Clinical manifestations

Presentation and course

Autoantibodies targeting the GAD enzyme can impair GABAergic neurotransmission and result in neuronal excitability disorders, the most characteristic being stiff-person syndrome. The other neurologic diseases associated with anti-GAD antibodies are autoimmune epilepsy, limbic encephalitis, cerebellar ataxia, and nystagmus and all collectively comprise the “GAD antibody-spectrum disorders” (55; 30). Overlapping symptomatology among all these GAD-positive disorders is common. GAD antibodies at lower titers directed at different epitopes are also found in about 80% of patients with diabetes mellitus type 1 (03), whereas up to 30% of patients with GAD antibody-spectrum disorders also have diabetes mellitus type 1. Patients with GAD antibody-spectrum disorders have elevated titers of anti-GAD antibodies in the serum, often also detected in the cerebrospinal fluid, ranging between 20,000 and 2,000,000 IU/ml as determined by a quantitative ELISA assay; in patients with diabetes mellitus type 1, the titers range from 5 to 1000 IU/ml (36; 02). The presence of high GAD titers in connection with true neurologic syndromes, compared to atypical or nonspecific entities and diabetes mellitus type 1, is a fundamental consideration for the diagnosis of “GAD antibody-spectrum neurological disorders” and highlight that anti-GAD antibody titers do matter, even though their pathogenicity is still unclear (93; 30).

Stiff-person syndrome. Stiff-person syndrome, first described by Moersch and Woltman in 1956 as “stiff-man syndrome,” is an autoimmune neurologic disorder characterized by progressive muscle rigidity and spasms, mainly in the truncal and proximal leg muscles (92; 11). It has been said to be rare, affecting one in 1 million people, but its precise frequency is unclear. Based on the large number of patients referred to us in the last 30 years (having seen, examined, and personally followed all of them), we believe that it is more common than has been thought but often still unrecognized or misdiagnosed. Stiff-person syndrome is most often seen in patients above the age of 20 with an average age of onset at around 30 to 35 years (36). It is twice as common in women than in men, hence, the reason it was originally labeled “stiff-person syndrome” instead of the original “stiff-man syndrome” (26).

Patients typically present with muscle spasms and stiffness, concurrently in the thoracolumbar paraspinal and abdominal muscles, resulting in difficulties turning and bending (36). When stiffness is severe, it resembles a “statue” or a “freezing”-like appearance; patients often describe that they walk like a “tin-man.” They often have an accompanied anxiety, which, when severe, is misdiagnosed as a primary anxiety disorder; task-specific phobias, especially fear of walking or falling, are also common (36; 04). Symptoms of muscle spasms and stiffness can be precipitated by unexpected stimuli, including sounds like a phone ringing or a siren, sudden touches, and emotional upset. In some cases, these events can cause severe and continuous painful spasms, along with stiffness in the thoracic muscles with breathing difficulties, tachycardia, and hyperhidrosis, a condition we have labeled “status spasticus,” requiring emergency admission for intravenous diazepam (38; 37). Electrophysiological studies have revealed continuous activity of motor unit firing at rest, confirming that stiffness is caused by co-contractions of agonist and antagonist muscles (89; 78; 108). Diabetes mellitus type 1 and other autoimmune diseases, such as vitiligo, pernicious anemia, celiac disease, or thyroiditis can be seen in up to 35% of patients (38; 37; 2022; 87).

In up to 80% of patients with stiff-person syndrome, autoantibodies to GAD are detected in the serum (38; 53). The antibodies may interfere in vitro with GABA production and in vivo with the entire GABAergic system (41; 77), resulting in unbalanced neurotransmission with enhanced excitotoxicity expressed as spasms and stiffness.

The diagnostic criteria for stiff-person syndrome were revised by Dalakas in 2009 to include the following: (1) stiffness of the axial muscles, particularly in the abdominal and thoraco-lumbar paraspinals, leading to hyperlordosis and unprotected falls; (2) superimposed painful spasms and “startle response” triggered by tactile, visual, or auditory stimuli; (3) electromyographic evidence of continuous motor unit activity concurrently in agonist and antagonist muscles; (4) absence of other neurologic findings that may suggest an alternative diagnosis; and (5) positive serology confirmed by immunocytochemistry, Western blot, ELISA, or radioimmunoassay (38). Although these criteria best describe “classic stiff-person syndrome,” it is now clear that some patients with positive anti-GAD antibodies may not exhibit all the typical stiff-person syndrome symptomatology, whereas a few others with clinically typical stiff-person syndrome may be negative for anti-GAD antibodies.

