Nucleoside analogue-related mitochondrial toxicity and link to lipoatrophy: effect of non-thymidine analogues abacavir and tenofovir

Simon Collins, HIV i-Base

The most significant research into lipoatrophy over the last few years have come from several Australian research groups. The plausibility for the link between mitochondrial toxicity and lipoatrophy is convincingly strengthened by research that correlates peripheral and facial fat loss with reduced mitochondria cell number in adipocytes).

HIV itself may have an impact. There is a correlation of CD4 count and mitochondrial depletion in PBMCs in treatment naïve patients that improves following ARV treatment that includes ddC and ddI, that suggests HIV itself may also reduce mitochondria in adipocytes.

Last year, David Nolan’s group reported mitochondria are similarly reduced in the following patient groups: d4T-treated > AZT-treated > ARV-experienced > ARV-naïve > HIV-negative with median mitochondria copies/adipocyte of 234, 537, 1169 and 1586 respectively, see also table below). Pathophysiological effects in adipocyte damage from fat biopsies in the same patients, showed progressively worsening adipocyte structure that correlated directly with reductions in mitochondria.

Longitudinal studies have shown differential impact of thymidine analogues. While BMI was 24kg/m2 in each group, DEXA scan confirmed percentage of leg fat to be 12%, 17% and 24% in the d4T, AZT and naïve groups respectively.

This year, Nolan presented new data from 190 biposies from just over 100 individuals, on the level of mitochondrial DNA content in adipocytes related to thymidine and non-thymidine analogs; with supporting chronology. Mitochondrial DNA levels at baseline had no correlation to CD4%, viral load, age or BMI.

Early mitochondrial depletion from d4T- or AZT-containing regimens (around –260 copies/cell per month; 60% reduction after 6-12 months), lead to adipocyte damage 3-12 months later, and subsequently symptomatic lipoatrophy over the next 10-40 months. This was statistically significant for both drugs, with the d4T causing twice the effect. Switching from d4T to AZT increased mtDNA.

Patients starting treatment with non-thymidine nucleosides, showed similar levels of mitochondria in adipocytes to ARV-naïve and HIV-negative patients. (non-significant reduction of –69 copies/cell/month; p=0.6) and people switching to from d4T to AZT or abacavir, or from AZT to abacavir saw mitochondrial number increase 3-11-fold over 1-24 months (p=0.01).

Adipocyte mtDNA copies/cell by NRTI exposure

Exposure group n no. mt/cell (range) log mt/cell
ART naïve 34 1427 (413-6570) 3.19 100%
AZT-containing 41 761 (94-2846) 2.89 49%
d4T-containing 35 250 (61-2287) 2.44 18%
non-thym (ABC or TDF) 19 1675 (916-4180) 3,23

Switch studies from d4T or AZT to abacavir (Carr et al) have already showed reversal of leg fat, and the Gilead 903 tenofovir study showed low incidence of peripheral fat loss, and both clinically support the benefit of non-thymidine nucleosides. This study provided additional data to support the non-significance of mitochondrial involvement of these newer drugs.

Mitochondrial protein was statistically higher in d4T- and AZT-containing regimens (p=0.032 and 0.009 respectively, compared to naïve patients) and PPAR-gamma significantly lower (p=0.021, d4T compared to naïve), Gene expression represent total adipose tissue rather than adipocytes (approximately 30% adipocyte and 70% stromal-vascular, including pre-adipocytes. ICE/Caspase1 expression, involved in the induction phase of apoptosis or pre-adipocytes correlated with mtDNA (p=0.02) and d4T exposure (p=0.04), but not to abacavir or PI.

Lipoatrophy should therefore be viewed as a dynamic event that is confounded by individual changes in tissue expression over time as it becomes progressively more severe. Progression can be very rapid (ie leg fat reduced to 10% over 10 months treatment with d4T) and may continue after any switch in treatment.

Clearly not all nucleosides are equal and in vitro and in vivo studies seem to have produced a consensus agreement of greatest negative impact of d4T > AZT > 3TC > abacavir = tenofovir,


We do not know the level of mitochondrial damage that triggers clinical symptoms, and there is still no data on the role of ddI.

While leg and arm fat may be reversible to some degree in some patients, this is a slow process dependent on individual regenerative capacity. Stankov and colleagues suggested that while both d4T and AZT affect adipocyte differentiation, only d4T is directly toxic to pre-adipocytes.


Nolan D. Differential effects of nucleoside reverse transcriptase inhibitor (NRTI) regimens on adipocyte mitochondrial DNA depletion in HIV-infected patients. 6th Lipodystrophy Workshop, Washington, 2004. Abstract 16.

Links to other websites are current at date of posting but not maintained.