There is no such thing as a "standard" breast implant. Silicone breast implants since 1962 have in common the presence of silicone in the shell with or without silicone in the contents of the shell. There are, however, a substantial number of characteristics that differentiate the more than 240 U.S.-made breast implants and expanders (Middleton, 1997). Many of these features are of slight consequence, but at least some of them have, or are reported to have, important influences on the biologic responses to, and complications of, implantation. Consideration should be given to the potential contributions of these influences as well as to the basic influence of silicone generically or of a specific silicone compound when assessing the consequences of breast implantation.
Unfortunately, the medical literature describing clinical experience with local and systemic complications of implants has often not specified the make and model of implants or their important characteristics. This has been improving, especially when reporting prospective trials (e.g., Burkhardt et al., 1986; Burkhardt and Demas, 1994; Burkhardt and Eades, 1995; Coleman et al., 1991; Hakelius and Ohlsen, 1992; Weinzweig et al., 1998). Many reports have been retrospective, however, and routinely women are unable to identify their implants. Medical records may be incomplete in this respect, or authors may not have appreciated how different implants can be and the value of taking the time to characterize implant populations by type, model, and manufacturer. Some implants
were custom made for individual plastic surgeons or were made in small numbers by even the major manufacturers. Breast implants in the United States have come from one of more than ten companies, most of which no longer exist and many of which changed names and ownership over the years. Implants also were made in less usual types such as triple-lumen, gel-saline single lumen, reverse double lumen, or gel-gel double lumen. Little attention has been paid to reporting these minor product lines, and each one is likely to constitute a small number in any reported series. In many of these instances, implant records, descriptions, and specifications may not exist or may be only partially complete from the most detailed clinical source or the manufacturers themselves (Middleton, 1998a).
Not all of the implant variations noted here will have important clinical implications, but some have the potential to, are hypothesized to, or have been reported to cause local or systemic health effects or complications that are relevant to the safety of silicone breast implants. Potential associations (e.g., shell thickness and rupture, shell texturing or coating and contracture, gel or saline fill and systemic effects, gel consistency and tissue penetration, gel composition and toxic response) are noted in the relevant chapters of this report. This chapter describes the kinds of implants as far as possible, and gives important product names and dates of market entry and exit to help identify the implants that are, or might be, involved in studies reported in the medical literature and to help understand the implications of these studies. More complete knowledge of the specific characteristics of implants in use at various times might provide some guidance in interpreting clinical reports. Such information would be useful prospectively. The committee concluded that as complete information as possible about the device itself would be helpful to an understanding of the safety of silicone breast implants.
There are a number of examples in the plastic surgery literature in which an appreciation of the implants being used could enhance an understanding of the medical and scientific implications of a particular report. Relatively thick shells in early-1970s implants may have retarded the outward diffusion of high-dose intraimplant SoluMedrol and explained the reported lack of steroid complications reported by Perrin (1976) as noted later (Cohen, 1978a; Perrin, 1978). Measurements of peri-implant, capsular tissue silicon or silicone levels vary independently of implant age and capsule pharmacological reactivity but might be related to shell (or gel) characteristics (Baker et al., 1982; Evans and Baldwin, 1996; Peters et al., 1995a; Teuber et al., 1995a). One report showed a "poor but positive" correlation of silicon levels with the age of a group of implants that presumably all had low-bleed shells (McConnell et al., 1997).
High frequencies of saline implant deflation have to be interpreted in the light of implant types (Grossman, 1973; Worton et al., 1980). Early
Simaplast, Klein-type, implants were fragile and had a 76% prevalence of leakage or deflation (Williams, 1972). Early HTV (high-temperature vulcanized) models from a number of manufacturers frequently deflated (Mladick, 1993). The Jenny, Heyer-Schulte models had a reported thickness of at least 0.016 inch, or 0.40 mm, that made them sturdy (Jenny, 1994; Schmidt, 1980). Modern RTV (room temperature vulcanized) models are reported to deflate infrequently (Gutowski et al., 1997; Mladick, 1993; and see discussion in Peters, 1997).
The high prevalence of shell rupture in one gel implant was said to reflect the unique process of dipping the gel into elastomer rather than filling a preformed elastomer rubber casing used with this particular model (MemeME, Aesthetech). Later models were said to correct this problem (MemeMP) and were certainly no longer made in this way (Middleton, 1998b; Middleton and McNamara, 1995). Early Ashley-type polyurethane-coated prostheses were also prone to rupture along the seam until this was modified (Cohney and Mitchell, 1997). Prevalence of rupture differing by brand of implant has been reported by Feng (IOM Scientific Workshop, 1998), and Peters and Francel have reported major differences in rupture for silicone gel implants of different vintages, up to 95% at 12 years' implantation with thin shelled, 1972-mid 1980s-implants (Francel et al., 1998; Peters et al., 1996). Others have also reported frequent rupture for these thinner-shelled implants, although the numbers could also reflect the effects of wear on long-duration implants and the biases inherent in patient groups that are identified because they present with problems that lead to explantation (De Camara et al., 1993; Harris et al., 1993; Malata et al., 1994a; Rolland et al., 1989a).
Separate descriptions of women with saline and gel implants in patient populations with systemic signs and symptoms (Cuéllar et al., 1995a; Dobke et al., 1995) might direct further inquiry, although some report similar autoantibodies in both saline and gel implant patients and atypical disease symptoms in saline implant patients (Byron et al., 1984; Martin et al., 1993; Miller et al., 1998; Vargas, 1979). Calcification has been associated with a particular make of implant, although it can occur around any implant (Peters et al., 1998; Rolland et al., 1989b). Implant rupture after closed compression capsulotomies is reported to depend on implant vintage (Lemperle and Exner, 1993). Other examples could be cited, but it is clear that complications such as capsular contracture, rupture, and silicone migration may vary with implants from different manufacturers based on factors that are not precisely identified, including different shells, different gel consistency and diffusion characteristics, different gel chemical composition and siloxane molecular weights, different shapes, and so forth (Ksander and Vistnes, 1985).
Implants come in a great range of sizes, generally from 80 to 800 cubic centimeters (cc) or milliliters in volume, although 1,000-cc implants or expanders have been used on occasion (R.A. Ersek, personal communication, 1998; Hester and Bostwick, 1990). The diameter, or largest dimension, of breast implants ranges from 7.5 to 16.8 cm and the projection, or profile, from 1.5 to 7.5 cm. These dimensions make breast implants dramatically larger foreign bodies in both surface area and volume than most other alloplastic implants, especially when placed bilaterally (Gold, 1983; Hester and Bostwick, 1990; Middleton, 1998a). Various shapes such as round, oval, teardrop, or contoured have been available.
There are a number of major construction types of implants. Single-lumen implants have a single silicone elastomer shell traditionally filled with silicone gel. The gel is composed of cross-linked silicone permeated by silicone fluid (as described in Chapter 2). The chemical composition and weight-average molecular weight of the gel differs from manufacturer to manufacturer and from time to time (Dorne et al., 1995). As measured by extraction studies, the fluid may comprise 30-80% (Council on Scientific Affairs, 1993). Changes in fluid content affect the consistency and feel of the implant. Saline (at salt concentrations similar to those of body fluids) is used to fill shells. Rarely, other materials such as dextran, polyvinyl-pyrrolidone, or recently, soybean oil have also been used.
Some of these implants (valved) may be inflatable, so that their volume can be changed during implantation. Expanders are specifically designed to be inflated incrementally postsurgically. This creates an enlarging tissue pocket either to accommodate a replacement implant after a period of tissue adjustment or, less frequently, to accommodate the expander permanently. Expanders are used most often in reconstruction. Expansion actually leads to the formation of new tissue after operative excision of tissues for cancer or other conditions. This usually entails epidermal thickening, proliferation of blood vessels, and some dermal thinning with thick collagen bundles and myofibroblasts and muscle cell degenerative changes and loss in both the experimental and clinical situation (Argenta, 1984a, b; Argenta et al., 1985a; Austad et al., 1982; Gur et al., 1998; Johanson et al., 1993; Kim et al., 1993; Leighton et al., 1988; Pasyk et al., 1982; Rowsell et al., 1986).
