Lipodystrophic Syndromes and Hyperlipidemia in a Cohort of HIV-1-Infected Patients Receiving Triple Combination Antiretroviral Therapy With a Protease Inhibitor

Objectives: To assess the frequency and features of lipodystrophic syndromes in HIV-1-infected patients receiving highly active antiretroviral therapy (HAART) with a protease inhibitor (PI), and examine whether clinical and biologic abnormalities are always associated in these conditions.

Cited Here...: Retrospective-prospective single-center observational study of 175 patients. Comparisons for continuous variables by t- test and paired t- test, and Kaplan-Meier analysis of time to onset of lipodystrophy were performed.

Cited Here...: In all, 51 patients (29%) had morphologic changes, after a mean HAART duration of 20.0 ± 6.1 months, and were categorized into pure lipoatrophy (n = 16), mixed syndrome (truncal fat accumulation and face or limb lipoatrophy) (n = 30) or pure truncal fat accumulation (n = 5). Because of the small number, the latter group was not analyzed statistically. No differences were found among patients with lipoatrophy, mixed syndrome, or no lipodystrophy, in terms of gender, CD4 count, and HIV RNA plasma load at time of HAART initiation, nor in response to treatment. Patients with a mixed syndrome were older. Patients with lipoatrophy had longer duration of HIV disease, pre-HAART exposure to nucleoside analog therapy, and HAART. Baseline and pre-HAART fasting triglyceride levels were higher in patients who developed lipoatrophy, whereas weight and fasting cholesterol were higher in patients who developed a mixed syndrome. After 12 and 24 months on HAART, triglycerides and cholesterol rose significantly in all patients, independently of lipodystrophy, whereas these parameters were not increased during nucleoside analog therapy.

Conclusions: Nucleoside analog exposure appears as a risk factor for lipoatrophy. Age and nutritional status (reflected by baseline weight, triglycerides and cholesterol) may influence the evolution to lipoatrophy or a mixed syndrome. Hyperlipidemia is observed in the absence of lipodystrophy and depends on PI exposure.

Highly active antiretroviral therapy (HAART) regimens allow a dramatic decrease of plasma HIV RNA and recovery of CD4 cells in HIV-infected patients. Concerns have been raised about those side effects of HAART that are generally termed lipodystrophy ^(1-6) and their potential association with increased vascular risk ^(7-10). These syndromes include fat wasting in the face and limbs, truncal visceral fat accumulation (buffalo hump) ^(1-6), insulin resistance, and elevation of blood lipid levels ^(2,6,11-16). Several pathophysiologic hypotheses have been proposed for these abnormalities ^(17-20) and a role for antiretroviral drugs, particularly protease inhibitors (PIs), is highly suspected. However, no exclusive relationship between the use of one particular class of drug and the occurrence of lipodystrophy has been established to date, because nucleoside reverse transcriptase inhibitors (NRTIs) may play a determinant role ^(21-23), possibly through mitochondrial toxicity ⁽²⁴⁾. In the present single center study, a cohort of HIV-1-infected patients receiving HAART regimens including a PI, were retrospectively and prospectively examined for clinical lipodystrophic syndromes and biologic abnormalities.

PATIENTS AND METHODS

A cohort of 175 patients (126 men and 49 women, mean age: 40.1 ± 10.0 years) receiving HAART regimens including a PI, and remaining on such a regimen throughout follow-up were examined regularly (every 3 months). Standardized questionnaires were filled prospectively from initiation of HAART for weight, clinical events, drug prescription, CD4 count, viral load, fasting blood glucose, cholesterol, and triglyceride levels, after patients gave written informed consent. Morphologic changes were systematically notified after the first reports of the lipodystrophic syndromes have been presented: facial and limb wasting, as well as thoracic and abdominal fat accumulation were systematically assessed semiquantitatively, according to patients' physical complaints and physical examination, which were independently recorded. For the period before HAART, biologic data were retrospectively examined and included in the longitudinal analysis, whenever available.

Measurements of fasting blood glucose, triglycerides, and total cholesterol were performed according to standard procedures (using the Hitachi 917 automat, Roche Diagnostics, Melan, France), in the same hospital laboratory. Viral load determinations (Roche Amplicor and Roche Amplicor Ultrasensitive assays, Roche, Basel, Switzerland) and CD4 count determination (flow cytometry, Becton-Dickinson FACscan, San Jose, CA, U.S.A.) met the requirements of interlaboratory quality control.

