Mitochondrial Abnormalities in HIV-Infected Lipoatrophic Patients Treated With Antiretroviral Agents

Background: Lipodystrophy is now widely described in HIV infected patients under antiretroviral regimen with important psychological impact. But physiopathology of loss of fat mass is still debated and the role of mitochondrial impairment is not clearly defined.

Objective: To correlate clinical lipoatrophy (LA) in HIV patients with long-term treatment by nucleoside reverse transcriptase inhibitors (NRTIs) and muscular impairment related to mitochondrial dysfunction.

Methods: Ten consecutive patients with clinical LA and 10 nonlipodystrophic (NLD) individuals on antiretroviral therapy were included. Patients underwent the following investigations: dual-energy x-ray absorptiometry (DEXA) scanning and lactate kinetics during standardized exercise. The mitochondrial respiratory complex activity (III and IV) and histoenzymatic abnormalities (classified as none, mild, or severe) were evaluated on muscle tissue obtained by biopsy in deltoid muscle.

Results: Mean NRTI exposure was longer in the LA group than in the NLD group (81 ± 30 months vs. 59 ± 15 months), but mean protease inhibitor exposure was identical in both groups. Mean fat mass distribution for leg in the LA and NLD groups was 860 ± 381 g versus 1895 ± 999 g, respectively. The lactic acidosis threshold during exercise was reached in the LA group at lower workloads (mean: 45 ± 17 W in the LA group vs. 68 ± 11 W in the NLD group), and maximum power output exercise was restricted in LA patients (mean: 115 ± 30 W vs. 153 ± 28 W). Total complex activities in muscular tissue were lower in LA patients: the median (range) for complex III was 67 (1-128) versus 112 (28-143), and the median (range) for complex IV was 28 (1-70) versus 42 (1-75). Six patients had severe histoenzymatic abnormalities in the LA group versus none in the NLD group.

Conclusion: Clinical LA, confirmed by DEXA, in long-term NRTI-treated patients was associated with muscular mitochondrial dysfunction as shown by rapid lactic acidemia increase, impairment of respiratory chain activity for complexes III and IV, and mitochondrial histoenzymatic abnormalities.

Highly active antiretroviral therapy (HAART) in HIV-infected patients has dramatically reduced the incidence of HIV-related opportunistic infections and has improved survival.¹ There is now evidence that eradication of the HIV under treatment would not be possible² and that structured treatment interruption is associated with rapid viral rebound, possible emergence of HIV resistance, and a significant decrease in CD4 cell count.^3,4 These findings suggest that patients will receive combination of antiretroviral agents for the rest of their lives. The occurrence of adverse events increases with long-term HAART, however, and has a marked psychologic impairment with a real effect on the quality of life. The lipodystrophy (LD) syndrome, considered a long-term toxicity of HAART, was extremely common in epidemiologic studies and could be regarded as the most specific problem associated with antiretroviral treatment.^5,6 Different clinical phenotypes occurred in the LD syndrome: accumulation of fat (lipohypertrophy) mainly in the abdomen, breasts (especially in women), or dorsocervical fat pad (buffalo hump); loss of fat from a facial pad, buttock, or limb (lipoatrophy [LA]); or a mixed syndrome with fat redistribution (decreases in limb and facial fat accompanied by increases in trunk and abdominal fat). LD syndrome is frequently associated with metabolic disorders, including insulin resistance, hypercholesterolemia, and hypertriglyceridemia.⁷

Fat abnormalities appeared in patients who started protease inhibitor (PI)-based therapy. Some hypotheses have been suggested, such as PI interaction with hepatic receptor-binding lipoprotein, which presents homology with HIV protease. PIs could inhibit these proteins and lead to peripheral adipocyte apoptosis.^8-10 The role of cortisol and dehydroepiandrosterone was also described.¹¹

LA has been described with nucleoside analogue reverse transcriptase inhibitors (NRTIs) alone or in antiretroviral therapy including 1 or 2 NRTIs. Major predictive factors for LA occurrence reported in several studies are length of antiretroviral therapy (mainly NRTI treatment) and use of stavudine.^12-16 These findings highlight the association between NRTI intake and the occurrence of LA. The physiopathology of this toxic effect is not clearly demonstrated, however, and mitochondrial involvement in LA has been debated. Brinkman et al¹⁷ hypothesized that DNA γ-polymerase, the enzyme responsible for mitochondrial DNA (mtDNA) replication, is inhibited by NRTIs, leading to mtDNA depletion and mitochondrial dysfunction, and therefore to the development of LA. To show evidence of mitochondrial impairment in patients, it is useful to measure lactate kinetics during exercise and to perform direct analysis on muscle biopsy (functional analysis and morphologic studies).

