HIV-associated neuromuscular weakness syndrome

Objective: To investigate progressive, severe neuromuscular weakness associated with lactic acidosis in some HIV-infected patients after exposure to nucleoside reverse transcriptase inhibitors (NRTI).

Methods: HIV-associated neuromuscular weakness syndrome (HANWS) was retrospectively identified and classified based on the level of diagnostic certainty: possible (progressive weakness owing to neuromuscular disease), probable (progressive neuromuscular weakness with documented exclusion of confounding causes), or definite (progressive weakness and electrophysiological or pathological evidence of neuromuscular pathology).

Results: Of 69 patients identified with HANWS, 27 had definite HANWS, 19 probable, and 23 possible. In 44 patients with documented follow-up, 16 required intubation and nine died. There was a marginal association between death and hyperlactatemia (P = 0.061). At onset of neurological symptoms, 68 were receiving antiretroviral therapy, including stavudine for 61 (89.7%). Serum lactate level was elevated (> 2.2 mmol/l) in 30/37 (81%), with a trend towards an association between hyperlactatemia and stavudine usage (P = 0.087). In 25, neurological symptoms occurred after antiretroviral therapy discontinuation (median, 14 days). Electrophysiological studies (n = 24) indicated sensorimotor neuropathy in 20 patients and myopathy in three. Nerve biopsy (n = 9) revealed axonal degeneration and inflammation in three, mixed axonal and demyelinating lesions in three, and primary axonal neuropathy in three. Of 15 muscle biopsies, three revealed inflammation and four mitochondrial abnormalities.

Conclusions: A severe neuromuscular weakness syndrome may occur in HIV-infected individuals. The association with hyperlactatemia and NRTI exposure supports mitochondrial toxicity as a pathogenesis. In some, the onset of neurological symptoms lagged significantly after discontinuation of antiretroviral therapy, suggesting that different etiological mechanisms may underlie these cases.

Introduction

Metabolic and neurological disorders are common complications of HIV infection and antiretroviral (ARV) therapy [^1-3]. Hyperlactatemia may occur in association with HIV infection, particularly in the setting of therapy containing a nucleoside reverse transcriptase inhibitor (NRTI). The frequency of hyperlactatemia varies widely among studies [^4-6] and has not been found to be a useful clinical marker for prediction of development of lactic acidosis syndrome (LAS). The use of NRTI appears to confer greater risk for the development of LAS [^4,5], suggesting that prolonged toxicity, presumably to mitochondria, leads to this syndrome.

The Food and Drug Administration (FDA) and several case series have reported a 'rapidly ascending neuromuscular weakness syndrome' associated with LAS in HIV-infected individuals [^7-9]. In the majority of these patients, dramatic motor weakness develops over days to weeks, resembling Guillain-Barré syndrome, and leads to respiratory failure and death in some patients. Diagnostic information from a limited number of patients has revealed evidence of severe axonal neuropathy. Systemic symptoms include nausea, vomiting, weight loss, abdominal distention, hepatomegaly, and lipoatrophy. From FDA surveillance, 22 of 25 patients developed this syndrome in association with stavudine (d4T) therapy [⁷], but muscle weakness worsened even after ARV discontinuation. These reports resulted in a 'Dear Healthcare Provider' letter and change in the d4T label. The present study reports on 69 HIV-infected individuals with progressive neuromuscular weakness, with emphasis on the electrophysiological, histological, and metabolic features of this syndrome.

Methods

Cases were identified retrospectively from Medwatch forms submitted to the FDA through the FDA Adverse Event Reporting System (AERS) and from collaborating authors. The FDA database was searched using the following search terms: neuromuscular weakness and lactic acidosis/symptomatic hyperlactatemia. These events were required to occur within 4 to 5 weeks of each other, in order to meet the FDA's case definition of 'ascending neuromuscular weakness syndrome'. The Medwatch forms were reviewed to supplement and confirm data present in the AERS database.

