amid regularly oriented fibrous connective tissue.
Capsules of textured implants are likely to have a palisaded secretory and phagocytic synovial multicellular layer without a basement membrane next to the implant, overlaid by a thicker, more disoriented, cellular and vascular connective tissue layer progressing to loose connective and adipose tissue in the surrounding breast tissue. Occasionally the synovial layer, especially in capsular infolds, assumes a papillary hyperplastic appearance (Hameed et al., 1995). This villous hyperplasia has been reported around 63% of textured implants at less than five years, decreasing significantly to 7% after five years (Wyatt et al., 1998). Capsular synovium is apparently of mesenchymal origin and is believed to be a reaction to the shearing movement of the implant. The synovium secretes proteoglycans, chondroitin 4-sulfate and keratan sulfate, which may lubricate the pros-
thesis-capsule interface, and there has been speculation that these substances may help to diminish contracture (Lin et al., 1994; Raso and Schulte, 1996). The synovial layer may diminish over time, as noted in Chapter 3, and as reported in a long-term study with findings of statistically significant decrease in synovial lining around textured and smooth implants over time (Wyatt et al., 1998). Synovium has been described around all kinds of breast implants including smooth saline implants (McConnell et al., 1997).
Fragments of the elastomer shell are seen in capsules around saline implants (Jenny and Smahel, 1981; Vargas, 1979), particularly from textured shells, but no droplets of silicone fluid. Silicone fragments, up to 1 mm in size, are found inside giant cells and in aggregates of giant cells called granulomas mostly in textured saline implant capsules. If these are permanent implants, not expanders, synovium may be less frequent with increased cellularity (Lesesne, 1997). Silicone shards and microfragments from other kinds of (nonbreast) implants provoke similar inflammatory reactions and granulomas, as do other kinds of polymers under similar circumstances (Needelman, 1995; Peimer et al., 1986; see Chapter 4). This cellular reactivity to elastomer may be due to the microfragment form, since bulk elastomer does not seem to provoke cytotoxic responses in cell culture (Lockhorn et al., 1996). With gel-filled implants, silicone fluid is seen at various depths of the capsule or outside it as small droplets ingested by macrophages or histiocytes or in extracellular spaces (Beekman et al., 1997a; Wickham et al., 1978). Silicone gel is present in the capsules around ruptured implants. Granulomas containing silicone gel or fluid may also be present. There are no obvious differences between a capsule exposed to fluid diffusion or to gel from an implant rupture (Luke et al., 1997), but vacuolated cells (macrophages) and silicone droplets differentiate capsules around gel from those around saline-filled implants. The effects of phagocytosed silicone gel microdroplets on individual macrophages are incompletely known, but appear relatively benign in some investigations of short term reactions (Azizsoltani et al., 1995). In polyurethane capsules, vacuolated cells, more giant cells, hemorrhage and hemosiderin pigment and fragmented polyurethane are seen (Smahel, 1978a).
Myofibroblasts, which are smooth muscle-like cells derived from fibroblasts, are characterized by 6 to 8 nanometer microfilaments, indented nuclei, desmosomes, and basal lamina and are implicated in capsule contracture and generally detectable in capsules in varying numbers (Rudolph, 1983). These cells contract and relax in response to smooth-muscle stimulants and relaxants, but their role in capsular contracture is still theoretical. They may cause contraction of the capsule, which then becomes fixed as fibrous tissue is laid down (Baker et al., 1981; Ryan et al., 1974), and they disappear after the contracture is mature and fibrotic
(Lossing and Hansson, 1993; Rudolph et al., 1977, 1978). Myofibroblast quantification was attempted in experimental animal wounds. Peak levels of these cells, about 75% of fibroblasts, were reached at 2 weeks after wounding, and myofibroblasts disappeared at 12 weeks. The frequency of these cells was apparently the same in all of the experimental wounds and independent of the degree of contracture (McGrath and Hundahl, 1982; Rudolph et al., 1977). In implanted women, however, myofibroblasts are seen over much longer periods of time, are present particularly in contracting capsules (where they are said to contain greater amounts of contractile [actin] protein) and to a much lesser extent in soft capsules, and are also seen in reoperated breasts and breasts with both saline and, more frequently, gel implants. Removal of the implant (foreign body) results in disappearance of the myofibroblasts and their associated pep-tide growth factors (Lossing and Hansson, 1993; Rudolph et al., 1978).
Macroscopic calcification (or mineralization) has been reported in 10-33% of capsules although these reports generally describe implant series that were problematic, symptomatic, or explants (Destouet et al., 1992; Peters and Smith, 1995; Peters et al., 1998; Rolland et al., 1989a,b). Capsular calcifications involve calcium phosphates and were not associated with silicone (or talc) deposits by electron probe microanalysis in one study (Raso et al., 1999). Calcification is particularly associated with implant shells with patches and more fibrous, long-duration capsules. It can occasionally be severe, causing pressure atrophy of breast and underlying muscle tissue, and it can be seen on mammography (Benjamin and Guy, 1977; Cocke et al., 1985; Fajardo et al., 1995; Frazer and Wylie, 1995; Gumucio et al, 1989; Hayes et al., 1993; Leibman, 1994; Luke et al., 1997; Redfern et al., 1977; Reynolds et al., 1994; Rolland et al., 1989b; Schmidt, 1993; Vuursteen, 1992; Yeoh et al., 1996; Young et al., 1989). Peters et al. (1998) have emphasized that calcification is associated with all first generation implants. Calcification surrounds most of the implant but does not always affect Dacron patches. In this large study of 404 silicone gel-filled breast implants of all three generations, calcification was related to first generation implants (100%), implant duration, and implant rupture. About 50% of second generation implants in place over 16 years, or ruptured and in place over 11 years, had associated calcification. Calcium phosphate was reported as hydroxyapatite in both heterotopic bone and spherulitic aggregates of crystal in the capsule near the implant surface. Although most reports of calcification, and the data in this IOM report refer to silicone gel implants, calcification in the form of hydroxyapatite crystals is occasionally described stuck to the surface of saline implants (Peters et al., 1998; Schmidt, 1993). Talc is also present in 49-71% of capsules (Kasper and Chandler, 1994; but any role for talc is purely speculative and unlikely, as noted by Peters, 1998). Calcification of deposits of
silicone injected into the breast has been reported as quite common (Inoue et al., 1983; Koide and Katayama, 1979), but in these mammography case series the silicone was very likely adulterated and calcification was more prevalent in association with paraffin injection.
During the process of formation and maturation of the capsule an array of chemical mediators, growth factors, enzymes, and various cellular factors are activated and inactivated at various times to effect the various stages and processes of the foreign body and inflammatory reactions (Anderson, 1988, 1993; Lossing and Hansson, 1993; Tang and Eaton, 1995; Ziats et al., 1988). Cellular behavior and appearance change on exposure to silicone (McCauley et al., 1990). In comparative studies, however, different polymers including silicone may have greater or lesser, but not categorically different effects on the release of these factors and on cell behavior (Anderson et al., 1995; Bonfield et al., 1989a, 1991; Cardona et al., 1992; James et al., 1997; Kao et al., 1994; Kossovsky et al., 1993; MacDonald et al., 1996; Miller et al., 1989; Miller and Anderson, 1989; Naidu et al., 1996; Petillo et al., 1994; Sevastianov and Tseytlina, 1984; Taylor and Gibbons, 1983; Wilsnack and Bernadyn, 1979). Also, when the silicone shell of a breast implant (expander) was coated with a 0.3-0.5 mm layer of pyrolyric carbon, the cellular types and proliferative activity of capsules underwent modest quantitative but not qualitative change (Bosetti et al., 1998).
An understanding of the host response to implants is important to experimental and clinical attempts to manage their undesirable and desirable aspects and to the overall safety of the implantation process. Some have argued for an adaptive immune component in silicone breast implant tissue reactions (Kossovsky, 1993). Others have tried to immunize animals to silicone gel, elastomer, or fluid using powerful adjuvants and have failed to observe a difference in the tissue reaction to subsequent silicone implants in either normal or immune deficient (nu/nu) animals (Brantley et al., 1990a,b; Klykken et al., 1991a,b,c). A study of expression of helper (CD4) T cells, suppressor (CD8) T cells, (CD11 b/c) macro-phages, and indicators of proliferation, DNA damage and apoptosis revealed no difference in acute and chronic capsules around Silastic or cellulose acetate (James et al., 1997). Other, investigators however, found restriction of receptors and cell types, suggestive of an immune response in capsules compared to control breast tissue (Ladin et al., 1994). The general tissue-polymer response was recently reviewed without supporting an adaptive immune response (Laurencin and Elgendy, 1994).
Capsules vary from 0.25 to 4 mm in thickness, with an average of 1.3-1.4 mm. (Emery et al., 1994; Ersek et al., 1991; Hardt et al., 1994; Raso and Greene, 1997; Rolland et al., 1989a; Williams, 1972). They are occasionally much thicker (13 mm, Baldt et al., 1994). If grossly abnormal with heavy
calcification or ossification, capsules may measure up to 2.2 cm (Peters and Pritzker, 1985; Peters et al., 1998). Contracted capsules vary in thickness. Some say they are thicker (Caffee, 1992c; Ersek et al., 1991five to tenfold difference, Class I versus Class III-IV; Lossing and Hansson, 1993; Rudolph et al., 1978), whereas others say they are not necessarily thicker but may in fact be thinner than uncontracted capsules (Gayou, 1979; Smahel, 1977; Vistnes and Ksander, 1983; e.g., mean thickness of 17 contracted capsules 0.47 mm, of 6 normal capsules 0.52 mm (Vistnes et al., 1981). The committee found no study in which manufacturer, type, and model were held constant and the only variables were contracture severity and capsule thickness. Grossly ruptured gel implants provoked significantly thicker capsules than intact capsules in experimental animals, however (Vistnes et al., 1977), and consistent with this presumed reaction to exposure to silicone gel and gel fluid, capsules around saline implants are thinner than those around gel and textured implants. Capsules around textured implants tend to take on the imprint of the pillared surface of these shells of those implants, with a greater likelihood of fluid accumulation in cysts within the capsule or in the space around the implant (Jenny and Smahel, 1981; Malata et al., 1997; Rothfuss et al., 1992); probably due to the presence of synovial linings with secretory function. Ahn et al. (1995a) found 0.2-20 ml of fluid within capsules around 15% of breast implants of all kinds in a small group of symptomatic women.
Some investigators have reported that smooth gel-filled implant capsules tend to thicken over time (Ersek et al., 1991; Wickman et al., 1993). Capsules around nontextured implants tend to become mature and stable after 9 to 12 months, and some have emphasized that most (89-93%) contractures are observable within the first postoperative year (Little and Baker, 1980; Malata et al., 1997; Moufarrege et al., 1987; Vogt et al., 1990). However, others have reported that, clinically, contractures continue to accumulate, although at lower frequencies, through the first 24 to 36 months after implantation and taper off thereafter (Brandt et al., 1984; Ersek, 1991a; Gayou and Rudolph, 1979; Rheingold et al., 1994).
