Lipoatrophy and mitochondrial DNA assays: see all, know all?

In 1999 we hypothesized that nucleoside analogue reverse transcriptase inhibitors (NRTI) might play an important role in the pathogenesis of lipodystrophy with mitochondrial dysfunction through the inhibition of DNA polymerase gamma [1]. By now, both cross-sectional and (retrospective analyses of) randomized studies have indeed shown that NRTI, in particular stavudine, are clearly associated with the development of lipoatrophy over time. Recently, the only prospective study of this kind, comparing tenofovir with stavudine, endorsed these earlier findings [2]. Furthermore, at adipose-tissue level, it was shown that especially in patients with exposure to stavudine, serious depletion of mitochondrial DNA (mtDNA) occurred [3–7], as well as mitochondrial dysfunction [8]. Finally, in three prospectively conducted studies, a moderate increase of peripheral subcutaneous fat was observed, when patients suffering lipoatrophy switched away from stavudine [9–11]. In this issue of the journal, McComsey et al. further elaborate on one of these switching studies (TARHEEL), providing the link between these clinical observations and the earlier observed stavudine-induced mtDNA depletion. In a substudy on 16 patients, they demonstrate in adipose tissue the restoration of depleted mtDNA after stavudine was switched to abacavir, as well as some improvement in mitochondrial function (complex I). Furthermore, they add further proof to earlier observations [12] that increased apoptosis might be the final pathway into lipoatrophy, which decreased after stavudine was changed. Studies like these are difficult to perform: material is obtained by invasive procedures and especially in prospective studies, this approach requires motivated collaboration of the patients involved. Furthermore, samples are very often too small to check for the texture of the tissue or to perform extra (duplicate) tests. Especially in patients with lipoatrophy, one might easily have an overrepresentation of connective tissue, since most adipocytes might have disappeared by then. Comparing samples between individuals (especially patients versus healthy controls) without exploring these heterogeneities, makes good interpretation delicate. Connective tissue cells (fibroblasts etc) contain much less mitochondria (and mtDNA) than adipocytes. In the present study, the fact that longitudinal samples were taken in the same patients, covers these caveats in part. To overcome this tissue sampling problem, many investigators have used more accessible tissue: blood. Studies have already been published [13,14], evaluating the effects of NRTI on mtDNA in peripheral blood mononuclear cells (PBMC) and expensive assays have been commercialized and advocated for routine patient care. However in a recent evaluation among laboratories, the same samples showed considerable variance of 96% (non-log transformed values) in measuring mtDNA content [15]. Furthermore, as was discussed for adipose tissue, PBMC also form a heterogeneous population of different cells (B- and T-lymphocytes, monocytes etc.), all in different stages of activation. Since the kinetics of mtDNA and mitochondrial proteins in these different cell-types are unknown, simply measuring mtDNA content in PBMC, without performing the appropriate checks for homogeneity, will too easily lead to conflicting results [7,16]. In fact, McComsey et al. found some increase of mtDNA content in PBMC after switching stavudine for abacavir, while Hoy et al. were unable to find such an improvement in a similar switching study [17]. Finally, untreated HIV-infected patients have been found to have ‘lower mtDNA’ content in PBMC than healthy controls [18] and spontaneous improvement of mtDNA in PBMC to pretreatment (or higher) levels was noted without any interruption of the previously mtDNA-depleting antiretroviral drug combination [19]. These reports underscore the need for controlling samples both for cell-type as well as for activation status before conclusions can be made. For the future, only studies, that analyse drug toxicity in detail at the level of the representative tissue involved will finally teach us the definite aetiology of lipoatrophy.