Immunology of Silicone
Silicone breast implants are associated with significant local complications. Some have studied whether these devices are also associated with systemic morbidity. Because experimental and clinical immune reactions to silicone have been said to be involved in such an association, the committee undertook an examination of the evidence for these reported reactions. An understanding of the basic immunology is also important in assessing the biologic plausibility of some reported clinical findings and some suggested associations, such as autoimmune or connective tissue disease or novel silicone associated systemic syndromes. In this chapter, the committee reviews and discusses reports from the peer-reviewed scientific literature of both animal and human immune responses, or absence of immune responses, to silicone in various forms. Included in this discussion is a brief description of a conceptual approach to investigating the immune response to silicone. Some reports on immune effects were reviewed but are not cited in this chapter. They can be found in the reference list of this report.
Immune Response to Silicone in Experimental Animals
Studying effects on the immune response in various experimental animals is an approach to investigating a substance that is often employed by basic and clinical scientists. Currently available experimental data indicate that silicone gel (or some higher molecular weight silicone
oils) can act as a weak adjuvant capable of enhancing antigen-specific immune reactions. Mice or rats exposed to various antigens emulsified in silicone gel produce a greater antibody reaction than if antigen alone is given (Hill et al., 1996; Naim et al., 1995a,b, 1996, 1997a; Nicholson et al., 1996). However, silicone gel was found to have weaker adjuvant activity than a widely used reference adjuvant, complete Freund's adjuvant. Furthermore, enhancement of immune responsiveness was observed only when silicone and an antigen were injected together as an emulsion in the same site. Injection of an antigen in one site and silicone gel (implant) in a different site did not augment the immune response (Bradley et al., 1994a,b; Klykken and White, 1996). Other animal studies of components of the immune response from several groups have shown that parenteral administration of silicone gel to animals induces a time-dependent decrease in natural killer (NK) cell activity (Bradley et al., 1994a,b; Wilson and Munson, 1996). The NK-cell system is an important part of the natural immune system that is believed to contribute to the initial response to infections as well to controlling the emergence of tumors. Significant reductions in tumor control or response to infections were not observed after silicone induced reductions in NK-cell activity in these studies.
Silicone gels given alone have been reported to produce disease in two different animal models. Silicone caused an acute arthritis when injected directly into the joints of one particular strain of rats, but no arthritis occurred if the gel was injected distal to the joints. Arthritis was not observed in joints that were not injected directly, and when joints were injected, local inflammation was observed, but not distant systemic effects (Yoshino, 1994). In a second animal model, Potter and colleagues (Potter and Morrison, 1996; Potter et al., 1994) induced monoclonal immunoglobulin producing B-cell tumors (plasmacytomas) in BALB/c mice following intraperitoneal injection of silicone gel. This is not a simple, silicone-specific disease model, however. These mice appear to be genetically predisposed to develop plasmacytomas on intraperitoneal exposure to other triggering agents. It has not been suggested that this model has implications for the induction of cancer in experimental animals or in humans.
Classical adjuvant arthritis does not appear to be inducible by silicones in rats or mice (Naim et al., 1995a,b; Schaefer et al., 1997), although both silicone gel and silicone oils can replace incomplete Freund's adjuvant in inducing collagen-initiated arthritis in DA rats if the collagen is mixed with the silicone (Naim et al., 1995b). Since the adjuvant arthritis model is a well-established, intensively studied animal model for inflammatory arthritis, the failure of silicones to activate similar clinical and pathologic features in experimental animals is an important finding. Likewise, exposure of 18,000 humans to mineral oil adjuvant was not followed
by excess connective tissue disease over 16-18 years of follow-up compared to 22,000 controls (Beebe et al., 1972, for a discussion of connective tissue or other systemic disease and silicone breast implants, see Chapter 8). The committee concludes there is no evidence for any human adjuvant disease, as asserted by some investigators (Miyoshi et al., 1964, 1973). For review and critique of additional animal studies, see Marcus (1996).
Possible Relationship to Autoimmune Disorders
Some investigators have suggested that current experimental animal data could support an association of silicone with immune effects in humans. Several ways in which silicone might activate an autoimmune disease in silicone breast implant recipients have been proposed and explored. First, since the major histocompatability (MHC) locus is critical to the way elements of the immune system sort out, recognize, and process foreign materials and antigens, a subset of women with implants could have a special human leukocyte antigen (HLA) that makes them particularly likely to process certain antigenic moieties in silicone gels in ways that activate T-cells to induce cell-mediated immune reactivity and initiate an inflammatory reaction. Secondly, cells of the immune system might be directly activated in patients with silicone breast implants. In other disorders, immune activation is usually indicated by obvious inflammatory cell infiltrates and damage within affected tissues or by deposition of specific antibodies within such tissues. Extremely high local concentrations of cytokines either in the serum or within local inflammatory reactions in involved tissues may also indicate immune activation. Thirdly, silicone breast implants might induce reactions to autologous or self-antigens. Such autoreactivity, if induced by breast implants, should be demonstrated by self T-cell reactivity or sensitization when T-cells are exposed to self-antigens or components of silicone gels.
