Silicone Breast Implants and Cancer
Induction or promotion of malignancies of the breast or other organs by silicone implants has been the subject of studies and comprehensive reviews with negative findings. Individual case reports have raised questions regarding the carcinogenicity of silicone breast implants. These questions include whether there has been an increase in primary or recurrent carcinoma of the breast associated with silicone breast implants, whether there has been an increase in breast malignancies other than primary carcinoma of the breast associated with implants, and whether there has been an increase in non-breast malignancies in women with implants, for example, solid tumors (carcinomas) of other organs, sarcoma, lymphoma, or myeloma.
The carcinogenicity of silicone or silicone implants is reviewed in Chapter 4. Although silicone formulated into implants of the proper size, shape, and surface characteristics can induce solid-state carcinogenesis in the susceptible rodent species associated with this phenomenon, it is not a specific response to silicone. Solid-state carcinogenesis occurs in rodents with exposure to a wide array of other substances. There is no convincing evidence that it is a human risk (Brand and Brand, 1980; Morgan and Elcock, 1995). Other well-designed and implemented experimental studies of the carcinogenicity of silicones reviewed earlier were negative.
A small series of case reports of breast malignancy associated with silicone injections is reviewed in Chapter 1. This series consists of relatively few reports, and as noted elsewhere, although case reports can be a basis for formulating hypotheses, they do not constitute evidence for an
association. Harris surveyed 184 surgeons performing breast implantation with a variety of pre-silicone (pre-1962) breast implants. No cases of breast malignancy were reported in 16,660 implants (Harris, 1961). Similar results were obtained by Snyderman and Lizardo (1960); only 4 of 500 plastic surgeons surveyed reported malignancies in women with pre-silicone implants. DeCholnoky's (1970) survey of 265 surgeons covered 10,941 patients; about one-third of these women had ''open pore," pre-silicone breast implants, and no cancers were found. Instances of cancer associated with silicone breast implants have also been the subject of a number of case reports (e.g., Benavent, 1973; Bingham et al., 1988; Bowers and Radlauer, 1969; Cammarata et al., 1984; Dalinka et al., 1969; Frantz and Herbst, 1975; Gottlieb et al., 1984; Hausner et al., 1978; Heywang et al., 1985; Holt and Spear, 1984; Hoopes et al., 1967; Lafreniere and Ketcham, 1987; Mendez-Fernandez et al., 1980; Paletta et al., 1992; Perras and Papillon, 1973; Shousha et al., 1994; Silverstein et al., 1990a-c; Stewart et al., 1992; Travis et al., 1984).
The committee has estimated that 70% of breast implantation is for the purpose of augmentation, that is, not performed after mastectomy for cancer; in the United States in 1997 there were 1.5 million to 1.8 million women with implants, or about 1 million to 1.3 million women with implants for augmentation. Given the incidence of breast cancer in the general population, tens of thousands of cases of breast cancer would be expected to occur over time in a cohort of this size. In fact, breast cancer is reported in association with implants in the epidemiological studies cited below or in examinations of the effectiveness of mammography (see Cahan et al., 1995, and other studies in Chapter 12) as an expected event.
In 1997, Brinton and Brown reviewed many studies relating to the carcinogenicity of silicone breast implants and concluded that these studies found no association of breast implants with breast carcinoma, although they noted that some potential outcomes, such as non-breast malignancies and breast sarcomas, or factors such as life-style, latencies, and others were not adequately addressed. Lamm (1998) also reviewed some of the epidemiological studies and, in a meta-analysis of four cohort studies, reported a standardized incidence ratio for breast cancer that was significantly less than one (0.70; 95% CI, 0.55-0.87), suggesting that breast implants were associated with a decreased risk of this disease.
