NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Institute of Medicine (US) Committee on the Safety of Silicone Breast Implants; Bondurant S, Ernster V, Herdman R, editors. Safety of Silicone Breast Implants. Washington (DC): National Academies Press (US); 1999.

Cover of Safety of Silicone Breast Implants

Safety of Silicone Breast Implants.

Show details

8Epidemiological Studies of Connective Tissue or Rheumatic Diseases and Breast Implants

Extensive recent reviews have focused on epidemiological reports studying the possibility of an association between silicone breast implants (primarily gel filled) and defined connective tissue disease (CTD) (Independent Review Group, 1998; Tugwell, 1998). A number of meta-analyses (or smaller reviews) of these reports have also been published (Hochberg and Perlmutter, 1996; Hulka., 1998; Lamm, 1998; Lewin and Miller, 1997; Perkins et al., 1995; Stein, 1999; Wong, 1996). These reviews and meta-analyses have considered a relatively standard body of literature, although there are some differences among them as to which individual reports were included or not included. The reviews also have considered connective tissue diseases, combined and individually, including particularly systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), Sjögren's syndrome (SS), systemic sclerosis or scleroderma (SSc), dermatomyositis/polymyositis (D/P) and occasionally others. Taken together, they do not support an association between connective tissue disease, combined or individually, or stated another way, an elevated relative risk for these diseases, in women with silicone breast implants. Although there are some informative abstracts and letter reports (e.g., Dugowson et al., 1992; Lacey, 1998; McLaughlin et al., 1994, 1995a; Wolfe, 1995a), with a few exceptions the committee limited its review of the epidemiology of connective tissue disease to full reports in the peer-reviewed, scientific literature, and focused on connective tissue diseases, combined and individual (primarily SSc, SLE, RA, D/P, and SS), in women with silicone breast implants of any kind. Although the committee was aware of numerous case reports of connective tissue disease in women with gel-, and a few with saline-filled, breast implants, a listing of such anecdotal evidence was not considered profitable. These case reports can be found in the references of this report. Many of them have been reviewed by Sánchez-Guerrero et al. (1994), and some deficiencies of these reports have been listed by Kurland and Homburger (1996), including: poor or absent case definition; failure to apply conventional diagnostic criteria; lack of identification of type of implant; silicone injections of questionable composition; unknown disease status prior to implant; lack of a consistent latent period; and inconsistent effects on symptoms after explantation (Kurland and Homburger, 1996).

The committee found 11 published cohort studies (Edworthy et al., 1998; Friis et al., 1997a; Gabriel et al., 1994; Giltay et al., 1994; Hennekens et al., 1996; Nyrén et al., 1998a; Park et al., 1998; Sánchez-Guerrero et al., 1995a,b; Schusterman et al., 1993; Weisman et al., 1988; and K.E. Wells et al., 1994), five case control studies (Burns et al., 1996; Englert et al., 1996; Hochberg et al., 1996; Strom et al., 1994; and Williams et al., 1997), and one cross-sectional study (Goldman et al., 1995), a total of 17 epidemiological studies of rheumatic diseases. The cohort studies were carried out in different locations in the United States and abroad: Alberta, Canada (Edworthy et al., 1998), Denmark (Friis et al., 1997a), Olmsted County, Minnesota (Gabriel et al., 1994), Amsterdam, Holland (Giltay et al., 1994), national samples in the United States (Hennekens et al., 1996 and Sánchez-Guerrero et al., 1995a,b), Sweden (Nyrén et al., 1998a), southeast Scotland (Park et al., 1998), Texas (Schusterman et al., 1993), San Diego (Weisman et al., 1988), and Florida (Wells et al., 1994). The five case-control studies were performed in Michigan (Burns et al., 1996), Sidney, Australia (Englert et al., 1996), San Diego, Baltimore-Washington, D.C., Pittsburgh (Hochberg et al., 1996), Philadelphia (Strom et al., 1994), and across the United States (Williams et al., 1997). The cross-sectional study was done in Atlanta, Georgia (Goldman et al., 1995). The principal results of these studies, that is, the risk of CTD associated with silicone breast implants, are summarized in Table 8-1.

TABLE 8-1. Studies of the Risk of Connective Tissue Disease (CTD) Associated with Silicone Breast Implants.


Studies of the Risk of Connective Tissue Disease (CTD) Associated with Silicone Breast Implants.

Cross-Sectional Study

Goldman et al. (1995) conducted a cross-sectional study using the computerized clinic records of 4,229 women, primarily with RA, seen in a rheumatology practice between 1982 and 1992. He found 721 women with connective tissue disease (12 with implants) and 3,508 control women (138 with implants) (85% of the implants were gel filled, 4% saline filled and 12% unknown). The adjusted odds ratio for all RA and connective tissue disease was 0.52 (95% confidence interval [CI], 0.29-0.92), even though six women who had implants after the diagnosis of RA or other CTD were included (Goldman et al., 1995). Rheumatoid arthritis and connective tissue disease overall were statistically significantly less frequently diagnosed in patients with silicone breast implants than in those without implants. It is possible that detection of breast implants was incomplete, since it depended on chart data in a practice where such information would have been an incidental finding. The prevalence of breast implants was higher than usually reported for unexplained reasons (150 out of 4,229 women, 3.5%), however.