The symptomatology is best highlighted in our longitudinal study of 57 anti-GAD-positive patients with stiff-person syndrome, which represents the largest series of patients examined and followed by the same clinicians every 6 months for a 2-year period to assess disease progression (104). The most common initial symptom was the insidious onset of proximal leg stiffness followed by rigidity in the lumbosacral paraspinal muscles and rigidity in the thoracic and abdominal muscles. When first examined, axial muscle stiffness (truncal and proximal legs), lumbar hyperlordosis, and impaired gait were detected in 68% of the patients, with 28% of them also having various degrees of facial muscle stiffness. About 15% of the patients with typical stiff-person syndrome symptomatology also had ataxia, dysarthria, and dysphagia, overlapping with the cerebellar variant, as described below. This is an important distinction because such patients may not fully respond to immunotherapies (103). Exaggerated reaction to various external stimuli and “startle response” were present in all the patients of this series except two who did not have episodes of muscle spasms. Marked anxiety related to unprotected falls or in anticipation of physically challenging situations was seen in 52 of 57 patients and was a persisting symptom in our longitudinal series. Three patients had refractory depression that coincided with the onset of stiff-person syndrome, whereas 21 patients experienced chronic anxiety combined with intermittently depressed mood. Simple phobias, such as fear of walking in open and crowded places, crossing a street, or taking an escalator were reported by more than 10% of patients, with several also having task-related phobias, such as the fear of timely completing a task, especially when being observed, eg, in the clinic, among crowds, and while speaking or performing in public. Formal neuropsychiatric testing in 10 consecutive patients form this series, however, did not meet DSM-IV criteria for phobic disorder (04). It was felt that the patients perceived their fears and anxiety as realistic, arising from the possibility of falls caused by muscle cramps and stiffness, a rather surprising interpretation as we do not see this phenomenon in other neurologic disorders with spasticity, weakness, and falls. Several patients of our series had been earlier misdiagnosed with conversion or functional disorder because their falls were attributed to avoidant behavior and heightened mental anticipation. Other common misdiagnoses were myelopathies, dystonias, or parkinsonism. Three patients used alcohol as a muscle relaxant prior to diagnosis and eventually became alcohol-dependent; two others used cannabis and stimulants (crystal methamphetamine). Many patients reported muscle pain along with painful spasms, and some had been on narcotics. Neurologists should, therefore, be aware that some patients may manifest concurrent neuropsychiatric symptomatology that, when prominent, necessitates the need for psychiatric or psychological support. At the other end of the spectrum, patients with painful spasms due to functional neurologic disorder misdiagnosed as seronegative, or with very low-GAD titers, stiff-person syndrome have increasingly been referred to us over the years.

Apart from the antibodies against GAD, other antibodies may be detected in patients with stiff-person syndrome.

Autoantibodies against GABAa receptor-associated protein (GABARAP) have been found in about 70% of the patients (102), but the practical value of these antibodies, even with their potential to be pathogenic, has not been further explored. Another autoantibody found in 10% to 15% of patients with stiff-person syndrome is against glycine-a1 receptor (anti-GlyR) (01; 86). Glycine is a key inhibitory neurotransmitter and anti-GlyRa1 antibodies may have a pathogenic role as they recognize extracellular epitopes of the receptor expressed in the spinal cord, brainstem, and cerebellum. These antibodies were first described in progressive encephalomyelitis with rigidity and myoclonus (85; 63). In about 5% of cases, stiff-person syndrome can be a paraneoplastic manifestation associated with antibodies against amphiphysin (39; 49) and in a single case against gephyrin (16). Apart from glycine-a1 receptor, all targeted antigens are predominantly cytoplasmic, and it remains to be investigated whether they can transiently exhibit an extracellular domain during neurotransmission and exocytosis to exert a pathogenic effect (37; 30).

Cerebellar ataxia. Anti-GAD antibody-associated cerebellar ataxias is the second most frequently reported GAD-related neurologic disorder. It affects more women than men, often with comorbid diabetes mellitus type 1 or polyendocrine autoimmunity (106; 62; 69; 103; 12; 74). These patients exhibit gait and limb ataxia, nystagmus, severe dysarthria, and oculomotor dysfunction, usually overlapping with the typical stiff-person syndrome symptomatology, which enhances the overall clinical presentation (66; 103; 107; 53; 30). CSF analysis may show oligoclonal bands, without protein elevation, and intrathecal anti-GAD antibody synthesis (62; 103). The MRI imaging of patients with cerebellar ataxia is normal (103), with only rare instances of mild cerebellar atrophy (62), implying a functional blockade rather than a destructive process (38). In a retrospective study, it was suggested that serum GAD65 antibody titer greater than 500 nmol/L and cerebellar ataxia predicted poor outcome (14).

Whether the antibodies play a role in the pathogenesis of cerebellar ataxias is unclear (90). Studies have shown that a monoclonal GAD65 antibody interferes with GABAergic neurotransmission in brain slice preparations and in vivo elicits in animals neurophysiological and behavioral effects mimicking cerebellar ataxias (82). Intracerebellar administration of IgGs from CSF of patients with GAD-associated cerebellar ataxia impairs cerebellar modulation of motor control and contributes to lack of coordination (13; 83; 84; 67; 68). The anti-GAD antibodies seem to act on nerve terminals of GABAergic interneurons depressing the release of GABA, resulting in hyperexcitability and eventually loss of Purkinje cells with diffuse proliferation of Bergmann glia (64; 97). Furthermore, a human monoclonal GAD65 antibody elicits some pathogenic effects resembling those induced by the patients’ CSF IgG (84; 82).

Limbic encephalitis with GAD antibodies. Autoimmune limbic encephalitis with anti-GAD antibodies clinically presents like the classic autoimmune or paraneoplastic limbic encephalitis (57) with impaired working memory, psychiatric symptoms, seizures, or altered level of consciousness (54). In some patients there are oligoclonal bands in the CSF and intrathecal synthesis of GAD-Abs (81), but the causative role of GAD antibodies is still unclear.

