Clinical manifestations

Presentation and course

Antiviral nucleoside analogue therapy is generally associated with a painful, sensory, distal polyneuropathy, which is often the dose-limiting side effect (87; 56; 55; 13; 30; 74; 14; 51; 59; 43). This toxic neuropathy usually appears 2 to 3 months after initiation of treatment and may affect the hands early on. Progressive zalcitabine-induced neuropathy was confirmed in one patient by degenerative changes in a series of punch skin biopsies (52). There were no autonomic symptoms or evidence of cranial nerve dysfunction at any dose.

Sensory neuropathy characterized by pain and paresthesias has also been associated with didanosine (22; 63; 68; 03; 58; 43) and stavudine (15; 10; 60; 75).

Significant risk factors for the development of antiviral nucleoside analogue-induced peripheral neuropathy include age (72; 20), height (20), history of diabetes (14), nutritional deficiencies (59), previous exposure to antiviral nucleoside analogue compounds (47), co-treatment with interferon (31), acetylcarnitine deficiency (26), a history of heavy ethanol consumption (29), and a European mitochondrial haplogroup T (34). This group examined mtDNA single nucleotide polymorphisms (SNPs), called haplogroups, in 156 non-Hispanic black participants in the ACTG 384 study (Canter et 2010). These preliminary results suggest that belonging to mitochondrial subhaplogroup L1c (OR, 3.7 [95% CI, 1.1–12.0]; p=.03) is an independent predictor of peripheral neuropathy. This is in addition to multivariate analysis showing that randomization to ddI plus d4T and older age contribute to that risk.

In a prospective study, HIV-associated neuropathy, an elevated triglyceride level, and diabetes mellitus type 2 were each demonstrated to increase the risk of neuropathy (04). Surprisingly, the metabolic syndrome did not increase the risk of neuropathy. Specific aspects of the patient’s HIV status, including CD4+ cell counts, are also highly predictive of antiviral nucleoside analogue-induced neuropathies (59). Kelleher and colleagues report that there is a 17% increase in risk for every 100 cell/microL decrement in CD4+ cell counts (43).

The risk factors for peripheral neuropathy and symptomatic peripheral neuropathy were evaluated using longitudinal data from treatment-naive patients initiating combination antiretroviral therapy in 6 randomized AIDS Clinical Trials Group trials (24). In 573 patients who initiated neurotoxic therapy, peripheral neuropathy affected 27.4% and was significantly associated with older age and current protease inhibitor co-administration. Symptomatic neuropathy was less frequent (9.8%), with risk factors being older patient age, history of diabetes, protease inhibitor use, and, to a lesser extent, taller height. Older age also predicted lower odds of recovery after neurotoxic drug discontinuation. It is not clear if this protease effect is mediated by direct toxicity or indirectly through metabolic derangements such as insulin resistance.

There has been limited exposure of children to antiviral nucleoside analogue; however, studies to date indicate little evidence of induced peripheral neuropathy (46). In children, regimens containing zidovudine seem to be less neurotoxic than those containing d4T (79). One report of a 6-year-old girl with AIDS stated that didanosine was associated with the development of multiple, well-circumscribed retinal lesions characterized by hypertrophy or hypopigmentation of the retinal pigment epithelium (84).

Although the principal toxic insult associated with antiviral nucleoside analogues (other than AZT) has been peripheral neuropathy, muscle, pancreas, liver, and heart have also been affected (48). Pancreatitis has been especially associated with didanosine (38).

A study investigating the antiviral efficacy and safety of telbivudine (LdT) in combination with pegylated interferon (PegIFN) alpha-2a as compared to monotherapies in patients with chronic hepatitis B was prematurely terminated due to increased rates of peripheral neuropathy in the combination therapy group. Peripheral neuropathy occurred in 7 out of 50, 1 out of 54, and 0 out of 54 subjects assigned to combination therapy, LdT, and PegIFN groups, respectively. The onset of neuropathy in the combination group was 2 to 6 months after treatment onset and consisted mostly of peripheral neuropathy or sensory neuropathy, with recovery at last follow up in 50% and improvement in the other three cases following drug withdrawal. The seventh case on combination therapy experienced complete recovery from demyelinating neuropathy. The case of peripheral neuropathy in the LdT-only group had a delayed onset (14 months), compared well to the frequency of neuropathy from the full Novartis clinical study database (0.45%), and improved after stopping the drug (50). This hepatitis B drug was discontinued by Novartis in 2016.

