It has been estimated that over 1 million women in the United States have had breast implants. Approximately 80 percent of these implants were used for cosmetic breast enlargement, generally in relatively young, healthy women. Surprisingly, despite the increasing incidence of breast cancer in the United States during the period of breast implant availability, most of the attention of the scientific community, the media, and others has been directed toward the rare occurrence of autoimmune disease. Questions regarding the possible carcinogenicity of these devices form the basis of this investigation.
Since 1976 we have been monitoring the cancer occurrence in a cohort of women in Los Angeles County who had undergone cosmetic augmentation of the breast.1,2 Members of this cohort have experienced many years of implant exposure and are now entering age groups with high breast cancer risk. In addition to concerns about the carcinogenic potential of breast implants, their presence complicates mammographic and physical examination of the breast, generating concern that this may result in delayed detection and thus poorer prognosis following breast cancer.
Materials and Methods
Records of all non-Spanish-surnamed white women who received a breast implant and who resided in Los Angeles County at the time of implantation were abstracted from the practices of 35 board-certified plastic surgeons; patients with prophylactic subcutaneous mastectomy or breast carcinoma prior to implantation were excluded. The included patients received their implants between 1953 and 1980. Approximately 100 board-certified plastic surgeons practiced in Los Angeles County at that time, although it is unknown how many performed augmentation mammaplasty. Of 3500 patients so enrolled, 3182 were followed during the study period. Identifying [Illegible Text] demographic information was extracted from these medical records along with details of the surgical procedure and subsequent visits. Strict patient confidentiality was maintained at all times.
Breast Cancer Detection
The Los Angeles County Cancer Surveillance Program, a population-based cancer registry was used to identify invasive and in situ breast cancer cases occurring among the study cohort. The registry has complete ascertainment of all cancers diagnosed among Los Angeles County residents for the period January 1, 1972, through December 31, 1991.
The Los Angeles County Cancer Surveillance Program database was searched for study subjects by means of computerized record linkage. Using all available name, birth date, and Social Security Number information for each implant patient, up to five potential matches were sought. Each potential match was manually judged to be an exact match, possible match, or nonmatch. For exact and possible matches, the Los Angeles County Cancer Surveillance Program and implant patient records were further examined for evidence of match or nonmatch.
While the record linkage methodology minimized risk of breach of confidentiality, implant patients who had moved out of Los Angeles County might not have had subsequent cancer diagnoses recorded by the Los Angeles County Cancer Surveillance Program. For this reason it was necessary to determine the residence of each implant patient throughout the study period. Since direct contact with patients was not allowed to ensure protection of confidentiality and to preserve privacy, this was performed utilizing a variety of publicly available information sources.
Residence in Los Angeles County was confirmed at the time the cohort was formed by obtaining independent documentation of this fact through public records. This eliminated nonresidents who came to Los Angeles temporarily for the surgery and patients using pseudonyms. Further documentation of place of residence has been collected for each patient through 1991. We accomplished this by using public records, i.e., driver's and vehicle license records, voter registration rolls, property tax rolls, marriage and divorce records, as well as two large multistate credit bureaus. Only dates corresponding with an activity, e.g., renewing a driver's license or registering to vote, were accepted as valid; simply remaining on a roster was not considered sufficient evidence of residency or vital status. Failure to document continuing Los Angeles County residence constituted loss to follow-up beginning on the day following the last documentation of a Los Angeles County residence.
Expected Number of Breast Cancers
Having determined the period of postimplant follow-up for each patient, the period during which incident cases of cancer would be recorded in the Los Angeles County Cancer Surveillance Program was calculated, commencing with the date of implantation or January 1, 1972, whichever was later, and terminating on the date of breast cancer diagnosis, the date of last documented residence in the county, date of death, or December 31, 1991, whichever was earliest.
