Background: Avascular necrosis (AVN) of the hip has been increasingly recognized in HIV-infected persons. We, and others, reported a higher incidence of AVN since the introduction of HAART in 1995. Some studies suggest that AVN is associated with protease inhibitor (PI) use, while others have not. The cause of AVN in HIV-infected persons is not known. We assessed the incidence of symptomatic AVN and related factors in a large HIV clinical cohort that has collected comprehensive demographic, clinical, and therapeutic data longitudinally since 1990.

Methods: A nested case-control study was designed to assess risk factors associated with development of AVN. Five controls were matched to each AVN case by date of enrollment into the cohort and subsequent duration of follow-up (±60 days). AVN was confirmed in all cases by radiographic confirmation and record review. Mantel-Haenszel was used to examine categorical variables and Wilcoxon-rank sum test was used to examine continuous variables.

Results: Only 2 cases of AVN were reported from 1990–1995 with the remaining 27 cases occurring 1996 and later. AVN incidence after 1995 is shown below.

 

Year	Cases	Person-years of follow-up (PY)	Rate/1000 PY **
1996	1	1,241	0.81
1997	1	1,554	0.64
1998	4	1,811	2.11
1999	6	1,897	3.16
2000	5	1,919	2.61
2001–02*	10	2,969	3.37

* Through 9/02; ** p = 0.05 for trend

 

In cases diagnosed since 1996, no differences in age, sex, race, injecting drug use, or diagnosis of diabetes between cases and controls were found. There were significant (p < 0.05) differences in CD4 (mean 357 cells/mm³ cases vs 236 cells/mm³ controls), HIV-1 RNA (mean 59,0304 copies/ml cases vs 179,026 controls), total cholesterol (mean 198 mg/dl cases vs 159 mg/dl controls), triglyceride level (mean 235 mg/dl vs 197 mg/dl controls), PI use (74% cases vs 45% controls), and NNRTI use (56% cases vs 26% controls), but not in overall nucleoside RTI use. Anti-inflammatory steroid use was more frequent (37% cases vs 7% controls), but there was no difference in use of testosterone or megace. Hypertension was more common in cases (30% vs 14% in controls; p = 0.07).

Conclusions: Although symptomatic AVN is still uncommon, the overall incidence has increased since 1996. In addition to steroid use, our results suggest that higher cholesterol and triglyceride levels are associated with AVN. PI and NNRTI use also appear to increase the risk for AVN.