In the interim World Health Organization guidelines of December 2018, the preferred and alternative first-line regimen for adults and adolescents included tenofovir disoproxil fumarate and 3TC combined with a third agent (World Health Organization 2018). In a study of Tenofovir-based combination antiretroviral therapy, 120 neuropathy-free and antiretroviral therapy–naive patients with HIV were followed up for 6 months (61). Nine percent of cases developed asymptomatic neuropathy, and 8% had symptomatic sensory neuropathy at 6 months. Forty percent of cases were detected within 50 days, and the remaining cases by 140 days. These researchers found active tuberculosis to be independently associated with the risk of developing sensory neuropathy.

Prognosis and complications

In most cases, the antiviral nucleoside analogue-induced neuropathy is reversible on discontinuation of treatment (02). Significant recovery occurred in 67% to 83% of subjects with clinically evident zalcitabine-induced neuropathy; however, there was often a period of “coasting” in which the neuropathy intensified immediately following cessation of treatment (13).

Clinical vignette

A 40-year-old male with HIV infection was treated with zalcitabine. At 7 weeks, he developed a sensory polyneuropathy. The predominant and presenting symptom was bilateral burning or shooting pain in the feet. This was soon followed by complaints of paresthesias, numbness, and weakness. Hyporeflexia of the Achilles reflex was an early and consistent neurologic sign of neuropathy, followed by clinically evident sensory loss, gait disturbances, and weakness. Elevation of vibration threshold in the great toe was the initial laboratory abnormality. At the time of maximal symptoms, mild to moderate nerve conduction abnormalities, suggestive of axonal dysfunction, were evident.

Biological basis

Etiology and pathogenesis

The common mechanism of antiviral nucleoside analogue toxicity appears to be interference with mtDNA and defective mitochondrial gene expression (18; 07; 48). These deficits lead to alterations in mitochondrial function and ultrastructure and ultimately to failure of cellular processes that depend heavily on high energy consumption. The specific subcellular target may be mtDNA gamma polymerase, which is essential for DNA homeostasis (80). Mitochondria in select elements of the peripheral nervous system, such as axons and Schwann cells, are especially sensitive to zalcitabine, didanosine, and stavudine, whereas mitochondria in skeletal muscle appear more easily affected by AZT. The reason for this differential susceptibility remains unclear. In a retrospective cohort study from the AIDS Clinical Trials Group, large mtDNA deletions and intraepidermal nerve fiber density (IENFD) were assessed in skin biopsies (66). Both IENFD and sural nerve action potential amplitude correlated with mitochondrial mutation burden in HIV-associated sensory neuropathy.

Several antiviral nucleoside analogue compounds, such as zalcitabine, have also been shown to inhibit the synthesis of select phospholipids (eg, phosphatidylcholine) in permeabilized rat hepatocytes (23). These effects would reasonably alter the phospholipid components of neuronal membranes, affecting their biophysical properties and possibly contributing to deficits in neuronal transduction and conduction associated with antiviral nucleoside analogue neurotoxicity. Investigation has focused on viral protein R (Vpr), a cytotoxic HIV-1 accessory protein synthesized late in the viral lifecycle (01). Vpr contributed to HIV neuropathy in cultured rat and human neurons and in a transgenic mouse model engineered to express Vpr in an immunosuppressed background. Vpr caused dorsal root ganglion neuronal damage, likely through cytosolic calcium activation and cytokine perturbation. Nerve growth factor (NGF) significantly protected mice sensory neurons from the effects of Vpr (82). Nerve growth factor acts through the tropomyosin-related kinase A receptor to block the Vpr-mediated decrease in axon outgrowth. Comparison of plasma miRNA profiles in HIV/AIDS patients with and without symptomatic distal symmetric polyneuropathy displayed ab increase in miR-455-3p (09). It turns out that miR-455-3p targets multiple genes implicated in peripheral nerve maintenance, including NGF and related genes. Based on neuronal cell culture data, the authors suggested that miR-455-3p may be a biomarker and may also have a pathogenic effect.