A number of epidemiological studies, both cohort and case control, of the potential associations between breast (or other) cancers and silicone breast implants provide good evidence that these implants do not result in a higher frequency of breast cancer. Summary data from these studies are listed in Table 9-1. Two small case control studies (Malone et al., 1992) of two age groups of breast cancer patients with odds ratios less than one
No. of Patients (cancer or implant)
SIR, OR (adjusted)
Berkel et al., 1992
0.48 (no CI)
Brinton et al., 1996
0.6 (95% CI, 0.4-1.0)
Bryant and Brasher, 1995
0.76 (95% CI, 0.6-1.0)
Deapen et al., 1997
0.63 (95% CI, 0.4-0.9)
Friis et al., 1997b
1.0 (95% CI, 0.4-2.0)
Glasser et al., 1989
1.0 (95% CI, 0.3-3.3)
Kern et al., 1997
0.67 (95% CI, .02-2.17)
McLaughlin et al., 1998
0.7 (95% CI, 0.4-1.1)
Park et al., 1998
No cancers observed
NOTE: CI = confidence interval; OR = odds ratio; SIR = standardized incidence ratio.
are not included here, because they were reported only by letter with little detail.
Berkel et al. (1992) reported the association of breast augmentation with breast cancer for all women in Alberta, Canada, who had implants between 1973 and 1986. The expected number of cancers was estimated from data from the Alberta Cancer Registry. The implant cohort was compared to a cohort of all women in Alberta who developed breast cancer (N = 13,557). The average follow-up of the implant cohort was 10.2 years, and the average length of time from breast augmentation to the diagnosis of breast cancer was 7.5 years. The standard incidence ratio was 0.476, significantly lower than expected (p < 0.01) (Berkel et al., 1992).
Bryant et al. (1994) reported some problems with the study methods of Berkel et al. that tended to introduce a bias resulting in an underestimate of the standardized incidence ratio. In a subsequent report, this group reanalyzed the original data, and a number of new standardized incidence ratios were reported (0.76, 0.81, 0.85, 0.68), depending on induction periods of 0, 1, 5, and 10 years, respectively; these ratios did not differ significantly from each other, nor was the incidence of breast cancer shown to be significantly higher or lower than in the general population (Bryant and Brasher, 1995).
Brinton et al. (1996) reported a population-based case control study of 2,174 cases of breast cancer and 2,009 age- and geographically matched controls. The odds ratio for augmentation in breast cancer patients, 0.6 (95% confidence interval [CI], 0.4-1.0), after adjustment for a number of factors, including age, race, body size, and family history of breast cancer, among others, suggested a lower association of breast implants with breast cancer.
In a series of studies, Deapen and colleagues described the association with breast cancer in a cohort of breast implant patients in the Los Angeles area over a 14-year period (Deapen et al., 1986, 1997; Deapen and Brody, 1992, 1995). The 1997 study reports 3,182 women collected from private practices in Los Angeles who had breast implants (74% gel filled) between 1953 and 1980 (91% since 1970). Data on breast cancer were obtained from the Los Angeles County Cancer Surveillance Program through 1991. The average follow-up was 14.4 years (range 0.04-20 years), and the median interval from implant to diagnosis of breast cancer was 10.3 years. The standardized incidence ratio (SIR) was 0.63 (95% CI, 0.428-0.895), indicating a significant decrease in breast cancer in the women with breast implants (Deapen et al., 1997).
Using the Danish Hospital Discharge Registry, 1,135 women with cosmetic breast implantation were identified with an average age of 31 years and an average follow-up of 8.4 years (Friis et al., 1997b). There was no increase in the standardized incidence ratio for all cancers (N = 27) or for breast cancer (N = 8): SIR = 1.1 (95% CI, 0.7-1.6) and 1.0 (95% CI, 0.4-2.0) respectively. No cases of multiple myeloma were observed in the implant patient cohort (0.1 case expected). These data update the preliminary report of McLaughlin et al. (1994).
Glasser et al. (1989) reported 4,742 breast cancer patients 20-54 years of age diagnosed between 1980 and 1982 and 4,754 controls who were part of a large U.S. case control study of cancer and steroid hormone use. The mean interval from implantation to diagnosis or interview was six to seven years. The adjusted odds ratio for breast augmentation was 1.0 (95% CI, 0.3-3.3) (Glasser et al., 1989).