In exploring whether silicone breast implants cause an autoimmune disorder in breast implant recipients, the committee concludes that it is important to determine if there is an abnormal immune response in these women that is directly caused by the implant. When this is examined, the immune function and responses in healthy women with breast implants should be compared to those of symptomatic women with implants, as well as comparing symptomatic women with breast implants to symptomatic women without silicone breast implants to determine whether a specific immune system abnormality can be identified that is associated with clinically recognizable symptoms in women with breast implants.
These determinations should be supported by addressing the following questions:
1. Is there an abnormality in natural immunity?
(A) Can any evidence be found for monocyte or non-specific T-cell activation as illustrated by accurate reproducible cytokine assays?
(B) Is there any evidence that components of silicone breast implants can act as a T-cell superantigen, or as a T-cell superantigen-like molecule that is capable of activating large numbers of T-cells with antigenic specificities, by binding shared T-cell receptor epitopes?
(C) Is there any evidence for an effect of silicone breast implants on NK-cell activity?
2. Is there an abnormality in the immune response?
(A) Do women with silicone breast implants and symptoms share a particular HLA haplotype profile?
(B) Can it be demonstrated that silicone-specific T-cells are present and have been activated in women with breast implants?
(C) Can silicone-specific B-cell reactivity be demonstrated in women with breast implants?
(D) Is it possible to demonstrate T or B cell autoreactivity in women with breast implants?
Studies of the Immune Response
Cytokines Representing Products of Activated T Cells
Most of the reports that have measured cytokine levels in women with silicone breast implants and in control groups have examined serum or plasma levels which is less reliable than measuring concentrations in tissues. Most studies of cytokine levels in breast tissue are case reports. An exception is the study by Mena et al. (1995) in which soluble mediators of inflammation were measured in explanted capsular tissue from women with silicone breast implants, in skin scar tissues from women undergoing reverse augmentation mammaplasty, and in synovial tissue of patients with various forms of arthritis. Tissues were cultured for 24 hours in vitro and supernatants were examined for levels of interleukin-2 (IL-2), IL-6, tumor necrosis factor-alpha (TNFa) and prostoglandin E2 (PGE2). No significant difference was noted between capsular tissue and controls. In particular, cytokine production from breast tissues that had been exposed to components of silicone breast implants and from skin involved in previous surgeries was not significantly different. Moreover, no correlation was recorded between systemic symptoms and actual measured cytokine production by explanted capsular tissues. Such measurements, however, may not be meaningful compared to determinations of cytokines by quantitative polymerase chain reaction (PCR), enzyme-linked immunosorbent assay (ELISA), or immune histochemistry.
A study by Ojo-Amaize et al. (1994) examined IL-1b, which is a soluble mediator of inflammation, and IL-1 receptor antagonist, also elevated in inflammatory states, in the blood of women with breast implants with and without symptoms. Differences were found between women with silicone breast implants and healthy age-matched controls, but no valid conclusions were possible because symptomatic women with silicone breast implants were not compared to well women with implants to determine if implantation itself is associated with an increase in either IL-1b or IL-1 receptor antagonist (Ojo-Amaize et al., 1994). Other studies showed no differences between women with breast implants compared to age-and sex-matched surgical patient controls when TNFa and IL-6 as well as soluble TNF receptor were examined (Zazgornik et al., 1996). Moreover, Garland et al. (1996) examined IL-6 levels in women with breast implants and age-matched women without implants; no significant differences were recorded in IL-6. Another report by Blackburn et al. (1997) looked at IL-6, IL-8, TNFa, and soluble IL-2 receptor in women with silicone breast implants compared to healthy age-matched controls. Levels of all cytokines measured were below the range of detection of the various assays, but these investigators were able to detect elevated levels of the same cytokines in the blood of rheumatoid arthritis patients studied at the same time. Few of the experimental studies include controls or testing of biomaterials for endotoxin, which could significantly affect cytokine production (Cardona et al., 1992). In a generic discussion of tissue responses to implantation with a number of different biomaterials including silicone, Anderson noted changes in tissue cytokine concentrations as a general response to biomaterial implantation (Anderson, 1988, 1993, see Chapter 5 for more discussion of and references to cytokines). None of these studies provides sufficient evidence for immune system activation in women with silicone breast implants.
Superantigens Explaining Symptoms in Women with Silicone Breast Implants
Superantigens represent a class of molecules that bind in a non-immune fashion to T-cell receptors and activate a much larger proportion of T-cells than is stimulated by conventional T-cell reacting antigens. Ueki et al. (1994) looked at whether silicate might function as a superantigen. In this study, chrysotile (a silicate) was mixed with peripheral blood T-cells from three healthy subjects. In two of these subjects, an increase in Vb5.3 T-cells and in the other of Vb6.7 T-cells were recorded (Ueki et al., 1994). These numbers are too small to support any conclusions, and breast implant patients are not exposed to chrysotile. O'Hanlon et al. (1996) reported a study of 20 explanted breast implant capsule tissue samples.
PCR was employed to detect T-cell receptor V-region segments as a way of identifying preferential T-cell receptor gene expression. The T-cell receptor V segments were present in 14 samples, but many different T-cell V segments were expressed (O'Hanlon et al., 1996). These observations do not support a prominent role for T-cell superantigens in any immune responseabnormal or normalin women with breast implants. There is, at present, insufficient evidence for superantigen activation of T-cells in patients with silicone breast implants.