In vitro studies. Each of the antiviral nucleoside analogues evaluated have clear effects on mtDNA in tissue culture. Exposure of the neuronal cell line PC-12 to zalcitabine and didanosine results in a decrease in mtDNA, destruction of mitochondria, and an increase in lactate production (42). These effects can be partially rescued by the administration of uridine and pyruvate (41). Chronic treatment of dorsal root ganglion cultures to zalcitabine resulted in a reduction in mtDNA, which was associated with a decrease in the buffering of Ca2+ transient currents, which are partially mediated by mitochondria (83). Separate studies using a human lymphoblastoid cell line, CEM, have confirmed that zalcitabine is more potent than stavudine, which is in turn more potent than didanosine regarding the ability to alter mtDNA (19; 53). The ultrastructure of mitochondria was altered by each drug, including distortions and reductions in cristae and the presence of numerous vesicles. The induced changes appeared to display unique features for each compound evaluated (53). These differences may partially explain why various antiviral nucleoside analogues affect different targets (ie, Schwann cells, axon, muscle, pancreas) although each appears to principally alter mitochondrial function.

Animal studies. Marked morphologic and electrophysiologic changes were seen in New Zealand white rabbits treated with zalcitabine (50 to 250 mg/kg) for periods up to 18 weeks (06; 28). Myelin pathology consisted of severe myelin splitting, intramyelinic edema, myelin folding, and demyelination. Myelin changes were seen in the majority of large diameter fibers, and remyelination was prominent. Axonal changes were less evident and characterized by axonal loss and rare examples of shrunken axons or intralumenal phagocytes replacing axons within myelin sheaths. Pathology was present in both sensory and motor fibers, including cells in both the dorsal and ventral spinal roots. The changes were associated with profound reductions in both sensory and motor electrophysiologic measures (> 30% in velocity, > 70% in amplitude). No abnormalities of myelin, axons, or neurons were observed in sections of the cervical, thoracic, or lumbar spinal cord, brainstem at the level of the gracilis, retina, or brain (06).

In a subsequent study, mitochondrial changes were associated with 35 mg/kg of zalcitabine in the rabbit (27). Pathology consisted of the presence of cup-shaped mitochondria, which were arrayed in unusual patterns and complex aggregations. The earliest changes in mitochondria were seen at 16 weeks and were positively correlated with myelin pathology in individual animals. Neither didanosine nor stavudine were found to induce myelin, axonal or mitochondrial changes in rabbits, even at near lethal doses (eg, 1500 mg/kg) for periods of up to 24 weeks (81).

The behavioral, electrophysiologic and pathologic features of the zalcitabine-induced neuropathy in the rabbit appear to differ substantially from the clinical features of zalcitabine neuropathy in humans. The rabbit data suggest a myelinopathy of large diameter fibers, with only secondary axonal involvement. They also indicate a significant involvement of motor axons. In contrast, human exposure to zalcitabine is more consistent with a predominant small fiber, sensory, distal axonopathy. These differences may be related to the relatively high doses used in the rabbit study, to a possible unusual susceptibility of Schwann cell mitochondria in the rabbit, or to the interaction of HIV infection and zalcitabine toxicity in the human population studied.

Zalcitabine at doses of 3 to 10 mg/kg, produce electrophysiologic and axonal changes in Cynomolgus monkeys, with no evidence of demyelination (05). Initial changes consisted of a subtle slowing of maximal sensory nerve conduction velocity (approximately 10%) and an associated reduction in compound sensory amplitudes. No clear neuropathology was evident at the time of these early electrophysiologic changes. Following 38 weeks of treatment, axonal changes consisted of a statistically significant reduction in mean cross-sectional diameter of myelinated axons in the sural nerve and a reduction in the density of unmyelinated fibers. Didanosine has been reported to cause axonal neuropathy, electrophysiologic changes, and ultrastructural changes in axonal mitochondria in rats (67).

Epidemiology

In the early monotherapy studies, the neuropathy associated with zalcitabine was evident in all HIV-1-positive patients evaluated at the highest dose tested, ie, 0.06 mg/kg every 4 hours for a period of up to 12 weeks (13). Lower doses (0.03 to 0.005 mg/kg every 4 hours) were associated with a similar but less intense sensory neuropathy. The incidence of neuropathy was reduced to approximately 33% in subjects receiving 0.005 mg/kg every 4 hours (13) and 34% of subjects receiving 2.25 mg/day (14). Findings from the Delta trial suggest that peripheral neurotoxicity peaks at 90 days after the onset of nucleoside analogue treatment and that the risk is especially high in the zalcitabine/AZT combination recipients (08). Older age (35 to 44 years) and lower CD4 counts (< 150) were also associated with an increased risk of neuropathy.