Case Control Studies

Burns et al. (1996) attempted to identify all cases of SSc diagnosed in individuals from 1985 to 1991 and living in Michigan for at least 18 years. Data were collected between August 1991 and May 1993. Women with SSc were identified from hospital records, rheumatologists, and the United Scleroderma Foundation; 274 women participated in the study.

Controls were 1,184 women identified by random-digit dialing who were matched to the cases by age, race, and geographic area of the state. Medical records of cases were reviewed by rheumatologists; controls reported health information by telephone interview. The accuracy of self-reporting of breast implants was confirmed by telephoning a separate population and was found to be 94%. Subjects were not informed that the object of the study was to investigate the association of implants with scleroderma. Breast implants had been placed after mastectomy for cancer in 50% of SSc patients and 43% of controls. A separate analysis for cancer versus non-cancer patients was not done. Of SSc patients, 0.7% (two) had implants compared with 1.2% of controls. One SSc patient had had her implants for 1 year, and the other for 12 years before diagnosis of SSc, and the median duration of implants (to the time of interview) in controls was 8.8 years. The implant was silicone gel filled in both of the scleroderma cases and in 12 of 14 controls. The odds ratio, adjusted for age, race, and date of birth, was 0.95 (95% CI, 0.21-4.36) (Burns et al., 1996).

A retrospective, case control study of all cases of SSc or limited SSc in Sydney, Australia diagnosed prior to 1989 was the subject of a report by Englert and Brooks (1994) and a subsequent validated update by Englert et al. (1996). Cases were identified by death certificates, hospital records, and physicians' records. Controls were patients who had visited 29 randomly selected general practitioners in Sydney since 1990. Women were interviewed, and medical records were reviewed. In this way, 287 cases and 252 controls were identified. Three cases and three controls had silicone gel breast implants before diagnosis or prior to the selected date for controls. The self-reported implant status of patients was in 100% agreement with medical records, and the adjusted odds ratio was 1.0 (95% CI, 0.16-6.16) (Englert et al., 1996).

Hochberg et al. (1996) identified 837 cases of SSc from three university arthritis clinics, and 2,507 local controls, matched for age and race who were found by random-digit dialing. SSc patients filled out a mailed questionnaire; controls were interviewed by telephone. The response rate was 73% for patients and 59% for controls. Eleven (1.31%) of the cases reported silicone gel breast implants, a median of 11 years before diagnosis of SSc. In comparison, 31 (1.24%) of controls reported gel implants. A validation study assessed the accuracy of breast implant self-report in controls on a 5% random sample, and agreement was 96.7%. (Accuracy of breast implant self report is not always so high-89.3%, Garbers et al., 1998). An odds ratio of 1.07 (95% CI, 0.53-2.13) was calculated using multiple logistic regression models adjusted for age, race, and site. This study had a power of 80% to detect an odds ratio of 1.8 or more (Hochberg et al., 1996). Although there were differences in ascertainment between controls (telephone interview) and patients (questionnaire), and some SSc patients with silicone breast implants may have died, this study provides good evidence against an association of breast implants and scleroderma.

Strom et al. (1994) reported cases of SLE from the offices of 22 rheumatologists and the Lupus Foundation in Philadelphia seen between 1985 and 1987, who had been part of an earlier unrelated study. In 1992, the authors were able to find 76% of the patients and 77% of the controls from the prior study for telephone interviews. Of 133 lupus patients, one (0.75%) was found to have breast implants compared to no implants in the 100 controls (friends). Because no implants were found in these controls, the authors calculated an odds ratio using controls from a cancer and steroid hormone study conducted elsewhere in 1980-1982 (presumably the study used for the report of Glasser et al., 1989, see Chapter 9). Although, the odds ratio was 4.5 (90% CI, 0.2-27.3), this finding was not statistically significant (Strom et al., 1994). Shortcomings of this study include the use of an unmatched control group from an earlier period with a prevalence of breast implants of 0.17% (a lower odds ratio would usually have been expected in a study group with less than 1% implants) and the very low power to detect associations between breast implants and SLE.

Williams et al. (1997) reported experience with 323 women, 156 with defined connective tissue diseases (RA, SSc, and SLE) and 167 with undifferentiated connective tissue disease, first in an abstract and subsequently in a full report. (The abstract is cited because it contains information on the distribution of breast implants and an odds ratio not found in the full report.) In the total group, three women had breast implants, one after the onset of CTD symptoms. Of the 156 women with defined connective tissue disease, one (of 40 women with SSc) had breast implants. There was no concurrent control group. Instead a historical frequency of breast implants from the survey of Cook et al. (1993) was used, which gives a national prevalence of 0.81%. This prevalence is probably an underestimate (see Chapter 1), and calculations of odds ratios based on this ''control" are probably somewhat overstated. The odds ratio, which was not significantly elevated, would be even lower if it were adjusted using a higher control prevalence. The odds ratio for both defined and CTD groups combined, counting only implants that preceded symptoms, was 0.74, (80% (sic) CI, 0.2-2.02) and was 1.15 (95% CI, 0.23-3.41), counting all (three) implants (Williams and Weisman, 1994; Williams et al., 1997). The absence of a concurrent control group is a weakness of this study, as is its modest power.