Autoimmune epilepsy. Antibodies against GAD have been found in patients with pharmacoresistant epilepsy, most often presented as temporal lobe epilepsy (75; 42; 80). Patients have normal MRI (96) and higher frequency of autoimmune comorbidities (42; 80). They may present with epilepsia partialis continua or with refractory convulsive and nonconvulsive status epilepticus (94; 22; 72; 116).

In a retrospective series, anti-GAD antibodies were detected in 22% of patients with various epilepsies and a concurrent autoimmune association (100). In another cohort of 233 patients with all types of epilepsy, the percentage of GAD-Abs was only 2.3% (44). However, when considering only patients with focal epilepsy, GAD-Abs were present in 16% of the cases (96), whereas among patients with temporal lobe epilepsy the percentage may be even up to 21.7% (45). Among 80 children with epilepsy, anti-GAD antibodies were the third most common Abs, after antinuclear and anti-voltage-gated potassium channels (45).

In a series of 1510 epileptic patients, three had musicogenic reflex seizures (MRSs), with two of them having GAD-associated epilepsy. Musicogenic reflex seizures have also been reported in a patient with stiff-person syndrome comorbidity. Although this clinical manifestation is extremely rare, musicogenic reflex seizures may be a distinctive type of epilepsy highlighted by anti-GAD antibodies, necessitating the need to test for GAD antibodies in all suspected musicogenic reflex seizures cases, even with normal structural MRI (46). In a retrospective chart review of 16 patients with musicogenic epilepsy, nine tested patients were found to be GAD-antibody positive in both serum and CSF (109). These patients had temporal lobe epilepsy with epileptiform EEG abnormalities captured when seizures were induced by music. Only one of six patients partially responded to immunotherapy, raising doubts as to whether this epilepsy is of immune etiology.

The anti-GAD antibodies, by inhibiting GABAergic pathways, may result in hyperexcitability, which can explain epileptogenesis. At least 5% of patients with stiff-person syndrome have seizures (35; 10), although in our experience the epilepsy in stiff-person syndrome is not refractory but rather easily controlled.

Progressive encephalomyelitis with rigidity and myoclonus. In 1956, progressive encephalomyelitis with rigidity and myoclonus was described by Campbell and Garland and was considered a stiff-person syndrome-spectrum disorder (18). Progressive encephalomyelitis with rigidity and myoclonus is now a distinct syndrome characterized by muscle stiffness, spasms, myoclonus, and brainstem dysfunction with oculomotor abnormalities, dysphagia, gait ataxia (20), prominent autonomic involvement, and depressed level of consciousness. It seems equally present in men and women although in our small series most patients were men. The hallmark of this disorder is the presence of anti-GlyR antibodies. As mentioned earlier, however, up to 15% of patients with stiff-person syndrome with anti-GAD antibodies also harbor low titers of anti-GlyR-Abs (86; 01; 61). An underlying tumor, especially thymoma or lymphoma, can be detected in about 20% of patients with progressive encephalomyelitis with rigidity and myoclonus (121). Another autoantibody that has been detected in four patients with progressive encephalomyelitis with rigidity and myoclonus is the anti-DPPX (10; 53), characterized by diverse symptomatology including prominent gastrointestinal manifestations, seizures, encephalopathy, sleep disturbance, and dysautonomia.

Limited histological data on progressive encephalomyelitis with rigidity and myoclonus have demonstrated inflammatory and microglial changes and cell loss in the pons, medulla, cerebellum, spinal cord, and autonomic ganglia (11). Some patients with progressive encephalomyelitis with rigidity and myoclonus had increased T2 fluid-attenuated inversion recovery signal of spinal cord and brainstem on MRI (20).

Nystagmus and abnormal eye movements. Isolated oculomotor dysfunction is another manifestation in patients with anti-GAD antibodies, characterized mainly by downbeat nystagmus and saccadic intrusions or oscillations but rarely ophthalmoparesis. In our experience, oculomotor dysfunction is not unusual among all GAD-positive patients with stiff-person syndrome but is more prominent in those with cerebellar ataxia (114; 06; 05; 125; 30). The most common isolated GAD-positive oculomotor dysfunction is persistent horizontal or downbeat nystagmus, presumably related to excitability of vestibular nuclei with increased drive to motor neurons of the ocular musculature, resulting in an upward slow phase followed by a quick compensatory downward phase (06; 98; 08).