By use of presumably more sensitive and quantitative techniques (applanation tonometry; see below), slow continued subclinical contraction has been measured in the one- to five-year interval after implantation. The risk of contracture continued to increase each year for five years in the McGhan (1998) AR90 five-year observational study of gel implants following augmentation and for three years after reconstruction. Reporting on a series of 186 implants, Peters et al. (1997) observed that Class III-IV contractures continued to accumulate over time, reaching 100% around silicone gel-filled implants at 25 years. In a large, long-term study primarily of contracted capsules, Wyatt et al. (1998) noted that the presence of a dense collagenous capsular architecture increased over time around
smooth implants. After five years, the parallel collagen fibers tended to become disoriented, significantly so around textured implants. Also, pep-tide growth factors are reported to be continuously present in capsules around breast implants and to subside only on removal of the foreign body (Lossing and Hansson, 1993). It is likely, therefore, that contracture is a progressive phenomenon that increases slowly with time (Hakelius and Ohlsen, 1997; Handel et al., 1995; Lossing and Hansson, 1993; McCraw and Maxwell, 1988; Peters et al., 1997). Also, as noted elsewhere in this report, some technologies to reduce contracture may become less effective over time (barrier layers, polyurethane), and in such cases, contractures could accumulate over longer periods (Cohney et al., 1991, 1992; Handel et al., 1995).
A foreign body reaction and the formation of a capsule are expected consequences of breast implantation. The contraction of the fibrous connective tissue of this capsule to the point of discomfort and loss of cosmetic effect is an undesirable but common complication. This complication is essentially a cosmetic problem, but it is associated with health and safety to the extent that it leads to frequent operative interventions such as open capsulotomy, explantation or replacement with both operative and anesthesia risk, as well as the risk of infection or other complications that accompany surgery. In a recent explant series, 72.5% of implant removals were performed for the indication of contracture (Beekman et al., 1997b). Certain characteristics of capsulessuch as thickness, Class III or IV contracture, associated granulomas, calcifications, and infection among others also may obligate capsulectomy, which can be a significant procedure accompanying explantation or replacement, may require an hour of additional operative time, and may force difficult decisions about further plastic interventions if tissue cover is inadequate (Young, 1998). Contracture may also lead to other interventions that carry risk, such as closed capsulotomies with possible hematoma or rupture, migration of silicone gel and a need for further surgery, or it may impair the use of diagnostic technologies such as screening or diagnostic mammography for the detection of cancer and other conditions, by making it much more difficult or impossible to achieve adequate breast compression and visualization of much of the breast tissue. The foreign body reaction, formation of a fibrous capsule, and its contracture are likely no more common in breast than in other implants, but the soft tissue, cosmetic role of the breast implant means that these reactions or complications have a substantially greater effect on the safety and performance of this implant, as just discussed.
Assessment of Capsular Contracture
The foreign body reaction is intrinsic to human physiology. Contracture, however, is an excess of fibrosis that may go beyond the patient's usual biological response and occurs, at least in part, individually by breast, presumably influenced by local, poorly understood factorsone of which may be the presence of bacteria, which would be random (Burkhardt, 1988). It could also be argued that the loss of aesthetics with one hard, distorted breast is more disturbing than with two symmetrically affected breasts. Burkhardt has pointed out that contracture should be reported as the percentage of breasts, not of patients. Burkhardt's calculations show how lower contracture percentages, if due to random local factors, can give progressively more misleading impressions if reported as patient percentages. For example, 80% of breasts contracted would yield 90% of patients with contractures statistically, and percentages close to this have been observed in actual clinical practice; 30% of breasts would yield 51% of patients, and 16% of breasts would yield 30% of patients (Burkhardt, 1984). Not everyone agrees that contracture is random, however, even though contractures observed in patients in clinical practice seem to be a mixture of unilateral and bilateral, and in some reports all contractures are unilateral (Milojevic, 1983). The finding of Class III contractures on one side and Class I on the other in monozygotic twins with identical implants three years postoperatively is also interesting, although anecdotal, evidence (Poppi, 1985). Malata et al. (1997) have commented that in some instances, different implants may have been inserted and caused this variation and that, in any case, it does not seem logical that even if local factors were controlling they would not be somewhat similar in the same woman.
Repeated efforts have been made to devise scientifically rigorous clinical assessments of contractures. These include applanation tonometry to compare, over time or between breasts, the relative imprint area on a disk placed on top of the breast in the supine position (Asplund et al., 1996; Gylbert, 1989; Hakelius and Ohlsen, 1992) or, using a standard weight disc, to calculate intramammary pressure (Moore, 1979); mammometry to measure comparative softness after manual compression (Barker, 1978); calipers to measure the compressibility of the breast over time (Burkhardt et al., 1982, 1986; Gylbert and Berggren, 1989) or modified to allow force-distance calculations (Hoflehner et al., 1993); tonometers to measure the compression pressure of the breast at intervals in the clinical course (Gruber and Friedman, 1981; Hayes and McLeod, 1979; Mulder and Nicolai, 1990); a durometer to measure breast hardness (Truppman and Ellenby, 1978); and standardized pictures to be rated by observers (Asplund and Nilsson, 1984). Applanation tonometry is the most fre-
quently used of these techniques, but none has become universally accepted.
The Baker classification has continued as the most common standard, essentially as originally proposed (Baker, 1975), although modifications have been used by some investigators (Burkhardt and Demas, 1994; Gylbert et al., 1989) or suggested by Spear and Baker (1995). This system, though not quantitative, has the advantage of describing the actual clinical factors important to cosmetic effectiveness. Although undoubtedly subject to some bias when used by the operating surgeon on his or her own patients, it has good interrater agreement when applied by experienced personnel who do not have a personal interest in the results of care or research trials. Agreement ranges upward of 83% and is even better between the two major groupings (Groups I-II versus III-IV) at the border between acceptable and unsatisfactory (Coleman et al., 1991; Gylbert et al., 1989; Malata et al., 1997). This classification rates a breast as follows: Class Ithe augmented breast feels as soft as an unoperated one; Class IIthe breast is less soft, and the implant can be palpated but it is not visible; Class IIIthe breast is more firm, the implant can be palpated easily, and it (or distortion from it) can be seen; and Class IVthe breast is firm, hard, tender, painful, and cold. Distortion is often marked. Class I and II breasts are considered clinically satisfactory. Class III and IV are not.
As noted earlier in this report and in the literature, women have shown considerable tolerance for contracture, either by not seeking medical attention for Class III or IV contractures or by pronouncing themselves satisfied with their implants when surveyed (Gylbert et al., 1989, 1990a). The history of breast augmentation with implants (see Chapter 1) began with unsatisfactory substances that were often extruded or, in the case of paraffin, led to devastating tissue reactions. The twentieth century history, however, demonstrates a continuing willingness of surgeons (and women) to experiment with a series of technologies that have progressively improved contracture rates. The original ''open-pore" implants, polymer sponges such as Ivalon and others, produced 100% Class III-IV contractures (Broadbent and Woolf, 1967), and needless to say, augmentation was not in great demand at that time (see discussion of prevalence in Chapter 1). The introduction of silicone gel implants with Dacron patches in the early 1960s reportedly lowered the incidence of serious contractures to around 75% (Gylbert et al., 1989), and the elimination of patches and use of smooth single-lumen gel implants reduced contractures further to about 50% or less (51.5% in gel implants followed an average of 8.5 years, Fiala et al., 1993; see also Brandt et al., 1984; Domanskis and Owsley, 1976; Gylbert et al., 1990a; Moufarrege et al., 1987; Shapiro, 1989). From these beginnings the problem of contracture
has been addressed, but not eliminated, in a number of ways. These include reducing the exposure of breast tissue to silicone fluid by using saline implants or gel implants with barrier shells, addition of steroids to the saline lumen of single or multiple-lumen implants, measures to control infection and hematoma, positioning of implants under the pectoral muscle instead of the mammary gland or breast skin, and development of polyurethane and textured implants.
As with ruptures, the reported prevalence of contracture depends on a number of factors, such as varying detection, the assembly of cases which may or may not be symptomatic or have other biasing factors, the different placements of implants, and timing of assessment and length of follow-up. Studies that attempt to examine unselected groups of women and account for these factors can give a perspective on contracture prevalence. Such studies are cited in the discussion that follows. Reports of high modern contracture prevalence probably reflect selected case series, for example, Solomon's (1994b) report of 71% Class IV contractures in 639 women with silicone gel implants at 440 days on average after placement.
Tissue Exposure to Silicone and Contracture
There are some data defining the amount and characteristics of the silicone gel fluid diffusing through the barrier and nonbarrier shells of implants from some manufacturers, but they are far from complete (see Chapter 3 and below). The total amount and definition of molecular species diffusing over specific units of time in the actual clinical situation for a given implant are not known. Qualitatively, silicone droplets are often visible in capsular tissue, however, and their presence has been said variously to correlate (Barker et al., 1978; Domanskis and Owsley, 1976; Wilflingseder et al., 1974) or not to correlate with capsular thickness and contracture (Gayou, 1979; Rudolph et al., 1978; Thuesen et al., 1995). Measurements in these studies are too subjective to constitute reliable indicators of capsular silicone content. Silicone droplets (from gel implants) or fragments (from saline or textured implants) are also seen from time to time in breast tissue capsules, both intra- and extracellularly in axillary lymph nodes (Barnard et al., 1997; Thuesen et al., 1995; Vargas, 1979), and in the dermis of the scar at the operative insertion site, probably silicone gel fluid rubbed off the implant during implantation (Raso et al., 1996). Such droplets in capsular tissue and within distended macrophages have been proven to contain silicon (Winding et al., 1988). Quantitatively (what is presumed to be), silicone has been measured in the circulating blood or serum and in capsules, breast, and other tissues, mostly by technologies that measure the presence of elemental silicon, not a specific organic form such as siloxane. Attempts have been made to correlate these measure-
ments with the clinical findings, but they have been variable and inconclusive, as noted below.
The technologies for obtaining an accurate determination of silicone, as silicon, in blood and tissue have evolved continuously. Results have been reported in different units, and actual readings have varied enormously as different technologies have been used and increasing care to eliminate contamination of sample from ubiquitous environmental sources or loss from sample due to volatilization has been exercised (see review of Cavic-Vlasak et al., 1996; and also Peters et al., 1995c). This has made it difficult to interpret results with confidence. In fact, it is difficult to be sure in various reports that contamination has been completely avoided. Silicon intake in the diet and in water or other drinks (including silicone, Kacprzak, 1982) can be substantial and can vary widely (20-50 mg SiO2/day, Bellia et al., 1994; reflected as daily urinary silicate excretions in this range, Berlyne et al., 1986; from 0.68 to 17.3 mg per liter of mean silicon compounds in drinking water in various U.S. cities, Morykwas et al., 1991; 9-14 mg of silicon ingested per day, Cavic-Vlasak et al., 1996). Silicone parenteral, oral, dermal or inhalation exposures are substantial and variable, with sources including lubricants in syringes (see Chantelau and Berger, 1985; Chantelau et al., 1986; Collier and Dawson, 1985), antifoams in foods, surfactants, emulsions, polishes, water repellents, textile finishes, fluid and powder treatments in skin and suntan lotions and other cosmetic products, antiperspirants, hair care products, shaving cream, anti-foams in many pharmaceutical products, and so on as is often reported in the labeling of these common consumer products (Cavic-Vlasak et al., 1996; Shields et al., 1996; Silicones Environmental Health and Safety Council, 1994).
