| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
Laboratoire de Biochimie III and Institut National de la Santé et de la Recherche Médicale U326 (J.T., X.C., J.Fa., B.P.), Service des Maladies Infectieuses (E.B., P.M.), and Institut National de la Santé et de la Recherche Médicale U518 (J.-B.R., J.Fe.), Toulouse University Hospital, 31059 Toulouse-Cédex, France
Address correspondence and requests for reprints to: Bertrand Perret, Institut National de la Santé et de la Recherche Médicale U326, Hôpital Purpan, 31059 Toulouse-Cédex, France. E-mail: perret{at}purpan.inserm.fr
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In most reports, the dyslipemia associated with antiretroviral therapy has been assessed from measurements of lipids, cholesterol distribution, and of major apolipoproteins (apos) (3, 4). However, lipoproteins are heterogeneous in terms of apoprotein composition, enabling to define subclasses of lipoparticles, which differ in their pro- (or anti-) atherogenic potential (5). In front of a mild hyperlipemia, identification of the accumulating lipoparticles is helpful to precise the atherogenicity of the lipoprotein profile. For instance, increased levels of complex lipoparticles, containing apo C-III and/or apo E, and apo B have been found in coronary artery disease patients (6, 7). Several factors may contribute to the pathogenesis of hypertriglyceridemia, like an increased synthesis of triglycerides (TGs) and very low-density lipoproteins (VLDLs), a lipolysis defect, or an impaired hepatic clearance of remnant lipoproteins derived from VLDLs or chylomicrons. Apo C-III seems to be a key protein modulating both the synthesis and catabolism of VLDLs (8, 9). On the other hand, apo E plays a major role in the catabolism of remnant lipoproteins, being a high-affinity ligand for hepatic receptors (10). Thus, a precise characterization of the accumulating lipoparticles should help to identify the mechanisms underlying the development of dylipidemia.
Occurrence of a lipodystrophy (LD) has also been commonly described on antiretroviral treatment, the most frequent aspect associating a wastage of the face and limbs, and an increase in abdominal fat (1, 11, 12). Despite the elevated prevalence of this LD (between 30% and 70% among studies), the factors or metabolic indices associated with its development have not been clearly elucidated.
The present study demonstrates that the moderate hypertriglyceridemia observed on PIs is associated with marked increases in complex lipoparticles containing apo B associated with apo C-III and/or apo E and points out apo C-III as an important determinant of the associated LD.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Consecutive HIV+ patients attending the HIV follow-up
consultation of the Purpan Hospital in Toulouse, France, at the
beginning of 1999 were included. Seventy-eight patients, among them 63
males, were found eligible according to inclusion/exclusion criteria.
Informed consent was obtained from each participant. Among male
patients, 49 were under tri- or quadri-therapy including PIs, whereas
14 others received only two reverse transcriptase nucleosidic
inhibitors (RTNIs). Patients were excluded from the study under the
following circumstances: age under 18 yr, presence of a diabetes,
obesity with a body mass index (BMI) of 27 kg/m2
or more, patent hyperlipemia before setting up the antiviral therapy
(cholesterol 
6.8 mmol/L and/or TGs 2.3 mmol/L), a known
endocrinopathy (dysthyroidia, hypercorticism, etc.), hepatic
failure, presence at the time of the examination of an opportunistic
infection, and, for PI-treated patients, less than 4 months of PI
administration. Distribution of the patients among the different
clinical stages (A, B, and C) of the infectious disease was assessed
according to the classification of the Centers for Diseases Control
(1993) (13). Stage A includes asymptomatic patients and
those with a persistent generalized lymphoadenopathy. Stage B includes
minor opportunistic infections (essentially, oral candidiasis, oral
hairy leukoplakia, Herpes zoster infection, and peripheral
neuropathy). Stage C includes major opportunistic infections (among our
patients: esophageal candidiasis, extrapulmonary cryptococcosis,
cytomegalovirus retinitis, recurrent Herpes
simplex infection, HIV-associated dementia, Kaposi’s sarcoma,
non-Hodgkin’s lymphoma, Mycobacterium avium infection,
Mycobacterium tuberculosis, Pneumocystis carinii,
and brain toxoplasmosis).
