Neurologic Disease and Its Association with Silicone Breast Implants
Two recent epidemiological studies, carried out in Denmark and Sweden, used national hospital discharge registry databases to identify women with breast implants and compared them to age-matched women who had undergone breast reduction surgery. Nyrén et al. (1998c) studied three large cohorts in Sweden3,502 women with cosmetic breast implants, 3,931 women with implants for reconstruction, and 3,351 women who had undergone breast reduction surgery. On review of a 2,500-woman sample from the cosmetic implant group, 24% were found to have saline implants. A ten-year follow-up of these women was performed through cross-linkage within hospital discharge registers. Neurologic diagnoses examined included multiple sclerosis, neuritis of the optic nerve, amyotrophic lateral sclerosis, diseases of the nerve roots and plexus, mononeuritis of the upper extremity (median, ulnar, and radial nerves), mononeuritis of the lower extremity, Guillain-Barré syndrome, and Meniere's disease. Charts of all women with neurologic diagnoses were reviewed. Overall the relative risk for any of the neurologic diseases was 0.8 (95% confidence interval [CI], 0.5-1.4). Following removal of prevalent (preexisting) and misclassified cases identified through chart review, there was a significant deficit (relative risk, 0.5; 95% CI, 0.2-0.9) of multiple sclerosis and a marginally significant deficit of mononeuritis of the upper limb (relative risk, 0.5; 95% CI 0.2-1.0) in implant patients (Nyrén et al., 1998c).
The Danish study followed 1,135 women with breast implants and 7,071 women in the comparison breast reduction group. The same neurologic diseases were examined, with the addition of myasthenia gravis, other demyelinating central nervous system neuropathies, and motor neuropathy. Cases were not verified. No increased relative risk for defined neurologic disease in the implant group relative to the comparison group was found. The relative risk in the implant group was 1.7 (95% CI, 0.9-2.9), but a similar excess of neurologic disease was found in the breast reduction control group. The relative risks for several individual neurologic conditions were not significantly elevated, and on chart review, 38% of the cases (5 of 13) of neurologic disease were discovered to have had their onset before breast implantation (Winther et al., 1998).
Another recent publication addressed the issue of sensorineural hearing loss (Meniere's disease) associated with silicone breast implants. A group of 119 of 184 women with Meniere's disease or progressive hearing loss and 100 age-matched controls responded to questionnaires (64.7% response rate) and provided serum samples to measure the presence of the 68 kiloDalton (kDa) protein found in some forms of autoimmune hearing loss. There was no significant association of silicone breast implants with Meniere's disease or progressive hearing loss (odds ratio, 1.42). The presence of the 68-kDa protein was not significantly associated with the presence of silicone breast implants (Kim and Harris, 1998). The committee has concluded that these well-designed epidemiological studies provide limited evidence for the lack of association between breast implants and neurologic disease.
Case Series and Reports
Pathological findings in nerve and muscle biopsies from 55 women with breast implants were reported in an abstract by Vogel and Edmondson (1996). Biopsies were examined by light and electron microscopy and by teasing the nerve fibers. Pathology was observed in 6 of 55 biopsies, including 3 with axonal neuropathy, 1 with granulomatous neuritis and myositis, 1 with chronic inflammatory demyelinating polyneuropathy (CIDP), and I with Charcot-Marie-Tooth disease. The authors concluded that only conventional neuromuscular disease was found in these women (Vogel and Edmondson, 1996).
Additional pathological and toxicological reports are reviewed in Chapter 4. Hine et al. (1969) did not find evidence for neurotoxicity on injection of silicone syringe lubricant into the lumbar subdural space and cisterna magna of rabbits, monkeys, and rats. Silicone was implanted subdurally in the brain of rats with no observable effect (Agnew et al., 1962). No direct evidence of silicone gel toxicity to peripheral nerves was
observed when gel was injected directly into or around the sciatic nerve of rats, although an inflammatory response followed by fibrosis was seen (Sanger et al., 1992).
Also silicon concentrations were somewhat elevated in brains of rabbits exposed to gel implants that had been cut open, but not in rabbits with intact gel or saline implants. No pathological changes were observed in the brains of these rabbits (Marotta et al., 1996b). These results are reported briefly with no individual data and are unconfirmed. No accumulation of silicon in sural nerves of women with breast implants were found by Evans et al. (1996), and toxicological studies of silicone distribution after injection subcutaneously of gel distillate in mice showed, if anything, lower brain concentrations of cyclic compounds and moderate concentrations of linears (Kala et al., 1998; see Chapter 4).