Cohort Studies

Schusterman et al. (1993) carried out a cohort study of postmastectomy breast cancer patients between 1986 and 1992 at the M.D. Anderson Cancer Center in Houston, Texas. Records of 250 women who had breast implants and 353 who had autogenous tissue reconstructions were reviewed. Patients were also sent questionnaires regarding medically identified and treated "rheumatic diseases." The follow-up period was only 1.9 years for women with implants and 2.5 years for the control autogenous reconstructions. One patient in each group developed a connective tissue disease. The relative risk was 1.08 (95% CI, 0.1-17.2) (Schusterman et al., 1993). This study was well controlled, was from a single center, and investigates an interesting subset of implant patients. The small number in each cohort (and low power) and the very short follow-up limit possible conclusions.

Gabriel et al. (1994) reported a retrospective, population-based cohort study carried out in Olmsted County, Minnesota (the Mayo Clinic area) of all women who had breast implants between 1964 and 1991. There were two age-matched controls for each case, and these control women had a medical examination within two years of their matched case's implantation date. Connective tissue disease was self-reported with a follow-up review of medical records. The average postoperative follow-up time was eight years. There were 749 women with implants and 1,498 controls; 5 of the 749 women with breast implants and 10 of the 1,498 controls had a connective tissue disease. The adjusted relative risk was 1.10 (95% CI, 0.37-3.23) (Gabriel et al., 1994). In this study, clinical records were reviewed, limiting information on women with breast implants to this source, but Mayo Clinic records are reputed to be excellent, and experienced individuals reviewed the charts.

Giltay et al. (1994) studied women who had silicone gel breast implants between 1978 and 1990 in a clinical practice in Amsterdam. Surveys were mailed to 287 living and traceable women with implants and to the same number of age-matched controls with other breast surgery. The response rate was 82% (N = 235) in women with implants and 73% (N = 210) in controls. The average length of follow-up was 6.5 years. Evaluation was by self-report and, for those reporting CTD symptoms, a physician contact. There was no disease-specific or general relative risk reported, but 37% of women with implants and 21% of controls reported symptoms beginning after surgery, primarily painful joints and burning eyes. In a personal communication, Perkins et al. (1995) reported two CTD cases in the women with implants and four cases in the controls, with a relative risk of 0.44 (no CI reported). Not all patients were located, and some did not answer the questionnaire. A rheumatologist reviewing survey responses assessed the likelihood of rheumatic disease as 6 and 8% in controls and cases, respectively (Giltay et al., 1994). This study was done during the height of publicity about potential problems with breast implants. About a quarter of the implant patients had reconstruction after mastectomy for cancer, but cancer patients were not included in the group of age-matched controls.

Wells et al. (1994) carried out a retrospective cohort study of patients with silicone breast implants (222) or other cosmetic surgery (80) from one plastic surgeon's office in Tampa, Florida, between 1970 and 1990. In 1990 and 1991, 826 women from the practice were mailed a questionnaire inquiring about 23 signs and symptoms that may have occurred before or after surgery. This was followed by a telephone interview. The initial response rate was only 42%, and the exclusion of ineligibles reduced the final cohort further to 302 women with implants and cosmetic surgery controls without implants. The average follow-up of implant patients was four years. Medically diagnosed connective tissue disease included Raynaud's phenomenon in one woman and seven cases of arthritis. Only 3 of 27 possible outcomes differed between the two groups. Swollen axillary glands and tender axillary glands were more common in women with implants, odds ratios 7.1 (95% CI, 1.13-44.4) and 6.9 (95% CI, 1.75-27.15), respectively, and change in skin color was more common in women with other surgery, odds ratio, 0.13 (95% CI, 0.05-0.32). The odds ratio for arthritis was 1.16 (95% CI, 0.15-9.04). No patients reported scleroderma or lupus (Wells et al., 1994). The follow-up in this study was of somewhat limited duration, and results for patients who received implants for reconstruction were not separated from those receiving implants for augmentation in the analysis.