Bioclinical examinations included routine determinations of the viral load, total blood cell and CD4 lymphocytes counts, the electrolytic balance, and liver enzyme activities. Besides, a complete evaluation of the lipoprotein profile was performed, and parameters of glucose homeostasis were measured.
Among PI-treated patients, some of them presented a LD, as clinically assessed by at least two of the following criteria: leaning with wasting of sc adipose tissue on the face, upper and lower limbs, and buttocks. These signs had to be associated with an increase in abdominal obesity in males, and in women, with an enhanced breast volume. Fat accumulation in the dorso-cervical area (buffalo-humps), as reported elsewhere (14), was never observed in our patients. The clinical criteria for LD were confirmed by dual-energy x-ray absorptiometry (DEXA) (15), enabling to determine body composition and fat distribution. LD was characterized by a low ratio of lower limb fat/total body fat and a high ratio of abdominal visceral fat/total body fat, by comparison with reference values obtained from age- and gender-matched individuals.
To get control reference values for all investigated parameters, 63 male subjects, living in the same geographical area, were selected so as to be matched for age with the PI-treated patients. Those subjects were recruited in a population study carried out in the frame of the MONICA center. The MONICA project is a WHO-supported multinational collaborative program, surveying the level and evolution of cardiovascular disease risk factors in 38 reference centers in 28 different countries (16), and one of these centers is located in Toulouse. Among their activities, the French MONICA centers periodically realize population studies, including questionnaires, clinical examination, and biological measurements. The recruitment is a random sampling representative of the regional population, based on polling lists.
Analytical methods
Blood samples were taken after a minimum 10-h fast. CD4 cell counts were determined by flow cytometry. The viral load in plasma was measured by quantitative PCR and is expressed as number of HIV-RNA copies per milliliter (17). Plasma glucose, TGs, and total cholesterol were measured with enzymatic reagents on an automated analyzer (Dade-Behring, Les Ulis, France), as described in Ref. 18 . High-density lipoprotein cholesterol (HDL-cholesterol) was assayed following a prior precipitation of apo B/apo E-containing lipoproteins with phosphotungstic acid and magnesium (Roche-Diagnostics, Meylan, France). Low-density lipoprotein cholesterol (LDL-cholesterol) was calculated according to Planella et al. (19), because this equation integrates, for a large part, the apo B value and is more appropriate in case of significant hypertriglyceridemia. Apo A-I, apo B100, and lipoprotein (a) [Lp(a)] were determined by first order immunoprecipitation in an automated analyzer (Cobas-Mira; Roche-Diagnostics). Total apo C-III and the apo C-III associated with apo B-containing particles (Lipo B:C-III) were measured by electroimmunoassay combined with an immunoprecipitation of apo B-containing lipoproteins (Sebia, Issy-les-Moulineaux, France) (5). Similar techniques were used for the determination of total apo E and of the apo E associated with apo B-containing particles (Lipo B:E), and for Lipo A-I. Insulin, proinsulin, and C-peptide were measured with specific RIAs (bi-Insuline, RIA-Diagnostics Pasteur, Paris, France; Proinsulin and C-peptide-CTK, DiaSorin, Inc., Anthony, France). The intra- and interassay coefficients of variation for cholesterol, HDL-cholesterol, glucose, apo A-I, and apo B were less than 4%. It was less than 10% in the case of insulin, proinsulin, C-peptide, and lipoparticles measurements.
