HIV-Associated Distal Sensory Polyneuropathy in the Era of Highly Active Antiretroviral Therapy: The Manhattan HIV Brain Bank

Objectives  To examine distal sensory polyneuropathy (DSP) in a highly active antiretroviral therapy era, human immunodeficiency virus (HIV)–infected cohort, to determine whether clinical manifestations are affected by demographic or other clinical variables.

Patients  One hundred eighty-seven patients with HIV infection enrolled in the Manhattan HIV Brain Bank underwent baseline neurologic evaluations between January 29, 1999, and June 17, 2002. Distal sensory polyneuropathy was diagnosed if patients displayed abnormalities in 2 or more of the following: ankle reflexes or vibratory or pinprick perception. Patients were classified as symptomatic if they described pain, paresthesia, or numbness. Nonneurologic information was obtained by interview, laboratory testing, and medical chart review. Psychiatric and substance use disorders were elucidated by semistructured interview. In 36 patients, morphometric analysis was performed on autopsy-derived sural nerves.

Results  Of 187 patients, 99 (53%) had DSP. Patients with neuropathy were older than those without (mean ± SD age, 45.3 ± 0.7 vs 41.2 ± 0.8 years, P <.001), and DSP was significantly more common in men (58% [83/99]) than in women (37% [16/99]) (P = .02). The presence of neuropathy was not correlated with plasma viral load, decreased CD4 cell counts, or neurotoxic antiretroviral therapy. Twenty-six of 99 patients with DSP were asymptomatic. Asymptomatic neuropathy was correlated with histories of opiate and sedative abuse and dependence. Symptomatic DSP correlated with ethanol and hallucinogen syndromes, but not neurotoxic therapy. Sural nerve morphometric findings did not distinguish between patients with substance use syndromes and those without.

Conclusions  In contrast to populations before the era of highly active antiretroviral therapy, DSP in the Manhattan HIV Brain Bank cohort is not associated with increased viral load or decreased CD4 cell counts in this cross-sectional analysis. Symptoms in DSP are associated with substance use disorders, but no difference in morphologic structure is seen in nerves of patients with HIV infection with and without substance use histories. Previously reported virologic and immunologic underpinnings of DSP may be affected by highly active antiretroviral therapy. Furthermore, symptoms of DSP in substance users may be altered by central mechanisms of increased or decreased tolerance to sensory disturbance.

With the availability of highly active antiretroviral therapy (HAART), peripheral neuropathy has emerged as the most common AIDS-associated neurologic disorder.^1,2 Concurrent with expanding use of HAART, there have been shifts in the demographic composition of individuals affected by human immunodeficiency virus (HIV).³ Therefore, changing demographics and evolving therapies have altered manifestations of HIV-related diseases and, in doing so, may affect their pathogenesis. An example of this is AIDS-associated dementia, in which evidence suggests that traditional associations with cerebrospinal fluid viral load and immune activation may no longer pertain.⁴

The pathogenesis of HIV-related distal sensory polyneuropathy (DSP) is unclear. In pre–HAART era studies,^5-7 it has been associated with low CD4 cell counts and increased HIV plasma viral loads. Neurotoxic antiretroviral agents (zalcitabine, didanosine, and stavudine) may also produce neuropathy in the context of HIV.⁸ Analysis of the pre-HAART Dana cohort failed to show correlation between CD4 cell count or dideoxynucleoside therapy and neuropathy, although lower CD4 cell counts predicted neuropathy development over time.⁹ Analysis of a sizable HAART-era cohort has not yet been reported. We describe herein the baseline characteristics of patients from the Manhattan HIV Brain Bank (MHBB), a HAART era, neurologic, neuropsychologic, and psychiatric study of individuals with advanced HIV infection enrolled in an organ donation program.

