Immune responses to silicone gel breast implants

1. Introduction
In addressing the key question - Do silicone gel breast implants lead to the development of an autoimmune syndrome? - members of the IRG have considered evidence from several medical/scientific disciplines. In this section, the immunological evidence is discussed.

The Medical Devices Agency, and its predecessor, have already published two previous detailed reports on silicone gel breast implants (1, 2). The IRG considers these to be models of what such reports should be, in the comprehensiveness of the literature reviewed, the critical rigour with which this evidence was analysed, and the balance and fairness of the conclusions reached. For this reason, although papers published before December, 1994 are mentioned in the present review, the focus of attention is on newer information that has since emerged.

2. Antibody responses
Three principal questions need to be separated:

a)    Do the silicones used in breast implants provoke antibody responses against silicones or their biological breakdown products?

b)    Do silicones have an effect leading to antibody responses to body substances adsorbed to silicones?

c)    Do silicones cause inflammatory reactions that indirectly provoke immune responses to the recipient's own tissues?

A further question in relation to antibody responses has been recently raised:

d)    Do siloxane polymers provoke antibody responses to partially polymerised acrylamide?

---

a) Do the silicones used in breast implants provoke antibody responses against silicones or their biological breakdown products?

Theoretical aspects
Foreign (exogenous) substances, or antigens, that induce serum antibody responses are first bound to the surface of potential antibody-producing (B) lymphocytes by specific immunoglobulin receptors projecting from the cell surface of the B lymphocyte (antigen capture). The antigenic molecule then becomes engulfed by the lymphocyte's membrane, passing into specialised intracellular compartments in which the molecule is enzymatically digested and the resulting molecular fragments are loaded into the antigen-binding groove of the class II HLA molecules expressed by the lymphocyte (antigen processing). The molecular fragments bound to the HLA molecules are transported to the surface of the B lymphocyte where they are displayed. Subsequently a complex chain of events is set in motion leading to further differentiation of the antigen stimulated B lymphocyte and the formation and secretion of antibody.

In the case of IgG antibodies, an interaction between the B lymphocyte and a Thymus-Derived (T) lymphocyte is necessary. The interacting T lymphocyte makes use of its antigen-specific receptors to recognise and bind to the HLA-peptide complex expressed on the surface of the B lymphocyte. Other complementary cell surface molecules on the B and T lymphocytes add to the binding forces between the cells, and transduce signals across the cell membranes. These signals lead the B lymphocyte to undergo clonal expansion, and differentiation into plasma cells that secrete soluble antibody.

Siloxane polymers used in breast implants are both hydrophobic (insoluble) and stable substances. In section 3 (dealing with the chemistry of siloxanes) of the Medical Devices Directorate's report (MDD/92/42) published in February 1993, it is stated "The high bond strength of the inorganic silicon-oxygen backbone gives polysiloxane materials high thermal and hydrolytic stability and resistance to ultraviolet degradation, oxidation and ozone attack." Given these properties, it is difficult to see how highly insoluble polysiloxanes could bind to the IgM receptors at the surface of B lymphocytes in the first process of antigen capture. It would be necessary to invoke in vivo molecular degradation of the polysiloxane chain to a soluble fragment to allow antigen capture to take place. Even if fragments of polysiloxane chains were captured and gained entry to the intracytoplasmic compartment of B lymphocytes in which antigen processing takes place, given their chemical properties, it is unlikely that they will be susceptible to the enzymatic hydrolysis of antigen processing. Furthermore, it also seems unlikely that small fragments of the polysiloxane chains would have the necessary affinity to be successfully held in the antigen binding groove of class II HLA molecules. From what is known of the physiology of antibody production, the formation of specific IgG antibodies in response to exposure to polysiloxanes is contrary to what would be expected. A caveat to this conclusion is that IgG antibodies formed in response to another antigen might show some degree of cross reactive affinity for a polysiloxane.

IgM antibodies, because of their structure, have multiple antigen-binding sites. This property allows them to form weak, low-affinity attachments to a wide variety of molecules, particularly those molecules that are composed of repeating units. This characteristic, plus the fact that interaction with T lymphocytes, as described above, is not necessary for the induction of soluble IgM, raises the possibility that some individuals may have IgM antibodies in their sera induced by a third party antigen but which have low-affinity binding constants for polysiloxanes.

Experimental evidence
There are rather few studies in the literature designed to investigate directly whether siloxanes can induce the production of anti-siloxane antibodies in experimental animals.

Nosanchuck (1968; ref. 3 & cited in Appendix 3 of ref.1, the MDD report MDD/92/42) failed to find evidence of antibodies in guinea pigs following subcutaneous injection of silicone fluid mixed with Freund's complete adjuvant. The immunisation schedule was a potent one, likely to provoke responses against other types of antigens. Samples of sera were taken before and 5, 10, & 15 weeks after immunisation. Sera were tested for by two in vivo methods, capable of considerable sensitivity, passive cutaneous anaphylaxis, and immediate cutaneous hypersensitivity. One in vitro method (Ouchterlony gel diffusion) capable of moderate sensitivity was also used.