Clinical vignette

A 62-year-old woman presented with trunk stiffness and rigidity manifested as slowly progressive difficulty turning and bending, slow-step walking, impaired balance, and sudden falls, resulting in fear of walking alone, especially crossing a street or in crowded places (31). Her fears were interpreted as related to anxiety and depression, and she visited a neurologist who diagnosed “rigid parkinsonism.” Brain MRI was normal and a DaTScan was reported as suggestive of extrapyramidal disease. She was prescribed sertraline and carbidopa/levodopa, but there was no benefit. Over the ensuing 12 months, her symptoms worsened; the doses of both carbidopa/levodopa and sertraline were increased, and pramipexole was added, but again there was no benefit. When she came to see us, close to 2 years after symptom onset, there was prominent hyperlordosis with concomitant stiffness of both abdominal and lumbar paraspinal muscles. There was no cogwheel rigidity. She was talkative, with normal articulation, but very anxious, constantly emphasizing her fear to walk alone and her frequent falls, especially in public spaces or with unexpected stimuli. The constellation of these symptoms raised the strong suspicion of stiff-person syndrome. Carbidopa/levodopa was discontinued (over 1 month) and anti-GAD antibodies came back strongly positive at very high titers of greater than 1,000,0000 IU/ml. She also had history of thyroid disease with positive anti-thyroid antibodies. Pramipexole was also stopped. She was started on baclofen 10 mg 3 times daily along with diazepam 10 mg 3 times daily (first dose in the evening, increasing to 3 times daily over 3 weeks). After 6 weeks, she had clearly improved with better mobility, less stiffness, and reduced fear of walking. An increase to diazepam made her sleepy; instead, gabapentin 400 mg 3 times daily was added. She improved further, became able to walk easily without falling and move about freely in open spaces without assistance; she was also able to drive a car again. Sertraline was also discontinued. About 3 to 4 months after therapy, she became functional with minor limitations to the point that additional treatments were not deemed necessary. Because the improvement was clinically satisfactory and the drugs well tolerated, no need for immunotherapy was considered.

Biological basis

The pathogenesis of GAD-associated syndromes is still unclear. Despite the key role of the autoantibodies in defining a rather heterogeneous group of overlapping disorders, it is not yet certain whether GAD antibodies are pathogenic or a marker of aberrantly activated innate and acquired immunity.

Etiology and pathogenesis

Structural and functional features of GAD protein. GAD is the enzyme that catalyzes the decarboxylation of l-glutamate to gamma-aminobutyric acid (GABA), which is the main inhibitory neurotransmission in the central nervous system. GAD exists in two isoforms, GAD65 and GAD67, each encoded by a different gene (43). GAD65 is mostly utilized whenever there is a need for immediate GABA synthesis and release (95; 48). The linear sequence of GAD can be divided into three functional domains: an amino(N)-terminal domain, a middle PLP-binding domain where the catalytic center of the enzyme resides, and a carboxy (C)-terminal domain.

The crystal structure of GAD65 has been informative about its pathogenic potential. GAD65 possesses an accessible surface area for binding, whereas the C-terminal and catalytic loop residues have flexibility and mobility (48). These biophysical properties could explain why in diabetes mellitus type 1 the GAD65 isoform could be autoantigenic.

Immunological profile: Distinction of GAD antibodies between stiff-person syndrome and diabetes mellitus type 1 and significance of GAD antibody titers. Several immunological differences have been observed between stiff-person syndrome and diabetes mellitus type 1. Patients with stiff-person syndrome and other neurologic GAD-spectrum disorders normally have very high anti-GAD titers compared to very low titers seen in diabetes mellitus type 1. Several assays are being used to detect anti-GAD antibodies, including quantitative radio-immunoassays and enzyme-linked immunosorbent assays (ELISA) (106; 120; 35). These assays, initially developed to detect the low titers of GAD antibodies in patients with diabetes mellitus type 1, require adaptations with serial serum dilutions to ensure the accurate detection of high titers as observed in patients with stiff-person syndrome. Other qualitative assays, such as tissue immunohistochemistry, cell-based assays, or line blots, are less sensitive and may only detect structural epitopes of GAD65 antibodies.

Depending on the laboratory and the method used, reference values may be expressed in different units. A major clinical laboratory in the United States uses radio-immunoassays and defines high titers as 0.02 nmol/L or greater. According to their experience, these titers are found in classic stiff-person syndrome (93% positive) and in related autoimmune neurologic disorders, whereas values in patients who have diabetes mellitus type 1 without a polyendocrine or autoimmune neurologic syndrome, usually have titers 0.02 nmol/L or less. Other U.S. and European laboratories use ELISAs where the cut-off for positivity is greater than 5 IU/ml. According to various clinical studies in patients with stiff-person syndrome, titers are considered high when they are above 10,000 IU/ml. The same applies to our own laboratory where we use ELISA. Titers within the range of 5 to 2000 IU/ml are seen in diabetes mellitus type 1; only titers greater than 10,000 IU/ml are associated with a neurologic disorder. In one study, the significance of serum anti-GAD65 antibody titers in connection with true neurologic disease was reconfirmed by setting a cutoff value of 10,000 IU/mL in ELISA based on their specificity in concurrent testing by immunohistochemistry and cell-based assay. High (greater than 10,000 IU/ml) titers conferred specificity for an autoimmune neurologic disease in 94% of the patients, including those with stiff-person syndrome, cerebellar ataxia, chronic epilepsy, limbic encephalitis, or overlapping conditions; in contrast, lower concentration antibodies were seen in a broad and heterogeneous spectrum of disorders (93). The high titers were also associated with measurable anti-GAD antibodies in the CSF.