Eligibility criteria for the MHBB are that patients must be HIV-positive, consent to postmortem organ donation, and (1) have a condition indicative of advanced HIV disease or another disease without effective therapy, (2) have a CD4 cell count of no more than 50 cells/µL for at least 3 months, or (3) be at risk of near-term mortality in the judgment of the primary physician. Indicator conditions for category 1 are progressive multifocal leukoencephalopathy, lymphoma, disseminated Mycobacterium avium–intracellulare, wasting, AIDS dementia complex, cytomegalovirus end-organ disease, visceral Kaposi sarcoma, congestive heart failure, or a serum albumin level of less than 3.2 g/dL.

On recruitment, patients undergo a battery of neurologic, neuropsychologic, and psychiatric examinations. General medical information, plasma viral loads, CD4 cell counts, and antiretroviral therapy (ART) histories are obtained through patient interview and medical record review. The neurologic evaluation is performed by a neurologist and includes a comprehensive evaluation of motor and sensory function. The diagnosis of DSP is made if the patient displays 2 of the following abnormalities in a distal and symmetric fashion: decreased or absent ankle deep tendon reflexes, vibratory sense, or pinprick perception. Patients are classified as symptomatic if they report 1 or more of the following: paresthesia, dysesthesia, or numbness. Psychiatric and substance use histories are obtained through the Psychiatric Research Interview for Substance and Mental Disorders.¹⁰

On demise, as permitted by the patient, the sural nerve is sampled and trisected. One portion is frozen, another is fixed in 10% phosphate-buffered formalin for routine histologic examination, and a third is fixed in glutaraldehyde for processing, osmication, and embedding into plastic with sectioning at 1-µm thickness. Sections are evaluated by light microscopy and used for morphometric analysis. Morphometry is performed with IP Lab version 3M software (Scanalytics Inc, Fairfax, Va) running on a Macintosh Power PC (Apple Computer Inc, Cupertino, Calif). Images are captured from a microscope (Labophot-2; Nikon Corp, Melville, NY) with a Sony videocamera (HyperHAD CCD-IRIS/RGB color videocamera; Sony Corp, New York, NY). A representative nerve fascicle is chosen by light microscopy for analysis. The number of myelinated fibers is counted, and the total endoneural area is calculated to arrive at a myelinated fiber density. Myelinated fibers are manually outlined and their diameters calculated. A distribution of large-caliber (>7 µm in diameter) and small-caliber myelinated fibers is derived.

Statistical analyses were performed on an iMac computer (Apple Computer Inc) using StatView version 5.0.1 software (SAS Institute, Cary, NC). Characteristics were compared between the DSP groups by t tests, χ² tests, analysis of variance, or logistic regression analysis. Differences in nerve morphometric findings were analyzed by t tests or analysis of variance.

At baseline evaluation between January 29, 1999, and June 17, 2002, 99 (53%) of 187 patients displayed signs of DSP. The demographic composition and immunologic or virologic characteristics of the patients are displayed in Table 1. The presence of neuropathy correlated with older age (45.3 vs 41.2 years; P <.001), male sex (58% [83/99] of men vs 37% [16/99] of women; P = .02), and white race (68% [32/99] of white vs 52% [44/99] of black and 41% [23/99] of Hispanic individuals; P = .04). No significant correlation was seen with risk factors for HIV or the duration of infection (patient age did not correlate with the duration of infection). When demographic variables were analyzed by logistic regression, only age (95% confidence interval, 1.024-1.119) and sex (95% confidence interval, 1.003-4.359) were consistently associated with DSP, and race was no longer significant. Patients with DSP did not differ from those without in regard to hemoglobin levels or plasma viral loads. There was a significant correlation of DSP with CD4 cell count: patients with DSP had a mean ± SD CD4 cell count of 228 ± 29 cells/µL compared with 128 ± 22 cells/µL for those without. However, when CD4 cell counts, viral loads, sex, and age were combined in logistic regression analysis, only age and sex emerged as significantly associated with DSP.