Natural Killer Cell Functions
Natural killer (NK) cells can be distinguished from other cells of the immune system by the expression of several distinctive markers on their cell surfaces. Some investigators have suggested that NK-cell activity may be decreased in autoimmune diseases (Sibbitt and Bankhurst, 1985; Struyf et al., 1990), but no specific mechanisms have been elucidated to indicate how decreased NK-cell activity could favor development of autoimmune disease. Consistent with animal toxicology studies noted earlier, it appears that NK cells in humans might be affected by exposure to silicone gel, since removal of silicone breast implants was followed by an increase in NK-cell function in 50% of women studied by Campbell et al. (1994). However, among the remainder, NK-cell function decreased in 26% and remained unchanged in 24%. No data were given on the normal day-to-day variability of NK-cell function in the individuals tested. Moreover, parallel studies of control groups of women without breast implants and women with breast implants who were symptomatic or asymptomatic were not presented, which considerably weakens the study. In the data published to date, there is no clear evidence that changes in NK-cell activity have functional effects or explain the signs and symptoms that characterize women with silicone breast implants who have chronic and unremitting complaints. Moreover, previous studies have demonstrated that NK-cell activity can be altered by stress, sleep loss, and various medications including corticosteroids among otherwise healthy subjects (Irwin et al., 1994, 1996; Pedersen and Beyer, 1986; Pedersen and Ullum, 1994; Shepard et al., 1994; van Ierssel et al, 1996). Whiteside and Friberg (1998) observed that geographic location, gender, age, and even occupation of control populations can affect NK-cell activity and that low NK-cell activity is observed in chronic fatigue syndrome, which has been studied extensively without a consistent correlation with immune defects having been discovered. They noted that NK-cell assays are usually single time point assays in small cohorts and rarely performed under the stringent quality control measures necessary to ensure reproducibility (Whiteside and Friberg, 1998).
Studies of the Adaptive Immune Response
Role of HLA
Many studies have been undertaken during the past several decades in search of susceptibility genes that may predispose certain individuals to the development of certain diseases. Much of this work has centered on the role of the human HLA locus. The presence of a susceptibility gene does not guarantee the development of a disease, but probably indicates that the disease may be much more easily triggered than in the absence of that gene. Only a few studies have provided useful information on HLA associations with symptoms among breast implant patients. Morse et al. (1995) examined whether women with breast implants and scleroderma-like symptoms shared an HLA DQ motif with scleroderma patients without breast implants. HLA types in the scleroderma patients without implants had already been determined and found to have a diminished frequency of leucine at residue 26 of the DQb chain. A control group of healthy individuals without implants was also included for comparison. The study examined a very restricted sort of HLA polymorphism, that is, whether subjects with breast implants and scleroderma-like symptoms also had a lower frequency of leucine at residue 26 of the DQb chain than the control group. The finding of a decreased frequency in women with breast implants and scleroderma suggested that these women are similar to typical scleroderma patients. This study does not provide evidence for a role of silicone breast implants in scleroderma. Rather this study shows that scleroderma can occur in a particular susceptible population whether or not the subjects have breast implants (Morse et al., 1995).
A second carefully designed study by Young et al. (1995b) examined whether certain HLA class I and class II MHC molecules were present in symptomatic women with breast implants, although there is a significant error associated with serological methods of typing class II molecules. Four groups were included: group 1 consisted of 77 women with silicone breast implants and debilitating fibromyalgia-like symptoms; group 2 was composed of 37 women with breast implants, but few symptoms; group 3 contained 54 healthy women without implants; and group 4 had 31 women with fibromyalgia and no breast implants. No differences were recorded in HLA class I (A, B, or C) alleles among the four groups; groups 1 and 4 had an increased frequency of class II DR53 and DR7 suggesting that symptomatic women with breast implants share HLA DR alleles with women who develop fibromyalgia (Young et al., 1995b, 1996). There is no evidence to indicate that fibromyalgia is an autoimmune disease. Current evidence suggests that HLA haplotypes of symptomatic women with silicone breast implants resemble those of symptomatic women without breast implants.
T-Cell Activation in Women with Silicone Breast Implants
In cell-mediated immunity against a particular antigen or group of antigens, T-cell clones that recognize the antigen(s) are stimulated by, and will proliferate in the presence of, that particular antigen(s). T-cells also proliferate in the presence of various mitogens which are substances that cause polyclonal nonspecific T-cell division. Mitogens do not activate the antigen-binding sites of T-cells. To date, no specific autoantigen has been identified that will stimulate T-cells of women with silicone breast implants. Many of the studies which have sought to provide evidence for T-cell activation in woman with breast implants have focused on attempting to detect silicone-specific T-cells in such individuals, that is, T-cells that divide in the presence of silicone or silicone components.
When current reports of T-cell stimulation in women with breast implants are examined, a number of methodological difficulties emerge. These include uncertainty about the physical state of the silicon or silicone used for stimulation, the composition and validity of control populations, and the procedures used in analyzing various sets of data. A critical aspect of most of the studies of T-cell reactivity is their inability to differentiate whether silicone or components from silicone breast implants are recognized as true antigens by T-cell receptors or are functioning as T-cell mitogens.