Collaborative authors identified individual patients based on the following criteria: new onset of limb weakness with a neuromuscular cause in an HIV-infected individual, with or without sensory involvement, either acute (1-2 weeks) or subacute (> 2 weeks) and affecting either the lower limbs or both lower and upper extremities. HIV-associated neuromuscular weakness syndrome (HANWS) was classified as possible, probable, or definite. Possible HANWS was progressive weakness with clinical features consistent with peripheral nerve or muscle disease, with confounding causes of weakness present or without documentation of medical evaluation to exclude such conditions. Probable HANWS was appropriate clinical features, with a documented medical and neurological evaluation excluding other confounding causes of weakness. Definite HANWS was classified as clinical features of probable HANWS with electrophysiological or pathological confirmation of neuromuscular pathology. An exploratory analysis focused on estimation. Because of the exploratory nature of the study, potential associations were identified using a significance level of 0.10, without adjustment for multiple comparisons.

Results

Characteristics of the cohort

Sixty-nine patients with HANWS were identified: 32 (46%) women, 34 (49%) men, and gender was unspecified in three. There were 27 patients with definite, 19 with probable, and 23 with possible HANWS. The groups did not differ in age, serum lactate, bicarbonate level, arterial pH, CD4 cell count or plasma HIV RNA (Table 1).

Table 1

Systemic symptoms, including nausea, vomiting, and abdominal pain, were present in 38 (56%) patients, of whom 24 (63%) had documented hyperlactatemia (venous lactate > 2.2 mmol/l). Of the 30 (43%) subjects overall with hyperlactatemia, 19 (63%) had electrophysiological or pathological evidence of neuromuscular pathology: 16 (84%) with neuropathy, and three (16%) with myopathy. Of seven subjects with normal lactate levels, two had demyelinating neuropathy.

The most commonly used ARV agent at the time of presentation was d4T, which was taken by 61/69 (88%) patients in the total cohort and 24/27 (89%) patients with definite HANWS (Table 2). The median duration of d4T use was 10.5 months (range, 1-71 months). Dosage and adherence information was not available. The median lactate level of those on d4T (5.30 mmol/l; n = 33) was higher than in those not on d4T (2.55 mmol/l; n = 4) (P = 0.087). There was no association between lactate level and the duration of d4T use.

Table 2

Sensory symptoms (paresthesias, dysesthesias, and numbness) were reported in 22 (32%) patients. Eight (12%) had bulbar symptoms, including facial nerve palsy, eye movement abnormalities (e.g., ophthalmoplegia, nystagmus, ptosis), dysphagia, and dysarthria. Areflexia was present in 12 (17%). By definition, all patients had limb weakness. The rate of progression of weakness ranged from 1 to 200 days. Of the definite cases with adequate data, weakness was acute (≤ 2 weeks) in 14 patients and subacute in eight (range, 17-60 days). Neurological symptoms occurred after ARV drug discontinuation in 25 patients (median, 14 days), prior to drug discontinuation in 14 (median, 19.5 days), and on the same day as discontinuation in eight. Of 22 patients with LAS symptoms, 11 had ARV therapy discontinued on the day of presentation with these symptoms, while seven had ARV drugs continued for a median of 30 days (range, 5-90). Four patients had ARV therapy discontinuation prior to presentation of systemic symptoms.

Various treatments were provided for the metabolic and neurological symptoms, including intravenous immunoglobulin in 12 (17%), vitamins B₁ and B₁₂ in 10 (14%), corticosteroids in nine (13%), carnitine or acetylcarnitine in eight (11%), coenzyme Q₁₀ in three (4%), and plasmapheresis in three (4%). Of 44 patients with documented follow-up, 16 (36%) had improvement of their neurological condition. The recovery period ranged from 2 to 17 months (median follow-up, 3.75 months), and the degree of recovery varied from mild improvement to complete mobility. Nineteen patients had substantial residual neurological deficits, marked by severe limb weakness, even after a prolonged period (median follow-up, 3.5 months; range, 1-12). Sixteen patients had respiratory failure requiring ventilatory support. Nine (16%) patients died (median, 30 days; range, 3-104), of whom six had elevated lactate levels and two had definite HANWS. Mortality was associated with lactate level (death, 10.08 mmol/l; survival, 4.40; P = 0.061). Exploratory multivariate modeling revealed a marginal association between mortality and lactate levels > 6 mmol/l [estimated odds ratio (OR), 12.4], and corticosteroid use (estimated OR, 24.3).