Naim & van Oss (1992; ref. 4 & cited in ref. 2, the Medical Devices Agency Report of December 1994) also failed to detect antibodies in the sera of rabbits immunised with silicone mixed with Freund's complete adjuvant.

This evidence from experimental animals is consistent with a conclusion that polysiloxanes do not themselves provoke antibody responses, and is in line with the theoretical expectation discussed above.

Whether or not immunologically significant degradation of polysiloxanes occurs in vivo and whether any putative biotransformed products are capable of inducing antibody responses remain unanswered questions. The MDA report of 1994 (2) discusses, on p.15, evidence on the migration and biotransformation of silicone in rats, detected by NMR. Although these studies suggested to the authors that there was in vivo degradation of silicone to hydroxylated silica and "high co-ordinated" silicone products, they provide no information on whether or not the break-down products can stimulate an antibody response. There has been no independent confirmation of these findings and further published work on some aspects of this subject has led to contradictory results (Macdonald et al., 1995).

Evidence from human subjects
As there are no well validated examples of a foreign substance being antigenic in humans but non-antigenic in other mammals, reports of human antibodies to polysiloxanes need to be scrutinised with care.

There are a small number of reports in the scientific literature of studies demonstrating antibodies in human sera to silicone elastomers.

1)    The report of Goldblum et al., 1992 (5), has been briefly mentioned in the MDA report of 1994. The paper describes 2 patients with silicone elastomer ventriculoperitoneal shunts in whom repeated revisions of the shunts were necessary due to inflammation adjacent to the shunts. None of the repeated attempts to isolate micro-organisms from the fluids exuded from the shunt area gave positive results. In the case of the first patient, after the 9th shunt replacement, "precautions were taken to cover the track with intact tissue; it has been well tolerated for longer than 3 years."

The second patient had the shunt inserted at approximately 9 months of age for management of hydrocephalus. The shunt was revised when the patient was 5 1/2 years. Two months later, the shunt was extruded. A few months later an abdominal pseudocyst was noted. Local reactions appeared along the shunt. As in the first patient, no micro-organisms were found either by culture or histology. The reactions resolved spontaneously in 10 days.

The sera of both the above patients and 5 control patients with shunts but no clinical reactions to them were investigated for the presence of antibodies against silastic tubing. The assay involved incubating 1 ml. of patients sera (dilutions from 1/10 - 1/1000) with 1 cm. of silastic tubing. The silastic tubing was then washed 3 times in buffered saline, and further incubated in rabbit anti-human IgG conjugated with peroxidase. After further washing, the silastic tubing samples were exposed to a conventional peroxidase conjugated antibody assay. IgG of sera of both patients with clinical reactions were found by the above assay to bind to the silastic tubing more strongly than did the IgG in sera from the 5 control patients.

Additional experiments claimed that Fab fragments derived from the above sera also demonstrated specific binding, but the details given do not allow one to evaluate the robustness of the data. Similarly, a methylsiloxane- albumin conjugate was prepared, and used to absorb the sera. Absorbed and unabsorbed sera were subsequently assayed for binding to silastic tubing, and the absorption was claimed to remove the antibodies to the silastic tubing. Again, insufficient details were given to allow a critical evaluation of the validity of this conclusion.

Clearly, both patients suffered inflammation in the tissues around the shunts, but the cause of that remains uncertain. The observation that both patients who initially appeared to be intolerant of their shunts later accepted them would be unexpected if the intolerance were due to an immune reaction.

The important question of whether the patients sera contained antibodies specific for silicone elastomer cannot be clearly answered from the data reported. Taken at face value, the ELISA appears to demonstrate that IgG binds to the silastic tubing, but the evidence that this is specific binding through its antigen-binding sites is inadequate. It remains possible that the binding is non-specific (in the immunological sense) and is due to the physico-chemical properties of the polysiloxane material. More rigorously controlled absorption and elution experiments would be needed to resolve this question. An additional confounding feature was that one of the patients with inflammatory reactions around the shunt was demonstrated to have raised IgG serum levels; hypergammaglobulinaemia is one possible explanation for an increase in non-specific binding of IgG to the silastic tubing.

In 1995, the same research group (6) showed that absorption of IgG to silicone was strongly influenced by the serum albumin levels. This observation is an additional reason for concluding that Goldblum and co-workers' original interpretation of their data is not sound. Further studies by independent workers (22) did not confirm the observation of anti-silicone antibodies and noted the unreliability of the test system.

2)    A second paper that was also discussed in the 1994 MDA report is that of Wolf et al. (7). It describes findings made with sera from patients with ruptured/leaking silicone breast implants (n=19), patients with intact implants (n=15), diabetic women exposed to silicone in syringe lubricant (n=10), and healthy women with no silicone breast implants (n=67). An ELISA method was developed that detected IgG binding to silicone coated polystyrene plates. In each assay, replicates of patients' sera, diluted 1: 1000, were mixed in the plate wells either with bovine serum albumin or with an albumin: silicone complex. The differences in IgG bound by these wells were interpreted as specific binding. The amount of IgG bound per well was assayed by a similar method to that used by Goldblum et al. (5).