Ojo-Amaize et al. (1994) reported an enhanced T-cell response to silicon dioxide, silicon, or silicone gel in symptomatic women with breast implants. The actual physical state of the ''antigen" in the preparations studied is unclear. The silicone gel was subjected to extraction before it was used, the silicon was later said to be silicate (Ojo-Amaize et al., 1995), and silicates are normally present in the circulation. Two different sets of controls were studiedone set for silicon dioxide and another for silicon and silicone gel. Data analysis used in this report was not conventional. Any individual who responded to any of the three "antigens" employed was considered a responder. Only one patient was shown to have a positive response at all three concentrations of antigen tested (Ojo-Amaize et al., 1994). The committee could not interpret the results of this study.
A subsequent report by Smalley et al. (1995a) examined the T-cell response to silicon dioxide (silica) in symptomatic women with silicone breast implants. The symptomatic patients appeared to show higher T-cell stimulation indices than age matched controls. No control group of symptomatic women without breast implants was included, and the original data were not shown in this study. In addition silica is an immunologically non-specific stimulator of macrophages (Aalto et al., 1975; Chen et al., 1996; Davis, 1991; Mancino et al., 1983), no particulate controls were included, and there is no evidence that women are exposed to silica by
silicone breast implants. The observations of Garrido et al. (1994) by nuclear magnetic resonance (NMR) and the detection of silica by polarizing microscopy, on which these authors rely as evidence for the presence of silica in women with breast implants, have been seriously challenged, as noted elsewhere in this report.
Smalley et al. (1995b) also exposed lymphocytes from women with silicone breast implants, normal controls, and women with defined connective tissue disease or fibromyalgia to three mitogens and colloidal silica. In a subsequent, expanded study, they examined responses to silica in 942 symptomatic and 34 asymptomatic women with implants and 220 normal control women. In the initial study, similar responses to mitogens were observed in control and implanted women. Stimulation indices after silica were elevated in women with implants, but not to the levels observed after mitogen stimulation. In the expanded study, 91.3% (860) of the women with silicone breast implants were deemed to have mild, moderate or markedly elevated stimulation indexes compared to controls on exposure to silica (Smalley et al., 1995b). This report also includes only brief aggregate data, and the problems previously noted obtain here as well.
Testing of the technology used in these reports by Smalley et al. by submission to their laboratory on two occasions of samples from eight women with either silicone gel, saline or double lumen implants and six women who had never had any implants (but for whom false histories of breast implantation were provided) was carried out by an independent investigator. This analysis yielded an array of results that bore no relationship to the clinical status of the patients, including positive stimulation indices, from mild to marked, in all six of the unimplanted women on first testing, and reversion to negative on repeat testing in one woman in both the implanted and the unimplanted group; the mean stimulation indices of the implanted and unimplanted group on repeat testing were moderately elevated and quite similar, 92 and 87, respectively (Young, 1996b). The findings of these studies have not been confirmed by others (in fact, as noted, serious questions have been raised), and their reproducibility and biologic plausibility are questionable.
Another report by Ellis et al. (1997b), which lacked a symptomatic control group and has not been confirmed, attempted to find autoreactive T-cells by looking at T-cell stimulation by connective tissue components as well as reactions to implant biomaterial. Twenty-six symptomatic women with breast implants were studied in parallel with 23 age-matched healthy controls without implants. Of these 26 women, 15 (58%) had undergone explantation of their implants, suggesting that this was not a random group. The women with breast implants showed increased T-cell proliferation to collagen I, collagen II, fibrin and fibronectin, but no incre-
merit in reactivity to myelin basic protein, transferrin, bovine serum albumin, tetanus, octamethylcyclotetrasiloxane (D4), or silicone gel. Anti-nuclear antibodies were also present at titers of 1:40 or more in five (19%) of the implant recipients, not significantly different from the healthy controls (Ellis et al., 1997b).
A study by Ciapetti et al. (1995) compared T-cell responses to silicone gel in 22 women with breast implants to those of 10 women without implants who were an average of 15 years younger. The authors reported increased proliferation of T-cells from women with breast implants on exposure to an aqueous silicone gel extract. However, a standard stimulation index was not given; instead, stimulation in the women with implants was compared to stimulation in the control group, and the difference in stimulation between the groups was much less than twofold. For unexplained reasons, the lymphocyte stimulation of augmentation patients (N = 6) was significant, but the stimulation of reconstruction patients (N = 16) was not. These positive results appear to rest on six patients. Moreover, considering the insolubility of silicone gel in aqueous media, it is not known what compounds might be present in the extract (Ciapetti et al., 1995).