Population-based age-, race-, and socioeconomic-specific breast cancer incidence rates for Los Angeles County were used to estimate the expected number of cancers in the cohort. Since women with prior breast cancer were excluded from the augmentation mammaplasty cohort, the population incidence rates included only first breast cancers. To account for the changes in breast cancer incidence rates among non-Spanish-surnamed whites in Los Angeles County over the 20-year period, we calculated the incidence rates for each of four 5-year periods. Since breast cancer risk is positively associated with socioeconomic status and augmentation patients tend to be of higher socioeconomic status, each patient was assigned a score of 1 to 5 based on the average family income and average educational level as reported by the United States Census Bureau for their census tract of residence; all patients in the Los Angeles County Cancer Surveillance Program are classified in the same manner. The expected number of breast cancer cases was calculated by multiplying these age- and period-specific incidence rates within socioeconomic strata by the age-specific person-years of follow-up of the implant patients in each period. Standardized incidence ratios were calculated as the number of observed breast cancer cases divided by the age-adjusted expected number of cases multiplied by 100 percent. Thus the number of observed cases is expressed as a percentage of the expected; a value greater than 100 percent indicates that more cases were observed than expected. Exact 95 percent confidence limits for the standardized incidence ratios were calculated.3
Stage at Detection
The Los Angeles County Cancer Surveillance Program began collecting summary information on disease stage for cancer patients in Los Angeles County in 1976. We compared the stage at diagnosis among breast cancer patients in the cohort with population-based figures. Age-adjusted percentage distributions of breast cancers by stage at diagnosis among cohort members were standardized to the age distribution of all breast cancers occurring among non-Spanish-surnamed white women aged 35 to 74 years in Los Angeles County during the years 1976 to 1991.
Of the original 3500 patients, independent documentation of name or Los Angeles County address could not be found for 84 (Table I). No follow-up was accrued for 108 patients who were lost to follow-up immediately following the date of surgery. Another 126 patients were followed but moved out of the county, died, or were lost to follow-up prior to 1972. Although none of these 318 women were included in calculations of the expected numbers of breast cancers, all were included in the linkage with the Los Angeles County Cancer Surveillance Program; no breast cancer cases were found among them.
The augmentation surgeries were performed between 1953 and 1980, with 91 percent of the procedures performed between 1970 and 1980. Eighty-eight percent of the devices used at first implant were silicone gel or silicone/saline-filled (Table II). Sixty of the patients received polyurethane-coated silicone implants. Among the patients for whom the type of implant was classified as unknown, we have reason to believe that 9 more have received polyurethane devices. The median age of the patients at the time of first implant was 31.4 years.
The 3182 patients provided 37,439 person-years of postimplantation experience in the period from January 1, 1972 through December 31, 1991(median follow-up 14.4 years, range 0.04 to 20.0 years). When the years of exposure before 1972 are included, the range of implant exposure was 0.04 to 36.5 years.
As of December 31, 1991, 31 cohort members had been diagnosed with breast carcinoma. Based on the era-specific, socioeconomic status-specific, age-specific breast cancer rates for non-Spanish-surnamed white women in Los Angeles County, 49.2 cases were expected (standardized incidence ratio = 63.0 percent; 95 percent confidence limits: 42.8 and 89.5 percent)(Table III). No sarcomas of the breast were diagnosed among cohort members.
The median age at implant of the 31 women who were subsequently diagnosed with breast cancer was 37.4 years (range 23.3 to 62.1 years), 6 years older than the median age at implant of the entire cohort. The median age at diagnosis was 47.2 years (range 34.5 to 74.5 years), and the median interval from implant to diagnosis was 10.3 years (range 2.9 to 30.0 years). The distribution of type of implants among the breast cancer cases was similar to that of the entire cohort (see Table II). Six other cohort members are known to have been diagnosed with breast cancer. Three were diagnosed after they left Los Angeles County and subsequently returned(although their participation was censored at the date they last resided in the county), and three others were diagnosed after 1991.
Since the breast cancer patients were older than the study cohort as a whole, the analysis was repeated, restricting the cohort to those implanted after age 40. These 580 patients had 7187 person-years of follow-up. Fourteen breast cancers were diagnosed, and 18.8 cases were expected (standardized incidence ratio = 74.5 percent, 95 percent confidence limits: 40.6 and 125.3 percent). We also examined risk by years of exposure to determine whether breast cancer risk increases with length of implant exposure; no increase was found (see Table III).