Kern et al. (1997) studied 680 cases of breast implantation in women with no prior history of cancer using the Uniform Hospital Discharge Data Set from 34 Connecticut hospitals during 1980-1993. The Connecticut Cancer Registry was used to verify cancer cases. Women with implants were compared to 1,022 control women undergoing tubal ligations, a less-than-ideal control group in several respects. Mean follow-up of the women with implants was 4.6 years and of the control group 5.4 years. The implant group had relative risks for breast cancer and non-breast cancers of 0.67 (95% CI, 0.2-2.17) and 0.21 (95% CI, 0.07-0.60), respectively. No cases of multiple myeloma or sarcoma were observed (Kern et al., 1997).
McLaughlin et al. (1994, 1995a, 1998) briefly reported on cohort studies from Denmark and Sweden. The Danish study, first reported by letter (McLaughlin et al., 1994), found standardized incidence ratios less than one for breast cancer and for all cancers; the results were reported in more detail by Friis et al. (1997b), as reviewed earlier. The Swedish study was mentioned first in a letter (McLaughlin et al., 1995a) and then expanded
and reported in a brief communication (McLaughlin et al., 1998). This communication described a study that consisted of 3,473 women undergoing breast augmentation mostly after 1976 with an average age at implantation of 30 years and an average follow-up of 10.3 years. The standardized incidence ratios (based on Swedish national cancer rates) for all cancers (N = 74) and for breast cancer (N = 18) were 1.1 (95% CI, 0.8-1.3) and 0.7 (95% CI, 0.4-1.1), respectively. A small increase in lung cancer (SIR = 2.7; 95% CI, 1.1-5.6), no significant excess of lymphoproliferative malignancy, and one case of multiple myeloma (expected cases not reported, but presumably somewhat less than one) were observed (McLaughlin et al., 1998).
Park et al. (1998) analyzed data from two groups of women from southeast Scotland who had implant surgery for augmentation or reconstruction at two Scottish hospitals. These groups are analyzed for anti-nuclear antibodies and connective tissue diseases. The group receiving silicone gel implants for augmentation consisted of 186 women. Although only 110 were seen on follow-up for the study, data on breast cancer were available for all 186, and no breast cancers occurred.
These epidemiological studies of breast cancer and silicone breast implants are strikingly consistent in showing no association. Some of the studies have very small numbers (and therefore low power), some have control groups that may not be exactly comparable (e.g., tubal ligations), and others have follow-up intervals after implantation that may be short in relation to reasonably expected latency periods between exposure and the onset of malignancy. The committee concludes, however, that there are sufficient studies with consistent and convincing findings of no association between breast cancer and implants.
Several of the epidemiological studies reviewed here also collected data on all cancers, or non-breast cancers, and found no associations with breast implants (e.g., Friis et al., 1997b; Kern et al., 1997; McLaughlin et al., 1994 and 1995a). The Deapen and Brody (1992, 1995) reports also contained information on all cancers; in 3,112 implant patients, 45 cancers (versus 50 expected) were found, and a standardized incidence ratio of 0.90 (95% CI, 0.66-1.20) was calculated. Vulvar and lung cancers were increased. No cases of multiple myeloma were observed versus 0.6 expected (Deapen and Brody, 1992, 1995). Although these data are not conclusive, they are generally negative; occasional increases in a particular cancer are not consistent and are likely due to chance or confounding factors. The committee concludes, therefore, that there is limited evidence that silicone breast implants are not associated with non-breast cancers.
A number of investigators have studied recurrence of cancer or death due to breast cancer in patients with silicone implants for breast reconstruction after mastectomy. Birdsell et al. (1993) found that survival in
implanted and nonimplanted patients was similar. Breast cancer recurrence in 306 reconstruction patients (207 with submuscular implants) followed for a mean of 6.4 years was similar to that reported in the scientific literature (Noone et al., 1994a). Johnson et al. (1989) reported that their experience with recurrence of or death from breast cancer in 118 mastectomy patients after reconstruction with implants was similar to their experience with mastectomy patients without implants. Similar results with implant patients and recurrences were also reported briefly and then in a large series comparing relapse-free intervals in women with and without reconstruction with implants after mastectomy for cancer by Georgiade et al. (1982, 1985). Petit et al. (1994) compared 146 women with silicone gel implants for reconstruction to 146 matched cancer mastectomy patients without implants who had 9- and 12-year follow-ups, respectively, and reported a relative risk for local recurrence of 0.5 (95% CI, 0.3-1.1) and of death from breast cancer of 0.5 (95% CI, 0.3-1.0). Park et al. (1998) studied 289 postmastectomy implant reconstruction patients (176 matched to mastectomy controls without implants) and reported a relative risk of recurrence and death (all cases) of 0.83 (95% CI, 0.48-1.45) and 0.51 (95% CI, 0.23-1.11), respectively. These data present a consistent picture that implants do not increase breast cancer recurrence rates or decrease survival rates in patients after reconstruction with implants.