Statistical analysis
The statistical significance of the differences between groups
were analyzed using the 
2 test for categorical
variables and one-way ANOVA for continuous ones. An analysis of
covariance was performed to compare mean values of plasma
concentrations of lipids, apos, and lipoparticles after adjustment for
confounding factors such as age, BMI, and, when stated, duration of
RTNI treatment. Post hoc tests were done when the overall
test was significant (P < 0.05). The Scheffé
test was performed after ANOVA and analysis of covariance. Comparisons
between before/after fenofibrate treatment or before/after PI
introduction were performed using a paired comparisons Student’s
t test. For variables with a skewed distribution, analysis
was performed after logarithmic transformation. When a close-to-normal
distribution was not reached, a Kruskal-Wallis test was used. Stepwise
logistic regressions were performed to identify among PI-treated
patients, the clinical and biological parameters associated with the
occurrence of LD. Each parameter was considered at two levels
(i.e. below and above its median value), as determined in
our whole population of patients. Time of antiretroviral treatment was
forced in the model because it has been identified as a strong
determinant of LD onset in several studies (20). Analysis
of maximum likelihood estimates was performed, and odds ratios were
determined. Statistical analysis was conducted using the SAS
statistical software release 6.12 (SAS Institute, Inc.,
Cary, NC).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Subsequent patients attending the HIV follow-up consultation were recruited during a 3-month period. Because the metabolic parameters investigated here are highly dependent on gender, only male patients will be further considered. Among them, 14 were treated only with two RTNIs, whereas another 49 also received one (or two) PIs. Regarding RTNI, 70% of patients received Stavudine plus Lamivudine, whereas Didanosine, Zalcitabine, and Zidovudine were given more occasionally. With respect to PIs, most patients were given a single molecule (indinavir, Ritonavir, nelfinavir, saquinavir) and 10 (of 49) received Ritonavir + saquinavir in association.
Distribution of the patients among the clinical AIDS stages was: in the PI group, 41% were at stage A, 25% at stage B, and 34% at stage C; for patients maintained on dual therapy only, 90% were at stage A. To get reference values for all investigated parameters, a sample of 63 male subjects, paired for age with PI-treated patients, was selected from a population study carried out in the same regional area.
Patients treated with RTNI only were younger than in the two other
populations (35.8 vs. 42.0 yr; see Table 1
). Weight and BMI were significantly
lower in PI-treated patients than in the other two populations. The
total duration of RTNI administration was longest in the patients
receiving PI also, because introduction of the latter had generally
followed a period of treatment with RTNI only.
|
Regarding glucose homeostasis, levels of glycated hemoglobin were found in the normal range (4–8%) in all patients. Fasting glucose was lower in all treated HIV patients (4.45 and 4.60 mmol/L) compared with controls (5.60 mmol/L, P < 0.001), whereas insulin was significantly higher (15.0 mUI/L vs. 9.8 mUI/L, P < 0.001). The homeostasis model assessment index for insulin resistance (HOMA-IR) was not significantly different between groups (data not shown). However, the HOMA index indicative of ß-cell function was three times higher in patients than in controls (P < 0.001). Finally, plasma C-peptide was 50% higher in PI-treated than in PI-naive patients (P < 0.05).
Lipids, apos, and lipoparticles
Table 2 displays the crude values
for lipoprotein markers, as measured in our three populations. Because
the three groups were somewhat different in terms of age and BMI, the
statistical significance of differences was assessed after adjustment
on age and BMI. Actually, those adjusted values (data not shown) were
very close to nonadjusted ones. Regarding apos, both the total
measurements and the amounts specifically associated with a definite
particle were determined: the Lipo B:C-III and Lipo B:E particles
contain apos C-III and E, respectively, associated with apo B, whereas
Lipo A-I refers to particles containing apo A-I but no apo A-II.
|
. TGs, apo C-III, and Lipo
B:C-III were 60–100% higher in PI-treated patients than in PI-naive
patients and in control subjects. By contrast, total and
LDL-cholesterol, as well as apo B, were significantly lower in RTNI
only-treated patients than in the two other groups. The levels of those
markers were not different between PI-receiving patients and controls.
HDL-cholesterol and apo A-I, parameters of the reverse cholesterol
transport, followed a decreasing gradient from controls to PI-naive and
then to PI-treated patients, although the intergroup differences did
not always reach significance. On the opposite, apo E and Lipo B:E
particles displayed a positive gradient: apo E was 80% and 200%
increased in RTNI only- and in PI-treated patients, respectively,
compared with controls.