Although the risk of breast cancer in augmentation mammaplasty patients was not increased, it has been suggested that these radiologically opaque devices could hinder detection of malignancies,4-6 resulting in later stage at diagnosis and poorer prognosis. Breast cancers diagnosed among augmentation mammaplasty patients occur at essentially the same disease stage as those diagnosed among all non-Spanish-surnamed white women in Los Angeles County (Table IV).
The medical risks of breast implants have received intense scrutiny in recent years, with empirical evidence to support or reject putative associations only recently becoming available. Foreign-body neogenesis, first described in 1941,7 can be induced in certain susceptible rodents by a wide variety of smooth-surfaced materials such as nylon, cellophane, polystyrene, glass, and metals.8-10 This appears to be dependent on the form of the material rather than its chemical composition, with tumors developing in response to smooth surfaces, as opposed to rough or discontinuous materials. The relatively large, smooth-walled silicone- or saline-filled devices used for augmentation mammaplasty closely fit the profile of devices that pose the highest risk in animal studies.
Some 22 cases of sarcomatous lesions at the site of artificial implants have been reported11; none were associated with a breast implant. A case-control study of soft-tissue sarcomas in men revealed no increase associated with artificial implants (OR = 0.66, 95 percent confidence limits: 0.32 and 1.25).11
Since sarcomas are rare in humans, especially breast sarcomas, even a modest number of additional cases could result in a detectable increase in incidence. Using data from the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, a study conducted by the Centers for Disease Control found no change in the incidence rates of malignant fibrosarcoma or other malignant sarcomas of the breast between 1973 and 1986.12
A recent extended analysis of SEER data demonstrated no change in female breast sarcoma incidence rates from 1973 to 1990. The mean age-adjusted incidence rate was 0.13 per 100,000 women for 1973 to 1981 as compared with 0.12 per 100,000 women for 1981 to 1990, despite the fact that the latter time period contained an estimated ninefold greater number of person-years of exposure to breast implants.13 A review of all soft-tissue sarcomas seen between 1982 and 1986 at the Memorial Sloan Kettering Cancer Center identified 1113 cases; only 6 (0.5 percent) of these were located in the breast, and none of the patients had received a breast implant (M. F. Brennan, personal communication). No cases of breast sarcoma were diagnosed in our cohort of augmentation mammaplasty patients.
A pattern of remarkable consistency of breast cancer risk estimates among breast implant patients has emerged from published epidemiologic studies. In Los Angeles we observed 63 percent of the expected breast cancer incidence, and the incidence in older and longer-exposed subgroups remains below the expected rates. A much larger study of very similar design has reported similar results.14 Of 10,835 augmentation mammaplasty patients in Alberta, Canada, with 89,219 person-years of exposure, only 45 breast cancers or 76 percent of the expected incidence of breast cancer was observed after an average follow-up of 10.2 years. The risk dropped even lower when analyzed separately for invasive breast cancers to 72 percent (95 percent confidence limits: 51 and 98 percent). A record linkage study from the Danish Cancer Registry with a cohort of 824 cosmetic breast implant patients with an average follow-up of 6.9 years yielded 1 breast cancer as compared with 4.2 expected (standardized incidence ratio = 24 percent; 95 percent confidence limits: 0 and 31 percent).15 Similarly, a Swedish registry linkage of 1756 cosmetic breast implant patients with 20,610 person-years of exposure (average 11.7 years) detected 7 breast cancers with 11.2 expected (standardized incidence ratio = 63 percent; 95 percent confidence limits: 30 and 130 percent).16
Three breast cancer case-control studies provide similar results. In one, among women less than 45 years old at diagnosis, the relative risk associated with breast implants was 0.8 (95 percent confidence limits: 0.3 and 2.2).17 In another, among women aged 50 to 64 years at diagnosis, the risk was 0.2 (95 percent confidence limits: 0.1 and 1.3). In a large multicentered breast cancer study of 2174 patients and 2009 population controls, the estimated relative risk associated with prior implant was 0.6(95 percent confidence limits: 0.4 and 1.0)18. Finally, recently cited results from the very large Cancer and Steroid Hormone (CASH) study of 4730 breast cancer patients and a similar number of controls provide a relative risk of 1.0.19
A limitation of the cohort linkage studies is the inability to adjust risk estimates for some known breast cancer risk factors. The strongest risk factor for breast cancer by far is age.20 As described previously, our estimates of the expected numbers of breast cancers are based on incidence rates for women of the same ages as our study population. Breast cancer rates also vary significantly by race,20 but our cohort and the rates generating the expected numbers of cancers were limited to non-Spanish-surnamed whites. Breast cancer rates are also higher among women of higher socioeconomic status,20 and Census Bureau data revealed that the implant patients' census tracts of residence were, on average, of higher income and educational levels than the county population as a whole, hence our socioeconomic status adjustment.