Sánchez-Guerrero et al. (1995) carried out a study using the Nurses' Health Study Cohort to investigate the association between connective tissue disease and breast implants. The 87,501 eligible women in this cohort had responded previously to biennial mailings that included questions about physician-diagnosed defined connective tissue diseases, rheumatic conditions, and "connective tissue disease not further specified." Women identified in this way completed a subsequent survey to identify those with three or more signs or symptoms of connective tissue disease or two or more swollen joints of at least six weeks' duration (cases). Criteria for cases were met by 1,294 women. Medical records were then reviewed by two rheumatologists. A screening with less stringent criteria yielded an additional 904 cases. A total of 516 women with defined connective tissue disease were identified from record review. In all, 1,183 women were found to have implants, 75 of them in the various CTD groups. A small validation study found 99 of 100 self-reports of breast implant status to be accurate (Karlson et al., 1999). The age adjusted relative risk was not greater than one for any of the above case definitions. For defined connective tissue disease, the relative risk was 0.3 (95% CI, 0-1.9) in women with silicone gel implants and 0.6 (95% CI, 0.2-2.01) in women with any breast implant (14% of implants were saline). The data in this study came from women who reported these symptoms and connective tissue disease prior to 1990 when adverse publicity about implants began to appear. Mean follow-up between surgery and disease diagnosis was nearly ten years (Sánchez-Guerrero et al., 1995). This study involved large numbers of women, who were selected only in the sense that they were health professionals who agreed to a long-term health study, and it had significant power to rule out all but the smallest risk of connective tissue disease. Furthermore, information on other than defined connective tissue disease was collected and analyzed. These considerable strengths make this an important study that finds no relationship between connective tissue disease or rheumatic conditions and breast implants.

Questionnaires on health-related items, including breast implants and diagnoses of a number of connective tissue diseases were mailed to 1.75 million women health professionals in the United States and Puerto Rico between September 1992 and May 1995. A total of 426,774 (24%) women responded, and 395,543 women returned questionnaires usable for the purpose of this study. Of these, 2.74% reported having breast implants between 1962 and 1991. Risk estimates showed a small and not statistically significant elevation in risk for each defined connective tissue disease (RA, SLE, SS, D/P, and SSc). For "other connective tissue disease including mixed" and for "any connective tissue disease," that is, any of the defined diseases, the risk estimates were significantly elevated, 1.30 (95% CI, 1.09-1.62) and 1.24 (95% CI, 1.08-1.41), respectively (Hennekens et al., 1996). This study was conducted during the height of publicity on silicone breast implants, and the overall response rate of 24% was much lower than the 70% response rate for the Nurses' Health Study reported by Sánchez-Guerrero et al. (1995) or the 100% inclusion of the Mayo Clinic study of Gabriel et al. (1994) suggesting possible selection bias. The prevalence of implants in this study was twofold higher than that in the population at large (see Chapter 1), which suggests that women with breast implants Were more likely to respond to the questionnaire than women who did not have breast implants. The evidence for disease in these women rests on their own unverified self-reports, which may not be reliable. On the other hand, the large numbers of women in this study give it power to detect small changes in risk. The upper bound (1.41) of the 95% CI for "any connective tissue disease'' suggests that if there is an increased risk, it is modest.

Friis et al. (1997a) reported a retrospective cohort study of all women identified in the Danish Central Hospital Discharge Register who had been hospitalized for placement of breast implants between 1977 and 1992. The register identified 1,135 women with cosmetic implants and 1,435 with implants for reconstruction. Three control groups were identified: 7,071 women undergoing breast reduction surgery, 472 women having breast ptosis corrected, and 3,952 women with breast cancer without implant. An analysis for the presence of five defined connective tissue diseases and "other defined rheumatic conditions" was carried out. The record of each patient CTD was obtained for verification by experienced rheumatologists and for implant status. Expected numbers of each connective tissue disease were calculated by multiplying the number of per-son-years of follow-up in the groups by the sex-specific national hospital discharge rates for each five-year age group and the calendar period of observation. For connective tissue diseases combined, the observed-expected ratio for cosmetic breast implant patients was 1.1 (95% CI, 0.2-3.4) and for reconstruction patients, 1.3 (95% CI, 0.5-3.6). For nonclassical "muscular rheumatism," there was an excess of cases for each of the four groups—cosmetic and reconstructive implants, breast reduction, and cancer (Friis et al., 1997a). This carefully designed study involved relatively large numbers of women and considered both defined and nonclassical CTD. However, only hospitalized cases of CTD were included.

Edworthy et al. (1998) attempted to contact 16,600 women identified through the Alberta Health Registry as having a breast implant or other cosmetic surgery between 1978 and 1986. Only 20% of the women could be contacted, were willing to participate, and fulfilled the eligibility criteria. The response rate was 59% (5,822) of the approximately 9,900 contacted, of whom 46% agreed to participate. No breast cancer reconstruction patients were included. After additional exclusion of some ineligibles, the final study group consisted of 1,576 women with breast implants (1,112 gel filled) and a control group of 727 women with other cosmetic procedures, with 12 and 11 years of follow-up, respectively. All women completed a questionnaire and had a blood sample taken. Women with relevant symptoms, diagnoses, or medications were invited for an examination by a nurse clinician and a rheumatologist (who were blinded to breast implant status). Symptoms, such as cognitive problems, numbness in the extremities, muscle pain, headache, and hand pain, were significantly more common in breast implant patients. There was no difference in the incidence of specific (RA, SLE, SS, SSc) or combined connective tissue disease or signs of atypical autoimmune disease in the two groups. The relative risk for all CTD was 1.0 (95% CI, 0.5-2.2) (Edworthy et al., 1998). Only 40% of women responding to the initial contact participated in this study, but the large number of women with good follow-up gives it power to detect fairly small differences in risk. The significant differences between the control and experimental groups were limited to subjective symptoms.