Since the early 1980s, designs have been available with detachable reservoirs to allow the expander to be left as a permanent placement (Becker, 1984, 1986a; Gibney, 1989). About 14% of reconstruction implants were estimated to be permanent expanders (Zones, 1992). According to American Society of Plastic and Reconstructive Surgeons' (ASPRS, 1997) data, 31.66% of breast reconstructions involve the use of some kind of an
expander. Expanders can be replaced with various permanent implants. Some surgeons report advantages to augmentation patients in the use of permanent gel-saline or saline expanders with the inflation ports left in place (Berrino et al., 1998a). Inflatable implants usually are saline filled, but rarely they contain gel with provision for changing size by the addition of saline to the gel-filled lumen. Expanders contain saline in a single lumen or sometimes an expandable saline lumen inside a gel lumen. Some expanders are directionally expandable through shell modification or pig-gyback-bonded separate expanders (Hester and Bostwick, 1990).
Historically, single-lumen, gel-filled implants (that include polyure-thane-coated implants) have been the most commonly used, approximating 60-80% of devices implanted (Middleton, 1997; Zones, 1992). However, since the 1992 FDA moratorium on gel-filled implants, single-lumen saline implants, that historically comprised 5% of implants (Gabriel et al., 1994; Zones, 1992), have almost completely replaced gel implants (see Chapter 1). Saline as the filler for implants was originally considered, but discarded by Cronin in the early 1960s, presumably because gel implants ''remained normally expanded even when torn" (Cronin and Gerow, 1963) and "it seemed unlikely that leakage could be avoided for life" (Cronin, 1983a). Saline-filled implants were reported in France in 1965 (Arion, 1965) and in the United States in 1969 (Tabari, 1969initially filled with dextran 6%). Saline implants were made earlier (1965) in New York City according to some observers (R.A. Ersek, personal communication, June 10, 1998). Their lesser popularity has been ascribed to a number of factors. These include: high deflation rates (since corrected) and leaky valves (Capozzi, 1986; Lantieri et al., 1997; Peters, 1997); a weight said to be up to 8% heavier than a comparable-volume gel implant (Barney, 1974); a fluid wave ("slosh effect") with sounds that can be heard by the patient (Asplund, 1984); and thin consistency and wrinkling that is visible through the skin and/or palpable (Gylbert et al., 1990a; Melmed, 1998; Slavin and Goldwyn, 1995). These undesirable features are alleviated to a large extent by the more tissue-like consistency of gel and its lower tendency to shift with different patient positions. Placement of the saline implant deep to the muscles of the chest and slight overfilling also minimize some of these problems (Nicolle, 1996). They must also be balanced against the problems of gel, such as gel fluid diffusion through the shell into the tissues; axillary adenopathy secondary to silicone; release of gel on implant rupture, which can be removed only incompletely by surgery and includes the possibility of gel migration and granuloma formation; higher incidences of contracture; and greater radiopacity of gel.
Standard double-lumen implants have two shells either connected or patched together or floating freely one in the other. The inner lumen is gel filled and the outer saline filled. In the case of reverse double lumen
(implant expander), there is an inner saline and an outer gel-filled lumen. The idea here was that the thicker gel would minimize wrinkling, which led to leaks, while at the same time the expandable inner lumen would allow for some volume adjustment and potentially intraluminal medication (Colon, 1982). Double-gel or reverse adjustable double-lumen (gel-gel with provision to add saline to the inner gel) implants were rarely made (see company data below). The gel part of the standard double-lumen implant provides the cosmetic and other advantages of gel, and the saline part provides an inflatable or expander option and a reservoir for adrenal steroid, which, although not an FDA approved use, has been reported to minimize capsular contracture (Lemperle and Exner, 1993; Spear et al., 1991; and others, as noted in Chapter 5 of this report).
Aside from the expander and reservoir possibilities, the outer saline lumen was also supposed to form, in theory at least, an additional barrier against gel fluid diffusion or gel leakage or rupture and resulting silicone tissue residuals and granulomas (von Frey et al., 1992). This feature by itself was often ineffective (Melmed, 1998); in fact, no data could be found to support such a function. Yu et al. (1996) measured quantities of gel fluid diffusing from an explanted double-lumen implant that were consistent with amounts measured from implants of various kinds and were about twice as much as from a single-lumen gel implant explanted from the other breast of the same patient. Additional data from this group showed silicone gel fluid diffusion from double-lumen implants in greater amounts than from either barriered or unbarriered gel implants (Marrota et al., 1996a). Cocke and Sampson demonstrated silicone fluid droplets in cells of capsules of double-lumen breast implants biopsied four months after implantation by transmission electron microscopy and electron dispersion x-ray analysis (Cocke and Sampson, 1987).
The outer lumen of a double-lumen implant was designed to be punctured and deflated as a corrective to capsular contracture (Hartley, 1976), but this kind of implant did not appear to decrease the incidence of contractures (De Blecourt et al., 1989; Vasquez et al., 1987). The volume of the gel or saline outer lumen has been less than the volume of the inner lumen, varying roughly from 10 to 40% of the inner-lumen volume (Hester and Bostwick, 1990). At one time, standard double-lumen implants comprised about 12-15% of breast implants (Middleton, 1997; Zones, 1992); but since the 1992 FDA moratorium, they have given way to single lumen-saline implants and have been implanted 1% or less of the time (ASPRS, 1996). Triple-lumen implants, also seldom used, have three shells, the inner and middle filled with gel and the outer with saline.
Implants of gel alone (with no shell) were used, mostly, but not always outside the United States (Freeman, 1972, 1977). However, two gel implants without shells, the Cavon implant (CUI International) and an
Aesthetech implant are reported to have been made and sold in the United States (Middleton, 1998a). Some surgeons used small amounts of gel to fill in around implants (Middleton, 1998b), injected gel directly (Boo-Chai, 1969) or injected fluid with a catalyst to create RTV silicone gel within the breast (Conway and Goulian, 1963). Though unapproved, silicone fluid alone was formerly injected into breasts for augmentation in the United States (Ellenbogen and Rubin, 1975; Kagan, 1963; Vinnik, 1976a; see also Chapter 1 of this report).
Implant Shell Characteristics
Implant shells are made of silicone rubber, that is, elastomer with a filler. They vary in the composition and characteristics of the elastomer (e.g., approximately 21-27% amorphous silica filler in the elastomer for the shell and in shell patches, and 16.5% in barrier coats according to Dow Corning). Specifications of other manufacturers may vary. Amorphous silica is different in its physicochemical properties and in its biologic effects from crystalline silica, which is reported not to be present in measurable amounts in implant shells or gels (see Chapter 2; see also IARC, 1997; Iler, 1981). Shell thickness also varies, ranging from 0.13 to 0.75 mm, or 0.005 to 0.030 inch. Some shells have been even thicker, and areas of some implant shells lie outside this range (J. Curtis, Dow Corning, personal communication, February 17, 1998; P. Klein, Dow Corning, personal communication, August 10, 1998; Z.F.Twardochleb, McGhan Medical, personal communication, July 7, 1998). Most shells have had smooth elastomer rubber, but increasingly, some are textured with different surface features or shell projections of varying coarseness, depending on the manufacturer.