In response to inquiries to the American Academy of Neurology (AAN) regarding scientific reports that associated neurological disease with silicone breast implants, the AAN Practice Committee performed an extensive search of the scientific literature from 1975 on (Ferguson, 1997). This search discovered 14 reports focusing on a variety of painful syndromes, for example, involvement of the brachial plexus by a ruptured silicone breast implant (Collins et al., 1995). A Call for Comments in the AAN newsletter produced some anecdotal responses, a published abstract (Rountree et al., 1995), several published papers and unpublished manuscripts from Ostermeyer-Shoaib and colleagues, and an article from Rosenberg (1996) that reviewed medical records of 131 patients that had been examined by Ostermeyer-Shoaib et al. (Rosenberg, 1996).
The published abstract by Rountree (1995) reported 330 women who were injury claimants and 248 controls. Antinuclear antibody (ANA) frequency was no different for women with silicone breast implants and controls. The prevalence of antibodies to ganglioside M1 (anti-GM1) was 3.3% in patients with ruptured implants, not significantly different from 5.4% in patients with intact implants. Age-adjusted prevalence was 19.1% in patients with ruptured implants compared to 7.1% in patients with intact implants, a difference that was still not. statistically significant. It was proposed that rupture increases with age and this increases the risk of anti-GM1 antibodies. Two reports by Ostermeyer-Shoaib and colleagues (Ostermeyer-Shoaib and Patten, 1996a; Ostermeyer-Shoaib et al., 1994), discussed below, were critically reviewed in the report of the Practice Committee (Ferguson, 1997). Problems identified included the following: (1) both papers are case series whose findings have not been replicated by other groups; (2) clinical presentation, symptoms, and physical findings were not associated with laboratory data for the individual neurological diagnoses; and (3) as reported by Rosenberg (1996), a review of the medical records from 131 of these patients found little data to
support the neurological diagnoses. The Practice Committee concluded that existing studies do not support any association or causal relationship between silicone breast implants and neurologic disorders.
Ostermeyer-Shoaib and colleagues published numerous abstracts describing self- or attorney-referred women with silicone breast implants and neurological syndromes (Ostermeyer-Shoaib and Patten, 1993a,b,c; Ostermeyer-Shoaib et al., 1992). Results from the first 100 consecutive referrals from 1985 to July 1992 of women with silicone breast implants who developed symptoms after implantation were reported in 1994 (Ostermeyer-Shoaib et al., 1994). This group of 100 women had increased to 1,500 by December 1993. Subjects were studied by history, physical and neurological examination, and a plastic surgery consultation. All subjects had 20 to 30 symptoms. The most common were weakness (95%); fatigue, myalgia morning stiffness, joint pain, and memory problems (81%); sicca complex (71%); shortness of breath (63%); and joint swelling (58%). History and physical examination provided diagnoses of polyneuropathy in 83, multiple sclerosis-like syndrome in 10 (with peripheral neuropathy present in eight of these), motor neuron disease in 5, and myasthenia gravis in 2 patients. The criteria used to establish these diagnoses were not reported. Symptoms and physical signs were tabulated only in the aggregate and therefore could not be linked to a diagnosis. Magnetic resonance imaging (MRI) of the brain in 84 patients demonstrated multiple white matter lesions in 19 women, but the distribution of these lesions was not specified. Thirteen women had multiple small ischemic lesions; this type of lesion is frequently considered part of normal aging. Assay of 28 spinal fluid samples revealed oligoclonal bands in 13, but the number of bands was not specified. Axonal neuropathy was diagnosed by electromyography in 4 patients, and 53 had an abnormal sural nerve biopsy with significant loss of myelinated fibers. It is surprising for the electrodiagnostic findings to reveal so few abnormalities in the presence of significant pathology in the large myelinated fibers. Randomly elevated autoantibodies, such as ANA, rheumatoid factor, anti-GM1, anti-MAG (myelin-associated glycoprotein) and antisulfatide, were measured. The committee can find no pattern of results or of clinical or laboratory evidence that would allow an assessment of the accuracy of any of the diagnoses.