Definite HIV-associated neuromuscular weakness syndrome

Nerve conduction and electromyography studies in 24 patients revealed sensorimotor polyneuropathy in 20 (Table 3). Findings were primarily axonal in 13, demyelinating in two, mixed in four, and undefined in one. One patient had polyradiculopathy; two had myopathy, and one had mixed neuropathy and myopathy. Electrophysiological results for six patients are presented in Table 4. Nerve biopsy in nine subjects paralleled electrophysiological findings, with prominent axonal degeneration in three and demyelinative changes in three. Three nerve biopsies revealed inflammatory infiltrates (endoneurial in two and unspecified in one).

Table 3

Table 4

Muscle biopsy in 15 patients indicated generalized myofiber atrophy in three and inflammatory infiltrates in three (Fig. 1A). Four muscle biopsies revealed evidence of mitochondrial dysfunction, including ragged red fibers (Fig. 1B), abnormalities in respiratory chain enzymes, and mitochondrial DNA depletion. The patient with the latter findings (Patient 8, Tables 3 and 4) was the subject of a previous case report [¹⁰], describing an infant presenting with profound motor delay, hypotonia, and areflexia, with lactic acidosis. Muscle biopsy revealed poorly defined mitochondria, with disorganized cristae and dense irregular granules. There was 62-89% reduction in the activities of four respiratory chain enzymes on spectrophotometric muscle analysis, and a 79% depletion of muscle mitochondrial DNA on quantitative Southern blot analysis. These findings normalized after cessation of ARV therapy (didanosine, zidovudine, and nelfinavir). Two index cases are described.

Fig. 1

Case 1 (Patient 16)

Patient 16 was a 25-year-old HIV-infected woman who developed insidious progression of a painful distal neuropathy following 6 months of treatment with d4T, didanosine, and lopinavir/ritonavir (Tables 3 and 4). She also had abdominal pain and vomiting, for which she was hospitalized. Venous lactate level was 7.4 mmol/l (normal, < 2.2 mmol/l), CD4 cell count was 85 × 10⁶ cells/l, and plasma HIV RNA was 1.2 × 10⁶ copies/ml. She had acalculous cholecystitis, with mildly elevated hepatic transaminases. ARV drugs were discontinued and this was followed by improvement in gastrointestinal symptoms. During the next 3 weeks, she experienced progressive severe weakness.

On readmission, she had marked psychomotor retardation, inattention, and poor memory. There was end-gaze nystagmus in all directions and left-gaze paresis. She had profound proximal muscle weakness with absent reflexes. She needed assistance to stand and was unable to walk. Sensory examination revealed a moderate, symmetrical stocking-glove pattern reduction to all modalities. Vital capacity was 1.2 litres, with a negative inspiratory force of -28 cmH₂O.

Blood chemistry was normal, except for aspartate transaminase (53 U/l). Venous lactate was 5.7 mmol/l and creatine kinase was normal. Cerebrospinal fluid analysis and magnetic resonance imaging of the brain with gadolinium were normal. Nerve conduction studies revealed normal motor nerve values, with absent sensory nerve action potentials. Electromyography revealed fibrillation potentials and brief, small-amplitude motor unit action potentials, consistent with irritative myopathy. Sural nerve biopsy showed a prominent decrease of large and small myelinated fibers, with axonal degeneration. Muscle biopsy revealed moderate myofiber size variation, with scattered regenerating and degenerating fibers, and rare angulated atrophic fibers. There was also myofiber vacuolation, with ragged red fibers, and reduced-to-absent cytochrome oxidase staining (Fig. 1).