The results of the assays showed that the greatest amounts of IgG binding to the silicone coated plates were found in sera from the patients with ruptured implants. Lower amounts were found in sera from patients with intact implants, and the lowest were found in diabetic patients and women without implants.

The authors interpret the data as indicating that the IgG binding is through its specific binding sites.

The significance of the above assay marketed by Emerald Biomedical Sciences, Inc. (Emerald), The Woodlands, Texas, has been challenged. An account of this is given in an article - Silicone Breast Implants: Why Has Science Been Ignored? (8), published in 1997 by the American Council on Science and Health.

In this article, it is pointed out that Dr. Noel Rose, of the Johns Hopkins University, Baltimore, an internationally recognised immunologist of great experience, was unable to confirm the validity of the "Emerald" test. Emerald continued to claim that Dr. Rose's findings were confirmatory, until the Johns Hopkins General Counsel wrote to them.

Conclusions
It is concluded that claims that silicones themselves provoke antibody responses in vivo are not sustained by present evidence reported in the scientific literature.

Whether or not biological breakdown products of siloxane polymers are formed in the body and can stimulate an antibody response remains an incompletely investigated question, although there is no published evidence for such effects. Furthermore, the IRG heard additional evidence that the breakdown of silicones within the body is unlikely to occur since the conditions needed to produce such breakdown outside the body were extreme, requiring temperatures of approximately 500 degrees centigrade and considerable acidity. No published evidence has been found establishing clearly that breakdown of silicones occurs within the body resulting in transformation products that are capable of inducing an immune response.

---

b) Do silicones have an adjuvant effect leading to antibody responses to other body substances adsorbed to silicones?

Early experimental studies reviewed in reference 1 have documented that proteins adsorbed to silicones can provoke an antibody response, and that silicones either have a rather weak or no adjuvant action. The proteins against which antibodies are formed can either be foreign substances injected with the siloxanes, or they may be self-proteins that adhere to the implanted siloxanes.

Recent Experimental animal studies
Studies have been reported on the effect of silicone elastomers and gel on two mouse models of autoimmune disease (9, 10).

In the first model of Freund's Adjuvant induced arthritis, DBA/1 mice were implanted with silicone elastomers and gel, and subsequently immunised with type II collagen and Freund's adjuvant. Control mice were immunised with collagen or adjuvant alone, and there were sham implanted control mice as well. The mice were assessed for arthritis over 12 weeks. There was no increase in severity or incidence of arthritis in the silicone treated mice. However serological studies suggested that silicone treated mice developed autoantibodies against silicone bound proteins. But as there was no clinical difference between the silicone - treated mice and the sham implanted controls, the antibodies against silicone bound proteins appeared not to be of pathological significance.

In the second experimental model, a similar experiment was carried out in MRL/lpr lpr mice, a strain that spontaneously develops the autoimmune disease of systemic lupus erythematosus (SLE). Implantation of silicone elastomers & gel did not lead to any increase in the severity or incidence of SLE. However, as in the DBA/1 mice, the silicone treated mice developed autoantibodies against silicone bound proteins.

These studies have confirmed that mice develop antibodies against self-proteins that bind to silicones. However in the experimental models studied to date, the presence of these antibodies did not affect the severity, incidence, or rate of progress of the autoimmune disease.

Clinical Experience
The work of Kossovsky et al. (11) was also discussed in the MDA report of 1994 (2), but further information has been gained since then. Briefly, Kossovsky et al. had claimed that a sub-population of women with silicone breast implants had developed antibodies to fibronectin and laminin, but these antibodies were not present in the sera of healthy women or women with various rheumatic diseases. Reviewing the data, the 1994 MDA report concluded "It is .... not possible to draw any definitive conclusions from this work ...".

In the review by the American Council on Science and Health, published in 1997 (8), there is an account of the history of Dr. Kossovsky's test. The test, named "Detecsil", was marketed as being able to show whether or not an individual had developed an immune response to silicone-associated proteins. It was claimed by Kossovsky that the validity of the test had been independently confirmed by the Scripps Clinic and Research Foundation, La Jolla, CA. However, the Scripps experts found no differences in antibodies of women with implants who had autoimmune diseases and those without implants who had autoimmune diseases. Thus the claims of Dr. Kossovsky and co-workers were not confirmed by an experienced independent laboratory. Despite this, the sales of Detecsil with the claim of independent confirmation continued, until the US Food and Drug Administration (FDA) intervened, forbidding its sale.

Conclusions
Although experimental animal work indicates that self (and foreign) proteins adsorbed to silicone polymers can induce an antibody response, there is no evidence showing that the response causes any tissue damage.

The claim that the "Detecsil" test could identify individuals amongst women with silicone gel breast implants who have developed immune responses to silicone-associated proteins and who were, as a consequence, at special risk of developing autoimmune diseases has not been upheld by an experienced, independent laboratory.