Shell Polyurethane Coating
Texturing was a reaction to the success in reducing capsular contracture of the original 1- or 2-mm-thick poly(ester)urethane foam-textured coating of a regular silicone gel-filled implant (Ashley, 1970, 1972). The foam was produced from the polymerization of polydiethylene glycol adipate with a 4:1 mixture of 2,4- and 2,6-toluene diisocyanate and was secured by an RTV silicone adhesive primarily to single-lumen gel implants. Occasionally standard double-lumen, saline, and gel-saline implants were coated with polyurethane foam. The foam had 80-100 open pores of 200-500-mm in diameter per linear inch (Batich et al., 1989; Mishra, in FDA, 1990). About 1.35 g of foam covered an average implant (Szycher and Siciliano, 1991a).
The polyurethane coating of implants was popular. An estimated
110,000 women, or about 10% cumulatively (FDA, 1995) or 18.8% in a given year (e.g., 1990; Zones, 1992) had implants with polyurethane coating because it secured the implant and clearly reduced early contracture (Capozzi, 1991; Capozzi and Pennisi, 1981; Hester et al., 1988). The coating may not have amounted to more than a mostly temporary and no more effective form of texturing than that described below (Brand, 1984; Burkhardt and Eades, 1995; McCurdy, 1990). However, there are suggestions that polyurethane specifically inhibited fibroblasts, had some specific effects on immune cells (Bradley et al., 1994a), and either specifically or because of fragmentation into many small particles (see below and Goodman et al., 1988) was particularly effective in causing an acute and continuing chronic inflammatory response that postponed the mature fibrotic phase and accounted for delayed capsule formation (Brand, 1988; Devor et al., 1993; Sank et al., 1993; Smahel, 1978a; Whalen, 1988) with significantly less contracture than texturing of the silicone shell alone (Handel et al., 1995).
A number of reports established the success of polyurethane-coated implants in reducing the frequency of contractures in either the submammary or submuscular position in the first few years after implantation (Cohney et al., 1992; Handel et al., 1991a; Herman, 1984; Hester et al., 1988; Hoffman, 1989; Melmed, 1988; Pennisi, 1985, 1990; Wells, 1988), but it is difficult to be sure whether the long-term contracture result after disintegration of the coating was better than with other implants. For example, significant contractures occurred late in Cohney's large, long-term study (Cohney et al., 1992). An occasional synovial lining (see below) around a polyurethane implant was reported, which may also have influenced capsule formation, although that report described implants of a young age (2.5 years) (Raso and Greene, 1995). A tendency to seroma and the occurrence of late pain have also been reported (Jabaley and Das, 1986; Wilkinson, 1985). The cellular characteristics of any implant capsule most likely depend on a number of factors, some of which are as yet unknown (Hardt et al., 1994).
Polyurethane foam was said to undergo partial chemical degradation under physiologic conditions, releasing compounds that are, or could become, carcinogens in animals but are not known human carcinogens (i.e., 2, 4-toluenediamine [2, 4-TDA] or precursors) (Benoit, 1993; Chan et al., 1991a,b; Luu et al., 1994; NTP, 1978); however, these are reportedly released in very small amounts that would not present an unacceptable risk (Expert Panel, 1991; FDA, 1991b, 1995). Other evidence suggests that chemical degradation, although possible (Dillon and Hughes, 1992), may have been minimal (Amin et al., 1993; Hester et al., 1997; Szycher et al., 1991; Szycher and Siciliano, 1991a). Furthermore, epidemiological human and experimental animal evidence does not support an association be-
tween cancer and polyurethane (Brand, 1988; Cohney et al., 1992; Devor et al., 1993; Hagmar et al., 1993; Lemen and Wolfe, 1993; Sorahan and Pope 1993).
On the other hand, Sepai et al. (1995) reported levels of 2, 4-TDA in drainage from the implant operative site and, over a two-year period, in the plasma (> 4 ng/ml) of women with polyurethane-coated breast implants that were higher than had been considered previously in assessing cancer risks from these implants suggesting that the exposures and risks might be problematic. These risks had been estimated by the FDA at between 5 in 10,000,000 and 111 in 1,000,000 lifetime cancers in women with polyeurethane implants depending on the currently available data or a worst case 100% degradation to 2-4-TDA, respectively. The risk of 5 in 10,000,000 was considered reasonable by others (Expert Panel, 1991).
Given the relatively small number of women with polyurethane implants still in place, the natural breast cancer incidence in women, and the lack of evidence for polyurethane carcinogenesis, which implies at most a small effect, if any, of polyurethane in causing human cancer, it is unlikely that any study of patients with existing implants will be able to provide sufficient evidence of an association between these implants and cancer. At present, evidence is lacking to conclude that there is an association between polyurethane-coated implants and cancer, and the weight of existing evidence suggests that there is no such association. Since the implantation of polyurethane-coated breast implants within the United States is unlikely, these conflicting studies may never be reconciled.
In the human breast, the 1.5- to 2-mm foam coating separated from the implant surface and disintegrated physically as well as chemically beginning almost immediately and progressing over a few years; what remained was a mostly smooth implant with a capsule containing polyurethane fragments (Guidoin et al., 1991a,b; Hester et al., 1988; Shapiro, 1989; Slade and Peterson, 1982; Smahel, 1978a,b; Szycher et al., 1991) although there has not been complete agreement on this point (Szycher and Siciliano, 1991b). The same disintegration into scattered polyurethane fragments and migrating fibers has been observed in experimental female mice within 47 weeks (Devor et al., 1993), and half-lives for biodegradation of about 21 months or 2-3 years have been calculated, respectively, from human 2,4-TDA excretion data and explanted implants (Hester et al., 1997; Sinclair et al., 1993).
Although some studies reported that the adherent capsule was more resistant to infection (Merritt et al., 1979; Whalen, 1988), others reported major problems with eliminating infections in capsules because disintegrated foam acted as multiple foreign bodies and was often very difficult to remove (Berrino et al., 1990; Bruck, 1990; Guidoin et al., 1991a; Hardt et al., 1994; Hester, 1988; Melmed, 1988; Okunski and Chowdary, 1987;
Umansky and Wilkinson, 1985; Wilkinson, 1985). In addition to problems in removing foam fragments, the polyurethane implant itself was difficult to explant (Bruck, 1990; Cohney and Mitchell, 1997). There is insufficient evidence that infection was more frequent in these implants (Handel et al., 1991a; Hester et al., 1988; Melmed, 1988, Pennisi, 1990). The coated shell was also notably more radiopaque on mammography than an uncoated shell (Young et al., 1991a) and may have interfered with magnetic resonance imaging for rupture (Ahn et al., 1993). Domestic polyurethane foam-coated implants were discontinued in 1991.
Implant Shell Texturing
The form of texturing of silicone elastomer implant shells varied considerably by manufacturer. For example, Dow Coming Silastic MSI (Micro Structured Implant) had regular pillars 750 mm high, 250 mm in diameter, and 500 mm apart. McGhan Biocell uses an open pore network, 3.1 pores/mm2, average size 289 mm, height 500-800 mm. Mentor Siltex has surface irregularities 65-150 mm high and 60-275 mm wide. Bioplasty MISTI consisted of pores 20-800 mm wide (Barone et al., 1992; Dow Corning, 1991, p. 20021; Jenkins et al., 1996; Maxwell and Hammond, 1993). Most of the originally smooth major construction types were sold with texturing, beginning in the mid-1980s, to minimize capsular contracture (Hakelius and Ohlsen, 1997; Hammerstad et al., 1996; Maxwell and Hammond, 1993; Pollock, 1993).