She had progression of limb weakness, with stable respiratory function. Treatment included intravenous immunoglobulin and methylprednisolone, thiamine, folate, multivitamins, riboflavin, and coenzyme Q. She was restarted on lopinavir/ritonavir and amprenavir. Within 2 weeks, she had marked improvement in eye movements, respiratory function, and limb strength. Over subsequent months, strength improved to allow independent ambulation.

Case 2 (Patient 21)

Patient 21 was a 39-year-old HIV-infected woman, with a CD4 cell count of 167 × 10⁶ cells/l and HIV RNA < 40 copies/m who was started on d4T, lamivudine, and indinavir (Tables 3 and 4). After 1 year, she developed nausea, vomiting, and abdominal pain. Ten days later, she developed slurred speech and generalized weakness, for which she was hospitalized. Neurological examination was normal. Head computed tomography, electromyography, carotid duplex ultrasound, and cerebrospinal fluid analysis were normal. The gastrointestinal complaints persisted, prompting change of indinavir to nelfinavir; after 2 weeks, all ARV medications were suspended. After a further 6 days after, she presented with complaints of anorexia, weight loss, abdominal pain, fatigue, and dyspnea, in addition to persistent nausea and vomiting. She also complained of arm and leg numbness and tingling, which had been present for the past 2 weeks. Admission laboratory results revealed serum lactate 12.5 mmol/l, bicarbonate 10 mEq/l (10 mmol/l), and anion gap 28. Arterial blood gas analysis on room air showed pH 7.16, bicarbonate 2.7 mEq/l, and partial pressure of 8 mmHg for carbon dioxide and 145 mmHg for oxygen. She was treated with intravenous fluids and bicarbonate for lactic acidosis. Over the following days, she had episodes of tachypnea, which did not require intubation, tachycardia, and intermittent hypotension. Peak serum lactate was 19.1 mmol/l.

She had profound limb weakness early after admission. Neurological examination revealed severe weakness in all extremities; decreased pinprick, vibration, and proprioception with a distal predominance; and generalized areflexia. Nerve conduction studies revealed evidence of generalized, severe, mixed axonal and demyelinating sensorimotor polyneuropathy. She was treated with intravenous immunoglobulin. Her systemic symptoms improved, with reduced nausea and vomiting, and improvement in metabolic values. Lactate levels declined to 2.4 mmol/l. However, her neurological status remained unchanged, despite 10 days of intravenous immunoglobulin infusion. On the third week of hospitalization, she was discharged to a rehabilitation center, unable to ambulate independently.

Discussion

This study reports on a series of 69 HIV-infected patients with progressive weakness and metabolic abnormalities: HANWS. Most patients with electrophysiological or pathological evidence of nerve or muscle disease (i.e., definite HANWS) had hyperlactatemia (20/27), and the mortality rate was 13%.

Neurological manifestations were variable, reflecting a spectrum of neuromuscular pathology, including a rapidly progressive sensorimotor polyneuropathy associated with areflexia, resembling the 'ascending neuromuscular weakness syndrome' reported by Marcus et al. [⁷] of the FDA. Other patients developed subacutely progressive proximal muscle weakness, and elevated serum creatine kinase, consistent with a myopathic process. Electrophysiological and pathological findings confirmed primary pathology of peripheral nerve, muscle, or both in most subjects with studies performed. While several patients had features of primary demyelination, consistent with the most common form of Guillain-Barré syndrome (i.e., acute inflammatory demyelinating polyneuropathy), most had primary axonal pathology. This disorder resembles the axonal form of Guillain-Barré syndrome, which has a worse prognosis than the demyelinating type [¹¹].

Hyperlactatemia associated with HIV infection and ARV therapy has been widely reported and is a life-threatening disorder [^6,12,13]. The frequency of hyperlactatemia has been variable among series and is associated with prolonged use of NRTI therapy [^4,14]. The neurological features associated with hyperlactatemia and LAS are not well characterized. Several authors have provided limited neurological data in the context of the systemic and metabolic abnormalities of LAS [¹²]. Only one patient has had description of a detailed neurological examination, and few patients have had electrophysiological or pathological data reported [^6,9,12,15,16]. The results support a severe axonal neuropathy in most patients, although myopathic features were also noted on muscle biopsy in several others [^6,12,15,16].