Nyrén et al. (1998a) reported a retrospective cohort study using the Swedish National Inpatient Registry linked to three Swedish population registries to assemble cohorts of 7,442 women with breast implants (68% gel or double lumen) and 3,354 matched control women with breast reduction surgery discharged from hospitals between 1972 and 1993. Women in these cohorts with a subsequent hospitalization for a defined or possible connective tissue disease were identified. Diagnoses were validated by medical record review, and average postoperative follow-up was 8.0 and 9.9 years for the implant and control groups, respectively. Defined connective tissue disease (RA, SLE, SSc, SS, and D/P) was found in 29 of the 7,442 women with implants and 14 of the 3,354 controls. Standardized observed or expected hospitalization ratios were 1.6 (95% CI, 0.95-2.5) for connective tissue diseases in women with cosmetic implants, 0.8 (95% CI, 0.4-1.4) for reconstruction patients and 1.1 for all women with implants (95% CI, 0.8-1.6) compared to 1.3 (95% CI, 0.7-2.2) for connective tissue disease in women with breast reductions. No risk elevations in any of the groups were noted for any of the five defined connective tissue diseases individually or for various other unspecified CTD or rheumatic conditions such as fibromyalgia, polymyositis and others (Nyrén et al., 1998a). This study is similar to the Danish study in possibly missing cases by including only hospitalized women, but it also has relatively large numbers, considers other rheumatic conditions as well as defined connective tissue disease, and is carefully designed.

Park et al. (1998) carried out a retrospective cohort study of all patients in two comparison groups in southeast Scotland admitted to two Scottish hospitals. Women who had silicone gel breast implants for augmentation between 1982 and 1991 were compared with a control group of outpatients in a plastic surgery department. A second group, cancer patients reconstructed with silicone gel implants during the same decade, was matched with a control group of breast cancer patients without implants. All women underwent physical and clinical laboratory examinations. The study included 317 women with, implants and 216 control women with a combined average follow-up of 68 months. There was one case of RA among the women with implants and one case in the controls, yielding an odds ratio of 0.42 (95% CI, 0.01-15.63). The women with implants and the controls were also similar in prevalence of symptoms such as joint pain (odds ratio; 1.07; 95% CI, 0.42-2.74), muscle pain (odds ratio; 1.17, 95% CI, 0.36-3.88), and fatigue (odds ratio; 2.01, 95% CI, 0.74-5.56). This study had good controls for both augmentation and reconstruction patients. However, it had relatively small numbers of women and according to the authors, the lowest relative risk detectable for connective tissue disease was 3.2 for reconstruction patients and 16 for augmentation patients.

Medical records for all patients receiving silicone gel implants for breast augmentation between 1970 and 1981 in a private practice were reviewed by Weisman et al. (1988). A survey was sent to the 378 patients identified, and 125 of 199 (63%) with correct addresses responded. Of 38 women who reported health problems since surgery and were interviewed, 22 were examined. Of this group, 12 women had bursitis; 7, knee pain; 14 osteoarthritis; 3 fibromyalgia; and 2 back pain. No cases of inflammatory rheumatic or connective tissue disease were discovered (Weisman et al., 1988). This was an uncontrolled study of a small group of women.


As noted earlier, the committee examined a number of comprehensive reviews and meta-analyses of epidemiological reports investigating associations of combined and individual connective tissue diseases with silicone breast implantation. The committee also reviewed published reports from 17 individual epidemiologic studies, which are discussed here, and took note of additional abstracts and letters. These reports and analyses generally examined connective tissue diseases combined. Of the 17 independent reports, from 6 to 12—depending on the disease in question—looked specifically at one or more of the individual CTDs listed earlier (i.e., SSc, SS, RA, SLE and D/P). A number of these reports also provided data on "other" connective tissue diseases. In only one instance was a relative risk or odds ratio significantly elevated, however. That report, based on a large number of women, found a small association of implants with combined connective tissue diseases (Hennekens et al., 1996). However, among women who responded to the study questionnaire, the proportion reporting breast implants was more than twice the estimated national frequency of these implants, which suggests selection bias. Moreover, the evidence for disease in these women consists of unverified self-reports. Thus, this study probably overstated the risk of connective tissue disease associated with silicone breast implants. If its results are valid, they rule out large increases of risk. Excluding this report, a very substantial body of evidence, consisting of a number of independent investigations and other analyses, does not provide evidence for an association of silicone gel-or saline-filled breast implants with defined connective tissue disease. Although others (e.g., Hulka, 1998) including authors of reports themselves, have pointed out problems with individual epidemiological studies, the consistency of results among many reports is impressive. As was the case with antinuclear antibodies (ANAs), data and results were rarely segregated by gel-or saline-filled implants, so no conclusions regarding associations with connective tissue disease by breast implant type are possible [see, however, Sánchez-Guerrero et al. (1995) data suggesting a lower relative risk with gel implants]. The committee concludes that there is insufficient evidence to support an association of silicone breast implants with defined connective tissue disease. That is, given the repeated finding of no elevated risk, the evidence supports the conclusion that there is no association, and therefore no justification for the use of resources in further epidemiological exploration of such an association.

