Frederik Wolfe and Janice Anderson

ABSTRACT: Objective. The symptoms of what been called silicone implant syndrome

Beginning in the 1980s, a series of reports linked silicone (filled) breast implant (SBI) to connective tissue disorders such as scleroderma, systemic lupus erythematosus (SLE), and rheumatoid arthristis (RA) and later to what has been called "atypical" connective tissus disorders or silicone implant associated syndrome (SIAS). But epidemiological studies have failed to show associations between well recognized conditions such as scleroderma, SLE, and RA and SBI. Among the criticism of these epidemiological studies has been that they were not designed to identify "atypical" diseases or SIAS. On the order side of the question, there have been a number of case series of SBI patients in these series have usually been self-referred or attorney referred.

In 1995 we reported the lack of association between RA. Osteoarthritis (OA), fibromyalgia (FM), and SBI. We have recently reported that 37-55% of all patients with FM satisfy the Bridges criteria for SIAS and that SIAS does not appear to be a distinct rheumatic disease. Because the relationship of FM to SBI has not been settled, it appeared appropriate to present more formally the data of the 1995 abstract.

MATERIALS ANS METHODS

Patients in the present report had RA, OA or FM seen in a rheumatic desease clinic from 1991 through 1994. In this clinic all patient data are entered into a computer at the time of the patients'visit. In addition, patients with RA,OA and FM who agree to participate in an outcome study are also assessed by mailed questionnaires. Particulary germane to this report , the date fo onset of the first symptom of patient's disease is enterred into the computer at the first clinic visit. We also record all types of surgery that patients have undergone. New illnesses and surgeries are updated at each subsequent visit using a formalized and standardized protocol. The method of data capture and assessments have been reported.

Date of desease onset. The disease onset date was coded by the same method and same person (FW) beginning in 1974 when the data bank was established. For RA, the symptom date represented the first manifestation of RA - usually characteristic RA joint swelling. For OA, the symptom date reflected the first persistent manifestation of pain in the affected joint. Symptom dates are more complicated in FM. In general, the symptom date was the first manifestation of a pattern or a symptom complex thought to be typical of FM. For example. If a patient said she had always had pain since early adolescence, that was taken as date of first symptoms. Similarly, in a pattern of joint and muscle pain in different regions, one after another, the first joint pain or the time when other symptoms of distress began to be present was taken as the first symptom date. Isolated musculoskeletal symptoms were not taken as the onset date of FM. These determination were before there was anh thought to analyze the date for implant status, and in no instance dit implant status have anything to do with this determination.

Implant status. To determinate whether patients had breast implants, we examined the record of each patient who was listed as having breast surgery. All patiens who had an implant listed in the chart, or who were listed as having a mastectomy, or in whom the breast surgery was not precisely identified were contacted by telephone to confirm their implant status, date of the implant and to determine whether the implant was a silicone filled or a salin filed implant. To ensure no case was missed among those listed as implant negative, we also read every chart of the patient with RA, OA and FM.

Control subjects. Nonrheumatic disease controls were obtained form a list of community subjects who participated in a previous study. Briefly, a liste of 10,000 subjects was purchased from a source of Wichita names, address and telephone numbers. The list whicf was initially sorted by postal (zip) code, was placed in random order by a computer randomization program. From this list, consecutive women were contacted. Each was asked her age and whether she had undergone a breat implant and if she had she was further asked the type of the implant. Patients with OA served as rheumatic disease controls.

Data analysis. To exclude patients who were eigher too old to have had implants based on age or too old based upon when implantation was begun in the US, only patients 75 years of age or younger were studied.

We were interested in 3 sets of implants: all implants whether they were saline or silicone filled implants, and silicone filled implant that were performed prior to disease onset. A series of different controls were utilized. First, patients with RA and FM were compared with OA patients, then with communith controls, then with both control groups combined. Finally, patients with FM were compared with all OA,RA and community control together. The rationale for the use of these controls was that the OA patients represented clinic controls, the community patients represented nondisease controls, and the combined OA and community subjects represented the set of controls that have never been considered to be associated with SBI. Finally, since RA has never been shown to be associated with SBI in any study, and was not associated with SBI in this study, we used the combination of RA, OA, and community subjects as a complete set of controls.

Confidence intervals (CI) for the proportion of patients with implants were calculated using Poison assumptions (table 1). Odds ratios (OR) were calculated using logistic regression. In preliminary analyses we performed the logistic repression while controlling for age. The inclusion of age in the model yielded about the same results as the model without age. Because there was no substantial difference in results, we present the data in table 2 without additionally controlling for age. Data were analyzed with Stata Version 5.0. Statistical signifiance was set at the 5% level except as describeb in the text.