The pathophysiological mechanism underlying NRTI-associated LAS is not clear. Mitochondrial toxicity likely explains at least some components of this syndrome [^17,18]. Cote et al. [¹⁷] reported a reduced ratio of mitochondrial to nuclear DNA in HIV-infected subjects with NRTI-associated LAS. The relationship of HANWS and LAS is also unclear. In vitro and animal data suggest that peripheral neuropathy caused by NRTI toxicity may be mediated by mitochondrial mechanisms, particularly because of the propensity of these drugs to inhibit mitochondrial DNA gamma-polymerase [¹⁹]. Peripheral nerves of HIV-infected patients treated with zalcitabine had greater morphological and molecular evidence of abnormal mitochondria in the axoplasm and Schwann cells than seen in patients not taking this drug [²⁰]. Additionally, a higher percentage of mitochondrial DNA depletion (80%) was noted in the zalcitabine group. Brew et al. [²¹] reported that serum lactate levels are elevated in patients with d4T-associated neuropathy. Mitochondrial myopathy with lactic acidosis has also been reported with zidovudine [^22,23].

Fialuridine, a NRTI studied for the treatment of hepatitis B, produced a syndrome of severe multisystem toxicity, including lactic acidosis, hepatic failure, peripheral neuropathy, and myopathy [²⁴]. This agent caused irreversible, severe, mitochondrial and cellular changes on human myotubule cultures [²⁵]. Myopathy and peripheral neuropathy may occur in HIV-seronegative individuals with inherited or sporadic mitochondriopathy [^26,27]. Notably, four of our subjects had evidence of mitochondrial pathology on muscle biopsy. It is not clear why mitochondrial pathology was not evident in all muscle biopsies. It is possible that sampling variability or tissue-specific mitochondrial pathology may be responsible. Alternatively, genetic factors may predispose patients to additive neurotoxicity of mitochondrial toxins. For example, Cuban blindness syndrome is more likely to develop in patients with pre-existing mitochondrial mutations [²⁸].

While mitochondrial toxicity is likely to underlie NRTI-associated lactic acidosis, it is uncertain that the neuromuscular disorders are caused by this mechanism. There is potentially interplay between primary HIV-induced effects and those linked to ARV drug toxicity. HIV itself may trigger pathways inducing neuromuscular disease, particularly immune-mediated mechanisms. Both inflammatory demyelinating polyneuropathy and inflammatory myopathy occur in HIV-infected patients and were reported before the availability of ARV therapy [^29,30]. Notably, these disorders are most common at seroconversion or in early HIV infection, when immune activation is likely. HIV infection alone may impair mitochondrial function [¹⁷], and ARV effects may be additive or synergistic. Morgello et al. [³¹] reached a similar conclusion in an analysis of muscle biopsies in HIV-infected patients with myopathy, showing similar pathological evidence of mitochondrial abnormalities in subjects exposed or unexposed to zidovudine.

The time course of events in our subjects is notable. There was a delay from the time of LAS presentation, and consequent ARV drug discontinuation, to the onset of neurological symptoms in more than half of the patients for whom these data are available. This suggests that patients with hyperlactatemia or lactic acidosis must be closely followed for the development of neurological symptoms, particularly after discontinuation of ARV drugs, and improvement of systemic symptoms. It is possible that recovery in mitochondrial function following ARV drug discontinuation may trigger immunological mechanisms, such as cytokine dysregulation, leading to neuromuscular pathology [^32,33]. Notably, several nerve and muscle biopsies in our patients contained inflammatory infiltrates, consistent with an immune-mediated process.