Because the time of exposure to RTNI was clearly different between the
two groups of treated patients (42.6 vs. 21.9 months; see
Table 1), the comparisons between groups 1 and 2 were repeated on
values adjusted for age, BMI, and RTNI treatment time (data not shown).
Again, the values differed little from those presented in Table 2. The
most important changes concerned HDL-cholesterol and apo A-I, which
were no longer significantly different between both groups.
Some biological parameters were routinely measured just before setting
up PIs in the regimen. The comparisons for glucose, cholesterol, and
TGs before and after PI treatment are shown in Table 3. The pre-PI values were rather close to
those measured in the RTNI only group (compare with Tables 1 and 2).
After an average 17-month PI administration, very significant rises in
TGs and cholesterol were recorded while glucose levels were
unchanged.
|
). Regarding apo B, TGs, apo E, and apo
C-III, the proportions of patients above median values were three times
higher among PI-treated than among PI-naive patients. Similar
differences were seen for increased levels of Lipo B:E and Lipo B:C-III
(data not shown). By contrast, the percentages of subjects with low
HDL-cholesterol or elevated Lp(a) (data not shown) were not different
between the two groups.
|
Associated LD
Among the PI-treated patients, 23 of 49 (47%) had developed a
LD+, as clinically assessed. LD was confirmed by DEXA. No case of LD
was recorded in the RTNI only group. The prevalence of LD was compared
according to the median values for the different investigated
parameters, as defined above (Table 5).
Occurrence of LD was 2.5-fold more frequent in patients showing
elevated levels of apo B (P = 0.01) or apo C-III
(P = 0.002), and a trend to a positive association was
also observed for total cholesterol. By contrast, the prevalence of LD
was not found associated with levels of any other of the following
parameters: HDL-cholesterol, apoprotein A-I, apoprotein E, Lp(a),
glucose, and insulin. As well, no association was found between
occurrence of LD and age, BMI, the viral load, or the duration of PI
administration. Finally, there was a (nonsignificant) tendency toward
higher prevalence of LD for longer times of RTNI treatment.
|
Effectiveness of fenofibrate in reversing the lipoprotein alterations
As the abnormalities recorded on long-term PI therapy mostly
concerned the metabolism of TGs, 13 patients with severe
hypertriglyceridemia were consecutively given a fibric acid derivative
(fenofibrate, 200 mg/day), and lipoproteins were reevaluated 2 months
later (Table 6). Patients were not
advised to change their dietary habits during this period. Fenofibrate
had marked effects on plasma TGs (-40%), apo E (-40%), and Lipo B:E
(-50%; data not shown) and modestly on cholesterol, apo B, apo CIII
(-20%), and Lipo B:C-III (-20%; data not shown). Levels of Lp(a)
and of insulin were unmodified after fibrate administration.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although hypertriglyceridemia and LD have been repeatedly observed in patients under intensive antiretroviral therapy, the precise role of PI is still actively debated. Different reports suggest that the duration of exposure to treatments, and particularly RTNI, is a strong predictor of the metabolic abnormalities (20). In the present study, several arguments are in favor of the implication of PI: 1) differences between PI-treated and PI-naive patients persisted after an adjustment on treatment time; 2) cholesterol and TGs were significantly increased after the introduction of PI, in agreement with other studies (21); and 3) in our series, 50% of PI-treated patients presented with a LD vs. none in PI-naive subjects. Recent reports also confirm the role of PI: in the follow-up of almost 3200 patients under highly active antiretroviral therapy—the French Aquitaine Cohort—multivariate analysis demonstrated that PIs are a major contributor for increased TG levels (22).