Poly(dimethylsiloxane) is about 37.8% (Kala et al., 1997, Thomsen et al., 1990) or 38.3% (Garrido and Young, 1999) elemental silicon by weight, and the technologies that measure silicone concentrations in various tissues usually detect the element silicon in the silicone and express results as weight of silicon per milliliter of serum or fluid or per gram of tissue or tissue dry weight. These values could be converted, of course, to the approximate weight of PDMS by multiplying by 2.6. However, this should be done with care, since, although it is likely when silicone droplets are seenit cannot be assumed that the silicon is in the form of silicone; it may be in other silicon compounds. Some of these technologies measure silicon in gross blood or tissue samples, (e.g., atomic absorption spectroscopy, gas-liquid spectroscopy, inductively [or direct] coupled plasma atomic emission spectroscopy, and mass spectroscopy). Other technologies can locate and identify silicon in microscopic sections using energy dispersive x-ray analysis or scanning electron microscopy (Winding et al., 1988). Silicone compounds can be identified by NMR, Fourier transform
infrared microspectroscopy (FTIR) and laser Raman microprobe, the latter two useful in microscopy (see review of Cavic-Vlasak et al., 1996 for discussion of these and other technologies).
Normal serum silicon levels, which vary slightly from blood or plasma levels (Bercowy et al., 1994), have been determined mostly, but not always, from groups of women assembled as controls for implant studies or studies of renal disease. These analyses have been controlled with calibration curves against known standards of hexamethyldisiloxane and analysis of tissues spiked with this compound (Evans et al., 1994, 1996) or other silicon-containing standards, including bovine liver tissue standards, recovery of spiked samples, repeat assays, and the like. Average serum (unless otherwise noted) silicon levels reported include <0.2-68 (mostly <0.6) µg/ml (Bercowy et al., 1994); 0.265 µg/ml in plasma of men (Berlyne et al., 1986); 0.6 µg/ml (Dobbie and Smith, 1982a, 1986, converted from micromoles per liter, 35.6 µmole/L = 1 µg/ml); 0.17 µg/ ml in plasma (Gitelman and Alderman, 1992); 0.22 µg/ml (Hosokawa and Yoshida, 1990); 0.25 µg/ml (Jackson et al., 1998); 0.03-0.209 µg/ml (Leung and Edmond, 1997); 0.02542 µg/ml, mean, and 0.02175 µg/ml, median (Macdonald et al., 1995); 0.1498 µg/ml in blood (Malata et al., 1994a); 0.01-0.25 µg/ml (Marco-Franco et al., 1991); 0.11 µg/ml in blood (Mauras et al., 1980); 0.02528 µg/ml, mean, and 0.01705 µg/ml, median in blood (Peters et al., 1995); 6.24 µg/ml in blood (Sun et al., 1996), 0.14 µg/ml (converted from micromoles per liter, Roberts and Williams; 1990), 0.27 µg/ml and 0.23 µg/ml, mean 1 week and 1 month, respectively, postpartum (Tanaka et al., 1990, converted from micromoles per liter), and 0.13 µg/ml (Teuber et al., 1995a, 1996). These values have been compared to measurements from women with silicone gel breast implants. The serum or blood values from women with breast implants have been found by some to be not significantly different from normals listed above and reported at the same time0.1355-0.1680 µg/ml in women with intact or ruptured implants (Malata et al., 1994a), and a mean of 0.02711 µg/ml, = or median of 0.02531 µg/ml (MacDonald et al., 1995). Others report approximately double levels in women with breast implants compared to the normal levels reported (see above, e.g., mean, 0.03909 µg/ml, median 0.03345 µg/ml [Peters et al., 1995a], 16.16 µg/ml [Sun et al., 1996] and 0.28 µg/ml [Teuber et al., 1995a; 1996]). Much of this serum (or plasma) silicon in normal men and women, and presumably in women with implants, is in the form of silicic acid or magnesium or calcium silicate and is excreted as such by filtration (and in some instances tubular secretion) by the kidney (Adler and Berlyne, 1986).
Garrido et al. (1994, 1996) found blood values by NMR that were below their detection limits in normal women and several orders of magnitude higher (i.e., milligram versus microgram quantities) than those
cited above in women with breast implants (31-143 millimole, total silicon) and also detected silicone breakdown products. An effort to confirm these results by NMR in implanted women was unsuccessful. If the microgram values in blood and serum cited above are correct, these levels are below the detection limit of the NMR technology used (Macdonald et al., 1995; Garrido and Young, 1999; Taylor and Kennan, 1996). If the milligram values reported by Garrido et al. are correct then gram quantities of silicone are in the circulation at any given time. If such elevated blood levels are correct, one would expect that levels of at least the same order of magnitude would be seen in body tissues that must be in some sort of equilibrium with blood and, in factat least for lower molecular weight siliconesare known to be in equilibrium with blood (see descriptions of D4 and D5 toxicology in Chapter 4). Such levels have not been reported by any technology except NMR (Garrido et al., 1994) in other human tissue than capsules around gel implants, and studies reporting these capsular levels also report levels in tissue in the few microgram range (see below). It was recently suggested that the 1994 analysis of Garrido et al. (1998) might have to be rethought (Garrido et al., 1998; MacDonald, 1999). In addition, studies have not confirmed the presence of crystalline silica (suggested as a breakdown product of silicone) in breast capsular tissue (IRG, 1998; Pasteris et al., 1999). Serum or blood levels of silicon do not correlate with whether an implant is intact or ruptured (Jackson and Dennis, 1997; Malata et al., 1994a; Peters et al., 1995a; Teuber et al., 1995a, 1996) or the duration of implantation (Jackson and Dennis, 1997; Peters et al., 1995a; Teuber et al., 1995a, 1996). The marked difference between capsular levels (or even pericapsular breast tissue levels) and blood-serum or distant tissue levels raises a question about the feasibility of movement of silicone (specifically from gel and elastomer) into the bloodstream and its dissemination from such depots with high concentration gradients. This question requires further investigation.
Almost all studies have agreed that there are baseline levels of silicon in normal breast and other tissues. Values reported range from the detection limit of 2.00 ng/g to 9.46 µg/g dry weight of heptane-extracted organosilicon in breast in a comparative autopsy case series of implanted and nonimplanted women. Some higher values in spleen and lung tissues were seen in the nonimplanted than in the implanted cadavers (95th per-centlie, 134.4 and 45.22 µg/g dry weight, respectively, versus 58.92 and 16.00 µg/g dry weight). The majority of implanted women in this report had at least one positive axillary node (presumably) by light microscopy for extra- or intracellular silicone (Barnard et al., 1993). Other reports include the following: 0.5-6.8 µg/g tissue in breast, liver, spleen, and subcutaneous tissue in an autopsy case series of women without implants (Evans et al., 1994); 0.25-2.4 µg/g dry (breast) tissue in women without
implants (Leung and Edmond, 1997); median levels of 27 µg/g and mean levels of 60.5-64 µg/g dry (breast) tissue in an operative, breast reduction, control series, range 4-446 µg/g (McConnell et al., 1997; Schnur et al., 1996; Weinzweig et al., 1998); and 0.025-1.460 µg/g dry weight in heptane extracts of normal breast tissue (Peters et al., 1995a).
Elevated levels of silicon are reported in capsules around intact saline implants, but not in breast tissue beyond the capsule. Reports include capsule levels in saline tissue expanders of 44-1,380 µg/g tissue (Evans and Baldwin, 1996); median capsular levels of 7.7 µg/g dry (heptane extracted) weight (Peters et al., 1995a); median capsule levels of 71.5 µg/ g with a mean of 140.7 µg/g and median breast tissue levels of 28.0 µg/g with a mean of 56.5 µg/g dry weight (Schnur et al., 1996). In an extension of their 1996 report, Evans and Baldwin (1997b) reported control cadaver silicon tissue levels from various organs averaging 2.2 µg/g tissue, with a median of zero and undetectable levels in >50%. Levels in the capsules of silicone elastomer port-a-catheter chemotherapy implants averaged 8.04 µg/g tissue; levels in capsules of saline implants or expanders averaged 292 µg/g tissue with a median of 110 µg/g tissue; and levels in non-breast tissue sites of women with saline and gel implants averaged 3.2 µg/g tissue, with a median of 2.7 µg/g tissue and undetectable levels in 18%, that is, not significantly different from controls. Apparently tissue silicone levels are increased even around small short-term non-breast silicone implants. (Evans and Baldwin, 1997a,b). The saline inside an implant contains on average 10-12 µg/ml silicon, and when the implant deflates this silicon is released into the capsule (presumably as silicone) and some probably reaches the surrounding breast tissue (McConnell et al., 1997). Deflated saline implant capsule and tissue levels reported include the following: capsule, range, <5-2818 µg/g dry weight (McConnell et al., 1997); capsule, median, 198 µg/g and mean 883 µg/g; and breast tissue, median and mean, 116 µg/g dry weight (Schnur et al., 1996).
Very high silicon levels are reported in capsules around silicone gel implants and intact and ruptured implant capsules: 15-9,800 µg/g of PDMS in formalinfixed tissue (Baker et al., 1982); 75-9,000 µg/g tissue (Evans and Baldwin, 1996); average levels in capsule around silicone gel implants of 1,439 µg/g tissue, with a median of 490 µg/g tissue (Evans and Baldwin, 1997b); 29-496 µg/g dry weight in presumably intact implant capsules (Leung and Edmond, 1997); a median of 11,492 µg/g and a mean of 11,613 µg/g in intact implant capsules with a median of 85 µg/g and a mean 490 µg/g in breast tissue. Levels reported for ruptured implant capsules are: a median of 13,590 µg/g and a mean of 14,683 µg/g, with a median of 64 µg/g and a mean of 3,332 µg/g in surrounding breast tissue, and axillary lymph node levels of 11,879 µg/g dry weight (McConnell et al., 1997). Capsules of intact and ruptured implants mea-
sured 9,979 µg/g, range 371-152,000 µg/g dry (heptane extracted) weight (Peters et al., 1995a); 0.252-116.9 µg/gram dried capsule (Sun et al., 1996); a median of 8,118 µg/g and a mean of 13,685 µg/g in intact implant capsule, with a median of 73.0 µg/g and a mean of 265.3 µg/g in breast tissue, a median of 12,666 µg/g and a mean of 14,751 µg/g in ruptured implant capsule with a median of 216.0 µg/g and a mean of 2,430 µg/g dry weight in surrounding breast tissue (Schnur et al., 1996); capsules of intact, 1,400 µg/g, and ruptured or possibly ruptured implants, 5,600-6,800 µg/g tissue of heptane extracted silicone (Thomsen et al., 1990). Using NMR, levels of 0.05-9.8% silicone by weight were reported in excised capsules of gel implants (Garrido and Young, 1999). A number of reports have confirmed a two-to tenfold variability of silicon levels from location to location in capsular tissue, as well as some variation in capsules from each breast in the same woman. If silicone levels are related to contracture, the relationship appears to be unpredictable and random (Baker et al., 1982; Evans and Baldwin, 1997b; McConnell et al., 1997).