Statistical analysis was performed using SAS package v.6.12 (SAS Institute Inc., Cary, NC, U.S.A.). Comparisons for continuous variables were done by t- test, paired t- test and analysis of variance (ANOVA) with multiple comparison adjustment when appropriate. Time to onset of lipodystrophic syndrome was also analyzed as censored data. Cumulative probability of event was computed by the Kaplan-Meier method. Comparisons of cumulative probabilities were done by log-rank test: median values of continuous variables were used as cutoffs. Influence of pre-HAART nucleoside exposure was examined by Cox proportional hazards models, adjusted for baseline CD4 and plasma HIV RNA load. Patients with pure truncal fat accumulation were excluded from statistical analysis because of their small number (n = 5).

RESULTS

Some 51 patients (29%) developed lipodystrophic syndrome after a mean 20.0 ± 6.1 months on HAART: 16 were classified as pure lipoatrophy, 5 as pure truncal fat accumulation, and 30 as mixed syndrome of fat redistribution, associating peripheral lipoatrophy, and truncal fat accumulation. Mean times between HAART initiation and appearance of lipodystrophic syndrome in these three groups were 21.8 ± 6.6, 19.3 ± 5.7, and 19.2 ± 6.0 months, respectively (not significant [NS]). At 2 years of follow-up, probability of developing lipodystrophic syndrome was 0.24 (95% confidence interval [CI], 0.16-0.30) (Fig. 1).

Fig. 1

Four patients, 1 in the group with pure lipoatrophy and 3 with mixed syndrome, developed diabetes mellitus that required insulin therapy. One of them, who was also a heavy smoker, also had myocardial infarction in the setting of major hypertriglyceridemia. All these events occurred in the 3 months after the clinical diagnosis of lipodystrophic syndrome.

Table 1 shows the demographic features and treatment history of the study population, divided into three groups: patients who subsequently developed predominant lipoatrophy, mixed syndrome, and no lipodystrophy at time of analysis. There were no significant differences in terms of gender predilection. Patients with a mixed lipodystrophy syndrome were significantly older (p < .005). Patients with lipoatrophy had significantly longer durations of known HIV diagnosis (p < .05), pre-HAART NRTI therapy (p < .005), and HAART exposure (p <10^-4). They also had received a significantly greater mean cumulative number of nucleosides before HAART (p < 10^-4). Accordingly, the percentage of patients never exposed to antiretroviral therapy at time of HAART initiation was significantly higher among nonlipodystrophic patients (p < .005). Number of different HAART regimens and PI use were no different among groups. No difference could be found in the use of a single NRTI among groups (Table 2). No patient received glucocorticoids, anabolic steroids, or growth hormone during follow-up.

Table 1

Table 2

Of all patients, 74 (42.3%) were treatment-naive at the start of HAART: compared with nucleoside-pretreated patients, they were significantly less immunocompromised at that time (CD4 , 297.1 × 10⁶ cells/L vs. 216.3 × 10⁶ cells/L;p < 10^-4) but their plasma HIV RNA viral load was not significantly different (4.6 ± 0.73 log₁₀copies/ml vs. 4.4 ± 0.79 log₁₀copies/ml, NS). Pretreated patients had received nucleosides for a mean duration of 2.37 ± 1.65 years. Length of pre-HAART treatment was significantly associated with the occurrence of lipoatrophy, even after adjustment for baseline CD4 and plasma HIV RNA values (p = .02).

Immunologic and virologic features of patients are summarized in Table 3. Globally, at HAART initiation, patients with lipoatrophy were slightly more immunocompromised, as shown by mean pre-HAART CD4 counts, in a relationship to longer known duration of HIV infection, although this did not reach statistical significance. More importantly, no difference in the immunologic or virologic response to HAART could be found between patients with and without lipodystrophy. There were also no differences in terms of CD8 cell count at baseline and throughout evolution (data not shown). Of note is that lipodystrophy did not occur only in patients with complete viral load control.