Several groups have investigated the possibility of cellular immunity to silicone. Snow and Kossovsky (1989) suggested that 3 of 29 patients with ventriculoperitoneal shunts might have delayed type hypersensitivity to silicone on the basis of tissue eosinophilia associated with the shunts. However, some of the patients in this series were infected and one of these had marked eosinophilia. Moreover, the finding of eosinophilia as a marker for delayed hypersensitivity is not conclusive. Jimenez (1994) also concluded that the requirement for revision of three shunts was caused by delayed hypersensitivity. Some non-silicone replacement shunts also required revision, and infection complicated the clinical picture here also. Nosanchuk could not demonstrate delayed hypersensitivity in guinea pigs injected with Dow Corning 360 fluid (Nosanchuk, 1968a). Kossovsky et al. (1998) also examined this question in guinea pigs. The animals were injected intraperitoneally three times per week for four weeks with either: an equal volume of Dow Corning silicone fluid and sterile guinea pig serum that was mixed for 28 days at 37°C and emulsified in complete Freund's adjuvant and tissue culture medium; or tissue culture medium alone. One week after the last injection, the animals were skin tested with: silicone-serum pellet (after centrifugation) in medium; a saline-serum pellet in medium; pure Dow Corning silicone fluid; or purified protein derivative (PPD), 250 tuberculin units. The guinea pigs had a marked reaction to PPD. They also reacted to the silicone-serum pellet, but the response to silicone alone was no different than the response to saline-serum pellet. Naïve animals that received spleen cells from experimental
guinea pigs gave essentially similar results. Whether the response to the silicone-serum pellet represented a reaction to denatured serum proteins resulting from the 28 day incubation or to microbial contamination of the material is uncertain. However, the data appear to exclude a response to silicone itself. Narini et al. (1995) studied delayed hypersensitivity in sheep that were injected with McGhan implant silicone gel alone, silicone gel and complete Freund's adjuvant, saline or adjuvant controls. They concluded that there was delayed hypersensitivity to the silicone gel. However, there was a significant response to silicone in the adjuvant primed sheep, strongly suggesting that this response was not antigen-specific.
Investigators attempted to immunize animals against silicone gel and studied subsequent tissue reactions on exposure to silicone. They reported no changes in tissue cellular reactions in immunized compared to non-immunized, silicone-challenged animals, suggesting that cellular changes are implant wound, rather than silicone immune, related (Brantley et al., 1990a,b; Klykken et al., 1991a-e, see Chapter 5). In addition, lymphocytes of animals immunized with silicone gel and complete Freund's adjuvant did not respond to silicone gel challenge after four weeks or eight months, and T-cell helper suppressor ratios were unchanged (Brantley, 1995b).
A study by Katzin et al. (1996) examined T-cells from breast tissue and peripheral blood of women with breast implants. Peripheral blood of control individuals (not matched to sex or age) was studied in parallel. T-cells obtained from breast tissues showed an increased prevalence of an activation marker, HLA-DR antigen, compared to peripheral blood T-cells. A decrease in the numbers of CD4+, CD45+, and RO+ helper T-cells was noted in the implant associated T-cells as compared with the peripheral blood T-cells of these patients. Comparison of peripheral blood T-cells of patients and controls revealed only a slight decrease in CD4+, CD45+, and RO+ cells in patients. The cells of the patients with implants were sometimes stained after freezing and thawing. It was not clear whether the control cells were handled in a similar fashion. Differences in implant and peripheral blood T-cell phenotypes in the same breast implant patients do not constitute a strong indication for an in vivo adaptive immune T-cell activation or an ongoing abnormal T-cell disease process.
In general, studies addressing these issues are limited, and the technical problems associated with available studies are substantial. In view of these factors, the committee concludes that studies supporting a consistent pattern of marked T-cell activation or sensitization against autologous self-antigens or silicone in patients with silicone breast implants are inconsistent and unconvincing. At present there are no conclusive data showing that silicone or any components of breast implants represent clearly defined T-cell antigens or that any individual connective tissue
component has been transformed or adapted into a T-cell autoantigen capable of perpetuating a chronic inflammatory process.
Effects of Silicone: Granulomatous Inflammatory Reactions
Inflammation is characterized by local vascular and tissue reactions as well as generalized systemic effects including fever, leukocytosis, and production of cytokines, as noted above. Inflammation represents the body's local reaction to tissue injury. Acute inflammation is characterized by exudation of various proteins and other substances from blood and migration of polymorphonuclear leukocytes to the site of injury. With biocompatible materials such as PDMS, this phase is not prolonged (indicating that the polymer is not providing a stimulus for continued inflammation) (Anderson, 1988). In prolonged tissue exposures, as is the case with silicone breast implants, chronic inflammation seems to be the most relevant process. Accumulations of lymphocytes and monocytes, which later differentiate into macrophages, characterize chronic granulomatous reactions. The tissue reaction to silicone implants appears to be one of granulomatous inflammation. The general histological appearance of granulomas is variable but usually shows a central region of macro-phages, with or without caseation, surrounded by a zone of lymphocytes and more peripherally a zone of fibroblasts. The presence of multinucleated giant cells is a key histologic feature. Foreign body type giant cells, which are more common in silicone granulomas, have centrally arranged nuclei.