The objective of our study was to correlate the relation between LA and severe muscular mitochondrial dysfunction in HIV patients on an NRTI-based regimen.

The occurrence of mitochondrial abnormalities in LA patients receiving NRTI-containing regimens was assessed and compared with those observed from matched patients on HAART but without clinical LD syndrome.

This study was approved by the local ethics committee, and written informed consent was obtained from all participants. The study took place at Rennes University Hospital from April 2001 to May 2002, and patients were recruited from the outpatient clinic of the Infectious Diseases Department.

MATERIALS AND METHODS

Patients followed at the outpatient clinic were screened for LD disorders (n = 721). All patients with clinical evidence of LA were selected, and history of morphologic changes under HAART during follow-up was assessed (35% of all patients). LA patients were selected by their physician on the basis of the results of physical examination (ie, subcutaneous fat loss on limb and face). The first 10 patients who gave their informed consent were included if they (1) were older than 18 years of age; (2) were receiving HAART, including an NRTI, for at least 12 months; (3) had stable disease with nonprogressive opportunistic infection; (4) had an HIV viral load <4000 copies/mL on the latest blood sample before entry and at least 2 consecutive viral loads below 200 copies/mL during follow-up; and (5) had a CD4 cell count >200 cells/mm³. Patients with homeostasis disorders and a hemoglobinemia plasma level below 10 g/L were excluded. Ten other patients were included as controls. The controls had no LD disorders on physical examination by their physician. Criteria of inclusion were similar to those used for LA patients. Controls were matched with LA patients included in the study on the basis of age, gender, CD4 cell count, and HIV viral load.

The 2 groups of patients (LA patients and nonlipodystrophic [NLD] patients) included in the study were compared with regard to their mitochondrial abnormalities.

At inclusion, patients in the 2 groups were examined at an outpatient clinic on 2 consecutive days and had the following laboratory tests: complete blood cell count, CD4 and CD8 cell counts, viral load measurement (using Roche Monitor polymerase chain reaction [PCR] with an undetectable level of 200 copies/mL), fasting cholesterol, triglyceride, glycemia, lactate, and postprandial lactate after a standardized meal.

Body fat mass and regional distributions were measured by dual-energy x-ray absorptiometry (DEXA). The underlying principle of this method is that bone and soft tissue areas can be penetrated by 2 discrete energy peaks (70 and 140 kV in our study) from an x-ray source. The degree of penetration was analyzed by a scintillation detector, and dedicated software reconstructed an image of the underlying tissue to give an estimate of its mass (HOLOGIC QDR-1000TM). This method has been shown to be accurate and reproducible when compared with other standardized techniques.^18,19 These data allow the operator to determine specific body region fat distribution, including the trunk (chest and abdomen, excluding the pelvis), leg (including the entire hip, thigh, and leg), and arm (including the shoulder, arm, and forearm). Our results were expressed as a percentage of fat and total grams of fat.

Computed tomography (CT) was performed to assess abdominal and midthigh subcutaneous fat. Patients were imaged on a CT scanner, with the slice passing through the fourth lumbar vertebra and the middle of the thigh. Results were calculated by summing the area of pixels in the slice, with CT values ranging from -150 to -50 Hounsfield units as previously described;²⁰ they were expressed in square centimeters for visceral adipose tissue (VAT), subcutaneous adipose tissue (SAT), and total adipose tissue (TAT).

Mitochondrial toxicity assessment used analysis of muscle biopsy and lactate kinetics during a standardized incremental exercise on a cycle ergometer until exhaustion.

Muscle biopsies for biochemical and histoenzymatic studies were performed in deltoid muscle and immediately frozen in liquid nitrogen.