Texturing was assumed to work to reduce capsular contracture in ways similar to polyurethane; that is, tissue grew into the interstices of projections or pores, prolonging chronic inflammation, disorienting collagen fibrils, and weakening their contractile forces (Barone et al., 1992; Whalen, 1988). Some descriptions of textured gel- or saline-filled implant capsules also note the formation of peri-implant synovial tissue, that is, a joint lining-like tissue reaction that, along with a more cellular capsule, is said to check excessive contracture (Raso and Greene, 1997). The synovial cells around the implant, which appear on monoclonal antibody characterization to be the same as joint synovium, are reported to be more common around textured implants (Bleiweiss and Copeland, 1995; Copeland et al., 1994; Luke et al., 1997; Wickman et al., 1993). They are of monocyte or macrophage origin. There is suggestive evidence that some have phagocytic and transport functions that may have the capacity to transmit particulates such as silicone microdroplets or elastomer fragments (or, experimentally, colloidal carbon) outside the capsule to local lymph nodes and that others have a secretory function that may contribute to fluid surrounding the implant (Emery et al., 1994). Synovial lining has also been reported around smooth gel and saline implants (Burmester et al.,
1983; Chase et al., 1994a; del Rosario et al., 1995; Hameed et al., 1995; Luke et al., 1997; Raso et al., 1994a,b, 1995a,b).
Some studies have concluded that this synovial lining is a reaction to friction in a surgical cavity (i.e., a bursa) and is indistinguishable from the synovial lining of joints and normal bursae (Copeland et al., 1994; Emery et al., 1994). Recent analysis proposes that synovium is a transitional phase, inversely related to implant age and unrelated to other factors such as implant surface, placement, and capsule or gel fluid diffusion through the shell (Chase et al., 1996; Ko et al., 1996). Wyatt et al. (1998) recently reported 93 capsules from a variety of smooth and textured implants examined primarily during capsulectomy for contracture. The smooth implants had been in place for an average of 104 months, the textured for 67.7 months. Synovial cellular reaction decreased significantly with time around both smooth and textured implants (p = .003 and .051, respectively). Possibly synovium, beginning in bursae formed in response to friction from movement common to all breast implants, may mature over time to a fibrous tissue sheath with increasing predominance of fibroblastoid cells that were originally in the minority (Raso et al., 1994a). Other studies have concluded that synovium is a fundamental biological phenomenon in tissue spaces exposed to friction, which is seen in about 25% of implant capsules but is not particular to such sites (Schnitt, 1995). Still other reports put the percentage of synovial villous hyperplasia considerably higher (63%) in early textured implant capsules (Wyatt et al., 1998). Texturing may also cause more peri-implant fluid, in part at least due to the secretion of proteoglycans by synovial secretory cells, although some fluid around implants in general is fairly common. In a study of explants from symptomatic women, intracapsular fluid (0.2-20 ml) was found around 15% of implants, was not related to infection or any particular symptoms, but was more frequent around textured implants (Ahn et al., 1995a). Texturing is reported by some to be variably more (e.g., polyurethane, Biocell) or less effective in providing tissue fixation of implants (Maxwell and Hammond, 1993). Different patterns and depths of surface texturing seems to make a difference in cellular behavior and probably in clinical results. (For further discussion, see Chapter 5.)
In more recent times, many gel-filled implant models have had shells constructed to lessen the diffusion of silicone fluid compounds into the tissues. Either these models add one or two shell layers of diphenyl or other modified siloxanes, or they interpose a layer of fluorosilicone between the shell and the gel contents. In the case of Dow Corning, this fluorosilicone layer is reported to have limited silicone fluid "bleed" (or
diffusion, as measured by various in vitro techniques designed to promote flow) to an estimated 96 mg per year per 300-ml implant, compared to 487 mg per year per implant in older models (T. Lane, Dow Corning, personal communication, April 28, 1998). Other, independent measurements of various implant models (Yu et al., 1996) put silicone gel fluid diffusion at about 300 mg per year, with considerable variation depending on implant age and manufacturer. Figures of 60-100 mg per year for pre-barrier and 5-10 mg per year for barrier implants are quoted frequently in the literature (Independent Advisory Committee on Silicone Gel-Filled Breast Implants, 1992), but these are probably just early Dow Corning estimates. Other early measurements varied between 15 and 45 mg at 12 weeks, depending on implant make (Bergman and van der Ende, 1979). In an accelerated bleed test (300ºF, forced-air furnace) by Battelle, the Dow Corning Silastic II accumulated 14 mg of silicone gel fluid diffusing out through the shell at eight weeks. Implants of other manufacturers accumulated between 300 and 400 mg (Morey and North, 1986). Diffusion through the shell depends on the maintenance of a concentration gradient by the removal of fluid from the outside surface and through the capsule. Since the capsule appears to present a barrier to silicone movement, these estimates of the amount of fluid diffusion may be high, because they do not reproduce this feature (Beekman et al., 1997a; Yu et al., 1996).
In implants with a gel fluid number molecular weight of 13,840 g/ mol, gel fluid diffusion averaged a number molecular weight of 11,630 g/ mol (less than 0.05% low molecular weight cyclic siloxanes, D4-D6) or 6,194 g/mol (less than 0.4% low molecular weight cyclics, molecular weight of D4 = 296) depending on the absence or presence, respectively, of a barrier coat (IOM Scientific Workshop, 1998). A minority percentage of higher molecular weight siloxanes, up to 400,000, was also reported in diffusate from implants without barrier coats, possibly from uncross-linked shell silicones (Varaprath, 1991, 1992). A silicone gel implant is said to contain about 855 parts per million (ppm) of D4, or about 256 mg in an average-size (300-cc) gel implant (T.H. Lane and J.J. Kennan, personal communication, 1998). Using different methods, others have reported 1 mg per day of low molecular weight cyclic diffusion into surrounding hydrophilic media in vitro (Lykissa et al., 1997; E.D. Lykissa, personal communication, July 29, 1998). This figure seems high given the measurements cited above for total diffusion and the percentage of these compounds in the gel fluid. Presumably if this were to continue, the entire amount of D4 in an implant would diffuse out in a year, which seems unlikely.
The fluorosilicone layer, which is a thin (about 10-µm) barrier coat on the interior surface of the implant shell, slowed diffusion because it had a solubility parameter quite different from that of the gel fluid. This was
unfavorable to higher molecular weight silicones so diffusion was composed of relatively more low molecular weight silicone compounds, as noted above (Van Dyke, 1994; Van Dyke and Fowler, 1994). It also was not as strong as the usual elastomer rubber and typically added to shell thickness, although only slightly (about 4%). Reports of capsular and blood silicon levels in patients with implants from many different manufacturers have not substantially changed over time when shells of varying barrier effectiveness were used. In addition, several authors have reported that fluorosilicone barrier shells lose their effectiveness after two or three years, presumably due to fracture of this weaker elastomer (Baker et al., 1984; McConnell et al., 1997; Peters et al., 1995a,b, 1996; Yu et al., 1996).
Other Characteristics of Implant Shells and Gels
Rare shells, like the early polyurethane Ashley implant, had inverted Y-shaped internal dividers, presumably designed to control the shape of the gel and implant, keeping the gel from sagging to the dependent and central part of the implant. This feature was continued in the Optimam polyurethane implant until 1991. Patches in almost endless variety have been made of elastomer rubber, elastomer-Dacron, or Dacron alone [poly-(ethylene terephthalate)-based cloth]. Seal patches were used to seal holes and slits left in shells during manufacturing and to reinforce and seal the shell entries of valves or valve tubes used for filling inflatables and expanders. Fixation patches were designed in a vast array of sizes, shapes, and locations to be infiltrated by tissue and thereby keep the implant from sliding around within the breast. Fixation patches were found, or could be ordered, on almost all major implant types. Expanders meant to be replaced by permanent implants were rarely made with these patches. By the 1980s, however, fixation patches had fallen out or favor. They were felt to contribute to scarring around the implant (Williams, 1972), to a higher incidence of peri-implant calcification (Luke et al., 1997; Rolland et al., 1989b) and possibly to an increased likelihood of rupture (De Camara and Sheridan, 1993; Malata et al., 1994a). Patches as seals are still needed for manufacturing of modem seamless implants.