As polypharmacy is increasingly used for the treatment of AIDS, interactions among synergistic or additive compounds and the use of alternating versus intermittent dosing regimens will take on added significance. For instance, a much lower incidence of neuropathy was seen when the same total dose of zalcitabine was given in a monthly alternating rather than weekly alternating fashion (76). Neuropathy was evident in only 2 of 56 patients receiving combinations of zidovudine and zalcitabine at doses that were clinically effective (54), and no serious neuropathy was evident in a second study, which included didanosine (36). In antiviral clinical trials using combination therapies, the prevalence of antiviral nucleoside analogue-induced peripheral neuropathy has been reduced to approximately 10% (72). The World Health Organization recommended that countries phase out the use of stavudine because of its long-term, irreversible side effects (86). In a study evaluating the evolution of symptomatic sensory neuropathy in 184 participants after initiation of antiretroviral therapy, pre-antiretroviral therapy 16% had neuropathy, and pain overall improved after 18 months of treatment (17). In those without baseline neuropathy, 18 per 100 person-years developed symptomatic neuropathy. It is noteworthy that 60% of cases receiving stavudine predicted reduced tendon reflexes but not incident symptomatic neuropathy cases. In another study from a tertiary referral hospital in southern Brazil, the prevalence of neuropathy (31.3%) was even more so increased in older patients and those with higher CD4 lymphocytes levels (12). This study was unique in that it also detected that smoking increased the odds ratio for neuropathy (3.4; p < 0.01).

In a report from South Africa, 72 of 75 self-identified Africans were started on tenofovir, emtricitabine, and efavirenz as first-line therapy (32). The researchers used the Brief Peripheral Neuropathy Screen amended by inclusion of pinprick sensitivity. Twelve percent were diagnosed with sensory neuropathy before starting treatment, and another 27% developed HIV-related sensory neuropathy in the following 6 to 8 months. The calcium/calmodulin kinase 2 (CaMKK2), which induces production of IL-1beta and TNF via the p38-MAPK pathway, is also expressed in the nervous system and plays a role in neuronal growth and repair. Three CAMKK2 haplotypes (2, 3, and 4) were associated with neuropathy after adjusting for body weight, nadir CD4 T-cell counts, and prior tuberculosis. In a time-to-neuropathy event modifier analysis of the multicenter ACTG study, 108 individuals initiating therapy had sufficient serum for analysis, including 30 neuropathy cases (39). Neuropathy onset was based on clinical symptoms or signs of neuropathy. These researchers’ preliminary results indicate that higher iron stores, the iron-loading hemochromatosis gene variant 187C>G, and lower levels for high-sensitivity C-reactive protein predicted delayed onset of neuropathy among self-reported white individuals.

In a systematic review and meta-analysis of HIV-related peripheral neuropathy, age and treatment with nucleoside analogue reverse transcriptase inhibitors are risk factors for developing HIV-related neuropathy (Julian at al 2021). The pooled prevalence of neuropathy in treatment-naive patients is 29% (95% CI: 9% to 62%), and it increased to 38% (29% to 48%) at later disease stages, with 53% (41% to 63%) being symptomatic. In addition to symptom management, discontinuation of drugs such as didanosine can improve or resolve symptoms. In a longitudinal study of adults living with HIV, peripheral neuropathy occurred in 21% of patients (78). Using machine learning algorithms, variables predictive of the development of neuropathy were HIV disease duration, peak plasma viral load, age, and low CD4+ T-cell count.

Prevention

The prevention of antiviral nucleoside analogue-induced neuropathy is principally related to the following measures: (1) effective use of combination antiviral therapy to minimize the putative insult from each compound, (2) careful monitoring on the onset of signs and symptoms to adjust treatment prior to irreversible changes, and (3) consideration of the risk factors for antiviral nucleoside analogue neuropathy when prescribing therapy.

Furthermore, in vitro evidence suggests that erythropoietin and FK 506 have distinct neuroprotective effects on ddC neurotoxicity. The immunophilin ligand FK 506 ameliorates ddC-induced neuritic pruning, neuronal mitochondrial depolarization, and neuronal necrotic death (44). Erythropoietin also prevented sensory axonal degeneration of dorsal root ganglia exposed to ddC (45).

Interestingly, in a U.S. cohort of HIV-infected individuals on stable regimens, the use of nucleoside analogues did not increase the risk of worsening of neuropathic symptoms and signs (35). Another study showed the administration of glucose-lowering drugs in recently treated HIV patients may prevent progression from asymptomatic to symptomatic neuropathy (25).