Other Atypical Signs or Symptoms

The committee reviewed the literature cited earlier in this chapter and additional studies for descriptions of unusual constellations of signs or symptoms and clinical laboratory findings that might define undifferentiated connective tissue disease or a new syndrome associated with silicone breast implants, which has been proposed by some physicians (Borenstein, 1994a, cited in Tugwell, 1998; Silverman et al., 1996c; Solomon et al., 1995). Patients with undifferentiated connective tissue disease have been defined and followed as part of a multicenter project, and the association of this condition with silicone breast implants, its natural history and outcomes, and a description of its signs and symptoms have been published (Williams et al., 1997, 1998). Early undifferentiated disease was defined as at least one of the following: Raynaud's phenomenon; isolated keratoconjunctivitis sicca; unexplained polyarthritis; or at least three signs and symptoms (e.g., rash, central nervous system symptoms, pleuritis, pericarditis), and abnormal laboratory tests, (e.g., elevated erythrocyte sedimentation rate, biologic false-positive test for syphilis). At five-year follow-up, most patients with this diagnosis continue with signs and symptoms or remit, but approximately 20% were reported to have progressed to a defined connective tissue disease, usually SLE or RA (Williams et al., 1998).

In an abstract and a full report of a case control study, Williams and colleagues (Williams and Weisman, 1994; Williams et al., 1997) reported 167 women with either undifferentiated connective tissue disease (N = 89), Raynaud's phenomenon (N = 24) or unexplained polyarthritis (N = 54). As discussed earlier, 156 women with defined connective tissue disease were also reported. One woman with silicone breast implants was found in each group (specifically one woman with SSc and one with unexplained polyarthritis). Using an historical control group and counting only cases of disease occurring after implantation, the odds ratio for the combined groups was 0.74, as noted earlier. No significant risk for defined or undifferentiated connective tissue disease in women who had received silicone breast implants was found in this study, although small numbers limited its power to detect slightly increased relative risks (Williams and Weisman, 1994; Williams et al., 1997).

A case control study reported only in an abstract (Laing et al., 1996) found three women (1.46%) with implants among 205 women with undifferentiated connective tissue disease, compared to 27 women with implants in 2,220 controls (1.21%), which resulted in an odds ratio of 2.27 (95% CI, 0.67-7.71). This study also suffers from small numbers and limited power, and because it is reported in abstract form only, methodologic details are lacking, and the results have not been peer reviewed. Another abstract describing a meta-analyses of seven epidemiological studies reported a pooled odds ratio of 1.74 for non-CTD arthralgia. The absence of any data or even a history of the studies analyzed limits interpretation of this abstract (Kayler and Goodman, 1985).

Five reports (Edworthy et al., 1998; Friis et al., 1997a; Hennekens et al., 1996; Nyrén et al., 1998a; and Sánchez-Guerrero et al., 1995) discussed earlier did not study undifferentiated connective tissue disease and breast implants specifically, but did evaluate a range of conditions besides defined CTD, including early, mixed, atypical, and ill-defined connective tissue diseases; fibromyalgia; and constellations of up to 41 signs and symptoms that occur in defined and undifferentiated CTD and in rheumatic-like conditions. Another study by MacDonald et al. (1996) reported 35 cases of chronic fatigue syndrome (one breast implant in the patient and control group each), also characterized by some of these symptoms, such as fatigue, joint pain, myalgia and muscle weakness, memory difficulties, among others. These six studies collected and analyzed data on signs and symptoms that occur in these connective tissue and rheumatic-like conditions. In none of these studies was a statistically significant association of atypical disease with silicone breast implants discovered. The committee concludes that there is insufficient evidence for an association of silicone breast implants with undifferentiated connective tissue disease, and most studies suggest that there is no association.

A list of 48 signs and symptoms that might be associated with, or caused by, silicone breast implants can be found in Tugwell (1998) and 55 signs and symptoms are listed in Appendix B in this report. There is considerable overlap between these lists. Many of these signs and symptoms are also listed in reports (often abstracts) by a number of investigators (e.g., Baker, 1996; Borenstein, 1994b; Brawer, 1996; Bridges et al., 1993a; Chow et al., 1995; Cuellar et al., 1995a; Davis et al., 1995; Freundlich et al., 1994; Lewy and Ezrailson, 1996; Love et al., 1992; Mathias et al., 1996; Mease et al., 1995; Nardella, 1995; Osborn et al., 1993; Romano et al., 1995;Solomon, 1994a, b, 1996; Vasey, 1992a,b, 1994; Weiner and Paulus, 1996; Weiner et al., 1992).