A variety of access ports and valves accessible through the skin have been available for inflatables and expanders, both at the implant and at a distance. Some of these remain with the implant. Some are removed after the final size of the implant has been achieved. Some valves have been reported to be quite insecure or to leak, which is said to contribute to implant deflation or microbial contamination of the saline in an implant. Middleton (1997, 1998a) listed six types of implants or expanders with valves and characterized and extensively described the various kinds of valves and patches.
The silicone gel-filled implant has undergone a number of changes over time. Its physical qualifies have been altered, making it firmer or softer and more or less elastic. These changes and the addition of barrier layers and other shell changes may result in some different molecular species in gel fluid diffusion, although the fluid itself, at least in the case of Dow Corning, is said to have remained the same since 1975. Various catalysts have also been used as the manufacturing process has changed over time. Traces of these remain in the gel and could in theory diffuse through the shell. These include 1,3-diethenyl-1,1,3,3-tetramethyl-disiloxane-platinum complexes and methylvinylcyclosiloxane-platinum complexes, among others. These substances, bis-(2,4-dichlorobenzoyl) peroxide, and tin compounds such as stannous octoate or oleate or di-butyltin dilaurate (see Chapter 4 section on tin) were used in the elas-tomer or seal patch adhesive or RTV saline implant shells. The platinum compounds (said to be in the zero oxidative state) are generally reported to comprise 0.9 ppm of the gels and about 8-10 ppm of the shell and patch (J. Kennan, Dow Corning, personal communication, April 28, 1998; also NuSil, Compton, 1997) although this may vary with different gels or shells. The average Silastic I 305-cc implant contained 281 µg of platinum, 74% (207 µg) of which was in the gel presumably as a colloidal, elemental platinum residual (J. Curtis, Dow Corning, personal communication, April 28, 1998; Lewis et al., 1997). The conventional implant size of around 300 ml is based on company representations and actual experience with a 1,317-implant series averaging 284 cc (Middleton, 1998b) though lower averages have been reported (e.g., 247 cc, Fiala et al., 1993). Older model Silastic 0 implants contained more platinumabout 480 [µg (Dow Corning IOM Scientific Workshop, 1998). Another investigator reported 4.5 µg/g platinum in gel from a Dow Corning implant, or about 1.3 mg per average implant, a higher value with questionable biological significance (El-Jammal and Templeton, 1995).1 Analytical measurement of curing agents, solvents, and catalysts (process aids; see Chapter 2) by 48-hour dichloromethane and 24-hour 70ºC saline extraction gave values of 20 ppm or less for these substances from extractions except for butylcarbitol acetate (up to 703 ppm). These substances are present at very low levels (T. Lane and J. Kennan, Dow Corning, personal communication, April 28, 1998)2 and seem unlikely to contribute significantly to tissue levels by diffusion through the shell.
1 Since the value of 207 µg is the actual amount of platinum added to about 300 g of gel, according to Dow Corning, the higher measured value is confusing.
2Acetone, butylcarbitol acetate, 2,4,-dichlorobenzoic acid, ethanol, 1-ethynyl-1-cyclo-hexanol, 2-methyl-3-butyn-2-ol, 1,1,1-trichlorethane, xylene, platinum, tin, and zinc were the substances measured.
Since silicone elastomers and gels may be irradiated under clinical conditions in patients with breast cancer, their stability and effect on the radiation beam when subjected to the doses delivered to such patients are important considerations. Although tests of a full range of elastomers and gels of various manufacturers have not been reported, the relatively small differences in silicones of various breast implants probably have minimal, if any, effect on stability or transmission characteristics. Tests of Surgitek, McGhan, and Dow Coming products have been reported. These reports and more generic reviews indicate that silicone gel breast implants should have no more clinically significant effect on radiation therapy than an equivalent amount of breast tissue or saline. Their physical characteristics and silicones themselves should remain relatively stable at dose levels up to 7,500 rads, although some discoloration of the gels and loss of compliance has been reported after irradiation (Klein and Kuske, 1993; Krishnan and Krishnan, 1986; Krishnan et al., 1983, 1993; Kuske et al., 1991; Landfield, 1983; McGinley et al., 1980; Shedbalkar et al., 1980). The radiation stability of silicone, partcularly platinum cured and phenyl methyl, is considered good (Radiation Sterilization Working Group, 1996). The cosmetic results are not adversely affected in irradiated, implant-augmented breasts when radiation is administered with careful attention to technique, according to most investigators, although there are some reports of greater frequency of capsular contracture. Contractures are more frequent and results are in general less good in immediate postmastectomy reconstruction using implants and in implantation in previously irradiated breast tissue (Barreau-Pouhaer et al., 1992; Chu et al., 1992; der Hagopian et al., 1981; Dickson and Sharpe, 1987; Evans et al., 1995; Forman et al., 1998; Halpern et al., 1990; Handel et al., 1991b; Jacobson et al., 1986; Kraemer et al., 1996; Krishnan et al., 1993; Kuske et al., 1991; Lafreniere and Ketcham, 1987; Mark et al., 1996; Rosato and Dowden, 1994; Ryu et al., 1990; Spear and Maxwell, 1995; Stabile et al., 1980; Thomas et al., 1993; Victor et al., 1998; von Smitten and Sundell, 1992), although radiation does not appear to affect contracture frequency or severity in experimental animals with implants (Caffee et al., 1988; Whalen et al., 1994). As noted elsewhere in this report (see Chapter 12), silicone gel- and saline-filled implants interfere with the visualization of the entire breast on x-ray film mammography. Relative radiolucencies of various fillers for implants were reviewed by Young et al. (1993a), and silicone gel interfered with visualization the most, followed by progressively more radiolucency with saline, bio-oncotic gel, and peanut oil fillers.
In view of the many manufacturers, major construction types, varying and changing shell elastomer rubber, gel, and surface characteristics, barrier layers, and other less meaningful differences, it is easy to appreciate why there were hundreds of types of implants. In fact, if dimensions,
shape, and patch and valve characteristics are added to the variables, Middleton has estimated that as many as 8,300 different implants might have been available. Some of these can be identified by implant surface markings, which are sometimes radiopaque, or by other characteristics that are unique to a particular implant and identifiable either on explantation or by techniques such as film or MRI mammography. Identification can be useful in assessing the way implants might behave and has of course been useful in litigation (Middleton, 1997, 1998a). Presumably, gel, saline, or other filler, smooth or textured surface, barrier layer or standard elastomer shell, elastomer shell thickness, physical or chemical characteristics, other physical and chemical gel and gel fluid characteristics and compositions, and the presence and concentration of nonsilicone substances (e.g., catalysts or other substances remaining in the implant from the manufacturing process), would represent a minimum list of features that might have biomedical and health implications, either local or possibly systemic. Information on the product characteristics introduced over time by various manufacturers and distributors could help in analyzing these associations. This information, often considered in the nature of trade secrets, is not available in any detail. Even the information in this chapter was not easy to assemble and has not previously been assembled in this way.
Implants from Dow Corning
Dow Corning Corporation made the first silicone gel breast implants for Dr. Thomas D. Cronin (Cronin and Gerow, 1963) in 1962. This model reflected the ideas of Cronin, who had been considering ways to improve implants through the 1950s, and of his resident Dr. Frank Gerow, in consultation with Mr. Silas Braley of the Dow Corning Center for Aid to Medical Research. After a year of development, the first implant was placed in March 1962 (Cronin, 1983a). Dow Corning continued to make these implants through 1963 in a form similar to, or the same as, that distributed in 1964, the company's first full commercial year. By the early 1970s the Dow Corning Cronin Dacron patched implant had achieved a stunning popularity of 88% of all implantations according to one survey (Williams, 1972).