Differential diagnosis

Associated or underlying disorders

The differentiation of true antiviral nucleoside analogue-induced peripheral neuropathy is often difficult because the signs and symptoms are similar to those seen in patients infected with HIV. At least four distinct peripheral neuropathy syndromes have been identified in AIDS patients, with the most common being a distal painful sensory neuropathy (57). Antiviral nucleoside analogue-induced neuropathy is principally distinguished by its speed of onset within the first 90 days to a year of treatment initiation and by its recovery following cessation of treatment.

Diagnostic workup

The workup for antiviral nucleoside analogue-induced neuropathy is identical to that of a symmetrical, prominently sensory, polyneuropathy. Loss of sensation, detectable by quantitative sensory evaluation, loss of Achilles tendon reflex, and positive symptoms such as pain and paresthesias of the feet are valuable indexes of the onset and progression of the condition. At the early stages, electrophysiologic measures are relatively insensitive and non-specific.

Skin biopsy to measure intraepidermal nerve fiber density (IENFD) may be helpful in confirming the diagnosis when there are atypical neuropathic symptoms, neuropathic symptoms without classical signs, or normal nerve conduction studies. HIV-infected patients, with or without neurologic signs, demonstrated lower intraepidermal nerve fiber density and increased risk of progression to symptomatic HIV-associated distal sensory neuropathy 6 to 12 months later (33). However, intraepidermal nerve fiber density may not be a useful predictor of small unmyelinated nerve fiber damage and neuropathy risk in HIV across all ethnic groups (69). In a study of 150 Thai subjects, all HIV-positive participants had lower intraepidermal nerve fiber density than HIV-negative subjects. But intraepidermal nerve fiber density was similar among the HIV-positive groups whether they were antiretroviral therapy-naïve or on d4T with or without neuropathy or neuropathic symptoms.

Management

Supportive measures and rehabilitation are discussed in another article: peripheral neuropathies.

The management of antiviral nucleoside analogue-induced neuropathies includes prompt cessation of therapy (when possible) or adjustment in the combination therapies utilized. Additional management tools include the treatment of pain with simple analgesia, augmented with antidepressants or anticonvulsive agents (59; 37). However, most of these drugs, with the exception of lamotrigine, have been demonstrated to be no more effective than placebo in relieving pain in HIV-related peripheral neuropathy (73).

It is important to treat HIV-related neuropathic pain, as it has a negative impact on health status and physical function (49). Furthermore, depression increases with increased pain severity, resulting in substantial clinical and economic burden. In this study of HIV-associated neuropathic pain, three quarters of patients had moderate to severe pain. Most patients were prescribed at least one medication for neuropathic pain, and these were most commonly narcotic analgesics or antiepileptic drugs and much less frequently a variety of antidepressant agents.

A double-blind placebo-controlled multicenter study of painful antiretroviral toxic neuropathy in patients with HIV-1 infection demonstrated efficacy in pain reduction of acetyl L-carnitine 500 mg administered 2 times per day intramuscularly (88). The oral study phase of acetyl L-carnitine 1000 mg 2 times per day did not show a statistically significant improvement. In a follow up report, Youle demonstrated an improvement in cutaneous nerve fiber density in acetyl L-carnitine recipients, which was maintained over a 4-year period of treatment (88).

In a placebo-controlled 12-week study of painful HIV neuropathy, 33% of subjects placed on high-concentration capsaicin dermal patches (NGX-4010) experienced significant pain relief, whereas only 18% of controls had this effect (70). In another study, the mean percent change from baseline in Numeric Pain Rating Scale (NPRS) score to weeks 2 to 12 was not statistically significant between active drug and placebo, but trends toward pain improvement were detected after a single 30-minute NGX-4010 treatment (21). Out of the 307 patients completing the former 12-week, randomized, double-blind phase of the study, 272 entered a 40-week, open-label phase and received up to three 60-minute NGX-4010 treatments (71). Patients recorded their average pain for the past 24 hours each day using the NPRS. The percentage NPRS score reduction from baseline to weeks 2 to 12 after treatment ranged from 23% to 26%. Patient Global Impression of Change and Clinician Global Impression of Change questionnaires also demonstrated benefits and adverse events were mild to moderate and cutaneous. In a meta-analysis of randomized and controlled studies of 8% capsaicin patch, baseline pain intensity score of less than or equal to 4 and female gender were predictors of sustained and complete response (40).