In a second publication, Ostermeyer-Shoaib and Patten (1996a) summarized findings in 26 patients with silicone breast implants that they believed had a systemic disease involving the central nervous system. A referring neurologist had made the diagnosis of atypical multiple sclerosis with physical findings of optic neuritis, hyperreflexia, spastic paraparesis, stocking-glove sensory loss, nystagmus, and ataxia. MRI had demonstrated white matter lesions in 21 of these women. Visual evoked
responses were prolonged in 14 of 23. Lumbar puncture found oligoclonal bands in 18 of 23, and electromyography was abnormal in 9 of 19, including 4 with carpal tunnel syndrome, 3 with myopathic units, and I with denervation (Ostermeyer-Shoaib and Patten, 1996a). Despite these varied electrodiagnositc findings, all 15 sural nerve biopsies were abnormal. The committee notes that these results do not fit any known pathology, are clinically inconsistent, and have not been reported by any other group.
Rosenberg (1996) reviewed the medical records of 131 patients examined by Ostermeyer-Shoaib and Patten. He found all patients had non-neurologic symptoms that included fatigue (82%), numbness or paresthesias in the upper extremities (50%), headache (47%), numbness or paresthesias in the lower extremities (41%), back pain (37%), gait disturbance (35%), neck pain (35%), dizziness (34%), blurred vision (18%), and diplopia (8%). Neurological examination was normal in 66% or revealed abnormalities that were rated as mild or subjective. The most common finding, varying sensory abnormalities, was present in 23%. Muscle weakness was present in 18% in a variable distribution. Reflexes, abnormal in 8%, were increased in all but one patient who had absent ankle jerks. Both mental status and cranial nerve abnormalities were found in 3%. Gait disturbance was present in 2 patients, although 46 had complained of this. Twenty-one antibodies were tabulated and found to have a prevalence similar to that in the general population. Rosenberg noted one case of multifocal motor neuropathy with an elevated titer of anti-GM1, a condition described in the neurological literature. Decreased perfusion by brain SPECT, reported in 67%, showed no consistent pattern. Electro-diagnostic testing was abnormal in 12 of 48 women, but these were primarily entrapment syndromes or root compression. Electrodiagnostic findings of axonopathy and myopathy in three women were not supported by clinical findings. Diagnostic imprecision and preconceptions were exemplified by the use of the term ''silicone encephalopathy," which was applied whenever any cognitive complaint existed, even with normal mental status testing. White matter lesions found by MRI were scattered nonspecifically (27%), or the few lesions present were not in a periventricular distribution as found in multiple sclerosis. The diagnosis of chronic inflammatory demyelinating poyneuropathy (CIDP) in 23% did not meet the clinical, electrophysiological, or neuropathological criteria for this condition. Rosenberg found support for some specific diagnoses in the 131 patients: depression (16), fibromyalgia (9), radiculopathy (7), anxiety disorder (4), multiple sclerosis (4), other psychiatric disorders (3), multifocal motor neuropathy (1), dermatomyositis (1), and carpal tunnel syndrome (1). However this review of medical records did not uncover any causal link between silicone breast implants and neurological disease (Rosenberg, 1996).
A neuroimmunologic evaluation of 200 symptomatic patients with breast implants and 100 symptomatic patients with chronic fatigue tabulates only laboratory data and provides an extensive methodological description. Antibodies to myelin basic protein, myelin associated glycoprotein (MAG), asialoganglioside GM1, and sulfatide were measured, and lymphocyte subsets were defined. The absence of demographic or clinical information limits any meaning that might be assigned to these laboratory data (Vojdani, 1995a). An abstract by Shanklin and Smalley (1996a) reports their examination of sural nerve biopsies from women with silicone breast implants using polarizing microscopy. These authors reported detection of quartzite silica scattered throughout the nerve at the outer surface of the myelin and speculated on how silicone might be transformed to silica in the body (Shanklin and Smalley, 1996a). However, there is no credible evidence that silicone is degraded to crystalline silica under physiologic conditions. There is also no credible evidence that crystalline silica originates from breast implants or is found near implants (see Chapter 5 of this report). Furthermore, polarizing microscopy is not a reliable technology for detection of crystalline silica (IRG, 1998; see Young, IOM Scientific Workshop), and examination of sural nerves in cadavers of women with silicone breast implants did not reveal elevated silicon concentrations (Evans et al., 1996).
The available studies suggesting neurologic disease, with the exception of obvious local problems due to the physical presence of silicone gel which can compress nerves following implant rupture and migration of the gel, have defects that limit any conclusions to be drawn from them. Furthermore, basic toxicological and animal experimental studies do not find pathology that would support a causation of human neurologic disease by silicone breast implants. Two epidemiological studies suggest that there is no elevated relative risk for neurological disease in large cohorts of women with silicone breast implants. The committee finds that the evidence for a general neurologic disease or syndrome caused by, or associated with, silicone breast implants is insufficient or flawed.