The histological appearance of foreign body reactions is not specific for any particular disease process. Silicone granulomas are typical foreign body granulomas which show multinucleated giant cells and aggregates of epithelioid macrophages, surrounded by dense lymphocytic and neutrophil cellular infiltrates. Plasma cells (known to be the cells responsible for antibody production) are frequently also present (Hill et al., 1996). The granulomatous histologic lesions induced by silicone seem to vary considerably depending on the form of silicone that actually causes them (Travis et al., 1985). Silicone elastomer, for example from joint prostheses or hemodialysis tubing, usually produces an intense foreign body giant cell reaction, most prominent in adjacent lymph nodes (Christie et al., 1989; Lazaro et al., 1990). Experimentally, the inflammatory response to particulate silicone elastomer is localized to the exposed (injected) joint, other joints are not inflamed (Worsing et al., 1982). The tissue reactivity to silicone gel or oil shows less organized granulomas with cystic spaces, presumably containing the foreign material (Hardt et al., 1995b; Peimer et al., 1986; Raso and Greene, 1997).
Tissue Response to Silicone
Interactions of Silicone with Plasma Proteins
The inflammatory reaction to implanted foreign materials such as silicone is similar to other foreign body tissue reactions, as noted. This reaction includes interaction with plasma proteins, recruitment of inflammatory neutrophils and monocytes, differentiation of monocytes into macrophages, fusion of phagocytic cells to form giant cells and stimulation of fibrosis (Anderson, 1988). Hydrophobic materials such as silicone elastomer are coated with host proteins within a few hours of implantation. Even in the presence of dilute protein solutions the elastomers are at least 70% covered with host proteins within one hour (Butler et al., 1997). Thus, most inflammatory cells may never make direct contact with the "naked" biopolymer, but a wide variety of plasma proteins such as IgG, albumin, fibronectin and complement components can adsorb to silicone gels or elastomers and help recruit inflammatory cells to the site of the implant. Presumably, the inflammatory cells that enter the tissues do not respond to the foreign material itself but to a surface layer of adsorbed, partially denatured plasma proteins for which there are specific receptors on neutrophils and macrophages (Anderson et al., 1990, 1995; Bonfield et al., 1989a,b). Such responding macrophages may secrete twofold more Il-1b, Il-6 and TNFa than macrophages exposed to naked silicone (Naim et al., 1998).
Formation of Giant Cells and Effects on Fibroblasts
A foreign body giant cell reaction is typical of the tissue response to silicone elastomer, but is less commonly associated with inflammation in response to silicone gel or silicone oil. Studies by several groups of workers suggest that silicone induces IL-1, TNFa or IL-6 production by human monocytes. Moreover, there also seems to be a possibility that the bio-materials used to stimulate cytokine production could possibly themselves have been contaminated by microbial products such as endotoxin lipopolysaccharide (LPS), lipoteichoic acid or similar molecules, which may be responsible for some of the observed experimental tissue reactions. This has not been investigated adequately as yet (Anderson, 1993; Bonfield et al., 1992; Miller and Anderson, 1989; B.D. Ratner, personal communication, 1998). Silicone implants eventually become walled off by the formation in vivo of a fibrous capsule, which is a normal element of the response to a foreign body. Contracture of the capsule can become a problem in some patients with silicone breast implants, as discussed in Chapter 5. Monocytes or macrophages in contact with silicone appear to
produce cytokines that stimulate fibroblast growth, but this is not necessarily an adaptive immune response (Bonfield et al., 1991; Miller and Anderson, 1989).
Specific Autoantibodies Involved in Symptomatic Responses to Silicone Breast Implants
Antinuclear antibodies and specific autoantibodies, such as anti-Ro, anti-La, anti-RNP, and anti-Sm antibodies, as well as rheumatoid factor or anti-topoisomerase antibody often associated with classical connective tissue disorders such as systemic lupus erythematosus (SLE), systemic sclerosis (SSc), Sjojren's syndrome (SS), or rheumatoid arthritis (RA), have been measured in patients with breast implants. These are discussed in Chapter 7. Several other general classes of autoantibodies have also been examined with respect to whether they might be involved either directly in pathogenesis or indirectly as disease activity indicators in patients with silicone breast implants. Anticollagen antibodies have been subjected to scrutiny by investigators who concentrated on a relatively small group of patients (Bar-Meir et al., 1995; Rowley et al., 1994; Simpson et al., 1994; Teuber et al., 1993). Anticollagen antibodies were examined in breast implant patients because the symptoms experienced by these subjects often include arthralgias, skin complaints, and musculoskeletal symptoms. Anticollagen antibodies have been measured in some patients with rheumatoid arthritis and also have been implicated in some mouse models of induced arthritis.
The initial report by Teuber et al. (1993) examined anti-collagen antibodies in 46 women with silicone breast implants (2 women with previously diagnosed connective tissue disease, 38 symptomatic women, 6 asymptomatic women) and age-matched healthy women. An ELISA reading greater than three standard deviations above the mean of the controls was considered a positive assay for anticollagen antibodies. Of the 46 women with implants, 35% had positive tests for anticollagen antibodies (26% to collagen I and 15% to collagen II) in comparison to 9% of healthy controls (Teuber et al., 1993). With the exception of antibodies to denatured collagen II, these differences were significant (p < 0.05). Rowley et al. (1994) reported anticollagen antibody reactivity among 70 women with silicone breast implants. Many of these subjects had been included in the previous study by Teuber et al. (1993). Anticollagen activity was observed in 41% of symptomatic women with breast implants in comparison to 29% of SLE patients, 48% of RA subjects, and 6% of controls (Rowley et al., 1994).