---

c) Do silicone polymers cause inflammatory reactions that indirectly provoke immune responses to the recipient's own tissues?

This issue has been extensively discussed in the 1994 MDA report (2), which concluded that the balance of evidence was tilting against the hypothesised link between silicone breast implants and connective tissue disease. However, since then, further studies have been published (12, 13).

1) Rowley MJ et al. Antibodies to collagen: comparative epitope mapping in women with silicone breast implants, systemic lupus erythematosus and rheumatoid arthritis. J. Autoimmunity 7: 775-789, 1994.
This reports studies on the autoantibodies to collagen found in the sera of 70 women with silicone breast implants who did not suffer from "a specific autoimmune disease, according to the criteria of the American College of Rheumatology...". The patients were self-recruited women who were worried about potential health risks due to their implants. Only 5 of the 70 were without symptoms (fatigue, myalgia, arthralgia, miscellaneous), and 49 (70%) of the patients had had moderate to severe capsular contracture of their implants. Twenty six (37%) had suffered one or more ruptured implants.

ELISA tests were used to identify the women with autoantibodies to native and denatured collagen types I & II.

Twenty nine (41%) of the 70 patients with implants had anti-collagen antibodies. These women did not differ significantly from the anti-collagen antibody negative women with implants with respect to length of time since implant inserted, proportion of the groups with capsular contracture, or a history of rupture. There were no significant differences between the groups for frequency of positive ANA. None of the women had rheumatoid factor.

The sera of 82 women with SLE, 94 with RA, and 133 controls were also studied for the presence of autoantibodies to collagen types I & II. The specificities of positive sera were then further investigated by examining their reactivity (Western blotting) with cyanogen bromide cut fragments of the collagens. The patterns of reactivity against fragments of collagen type I appeared to differ; sera from 3 of 18 (17%) silicone implant patients reacted with fragment CB8, and 9 (50%) reacted with fragment CB6. In contrast, sera from 17 of 18 (94%) patients with SLE reacted with CB8, and 2 (11%) reacted with CB6.

The authors discuss the possibility that there is a slow degradation of silicones in vivo leading to the production of silica which in turn, through its property of causing inflammatory reactions, may cause a breakdown in self-tolerance. They speculate that this may give rise to anti-type I collagen antibodies. The authors point to a reported association between exposure to silica dust and scleroderma (Haustein et al. 1990), and also that anti-type I collagen antibodies have been found in patients with pulmonary silicosis (Nagaoka et al. 1993). From this they conclude that silicone or its biodegradation products might act as adjuvants in situ to enhance the immunogenicity of type I collagen, or protein-silicone conjugates.

Commentary
The Review group considered the above hypothesis with great care. It is their judgement that the presently available evidence is not sufficient to sustain it. Further evidence would be needed to support this hypothesis.

2) Multiple autoantibodies in patients with silicone breast implants. E.Bar-Meir, S.S.Teuber, H.C.Lin, et al. J. Autoimmunity (1995) 8, 267-277 This reports the frequencies of autoantibodies against 20 autoantigens found in the sera of 116 women with silicone gel breast implants and 134 control women, without breast implants. A significantly increased frequency of autoantibodies against 15 of the 20 autoantigens was found in the sera of the women with implants when compared with the frequencies of these same autoantibodies in the control group of women.

Commentary
Although the matter is not discussed in the paper, it appears that the group of women with silicone gel breast implants were selected. Figure la of the paper shows a "pie chart" giving the proportions of the group with implants who had local complications involving the implants, e.g. rupture, contracture, hardening, changes that led to the implant being removed or replaced. The pie chart shows that only 4% of the women in this group were asymptomatic in this respect.

The question of how the patients were recruited (not reported on) is clearly very important. It seems probable that the group was not representative of an unselected group of women with implants. This conclusion may explain why the authors' results differ from the conclusions of others, as they acknowledge in the discussion.

In the absence of clear information about how the patients were recruited to this study, firm conclusions cannot be reached on whether or not an unselected population of women with silicone gel breast implants have a normal or increased frequency of autoantibodies. But it raises a further question - do women with severe local complications involving their implants develop a higher frequency of autoantibodies than women without local complications?

Conclusions
The question Do siloxane polymers cause inflammatory reactions that indirectly provoke immune responses to the recipient's own tissues? remains to some extent incompletely resolved. While there is no unambiguous, published evidence showing that the majority of recipients of silicone gel breast implants do develop immune responses to their own tissues as a consequence of the implant rather than through other factors, the possibility that there is a sub-group of the recipients who do so has not been formally disproved. None of the studies claiming an increased frequency of auto-antibodies in women with silicone gel breast implants has been adequate to permit this conclusion. Further studies would be necessary to identify:-

a)    any sub-group of recipients at risk;

b)    the auto-antibodies provoked and their target antigen.

One suggestion that has been made is that the age of the implant is a relevant factor, and that aged implants are more liable to leak or rupture. Thus women with aged implants may be a sub-group at risk.