In an attempt to aggregate these signs and symptoms to define a unique syndrome, a group of physicians proposed criteria for the diagnosis of a new condition [see earlier references, e.g., Borenstein (1994a) cited in Tugwell (1998)]. If such a condition could be defined and this definition applied to a reproducible set of signs, symptoms, and clinical laboratory criteria, then epidemiological studies to explore this condition in relation to silicone breast implants could be carried out. A definite diagnosis of this disease was proposed by the reference group of physicians to require the presence of a silicone gel filled breast implant and the presence of local disease, that is, any of the following: capsular contracture, rupture, more than six weeks of chest wall pain, more than six weeks of breast pain, axillary adenopathy, entrapment neuropathy, immune skin rash or immune granulomas in the implant capsule. Objectively verifiable major and minor criteria were also proposed; these include, as major criteria, symmetrical myalgia with 4-11 tender points, chronic fatigue of six-month duration, cognitive dysfunction of six-month duration, and objective sicca complex; and as minor criteria, arthralgia, enthesopathy, subjective sicca complex, cerebellovestibular dysfunction, hair loss, Raynaud's phenomenon, photosensitive skin rash, immune-mediated skin rash, improvement of two major or one major and four minor criteria within 18 months of explantation, positive ANA titer of 1:40 or more on HEp-2 cells, elevated erythrocyte sedimentation rate, and abnormal quantitative immunoglobulin G (IgG; Borenstein, 1994a cited in Tugwell, 1998; Silverman et al., 1996c; Soloman et al., 1995).

The requirement for the presence of a silicone breast implant at some time in the patient's life to make the diagnosis of this proposed disease means it cannot be diagnosed as an independent condition that can be assessed objectively. Investigations of its relationship to breast implants would be problematic at best since the condition could not exist absent a gel implant. The requirement for an implant is more consistent with a toxic reaction to silicone gel, but this perspective is not supported by available toxicological evidence (see Chapter 4). It is also of interest that there is no requirement for a saline-filled silicone implant, although atypical signs and symptoms have occasionally been reported to be associated with these as well (Byron et al., 1984; see also earlier description of CTD and Appendix B of this report). Moreover, the required presence of local disease will be fulfilled frequently (see Chapter 5), creating a large group of women eligible for this diagnosis if they display several of the required signs and symptoms (major and minor criteria), which are semiquantitative, subjective, and common in the general population.

In assessing this proposed new disease, others have also noted that there are few objective signs and that the majority of these signs and symptoms are common in the general population or frequently seen in other defined diseases (Tugwell, 1998). Hyams (1998), in a recent review of symptom-based conditions, noted that a similar constellation of signs and symptoms characterizes a number of diagnoses, such as chronic fatigue syndrome, multiple chemical sensitivities, fibromyalgia, and others, including a novel silicone-associated condition. He listed the shared features of these conditions (see Box 8-1) and, in addition to the frequency of the symptoms in the general population, reinforced the significance of the absence of a specific case definition, and the lack of established diagnostic criteria that exclude well-recognized causes of chronic somatic symptoms or distinguish various symptom-based conditions from each other. This is not to question that such patients are ill, often severely so, only to suggest that whether afflicted patients are suffering from unique scientifically defined illnesses has not been established. Others have recently stressed the prevalence of this constellation of signs and symptoms, the reliance of research on self-report data, and the uncertain boundaries of the multiple associated diagnoses (Lloyd, 1998).

Box Icon

BOX 8-1

Shared Features of Symptom-Based Conditions. Characteristic symptoms potentially involve multiple organ systems and not a recognizable pattern of complaints. Characteristic symptoms are not consistently associated with objective physical signs or laboratory (more...)

Many of these criteria have also been reviewed in existing studies of women with breast implants. The reports listed earlier are also, with few exceptions, uncontrolled case series descriptions. Although some describe thousands of women, by lacking controls they are essentially case reports of a selected group and therefore constitute only weak evidence of an association of implants with the described health conditions. These case or case series reports have gathered women with implants who have complaints, many of them self-or attorney referred, or patients who are seen in practices of rheumatologists who have evaluated women with implants for injury claims; that is, they do not represent an unbiased cross section of the population of women with implants.

Other reports have assessed various symptoms and signs, including some proposed for the new disease, as part of cohort studies (e.g., Edworthy et al., 1998; Friis et al., 1997a; Gabriel et al., 1994; Giltay et al., 1994; Kim and Harris, 1998; Nyrén et al., 1998c; Park et al., 1998; Wells, 1994; and Winther et al., 1998). These studies are discussed briefly above [except for those of Kim and Harris, 1998; Nyrén et al., 1998c; and Winther et al., 1998, which focus on neurologic symptoms and are discussed in Chapter 10]. They also collected data on individual signs and symptoms. For example, in the cohort study of Giltay et al. (1994), no difference was found between the prevalence of Raynaud's phenomenon in implant (3.69%) and control (3.33%) patients. Likewise in Park et al. (1998) one patient with Raynaud's phenomenon was found among 110 women with augmentations (0.91%) and three patients among 129 controls (2.34%), and there were seven cases of Raynaud's among 207 reconstruction patients (3.38%) and five of 88 controls (5.68%). Wells et al. (1994) found one case of Raynaud's phenomenon in 222 women with implants and no cases in 80 control women. The prevalence of Raynaud's phenomenon in a population of normal adult women 53.8 years of age on average was 9.6% according to Fraenkel et al. (1999).