Other factors such as history of breast cancer in first-degree female relatives, younger age at menarche, later age at first full-term pregnancy and fewer full-term pregnancies, and later age at menopause are associated with increased risk.20 However, data on these factors were not available from the patients' medical records, and we were not permitted to contact patients directly. Augmentation mammaplasty patients could be significantly different from the general population in regard to these factors. Use of population-based breast cancer rates that are unadjusted for such factors would possibly overestimate or underestimate the expected number of breast cancers in our cohort.
These breast cancer risk factors also may have influenced the decision of whether to perform the augmentation mammaplasty procedure. If a woman contemplating augmentation mammaplasty was known to be at increased risk, e.g., had a first-degree relative with breast cancer, it is possible that she or her surgeon might have declined the procedure in the belief that the implant would further increase risk or inhibit detection. If this occurred, the augmentation mammaplasty patients would be a select population at somewhat lower risk of cancer than the general population, and the expected number of breast cancer cases would be too high. However, the result from the above-cited large breast cancer case-control study18 is adjusted for age, race, family history of breast cancer, body size, and screening history; the resulting 0.6 relative risk is very similar to the linkage study results, providing strong evidence that the observed reduction in breast cancer risk among augmentation mammaplasty patients is not explained by selection bias or confounding.
Perhaps it is still too early to detect a breast implant-related increase in breast cancer incidence. A literature review of anecdotally reported foreign-body-associated malignancies yielded 43 cases with an apparent trimodal pattern of incidence, with peaks at about 1 year, 14 years, and 28 years.21 The latency period for breast cancer is not well characterized and probably differs by etiologic agent; atomic bomb survivor data and risk associated with factors such as menarche and age at first birth suggest at least a 15-year latency period.22 Although the patients with 15 or more years of exposure in the present study experienced only 70 percent of the expected number of breast cancers, with a median of 14.4 years of exposure for the entire cohort, additional follow-up will be important.
The fact that the observed number of breast cancer cases in the cohort remains at a level well below our best estimate of the expected number offers some assurance that a truly significant increase in breast cancer is not being missed. Given that 49.2 breast cancer cases would be expected in the cohort, 64 breast cancers would have to have occurred for this cohort to be at significantly increased risk, 33 more than were actually observed. All research-based estimates of breast cancer risk among breast implant patients fall at 1.0 or below.
We have continued to follow cohort members who received implants in the period 1953 to 1971 even though we were unable to ascertain breast cancers that occurred before the inception of the Los Angeles County Cancer Surveillance Program in 1972. These 607 patients are of particular interest because they represent the longest exposure to implants, and their 1737 person-years of exposure before 1972 (median 1.9 years) are not utilized in our analysis. To estimate the number of breast cancers that would have been expected in this subgroup prior to 1972, we applied the earliest Los Angeles County breast cancer incidence rates (1972-1976) to the age-specific person-years of exposure. Based on these population rates, 1.1 breast cancers would have been expected prior to 1972. Among all 31 breast cancers observed in the full cohort by 1991, the shortest interval from implant to diagnosis was 2.9 years. Thus, given the low number of breast cancers expected before 1972 and the short (1.9-year) median interval from implant to 1972, we believe that very few, if any, breast cancers were missed from this period.