A real difficulty in addressing this problem is the lack of quantitative information about the incidence, amount, and rate at which silicone polymers escape from the different types of implants, particularly in the case of implants inserted more than 7 - 10 years previously. This information has potential importance because siloxane polymers injected directly into tissue can produce an inflammatory reaction. This point will be taken up again elsewhere in this report.

---

d) Do siloxane polymers provoke antibody responses to partially polymerised acrylamide?

Tenenbaum S. A, et al. 1997. Use of antipolymer antibody assay in recipients of silicone breast implants. The Lancet 349: 449-454, and the ensuing correspondence also published in The Lancet.
This publication (14) describes the finding of antibodies specific for partially polymerised polyacrylamide in the sera of patients with silicone gel breast implants. Sera were investigated from patients with silicone gel breast implants, from control patients suffering autoimmune diseases, and from healthy female volunteers. In the above publication the antibody detected is abbreviated to APA, short for anti-polymer antibody, but details of its binding specificity are not given in the paper. For fuller details it is necessary to consult an abstract (15) and U.S. Patent No. 5,620,859 filed on April 15th, 1997 by Robert F. Garry: Scott A. Tenebaum, & Douglas R. Plymale.

The paper reports that sera from silicone gel breast implant patients with severe symptoms were positive for the antibody significantly more frequently than sera from implanted patients with mild or no symptoms, or the healthy female volunteer group. The study is described as a blinded one, in which numbered sera from unidentified patients & controls were sent to the laboratory of Dr. Tennenbaum (Tulane University School of Medicine, New Orleans, LA) for antibody testing. The patients & controls were recruited and assessed at the Arizona Rheumatology Centre, Phoenix, AZ.

The paper states that APA were found in 1 of 34 (3%) implanted patients with limited symptoms, 2 of 26 (8%) with mild symptoms, 7 of 16 (44%) with moderate symptoms, and 13 of 19 (68%) with advanced symptoms. In contrast, APA were found in 4 of 23 (17%) healthy control women without implants and 2 of 20 (10%) women without implants but with an autoimmune disease.

It was concluded that the APA assay can distinguish between breast implant recipients with limited/mild signs and symptoms and those with more severe ones. It was also concluded that it was of use in distinguishing implanted patients with severe signs and symptoms from patients with autoimmune diseases meeting criteria of the American College of Rheumatology.

Subsequent to the publication of this paper, there was an extensive correspondence published in The Lancet. Criticisms were levelled at:-

a)    The ascertainment of the patients, and the low participation rate of those approached. Only 15% of breast implant recipients were enrolled, throwing doubts on whether this group was representative of all recipients.

b)    The classification categories of the patients. There appeared to be little justification for classifying patients with poor functional status and mild or no symptoms in the mild or limited group of the study. Similarly, patients with moderate or severe symptoms, but with limited or mild functional impairment were classified as moderate or advanced for the purpose of the study.

c)    The fact that Tenenbaum and co-workers had previously reported (16) a high positivity rate for APA in patients with fibromyalgia. This finding was not referred to at all in the paper reporting findings in the SBI recipients.

d)    A letter from two correspondents described how serum samples from 4 of 5 females none of whom had breast implants, had been reported as positive for APA by a "Tulane researcher" (unidentified). Doubt was therefore thrown on the figure of 17% positive for APA in the group of healthy women without implants in the Lancet paper.

Dr. Tenenbaum responded to these criticisms by asserting that a different version of the assay was used in the study on fibromyalgia patients. The implanted patients with symptoms did not all have fibromyalgia, and therefore it could not be concluded that APA positivity was associated with fibromyalgia.

Commentary
The publication by Tenenbaum, et al. is not conclusive. The criticisms mentioned in the correspondence are legitimate. Only further studies can determine whether or not the findings reported are reproducible.

However, there is no reason to expect antibodies specific for partially polymerised acrylamide to be induced in women with silicone gel breast implants. The chemical structure of acrylamide is unrelated to dimethyl siloxanes and it would be necessary to postulate a strong cross-reactivity. Any such cross-reactivity should be readily demonstrable in vitro, but such cross-reactivity has not been reported. Tenenbaum & co-workers do not advance any immuno-physiological explanation for their unexpected observations, nor do they provide any immuno-histological evidence that the APA antibodies react with any human tissue. Given this situation, independent confirmation of the Tenenbaum report is essential before one could accept its conclusions.

3. Tlymphocyte activation
The sub-population of T lymphocytes known as T helper cells plays a crucial role in immune responses. T lymphocytes not only respond to appropriately presented antigenic material themselves, but also interact with B lymphocytes, enabling the latter to differentiate into plasma cells that synthesise soluble antibodies. For this reason, the recent claims by Smalley, Shanklin et al. (17, 18, 23) that T lymphocytes of patients with silicone gel breast implants are activated when cultured in the presence of silica have considerable potential significance in relation to the question: Does exposure to silicone lead to the provocation of a lymphocytic response?