Breast Sarcomas and Other Tumors
Because silicone solid-state carcinogenesis results in sarcomas in susceptible rodents, the prevalence of breast sarcomas in women has been explored. It is difficult to assess whether there has been an increase in tumors of the breast other than primary carcinomas because malignancies other than primary carcinomas are rare, have not had uniform classification over the years, and probably have not been reported consistently in tumor registries (Callery et al., 1985). Tumors of stromal or fibrous origin arising in the capsule of a breast implant in humans would, presumably, be analogous to those caused in solid-state carcinogenesis in rodents. Such tumors with fibrous, myeloid, and fatty tissue patterns do occur in the breast but are rare, constituting 0.5-1% of primary breast malignancies (Tang et al., 1979). These malignant tumors with the potential to metastasize have been grouped under the term "stromal sarcomas." At present, there is no evidence that breast sarcomas have increased in frequency or are occurring unusually in women with silicone breast implants. Only two reports were found in a review by Lorentzen (1988) for the Public Health Service: Kobayashi et al. (1988); (a stromal sarcoma) and Morgenstern et al. (1985); (a pseudosarcomatous carcinomafelt by a departmental scientist to be an undifferentiated sarcoma), both after sili-
cone injections. Sarcomas were not found in the epidemiological studies reviewed earlier of all, or non-breast, malignancies occurring in women with implants. Deapen et al. (1997) found no sarcomas in their own study and reported a review (personal communication of M. F. Brennan) of sarcomas occurring at Memorial Sloan-Kettering Cancer Center of which 0.5% were in the breast but none in women with implants (Deapen et al., 1997). The National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) data have been reviewed for changes in the incidence of breast sarcoma from 1973 to 1986 (extended to 1990 by Engel et al., 1995) that might have occurred consequent to the increasing prevalence of breast implantation during that time. May and Stroup (1991) found no increase, and Engel and Lamm (1992) also found no increase on reanalysis of the data to allow for a ten-year latency between exposure to silicone and appearance of sarcoma. Breast sarcoma was and remained extremely rare, 0.12-0.13 case per 100,000 woman-years (Engel and Lamm, 1992; Engel et al., 1995; Lamm and Engel, 1989; May and Stroup, 1991). This evidence consistently fails to support an increase in breast sarcoma associated with silicone breast implants, although analysis of the national data would not be expected to detect small increases in breast sarcoma because of the rarity of that condition.
Another rare fibrous tumor that can infiltrate extensively into surrounding structures but does not metastasize is classified by the term "desmoid." Rosen and Ernsberger (1989) reported their experience with 22 cases of breast desmoids. One of these cases was associated with saline implants that had been in place for several years. Four other cases of desmoids in association with breast implants have been reported (Dale and Wardlaw, 1995; Jewett and Mead, 1979; Schiller et al., 1995; Schuh and Radford, 1994). Dale and Wardlow (1995) reviewed the literature and found less than 75 other cases of desmoids reported in women without breast implants. The relation of these tumors to previous trauma, scars from previous surgery, fibroadenomas and fibrocystic disease was noted. Desmoids occur rarely; there is no evidence of an increase in frequency, and they are apparently associated with fibrosis, which occurs in the breast in conditions other than implantation. It is possible that a desmoid could occur very rarely in association with the fibrotic response that forms the capsule around a breast implant, but there is no evidence to conclude that this occurs with increased frequency in the presence of silicone breast implants.