Collectively, practicing neurologists should be aware that anti-GAD antibody titers do matter (29; 30; 93). If high (greater than 10,000 IU/ml), they are diagnostic for a true GAD antibody-spectrum disorder, necessitating immunotherapy when there is significant disease severity. Lower (less than 10,000 IU/ml) titers are associated with atypical or nonspecific neurologic disorders requiring further investigation and testing the CSF for GAD antibodies. Very low titers (less than 2000 IU) are typically seen in diabetes mellitus type 1 or are of unclear significance. Importantly, GAD antibodies can also be detected within the various IVIg preparations as part of the natural antibody repertoire derived from normal donors. Consequently, anti-GAD antibodies can be detected in the serum of patients receiving IVIg (40). However, there is no association between GAD-Ab titer and disease severity, and no significant titer reduction has been documented after immunotherapies with either IVIg or rituximab based on two controlled studies we have performed (34; 33; 30).

Patients with diabetes mellitus type 1 harbor antibodies directed against conformational epitopes exclusively located in the PLP- and C-terminals domains (76; 47). In contrast, patients with stiff-person syndrome predominantly recognize linear epitopes in all three domains on GAD65 and GAD67 (17; 36; 70), and specifically in the first 100aa, which constitute the regulatory sequence in the N-terminal GAD65 domain that is not recognized by the diabetes mellitus sera (73; 124; 36; 99; 101; 118; 21). The GAD Abs in stiff-person syndrome, therefore, exhibit a different epitope pattern of antibody reactivity with distinct biological effects compared to diabetes mellitus type 1.

Whether different epitope patterns exist among GAD-related syndromes is unclear. In one study, GAD-Abs from patients with limbic encephalitis were more likely to recognize epitopes in the N-terminal domain, compared to those with stiff-person syndrome, cerebellar ataxia, or epilepsy, with the latter showing more reactivity to the C-terminal domain of the enzyme (82; 79). However, in our study of 27 patients with diverse GAD-related syndromes, we found no differences in epitope specificities, except in three patients with epilepsy (50). Accordingly, the current data cannot explain the diverse clinical presentation based on different epitope-binding patterns.

Intrathecal synthesis of GAD antibodies. There is strong evidence that in stiff-person syndrome, as well as in patients with the other GAD-associated neurologic syndromes, there is intrathecal synthesis of GAD antibodies. Using the Link formula, as proposed by Dalakas and colleagues, the ratio of GAD antibody concentration in the CSF to that in the serum was divided by the ratio of albumin concentration in the CSF to that in the serum; values greater than 1 have been indicative of robust intrathecal synthesis (35; 103). In clinical practice, when the serum GAD antibody titers are above 10,000, GAD antibodies are also detected in the CSF (35; 93); in these circumstances, a diagnostic lumbar puncture may not be necessary, especially in patients with stiff-person syndrome, where the stiffness in the lumbar paraspinals requires a radiology-guided puncture (30).

The demonstration, however, of intrathecal GAD antibody synthesis comprises the strongest evidence linking a neurologic syndrome to autoimmunity, as correctly suggested by Graus and colleagues (53). In clinical practice, testing the CSF for GAD antibodies is essential in patients with serum titers below 10,000 and in patients with seronegative GAD-spectrum disorder, especially those with encephalitis, and in patients with a seemingly functional disorder resembling stiff-person syndrome symptomatology.

Pathogenicity of GAD-Abs: the uncertainty of experimental animal models. Whether anti-GAD antibodies are pathogenic, considering that they target an intracellular antigen, or are only a disease marker remains unanswered. Rats treated intracerebroventricularly with stiff-person syndrome-IgG showed a stiffness-like behavior, a decline of motor function as measured by time on the Rotarod test, and a decrease in forelimb grip strength as compared to control IgG-infused rats. Additional studies of passive transfer of GAD-Abs from patients into rats or mice have shown continuous motor activity with repetitive muscle discharges and abnormally enhanced reflexes with increased excitability of anterior horn cells (83; 84; 59). In vitro, GAD antibodies interfere with GABA production (41; 65; 66), whereas in vivo, they affect the function of GABAergic neurons and interfere with GABA synthesis, resulting in impaired inhibitory neurotransmission without causing structural brain changes (21; 82). These experimental data are consistent with normal MRI imaging and the reduction of GABA level found with MRS spectroscopy in the brain of patients with stiff-person syndrome (77) and in their CSF (34), suggesting that these antibodies may be of some significance in inducing a neuronal functioning blockade but not neuronal destruction (30). Whether these effects are related to anti-GAD or other synaptic antibodies directed at different synaptic antigens is unclear. Stereotactic injection of GAD-Abs into the hippocampus of rats in vivo did not alter spontaneous and evoked GABAergic synaptic transmission (113; 58). However, in contrast to animals treated with anti-GAD antibodies, animals treated with IgG anti-amphiphysin antibodies, either intraperitoneally (112) or intrathecally (51), have exhibited stiffness-like behavior.