Shanklin speculates that there is breakdown of silicones to silica in vivo and that the silica induces an autoimmune process. Unfortunately, virtually all the important data in the publications used to support this view are reported in the form of a ratio, known as the stimulation index, rather than as direct measurements of the uptake of labelled nucleotides by the T lymphocytes. Smalley et al. (17), for example, report an adaptation of a conventional lymphocyte stimulation test, using colloidal silica (silicon dioxide) as the test antigen. An initial set of stimulation assays was carried out on lymphocytes harvested from 70 women with breast implants, 76 normal controls without breast implants or any symptoms, and from 18 patients with rheumatic diseases and no breast implants. A follow-up study was then carried out on larger groups which included 220 normal controls without implants, 942 patients with implants and who had symptoms that have been attributed to breast implants, and 34 patients with implants but who did not have symptoms.

The authors report most of their data as stimulation indices (SI). They state that for the initial study the mean SI of T lymphocytes from implant patients (195 +/- 19.3) was approximately 18-fold higher than that of T cells from normal controls (p < 0.00001). The mean SI of T cells taken from rheumatic patients was similar to that of the controls. In the follow-up study, the mean SI of the controls was 10.0 +/- 0.41. The authors claim that over 90% of symptomatic patients with implants had SI significantly above those of the normal controls.

As indicated above, the question of whether T cells of women with silicone gel breast implants are activated by siloxane polymers or their breakdown products is a very important one. In these circumstances, it is essential that full details are given, both of the methods, and the results. It is unfortunate that the report is deficient in both respects. How these experiments were carried out, the numbers of experiments performed, and the reproducibility of the assay are all aspects that are inadequately described, even in additional evidence provided specifically for the Independent Review Group (23).

Most important of all is the deficiency in how the results are reported. In the paper, virtually all the results are given as stimulation indices (SI), the counts per minute (cpm) of thymidine taken up by cells in the experimental cultures divided by the corresponding cpm of the control cultures. The effect of presenting the data in this way is that the actual figures of thymidine incorporation for experimental and control cultures are concealed from the reader. In these circumstances, it is not possible to judge whether or not a high SI is due to (unexplained) low counts in the control cultures, which may occur in technically unsatisfactory assays. Thus the significance of the high SI figures reported remains quite uncertain.

There is one sentence in the results section that does mention directly the cpm of thymidine incorporation. It states that "The raw counts for implant patients after stimulation with silicon dioxide varied from 500 to 3912....". Such incorporation figures are so low that strong doubts must exist as to their significance. Similar criticisms are made by Marcus (19) when reviewing the experiments of Ojo-Amaize et al. (20) in which lymphocyte proliferation in response to silicone-containing compounds was claimed.

Smalley et al. (18) report a similar study but in this case carried out to determine the responsiveness to silica of T lymphocytes harvested from children of women with silicone gel breast implants. The authors claim significant activation by silica of the lymphocytes from these children. However, the criticisms given above also apply to this paper.

A recent study by Ellis et al. (21) reports proliferation responses of lymphocytes from women with silicone gel breast implants against several autoantigens and compares these with the responses of cells from a suitable control group of healthy women. Significantly higher proliferation occurred in response to Collagen types I and III, and to fibrinogen and fibrin. This work needs to be independently confirmed and the relationship of the proliferative responses to patient reported symptoms also needs to be investigated.

Conclusions
The claim that T lymphocytes from women with silicone gel breast implants and the offspring of recipients of silicone gel breast implants are activated in vitro by culturing the cells with colloidal silica is open to a number of technical criticisms. The Review Group concluded that, with the exception of Ellis et al.(21), the above reports were so deficient that valid conclusions could not be drawn from them. Given the way that the data were reported, which would not be acceptable to the main immunological journals, a critical assessment of these studies was not possible. In addition the reports contain insufficient details of methodology, reproducibility and the numbers of experiments performed.

The issue is an important one, and should be addressed in further experiments by one or more independent laboratories experienced in the field of T lymphocyte activation. Given the biological significance of the issue, it is essential that one or more independent laboratories, experienced in T cell culture technology should be commissioned to re-examine the claim before any conclusions can be drawn. Furthermore, attempts to clone any specifically reactive T cells should be made.

4. Associations of autoimmune diseases with particular HLA variants
T lymphocytes, which have several critical functions in immune reactions, recognise foreign antigens and auto-antigens as molecular fragments that are bound in a specialised "pocket" of HLA proteins. Because there is immense variation of the HLA genes, the corresponding HLA proteins also have great variation between different individuals. It has been shown that most of the variations reside in amino acids lining the "pocket" of the HLA molecule and affect which molecular fragments can be held in the "pocket". This is one and perhaps the major reason why the ability of an individual to react to a given antigenic molecular fragment depends upon the HLA gene variants that (s)he inherits. It is also a widely favoured explanation for the well established observation that the distribution of HLA gene variants is abnormal in patient groups in all human autoimmune diseases. This phenomenon has its counterpart in all animal autoimmune diseases. Indeed, if the distribution of HLA gene variants in a group of subjects with a disease of unknown pathogenesis is the same as that of the normal population of the same ethnicity, then the disease almost certainly is not due to autoimmune pathogenesis.