Several case reports and review articles indicate that the incidence Of primary breast lymphoma is 0.05-0.53% of primary breast malignancies (Petrek, 1987a,b) In recent years there have been a few case reports of lymphoma developing in breasts in relation to silicone implants (Benjamin et al., 1982; Cook et al., 1995; Duvic et al., 1995; Krech, 1997). These
lymphomas were of several varieties including follicular cutaneous T-cell (mycosis fungoides and Sézary syndrome) and neoplastic T-cell lymphoma. Proximity to the implants and foreign-body reaction with giant cells were noted. Significant increases in lymphomas were not found in several epidemiological studies (Friis et al., 1997b; Kern et al., 1997; McLaughlin et al., 1998). The committee also noted two reports of squamous cell carcinoma arising in the capsule of a breast implant (Kitchen et al., 1994; Paletta et al., 1992).These case reports are few in number and do not constitute evidence for an association of these malignancies with breast implants.
The question of a relationship between exposure to silicone and subsequent development of monoclonal gammopathy of undetermined significance (MGUS) or multiple myeloma in humans was raised by Salmon and Kyle (1994) in commenting on the induction of plasmacytomas after intraperitoneal injection of silicone gel in genetically susceptible strains of mice (Potter et al., 1994). This question is also discussed in Chapter 6 of this report. Salmon and Kyle emphasized, as does this report, that experimental plasmacytomas can be induced only under special conditions in genetically susceptible mice, that they differ from multiple myeloma, and that agents other than silicone can induce this response in susceptible mice.
Garland et al. (1996) reported five Florida women diagnosed with immunoglobulin G myeloma from 1990 to 1993 after 2-12 years' exposure to silicone gel implants. Three of these women were 45 years of age or less. Garland concluded that these three cases of multiple myeloma in women with implants were several times the number expected in the State of Florida (Garland et al., 1996). Tricot et al. (1996) reviewed 114 women with multiple myeloma seen from 1992 to 1995; 9 (7.9%) of these women had silicone breast implants. Silverman et al. (1996a) reported three women with silicone breast implants and multiple myeloma; 2 were from a small series of 34 multiple myeloma clinic patients. Although these reports raise the question of an association, as uncontrolled case reports they cannot support a conclusion.
Garland et al. (1996) also studied immunoglobulin levels in a small sample of referred sera from women with silicone breast implants and found 30% of these women to have elevated levels. Silverman et al. (1996a) reviewed immunoglobulin levels in 630 symptomatic women with silicone breast implants of 14 years' mean duration and found elevated immunoglobulins in 23%. In neither of these studies was the frequency of other conditions that might be associated with increased blood levels of immunoglobulins fully assessed in the patients with elevated values. Five women with MGUS were found; two of these women (of four contacted) reverted to normal immunoglobulin levels after removal of their silicone
gel implants. Other studies that tested immunoglobulin levels in women with implantssome finding an increase, and others nothave been discussed in Chapter 6. Also, a large study, which was part of a major epidemiological cohort study of connective tissue disease and rheumatic symptoms in women with breast implants, did not find elevated immunoglobulin levels in women with implants compared to healthy controls without implants (Karlson et al., 1999).
Epidemiological studies reported by Deapen and Brody (1992, 1995), Friis et al. (1997b), Kern et al. (1997), and McLaughlin et al. (1998), discussed earlier, have not observed significant (or any) numbers of myeloma cases in women with breast implants. The committee concludes that evidence for an association between silicone breast implants and multiple myeloma or MGUS is insufficient.
There is a consistent, substantial, long-term base of scientific evidence bearing on the experimental carcinogenicity and clinical breast or other cancer experience with silicone and silicone breast implants. Based on its review of this evidence, the committee concludes that the available evidence does not support an association of silicone or silicone breast implants with experimental carcinogenesis (other than rodent solid-state carcinogenesis), primary or recurrent breast cancer, breast sarcoma or other solid tumors, lymphoma, or myeloma. If anything, evidence (though limited) suggests a lower risk of breast cancer in women with silicone breast implants.