One study detected silicone degradation to silica and "high coordinated silicon complexes" inside gel implants (Pfleiderer et al., 1993a), and some reports speculate that electron microscopy or FTIR may indicate some degradation forms of silicone in tissue (Greene et al., 1995a; Hardt et al., 1994). Other investigators, however, using NMR and laser Raman microprobe have reported that the silicone inside the gel implant and in the capsular tissue is chemically the same, i.e., only PDMS (Centeno et al., 1994; Garrido and Young, 1999), and that implant gel is stable in vivo over prolonged periods (Dome et al., 1995). The weight of current evidence does not support the detection of silica or other silicone degradation products by the reported technologies. There is no correlation between implant age and capsular silicon level in most reports (Baker et al., 1982; Barnard et al., 1997; Evans and Baldwin, 1997b; Peters et al., 1996; Schnur et al., 1997), although one analysis of intact gel implants found a significant correlation between implant age and silicon levels (McConnell et al., 1997). Also, a qualitative analysis of silicone migration through the capsule found greater migration with implant age (Beekman et al., 1997a). A few investigators have reported that there is no correlation between the integrity of the implant and silicon capsular levels (Evans and Baldwin, 1996; McConnell et al., 1997; Peters et al., 1995a). Not surprisingly, silicone is found microscopically or analytically in tissue around non-breast silicone elastomer implants (Evans and Baldwin, 1997a; Needelman, 1995; Peters et al., 1995a). A preliminary report of implanted rabbits described more silicone per gram of tissue in capsules from ruptured than from intact gel implants, and tissue silicone levels in organs of implanted animals were the same as controls except for brain and capsule in which significantly higher levels were measured (Marotta et al., 1996b).
These data define in part the exposure of women to silicone from breast implants, which appears to be primarily and usually from the implant, its capsule, and the immediately surrounding breast tissue and axillary lymph nodes. Silicon found in distant tissues apparently reflects human exposure to ubiquitous silicon or silicone from the environment, and concentrations in tissue and fluids of women with implants are not significantly higher than control values from women without implants. This does not mean that all implant silicone is accounted for. Small amounts of low molecular weight compounds from gel fluid are likely to diffuse or be transported away from their source and to be subject to lung, hepatic, or renal clearance, as some elevated blood levels (if accurate) might imply. This is consistent with animal studies using carbon-14 (14C) labeled silicone. Only 33-47 µg of 500,000 µg of silicone gel (with 80% extractable 1,000 centistoke, 14C-labeled PDMS fluid) implanted without a containing shell in the backs of mice left the implantation site over 20 weeks the majority of which was excreted in urine, most of the balance being found in regional nodes (FDA, 1992a, vol. I, pp 86-87, vol. II, pp 94-96, Isquith et al., 1991, see also the review of distribution and pharmacokinetics of D4 and to some extent D5 in Chapter 4). The substantial variation in silicon measurement levels reviewed here, however, raises a question of the reliability of some of these data, which appear to be outliers and suggests caution in extrapolating from these results. Because one cannot be sure how reliably contamination or volatilization was ruled out in reports, and therefore how accurate silicon determinations were, the committee believes that the situation calls for an effort to develop and agree on standardized technologies and normal biologic measurement values that can be used as accepted references in research and clinical medicine.
Silicone exposure, as measured by capsular silicon levels, was not found to be associated with connective tissue or autoimmune disease-like signs or symptoms (Evans et al., 1996; Evans and Baldwin, 1997b; Weinzweig et al., 1998). These data also provide a context for considering the association between implants and local complications. Evidence for a relationship of tissue silicon concentrations and changes in the breast, including capsular contracture, is insufficient. Silicon levels were correlated with microscopic changes such as foamy histiocytes and vacuoles; that is, the levels were associated with microscopic signs of silicone in the tissue, but not with inflammation, giant cells, or calcification (McConnell et al., 1997). Silicone levels were also correlated with an abundance of fibroblasts and lymphocytes, silicone droplets in the tissues, and sparseness of plasma cells (Thomsen et al., 1990). Thomsen's report stands alone in suggesting a direct relationship, measured by quantitative analytic techniques of silicone equivalents, between increasing silicone capsular tissue levels and increasing fibrosis. Using a semiquantitative technique, energy
dispersive x-ray analysis (which suffers from the sampling problems of electron microscopy), Jennings et al. (1991a,b) reported that tissue silicon levels were lower in more severely contracted than in soft capsules. However, these levels were higher in capsules around gel than around saline implants and diminished rapidly with distance from the implant (Jennings et al., 1991a,b).
Other evidence for the relationship of silicone exposure to contracture depends on the lesser frequency of contracture observed with implants that deliver less silicone to the tissue, (e.g., saline and, at least initially, barrier implants). Capsules around saline implants have been reported to have less microscopically detectable silicone, as noted earlier in this chapter. Barrier implants allow diffusion of smaller quantities of silicone gel fluid through the shell in vitro, but the actual measured implant capsular tissue silicon levels between one manufacturer's barrier and nonbarrier implants did not reveal a difference (Peters et al., 1996), and one study comparing gel fluid diffusion amounts among different implants removed from patients did not find differences between barrier and nonbarrier shells (Marotta et al., 1996a). This may reflect the known individual variability of silicon tissue levels or the loss of barrier effectiveness over time as suggested in this and other reports.
Frequency of Capsular ContractureSaline Implants Compared to Gel Implants
As noted earlier, reports of the frequency of capsular contracture suffer from many of the same problems as studies of breast implant rupture in the plastic surgery literature. Reports use different units (percentage patients, percentage breasts); some give Baker class, others do not, some use Class II-IV, others Class III-IV. In addition, there are many variables such as submammary or submuscular placement, use or nonuse of steroid, mixing of different makes of implants, variable and too short follow-up periods, and noncomparable study and control groups. In general, however, the results seem to support a lower frequency of contracture around saline implants compared to gel. The following frequencies refer to Class III-IV contractures, unless otherwise noted.
Asplund reported contractures in 55% of capsules around gel and 20% of capsules around saline implants in women with reconstructions (Asplund, 1984). Similar figures, 50% in gel and 16% in saline implants, were reported for this same group of women five years later by Gylbert (Gylbert et al., 1990b). McKinney and Tresley (1983) reported contracture frequencies of 36% around gel and 24% around saline implants in breasts of augmentation patients. Hetter (1979) reported gel implant capsular contractures of 64% and saline implant capsular contractures of 40% in
women augmented in multiple surgical practices with different models of implants. Cairns and de Villiers' (1980) contracture results for augmentation were 81.1% with gel, and 8.3% with saline implants. Reiffel et al. (1983) reported "firmness" in a group of 307 women from three plastic surgery practices receiving gel and saline breast implants primarily in the 1970s. There was a statistically significant greater frequency of firmness in the whole group after augmentation with gel implants (61%) than with saline implants (23%), and the differences were similar and statistically significant in each practice taken separately. In this study follow-up was generally short, from 6.8 to 29 months. Edworthy et al. (1998) reported 36.3% Class III-IV contractures in left breasts and 40.7% in right breasts of women with gel-filled implants (N = 1,112) and 18.2% Class III-IV contractures in left and 22.3% in right breasts of women with saline-filled implants (N = 352) in a large survey series of augmentations. Their aggregate figures for contractures per patient were 56.8% of women with gel implants and 40.5% of women with saline implants.
Other substantial numbers of Class III-IV contracture of capsules around gel-filled implants have been reported: 79% at 15-21 years after submammary augmentation (Gylbert et al., 1989); 44% of gel submuscular implants (Hakelius and Ohlsen, 1997); 40% of augmented breasts with "firmness" leading to reoperation in more than a third (Domanskis and Owsley, 1976); 45% of augmented breasts with Class II-IV contractures needing capsulotomies (Brandt et al., 1984); 56% Class III-IV contractures of 60 breasts augmented with double-lumen implants in the placebo arm of an antibiotic treatment trial (Gylbert et al., 1990a), and 14.6% of 1,454 breasts requiring reoperation in 749 women (Gabriel et al., 1997). Very low rates of contracture around saline implants have been reported in two large series: 1.3% (Lavine, 1993); and 1.1% (Mladick, 1993). Also 1.9% contractures were reported in a smaller experience (Capozzi, 1986). However, in the large series of Gutowski et al. (1997), Class III-IV contracture (as graded by the patients) plus the number of patients requiring open or closed capsulotomy totaled 20.4% of patients; the prevalence of Class III-IV contracture reported by Rheingold et al. (1994) was 9.5%, and by Francel et al. (1993) in immediate and delayed reconstruction using implants or expanders was 11% and Worseg et al. (1995) reported 37.6% contracture. All of these saline implant series had adequate to good length of postoperative follow-up.
The Class III-IV contracture frequencies for mostly textured, saline implants at 12 months cumulative follow-up in the modern McGhan LST of 2,855 women were 6.2% overall, 5.5% for augmentation, 10.6% for reconstruction, and 8.8% for revision. Undoubtedly, contractures will continue to occur beyond 12 months. The figures for gel implants in the McGhan AR90 five-year clinical study were 9.3% of augmented breasts
and 4.9% of reconstructed breasts. The better value for reconstructions at five years may be explained in part by the predominance of textured implants used for reconstruction (94.4%) versus augmentation (62.1%) and in part by the submuscular placement used for reconstruction (98.1%) versus augmentation (50.9% submuscular, 49.1% submammary). Historically, contracture has been found to be more frequent following reconstruction (reviewed in Environ, 1991). The 1990 Dow Corning figures at two years were 17.6% Class III-IV contracture for Silastic II, and 8.6% for Silastic MSI implants in augmentation (Bowlin et al., 1998). All of these studies have so many variables, such as different vintages and manufacturers of implants, follow-up, placement, texturing, and indications for implantation among others, that it is not possible to draw a firm conclusion about the frequencies of contracture in capsules around saline- or gel-filled implants, but the evidence suggests that women can expect more contractures around gel implants than around saline implants if these are the only variables.