Spectrophotometric analyses of the respiratory chain complexes were performed on homogenates from the frozen samples as described by Rustin et al.²¹ One fragment was tested for function of mitochondria oxidative phosphorylation using spectrophotometric measurement to assess the activity of the respiratory chain complexes, including complex I, II, III, and IV substrates: (1) NADH-ubiquinone oxidoreductase (complex I), (2) succinate ubiquinone reductase (complex II), (3) ubiquinol-cytochrome c oxidoreductase (complex III), and (4) cytochrome c oxidase (COX; complex IV), as well as citrate synthetase activity (Uvikon XL; Biotek Instruments, Winooski, Vermont).

A second fragment was processed for histoenzymologic studies to detect mitochondrial disorders: histoenzymatic reactions on cryostat section; Gomori trichrome staining, which detects the presence of ragged-red fibers (RRFs); and COX and succinate dehydrogenase reactions, which identify RRF equivalents. The results of morphologic mitochondrial irregularities were classified as none, mild, or severe by the anatomopathologist without knowledge of patient characteristics (blind examination).

After a biopsy was performed, patients completed a standardized incremental exercise on an electromagnetically braked cycle ergometer (ER 800 Jaegger). An electrocardiogram was recorded before and during exercise; the test started at 30 W of power, with 20-W increment increases every 2 minutes until exhaustion. Arterialized capillary ear blood samples were drawn at rest, at the end of each 2-minute step, and at the end of the test. Samples were immediately analyzed using an electrochemical enzymatic method (Kontron Lactate Analyser LA 640), which was first validated by Geyssant et al.²² According to the precision of the method, results are given to the closest 0.1 mmol/L.

Gas exchange and ventilatory variables were simultaneously analyzed breath by breath using a calibrated computer-based exercise system (Oxycon Pro Jaegger). All tests were performed in the same laboratory. Patients were asked to pedal to the limit of tolerance, with active encouragement from the same investigator. The following criteria were used to establish a maximum effort: maximum heart rate (HR) >90% of age predicted (220 - age), maximum respiratory exchange ratio (R) >1.10 or a plateauing of oxygen consumption per unit time (Vo₂), patient's exhaustion, and lactatemia >8 mmol/L. The following data were automatically calculated using standard formulas²³: oxygen uptake (Vo₂ L/min), carbon dioxide output (Vco₂ L/min), respiratory exchange ratio (R = Vco₂/Vo₂), minute ventilation (Ve L/min), ventilatory equivalent for O₂ and CO₂ (Ve/Vo₂ and Ve/Vco₂, respectively), HR in beats per minute (bpm), and oxygen pulse (Vo₂/HR mL/bpm). The average Vo₂ for the last 15 seconds of the test was considered to be representative of the subject's actual maximum Vo₂ and was compared with the predicted values. The anaerobic threshold (AT) was estimated from the lactate kinetics curve (first elevation of lactatemia above resting level: approximately 2 mmol/L) and from the ventilatory method (when Ve/Vo₂ increased and Ve/Vco₂ remained stable). The determination of AT is interesting as a marker of the transition between moderate and heavy exercise because it demonstrates the ability of the patient to tolerate a given work rate and usually occurs at approximately 55% of the predicated (or actual) maximum Vo₂ in an active healthy subject.²⁴

Statistical analysis used analysis of variance and the nonparametric Mann-Whitney U test, and associations between variables were determined using the Spearman correlation coefficient. Comparisons of the 2 groups for continuous variables during the different steps of exercise used linear model regression for repeated measures. All tests were run using SPSS (version 10.0; SPSS, Chicago, IL). Data are presented as mean ± SD unless otherwise specified. Results are considered significant at P < 0.05.