These studies indicate that some symptomatic women with breast implants may have antibodies reacting to collagen I and II. No overall consistent pattern of reactivity is apparent by Western blotting. No data were presented by this group on healthy women with silicone breast implants or on women with functional musculoskeletal complaints without breast implants. The relevance of these antibodies to clinical symptoms was not discussed. Similar results in similar patients were reported in abstract form by Simpson et al. (1994). No clear role for anticollagen antibodies is yet apparent in any connective tissue disease.
Whether individuals with silicone breast implants ever develop antibodies to silicone is relevant to the question of the immunogenicity or potential disease producing capacity of silicone. Investigation of this question has turned out to be rather difficult because of marked technical problems in constructing a reliable assay for antisilicone antibodies. Silicone itself is sticky and viscous and can bind non-specifically to proteins or surfaces. When a substrate that is naturally sticky and gummy is used, high levels of immunoglobulins will produce falsely high ELISA readings because they adhere to plates coated with such a sticky antigenic target.
Wolf et al. (1993) developed an assay for antisilicone antibodies using bovine serum albumin bound to polystyrene plates and then coated with hydroxylsilicone. Using this assay, these workers concluded that anti-silicone antibodies were increased in women with silicone breast implants, and moreover, that the highest levels of antisilicone antibodies were found in implant recipients whose implants had ruptured. A control group of symptomatic women without breast implants was lacking, and whether the women studied were symptomatic or not was not reported.
Later Rose et al. (1996) employed the same type of assay to show that women with connective tissue diseases as well as women with silicone breast implants had elevated antisilicone reactivity. Rosenau et al. (1996) could not confirm the work of Wolf et al. and could not develop an ELISA test for antisilicone antibodies using a number of variations of silicone and other conditions. Others confirmed that proteins, including antibody proteins, bind non-specifically to silicone. They failed to uncover any evidence for antisilicone antibodies (Butler et al., 1996; Van Oss and Naim, 1996). Goldblum et al. (1992) provided an example of the difficulties in determining the presence of antisilicone antibodies. An initial report of antibody to silicone (Silastic) ventriculoperitoneal shunts was retracted when it was discovered that experimental and control sera both had similar IgG binding to silicone, the variable being differential modulation by albumen (Goldblum et al., 1995). Others were unable to find antisilicone
antibodies in women with silicone breast implants using similar technology (Rohrich et al., 1996), including the technology of Rosenau et al. (1996) as applied to 200 women with silicone breast implants and 500 controls (Karlson et al., 1999).
Kossovsky et al. (1993) employed the DetectSil assay for sera binding to antigens which had been modified by silicone. In this assay microtiter plates were coated first with silicone and then with various self-antigens such as fibrinogen, collagen, and fibronectin prior to applying a dilution of test serum. Binding of sera to such plates without antigen was greater than binding to plates coated with antigens, which was consistent with the known sticky qualities of silicone. Other investigators attempting to produce antisilicone antibodies in animal models have not been able to do so (Naim and van Oss, 1992; Nosanchuk, 1968a). It currently appears that a reliable assay for antisilicone antibodies is not available, and the existence of antisilicone antibodies is unproven and biologically unlikely. The Centers for Disease Control and Prevention (CDC) have expressed skepticism about such diagnostic tests for silicone breast disease (CDC, 1996). For these reasons, the committee has concluded that evidence that silicone or silicone-modified proteins are antigenic and capable of eliciting an adaptive immune response is insufficient or flawed. The evidence suggests that specific antibodies are not produced in response to exposure to silicone.
In an attempt to develop an assay specific for a silicone breast implant associated disease, Tenenbaum et al. (1993a, 1997a), Gluck et al. (1996), and Wilson et al. (1999) have reported results with an antipolymer antibody test (APA). This assay is said to measure the binding of IgG to partially polymerized polyacrylamide immobilized on nitrocellulose strips (as described in Wilson et al., 1999). This is a non-silicon-containing compound; it is unlikely that it has any antigenic similarity to PDMS or other breast implant silicone compounds. In an early abstract report of APA testing of 97 symptomatic women with silicone breast implants, 19 asymptomatic women with implants, 23 healthy controls without implants and 15 women with classical connective tissue disease without implants, APA positivity was related to severity of illness (Tenenbaum et al., 1993a).
In an unblinded survey of 667 symptomatic women with breast implants, 54.4% (N = 343) APA positivity was subsequently reported. This was followed by a blinded study of a highly selected group of women with silicone breast implants (50% of a clinic group of implanted women was solicited, 70% of these declined to participate). Five groups of women
were tested: 34 breast implant patients with limited symptoms and disability; 26 with mild problems; 16 with moderate symptoms; and 19 with advanced symptoms and functional disability; and 15 patients with classical connective tissue disease and implants. APA positives in these five cohorts were 3, 8, 44, 68, and 20%, respectively. Twenty-three healthy clinic personnel controls were 17% positive, and 10% of 20 classical connective tissue disease patients without implants were positive. Simultaneous testing for antinuclear antibodies was carried out. There was no relationship between the two assays (Tenenbaum et al., 1997a). The division of these implant patients into severity groups was subjective. Part of the study was unblinded, and the patient group appeared to be selected in unknown ways from a group already referred for problems. A number of methodological controls were absent (for example, albumen, other polymers). Similar criticisms of this work were made at the time the report appeared (Angell, 1997; Edlavitch, 1997; Korn, 1997; Lamm, 1997).