To further investigate whether inclusion of the patients implanted before 1972 affected our overall breast cancer risk estimate, we have conducted a separate analysis excluding them. When we include only those patients who received implants after 1971, for whom the postimplant follow-up is covered by the Los Angeles County Cancer Surveillance Program (2575 women with 29,067 person-years, median follow-up 13.8 years), 21 breast cancers were observed and 36.4 were expected (standardized incidence ratio = 57.7 percent, 95 percent confidence limits: 35.7 and 88.2 percent).
Conjecture that polyurethane could degrade to 2,4-toluene diamine, a carcinogen in cancer-prone rats, has raised questions as to whether devices with polyurethane covers might be associated with increased risk of cancer in humans. In our study, none of the breast cancer patients had received silicone gel-filled polyurethane-covered implants. However, one patient with polyurethane sponge implants developed breast cancer during the study period.
Some experimental work suggests possible anticarcinogenic effects of silicone implants. A recent small in vitro study showed inhibited growth of human breast adenocarcinoma cells when cultured with plasma of women with silicone gel implants as compared with controls.23 The reduced risk of breast cancer observed in our study is also similar to that found experimentally in a rat study by Dreyfuss et al.24 Three hundred female Sprague-Dawley rats were randomized to receive either a silicone implant, silicone-free gel, a silicone sheet, a polyurethane sheet, or a sham operation. Each group was further randomized into three subgroups to receive N-methyl-N-nitrosourea as a carcinogenic stimulus at the time of implant, N-methyl-N-nitrosourea 14 days after implantation, or saline. No tumors were observed in the saline-injected group. Among the N-methyl-N-nitrosourea-stimulated animals, a significantly lower incidence of cancer was observed only among the silicone gel-implanted rats who had received the delayed N-methyl-N-nitrosourea injection. The authors concluded that “it appears that the silicone implant provides a protective effect or at least a delay in tumor formation.” In another animal study, Fischer rats were injected intradermally with carcinoma 13762 mammary carcinoma cells and randomized to four experimental groups: non-operated controls, sham operations, implanted with a tissue expander unexpanded, and implanted with an expanded tissue expander.25 Average tumor weights for the four groups were 8.7, 9.5, 1.9, and 0.8 gm, respectively, demonstrating statistically significant reduction in tumor growth in the presence of expanders, even when the devices were not expanded.
Two additional studies have addressed the possibility that breast implants may actually reduce cancer incidence. Sprague-Dawley rats were randomized to silicone implants or sham operations in one of three anatomic locations and a subsequent stimulation with N-methyl-N-nitrosourea.26 Again, when implants were placed beneath the mammary gland, significantly fewer rats(11.5 percent) developed cancer than the sham-operated rats (64.3 percent). The animals receiving implants in the dorsum or intraperitoneal cavity also developed fewer tumors than animals with sham operations at these locations. Furthermore, mammary gland site specificity is suggested by the fact that animals with mammary implants developed significantly fewer tumors than the animals with abdominal or dorsal implants. The second experiment was designed to determine whether reduced tumorigenesis would be observed even in the absence of exogenous chemical carcinogen stimulation.26 C3H/OuJ mice were randomized to implantation of a silicone implant, free silicone gel, silicone sheet, or a sham operation. Only 17 percent of implant-exposed animals developed carcinomas compared with 45 percent of those exposed to free gel, 42 percent with silicone sheets, and 50 percent with sham operations. No sarcomas were observed.