With the counterintuitive association of DSP and higher CD4 cell counts, we examined whether DSP was associated with ART. At the time of evaluation, 127 (68%) of 186 patients were receiving ART (Figure 1). Nine (5%) were taking 2 ART medications, 73 (39%) were taking 3, 34 (18%) were taking 4, 8 (4%) were taking 5, and 3 (2%) were taking 6. Of the patients not receiving ART at the time of evaluation, 93% (55/59) had previously received ART (mean ± SD number of previous medications, 4.8 ± 3.3 [range, 0-15]). Therefore, the patients not receiving HAART had high levels of previous use of ART. Furthermore, the patients not receiving HAART did not differ significantly from those who were with regard to the prevalence of DSP (58% [34/59] of those not receiving HAART vs 52% [66/127] of those receiving HAART; P = .65, contingency table analysis) or whether the DSP was symptomatic (73% [43/59] of those not receiving HAART vs 74% [94/127] of those receiving HAART; P >.99, contingency table analysis).

The presence of neuropathy was not significantly related to concurrent use (P = .34), exposure within 2 months (P = .43), or remote exposure to neurotoxic nucleoside reverse transcriptase inhibitors (NRTIs) (P = .12, contingency table analysis for all). Neuropathy was not correlated with the total number of ART medications a patient was taking or to any classes of drugs. Neuropathy did not significantly correlate with the cumulative number of months of exposure to neurotoxic NRTIs, although patients with DSP had a mean exposure of 19.3 months, compared with 14.1 months for patients without DSP (P = .17, t test). Of the 99 patients with DSP, 15 displayed onset of symptoms in conjunction with neurotoxic NRTIs and were classified as having combined HIV-ART neuropathy, as symptoms had improved with the cessation of drugs. Follow-up evaluations in the first year of study were available in 12 of these 15 patients (mean follow-up, 10 months [range, 3-12 months]). Of the 12 patients, only 1 had complete resolution of neurologic deficits with discontinuation of neurotoxic ART.

With the counterintuitive association of DSP and higher CD4 cell counts, we examined the potential for non–HIV-related factors to contribute to neuropathy. Of 32 individuals with DSP and CD4 cell counts greater than 200 cells/µL, 21 (66%) had factors deemed potentially contributory, and 11 (34%) were at risk for neuropathy on the basis of HIV alone. The most common confounders were alcohol use (9 patients), neurotoxic ART (5 patients), chemotherapy (2 patients), diabetes mellitus (2 patients), complex neurologic disease precluding definitive diagnosis (2 patients), and renal disease (1 patient). Of the 11 patients with no identifiable risk for neuropathy other than HIV, 6 had remote exposures to neurotoxic ART medications.

Psychiatric and substance use disorders were elucidated in 164 patients (89 with DSP and 75 without) (Table 2). Neuropathy was associated with lifetime diagnoses of depressive disorder, and ethanol syndromes were present in 54% [48/89] of patients with DSP and 36% [27/75] of patients without. Active ethanol syndromes within 2 years of the examination were present in 19 patients with DSP and thus constituted confounders for attribution to HIV disease. Lifetime history of sedative syndromes was also correlated with DSP, present in 16% [14/89] of patients with neuropathy and 4% [3/75] of patients without. Patients with neuropathy had a greater number of diagnosed substance use syndromes; however, this difference approached but did not attain significance (mean ± SD, 2.4 ± 0.2 substance use syndromes per patient with DSP vs 1.9 ± 0.2 syndromes without DSP; P = .07).

Of the 99 patients with DSP, 26 were asymptomatic. Asymptomatic patients did not differ from symptomatic patients with regard to demographic variables, exposure to neurotoxic NRTIs, plasma viral load, or CD4 cell count. When homosexual or bisexual, heterosexual, and intravenous drug use (IVDU) risk groups were examined, patients with IVDU had proportionately greater amounts of asymptomatic neuropathy (Table 3). Seventy-three percent (19/26) of all patients with asymptomatic neuropathy had IVDU risk for HIV infection vs 47% (32/68) of those with symptomatic DSP and 43% (35/81) of those with no neuropathy (P = .05, contingency table analysis after eliminating 12 patients with other or unknown risks for HIV). Of the 73 symptomatic patients, 61 (84%) had positive symptoms (pain or paresthesia) and 12 (16%) had numbness only (negative symptom only). Within the symptomatic group, those with IVDU were more likely to exhibit numbness only (9 of 32 symptomatic patients with IVDU) than non-IVDU risk groups (3 of 41 symptomatic non–drug users) (P = .04, contingency table analysis).