Dow Corning gel used in the early Cronin and Silastic implants and officially from 1964 to 1969 and 1969 to 1975 respectively, was thick and firm. Information on these and later gels, elastomers, adhesives, and patches can be found in Chapter 2. In general, catalyst mixtures contained 0.9% platinum complexes or about 0.2% platinum (J. Kennan, personal communication, April 28, 1998), and the swelling fluid that was added to the gel was changed to (and remained) 1,000 centistoke (cS) DC-360 fluid
in 1975 (T.H. Lane and J.J. Kennan, Dow Corning, personal communication, August 28, 1998). Also in 1975, the gel in the Silastic I model was changed so as to be softer and more compliant. This gel was used in the Silastic I, Silastic II (1981-1992), and textured models, Silastic MSI (1990-1992). Double-lumen models, first introduced in 1979, lagged in changing from Silastic I to II or adding MSI, but used the same responsive gel. Less compliant gel was partly reintroduced with shell changes in 1978 (522 and 529 series implants only) and 1981 (PO17 teardrop series implants only).
From 1964 to 1967, Dow Corning silicone elastomer rubber shells were thick and seamed. They were less thick and seamless from 1968 to 1973 (Cronin and Greenberg, 1970). As noted, chemical details can be found in Chapter 2 of this report. These two shells were, respectively, about 0.030 inch (0.75 mm) and 0.010 inch (0.25 mm) thick. The latter shell was made thinner and softer from 1973 to 1978 and was 0.005 inch (0.13 mm) thick on average. These dimensions presumably were described by Weiner et al. (1974) and may explain the differences in rupture rates reported by Peters and others, noted earlier. From 1979 to 1992 a high-performance elastomer shell was produced by dip coating. This shell, first used (1979-1981) only for saline inflatables, measured approximately 0.011 or 0.012 inch (0.28-0.30 mm) on average, depending on implant size, except when used in the textured MSI implants, where it was 0.020 inch, or 0.50 mm, thick. The fumed silica or silica aerogel content of these elastomers varied from about 24 to 37%.
In 1981, a 0.010-mm fluorosilicone barrier layer was added to the interior walls of implant shells filled with silicone gel. The barrier coat consisted of much the same components as the high-performance elastomer but used methylvinyl-co-trifluoropropylsiloxane instead of vinyl-terminated poly(dimethylsiloxane) (PDMS). As noted earlier, this barrier layer was to limit silicone fluid diffusion through the shell by changing shell solubility characteristics. It also was not as strong as usual elastomer rubber and so it could not spare elastomer thickness. Thus, the fluoro-silicone layer added to the shell thickness. This barrier was said to be more effective than others (Morey and North, 1986) and was also said to be quite different from other barrier technologies and more effective at higher (nonphysiologic) temperatures (Caffee, 1992a).
After 1968, all shells were seamless. Seamless shells were prepared by dipping a form (mandrel) into an elastomer dispersion. After removal from the mandrel, the hole where the support arm held the mandrel was patched with a silicone elastomer material. Next, silicone gel was injected into the shell. For the responsive gel-filled product, the small perforation left after injection of the gel was sealed with a drop of silicone adhesive containing methyltriacetoxysilane and ethyltriacetoxysilane as cross-link-
ers and stannous oleate as a catalyst. This adhesive was not used to bond fixation patches; a different form of elastomer was used for that purpose.
In 1968, Heyer-Schulte Corporation became the first U.S. manufacturer of saline-filled breast implants. These implants had a shell thickness of 0.016 inch or 0.40 mm. The company was acquired by American Hospital Supply in 1974 and, after name changes, by Mentor in 1984. In 1968, Heyer-Schulte manufactured a polyurethane-coated silicone gel prosthesis with an internal Y-shaped baffle called the ''Natural Y" prosthesis, which had previously been marketed by Poly Plastics Silicone Products (Ashley, 1970, 1972) pursuant to a 1968 patent by the plastic surgeon W. Pangman. Between 1971 and 1984, Heyer-Schulte introduced various additional models of single-lumen gel, inflatable saline, and double-lumen implants. These were mostly smooth surfaced. However, a polyurethane-coated adjustable gel-saline model was introduced briefly in 1973 (described by Jobe, 1978), and the Capozzi model polyurethane implant, which had no foam on its upper part in an effort to avoid fixation of the top of the implant and vertical wrinkling (see Chapter 5), was also available in the 1970s (Jobe, 1978). Data on shell thickness and gel characteristics were not made available by the company. Dorne reported weight-average gel molecular weights of 83,500 from a model 2000 (supposedly mid-1970s) implant, considerably higher than modern Silastic II gels (about 55,000). Slight differences in nuclear magnetic resonance (NMR) were detectable among gels from different manufacturers (Dorne et al., 1995); this was probably related to the firmer early versus the more compliant later gel.
Saline implant shell changes from HTV platinum catalyzed to RTV tin catalyzed toward the end of this period are noted to have produced dramatic improvements in abrasion resistance and durability (Gutowski et al., 1997; Mladick, 1993). Improvements in valves are also cited (see comments in Cocke, 1994; Gutowski et al., 1997; Lantieri et al., 1997; Mladick, 1993; and Peters, 1997, concerning the improved series 1600 and the problematic, leaky series 1800). Becker (1984) reported the development of his Mentor permanent expander at about this time and subsequently described an attached microreservoir that could be buried subcutaneously for long periods (Becker, 1986a).
Mentor continued the gel double-lumen and inflatable saline implant lines on acquisition of the company in 1984. In 1985, Mentor introduced a smooth reverse double lumen implant with gel outside and saline inside (the Becker implant, popular to this day as Siltex for reconstruction; Becker, 1987a, 1988) and saline expanders (in 1982, Radovan reported he
had been using a Heyer-Schulte expander presumably for some time). In 1987 and 1989 the Siltex (textured) lines of implants were introduced, including the Siltex single-lumen gel, and then the Siltex (Becker) textured reverse double-lumen inflatable and textured saline inflatable implants. Clinically, the Siltex saline implant had a noticeably thicker shell than the smooth model. Women were said to prefer the smooth model, despite its higher frequency of contracture, to the often palpable and visible Siltex model (Burkhardt and Demas, 1994). Barrier shells were introduced on gel-filled models, presumably in the late 1970s. McGhan licensed barrier technology to Mentor in 1990 (Z.F. Twardochleb, personal communication, December 14, 1998).
In 1992, new (post-FDA moratorium) models of Siltex expander-im-plants and contoured saline implants with various valves were introduced. The elastomer from these models was used for immune adjuvant studies reported in 1996 (Hill et al., 1996). Presumably, it was improved over earlier elastomers, but specifics are not available from the company. Like those of other manufacturers, the Heyer-Schulte and Mentor lines offer various patch and valve designs. Gel suppliers for these companies were General Electric from 1970 to 1976, Dow Corning from 1976 to 1992, and Applied Silicone from 1988 to the present.
Polyurethane History and Medical Engineering Corporation (Surgitek) Implants
The polyurethane Natural Y line was continued by Heyer-Schulte through 1978. It was then manufactured by Cox-Uphoff International until 1981. In that year, the Aesthetech Corporation was formed and began manufacturing the device. It was known as the Optimam from 1982 to 1991. The MemeME was introduced as a polyurethane-coated, single-lumen gel implant without the Y septum from 1982 to 1985. The Vogue model with a Y septum was also made from 1982 to 1985 and replaced by the MemeMP (Moderate Profile) without the Y septum from 1985 to 1991. The Replicon polyurethane implant without septum was made from 1984 to 1991. These lines were acquired by Cooper Laboratories until December 1988, when the Aesthetech Corporation was merged into the Medical Engineering Corporation (MEC), which continued to manufacture the Meme, Replicon, and Optimam implants until sales were suspended in 1991. Elastomer shell thickness reported for Meme was 0.003 inch, or 0.075 mm; for Replicon, 0.009 inch, or 0.23 mm; and for MemeMP, 0.009 inch or 0.23 mm (dates provided by Mentor Corp.; Tobin and Middleton, 1987). These dimensions probably accounted for the changes in rupture rates of these different models noted earlier. The Microthane polyurethane foam used for Aesthetech and MEC (Surgitek) implants was fabri-
cated from polyester foam supplied by Scotfoam (SIF 100 ppi Z M). The polyurethane was reported to have been made from an ethylene adipate triol with a molecular weight of 3,000 and toluene diisocyanate (20% 2,4-and 5% 2,6-toluenediisocyanate by weight). The molecular interface was composed mainly of polyester linkages and very few urethane linkages and was reported to yield very little TDA on enzymatic attack (Delclos et al., 1996; Szycher and Siciliano, 1991a).