Gluck et al. (1996) followed up this study of silicone breast implant patients by testing 48 patients with fibromyalgia, 16 patients with osteoarthritis, and 14 patients with RA. Positive APA tests were observed in 48, 19, and 7% of these groups, respectively. Wilson et al. (1999) recently reported results similar to those of Gluck et al., but assessed patients for severity and found more positive APA tests in fibromyalgia patients with severe (61%) than mild (39%) disease. They concluded that the APA test may be a marker for severe fibromyalgia although the committee is not aware of any standards or criteria for grading the severity of fibromyalgia. An agency of the Dutch government found this test to be reproducible and plans to carry out investigations on its clinical implications (de Jong et al., 1998). The specificity and clinical significance of the APA assay remain to be defined, but it is unlikely that it detects antisilicone antibodies.
In addition, Praet et al. (1999) in an abstract reported that APA assay results from the United States and the Amsterdam Red Cross Blood Transfusion Services Central Laboratory in 24 women with silicone breast implants undergoing explantation matched completely. Five preoperative and eight postoperative samples were positive. There was no correlation with autoantibodies, signs of rheumatic disease, general health, implant type or capsular contracture. There appeared to be a higher frequency of APA positivity in women with extracapsular silicone leakage than in those without, and surgery itself appeared to result in more APA positives and higher circulating APA concentrations. The statistical significance of these results was not reported.
Induction of Hypergammaglobulinemia by Silicone
Some evidence has been presented that silicone exposure may stimu-
late a polyclonal hypergammaglobulinemic state in some individuals. Ostermeyer-Shoaib et al. (1994) reported elevations but also some decreases in IgG levels. Primary data were not shown, so the significance of these findings is not clear. Elevations of IgG and IgM levels were reported by Brunner et al. (1996) in, respectively, 12.6% and 7.5% of 239 patients with silicone breast implants. The frequency of increases in IgG or IgM concentrations was similar in patients with gel-filled or saline implants. In this study, however, matched controls were not examined, and information about actual levels of absolute IgG or IgM was not reported, making exact correlation and analysis difficult.
In another series, no differences in immunoglobulin levels between women with or without breast implants were found (Bridges et al., 1993a). The well controlled analysis of blood samples from the Nurses' Health Study by Karlson et al. (1999) also found no increases in IgA, IgG or IgM levels in 50 women with implants compared to 50 healthy control women without implants. Additional studies of immunoglobulin levels with some elevations (Garland et al., 1996; Silverman et al., 1996a) are discussed in relation to multiple myeloma in Chapter 9. In summary, the data on immunoglobulin levels are inconsistent, but a controlled study of an apparently random group of women did not find elevations in women with breast implants compared to matched controls.
In this connection, several groups are currently in the process of studying whether subjects with silicone breast implants are more likely than the general public to develop multiple myeloma, a lethal malignant proliferation of monoclonal plasma cells considered to be a cancerous condition. Previous animal studies by Potter et al. (1994) have demonstrated that certain susceptible strains of mice develop intraperitoneal plasmacytomas in a high proportion of instances after exposure to silicone gel from breast implants, but this is not necessarily a model for human multiple myeloma as noted in Chapter 9. Silverman et al. (1996a) reported three cases of multiple myeloma in women with silicone breast implants. These women were found among 34 patients seen in the myeloma clinic at University of California at Los Angeles (UCLA) representing 5.9% of the clinic population. In subsequent studies of different larger cohorts of women with (N = 108) and without (N = 176) breast implants, elevated immunoglobulin levels and five instances of monoclonal gammopathy were detected. Monoclonal gammopathies disappeared in two women after removal of their implants, in a third there was no change after explantation, and two women were not retested. These preliminary reports raise the possibility that silicone implants may increase the risk of monoclonal gammopathy and/or multiple myeloma (Silverman et al., 1996a). However, since benign monoclonal gammopathy is relatively frequent in the healthy population (Kyle, 1996), and since several case series
and epidemiological studies do not confirm these findings, these reports provide insufficient evidence for an association (see Chapter 8).
Based on the data available, the committee concludes that there is no convincing evidence to support clinically significant immunologic effects of silicone or silicone breast implants. This includes: insufficient evidence for an association of a particular HLA type in women with breast implants and health conditions; insufficient evidence for silicone as a super-antigen; insufficient or flawed evidence that silicone produces immune activation of cells of the immune system, silicone antibodies, delayed type hypersensitivity to silicone, cytokines as an immune response, antigen specific immune cellular infiltrates; and insufficient evidence for autoantibodies or T-cell self antigen activation. The paucity of significant, well controlled studies examining these questions is responsible for these conclusions. The committee finds that there is conclusive evidence that some silicones have adjuvant activity, but there is no evidence that this has any clinical significance. The committee has also concluded that evidence from experimental studies of the immunology of silicone does not support, or lend biologic plausibility to, associations of silicone breast implants with immune related human health conditions.