In general, the prevalence rates of LD were reported with ranges from 7 to 84% (average of 42%) in patients receiving PI-based cART, and from 0 to 38% (average of 13%) in those receiving NNRTI-based cART from various populations and countries with different clinical and metabolic characteristics [19
]. This study, based on data from a multicenter observational database in the Asia-Pacific region, showed a lower prevalence of LD than the rates of 43 to 53% seen earlier in the cART era in large cohorts in Europe, Australia, and the United States [20
]. However, in a Spanish cohort reported more recently, the prevalence of LD was 17.8% (420 of 2358) [23
] and, according to data from a Swiss HIV study, patients starting cART in 2003–2006 were significantly less likely to experience LD than those starting between 2000 to 2002 [24
]. As treatment patterns change with a decrease in thymidine analogue (d4T/AZT/ddI) use and an increase in tenofovir disoproxil fumarate (TDF) use, it is likely that LD rates will also decline [24
]. To our knowledge, our analysis is the first regional cohort study of LD in the Asia-Pacific. Previous singleinstitution studies in Thailand, Singapore, and South Korea reported the prevalence of LD as ranging from 3.5% to 66.1% [11
]. Because LD was identified clinically using a high threshold (i.e., severity grade ≥ 3) and without the use of quantitative tools such as dualenergy X-ray absorptiometry (DEXA) or computerized tomography (CT) scans, the true prevalence of LD may be higher than what we have reported. Also, the lower prevalence of LD in our sites might have been caused by different host factors such as unknown genetic background and insufficient concern for LD in several resource-limited settings. Actually, a few previous studies showed that LD was infrequent in non-white races as compared to caucasians [22
], and that genetic variations can also influence the emergence of LD [7
]. Further studies are warranted to confirm a more objective prevalence of LD through universalized and validated case definitions in the Asia-Pacific region.
We confirmed that the use of d4T is a strong risk factor for the development of LD in the Asia-Pacific region. Our results are consistent with other studies that show that treatment with NRTIs, and particularly with d4T, which has the greatest mitochondrial toxicity in the class, and longer duration of cART are key risk factors for the development of LD [20
]. The finding that patients receiving AZT for longer than the MD of treatment were less likely to have LD may have been due to their relatively shorter exposure to d4T.
Our finding that patients living in high-income countries had a tendency toward a lower prevalence of LD than those in low-income countries might have been related to the wider range of available antiretroviral drugs and lower reliance on d4T in highincome countries.
It was unexpected that patients with longer PI treatment would have lower risk of developing LD, regardless of the duration of d4T treatment. Treatment with PIs is known to be an important attributable factor for the development of LD, and NNRTIs were not traditionally considered to be associated with the development of LD [24
]. However, recent clinical trials have revealed that limb fat gain in patients receiving EFV was lower than in treatment with ritonavir-boosted lopinavir (LPV/r) [29
]. In addition, a recent AIDS Clinical Trials Group (ACTG) A5142 study showed that lipoatrophy was more frequent with EFV than with LPV/r when combined with d4T or AZT [27
]. These findings suggest a protective role of ritonavir-boosted PIs for LD. The mechanisms for the protective role of ritonavir-boosted PIs are not well understood, but a potential explanation is that ritonavir may mitigate the mitochondrial damage caused by thymidine analogues [27
]. Although PIs, especially ritonavir-boosted, may have potential protective roles for LD, other metabolic complications, such as dyslipidemia and insulin resistance, and the tolerances of drugs associated with PIs should be prudently considered when choosing an antiretroviral drug [27
Our study had several limitations. The lack of a uniform, objective, diagnostic method for identifying LD could have led to a selection bias. In this study, we tried to capture the rate of LD in the Asia-Pacific sites that are capable of diagnosing LD. However, the sites themselves that have ever reported LD may add confounding factors such as antiretroviral regimens, race, and HCV infection. Although we limited the inclusion criteria to higher-grade LD to reduce this risk, cases of LD could have been missed, due to local variations in diagnosis and reporting. As this is an observational cohort across centers with varying levels of clinical and monitoring capacity, not all patients had baseline levels for all possible variables. We excluded pre-ART lipid and glucose tests from our final analysis for this reason, but acknowledge that other missing data could have impacted our findings.
Stavudine is one of the most commonly used NRTIs in the world and in the Asia-Pacific region [5
]. TDF is unavailable at many sites in the Asia-Pacific region, so d4T or AZT in combination with 3TC is usually the only NRTIs available for standard first-line regimens [5
]. Our findings emphasize the importance of phasing out d4T use and increasing access to TDF to minimize the risk of developing LD.