Barrier Implants and Contracture
Barrier-coated shells also decrease tissue exposure to silicone by slowing the diffusion of gel fluid through the implant shell at least for some years. Animal experiments in mice, guinea pigs and rabbits showed qualitatively less silicone in tissue around ''low bleed" implants and softer, less contracted capsules (Barker et al., 1981, 1985; Caffee, 1986a). Implants with a McGhan barrier to gel fluid diffusion (presumably the dimethyl diphenylsiloxane, Intrashiel, technology) were compared to standard gel-filled implants in rabbits. Silicone was observed and confirmed by scanning electron microscopy or electron dispersive x-ray analysis in significantly more (11 of 20) capsules around standard implants than in capsules (1 of 10) around barrier implants, although there was no difference in capsule firmness (Rudolph and Abraham, 1980). In a very preliminary study, Price and Barker et al. (1983) expressed the opinion that Silastic II barrier implants appeared to be lessening contracture. They were less sure about Intrashiel implants. Their findings consisted of eight "contractions" of unspecified severity in 170 (4.7%) breasts after very short follow-up. Chang et al. (1992) compared conventional implants from several manufacturers and low-bleed gel Silastic II implants in women augmented submuscularly. At more than a year of follow-up, with conventional implants the Baker scores averaged 1.65. There were 16% Class HI-IV contractures. The low-bleed implants had a Baker score average of 1.07. There were no Class III-IV contractures. These were significant differences (Chang et al., 1992). Biggs et al. (1993) compared their low-bleed and conventional results at more than a year follow-up. There was no signifi-
cant difference between the percents of patients with Class I soft breasts, but in submuscular cases the percent Class III-IV contractures was minimally significantly lower with low-bleed implants. In women with submammary implantation, the low-bleed implants produced significantly fewer severe contractures than the conventional smooth single-lumen gel implants.
The definite proof of a relationship between tissue silicone and contracture in humans is lacking since no study of adequate power has held all other variables constant and compared actual tissue silicon measurements to contractures. There is considerable inconsistency among various reports as noted earlier. In the reported literature, qualitative assessments of silicone droplets or a few measurements of silicon may or may not correlate with contracture severity. On the other hand, silicone fluid injected into breasts causes fibrosis and walling-off of silicone deposits, and gel implants are associated with much higher capsule and tissue silicon measurements. Saline-filled and barrier-coated implants appear to be associated with lower tissue silicone exposure and fewer and less severe contractures compared to conventional gel implants in a preponderance of the studies cited above. Fibrous capsules form around any foreign body, and contracture of these capsules is undoubtedly multifactorial, however. Until definitive studies are carried out, it seems reasonable to assume, based on current evidence, that silicone fluid and gel may contribute to contracture rate and severity and that this can be beneficially influenced by barrier technology or by substituting saline filler.
Effect of Implant Surface on Contracture
Experimental work on the effects of surface characteristics of foreign bodies and clinical experience with polyurethane foam-covered implants suggested that providing breast implants with a "rough" or textured surface might result in fewer and less severe contractures. Capsular reactions to texturing and polyurethane have been described earlier in this chapter and in Chapter 3. Some experiments in rats and rabbits failed to show an advantage of textured implants or showed some advantage only with expanders (Barone et al., 1992; Bern et al., 1992; Bucky et al., 1994; Caffee, 1990). Other studies in the same kinds of experimental animals revealed decreased capsular contractures around the textured implants which were related in some cases to the depth and spacing of the texturing; in other cases there were strikingly different prevalences (Brohim et al., 1992; Cherup et al., 1989; Clugston et al., 1994; Maxwell and Perry, 1995; Smahel et al., 1993). Texturing seems to have differing effects on capsule characteristics depending on the characteristics of the texturing. In some cases, almost no effects are noted if surface deformities are shallow (den Braber
et al., 1997), but in other instances where (human) cellular activity was studied on elastomer surfaces with shallow (0.5 m) grooves of varying width and spacing, fibroblast proliferation and orientation differed as the surface changed (van Kooten et al., 1998). In one study, lower reactivity of capsular tissue to smooth muscle stimulants was also observed (Malata et al., 1993).
With few exceptions, a number of clinical trials or observational studies have supported the association of texturing with less severe capsular contracture. Asplund et al. (1996) found 3-9% Class III-IV contractures around textured implants and 10-20% around smooth-surfaced gel implants from the same manufacturer, measured by three techniques (two of which provided a statistically significant difference) in submuscular augmentation. In a study directed primarily at exploring the role of infection in contracture, texturing provided a significant enhancement of contracture control for the saline inflatable implants of one manufacturer (Burkhardt and Eades, 1995). A study designed to further explore antibacterial effects in submammary augmentation compared textured and smooth saline implants of another manufacturer and found, respectively, 2 and 40% Class III-IV contractures (Burkhardt and Demas, 1994).
A prospective controlled trial using submammary gel-filled, low-bleed implants that were identical except for texturing produced Class III-IV contractures in 58% of smooth and 8% of textured implants (Coleman et al., 1991). A five-year follow-up of these women produced essentially the same highly significant results11%, and 59% Class III-IV contractures in the textured and smooth devices, respectively, and 31% replacements for the smooth implants (Malata et al., 1997). Using implants of yet another manufacturer, Ersek (1991b) reported double-lumen (with steroid) implant Class III-IV contractures of 34.5% for smooth submammary, 7.9% for submuscular, 2.5% for textured submammary, and 0% for textured submuscular implants. Hammarsted et al. (1996) reported that postmastectomy reconstruction patients implanted with double-lumen textured and smooth gel implants (from different manufacturers) with intraluminal steroid had, respectively, 9 and 24% Class III-IV contractures. McCurdy (1990) compared two different textured gel implants with polyurethane-coated implants and smooth double-lumen implants and found that texturing was as effective as polyurethane in eliminating contractures, whereas smooth implants produced 25% contractures.
In a study with a short follow-up of women with breast augmentation using gel implants, Pollock (1993) reported 21% of patients with Class II-IV (13%, Class III-IV) contractures around smooth, barrier shell implants and 4% with Class II-IV (2%, Class III-IV) contractures around textured implants with otherwise the same shells from the same manufacturers. These results are biased against finding a beneficial effect of texturing
since the smooth implants were more often (24% versus 1%) placed in the submuscular position, which should have lessened contractures of their capsules. In a comparison study of patients with textured and smooth surfaced, but otherwise identical, gel-filled implant placed in opposite sides, the textured implant was unequivocally preferred by the women and rated better by surgeons in terms of contracture (Hakelius and Ohlsen, 1997). The Dow Corning 1990 multisite study of its smooth and textured gel implants reported half as many Class III-IV contractures around the textured as around the smooth implants (Bowlin et al., 1998). Vogt et al. (1990) reported a multicenter survey using textured double-and single-lumen implants, some with steroid or antibiotics, compared to historical controls. After 12 months the contractures around textured implants were 1.8% overall, compared to the historical controls of 25% and 22% (Little and Baker, 1980; Moufarrege et al., 1987; Vogt et al., 1990).
In a rare negative study, Handel et al. (1995) used a corrective factor for different follow-up periods and reported similar contracture frequency around various saline and gel smooth, textured and polyurethane implants from a number of manufacturers. In a second negative study, 20 consecutive patients at least two years after unilateral mastectomy were given either textured or smooth expanders, followed shortly thereafter with textured or smooth gel-filled implants. There was no difference in contracture or the thickness of capsules between the two groups, although the textured implant capsules contained many silicone fragments (Thuesen et al., 1995). In another, more impressive negative study, Tarpila et al. (1997) augmented a small group (N = 21) of women in the submammary position with a randomly placed textured saline implant in one breast and a smooth saline implant in the other. The implants were the same size and shape from the same manufacturer. Class III-IV contractures were 38% around the smooth and 29% around the textured pros-theses, results that were not significantly different statistically (Tarpila et al., 1997).
These authors speculated that textured (and thicker) shells may reduce gel fluid diffusion from gel-filled implants which would explain the reduction in contracture around these implants. However, it would not explain how textured saline implants reduce contractures in most studies, where there is no gel or gel fluid diffusion, and also where fragments of silicone elastomer in tissue are more frequent than in smooth implants. There is no evidence for lower silicon levels in capsules around textured implants, and as noted earlier, these capsules have clearly different cellular characteristics that most probably play a role in their effect on contracture. Although studies that control all variables except texturing and have adequate numbers are not available, this evidence suggests that capsular
contracture is less with textured implants than with smooth surfaced implants.
Effect of Steroids on Contracture
Local adrenal cortical steroid treatment may also play a role in contracture and in the safety of silicone breast implants. Steroids have been placed in the lumen of saline implants, in the saline lumen of double-lumen implants, or in the tissue pocket that receives an implant. It is speculated that steroid in single-lumen implants may deliver higher tissue concentrations and thus provoke more steroid complications than in double-lumen implants. In the latter the diffusion may be divided into both an outward to the tissue and an inward to the gel lumen direction. This presupposes that steroid concentrations are made equal in the unequal saline volumes of the two implant types by adjusting total dosage, and also assumes that the effects of such adjusted total dosage depend on concentration relationships. These are unproved assumptions. Exactly what happens to forms of methylprednisolone (SoluMedrol) or triamcinolone (Kenalog) placed in the lumen of saline implants could not be determined from reports found in the breast implant literature. Conflicting in vitro studies found that steroid diffusion out of the implant was very slowa few months to years in durationand likely varies with the physical and chemical characteristics of the implant shell. Continuing steroid diffusion probably results in exposure of surrounding capsular and breast tissue to pharmacologic concentrations of steroids over prolonged periods of time as clinical experience suggests. Berman et al. (1991) concluded that about 30% of the methylprednisolone might be in the shell at any one time depending on its thickness, or depending on the drug's chemical formulation, it might be variably subject to hydrolysis or crystallization from solution (Berman et al., 1991; Cohen, 1978a; Cucin et al., 1982; Morykwas et al., 1990; Perrin, 1976; Spitalny et al., 1981). In addition, as Gutowski et al. (1997) recently noted, steroid inside saline implants may be, in their experience, a risk factor for twofold increments in deflation rate. Manufacturers have pointed out that the interaction of this chemical with the implant shell has not been investigated adequately, and its use cannot be recommended (Gutowski et al., 1997). In addition, delivery of steroid from a breast implant is not an FDA-approved usage.
Because of the known effects of steroids on scar formation and inflammation, Peterson and Burt (1974) instilled 60 mg of triamcinolone into the pocket on one side of eight bilaterally augmented women and noticed that the treated side was consistently softer than the other over a follow-up period that ranged from a few months to a year. This study had insufficient follow-up, small numbers and unblinded assessment. Some
subsequent communications reported failure of this technique (Brownstein and Owsley, 1978; Kaye, 1978; Price, 1976). For example, Biggs and Yarish (1990) reported that 14% of breasts treated with periprosthetic steroids had Class III-IV contractures and only 4% of breasts without steroids had this complication and Vasquez et al. (1987) also found that steroid-treated breasts had more, but not significantly more, contractures: 40.7% with, versus 31.7% without periprosthetic triamcinolone.