Edworthy et al. (1998) found hand pain in 26% of women with implants and 18% of controls (a significant difference). Gabriel et al. (1994) found swollen joints in 3.34% of women with implants and 2.60% of controls, a relative risk of 1.35 (95% CI, 0.81-2.23). Giltay et al. (1994) reported painful joints for at least three months in 19.57% of implant patients and 8.57% of controls. In the cohort study of Park et al. (1998), 11 of 110 augmentation patients (10.0%) and 12 of 128 controls (9.38%) had joint pain, as did 31 of 207 reconstruction patients (14.98%) and 13 of 88 controls (14.77%). The prevalence of arthralgia was not significantly different in the patients studied (11%) and in controls (5%) in the cohort reported by Wells et al. (1994); swollen axillary glands were more common in the patients than in controls. The cohort studies reviewed here did not find sicca complex, rash, myalgia, skin tightening, or thickening significantly more frequently in women with breast implants compared to controls.

In a study that evaluated a constellation of 23 signs and symptons, there was no difference between experimental and control groups for 20 of the 23 (Wells et al., 1994). When 1,016 women with gel implants were compared with 309 women with saline implants and 609 cosmetic surgery controls by a survey questionnaire that sought information about memory difficulty, myalgia, arthralgia, numbness and headache, among others, the frequencies in the three groups were similar (Barr et al., 1998). When 105 women with breast implants were evaluated by Abeles and Waterman (1995), only fatigue, cognitive disorder, and arthralgia were present in 50% or more of the women. Others of the 14 "key" signs and symptoms of the proposed new disease, or those commonly attributed to patients with breast implants, were infrequent (5-25%). Women averaged only 2.3 signs and symptoms and a consistent clinical picture could not be defined (Abeles and Waterman, 1995). Similar findings were reported by Weisman et al. (1998) in a study described earlier in this report. Of 125 women with implants, 38 (30%) were found to have signs and symptoms, including fibromyalgia, osteoarthritis, bursitis, or isolated back or knee pain. No atypical disease was found (Weisman et al., 1988).

Blackburn and colleagues examining a selected group of 70 physician-referred women with silicone gel implants in place for an average of 10.2 years, found one case of postimplant SLE and one case of SS and also fibromyalgia, osteoarthitis, or soft-tissue rheumatism (Blackburn and Everson, 1997; Blackburn et al., 1997). Moreover, measurement of interleukin-6 (IL-6), IL-8, soluble intercellular adhesion molecule-1 (sICAM-1), soluble IL-2 receptor (sIL-2R), and tumor necrosis factor-alpha (TNFα) in peripheral blood of varying numbers of these women did not reveal elevations compared to ten concurrent and an unknown number of historical controls (see the discussion of cytokines in Chapter 6). In 13 of these 70 women (19%), ANAs were detected at titers of 1:40 or more. These investigators also concluded that an atypical disease could not be demonstrated in these patients (Blackburn and Everson, 1997; Blackburn et al., 1997).

Considerable variation in the prevalence of individual signs and symptoms was reported when the frequencies described in case series by Vasey et al., 1992b; Solomon et al., 1994a; Freundlich et al., 1994 and Borenstein were compared (Borenstein, 1994b). Todhunter and Farrow (1998) analyzed the overlap of signs and symptoms often, but variably, described in the multiple reports and abstracts listed and discussed above, which were also examined in some of the studies of defined connective tissue disease discussed earlier. They found these signs and symptoms covered within the endpoints examined by the epidemiological studies, and these studies did not show a statistically significant association of the signs and symptoms (or defined CTD) with breast implants. They concluded that this makes a specific atypical syndrome unlikely (Todhunter and Farrow, 1998).


The committee finds no convincing evidence for atypical connective tissue or rheumatic disease or a novel constellation of signs and symptoms in women with silicone breast implants. Case reports, of which there are many, do not provide evidence, although they may suggest hypotheses that can be tested, as has been possible for defined CTD. A defined and testable disease is a precondition for any type of study. Given the frequency of local complications in women with silicone breast implants and the frequency and subjective nature of the symptoms that have been proposed by some to characterize a hypothetical novel disease, a large group of women would meet the criteria for this disease if such a definition were accepted. The diagnosis would ultimately depend on conditions such as fatigue, cognitive dysfunction, arthralgia, and the like which are nonspecific and common. A new disease would then be created by the discovery of an implant and its common local complications in women who had signs and symptoms prevalent in the general population or in fibromyalgia, chronic fatigue syndrome, multiple chemical sensitivities, or other less well-defined conditions. As noted earlier, silicone toxicity is conceptually more straightforward, although it is not supported by the toxicologic data reviewed in Chapter 4.

The evidence for an atypical disease or a novel syndrome is insufficient or flawed. It consists of selected case series, few of which describe a consistent and reproducible syndrome. The controlled epidemiological studies cited provide stronger, contrary evidence. In view of the paucity, weakness, and conflicting nature of the evidence, the committee concludes that there is no rigorous, convincing scientific support for atypical connective tissue or any new disease in women that is associated with silicone breast implants. In fact, epidemiological evidence suggests there is no novel syndrome.

Copyright © 2000, National Academy of Sciences.
Bookshelf ID: NBK44786


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...