RESULTS

About 0.8% of persons in this study underwent breast implantation compared to about 1% noted nationally and in a study at the Mayo Clinic. OF the 14 implants. 12 were sicilicone filled and 2 were saline filled. Or the 6FM patients with silicone filled implants, 2 of the silicone filled implants postdated the onset of FM and 4 preceded the development of FM.

We assessed the association of implantation as noted in table 2. Regardless of the controls used and the implantation type, no association was noted between RA and implantation, as evidenced by the large p values and wide CI.

No association between predisease silicone filled implantation and FM was detected regardless of control group that was used. When all non-FM subjects were considered as controls the OR for the effect of implantation on FM was 1.22 (CI 0.30, 4.89) (p=0.781). However, when all implants were considerex, OR for the relation between implants and FM was 2.45 (CI0.86,703), p=0.095.IF90% rather than 95% CI are applied, the CI for this relationship is 1.01, 5.94.

DISCUSSION

There is general agreement that the best way to untangle the relationship between SBI and subsequent rheumatic disease is to follow prospectively those persons who receive implants, taking appropriate care that they are desease-free at the time of the implant, while monitoring the development of new diseases. Some sort of control group will be needed, and control groups can be derived from population data or from simultaneous control groups. TO date, no such studies have been reported.

The classic epidemiological assessment was performed by Gabriel, et all. In their effort they used the Mayo Clinir epidemiologiy study to investigate the rates of disease in persons with and without SBI. They did not find evidence of an increase in well defined rheumatic diseases such as RA, SLE, or scleroderma in their careful, retrospective assessment. In this suty they relied on clinicians diagnoses, but verified each case by reviewing medical records. No estimate was made of SIAS or FM.

A similar study desing was used by Hennekens et al in the Harvard Nurses study. This effort found no association between RA, SLE, and scleroderma and SBI, although a slight,

Overall association was found with connective tissue diseases in general. One problem with this eport was that diagnoses were determined by self-report and therefore that over-reporting may have occurred. They also did not assess SIAS or FM.

Another type of study desing involves examining persons with various rheumatic diseases and comparing the rates of SBI in those with desease to those without disease. Using this method, which is also the method of the current report, Goldman and all found no increase in SBI in patients with RA compared to controls. Golman also reported that 37 of 43 FM patients with implans were diagnosed after implatation. But in that report the relation between onset of FM and implantation was not describeb, making interpretation difficult. Also the number of patients withFM who had implants appeared to be increased compred to patients with other rheumatic deseases in that study, although the number of patients studied was not given.

Our study desing has a number of limitations. First, the results might be influenced by the referral characteristics of the clinic in that in some clinics many SBI cases with rheumatic symptoms might be referred while in other clinics few cases might be referral. Referral bias, if prensent, might be based on differential self-referral by groups of women with SBI, by attorneys, or by the implants surgeons. A second weakness of this design that is not present when persons with SBI are folloed is that the risk of SBI is a function of age, since the implantation was not done or wa less common many years ago. To control for this potential problem we excluded patients over the age of 75 years. This studies cited above. Second, it has the abality to identify cases of FM, something that cannot be done by selp-report or case review. This ability is important because FM closely resembles SIAS and mught even be the same illness.

In agreement whth all previous studies, we did not find a significant association between RA and breast implantation, regardless of the type (OR 1.66, CI 0.33, 8.23. p=0.538). for RA compared to the combined control groups. Nor did we find an association between SBI and FM (OR 1,52, CI 0.31, 7.56, p=0610) for OA and community controls and (OR 1.22, CI 0.30, 4.89, p=0.781) for all subject controls when only SBI that occured after FM onset were considered. Of interest, however, was that when all implants were considered, regardless of their timing in relatin to disease onset, the OR and p value were 2.45 (95% CI 0.86, 7.03).p=0.095, (90% CI 1.01,5.94). These data suggest that SBI is not causally related to FM development, but that there might be an overall non-causal association with FM, an association that is significant at the 9.5% level rather than the conventional 5% level.

It is possible that psychosocial features noted in patients with FM and in persons with implants could be the common denominator. IF this were true then one could suggest that perhaps it is the characteristics of FM that lead to implants rather than implants leading to FM. Our data cannot answer this question. But it may be that furure prospective studies of implantation can provide more details of this interesting hypothesis.

References: Wolfe and Anderson: Silicone implants FM.RA.; p. 2025 at 2028

The Journal of Rheumatology 1999; 26:9p.