We analyzed how individual substance use disorders might relate to symptoms in DSP. Lifetime diagnoses of opiate and sedative syndromes were significantly correlated with asymptomatic DSP, and lifetime diagnoses of ethanol and hallucinogen syndromes with symptomatic DSP. When all substances were combined in logistic regression analysis, opiate and sedative syndromes maintained significant correlation with asymptomatic DSP (symptomatic vs asymptomatic, P = .03 for opiates; P = .02 for sedatives). When substance use syndromes were examined for relationship to positive and negative symptoms in symptomatic patients, opiate disorders accounted for more than the expected numbers of patients with numbness only (31% [8/26] of symptomatic opiate users had numbness only vs 8% [3/39] of individuals without opiate diagnoses; P = .04, contingency table analysis). The distribution of patients by substance use syndromes and neuropathy categories is in Table 4.

Morphometric analysis for myelinated fiber density was undertaken on sural nerves obtained from 36 patients. The mean ± SD myelinated fiber density was 5627 ± 306 fibers per square millimeter (range, 1576-10 470 fibers per square millimeter), with large fibers (>7 µm) 1133 ± 112 fibers per square millimeter [range, 0-2331 fibers per square millimeter]) and small fibers 4493 ± 256 fibers per square millimeter [range, 1190-8480 fibers per square millimeter]) present. There were no correlations of myelinated fiber density with sex, risk group, or race. A correlation of myelinated fiber density with age approached significance (r = 0.32, P = .06); this correlation was more pronounced for small fibers (r = 0.34, P = .04) than for large (r = 0.09, P = .59). Patients without DSP had a mean ± SD fiber density of 5902 ± 531 fibers per square millimeter compared with 5612 ± 400 fibers per square millimeter in those with DSP; this difference was not significant (P = .74), and the effect of age on fiber density in multiple regression analysis was more pronounced than the presence of DSP. Similarly, despite differences in clinical manifestations and substance use histories, there were no differences in myelinated fiber densities from patients with histories of opiate, hallucinogen, sedative, or ethanol syndromes and others. There were no CD4 cell count or viral load–associated differences in fiber density. Finally, the mean fiber density did not correlate with overall cumulative exposure to neurotoxic ART medications, to remote histories of exposure, to exposure at death, or within 2 months of death.

Peripheral neuropathy is the most common HIV-related neurologic disorder in the HAART era and has been studied as an important complication of HIV since the onset of the AIDS epidemic.^1,2,11,12 Among all forms of AIDS-associated neuropathy, DSP is the most common and is most often seen in advanced HIV.² The MHBB cohort constitutes a population at risk for DSP by virtue of a large percentage of individuals with advanced HIV. Accordingly, it is not surprising that 53% [99/187] of individuals in the MHBB displayed DSP on study entry. In a recent description of the Dana cohort, an advanced, pre-HAART group, 55% (147/268) of baseline evaluations revealed DSP.⁹ Therefore, it would seem that, regardless of the therapeutic era, more than half of individuals with advanced HIV infection display signs of neuropathy.