On the other hand, Hipps et al. (1978) found Class II-IV contractures in 26% of women with implants for augmentation in a series treated with 60 mg of triamcinolone in the pocket around smooth gel implants, compared with 35% contractures in the no-treatment group. Baker (1975) reported a 10% decrease in Class II-IV contracture when 20-40 mg of triamcinolone was placed in the tissue pocket. And Lemperle (1980) commented that 20 mg of triamcinolone crystals in the pocket on one side seemed to produce a softer augmented breast, but also a number of perforations due to tissue atrophy at the location of the crystals. Local instillation of triamcinolone around implants in experimental animals failed to produce a marked effect on capsules (Vistnes et al., 1978), or, in fact, any effect on capsules or intraprosthetic pressures (Moucharafieh and Wray, 1977). Reduction in contractures to 5% was noted after placing 62.5 mg of methylprednisolone in each saline prosthesis (Perrin, 1976), and recommendation of this treatment was repeated by Hartley (1976). Addition of 40 mg of triamcinolone to the lumen of saline implants produced marked thinning of the overlying tissue, inferior displacement of the implant, and ptosis of the breast with impending extrusion after a few months. These complications did not appear when 20 mg of methylprednisolone was substituted (Ellenberg, 1977). A case of late erosion of a medium-sized arterial branch with substantial hemorrhage and implant loss has also been reported after triamcinolone at 40-mg dosage (Georgiade et al., 1979).
Experimental results reported by Ksander et al. (1978) using high dose triamcinolone inside implants produced a number of extrusions and disorganized, loosely knit, thinner capsules in the steroid treatment group, although there was no measurable effect on hardness (Ksander et al., 1978). In a later study, methylprednisolone at two dose levels (0.1 and 1 mg/ml) had a significant effect on capsule histology and compressibility (Ksander, 1979). Subsequent reports of the use of 62.5 mg methylprednisolone confirmed steroid complications of severe ptosis of the breast, inferior displacement of the implant, atrophy and bluish discoloration of overlying tissue, and implant extrusion as long as two years postoperatively and the need for replacement of 70% of the implants in one series (Cohen, 1978a; Cohen and Carrico, 1980; Oneal and Argenta, 1982; Persoff, 1978). Lemperle (1980) found that using 50 mg in the outer lumen of
double-lumen prostheses resulted in the need to replace the implants with single-lumen gel implants after a year, although contractures fell from 57 to 16% in his patients.
Carrico and Cohen (1979) reported that methylprednisolone in the tissue around the implant had no effect on contracture (control and treatment frequencies both 50%) and provoked no steroid complications. Methylprednisolone at doses greater than 20 mg within the saline implant, however, led to steroid complications in 61.5% of breasts and to 4% Class III-IV contractures, while doses of 20 mg led to steroid complications in 8.3% of breasts and 4.2% contractures. Because the same total dose may be contained in different volumes of saline used to inflate the outer lumen of the implant, concentrations of the drug may vary, and in this study, higher concentrations appeared to correlate with increasing steroid complications (Carrico and Cohen, 1979). Ellenberg and Braun (1980) reported their results with 20 mg or less methylprednisolone in double-lumen implant augmentations compared with a control group of gel implants. The control contracture rate was 67.6%. Those with 5-15 mg of methylprednisolone experienced 11.9% contractures and 2.4% steroid problems, those with 20 mg methylprednisolone experienced 9.9% contractures and 3.6% steroid problems (Ellenberg and Braun, 1980).
In a study of a large number of patients comparing single-lumen gel implants with double-lumen implants containing 12.5 mg of prednisolone, contracture frequencies were 19% in augmentations, 54% after subcutaneous mastectomy, and 64% in post mastectomy reconstructions with the gel implants without steroids. Contractures decreased to 4%, 14.9% and 24.4% in these same categories when steroid-containing double-lumen implants were used (Lemperle and Exner, 1993). In a randomized and controlled study, Spear et al. (1991) compared the Class II-IV contractures and steroid complications of smooth double-lumen implants with and without 16 mg of methylprednisolone, assigned randomly to two well-matched groups of women undergoing submuscular reconstruction, 44 breasts with steroid and 45 without, followed for a minimum of three years. Contractures were 14% in the steroid group compared to 44% in the non-steroid group, and there was no difference in complications between the two groups.
McCurdy (1990) compared textured gel implants from two manufacturers with polyurethane-coated implants and smooth double-lumen implants with and without 20 mg methylprednisolone in submammary augmentation. Although the follow-up was short in some groups, in general the Class III-IV contractures were zero and 3.9% around polyurethane-coated implants and steroid-added double-lumen implants compared to 25% with the no-treatment smooth double-lumen implants. The prevalence of local steroid complications was 17.1%, however.
Although the total dose appears to be an important variable, it is likely that the intraluminal concentration of steroid also plays a role, as suggested by the data of Carrico and Cohen (1979) and the subsequent reanalysis and discussion by Cohen and Carrico (1980). The much higher complication rate reported by McCurdy (1990) than by Spear et al. (1991) whose doses were similar (20 versus 16 mg) but whose concentrations were markedly different, (i.e., 100 mg versus 40 mg per 100 ml of saline), is also suggestive. In an inflatable saline implant, the steroid is contained in a much larger volume and thus is present at a much lower concentration. Some authors report that the total dose of 20 mg in these cases never causes a problem (Guthrie and Cucin, 1980). It is logical to assume that the higher concentration gradient associated with the more concentrated steroid would expose the tissue to a higher dosage of the drug, although probably for a shorter time. In an analysis of their data by concentration of steroid in the implant, Cohen and Carrico (1980) concluded that methylprednisolone should be administered according to concentration in the implant and not according to total dose. Based on their clinical findings, they recommended 5-10 mg/100 ml of saline as a reasonable level that would minimize, but not eliminate, steroid complications and at the same time have an effect on contractures.
This suggestive evidence for an effect of intraimplant adrenal cortical steroid in decreasing contracture has to be balanced against the occurrence of steroid complications, the possible weakening of implant shells, the availability of other modalities to reduce contracture, and the nonapproved status of this intervention. The studies cited have, in general, design problems, including small numbers, lack of controls, varying dose levels, and placement of the steroid in implants with varying shell characteristics, among others. The committee believes that, at a minimum, before this treatment can be recommended, the behavior of a particular implant as a delivery vehicle and the therapeutic results and complications of a defined dosage to tissue in properly controlled and randomized studies would have to be determined.
Role of Infection and Antimicrobial Treatment
The safety of breast implants is affected by infections in a number of ways. Surgical wound infections necessitate medical and surgical interventions. Infections of the implant or implant pocket often require extensive treatment, including removal and replacement of the implant (e.g., Rheingold et al., 1994). A small number of these infections are caused by unusual and recalcitrant microbes, including various fungi, mycobacteria, and clostridia that resist rapid resolution or resolution without implant removal. These organisms differ from the usual bacteria found in
wounds or infecting implants perioperatively, such as Staphylococcus aureus b hemolytic streptococci, or less virulent staphylococcal species. A diverse array of bacteria can be cultured from the surface of, or from breast tissue around, implants often with no clinical signs, such as S. epidermidis and related species, Propionibacterium acnes and related species, S. aureus, anaerobic diphtheroids and more rarely Streptococcus A and B, Escherichia coli, Enterococcus, Corynebacterium, Klebsiella, Pseudomonas and infrequently others (Ablaza and LaTrenta, 1998; Ahn et al., 1996; Brand, 1993; Clegg et al., 1983a; Coady et al., 1995; Courtiss et al., 1979; Dobke et al., 1995; Foster et al., 1978; Gylbert et al., 1990b; Hunter et al., 1996; Lee et al., 1995; Netscher et al., 1995b; Peters et al., 1997; Truppman et al., 1979; Virden et al., 1992; Williams et al., 1982; Young et al., 1995a). Some of these latter organisms, and occasionally fungi, may also be found within saline expanders and inflatable implants, where they can survive and even proliferate possibly supported by glucose that diffuses into, and has been measured within, the implant (Blais, IOM Scientific Workshop, 1998; Chen et al., 1996; Coady et al., 1995; Nordström et al., 1988; Young et al., 1997). Presumably they enter through the punctures in the inflation ports (Liang et al., 1993). These periprosthetic organisms are usually discovered on aerobic and anaerobic culture of implants, pockets and capsules. They are often not involved in clinically apparent perioperative infection problems, which for the most part are caused by S. aureus, hemolytic streptococci or some less virulent staphylococci, are infrequent, and occur within a month after surgery (Courtiss et al., 1979). In general, studies of infection suffer from the use of varying culture technologies, some failing to culture anaerobically or for a long enough time, some with greater or lesser vigor and thoroughness in sampling the implant surface or peri-implant tissue (Virden et al., 1992).
Local, perioperative infections are generally treated with antibiotics and resolve, although they may contribute to pain or other complications. The frequency of these infections is reported in many case series and ranges around 1-4% after augmentation and significantly higher after reconstruction (e.g., 13%, Bailey et al., 1989; 6%, Courtiss et al., 1979; 5%, Crespo et al., 1994; 7%, Eberlein et al., 1993; 5.8%, Furey et al., 1994; 2.5%, Gibney, 1987; 2%, Noone et al., 1985; 3%, O'Brien et al., 1993; 5%, Slade, 1984; 13%, Vinton et al., 1990) including one report of very high numbers of infections, 8 of 15 patients (53%), in women undergoing immediate postmastectomy reconstruction with expanders (Armstrong et al., 1989). Gabriel et al. (1997) reported a combined total of 1.1% of breasts, implanted primarily for augmentation, reoperated for infection. Brandt et al. (1984) reported 3.9% infections in gel augmented breasts. Biggs et al. (1982) reviewing an 18-year experience reported 2-7.6% of patients reop-
erated for removal of infected implants at various stages in the evolution of this practice (Biggs et al., 1982).
The McGhan LST found 1.1 and 6.9% of breasts with infections after saline augmentation and reconstruction, respectively, and the McGhan AR90 (1998) five-year experience of infection with gel implants was 0.7% of augmented breasts and 0% of reconstructed breasts. The Mentor adjunct study (1992) found 4.3 and 1.3% infections at the three-year follow-up of gel implantation for reconstruction and augmentation, respectively. Very infrequent infections in saline implant augmentation were reported by Gutowski et al. (1997), 0.2%; and by Mladick (1993), 0%. A frequency of infections in mostly inflatable saline implant augmentations of 1.5% was reported by Rheingold et al. (1994). A survey by Brand of 73 plastic surgeons using a diversity of implants found frequencies of infection of 0.06-0.16% for implants in augmentations and 0.3-6% for implants in reconstructions. Since a long time interval was covered and "only severe infections" were reported, considerable underreporting is probable in this survey (Brand, 1993). In a Centers for Disease Control and Prevention (CDC) survey of 2,734 plastic surgeons with a 67% response rate, wound infection after augmentation was reported in 0.64% of patients (Clegg et al., 1983b).