Table 3

Table 4 details the longitudinal evolution of weight, fasting glycemia, blood triglyceride, and total cholesterol at baseline (the oldest available values available in the charts before any antiretroviral treatment), at time of HAART initiation, at time of lipodystrophic syndrome when applicable, and at months 12 and 24 of HAART. For all these variables, values at baseline and at time of HAART initiation were not statistically different between groups. Although some patients gained weight during the first months on HAART, mean weight was not significantly modified during the follow-up period. Similarly, mean body mass index did not vary significantly from baseline to time lipodystrophy was diagnosed in any group (from 21.6 ± 2.6 to 20.9 ± 2.6 kg/m², from 24.1 ± 3.7 to 23.7 ± 3.4 kg/m², and from 20.7 ± 1.5 to 20.2 ± 2.6 kg/m² in patients who developed pure lipoatrophy, a mixed syndrome or pure truncal fat accumulation, respectively, [NS]). Excluding the 4 patients who developed overt diabetes, mean fasting glycemia did not vary significantly during follow-up of lipodystrophic or nonlipodystrophic study subjects.

Table 4

Blood lipid levels rose significantly in lipodystrophic patients, as well as in nonlipodystrophic patients, in comparison with pre-HAART values. Triglycerides and cholesterol tended to be higher in lipodystrophic patients at the time of lipodystrophy diagnosis than in nonlipodystrophic patients at months 12 or 24 of HAART, the highest values for triglycerides and cholesterol having been reached in patients with lipoatrophy and truncal fat accumulation, respectively. Interestingly, as shown in Table 5, distribution of triglyceride and cholesterol values in lipodystrophic patients indicates that a relatively high proportion of patients still had values within normal limits and some patients had no significant elevation of these parameters. Of special note is that no significant variation of triglyceride and cholesterol levels were observed during the time on NRTIs, before HAART initiation in either group (Table 4 : comparison between baseline and HAART M0 values; in all cases, p > .05).

Table 5

In addition, as shown in Table 6, baseline and pre-HAART weight and cholesterol were significantly higher in patients who eventually developed a mixed syndrome (p < .05 in both cases), whereas baseline triglycerides were higher in patients who were to develop lipoatrophy (p < .005).

Table 6

DISCUSSION

Lipodystrophic syndromes in patients receiving antiretroviral therapy are probably heterogeneous because clinical descriptions indicate a variable association of lipoatrophy, fat accumulation, and redistribution of subcutaneous fat towards visceral adipose tissue ^(1-6). The current study confirms that three syndromes can be individualized on a purely clinical basis: lipoatrophy of face and limbs, abdominal and/or thoracic adiposity, and a mixed fat redistribution syndrome with features of lipoatrophy and central adiposity. Because we did not perform precise anthropometric measurements, we may well have missed moderate morphologic changes, and such measurement should definitely be included in further studies, but we feel that careful record of patients' complaints and clinical examination in a homogeneous population followed in a single center by a small number of clinicians limits this possibility. The present analysis took into account only overt clinical changes noticed by patients and physicians, and did not include symptoms related to PI toxicity and sometimes assimilated with lipodystrophic syndromes, such as dry skin, hair loss, or ingrown toenails ⁽²⁵⁾, and biologic abnormalities were analyzed independently: in the absence of a recognized case definition, all these factors can explain the relatively low frequency of lipodystrophy in this cohort, as compared with others in similar settings ^(2,15).

However, cumulative probability of presenting with lipodystrophic syndrome sharply increased with time in this cohort, as in others, reaching 0.42 (95% CI, 0.31-0.50) at 2.5 years of follow-up. Moreover, we did not perform quantitative measurements of visceral fat by computed tomography (CT) scan and the possibility remains that patients with pure clinical lipoatrophy may also have some degree of visceral fat accumulation ⁽²¹⁾. However, in the group of patients studied, we did not note postdiagnostic modifications with time of the pattern of lipodystrophic syndromes that would indicate an evolution from pure lipoatrophy toward a mixed syndrome of fat redistribution.

Interestingly, we found that patients with a mixed syndrome of lipodystrophy were older than patients with lipoatrophy or no lipodystrophy, which may reflect the role of age-dependent modifications of lipid metabolism and insulin sensitivity in the response to metabolic perturbations induced by antiretroviral drugs, as previously suggested ⁽⁶⁾. These patients also had higher weights and fasting cholesterol values at baseline (i.e., before any treatment) and just before the initiation of HAART, whereas patients who eventually presented with lipoatrophy were characterized by higher baseline and pre-HAART fasting triglycerides. This may reflect nutritional status before active antiretroviral treatment, because it is well known that wasting, hypocholesterolemia, and hypertriglyceridemia are features of untreated HIV infection, as a result of the uncontrolled production of proinflammatory, catabolic cytokines ^(26,27). This would suggest that patients with more advanced HIV infection and metabolic consequences thereof are more likely to develop predominant lipoatrophy. Association of lipoatrophy with longer known duration of HIV infection adds credence to this hypothesis. Conversely, patients with a better metabolic pre-HAART status appear more likely to present with a mixed syndrome, including the fat accumulation component.