Women are over-represented in several series of NRTI-associated lactic acidosis compared with the general population of HIV-infected patients [¹²]. Similarly, 46% of the patients in our series are women. Additionally, women constituted 50% of patients with hyperlactatemia, and 42% of those with both hyperlactatemia and definite HANWS. Reasons for this gender preference are not clear.

There are several limitations of the current study. Since this an initial description of an entity that is not well understood, we included subjects that met our minimal clinical definition of the neurological aspects of the syndrome. Information was retrospectively collected from multiple sources, resulting in non-standardized information. Complete datasets are not available for all patients, particularly lactate level, precluding adequate comparison between those with and without hyperlactatemia. Recognizing that the precision of diagnosis varied between subjects, based on the availability of ancillary tests and exclusion of confounds, we performed separate analyses of the definite and probable groups alone. These results are similar to those of the entire cohort, including those with possible HANWS. We explored several associations, each using a 0.10 significance level without adjustment for multiple comparisons. Therefore, results should be interpreted with caution. Data summaries and analyses were performed on non-missing data with no imputation. Systematic missing data may have influenced our results.

Importantly, most of our patients had exposure to ARV drugs, including NRTI. We did not explore comparative features of the comparison groups (i.e., ARV drug-treated patients without neuromuscular weakness, or ARV drug-naive HIV-infected individuals with neuromuscular disease). The study of such comparison groups is of particular importance since neuromuscular disorders, such as Guillain-Barré syndrome and myopathy, may occur in HIV-seropositive individuals independent of ARV therapy. The over-representation of subjects with hyperlactatemia and NRTI exposure in our series may, in part, reflect acquisition bias, particularly since the FDA, which is the source of several of our cases, mandated that lactic acidosis must be present with neuromuscular weakness in order to meet their case definition [⁷]. Furthermore, the reported association of d4T with lactic acidosis may have led to more aggressive screening of lactate levels in patients receiving d4T.

There are no prospective studies or case-control series documenting the occurrence of HANWS, with or without lactic acidosis. If one speculates an association with NRTI therapy, it is notable that these cases have been reported only since 1998. Didanosine was approved by the FDA in 1991 and d4T in 1994, and both have been in widespread use since that time.

The pitfall of attributing organ system toxicity to a specific ARV agent is illustrated by experience with zidovudine and myopathy. In 1990, a case series was reported of HIV-infected patients with a polymyositis-like syndrome [³⁴]. All were receiving zidovudine. Muscle biopsies revealed ragged-red fibers and other evidence of mitochondrial pathology. However, this and several other similar case series were retrospective, and few included zidovudine-untreated control groups. In two placebo-controlled trials of zidovudine therapy, there was no difference in the occurrence of myopathy in subjects regardless of zidovudine exposure [^35,36]. These results indicate that one must be cautious in attributing toxicity to a specific ARV agent based solely on selected case series.

There are numerous central and peripheral neurological disorders that may present with rapidly progressive weakness. These may be associated with HIV infection, independent of ARV therapy, and have different pathogeneses and treatment. HANWS appears to represent a syndrome rather than a distinct pathological entity, with features spanning a spectrum of neuromuscular disorders, including peripheral neuropathy, myopathy, and mixed pathology. This is reminiscent of a syndrome that occurs in severely ill patients, termed critical illness neuromyopathy [^37-40]. It must be determined from prospective and case-control studies how HANWS fits into the spectrum of neurological disorders that occur in HIV infection, and whether NRTI exposure and lactic acidosis represent specific distinguishing features. This understanding is crucial in guiding clinicians in the recognition and management of these disorders. A data collection effort for cases of HANWS, available to all clinicians, is ongoing in the Neurological AIDS Research Consentium (http://www.neuro.wustl.edu/narc/).

Acknowledgements

We thank Drs Alan Pestronk for pathological specimens, Susan Morgello for review of biopsies, and John Griffin for his review of the manuscript.

Sponsorship: This study was supported by NIH grants K24NS02253 (DS), NS44807 (JCM), NS26643 (JCM), NS44807 (JCM) U01NS32228-08A1, and U01A127667.