There is one detailed publication on HLA variants in women with breast implants (24). In this study, 199 women classified into one of four groups were HLA typed. There were 77 women with breast implants who suffered one or more of the following symptoms - breast pain, chronic fatigue, generalised pain of the torso, myalgia, arthralgia-persisting for at least 4 months. Thirty seven women with breast implants but without these persistent symptoms were also studied. Lastly there were 54 women without breast implants or symptoms, and a group of 31 women without breast implants, but who had symptoms like those in the first group. The authors found what they claim to be a statistically significant increase in the frequency of HLA-DR53 in the groups of women with symptoms, with or without SBI. In addition, the paper reports a statistically significant association between the presence of HLA-DR53 and a serum antibodies against B lymphocytes, indicating an autoimmune response.

Commentary
In the judgement of the Independent Review Group, the conclusion of a statistically significant association between HLA-DR53 and persistent symptoms is not justified by the data. The authors failed to apply the appropriate correction factor normally used when multiple comparisons are made in this type of study. The authors' argument for not applying the correction factor - that linkage disequilibrium of HLA region gene variants makes the frequencies of many of the variables interdependent - is not a valid one. If a proper correction factor is applied to the data, no statistically significant difference is observed.

It is also relevant that the frequencies of HLA-DR53 in the group of women with SBI and symptoms do not differ significantly from those in the group of asymptomatic women without SBI, or the group of women with symptoms but no implants. The only substantial differences observed was a low frequency of HLA-DR53 in the group of 37 women without persistent symptoms but who had received implants. Given the inadequate statistical rigour of this study, further investigations would be needed before any valid conclusions could be reached.

Conclusions
The Review Group interprets the absence of any confirmed HLA association with a defined symptom complex resulting from silicone breast implantation as evidence against the hypothesis that silicone breast implants induce autoimmune tissue damage.

5. General conclusions
1.    Claims that silicones themselves provoke antibody responses in vivo are not sustained by any reproducible studies in the scientific literature.

2.    Whether or not the biological break-down products of silicone gel breast implants induce antibody responses has not been adequately investigated; consequently, the question remains unanswered, although the breakdown of silicones in vivo is unlikely.

3.    Although experimental animal work shows that self and foreign proteins adsorbed to silicones may induce an antibody response to the protein, there is no published evidence demonstrating that the response causes tissue damage.

4.    The question - Do silicones cause inflammatory reactions that indirectly provoke autoimmune responses? - remains incompletely resolved. At present none of the studies addressing this question has been adequate to support the conclusion that an increased proportion of recipients of silicone gel breast implants develop autoimmune responses to their own tissues, but the possibility that there is a sub-group of recipients who do so has not been formally disproved.

5.    There is a lack of quantitative information on the incidence, amount, and rate at which silicones escape from different types of implants, particularly in the case of implants inserted more than 7 years previously. This information is of relevance to questions concerning its possible pathogenic role.

6.    The claim of Tenenbaum et al. that there is an increased frequency of antibodies to partially polymerised polyacrylamide in the sera of patients with silicone gel breast implants is open to a number of technical criticisms. This claim needs verification by one or more independent laboratories.

7.    The claim of Smalley et al. that T lymphocytes harvested from patients with silicone gel breast implants are activated in vitro by colloidal silica dioxide and display immunological memory to silica dioxide is open to technical criticism. This claim needs verification by one or more independent laboratories.

8.    There are no confirmed published observations demonstrating a statistically significant association of a particular HLA phenotype with the "atypical autoimmune syndrome" that has been claimed to be due to silicone breast implants. The absence of such an association is evidence against the concept of an autoimmune pathogenesis.

6. Recommendations
1.    Research should be commissioned in independent laboratories with appropriate expertise to verify the conclusions of Tenenbaum et al. In addition, coded samples of sera from recipients of silicone gel breast implants and controls, prepared in the U.K. should be sent for testing at Dr. Tenenbaum's laboratory. A small group should be formed to plan and supervise the progress of the commissioned research, and to analyse the data resulting from the antibody assays at Dr. Tenenbaum's laboratory.

2.    Research should be commissioned in independent laboratories with appropriate expertise to verify the conclusions of Smalley et al. The same small group proposed above in paragraph 1 should also plan and supervise the progress of this research.

3.    Research should be commissioned to investigate quantitatively the incidence, amount, and rate at which siloxane polymers escape in vivo from different types of implants, particularly in the case of implants inserted more than 7 years previously.

4.    The IRG should meet to consider the results of the commissioned research as soon as is feasible, and make further recommendations as necessary.