Together, the human and animal evidence suggests that the reduction in breast cancer in the presence of implants is not explained by unmeasured confounding. How, then, could breast implants reduce the risk of cancer and slow the growth of tumors when they do occur? Although it is beyond the scope of this report to detail the evidence for and against each, at least three mechanisms have been suggested. Breast implants are very large foreign bodies and clearly stimulate a foreign-body response. A chronic enhanced local immune surveillance could provide a mechanism in which both carcinogens and transformed cells are more actively destroyed. A pressure or tissue-compression mechanism also has been suggested. The weight and volume of the device may compress the glandular tissue, decreasing blood supply, or in some other way reduce the rate of cell proliferation. This hypothesis is also consistent with the observation of smaller, apparently slower-growing tumors in the presence of implants. Third, some patients report their breasts to be colder following augmentation mammaplasty. The presence of a large, nonvascularized mass largely separating the breast tissue from the body could lower the temperature of the tissue, which could cause a long-term reduction in the metabolic rate. Further investigation of these and other mechanisms could provide important clues for prevention of human breast cancer.
When breast cancer does occur in the implanted breast, is detection delayed? We found no difference in breast cancer stage at diagnosis among implanted women as compared with the general population. In a population-based report from Alberta, Canada,27 41 augmentation mammaplasty patients who subsequently were diagnosed with breast cancer were not diagnosed at a later stage nor did they experience impaired survival when compared with breast cancer patients without implants.
Thus, in our study of cancer following cosmetic augmentation mammaplasty, we find (1) a significant reduction in risk of breast cancer, (2) no breast cancers among the small number of patients with polyurethane-covered implants,(3) no evidence of sarcomas produced by solid-state carcinogenesis as is observed in rodents, and (3) no delay in the detection of breast cancer.
Dennis Deapen, Dr.P.H.
Department of Preventive Medicine; School of Medicine; University of Southern California; 1540 Alcazar Street, Suite 204; Los Angeles, Calif. 90033
Malcolm C. Pike, Ph.D., and John T. Casagrande, Dr.P.H., participated in the design and implementation of this study. We wish to thank Janice Schaefer, R.N., for data management, Richard Pinder for record linkage, Sandra Finalet for patient tracing, Angela Miu, M.S., for technical support, and the staff of the USC Cancer Surveillance Program. We continue to be indebted to the 35 participating plastic surgeons and their office staffs for their trust, cooperation, and assistance in making the records available.
This work was supported in part by NIH Grants CA-14089 and CA-17054 and grants from the Dow Corning Corporation, Cooper Surgical, McGhan Medical Corporation, and Mentor Corporation. Cancer incidence data have been collected under Subcontracts 050A-8709 through 050E-8709 with the California Public Health Foundation. The subcontract is supported by the California Department of Health Services as part of its statewide cancer reporting program, mandated by Health and Safety Code Section 210 and 211.3. The ideas and opinions expressed herein are those of the [Illegible Text] and no endorsement of the State of California, Department of Health Services, or the California Public Health Foundation is intended or should be inferred.
1. Deapen, D. M., Pike, M. C., Casagrande, J. T., and Brody, G. S. The relationship between breast cancer and augmentation mammaplasty: An epidemiologic study. Plast. Reconstr. Surg.
77: 361, 1986.
2. Deapen, D. M., and Brody, G. S. Augmentation mammaplasty and breast cancer: A 5-year update of the Los Angeles study. Plast. Reconstr. Surg.
89: 660, 1992.
3. Breslow, N. E., and Day, N. E. (Eds.). Statistical Methods in Cancer Research
, Vol. II: The Design and Analysis of Cohort Studies
. London: Oxford University Press, 1987. Pp. 69-71.
4. Silverstein, M. J., Handel, N., Gamagami, P., et al. Breast cancer in women after augmentation mammoplasty. Arch. Surg.
123: 681, 1988.
5. Hayes, H., Jr., Vandergrift, J., and Diner, W. C. Mammography and breast implants. Plast. Reconstr. Surg.
82: 1, 1988.
6. Silverstein, M. J., Gamagami, P., and Handel, N. Missed breast cancer in an augmented woman using implant displacement mammography. Ann. Plast. Surg.
25: 210, 1990.
7. Turner, F. C. Sarcoma at sites of subcutaneously implanted Bakelite discs in rats. J. Natl. Cancer Inst.
2: 81, 1941.