RESULTS

Patient Characteristics

Twenty HIV-infected patients, 10 with LA and 10 without LD (NLD = controls), were included in the study during a period of 6 months. All patients were on HAART at inclusion: 9 were on a regimen containing 2 NRTIs and 1 PI (6 and 3 patients in the LA and NLD groups, respectively); 2 were on 2 NRTIs combined with 1 NNRTI (1 in each group); 4 were on 2 NRTIs, 1 nonnucleoside reverse transcriptase inhibitor (NNRTI), and 1 PI (1 and 3 patients in the LA and NLD groups, respectively); 4 were on 2 NRTIs and 2 PIs (1 and 3 patients in the LA and NLD groups, respectively); and 1 LA patient was on 3 NRTIs and 1 NNRTI. It was the first line of treatment for antiretroviral therapy in 4 patients (1 and 3 patients in the LA and NLD groups, respectively); the second line of treatment, including dual therapy with NRTIs, for 9 patients (4 and 5 patients in the LA and NLD groups, respectively), and the third or more line of treatment in 7 patients (5 and 2 patients in the LA and NLD groups, respectively). Changes in antiretroviral treatment resulted from failure, intolerance, or a switch from a PI to an NNRTI for better compliance in patients with an undetectable HIV viral load in plasma. All NRTIs were used during the course of treatment: 15 patients received zidovudine (8 patients in the LA group and 7 in the NLD group), and 10 received stavudine (9 patients in the LA group and 1 in the NLD group) for at least 2 cumulative years.

Characteristics of patients at inclusion were similar for age, gender, clinical Centers for Disease Control and Prevention (CDC) stage and proportion of AIDS stage, CD4 cell count, and HIV viral load (Table 1). LA patients experienced longer exposure to NRTIs than the control group (81 ± 30 months vs. 59 ± 15 months, respectively) even if it did not reach statistical significance (P = 0.058). The NRTI exposure was similar in the 2 groups of patients for zidovudine, lamivudine, and abacavir but longer in the LA group for didanosine and stavudine as shown on Table 2. Total exposure to PIs was similar in the 2 groups, but if we considered each PI individually, indinavir was more likely prescribed in the LA group: 20 ± 22 months versus 6 ± 14 months in the NLD group (P = 0.05). HIV disease has probably been more pronounced in LA patients: they had a lower CD4 cell count at least once during the follow-up (nadir) period and were also known to be infected with HIV for a longer time and to have a shorter period of undetectable viral load <200 cells/mm³. Weight and body mass index (BMI) were lower in the patients with LA than in patients in the NLD group with a similar height: 1.68 (±0.11) m in the LA group versus 1.72 (±0.10) m in the NLD group (P = 0.5). Eight patients in LA group had an increase in serum triglycerides at inclusion, but only 5 patients in the NLD group had an increase in serum triglycerides at inclusion. Plasma cholesterol was similar in the 2 groups. Fasted lactate and postprandial lactate measures were significantly higher in LA group compared with controls (P < 0.05). Median fasting glycemia was 5.5 (range: 4.1-6.8) mmol/L for LA group and 5.5 (range: 4-5.9) mmol/L for the NLD group (normal value range: 3.9-6.1 mmol/L).

Table 1

Table 2

Fat distribution in LA patients and controls was assessed by DEXA, which confirmed physical examination findings in all patients but 2 (1 in each group). The mean ± SD is reported in Table 3. The 2 discordant patients have been examined again by their physician and by an independent medical doctor. They were confirmed and analyzed in their respective group as intent-to-treat in real-care medicine. Analysis was also performed with regard to the results of DEXA, but global results did not change (data not shown).

Table 3

CT has shown results similar to those of DEXA for fat distribution assessment in all patients. The correlation between these 2 examinations was higher in the NLD group compared with the LA group (R² = 0.79 vs. R² = 0.94, respectively).

Cardioventilatory and Exercise Parameters

All but 1 LA patient could complete the test until exhaustion. No electric or clinical sign of coronary impairment was observed in any patient before or during exercise, and the relation between HR and power output was similar in both groups as shown in Figure 1.

Figure 1

Ventilatory parameters have shown 2 mild restrictive syndromes, 1 in each group, and 2 obstructive syndromes, both in the NLD group.

Median (range) maximum Vo₂ compared with the predicted values was not significantly reduced when analyzed for the whole population: 98% (63%-115%). Nevertheless, results in the LA group trended to be slightly lower, with a wider range than in the NLD group: 91% (63%-115%) versus 104% (89%-114%), respectively, even if the difference was not statistically significant (P = 0.1).

