Reduced bone mineral density in HIV-positive individuals

A total of 105 HIV-positive patients underwent dual-energy X-ray absorbtiometry (DEXA) scan to assess bone mineral density (BMD). The prevalence of reduced BMD was found to be 71% and was higher in patients who had ever been treated with protease inhibitors (PI). Our results suggest a possible association between PI and reduced BMD, and further complicate the debate regarding when to commence treatment of HIV and with what agents to start.

Highly active antiretroviral therapy has radically improved the prognosis of patients infected with HIV ^[1,2]. Metabolic complications related both to living with HIV long term and the use of antiretroviral treatment have increasingly been noted ^[3,4], and several groups have reported reduced bone mineral density (BMD) in HIV infection ^[5–8], some suggesting an association with protease inhibitors (PI) ^[7]. Dual-energy X-ray absorbtiometry (DEXA) scan is used to determine BMD, and results may be categorized by T score (which compares an individual's BMD with the mean peak bone density for that sex in the general population, the WHO defining osteopenia as a T score of between 1 and 2.5 SD below the mean and osteoporosis as a T score of > 2.5 SD below the mean) ^[9]. Our study aimed to establish the prevalence of reduced BMD in a single clinic population of HIV-positive individuals, and to identify factors associated with this abnormality.

Subjects were recruited between October 1999 and November 2000. Data collected included demographic information, treatment history and AIDS-defining illnesses, body mass index (BMI), CD4 cell count (measured by flow cytometry) and plasma RNA level (Roche Amplicor, Basel, Switzerland).

DEXA scans were performed by the same radiographer on single scanner calibrated daily. Each patient was given a summary ‘normal', ‘osteopoenic’ or ‘osteoporotic’ result according to the minimum T score of the neck of femur, total femur or L1–4. For the analysis of associations with different variables those with summary scores greater than 1 SD below the mean were classified as ‘abnormal'.

Logistic regression analysis, performed using the Stata 6 statistical package, was used to derive odds ratios for the relationship between all variables and abnormal BMD.

The study consisted of 105 patients, 71% of whom were men and 75% were Caucasian. Fifty-four per cent of patients had ever smoked and 73% of men were homosexual. The median age was 40 years (range 26–60 years) and BMI was 24.4 kg/m² (range 17–35 kg/m²). The median time since HIV diagnosis was 5.6 years, and 86% had received antiretroviral therapy [69% PI, 45% non-nucleoside reverse transcriptase inhibitors (NNRTI)]. Among those who had been exposed to PI the median time since the start of therapy was 33 months (range 2–73 months). Thirty-six per cent of patients had a history of AIDS and 30% a history of self-reported lipodystrophy. The median CD4 cell count was 419 × 10⁶/l and viral load was 50 copies/ml. The median T score at the L1–4 site was −1.36, total hip −0.53 and neck of femur −0.75.

The prevalence of reduced BMD was 71% with 58% of patients having a T score in the osteopenia range and 13% in the osteoporosis range. Thirty-one (30%) patients had an abnormal T score at one site, 20 (21.5%) at two sites and 24 (19%) at all sites.

Neither age, sex nor smoking were significantly associated with the outcome (Table 1). Notably there was no significant relationship between race and reduced BMD. The nadir CD4 cell count was, however, significantly associated with reduced BMD, those with very low nadirs (< 50 × 10⁶/ml) having greater odds of an abnormal result than those with nadirs greater than 200 × 10⁶/ml [crude odds (OR) 2.19, P = 0.02], although this association did not persist in multivariate analysis. A history of AIDS was also associated with reduced BMD (crude OR 3.77, P = 0.02), but again this relationship did not persist after adjustment. There were no significant associations with the CD4 cell count or RNA level at the time of scan, BMI, duration of known HIV positivity, or self-reported lipodystrophy. Experience of treatment with any antiretroviral agent was associated with a low BMD (crude OR 2.66, P = 0.09), whereas those who had received NNRTI had a reduced odds of an abnormal BMD (crude OR 0.69, P = 0.41). Treatment with PI was significantly associated with an increased odds of abnormal bone density, both in terms of current treatment (crude OR 2.68, P = 0.03) and ever having had treatment (crude OR 3.39, P = 0.01). After adjustment, however, these relationships did not remain statistically significant.

Table 1: Odds ratios for associations with reduced bone mineral density.

Our results, in common with other studies ^[7], suggest a possible association between the use of PI and reduced BMD in HIV-positive individuals, although as all patients on therapy also receive nucleoside analogues, it is possible that the effect may be a consequence of nucleosides and PI combined. Other studies have suggested that the duration of HIV infection may also be important ^[3], but the date of seroconversion was unknown in our cohort and the duration of known HIV seropositivity was not found to be associated with reduced BMD (although the date of HIV positivity is unlikely to be an accurate marker for the date of seroconversion). A very low CD4 cell nadir and previous AIDS were associated with an increased prevalence of reduced BMD in univariate analysis, and it may be that lower CD4 cell nadirs or AIDS reflects a greater duration of infection, particularly in those diagnosed in the pre-highly active antiretroviral therapy era.

We are unaware of any data showing that reduced BMD translates into an increased risk of fracture in HIV-positive individuals. This may be because the majority of patients are young, have few visual or balance problems and are not prone to falls. Alternatively, it may be that we have unearthed a potential problem before the clinical consequences have become apparent, and we may in future see an increase in the fracture rate.

We believe that screening for reduced BMD in the HIV-positive population cannot yet be justified. Whereas reduced BMD is a potentially serious condition, its clinical significance within this patient group is uncertain, scans are expensive and we neither know the aetiology nor have any definitive treatment to offer. Early knowledge of this diagnosis may be harmful to patients who already have to cope with a chronic, incurable condition. Notably, for postmenopausal women (known to be at risk of osteoporotic fracture and for whom treatments are of confirmed benefit) the consensus is that large-scale screening cannot be justified ^[10].

In summary, our data suggest a possible association between PI and reduced BMD. The benefits conferred by PI, however, both in terms of the prevention of HIV-related disease and increased longevity, are undisputed. However, the continuing emergence of possible toxic side-effects serves further to complicate the ongoing debate regarding when to start antiretroviral treatment and which agents to use.

Antonia L. Moorea

Arvind Vashishtb

Caroline A. Sabina

Amanda Mocrofta

Sara Madgea

Andrew N. Phillipsa

John W. W. Studdb

Margaret A. Johnsona