The Medical Engineering Corporation (Surgitek) began distributing implants in 1971, originally various lines of gel-filled single-lumen implants, to which were added inflatable saline implants and expanders in 1974, double-lumen implants in 1977, and single-lumen gel-saline implants in 1978 (original Perras Papillon and later models). More compliant shells and gels were introduced in 1972. An unusual gel inflatable was made in 1973 (Dahl et al., 1974). The corporation was acquired in 1982 by Bristol Myers Squibb, which added high-profile models in 1982 and SCL (Strong shell, Cohesive gel, Low bleed) gel implants in 1986. The Natural Y polyurethane line was acquired by Bristol Myers Squibb in 1988. The polyurethane models were the only non-smooth-shelled models.
The first shells in 1971 were made of a General Electric (GE) Company copolymer of silicone and Lexan polycarbonate. Because of its stiffness, this was replaced by a GE silicone elastomer in 1973. From 1971 to 1976 the gel was GE 6193 MEC 122. From 1976 on, the shells were Dow Corning elastomer (Q7-2245, Q7-4735, and Q7-4750 at various times with various models, and Dow Corning 2213 on polyurethane models in 1981-1988), with an interior McGhan NuSil phenylsiloxane barrier on the SCL models. After 1976 the gels were supplied by Dow Corning and were similar to those used in its implants (Dow Corning Q7-2167/68). Peters reported that the low-bleed feature (SCL) introduced in 1986 did not seem to provide lasting protection, at least as determined by capsular and blood silicon levels (Peters et al., 1996). The company stopped distributing breast implants in 1991.
Inamed-McGhan Medical and CUI Implants
McGhan Medical Corporation began marketing breast implants in 1974. The company was merged with Minnesota Mining and Manufacturing Company in 1980 and subsequently acquired by First American in 1985. The company is now wholly owned by Inamed Corporation. Its implants have been sold in the United States and worldwide under the McGhan brand name. In 1975 the product line included smooth, single-lumen saline implants and expanders, single-lumen gel implants and combination gel-saline implants, and the first double-lumen implants (Hartley, 1976). Except for the single-lumen gel implants, all of these mod-
els had been updated by 1981. Individually customized inflatable implants for reconstruction after radical mastectomy were reported (Birnbaum and Olsen, 1978). Reverse double-lumen implants and expanders were added in 1987, including a double-gel model (with gel in both lumens) in 1991 and a triple-lumen model (with saline in the outer lumen and two gel lumens) in 1979. New shells were introduced periodically: a "silica free" (the exact meaning of this term is not clear) low-bleed, outer layer (Natrashiel) in 1977; a similar low-bleed shell (Natrashiel II) and a double-layer, increased-strength, decreased-bleed shell (UHP, Ultra High Performance) in 1978, and a low-bleed model containing two high-performance elastomer layers with a dimethyldiphenylsiloxane barrier layer (Intrashiel) in 1979. Gel-filled, textured and smooth models using this technology remain available.
A textured (Biocell) shell was announced in 1987 (as described earlier, Maxwell and Falcone, 1992; Maxwell and Hammond, 1993). The Intrashiel shell was the subject of an experimental study in rabbits reported in the medical literature and appeared to control silicone gel fluid diffusion from implants (Rudolph and Abraham, 1980). New shells were used on all models (except triple lumen) by 1979. In 1994, after the FDA moratorium, additional saline models were introduced. The company now sells Intrashiel, Biocell and smooth single-lumen gel, and smooth and Biocell saline models with diaphragm valves. The company's protocol for evaluation leading up to an application for premarket approval includes also a standard double lumen and a triple lumen with a gel cone within the combination gel-saline design. The company's saline tissue expander with a MagnaSite injection site, introduced in 1984, continues on the market in smooth and Biospan textured form; smooth and textured tissue expanders without the MagnaSite injection site are also available. Because the expander with the MagnaSite injection site has a magnetic locator, MRI is contraindicated in patients with these devices.
Some details of the specifications of shells and gels are considered proprietary by the company. However, measurements may be available from other sources (e.g., Intrashiel shell thickness measured 0.005-0.014 inch, 0.13-0.35 mm, in 1985; Morey and North, 1986), and other details were disclosed for this report by the company in a redacted report of confidential business information (gel implants only) that had been provided to the Independent Review Group (IRG, 1998). Shells were reported in 1994 to consist of multiple layers of modified PDMS incorporating vinyl substituents and/or modified PDMS incorporating phenyl or trifluoropropyl and vinyl substituents, the phenyl- (style 40) and trifluoropropyl- (style 246) modified silicones being the barrier layers. Treated amorphous silica is included in the shell formulation. Platinum catalysts are used, and solvents in which the shell components are dispersed dur-
ing manufacture include trichloroethane or xylene (removed after dip casting). Also, in 1994 it was reported that the basic cure for the shell and silicone gel involves the reaction of a linear PDMS containing vinyl substituents with a linear, PDMS-containing silicon-hydrogen bonds catalyzed by classical platinum catalysts. Patch and injection port materials are cured using a peroxide catalyst that decomposes with the heat of the vulcanization step. The gel contains only PDMS polymers, a significant proportion of which are not bound but are entangled in the cross-linked polymer network. Approximately 2-8% of the shell and 50% of the gel by weight are extractable by hexane, slightly more by dichloromethane. Hexane extracts of shell and gel contain low concentrations of cyclic PDMS (about 520 µg/g, D3-D6, of which 38-99 µg/g is D4 in gel and shell respectively, i.e., somewhat less than the Dow Corning extracts [0.05%] noted earlier) and linear PDMS along with trace amounts of solvents. Preliminary data suggest that the barrier shells reduce gel fluid diffusion by at least an order of magnitude compared with a McGhan nonbarrier shell (Eschbach and Schulz, 1994redacted).
Silicone gel suppliers to McGhan Medical were General Electric 1975-1976, Dow Corning 1976, McGhan 1977-1984, McGhan Nusil 1984-1992, and thereafter NuSil Technology. Although undoubtedly a generic (see Dow Corning above) rather than a specific silicone gel property, a NuSil gel was reported to have adjuvant and other immune properties (Naim et al., 1993, 1995a,b). This and another gel from McGhan implants and one Dow Corning gel were later compared for strength of adjuvancy by a similar test procedure, which presumably does not reflect what happens in women with implants. Gels with greater emulsifying ability proved stronger adjuvants. This is an indication of the different characteristics of gels that can have biological implications, although it is not clear what relationship it bears to the in vivo situation in women with implants and thus what, if any, clinical significance this particular effect might have (Naim et al., 1997; see also Chapter 6 of this report).