Apo C-III seems to be a key apopoprotein regulating both the synthesis and catabolism of TG-rich lipoproteins and of their remnants formed through action of lipoprotein lipase. In transgenic mice, an overexpression of human apo C-III of only 40%, above endogenous levels, was associated with a 3-fold increase in plasma TG (9). Apo C-III inhibits the apo C-II-activated lipoprotein lipase (23). Besides, apo C-III impairs the interactions of the neighbor apo E with the LDL receptor or with the LDL receptor-related protein (LRP), a receptor involved in the liver uptake of remnant particles (24). It has also been demonstrated that the presence of apo C-III on a lipoprotein particle decreases the affinity of the neighbor apo B toward receptors (25). As a matter of fact, apo C-III impairs the in vivo hepatic uptake of remnant particles (26). With respect to the lipid-associated vascular risk, Lipo B:C-III would represent cholesterol-enriched remnant particles, potentially atherogenic and associated with the formation of small and dense LDL (27). In several interventional trials, documented by coronary angiography, the Lipo B:C-III particles were related to the progression of the atherosclerotic lesions (28, 29).
Apo C-III synthesis is transcriptionally regulated by several
mediators (30). It is down-regulated by insulin
(31) and by the 
peroxisome proliferator-activated
receptor (PPAR), a nuclear receptor with multiple target genes
involved in lipid metabolism and fatty acid oxidation
(32). The insulin-responsive element and the PPAR
recognition element are both located in the proximal (-500 bp)
promoter of the apo C-III gene (30). Functionally, PPAR
heterodimerize with the nuclear retinoic acid receptors, called RXRs
(32). Carr et al. (33) have
proposed that PIs would impair the activation of the PPAR/RXR
heterodimer. This hypothesis is based on the homology between the HIV-1
protease and CRABP, a binding protein involved in the cytoplasmic
transport of retinoic acid to the mitochondria, where it is activated
to cis-retinoic acid, an obligatory step for RXR activation
(32). Thus, antiproteases, by interacting with CRABP,
would impair the activation of retinoic acid and the downstream
function of RXR-PPAR. In support of this proposed mechanism is a
recent report describing the toxicity of isotretinoin, when
coadministered with PI (34). Concordant also are our
observations that TGs, apo C-III, apo E, and the corresponding
lipoparticles were effectively reduced after fenofibrate, which acts
through direct activation of PPAR. These reductions likely reflect
the effectiveness of fenofibrate, although we did not strictly control
for absence of dietary change during the period of the trial.
The most dramatic changes in the lipoprotein profile on PI administration concern apo E and the Lipo B:E lipoparticles. Carr et al. (33) have also reported on the relative homology between the HIV-1 protease and the LRP, a receptor that binds apo E (10). They suggested that anti-proteases might impair the function of LRP, leading to the accumulation of remnant particles. However, as discussed above, the presence of excess apo C-III on a particle, may impair the normal interaction of apo E with receptors and, thus, cause accumulation of those apo E-containing particles. Apo E-rich particles may be highly atherogenic because the presence of multiple copies of apo E renders them avidily taken up by macrophages (35).
Impaired glucose tolerance and hyperinsulinism have been commonly described as side effects of PI therapy (1, 4). In our present observations, fasting insulin was 50% higher in both groups of patients compared with controls, whereas glucose levels were significantly lower. These data suggest a ß-cell dysfunction on prolonged antiretroviral treatment, as demonstrated previously (4). The levels of plasma insulin that we found are rather comparable with those recently reported in HIV patients treated either with RTNI alone or with PI (36). Nevertheless, C-peptide, an indicator of insulin secretion, was found higher in PI-treated than in PI-naive patients. Among the former, elevated C-peptide levels (>2.5 ng/mL) were associated with the highest increases in TGs, cholesterol, and apoproteins B, C-III, and E (data not shown). Thus, several elements of the metabolic syndrome are present in PI-treated patients (hyper-TG, low HDL, hyperinsulinism). It was recently demonstrated that PIs lower the expression of insulin receptors on adipocytes (37). This may favor the release of free fatty acids from adipocyte stores and the further synthesis of TGs by the liver.