Four of the 10 LA patients reached a maximum Vo₂ lower than 80% of the predicted value; in addition, 1 patient did not complete the maximal exercise criteria and this patient's Vo₂ peak (not included in the maximum Vo₂ calculations) was 50% of the predicted maximum Vo₂ value. Conversely, the lowest maximum Vo₂ in the NLD group reached 89% of the predicted value. Lactate production was higher in the LA group than in the NLD group at the same level of maximum Vo₂, however (Fig. 2; P < 0.05).

Figure 2

In the same way, the maximal power output on the cycloergometer was reduced in the LA group: 137 (range: 70-190) W for the whole population and 119 (range: 70-170) W for the LA group versus 153 (range: 100-190) W for the NLD group (P = 0.01); the power output was associated with higher lactate production in the LA group up to the 110-W step; for higher power steps, the number of patients was too small to detect any statistical difference (Fig. 3).

Figure 3

The AT occurred early for the whole population: 47% (range: 33%-62%) of the predicated maximum Vo₂, with no statistical difference between the groups (Fig. 4A; P = 0.1) but earlier in the LA group with regard to the power output (see Fig. 4B; P = 0.005) and HR (see Fig. 4C; P = 0.05).

Figure 4

These results provide evidence that all patients presented with abnormal lactate production; LA patients produced higher blood lactate concentrations than NLD patients for the same levels of exercise even if the maximum Vo₂ values were not significantly different between the 2 groups.

Muscle Biopsy

The biopsy was done in deltoid muscle and was divided into 4 pieces: 2 for functional analysis and 2 for histoenzymatic studies. No complications (hemorrhagic or infectious) were seen after muscle biopsy in any patient.

Functional Analysis by Spectrophotometry

The 4 mitochondrial respiratory complex activities were measured (complexes I, II, III, and IV) and expressed in nanomoles per minute per milligrams of proteins. Values represented the total activity of the enzymatic chain reaction and were compared in each group with the normal values.

One LA patient had undetermined results for the 4 complexes (equal to 0, corresponding to a negative slope for related activity of the complex in optic density) and was excluded from the analysis. Two other LA patients had values under the level of interpretable signal for complex IV, one LA patient had values under the level of interpretable signal for complexes III and IV, and one NLD patient presented with undetermined values for complexes I and IV. For these patients, the value 1 has been given in the place of undetermined results.

For complex I, the median (range) in LA patients was 17 (12-40), and it was 24 (1-40) in NLD patients (P = 0.2), ranging from 21 to 49 for normal values. For complex II, the median (range) in LA patients was 24 (16-37) and 20 (10-37) in NLD patients (P = 0.3), ranging from 22 to 40 for normal values; there were no differences between the 2 groups (Table 4).

Table 4

For complex III, however, the median (range) in LA patients was 67 (1-128) and 112 (28-143) in NLD patients (P = 0.03), ranging from 114 to 192 for normal values. For complex IV, the median (range) in LA patients was 28 (1-70) and 42.5 (1-75) in NLD patients (P = 0.03), ranging from 45 to 93 for normal values. Lower activity was shown in the LA group compared with the NLD group, which reached statistical significance as shown in Table 4.

Morphologic and Histoenzymatic Analysis

Eight patients in the LA group presented with histologic and enzymatic abnormalities related to mitochondropathy (Table 5): 6 with severe defects (presence of COX-negative fibbers plus morphologic abnormalities) and 2 with mild defects (presence of COX-negative fibers or morphologic abnormalities on standard coloration). Five patients in the NLD group presented with mild abnormalities (presence of COX-negative fibers or morphologic default); patient 19 had both.

Table 5

LA patients who presented with morphologic and/or enzymatic disorders classified as severe or mild (8 patients) experienced worse exercise performance compared with LA patients with no abnormalities (2 patients): HR (bpm) at AT: 119 (±19) versus 105 (±8), percentage of maximum Vo₂ at AT: 43(±8) versus 57 (1 patient), and power output at AT (watts): 44 (±19) versus 50.

When correlation tests were performed (Spearman correlation coefficient) in all the study patients, a positive correlation was found between the time exposure to stavudine and the decrease in activity in complexes III and IV (P = 0.04 and P = 0.03, respectively) and between the time exposure to didanosine and the decrease in activity in complex III (P = 0.04), but no differences were found when time exposure to other NRTIs and other complex activity were considered.