8. Oppenheimer, B. S., Oppenheimer, E. T., and Stout, A. P. Sarcomas induced in rodents by imbedding various plastic films. Proc. Soc. Exp. Biol. Med.
79: 366, 1952.
9. Oppenheimer, B. S., Oppenheimer, E. T., Danishefsky, I., et al. Further studies of polymers as carcinogenic agents in animals. Cancer Res.
15: 333, 1955.
10. Oppenheimer, E. T., Willhite, M., Stout, A. P., et al. A comparative study of the effects of imbedding cellophane and polystyrene films in rats. Cancer Res.
24: 379, 1964.
11. Morgan, R. W., and Elcock, M. Artificial implants and soft tissue sarcomas. J. Clin. Epidemiol.
48: 545, 1995.
12. May, D. S., and Stroup, N. E. The incidence of sarcomas of the breast among women in the United States, 1973-1986 (Letter). Plast. Reconstr. Surg.
87: 193, 1991.
13. Engel, A., Lamm, S. H., and Lai, S. H. Human breast sarcoma and human breast implantation: A time trend analysis based on SEER data, 1973-1990. J. Clin. Epidemiol.
48: 539, 1995.
14. Bryant, H., and Brasher, P. M. A. Breast implants and breast cancer: Reanalysis of a linkage study. N. Engl. J. Med.
332: 1535, 1995.
15. McLaughlin, J. K., Fraumeni, J. F., Jr., Olsen, J., and Mellemkjaer, L. Breast implants, cancer, and systemic sclerosis (Letter). J. Natl. Cancer Inst.
86: 1424, 1994.
16. McLaughlin, J. K., Fraumeni, J. F., Jr., Nyren, O., and Adami, H.-O. Silicone breast implants and risk of cancer? (Letter). J.A.M.A.
273: 116, 1995.
17. Malone, K. E., Stanford, J. L., Daling, J. R., and Voigt, L. F. Implants and breast cancer (Letter). Lancet
1: 1365, 1992.
18. Brinton, L. A., Malone, K. E., Coates, R. J., et al. Breast enlargement and reduction: Results from a breast cancer case-control study. Plast. Reconstr. Surg.
97: 269, 1996.
19. Brinton, L. A., Toniolo, P., and Pasternack, B. S. Epidemiologic follow-up studies of breast augmentation patients. J. Clin. Epidemiol.
48: 557, 1995.
20. Kelsey, J. L. Breast cancer epidemiology: Summary and future directions. Epidemiol. Rev.
15: 256, 1993.
21. Brand, K. G., and Brand, I. Risk assessment of carcinogenesis at implantation sites. Plast. Reconstr. Surg.
66: 591, 1980.
22. Committee on the Relationship Between Oral Contraceptives and Breast Cancer. Oral Contraceptives and Breast Cancer
. Washington: National Academy Press, 1991. P. 96.
23. Garrido, L., Bogdanova, A. Y., Pfleiderer, B., and Brady, T. J. Degradation and Bioactivity of Implanted Silicones (Abstract). In Proceedings of the Second Annual Meeting of the Society of Magnetic Resonance
. 1994. P. 1468.
24. Dreyfuss, D. A., Singh, S., Dowlatshahi, K., and Krizek, T. J. Silicone implants as an anticarcinogen. Surg. Forum
38: 587, 1987.
25. Ramasastry, S. S., Weinstein, L. W., Zerbe, A., et al. Regression of local and distant tumor growth by tissue expansion: An experimental study of mammary carcinoma 13762 in rats. Plast. Reconstr. Surg.
87: 1, 1991.
26. Su, C. W., Dreyfuss, D. A., Krizek, T. J., and Leoni, K. J. Silicone implants and the inhibition of cancer. Plast. Reconstr. Surg.
96: 513, 1995.
27. Birdsell, D. A., Jenkins, H., and Berkel, H. Breast cancer diagnosis and survival in women with and without breast implants. Plast. Reconstr. Surg.
92: 795, 1993.