During the past 2 decades, the HIV epidemic shifted into women and minority populations with significant comorbidities.³ There has also been a survival benefit coincident with improving therapy.¹³ Therefore, should demography affect manifestations of DSP, HAART-era cohorts may show an altered profile of disease. In this study, we demonstrate a significant association between male sex, increased age, and DSP. Advancing age has been previously noted as a predictor of DSP.⁶ Less clear has been the sex-relatedness of the disorder. Ninety-eight percent (263/270) of patients in AIDS Clinical Trials Group Team 291, a trial of nerve growth factor for HIV-associated DSP, were male.^5,9,14 In a study⁷ of a neurology clinic population at our institution, 84% (53/63) of individuals diagnosed with DSP were male, but concern was raised that this reflected recruitment bias. In a recent analysis of the pre-HAART Dana cohort, there was no significant difference in the percentage of men in the group with DSP (80% [118/147]) and in the group without (75% [91/121]) (P = .45).⁹ In contrast, in our cohort, 84% (83/99) of patients with and 69% (61/88) of those without DSP were male, a significant difference of 15% (P = .03) compared with the 6% (P = .45) in the Dana study. Furthermore, the significance of the difference persisted in our cohort with adjustment for the variation in other demographic factors. It is unclear why our cohort demonstrates a sex preference for DSP. Regardless of the reason, it seems that increasing age and higher percentages of women may have opposing effects on the overall prevalence of HIV-associated DSP.

Another unexpected finding in our study was the lack of correlation between DSP, decreased CD4 cell count, and increased viral load. In pre-HAART era studies,^5,7,9,15 elevated plasma viral load and decreased CD4 cell count have been predictive of development of DSP and correlate with the severity of symptoms. Although we did not test the predictive value of these laboratory measures, their lack of correlation might be a function of several possibilities. One is that our patients may have acquired DSP at prior nadir values in immunologic or virologic status and that, with HAART, variable degrees of immune reconstitution have masked traditional associations but have not improved neuropathy. We do not have nadir CD4 cell count and peak viral load values available and cannot test this hypothesis. Another possibility is that, in the HAART era, immune reconstitution potentiates DSP, as in other immune reconstitution disorders. It is also possible that confounding conditions not dependent on immunologic or virologic status, such as neurotoxic ART, diabetes mellitus, and other medication or substance use, are becoming more important in the pathogenesis of DSP in the HAART era. This would be consistent with the concept of increasing importance of comorbidities in HIV disease. Whatever the hypothesis, these possibilities reflect a theme that has emerged in the HAART era, namely, that immunologic and virologic variables of neurologic disease no longer show their traditional associations. An example of this is seen with the recent report of the North East Dementia cohort, in which cerebrospinal fluid viral loads no longer discriminate between patients with and without cognitive impairment.⁴ Finally, another possibility for the lack of traditional associations of immunologic and virologic variables in our population is that they may segregate with other variables associated with DSP. This is supported by the fact that, after adjustment for the significant demographic variables of sex and age in our study, associations between DSP and CD4 cell counts lost significance.

Neurotoxic NRTIs were not associated with DSP in our patient population. This is similar to findings in the Dana cohort.⁹ Neurotoxic ART medications were also not related to symptoms in our study, and 1 in 12 individuals characterized as having combined HIV-ART neuropathies had complete resolution of neurologic signs during a follow-up evaluation. This suggests that, while neurotoxic drugs contribute to the pathogenesis of DSP, their effects are predicated on the underlying metabolic and immunologic status of the individual. As in these advanced cohorts, neurotoxic ART exposures do not account for the presence or persistence of neuropathic signs. This possibility has been raised previously, with the underlying substrate of advanced immunodeficiency postulated to constitute the major susceptibility to ART-induced DSP.¹⁶ Alternatively, it may be that damage sustained in toxic neuropathy may be so severe as to preclude complete reversal of abnormal neurologic signs.