7. References
Note: Only primary references are listed here.

1.    Tinkler JJB, Campbell HJ, Senior JM, Ludgate SM. Evidence for an association between the implantation of silicones and connective tissue disease. Medical Devices Directorate Report No. MDD/92/42; 1-65. 1993

2.    Gott DM, Tinkler JJB. Silicone implants and connective tissue disease. Medical Devices Agency Report; 1-62. 1994

3.    Nosanchuk JS. Injected dimethylpolysiloxane fluid: a study of antibody and histologic response. Plastic & Reconstructive Surgery; 42:562-566. 1968

4.    Naim JO, van Oss CJ. The effect of hydrophilicity-hydrophobicity and solubility on the immunogenicity of some natural and synthetic polymers. Immunol. Invest.: 21: 649-662. 1992

5.    Goldblum RM, Pelley RP, O.' Donell AA, Pyron D, Heggers JP. Antibodies to silicone elastomers and reactions to ventriculoperitoneal shunts. Lancet: 340: 510-513. 1992

6.    Goldblum RM, Pyron D, Shenoy M. Modulation of IgG binding to silicone by human serum albumin. FASEB J; 9: A1029 (abstract 5967). 1995

7.    Wolf LE, Lappe M, Peterson RD, Ezrailson EG. Human immune response to polydimethylsiloxane (silicone): screening studies in a breast implant population. FASEB J; 7: 1265-1268. 1993

8.    American Council on Science and Health. Silicone breast implants: why has science been ignored? ACSH, New York 1996

9.    Schaefer CJ, Knapp T. Andrews A, Delustro F. Wooley PH. Silicone implantation induces antibodies to silicone bound proteins during murine type II collagen-induced arthritis. Arthritis and Rheum. 39 (9, suppl.) Abstract 1996

10.    Schaefer CJ, Wooley PH. The influence of silicone implantation on murine lupus in mrl mice. Arthritis and Rheum. 39 (9, suppl.) Abstract 1996

11.    Kossovsky N. Zeidler M, Chun G. Papasian N. Nguyen A, Rajguru S. Stassi J. Gelman A, Sponsler E. Surface dependent antigens identified by high binding avidity of serum antibodies in a subpopulation of patients with breast prostheses. J.Appl. Biomat.; 4:281-288. 1993

12.    Rowley MJ, Cook AD, Teuber SS, Gershwin ME. Antibodies to collagen: Comparative epitope mapping in women with silicone breast implants, systemic lupus erythematosus and rheumatoid arthritis. J. Autoimmunity; 7: 775-789. 1994

13.    Bar-Meir E, Teuber SS, Lin HC, Alosacie I, Goddard G. Terybery J. Barka N. Shen B. Peter JB, Blank M, Gershwin ME, Shoenfeld Y. Multiple autoantibodies in patients with silicone breast implants. J. Autoimmunity; 8: 267-277. 1995.

14.    Tenenbaum SA, Rice JC, Espinoza LR, Cuéllar ML, Plymale DR, Sander DM, Williamson ll, Haislip AM, Gluck OS, Tesser JRP, Nogy L, Stribrny KM, Bevan JA, Garry RF. Use of antipolymer antibody assay in recipients of silicone breast implants. Lancet; 349: 449-454. 1997

15.    Tenenbaum SA, Cuéllar ML, Citera G. Silveira LH, Garry RF, Espinoza LR. Identification of a novel antigen recognised by silicone breast implant recipients. Arthritis and Rheum.: 36 (9): Abstract 123. 1993

16.    Gluck OS, Tesser JRP, Tenenbaum SA, et al. Development of a laboratory marker for fibromyalgia. Arthritis and Rheum; 39 (S90): 381 abstract. 1996.

17.    Smalley DL, Shanklin DR, Hall MF, Stevens MV, Hanissian A. Immunologic stimulation of T lymphocytes by silica after use of silicone mammary implants. FASEB J: 9: 424-427. 1995

18.    Smalley DL, Levine JJ, Shanklin DR, Hall MF, Stevens MV. Lymphocyte response to silica among offspring of silicone breast implant recipients. Immunobiol.; 196: 567-574. 1996/97

19.    Marcus DM. An analytical review of silicone immunology. Arthritis and Rheum.; 39 (10):1619-1626. 1996

20.    Ojo-Amaize EA, Conte V, Lin H. Brucker RF, Agopian MS, Peter JB. Silicone-specific blood lymphocyte response in women with silicone breast implants. Clin. Diagn. Lab. Immunol.; 1: 689-695. 1994

21.    Ellis, T. M; Hardt, N. S; Campbell, L; Piacentini, D. A; and Atkinson, M. Cellular immune reactivities in women with silicone breast implants: A preliminary investigation. Annals of allergy, asthma and immunology; 79: 151-54. 1997

22.    Rohrich RJ, Hollier LH, Robinson JB. Determining the safety of the silicone envelope: In search of a silicone antibody. Plastic & Reconstr. Surg.; 98: 455-458. 1996

23.    Shanklin DR. Personal communications to the Independent Review Group. 1997

24.    Young VI., Nemecek JR, Schwartz BD, Phelan DI., Schorr MW. HLA typing in women with breast implants. Plastic & Reconstr. Surg. 96; 1497-1519. 1995