Correlations were also found between the presence of morphologic abnormalities in the muscle biopsy and a higher HR at the AT (P = 0.001) for the whole population and between lower complex III activity and a higher percentage of maximum Vo₂ at the AT (P = 0.03) for the NLD group.

DISCUSSION

We investigated the relation between LA and mitochondrial abnormalities occurring in the muscle of HIV-infected patients on NRTI HAART therapy. We have shown that LA was associated with severe mitochondrial dysfunction on muscle compared with that in NLD patients on HAART. Function alteration of the respiratory chain measured by spectrophotometry (decrease in complex III and IV activity in LA compared with NLD; P = 0.03) and histoenzymatic study on muscular biopsy (severe defect in 6 of 10 LA patients with the presence of COX-negative fibers compared with 5 of 10 NLD patients having a mild defect and 5 of 10 NLD patients having a normal examination) were correlated to lactate production during exercise, which occurred earlier and at a higher level in LA patients (P = 0.05). In NLD patients with no or only a mild mitochondrial disorder, exercise performance and production of lactate were close to normal but below standard for sedentary people of the same age. Hyperlactatemia in routine investigation is possibly related to mitochondrial dysfunction.²⁵ Fasted lactate was normal in all patients and increased moderately after a meal. The increase was greater in LA patients. Nevertheless, only exercise could detect early and abnormal production of lactate; this was observed in all the study patients, with higher production in patients who presented with functional and morphologic mitochondrial abnormalities on muscular biopsy. Nevertheless, hyperlactatemia could be present apart from mitochondrial dysfunction in HIV patients secondary to poor tissue perfusion occurring because of circulatory insufficiency, severe anemia, or as a result of aerobic disorders, malignancies, diabetes mellitus, renal or hepatic failure, gastrointestinal disorders, or drug intake (biguanides, ethanol, methanol, and isoniazid).²⁶ In our patients, there was no anemia and the median (range) hemoglobinemia was 14.05 (12.6-15.9) in the LA group and 14.45 (12.1-16) in the NLD group. No patients experienced diabetes mellitus (normal fasted glycemia in all study patients) or organ dysfunction (HR on physical examination and renal or hepatic biologic results). There were no alcoholic patients in this study, and no other drugs were prescribed apart from the antiretroviral therapy in the study patients.

Exercise limitation has also been described in HIV-infected patients before the era of HAART.^27,28 These patients on antiretroviral therapy presented with more important muscular tiredness, showing a worse adaptation to effort, which had led to the hypothesis of heart diseases in these populations; in our patients, there were no differences between the 2 groups in terms of HR and power output (see Fig. 1), excluding heart disease involvement in the exercise limitation. Testing HIV patients during exercise is useful for assessing the AT, and determination of oxidative capacity limitations in this population correlated with mitochondrial dysfunction.^29,30 The LA patients reached the AT at much lower workloads and Vo₂ oxygen uptake than the NLD patients, suggesting impairment of mitochondrial function as previously described in non-HIV patients.³¹ Lactate kinetic level during exercise was not a sensitive examination for mitochondrial dysfunction, however, and measurement of oxidative capacity is preferred.³² Our results have shown impairment in such exercise in all patients compared with healthy individuals (Sport Biology Unit experience), suggesting the influence of HAART treatment if we consider that HIV infection is controlled in our study patients (low HIV viral load and high CD4 cell count).

Our results are not completely confirmed by those of other published studies. Roge et al³³ have reported a lower working capacity in LA patients on ergonometer cycle exercise but found no difference in mitochondrial function compared with HIV-negative controls (activity of citrate synthetase, activity of hydroxyacyl-coenzyme A dehydrogenase, concentration of inosine monophosphate, and glycogen content) with a different method of measurement using the Cobra Fara 2 (Roche Diagnosis) and chromatography. It has been suggested that measurement of enzymatic activity of the respiratory chain complexes provides a more suitable evaluation for mitochondrial activity;²¹ moreover, the authors had not performed histoenzymologic studies. Others have found mitochondrial dysfunction in LD patients using spectrophotometry analysis but did not select the studied patients on LD phenotypes and pooled LA and lipohypertrophy patients, which led to difficulty in interpretation when considering different physiologic pathways in the occurrence of the 2 LD phenotypes.³⁴