The finding that lipoatrophic patients had longer exposure to nucleosides analogs before HAART and longer HAART exposure argues for a role of the cumulative toxicity of antiretroviral drugs in the loss of adipocytes alone or with modifications of their functions. Exposure to NRTIs could play a definite role, as suggested by the overrepresentation of the nucleoside-pretreated patients in the group who developed lipodystrophy, which cannot be accounted for by HIV disease history only because their mean viral load was no different from that of the study subjects who were treatment naive at the time of HAART initiation. Lipodystrophy has been described initially in patients receiving PIs ^(1-6), but use of these drugs in HAART regimens has resulted in improved long-term survival and much longer use of NRTIs as they became available in clinical practice. Other studies have found that nucleosides may be responsible for lipodystrophic syndromes, and some researchers have even suggested that stavudine or lamivudine were strongly associated with lipoatrophy ^(21-23). The introduction of PIs to therapeutic regimens may also accelerate the pathophysiologic mechanisms leading to clinical lipodystrophy ⁽²⁸⁾.

We could not find an association between the occurrence of lipodystrophy and the biologic response to HAART, and thus we would not favor the hypothesis according to which adipocyte and lipid metabolism may be influenced by the correction of immunologic and/or virologic disturbances, although we acknowledge that careful examination of, for instance, proinflammatory cytokines at the cell or tissue level remains necessary to clarify this issue, as suggested by recent findings ^(29,30).

Of interest is that elevation of blood triglyceride levels was noted in all groups of lipodystrophic patients, who had reached the highest values, but also in nonlipodystrophic patients. Classification of patients into groups with normal and abnormal lipid values may be relevant for the assessment of vascular risk (Table 4). Examining the evolution of the mean lipid values allows the description of the impact of antiretroviral drugs on these parameters (Table 3): in this cohort, hypercholesterolemia and hypertriglyceridemia appeared only with prescription of PIs in HAART regimens and were not seen during pre-HAART nucleoside analog therapy. That lipid abnormalities were not constant, even in lipodystrophic patients, indicates that clinical and biologic abnormalities are not always associated. Whether they represent a different degree of perturbation of lipid metabolism resulting from the same pathogenic process remains to be determined. Of note is that cholesterol and triglycerides rose only after the introduction of PIs: if one admits the influence of nucleoside exposure on the development of lipoatrophy, then one could probably dissociate hyperlipidemia and clinical abnormalities, both in terms of mechanisms and origin.

Glycemia was not modified in patients receiving HAART , except for those who developed overt diabetes. However, it has clearly been demonstrated that insulin resistance is prevalent in this patient population, even in the setting of normal fasting glycemia ⁽¹⁵⁾. Considering the lipid abnormalities in HAART-exposed HIV patients and their similarities with those observed in other forms of congenital or acquired lipoatrophy ⁽³¹⁾, insulin-resistance may very well be a key element for the understanding of this syndrome.

In conclusion, we found in this population that elevation of blood triglycerides and cholesterol is independent of clinical lipodystrophic syndromes and NRTI use and appears to depend on PI exposure ⁽³²⁾. Data from the current study also suggest that among lipodystrophic syndromes of patients receiving HAART, pure lipoatrophy and mixed fat redistribution syndrome can be distinguished on the basis of age, duration of HIV infection, durations of pre-HAART exposure to NRTIs and HAART, and pretreatment nutritional status as reflected by weight and by measurement of levels of cholesterol and triglycerides. Cumulative toxicity of nucleosides may play a determinant role in lipoatrophy, and mitochondrial toxicity ⁽²⁴⁾ should therefore be examined as a pathogenic hypothesis.

Acknowledgments:

The authors thank Marianne Burgard, Marie-Laure Chaix, and Christine Rouzioux, at the Virology Laboratory of the Necker Hospital, for their continuous support and the very high quality of the medical virology work performed on patients; José Timsit, at the Clinical Immunology Department of the Necker Hospital, for fruitful discussions and support, and Aline Maignan for valuable help in data management.