Several factors predisposing to the occurrence of LD on PI therapy have
been proposed (11, 12, 20, 33). In agreement with earlier
reports, neither the viral load nor the CD4 lymphocyte counts were
different between LD+ and LD- patients (4, 11, 12).
Again, recent reports have questioned the precise role of PIs in the
pathogenesis of LD and stressed the implication of RTNIs. For instance,
lipodystrophic syndromes have been described in patients receiving only
long-term nucleoside analogs, particularly Stavudine and/or
Lamivudine (38, 39). A precursor phenotype of LD
was recently documented by DEXA, in patients receiving only
RTNIs (36). An interesting hypothesis is that RTNIs
might be deleterious toward mitochondrial DNA, through interference
with the 
polymerase, and cause mitochondrial dysfunction
(20). In vitro, an effect Zidovudine was
demonstrated on hepatoma cells, inducing an increase in lactic acid and
an inhibition of mitochondrial enzymes (40). Analogies
between the antiretroviral-triggered LD and other lipomatoses,
associated with mitochondrion alterations, lends support to this
hypothesis. Finally, data from observational cohorts show a strong
association between the time of exposure to RTNIs and the onset of
lipoatrophy (20). However, the development of LD seems to
be considerably accelerated by the addition of antiproteases
(41). A role for PPARs in the pathogenesis of LD can also
be foreseen (42). Indeed, in adipose tissue, the PPAR
isoform is predominantly expressed, playing a role in adipocyte
differentiation, proliferation, and apoptosis (43). In
line with the hypothesis discussed above for (liver) PPAR, a defect
in PPAR activation on PIs could affect fat distribution. Moreover, a
massive apoptosis of sc adipocytes could lead to enhanced flux of free
fatty acids to the liver and secondary VLDL synthesis.
Looking for possible determinants of LD occurrence, and taking into consideration the duration of antiretroviral therapy, we found that a more than 30 mg/L apo C-III level was a strong predictor. This result suggests a relationship between both syndromes: dyslipemia and LD.
In conclusion, measurements of apo C-III, apo E, and the corresponding lipoparticles seem to be important in the follow-up of PI-treated patients. Prospective studies, including these markers, are needed to further explore the possible association of these markers with the development of a lipodystrophic syndrome.
Received December 28, 1999.
Revised September 14, 2000.
Accepted October 4, 2000.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2-macroglobulin receptor. J Biol Chem. 266:13936–13940.This article has been cited by other articles:
 |
|
 |
|
  S. Grinspoon and A. Carr Cardiovascular Risk and Body-Fat Abnormalities in HIV-Infected Adults N. Engl. J. Med., January 6, 2005; 352(1): 48 - 62. [Full Text] [PDF]
|
|
|
|
 |
|
 C. Fisac, N. Virgili, E. Ferrer, M. J. Barbera, E. Fumero, C. Vilarasau, and D. Podzamczer A Comparison of the Effects of Nevirapine and Nelfinavir on Metabolism and Body Habitus in Antiretroviral-Naive Human Immunodeficiency Virus-Infected Patients: A Randomized Controlled Study J. Clin. Endocrinol. Metab., November 1, 2003; 88(11): 5186 - 5192. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Nolan, G.F. Watts, S.E. Herrmann, M.A. French, M. John, and S. Mallal Endothelial function in HIV-infected patients receiving protease inhibitor therapy: does immune competence affect cardiovascular risk? QJM, November 1, 2003; 96(11): 825 - 832. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Barbaro HIV infection, highly active antiretroviral therapy and the cardiovascular system Cardiovasc Res, October 15, 2003; 60(1): 87 - 95. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Leung, R. A. Hegele, and G. F. Lewis Rapid Development of Massive Tendon Xanthomas following Highly Active Antiretroviral Therapy Ann Intern Med, October 1, 2002; 137(7): 624 - 624. [Full Text] [PDF]
|
|
|
|
 |
|
 L. A. Panther How HIV infection and its treatment affects the cardiovascular system: what is known, what is needed Am J Physiol Heart Circ Physiol, July 1, 2002; 283(1): H1 - H4. [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