The data from experiments conducted in cultured neurons are also inconsistent. Hippocampal cultured neurons treated with sera from epileptic GAD-positive patients showed an increase of postsynaptic inhibitory potentials compared to negative controls (119). Further, when rat cerebellar slices were exposed to serum or CSF from patients with stiff-person syndrome or cerebellar ataxia, a decrease of postsynaptic inhibitory currents of Purkinje cells was observed, compared to GAD-negative sera form ataxic patients (65; 91; 114). Some studies have also shown epitope-dependent pathogenic actions of GAD-Abs in histological brain sections and in vivo preparations (82; 110), whereas others showed lack of internalization into hippocampal cultured rat neurons (55). It remains unclear how GAD-Abs can cause the GABAergic dysfunction in stiff-person syndrome if not internalized into neurons. The possibility that antigens during synaptic transmission transiently expose extracellular epitopes, recognized by the immune system, remains hypothetical.

Circulating GAD-reactive B cells that can differentiate into antibody-producing cells also have been detected in the peripheral blood and bone marrow of patients with GAD-Ab-associated neurologic syndromes. Interestingly, the presence of GAD was not required for induction of GAD-antibody-producing cells, and GAD Ab production by stimulated peripheral blood cells did not correlate with GAD Ab serum levels, suggesting an additional source of GAD Abs. This study implied that targeting both memory B cells (ie, with rituximab) and plasma cells (ie, with bortezomib) might be a potential treatment option (115).

Finally, cytotoxic T lymphocytes have been found in histological preparations of temporomesial tissue from patients with pharmacoresistant epilepsy associated with GAD antibodies who underwent temporal lobectomy (19). These T cells may release perforin and granzyme, leading to necrosis, apoptosis, or electrical silencing of the respective neurons. It was suggested that GAD-associated epilepsy may be caused by potentially neurotoxic CD8+ cells against GABAergic interneurons (24).

Epidemiology

No formal epidemiological studies have been done. The stated prevalence of stiff-person syndrome as “one case per million people” (88) was not based on true epidemiological data; rather, it was a remark to denote that the disease is very rare. Considering that the GAD spectrum disorders are not limited to stiff-person syndrome, as outlined earlier, their incidence requires a systematic study. A relevant example is the association of GAD antibodies with drug refractory temporal lobe epilepsy; since first noticed in 1998 (52), more than 200 cases with chronic pharmaco-resistant epilepsy have been now reported. The importance of being cognizant in recognizing these syndromes in a timely manner has been repeatedly emphasized as we consider GAD-positive patients to have a potentially treatable autoimmune disorder with early immunotherapy initiation (25).

Differential diagnosis

See discussion in the following articles: stiff-person syndrome, nonparaneoplastic autoimmune cerebellar diseases (for discussion of cerebellar ataxia), anti-LGI1 encephalitis (for discussion of limbic encephalitis with GAD antibodies), and progressive encephalomyelitis with rigidity and myoclonus and glycine receptor antibodies (for discussion of nonparaneoplastic autoimmune cerebellar diseases).

Diagnostic workup

The diagnostic work-up for stiff-person syndrome requires clinical judgement based on specific clinical criteria, electrophysiological assessment, the presence of other autoimmunities or diabetes type 1, and, importantly, a search for the relevant autoantibodies (38; 37; 30; 36). The main clinical diagnostic criteria include the following: (1) stiffness of the axial muscles, particularly in the abdomen and thoracolumbar paraspinals, leading to hyperlordosis, abnormal gait, and falls; (2) superimposed episodic spasms precipitated by sudden movement, emotional upset, or unexpected auditory, visual, or aesthetic stimuli; and (3) absence of other neurologic findings that may suggest an alternative diagnosis, such as absence of brainstem, pyramidal, extrapyramidal, and lower motor neuron signs, sphincteric and sensory disturbance, and cognitive involvement (53). Functional disorders should be also considered, especially in seronegative patients or those with low anti-GAD titers. Electrophysiological data are helpful in the diagnosis by demonstrating continuous motor unit firing activity at rest from agonist and antagonist muscles concurrently. Antibody testing and precise determination of anti-GAD antibody titers is fundamental, with the cut-off positivity titers greater than 10,000 IU/ml titers by ELISA. The main diagnostic challenge remains seronegative stiff-person syndrome, which represents close to 20% of patients. Adherence to strict clinical criteria, neurophysiologic testing, and consideration for neuropsychiatric assessment to exclude a functional disorder are critical. An empirical trial with diazepam is often used, but in our view, it does not ensure diagnostic accuracy because it cannot differentiate an organic from a functional disorder.

Although not considered diagnostic because routine brain MRI is normal or nonspecific, some specialized imaging techniques have been used as a tool to explore and improve our understanding of the syndrome. Most informative has been the magnetic resonance spectroscopy study performed in a large series of well-characterized GAD-positive patients with stiff-person syndrome (77). This study showed prominent and significant reduction in GABA level, predominantly in the sensorimotor cortex and, to a lesser degree, in the posterior occipital cortex, indicating involvement of the inhibitory GABAergic pathways, which is consistent with the reduced GABA levels observed in the cerebrospinal fluid of the same patients (35). In another patient series, the morphological MRI of 26 patients (T1-weighted and FLAIR-weighted images) was analyzed at the initial stage of diagnosis. Using measures to assess brain cortical thickness, cerebellum thickness, and hippocampal volume, a peculiar profile of atrophy was found. Patients with anti-GAD antibodies were reported to have atrophy in the temporal and frontal lobes and a focal cerebellar atrophy of the V-lobule, independently of the anti-GAD phenotype, suggesting that the different neurologic anti-GAD phenotypes should be considered as a continuum (23). Because the study group was very heterogenous and the routine brain MRI in these patients was normal, the information is of unclear significance.