Some wound infections are not treated successfully with antimicrobial therapy and result in loss of the implant (Courtiss et al., 1979). Very rarely, there are very serious or lethal complications such as staphylococcal, streptococcal, or other bacterial toxic shock syndrome (Barnett et al., 1983; Bartlett et al., 1982; Brown et al., 1997a; Giesecke and Arnander, 1986; Holm and Muhlbauer, 1998; Oleson et al., 1991; Poblete et al., 1995; Tobin et al., 1987; see also Walker et al., 1997, for a case after explantation and granuloma excision). Very rarely also, an infection may occur in an otherwise well-tolerated implant many years after surgery without an apparent inciting event (Ablaza and LaTrenta, 1998).
In addition there is evidence that infection is associated with increased frequency and severity of implant capsular contracture and thus with the interventions that accompany this complication. The tissue of the breast is open to the environment through the lactiferous ducts, which are colonized extensively by normal skin flora, both aerobic and anaerobic bacteria, primarily S. epidermidis, P. acnes and anaerobic diphtheroids. As a result, bacteria can be recovered from 91.6% of female breasts, usually bilaterally (primarily coagulase-negative staphylococcal and propionibacterial species, Ransjö et al., 1985). Implants themselves, implant pockets, or capsules and nipple secretions have yielded 23.5-89% positive bacterial cultures, using various techniques (Ahn et al., 1996; Burkhardt et al., 1981; Courtiss et al., 1979; Dobke et al., 1995; Gylbert et al., 1990b; Netscher et al., 1995b; Peters et al., 1997; Thornton et al., 1988; Virden et al., 1992).
These bacteria, particularly gram positive bacteria, have been shown in vitro to be able to adhere within two minutes to, and colonize, all types of silicone breast implants (Jennings et al., 1991a,b; Sanger et al., 1989). They are often located in a bioslime film on the surface of the implant (Dougherty, 1988), where they are largely protected from antibiotic action (Evans, 1987), and they presumably contribute to infections after implant surgery. Thornton et al. (1988) found that some postoperative breast infections were associated with the same organisms that they had cultured at the time of surgery (both for implantation and for breast reduction) from deep within the breast, primarily coagulase negative staphylococci. In this series of 30 patients (19 with breast reductions), contracture was associated with positive cultures, but the numbers were too small to achieve statistical significance (Thornton et al., 1988).
In rabbits with implants contaminated with S. epidermidis compared to sterile controls, the contaminated capsules were Class III-IV and two to three times thicker with more dense collagen than the control Class I-II capsules (Shah et al., 1981). In a subsequent study, the effect of intraluminal cephalosporin was evaluated in this protocol, and the thickness of the capsules around contaminated, antibiotic-containing implants was significantly reduced (Shah et al., 1982). At about this time, cephalosporin (and gentamycin) had been found to diffuse from Heyer-Schulte saline implants in significant concentrations for up to six months (Burkhardt et al., 1981). Guinea pigs formed capsules more rapidly after experimental implants were dipped in staphylococcal broth cultures overnight (Kossovsky et al., 1984). Quantitative data in this report were sparse, and the effect of coating an implant with broth before placement may be an uncontrolled, confounding variable. Others have experimented with iodinated silicone implants. Implants containing povidone-iodine (Betadine) were found to inhibit bacterial growth in vitro due to the diffusion of free iodine through the shell. Saline implants placed in mouse tissue pockets contaminated with S. aureus had capsules 2.8 times thicker than povidone-iodine implants similarly placed or saline implants placed in sterile pockets (Birnbaum et al., 1982; Morain and Vistnes, 1977). Since iodine degrades the silicone shell, this is not a clinically useful observation (Morain, 1982).
Broadbent and Woolf (1967); Burkhardt et al. (1986); Courtiss et al., (1979); and Dobke et al. (1995) reported clinical associations of positive cultures with contracture, and Netscher et al. (1995) found a significant association of Class IV contractures with positive periprosthetic explant capsule cultures. Virden et al. (1992) performed routine and special research cultures with 55 silicone implants (38 gel- or saline-filled implants and 17 expanders) removed from 40 women. Class III-IV contracture was observed around 24 (63%) implants and 3 (18%) expanders, and cultures
(mostly research not routine) were positive (primarily S. epidermidis) from 56% (15 of 27) of implants with contractures and only 18% (5 of 28) of implants without contracture, a statistically significant (p < .05) difference. Similar to the findings of Parsons, 91% of painful contractures were associated with positive cultures (Virden et al., 1992).
Burkhardt et al. (1981) originally noted a decrease in Class III-IV contractures to 3% of breasts with implant intraluminal Keflin or Garamycin in a short follow-up study, compared to a historical control rate of 37%. They subsequently conducted a prospective randomized trial using single-lumen saline inflatable implants in the submammary position that compared a control group with four groups using a variety of antibacterial treatments, including local irrigation with povidone-iodine, antibiotic foam, and intraluminal cephalothin. They demonstrated a significant improvement in Class III-IV contracture from a control value of 41% to a combined experimental group value of 19% (Burkhardt et al., 1986). In a subsequent prospective, randomized study that looked at both texturing and povidone-iodine, the antibacterial irrigation failed to have any effect on contracture (Burkhardt and Demas, 1994), and Gylbert et al. (1990b) could show no effect on contracture of preoperative infusions of antibiotics that dramatically lowered the culture positivity of the implant pocket. This latter result is consistent with the generally held conclusion that preoperative prophylactic antibiotics are of little value (Courtiss et al., 1979) and may also reflect the fact that subclinical implant infections in a slime layer around the implant are protected from antibiotic action (Virden et al., 1992). Gutowski et al. (1997), however, reported that implants containing antibiotics experienced a lower frequency of contracture and in a final prospective randomized study that compared implant texturing from another manufacturer and povidone-iodine irrigation of the implant pocket with untreated smooth implants, a significant effect of antibacterial treatment on Class III-IV contracture was observed (Burkhardt and Eades, 1995). Dobke et al. (1995) in culturing a series of 150 explanted gel and saline (19 implants, 26% culture positive) breast implants, noted that 76% (62 out of 82) of those with Class III-IV contractures, but only 28% (19 of 68) of those without contracture were culture positive, primarily with S. epidermidis. This difference was statistically significant (p < .05) (Dobke et al., 1995). According to Burkhardt, (1988) infection explains the varying occurrence of contracture, that is, its frequent appearance unilaterally as well as bilaterally in proportions that statistically appear to represent a random (infectious) event. More recently, Peters et al. (1997) reported no association of capsular culture positivity (of 42%) with severe contracture in a series of 100 women whose implants were removed.
Dowden (1994) suggested that the presence of the subclinical infec-
tions or contaminations described above may contribute to systemic signs and symptoms such as fatigue, myalgia, diarrhea, and arthralgia, among others, in implanted women. He reported seven women, five of whom had positive cultures for S. epidermidis or Propionibacterium acnes, whose symptoms resolved and whose health returned soon after explantation (Dowden, 1994). In a study that compared women with implants without general health problems to women with implants and a similar, but somewhat more extensive constellation of signs and symptoms, including arthralgia, dry mucous membranes, fever, hair loss, and cognitive problems, Dobke et al. (1995) found health problems to be associated with positive cultures. Among women with these symptoms, implants were 81% culture positive compared to 28% positivity in those without such signs and symptoms, and among women with both the three systemic signs and symptoms and Class III-IV contracture, 95% (19 of 20) of patients had culture-positive implants.
Earlier, the same group had tested the hypothesis that a painful pros-thesis signified subclinical infection. Painful breast and penile prostheses were cultured at explantation and compared with cultured expanders (removed for replacement with a permanent implant) or with cultures of malfunctioning penile implants. In the aggregate 26 out of 28 (93%) painful prostheses and 4 out of 31 (13%) devices that were not painful were infected, mostly (> 90%) with S. epidermidis. Replacement of infected and painful devices with sterile devices while giving antibiotics resulted in pain-free devices in nine of ten instances (Parsons et al., 1993). Others evaluating culture-positive explants have not found associations with the health problems noted above, although little in the way of description is provided (Ahn et al., 1996), and the evidence for an effect of infection on symptoms remains limited.
Also, the important data of Burkhardt would be more persuasive if the comparison control groups of smooth saline implants were not at the upper ends (27-41%) of the Class III-IV contracture rates for modern saline implants, and some studies have been negative (e.g., Peters et al., 1997). The differences in contracture frequency with saline versus gel and textured versus smooth implants are not readily explained by a bacterial theory of causation. Nevertheless, the evidence for a relationship between the presence of bacteria around the implant and contracture, although not conclusive, is certainly suggestive.
Role of Hematoma
Collection of blood, hematoma, or tissue fluids, (seroma), around implants is very like overt infection in that it complicates a small number of implantations, often requires an invasive intervention, although some
resolve (or drain) spontaneously, and has been suggested as a factor in contracture. Since the frequency of clinically apparent hematoma or seroma is usually much lower than that of significant contracture, this complication is, at the most, a small contributor to contracture. Hematoma was the indication for reoperation in 3.5% of the breasts in the Mayo Clinic series (Gabriel et al., 1997). In the experience of one plastic surgery clinic 5-10.3% of patients were reoperated for hematoma over an 18-year period (Biggs et al., 1982). Hematoma or seroma was not reported in the McGhan LST or AR90 observational studies. This complication probably is reported quite variably, and often it is mentioned in case series reports only in a cursory fashion, if at all. Some of these reports of hematoma or seroma frequencies and the instances of accompanying operative interventions to provide open or needle drainage have been noted earlier in this report. Plastic surgeons vary in the use of drains (which some report prevent contracture to a meaningful extentBrandt et al., 1984; Hipps et al., 1978) and other operative precautions to prevent or manage bleeding and the collection of blood around implants. Many surgeons use drains when implanting textured-surface implants to prevent seroma formation. Also, hematomas around implants may become infected or be associated with infections (Courtiss et al., 1979).
A hematoma frequency of 2%, most requiring reoperation, was reported by Rheingold et al. (1994) and Baker et al. (1975). Additional reports include 1.4% hematoma (Biggs et al., 1990), 5.9% (Brandt et al., 1984), 4.5% with implant loss (Artz et al., 1991), 0% hematoma or seroma (Bayet et al., 1991), 6% hematoma (Capozzi, 1986), 20% ''postoperative bleeding" (Gylbert et al., 1989), 1.1% hematoma (Lavine, 1993), 0.48% of hematomas (Mladick, 1993), 2.1% hematomas (Williams, 1972) and so on, for augmentation and reconstruction with both gel and saline implants. These reports are typical for hematomas that are observed within days after implantation. Rarely hematomas occur years later in association with contracture, due presumably to microfractures of the stiff fibrous capsule. These can pose significant problems requiring more extensive surgery (Cederna, 1995; Frankel et al., 1994; Marques et al., 1992). Conversely, there are those who believe that events such as trauma, which could produce hematoma, may cause late-onset contracture (Ashbell, 1980). Observations of hematoma associated with contracture are mixed. Some report the absence of an association (Asplund, 1984; Coleman et al., 1991; Hakelius and Ohlsen, 1992), but these reports involve very small numbers of hematomas and were not designed to study the issue. Others have found a significant association between hematoma and contracture in their clinical studies (up to two- or threefold greater prevalence of contracture in implants with hematoma than in those without (Handel et al., 1995; Hipps et al., 1978; Wagner et al., 1977; Watson, 1976). In a study
involving baboons, with numbers too small to have any significance, implants with blood around them had thicker and harder capsules. In another larger study with rats, hematoma had no effect on either capsular thickness or intraimplant pressure, although a combination of hematoma and steroid did elevate pressures (Moucharafieh and Wray, 1977). Caffee (1986b) also found no effect of hematoma on capsular contracture in rabbits. These studies are inconclusive. The safety implications of hematoma involve primarily the few percent extra interventions required to resolve these complications, the suggested association of infection, and the limited evidence that the incidence of contracture and its accompanying problems might be somewhat higher around implants with hematomas.