One of the novel observations in our cohort is the association of substance use syndromes and symptoms of disease. A large percentage of MHBB patients with asymptomatic neuropathy were accounted for by IVDU, and there was a significant correlation of opiate syndromes with asymptomatic DSP and with symptomatic DSP manifest by only negative symptoms. This is unexpected, as there are reports of greater sensitivity to pain in individuals receiving methadone hydrochloride maintenance.¹⁷ In contrast, there is evidence that subjects with former opiate addiction who are completely drug-free and not taking methadone have pain tolerance comparable to that of control subjects without history of addiction.¹⁷ In AIDS cohorts, there are reports of no difference in the prevalence of pain between patients with and without IVDU.¹⁸ Most patients in MHBB with opiate addiction are receiving methadone maintenance therapy. Therefore, if prior reports describing enhanced sensitivities in persons receiving methadone therapy are correct, one would expect heightened symptoms of disease in our cohort. However, with DSP, normal pathways of pain perception have been damaged, and this may bear on the phenomenon we observed. Although relative contributions of central and peripheral circuits in the generation of neuropathic pain are not clear, we were unable to detect any morphometric differences between the peripheral nerves of patients with and without opiate addiction. Therefore, we hypothesize that the opiate-related effect we observed is central in origin, reflecting altered neurobiology or concomitant medication or substance administration.

In summary, our study of DSP in the MHBB cohort suggests that, with evolution of the AIDS epidemic, manifestations of this disease are shifting. The lack of its traditional association with reduced immunologic status may indicate that therapeutically induced immune reconstitution fails to ameliorate underlying neurologic deficits and may increase the focus on comorbidities in its pathogenesis. Furthermore, as the epidemic shifts into cohorts with more substance use, DSP-related symptoms of disease may be affected by central, and not peripheral, mechanisms.

Corresponding author and reprints: Susan Morgello, MD, Department of Pathology, The Mount Sinai Medical Center, Campus Box 1134, 1 Gustave L. Levy Pl, New York, NY 10029 (e-mail: susan.morgello@mssm.edu).

Accepted for publication December 3, 2003.

Author contributions: Study concept and design (Drs Morgello, Simpson, and Geraci); acquisition of data (Drs Morgello, Estanislao, Geraci, DiRocco, Ryan, Yakoushina, Khan, Mahboob, Dorfman, and Sharp and Messrs Gerits and Naseer); analysis and interpretation of data (Drs Morgello, Estanislao, Simpson, Ryan, and Dorfman); drafting of the manuscript (Drs Morgello, Estanislao, Yakoushina, and Mahboob); critical revision of the manuscript for important intellectual content (Drs Morgello, Estanislao, Simpson, Geraci, DiRocco, Ryan, Khan, Dorfman, and Sharp and Messrs Gerits and Naseer); statistical expertise (Drs Morgello, Ryan, and Dorfman); obtained funding (Dr Morgello); administrative, technical, and material support (Drs Morgello, Estanislao, DiRocco, Yakoushina, Khan, Mahboob, and Sharp and Mr Gerits); study supervision (Drs Morgello, Simpson, Geraci, and Ryan and Mr Gerits).

This project was supported by grant R24 MH59724 from the National Institute of Mental Health, National Institute of Neurological Disorders and Stroke, and National Institute on Drug Abuse, National Institutes of Health, Bethesda, Md, and by grant M01 RR000071 from the General Clinical Research Center, Mount Sinai School of Medicine, New York.

We thank the patients and staff of the Manhattan HIV Brain Bank, a member of the National NeuroAIDS Tissue Consortium.

Investigators and staff of the MHBB include Laurie Abromowitz, CSW, Sherly Altidor, PA, Laura Banks, MD, Yvonne Brown, RN, Jacqueline Crittendon, BS, Colleen Dowling, RN, Yan Ling Gao, MD, Tauseef Haider, MD, Deborah Hesketh, RN, Talha Idrees, MD, Keren Isaacs, MS, Geraldine Joseph, PA, Victoria Kozlowski, RN, Damien Laudier, BS, Lalitha Mantha, RN, Aleks Maryanchik, MS, Natalie Massenberg, BS, Christine Mondragon, RN, Jacinta Murray, BS, Phyllis Ristau, RN, Amy Scarano, BS, JoAnne Sweeney, RN, Michele Tagliati, MD, Enrique Wulff, MD, and Mohammad Zaidi, MD.