Cox Uphoff International (CUI) began marketing implants in 1976, introducing smooth single-lumen gel (1976-1991), double-lumen (1977-1991), reverse double-lumen (1982-1993), triple-lumen (1983-1988), gel-saline adjustable (1987-1991), various tissue expanders including a permanent gel/saline model (Gibney, 1989) (1976-present), smooth dimethyl saline (1977-1991) and RTV smooth (1991-1994) and RTV textured (Microcell) (1992-1994) saline models. In 1980 and 1981, the Cavon silicone implant was made by Cox Uphoff (and later, in 1981-1985, by Aesthetech). By 1984, a DRIE (diffusion rate inhibiting envelope) low-bleed shell was introduced. Silicone gel suppliers to Cox Uphoff were Dow Corning (1976-1983), International Silicone (1981-1984), McGhan NuSil (1987-1991), Admiral Materials (1985-1989), Polymer Technologies
(1989-1990), and Applied Silicone (1990). The company was acquired by Inamed Corporation in 1989. Details on specifications of gels and shells are considered proprietary by Inamed.
Bioplasty and Other Manufacturers
After acquiring Roger Klein Mammatech in 1987 (R.A. Ersek, personal communication, March 19, 1998), R.A. Ersek, a plastic surgeon, began the Bioplasty line in 1988. The line originally consisted of MISTI (Molecular Impact Surface Textured Implant) textured single-lumen gel and double-lumen models, which were developed in 1987 and introduced in 1988. MISTI GOLD textured single-lumen polyvinylpyrrolidone (PVP, known as bio-oncotic gel) implants were introduced in late 1990 under FDA 510k provisions, followed by single-lumen PVP and saline MISTI GOLD and MISTI GOLD II textured PVP and saline prefilled and inflatable implants in 1991. Gel suppliers were Dow Corning (1987-1988), Admiral Materials (1987-1988), and Applied Silicone (1988-1992). The PVP used was low molecular weight (average molecular weight = 13,700; Beisang and Geise, 1991). Higher molecular weight PVP was used in more than 500,000 cases during World War II as a plasma expander (Bischoff, 1972). PVP was said to have much greater lubricating properties for the silicone shell, thereby decreasing friction and fold flaws; to diffuse rapidly from tissues; and to be quickly eliminated by the kidney, thus eliminating residuals or granulomas after implant rupture. These implants also had textured shells to reduce contractures and were reported to be radiolucent enough not to block mammographic visualization (Ersek and Salisbury, 1997; Ersek et al., 1993), although they were less radiolucent than triglyceride (peanut oil) filler and tended to degrade when subjected to irradiation (Klein and Kuske, 1993). Bioplasty entered bankruptcy in 1992, and its assets were purchased by NovaMed in Germany, which plans to reintroduce the slightly thicker polyvinylpyrrolidone lines to the U.S. market as Nova Gold (R.A. Ersek, personal communication, April 6, 1998). At present, NovaMed has begun domestic operations as a small company in Minnesota with plans to begin sales with saline implants first. The company has a marketing alliance with McGhan Medical Corporation (Z.F. Twardochleb, McGhan Medical Corporation, personal communication, 1999).
A number of foreign manufacturers distributed implants in the United States. The first saline implants and inflatables by Simaplast (France) were distributed as Roger Klein Mammatech (1968-1978, although some say earlier; R.A. Ersek, personal communication, June 10, 1998). Smooth single-lumen gel, saline inflatable, double-lumen, and some poly-urethane-coated implants were introduced in 1987 by Laboratoires Sebbin
(Unimed, Omega label). Koken (Japan), also known as Porex, presumably distributed gel implants with very fluid, compliant gel in the 1980s. Other companies include Unimed (Germany), PMT (Progress Mankind Technology, GermanyIntegra label), Polytech (Germany, associated with Brazilian SilimedOpticon, Mesmo, Vogue labels), Nagor (Britain), Eurosilicone (France, acquired by Polytech), and Silicone Medicale Paris (France) (Middleton, 1998a). Currently some foreign manufacturers continue to have small minority positions marketing saline implants under 510(k) provisions in the U.S. breast implant marketHutchinson International, Inc., representing Biosil, a British company, Poly Implant Prostheses, S.A. (PIP), a French company and Silimed, a Brazilian company. Little information is available about their products in the medical or other literature, and neither Biosil nor PIP responded to requests for information for this report (E. March, Medical Devices, personal communication, July 22, 1998).
The universe of implants is large, and the variation among them is substantial. In women being studied, it is often difficult to know what implants might be in place and what their characteristics might be. Companies apparently made implants for a number of plastic surgeons and often introduced designs conceived (and patented) by individual surgeons, presumably without further testing. The business was competitive, and companies introduced changes such as softer gels; barrier low-bleed (low-diffusion) shells; greater or lesser shell thickness and durability to reduce rupture and leaking and/or to enhance softness; texturing to reduce contracture; and various sizes, contours, shapes, and multiple lumens in the search for better aesthetics. These changes were introduced at somewhat different times and usually affected some but not all of a company's products. It was not possible to locate much in the way of clinical pretesting of these changes, some of which had unintended consequences. In fairness, it should be noted that testing was not required by the FDA until recently. In general, early implants were thick shelled and contained firm gel. More compliant gels were introduced between 1972 and 1975 by various companies, although the gels were primarily General Electric (departed the medical marketplace in 1976) and Dow Corning. Thinner elastomer rubber shells entered the market from 1972 to 1977, depending on the manufacturer. Various high-performance shells were marketed between 1978 and 1986. It is not clear what chemical changes made these shells high performance, although the performance referred to was greater resistance to tearing. Barrier coating, low silicone gel fluid diffusion features were added on various lines between about 1977 and
1986, and texturing was added in 1987 and 1988 (except for polyurethane coating, which was available from around 1970). The dates indicating market entry of significant new general characteristics could be helpful in considering clinical reports that do not describe implant types and specifications.
It is tempting to assume, as informal comments from some company officials and some comparative analyses imply, that implants of companies using Dow Corning or General Electric silicone gel and elastomer rubbers were made with chemical compounds and characteristics that were quite similar (Thompson et al., 1979). Clearly, gels changed over time and differed somewhat among different manufacturers who entered and left the market at various times. Thus, women were undoubtedly exposed to varying concentrations of the different molecular types and molecular weights or viscosities of silicone, but there were also surely many similarities among products from the same or very few basic manufacturers. This is the assumption that appears to underlie one recent analysis of soft-tissue responses to "approximately fourteen" precursor components of breast implants (Picha and Goldstein, 1991). With respect to elastomers, they were identified only 45% of the time in the series of up to 50 reports reviewed by Foliart (1997).
In general, it can be concluded that there have been at least three "eras" (referred to as generations by others, as noted later in this report) and a number of lesser variations in breast implant manufacture. In the first era, dominated by Dow Corning, there were thick-shell, thick-gel, patched, smooth-surfaced implants with low rupture rates, high contracture, and probably moderate to high gel fluid diffusion rates; in the second era, thin-shell, compliant gel, smooth-surfaced gel and HTV saline implants with high rupture and deflation rates, high contracture, and high gel fluid diffusion rates; and in the third era stronger-shell, barrier-layer, compliant gel, textured gel- and saline-filled and stronger RTV saline implants with as-yet incomplete data, but presumably lower rupture and deflation rates (although not enough time has elapsed to predict this with confidence), lower gel diffusion rates, and lower contracture rates. This latter era may continue to the present or may have given way to a fourth era with changes resulting from the FDA moratorium on gel implants, the predominance of saline implants from the two remaining manufacturers, and the requirements for study protocols. In any event, differences in local and perioperative complications caused by these implants from different eras clearly have implications for the safety of silicone breast implants as described further in Chapter 5. It seems reasonable to conclude that both the physical and the chemical characteristics of implants should be spelled out clearly in product changes, introductions, and investigations because they may influence patient reactions and pa-
tient health. Moreover, as noted in Chapter 5, it would be desirable to accumulate information on the safety, complications, and health effects of a stable group of implants and not make changes until the safety, complications, and health implications of these changes are known.