Association of osteonecrosis and peripheral neuropathy in HIV-1-infected patients

Abstract

Atraumatic osteonecrosis affects about 1% of HIV-infected persons when asymptomatic and about 5% when symptomatic [1]. Two clinical circumstances suggest that nerve growth factor (NGF) might be involved in the excessive prevalence. First, patients with hereditary sensory and autonomic neuropathy (HSAN) have a remarkable (5–23%) prevalence of osteonecrosis [2–5]; the genetic basis of HSAN type 4B is missense mutations in the trkA receptor for NGF [6,7]. Second, 16 patients who received a monoclonal antibody that blocks interaction of NGF with its receptor, developed osteonecrosis [8]. Moreover, NGF is both osteogenic [9] and angiogenic through upregulation of vascular endothelial growth factor (VEGF) [10]. The above circumstances suggested the hypothesis that HIV-infected persons with osteonecrosis might have more peripheral neuropathy than those without osteonecrosis and that impaired actions of NGF and, probably, of VEGF might contribute to the excess of osteonecrosis. We identified 106 patients with osteonecrosis among 11 506 HIV-infected persons; characteristics of these 106 individuals have been presented [11]. The diagnosis of peripheral neuropathy was taken from the electronic medical record (EMR). This was the only basis for that diagnosis in the 11 400 patients without osteonecrosis. For the 106 patients with osteonecrosis, we also required written documentation of symptoms and signs confirming the diagnosis of peripheral neuropathy. Unilateral symptoms or symptoms confined to the upper limbs were not accepted. Symptoms were generally recorded as burning, pain or numbness, or a combination of these, in the feet. Signs typically included impairment of pin-prick, light touch or vibratory, sensations in the toes or feet, and impairment or loss of tendo-Achilles reflexes. The diagnosis of diabetes mellitus required the use of antidiabetic medication, fasting blood glucose of more than 125 mg/dl on two separate occasions or glycated hemoglobin more than 6.4%. The diagnosis of alcohol abuse was taken from the EMR. Pharmacy records were reviewed for the use, before the diagnosis of peripheral neuropathy, of zalcitabine (ddC), didanosine (ddI) or stavudine (d4T). Statistical analyses used χ2 or Fisher's exact tests. Significance was two-tailed. Among the 106 patients with osteonecrosis, there were 44 diagnosed with peripheral neuropathy involving the lower limbs. We excluded 13 of these for various reasons. Peripheral neuropathy in the remaining 31 (29.2%) of the 106 osteonecrotic patients was 1.8-fold more frequent than in the 1856 (16.2%) of the 11 400 without osteonecrosis (odds ratio 2.1, 95% confidence interval 1.4–3.2, P < 0.0003). We did not seek written documentation confirming the diagnoses of peripheral neuropathy for the 1856 without osteonecrosis; were that number reduced, by a factor of 13/44, in order to provide the same proportion of excluded patients as in those with osteonecrosis, then the 31 (29.2%) of 106 would be compared with 1308 (11.5%) of 11 400 (P < 0.0001), a 2.5-fold difference. In the following analyses, group 1 comprises the 31 patients with peripheral neuropathy before osteonecrosis and group 2 excludes four with unobtainable records and comprises the 71 with osteonecrosis either without peripheral neuropathy or with peripheral neuropathy diagnosed after osteonecrosis. ddC, ddI or d4T were used by 90.3% of group 1 and by 69.0% of group 2 (P = 0.02). Diabetes mellitus was present before the diagnoses of peripheral neuropathy in 9.7% of group 1 and in 15.5% of group 2 (P = 0.54). Alcohol abuse was present before the diagnoses of osteonecrosis in 16.1% of group 1 and in 11.3% of group 2 (P = 0.53). The above data confirm that HIV-infected persons with osteonecrosis have between 1.8-fold and 2.6-fold more peripheral neuropathy than those without osteonecrosis and support the inference that low levels of NGF could be among the biological underpinnings of osteonecrosis. Antibodies against NGF impair nerve function and, in a clinical trial, resulted in osteonecrosis [8]. Low levels of NGF were seen in a group of 131 HIV-infected patients [12], and have been described in patients with diabetic peripheral neuropathy [13]. NGF recognizes a variety of ligands relevant to bone regeneration, including VEGF receptors and receptors on osteoblasts. The role of reduced levels of VEGF in the pathogenesis of osteonecrosis is shown by the observation of osteonecrosis in some patients given a monoclonal antibody against VEGF [14]. Corticosteroids and alcohol abuse, which together are associated with 90% of cases of atraumatic osteonecrosis of the femoral head [15], are both associated with low levels of VEGF [16–20]. In summary, excessive peripheral neuropathy in patients with osteonecrosis not only suggests that NGF and VEGF contribute to the pathogenesis, but also supports the hypothesis that impaired regenerative capacity of osteoblasts, imposed by these cytokines, is responsible for the excess of osteonecrosis in HIV-infected patients. Acknowledgements Conflicts of interest There are no conflicts of interest.