In our study, LA as mitochondrial dysfunction on muscle was associated with a longer NRTI exposure (81 ± 30 months vs. 59 ± 15 months in NLD patients) and the use of stavudine, but PI exposure was equivalent in the 2 groups. This was previously reported in multivariate analyses as risk factors of LD.^35-38 Stavudine is one of the NRTIs (with zalcitabine) that had a high affinity for mtDNA γ-polymerase compared with the other NRTI-using enzyme assays and cell cultures,³⁹ and the use of stavudine is associated with a 2.5-fold higher risk of developing LA than the use of zidovudine.¹⁴

In addition, improvement of LA was demonstrated in patients who experienced an NRTI switch from stavudine to abacavir or zidovudine,^40,41 which indirectly supports the hypothesis that the use of a toxic nucleoside analogue (eg, stavudine) for mitochondrial function is associated with the occurrence of LA.

The widespread use of antiretroviral therapy in HIV infection has included several descriptions of the toxic effect on mitochondria caused by nucleoside analogues,⁴² zidovudine-associated myopathy,⁴³ zalcitabine- and didanosine-associated neuropathy,⁴⁴ and mitochondrial disorders in infants born to nucleoside analogue-treated mothers.⁴⁵

The role of stavudine and zidovudine in adipose cell dysfunction was also recently described by an increase in cell apoptosis in culture;⁴⁶ whereas PIs were mainly responsible for the occurrence of LD and metabolic disorders by reducing the expression of sterol-regulatory-element-binding protein-1c (SREBP-1c), which is an essential step in adipocyte differentiation. These findings suggest that synergistic action may occur in LA as previously reported.¹⁴

Our study supports the hypothesis that mitochondrial impairment in muscle from long-term NRTI-treated patients could lead to exercise limitation, especially in LA patients. When correlation tests were performed in all patients, a positive correlation was found between the time exposure to stavudine and defects in complex III and IV activity and between the time exposure to didanosine and defects in complex III activity. These findings point out that clinical function impairment was directly caused by mitochondrial dysfunction and was especially marked in LA patients receiving stavudine and/or didanosine.

Our study has limitations, however, including the following. First, the physical examination was not reliable enough for detecting and assessing the LD syndrome. One male subject (patient 13) classified as NLD by clinical examination presented with fat mass distribution abnormalities on the DEXA results, and a female subject (patient 11) who presented with LA on physical examination was not confirmed by DEXA. Second, 3 patients in the LA group had no detectable complex III and/or IV activity (value <0). Two of these 3 patients have shown severe histoenzymologic abnormalities and a low percentage of predicted maximum Vo₂ at AT (37 and 43), suggesting dramatic mitochondrial dysfunction. We then considered a complex value equal to 1 for the statistical analysis in the comparison of the 2 groups.

Spectrophotometry was used to determine complex activity and showed certain limits.⁴⁷ Even if controlled tests (parametric tests) were successful in each patient's muscle biopsy, we could not exclude technical problems when no activity was detectable. These 3 1 values have dropped the median of the total complex activity in the LA group. It has been demonstrated that mitochondrial function investigation by muscle biopsy must be undertaken with caution because of the possible discrepancy between different available techniques for assessment and nonhomogeneous distribution of defective mitochondria in the tissue,⁴⁸ which could explain why we found normal histoenzymatic examination results in 2 patients in the LA group despite LA and earlier production of lactate during exercise.

The mtDNA depletion was not assessed in this study and would be relevant in the adipose tissue.⁴⁹ No reports have described the assessment of mitochondrial depletion in muscular biopsies.

In conclusion, the results of the present study show that muscular mitochondrial dysfunction could appear in all patients on an NRTI HAART regimen; these abnormalities were significantly associated with clinical LA but were also found in patients without LA syndrome. Muscular impairment should be then considered in LA patients with physical asthenia.

We cannot make conclusions on the physiopathology process of LA based on this study but have acquired increased knowledge about NRTI toxicity on mitochondrial chain reaction as a cause of muscular dysfunction involved in HIV patients, especially those who present with LA syndrome.

ACKNOWLEDGMENT

The authors thank the patients who participated in the study and the Anatomopathology Department of Rennes University Hospital; they also thank Bruno Turlin for technical assistance.