The Influence of Implant Location
The placement of implants in the submuscular position, which was originally reported by Dempsey and Latham (1968) and modified to partial muscular coverage by Regnault and colleagues has a salutary influence on the incidence of contracture, decreasing it in the latter report from 30% in the submammary group to 10% in the submuscular group (Regnault, 1977; Regnault et al., 1972). This effect is reported in a number of additional studies that cite significant decreases in contracture when comparing women with submuscular implants to women with submammary implantation of different kinds of gel-filled implants. These include decreases from 11.1% to 3% of Class III-IV contractures with some standard and some low-bleed gel implants (Biggs and Yarish, 1990); 40% to 5% of patients with severe contracture (Mahler and Hauben, 1981); 83.8% to 27.1% of Class III-IV breasts with gel implants of 12 years' duration or less (Peters et al., 1997); 41% to 8% of Class III-IV contractures with gel-filled implants (Puckett et al., 1987); improvement from 30% Class I contracture to 95% Class I contracture around gel-filled implants (Scully, 1981); average self-assessed Baker score at five years of 2.9, submammary to 2.1, submuscular using gel-filled implants (Fiala et al., 1993), and 58% submammary and 9.4% submuscular gel implant contractures (Vasquez et al., 1987).
Other reports, cited earlier, describe low rates of contracture in patients with submuscular implants studied and reported for other reasons (e.g., Chang et al., 1992). A review of the literature by Puckett, cited in another report, concluded that Class III-IV contractures occurred in 43% of breasts with submammary and 6% of breasts with submuscular implants (Biggs and Yarish, 1988). Hetter (1991) repeated his 1979 survey and reported that the contracture (firmess) rate had dropped from 64% to 8% since he had changed from the submammary to the submuscular approach. In a study of saline inflatables, Cocke reported 44% noticeable
firmness in submammary implants and 19% in submuscular implants (Cocke, 1994). In a review of a large experience with polyurethane-coated gel implants, 22 of 658 (3.3%) implants in the submuscular position had Class II-III contractures (only two were Class III) compared to 14 of 237 (5.9%) submammary placements with Class II contractures (Hester et al., 1988).
A few studies compared contractures after subcutaneous implantation with those after submuscular implantation of gel-filled breast implants. Two found firmness of 31-50% in submuscular and 80-100% in subcutaneous implantation (Slade, 1984; Ringberg, 1990). A third study found 0 and 7% of breasts with Class IV and Class III contractures, respectively, in submuscular (includes both pectoral and serratus coverage), and 7 and 33% of breasts with Class IV and Class III contractures, respectively, in subcutaneous implantation in delayed reconstruction followed from one to five years (Gruber et al., 1981). There appears to be sufficient evidence to conclude that submuscular rather than subglandular or subcutaneous placement of the implant is associated with a lower incidence of severe contracture. This observation should be among the several factors considered by both the patient and surgeon in deciding between submuscular and other placement of the implant.
Not everyone agrees with a policy of routine placement of implants under the muscles of the chest wall for augmentation (Ashbell, 1980; Courtiss et al., 1974), and plastic surgeons still use the submammary approach in 32% of augmentations (ASPRS, 1997) presumably because of the better aesthetics of this placement in breasts with adequate tissue cover, and possibly because submuscular implantation has been associated with more pain. Placement has relevance for the issue of safety primarily because of its effect on contracture, which lessens the need for interventions secondary to this complication. The submuscular position may also facilitate examination of the breast for cancer, since the glandular tissue lies above the implant and is all available for palpation (Little et al., 1981, see also Chapter 12 ). Although the submuscular operative approach is technically somewhat more demanding, the rates of rupture, deflation, infection, hematoma, and other complications do not seem to differ significantly between submuscular and submammary placement. Occasional speculation about the submuscular position, noted below, does not have convincing nonanecdotal, experimental, or clinical support in the studies cited. It cannot be concluded that submuscular implants, being further away from potentially contaminated breast glandular and ductal tissue, are less prone to infection. There is no evidence that such implants are in some way less sensitive to silicone droplets, or might benefit from the massaging action of overlying musculature. Although the theory is intuitively attractive, there are no data in the literature avail-
able to the committee to show that placement of implants with muscular cover between them and glandular tissue results in earlier diagnosis c)f breast cancer by palpation or mammography, or that contracture might not be less frequent but merely more difficult to detect in this position.
Other Complications and Their Relevance to Safety
At the beginning of this chapter, local and perioperative complications were discussed. Not all of these have numerically significant, or medically and clinically important, safety implications. Although some of these conditions may be mentioned in other chapters of this report, the committee finds that for its purposes here the major influences on safety have been discussed. For example, although the reference list includes about 30 citations on the effects of radiation therapy in women with breast implants, implants themselves have good stability to clinically relevant dose levels of irradiation, they do not significantly interfere with the radiation beam and radiation therapy, and evidence that radiation can increase implant capsular contracture is limited (see Chapter 3 for discussion).
One additional problem may merit some attention. Pain associated with implantation is common enough to be considered. Localized pain results in requests for implant removal (e.g., as an indication in 19.2% of explants, Beekman, 1997b, as one of the major reasons for a 13% prevalence of replacement, Bright et al., 1993) or in interventions for relieving pain associated with contracture. Some authors have reported complaints of pain in the great majority of women with implants (107 of 114 consecutive patients, Silver and Silverman, 1996), pain manifesting as a new "chest wall syndrome" in 68% of women with implants (Silver et al., 1994), or pain in 36% of women explanted (Peters et al., 1997), but these were not representative samples of the population of women with implants. Most reports of complications do not include much if any detail on pain. It was included without discussion in the list of indications for reoperation and was the indication for secondary surgery in 1.1% of the Mayo Clinic group of 749 implanted women (Gabriel et al., 1997), and reports often cite rather low frequencies.
Wallace et al. (1996), however, reviewed the subject of pain after breast surgery using a questionnaire with a 59% return rate (282 women). Although the response rate might indicate a bias toward complaints, this group reported substantial local pain after reconstruction (up to 50%) and augmentation (38%). Pain was also more common after submuscular (50%) than submammary (21%), and after saline (33%) than gel (22%), implantation. Since the pain was worse after implantations than after
reduction surgery or mastectomy alone, these authors assumed it was related, at least in part, to the implants, although significant prevalence of chronic postmastectomy pain has been reported in other surveys [12.7-17.4% of patients at various times up to a year postoperative in Kroner et al. (1992), 20% of patients in Stevens et al. (1995)]. Others have reported that pain often recedes after explantation (Peters et al., 1997). Of the augmented and reconstructed patients with pain, 20-29% required pain control medication, though for how long is not clear. Pain is one of the indications for implant removal. Capsule formation, especially under the muscles, may result in nerve compression and pain leading to a requirement for secondary procedures. Other late pain may be due to muscular compression (Huang, 1990). Usually, pain with late onset (8% and 30% of reconstruction and augmentation patients, respectively) represents contracture pain (Wallace et al., 1996). Pain is also associated with some gel implant ruptures, up to 93% in some reports (Ahn et al., 1994b; Andersen et al., 1989), is reported in association with polyurethane implants (Jabaley and Das, 1986; Smahel, 1978a; Wilkinson, 1985), and is reported in association with positive implant bacterial cultures (Parsons et al., 1993; Virden et al., 1992) or calcification around the implant (Peters et al., 1998).
There are a number of specific reports of breast pain associated with implantation (Cuéllar and Espinoza, 1996; Huang, 1990; Jabaley and Das, 1986; Janson, 1985; Lu et al., 1993, 1994; Sichere et al., 1995). These reports describe some severe chest, subscapular and arm pain syndromes, and unusual presentations in women with implants, and they list some of the indications for explantation. However, as others have pointed out, chest pain is a common complaint, and evidence to support the association of pain with implants in some of these cases, which come from highly selected groups, is not persuasive (Kulig et al., 1996; Mogelvang, 1996). As Wallace et al. (1996) discuss, pain, like sensory change, which is of similar frequency, is not surprising given the damage to the nerves to the breast and nipple during implantation and reconstruction surgery and the routine injury to the nerves including the intercostobrachial nerve (to the arm) during mastectomy with axillary dissection (Benediktsson et al., 1997 and reviewed in Courtiss and Goldwyn, 1976); see values of 41.6% permanent nipple sensory changes (Fiala et al., 1993); and 41% change (Hetter, 1979); 18% decrease in sensation (Hetter, 1991); nerve damage and paresis (Laban and Kon, 1990; Wallace et al., 1996); and partial to complete sensory loss in the nipple of 70% and in the whole breast of 12% after augmentation, although it should be noted that this was an explant series with a high (65%) incidence of breast pain (Peters et al., 1997).
The frequency of local and perioperative complications has been substantial in both augmentation and reconstruction of the breast with either saline- or gel-filled silicone implants. These complications have safety implications, because they may have health consequences of their own and because they may result in further operative or medical interventions that may also have health consequences. The committee sees little justification for some of these interventions, for example, closed capsulotomies or the use of steroids.
Much information in this chapter may not apply to the present and may not provide a basis for decisions concerning future experiences because past reports of complications reflect experience with implants having physical and chemical characteristics that differ from current implants and surgical practices that differ from current practices. Although the present state of knowledge does not allow definite conclusions to be drawn about the prevalence or incidence of some complications, some of the more common complications such as rupture, deflation, and contracture may be becoming less frequent due to operative and technological improvements. Information to permit conclusions about the frequency, causes, and management of complications has to be gathered based on research on a stable population of standardized devices. Much remains to be learned about the basic biology of foreign body, silicone, and other polymer interactions with tissue, although progress has been made recently.
The committee drew conclusions about ruptures and deflations, the role of silicone in contracture, saline versus gel implants, barrier shells and shell texturing, submuscular placement of implants, the roles of infection and hematomas, the use of adrenal steroid, pain and other outcomes that can affect reoperations and local and perioperative complications. In general, however, the frequency of reoperations and local complications is sufficient to be of concern to the committee and to justify the conclusion that this is the primary safety issue with silicone breast implants, and it is certainly sufficient to require very careful and thorough provision of the kind of information contained in this chapter to women considering breast implant surgery. The committee concludes that many of these risks continue to accumulate over the lifetime of a breast implant.