In a retrospective study, fluorodeoxyglucose positron emission tomography (FDG-PET) findings in the brain and muscles of patients with stiff-person syndrome (SPS) spectrum disorders were reported (122). As expected, nonspecific metabolic abnormalities were seen in various cortical regions in both stiff-person syndrome and cerebellar ataxia, whereas 62% also demonstrated muscle hypermetabolism. Neither brain nor muscle metabolism correlated with functional outcomes or treatments (122). Although the impetus of the study was to look for hidden neoplasms, FDG-PET does not have any diagnostic or disease monitoring value for patients with stiff-person syndrome spectrum disorders.

Management

Patients with stiff-person syndrome experience severe anxiety and depression due to phobias of falling and completing even simple physical tasks. Patients with significant symptoms that do not improve concurrently with the physical symptomatology need clinical and psychological support both at home and at work. Their phobias often lead to depression and sometimes to addiction to drugs such as benzodiazepines or narcotics, highlighting the need for multifactorial care.

For stiff-person syndrome and GAD-associated disorders, two strategies of treatment can be implemented: symptomatic or immunologic, either independently or in combination, depending on symptom severity (117; 25; 38; 31). Symptomatic relief is the first treatment strategy, and it is often achieved with agents that enhance GABAergic transmission, such as benzodiazepines. The commonest therapeutic option is diazepam, a GABAA agonist. This drug can help most patients, although the high doses sometimes required cannot be tolerated and may lead to addiction. Other similar compounds include clonazepam, alprazolam, lorazepam, and temazepam. The second category of drugs are antispasticity agents, such as GABAB agonists; because of better tolerance, we have been using then as first-line therapy in lieu of benzodiazepines. The most effective among them is baclofen, considered the second most useful drug after diazepam, now even replacing it in order. Sometimes high doses are required to induce improvement, causing some cognitive effects. Antiepileptic drugs that enhance the brain’s GABAergic transmission also improve symptoms, either alone or in conjunction with the baclofen and benzodiazepines. In our experience, the most helpful agents in this family are gabapentin and vigabatrin, which act by inhibiting GABA-transaminase and increasing GABA. Tiagabine, an inhibitor of GABA reuptake, and levetiracetam, which facilitates inhibition of GABAergic transmission, may offer benefits if well tolerated. Other drugs include tizanidine, a centrally acting alpha2 adrenergic receptor, and dantrolene, a muscle relaxant.

If the above agents do not offer a satisfactory benefit, we proceed to immunotherapy. The most widely used agent in this category is intravenous immunoglobulin (IVIg) after its proven efficacy. In a randomized, double-blinded, placebo-controlled crossover trial we conducted in GAD-positive stiff-person syndrome, IVIg resulted in significant improvements in objective stiffness parameters and activities of daily living (35). The duration of efficacy after each monthly IVIg infusion ranges from 4 to 12 weeks in most patients. IVIg remains the only immunomodulatory therapy with proven benefit in stiff-person syndrome. Subcutaneous immunoglobulin may be also an option in patients with poor venous access or when there is a demonstrable early wearing off effect to ensure sustained benefit (27). Plasmapheresis has been of limited and transient benefit, and we do not routinely use it despite some anecdotal case reports (60). Corticosteroids are surprisingly ineffective in our experience with many patients, although one anecdotal report had shown limited benefit (88). Furthermore, triggering or exacerbating diabetes is a serious consideration that limits further its use in stiff-person syndrome. Of paramount importance is the control of diabetes, which requires insulin most of the time and, if uncontrolled, seems to worsen the neurologic symptomatology.

Since the impressive improvement noted with IVIg in the 3-month controlled trial conducted more than 20 years ago (35), IVIg has been used monthly, but rather liberally, for the chronic management of stiff-person syndrome without evidence that it sustains stability over time or arrests disease progression. Importantly, the commonly encountered conditioning effect seen in 40% of all patients on chronic IVIg therapy who continue treatments even without exhibiting objective benefits due to the fear that it may worsen without it (32) has not been taken into account in patients with stiff-person syndrome, generating concerns of chronic IVIg overuse. A large study has explored the benefits of IVIg as chronic maintenance therapy in 36 GAD-positive patients with stiff-person syndrome treated over a 10-year period (30). The long-term effects of IVIg were assessed based on improvement in mRS scores, physician-assessed stiffness, balance, and gait and were confirmed with dependency trials prolonging infusion frequency or wearing-off effects in between doses. It was observed that 67% of patients had a clinically meaningful response over a median 3.3 (1-18)-year period, with improved gait, posture, and balance and decreased stiffness, spasms, and startle response. Importantly, 37.5% of the responders had long-term stability without disease progression for a 4.3 (1–18)-year period. However, although 29.1% continued to experience improvement, they also showed diminishing benefit after a 3.3-year period due to disease progression, highlighting the need for more effective long-term therapies (30). Of interest, 12.5% exhibited a conditioning effect after 5 (2–18) years, emphasizing the need to perform periodic dependency trials to avoid overuse, as noted in other neurologic disorders on chronic IVIg therapy (32).

Immunosuppressive agents, such as azathioprine, methotrexate, cyclophosphamide, and mycophenolate mofetil, are equally disappointing in our experience in spite of rare anecdotal reports (60; 09). The most useful drug in this category is rituximab. A randomized controlled trial we conducted in patients with stiff-person syndrome demonstrated lack of efficacy of rituximab compared to placebo owing to a strong placebo effect (33). In this series, however, seven patients improved, and four of them with severe disease demonstrated meaningful to impressive improvements. On this basis, we believe rituximab is a useful drug for a subset of patients who have failed therapies with GABA-enhancing drugs and IVIg. It should be stressed that anti-GAD antibody titers may drop but not at a statistically significant level, and the titers do not correlate or predict improvement.

Some patients with stiff-person syndrome who failed conventional immunosuppressive therapy have experienced benefit after autologous hematopoietic stem cell transplantation (auto-HSCT). In one small study, three patients with stiff-person syndrome and one with progressive encephalomyelitis with rigidity and myoclonus were initially treated with cyclophosphamide (Cy) 2 g/m2 + granulocyte-colony stimulating factors (G-CSF) and then conditioned with Cy 200 mg/kg + plus anti-thymocyte globulin (ATG) followed by auto-HSCT. All patients tolerated the procedure well and showed improved physical performance. One patient’s walking distance improved from 300 meters to 5 miles, and one other’s ambulation improved from being confined to a wheelchair to being able to walk with a frame. Two patients became seronegative for anti-GAD antibodies and normalized their neurophysiological abnormalities. Although auto-HSCT was thought promising, a large study aiming at 40 patients was terminated early after recruiting 23 patients because of lack of efficacy or transient benefits, taking into account potential serious complications (15; 71). One of the many limitations of this study, as pointed out, was the enrollment of patients with advanced disease, drawing attention to the need for using objective measurements of quantifying stiffness with validated scales, taking into account a strong placebo effect if a new hematopoietic stem cell transplantation (HSMT) trial in stiff-person syndrome patients is considered (28).

Conclusions. The pathogenetic potential of anti-GAD antibodies remains unsettled. The high rate of intrathecal synthesis of anti-GAD-specific IgG in some patients signifies B-cell in-situ stimulation in the CSF compartment and possibly in-situ action of antibodies within the CNS, but it is unclear what drives their CNS persistence (12). The reason for the clinical heterogeneity among GAD-antibody-associated syndromes is also uncertain. It was thought to be related to the variable susceptibility of GABAergic neurons to anti-GAD or other still unidentified autoantibodies (107; 74; 90), but data from two independent studies indicate that all anti-GAD antibodies from the various hyperexcitability syndromes recognize the same dominant GAD epitope (68; 82). The uncertainty of whether anti-GAD antibodies are simply disease markers or pathogenic has an impact on therapeutic approaches. First-line therapies are symptomatic using GABA-enhancing drugs, followed by IVIg as the only proven immunotherapy effective in patients with stiff-person syndrome who do not optimally respond to first-line drugs (30). IVIg also provides sustained benefit in 67% of patients over a median 3.3 (1–18)-year period, but in 29%, the benefits are diminishing after a mean 3.3-year period due to disease progression (30), highlighting the need for more effective long-term therapies. It is anticipated that a better understanding of the immunopathogenesis of the syndromes will greatly inform our therapeutic approaches.

Outcomes

Not all the neurologic syndromes associated with GAD antibodies adequately respond to present immunotherapies. Whether such variability relates to the fact that these antibodies target a nonpathogenic intracellular antigen is unclear. Of interest, patients with progressive encephalomyelitis with rigidity and myoclonus with GlyR antibodies, which are pathogenic based on in vivo models (105), seem to have a better response to immunotherapies based on very small anecdotal series (20).

In a study of 57 patients with stiff-person syndrome followed for greater than 5 years, 80% exhibited some degree of disability progression (104). In patients with cerebellar ataxia and GAD antibodies, the response to immunotherapy is clearly suboptimal. In a retrospective study of 25 patients with a median follow-up of 5.4 years, only 11 (44%) had a good functional status at the last visit (07). The prognosis of epilepsy associated with GAD antibodies is also variable. Based on our experience, some patients have excellent clinical response, but in one series, only occasional patients had clinical improvement with immunotherapy or epilepsy surgery (19).

Special considerations

Pregnancy

For pregnancy considerations in patients with stiff-person syndrome see discussion in article titled Stiff-person syndrome. For nonparaneoplastic autoimmune cerebellar diseases, see discussion in Cerebellar ataxia. For antibody-mediated epilepsies, see discussion in Autoimmune epilepsy.

Anesthesia

For anesthesia considerations in patients with stiff-person syndrome see discussion in article titled Stiff-person syndrome. For nonparaneoplastic autoimmune cerebellar diseases, see discussion in Cerebellar ataxia. For antibody-mediated epilepsies, see discussion in Autoimmune epilepsy. For progressive encephalomyelitis with rigidity and myoclonus and glycine receptor antibodies, see discussion in Progressive encephalomyelitis with rigidity and myoclonus.