Share this article on:

Ten-year diabetes incidence in 1046 HIV-infected patients started on a combination antiretroviral treatment

Capeau, Jacquelinea; Bouteloup, Vincentb; Katlama, Christinec; Bastard, Jean-Philippea; Guiyedi, Vincentd; Salmon-Ceron, Dominiquee; Protopopescu, Cameliaf; Leport, Catherined; Raffi, Francoisg,*; Chêne, Genevièveb,*the ANRS CO8 APROCO-COPILOTE Cohort Study Group

doi: 10.1097/QAD.0b013e32834e8776
Clinical Science

Objective: To evaluate the incidence and determinants of diabetes in a cohort of HIV-infected adults initiated with combination antiretroviral treatment (cART) in 1997–1999 and followed up to 2009.

Design: Prospective study of 1046 patients at 47 French clinical sites.

Methods: Potential determinants of diabetes occurrence, defined by confirmed increased glycemia and/or initiation of antidiabetic treatment, were assessed by a proportional hazards model, including time-updated metabolic parameters and ART exposure.

Results: Among the cohort, representing 7846 person-years of follow-up (PYFU), 54% received indinavir, 75% stavudine and 52% didanosine. Overall, 111 patients developed diabetes, with an incidence of 14.1/1000 PYFU (14.6 in men, 12.6 in women). Incidence peaked in 1999–2000 (23.2/1000 PYFU) and decreased thereafter. The incidence of diabetes was associated [adjusted hazard ratio (aHR), all P < 0.02] with older age (hazard ratio = 2.13 when 40–49 years, hazard ratio = 3.63 when ≥50 years), overweight (hazard ratio = 1.91 for a BMI 25–29 kg/m2, hazard ratio = 2.85 >30 kg/m2), waist-to-hip ratio (hazard ratio = 3.87 for ≥0.97 male/0.92 female), time-updated lipoatrophy (hazard ratio = 2.14) and short-term exposure to indinavir (0–1year: hazard ratio = 2.53), stavudine (0–1year: hazard ratio = 2.56, 1–2years: hazard ratio = 2.65) or didanosine (2–3years: hazard ratio = 3.16). Occurrence of diabetes was not associated with HIV-related markers, hepatitis C, hypertension or family history of diabetes. Insulin resistance was predictive for incident diabetes.

Conclusions: In this nationwide cohort, followed for 10 years after cART initiation, diabetes incidence peaked in 1990–2000, was markedly higher than that reported for European uninfected or other HIV-infected populations (4–6/1000 PYFU) and linked with age and adiposity. Adiposity and glycemic markers should be monitored in aging HIV-infected patients.

aUPMC Univ Paris 06, UMR_S938, Inserm, UMR_S938, CDR Saint-Antoine, F-750012, AP-HP, Hôpital Tenon, Paris

bInserm, U897; Université Bordeaux Segalen, ISPED, Bordeaux

cAP-HP, Hôpital Pitié-Salpêtrière, Service des Maladies Infectieuses, UPMC Univ Paris 06, UMR_S943, Inserm

dUMR_S738, Inserm, Université Paris Diderot Paris7, CHU X-Bichat, AP-HP

eAP-HP Hôpital Cochin, Service de Médecine Interne, Paris

fInserm, U912 (SE4S), Université Aix Marseille, IRD, UMR_S912, ORS PACA, Observatoire Régional de la Santé Provence Alpes Côte d’Azur, Marseille

gHotel-Dieu, Nantes, France.

*Francois Raffi and Geneviève Chène contributed equally to the work.

Correspondence to Jacqueline Capeau, Faculty of Medicine Pierre and Marie Curie, Inserm UMR_S938, 27 rue Chaligny, 75571 Paris Cedex 12, France. Tel: +33 1 40011332; fax: +33 1 40011432; e-mail:

Received 2 July, 2011

Revised 23 September, 2011

Accepted 24 October, 2011

Back to Top | Article Outline


Human immunodeficiency virus (HIV)-infected patients treated with combination antiretroviral therapy (cART), including nucleoside analog reverse transcriptase inhibitors (NRTIs) with protease inhibitors or non-NRTI, have markedly increased their life expectancy. The incidence of metabolic disorders – including type 2 diabetes – among these patients is, however increasing. The causes of these metabolic disorders may be linked to the use of antiretroviral drugs [1,2] and to cART-related lipodystrophy [2–4] in addition to conventional determinants like ageing, male sex, adiposity markers, dyslipidemia, hypertension, infection with hepatitis C virus (HCV) and genetic characteristics [2,5].

Soon after the introduction of first-generation protease inhibitors in 1996, various cohorts of HIV-infected patients, including ours, were found to show a high prevalence of diabetes [6–8]. More recently, these data were updated by the Swiss HIV Cohort Study (SHCS) and the Data-Collection-on-Adverse-Events of Anti-HIV-Drugs (D:A:D) study, which reported an incidence of 4.4 and 5.7 per 1000 person years of follow-up (PYFU) [9,10], respectively. These estimations are in the range of those reported for the European general population of the same age and sex (4–6 per 1000 PYFU) [11] and in the French HIV-uninfected population (4–5 per 1000 PYFU) [12].

Very few longitudinal studies can simultaneously explore all the potential determinants that may be involved in the incidence of diabetes among HIV-infected patients. Our ANRS-CO8 APROCO-COPILOTE cohort provided a unique opportunity to evaluate these dimensions in patients who started cART, including their first protease inhibitor regimen, in 1997–1999 and were then prospectively followed up to 2009.

Back to Top | Article Outline


Patients: the APROCO-COPILOTE cohort

Between 1997 and 1999, 1281 HIV-infected adults at 47 French clinics were enrolled in a prospective study. These patients had not previously received protease inhibitor and were starting cART, based on protease inhibitor and NRTIs. At that time, non-NRTI treatments were not available. Patients underwent physical and laboratory examinations at enrolment, then after 1 and 4 months of cART, and every 4 months thereafter. Their height was recorded at the start of the study and weight was recorded at each visit to calculate BMI. At the 12-month or 20-month (M12/M20) examination, and annually thereafter, physicians assessed lipodystrophy, waist and hip circumferences were measured with a tape, blood pressure (BP), fasting glucose and fasting lipid parameters were measured, and a 2-h oral glucose tolerance test (OGTT) was also performed. Only patients with at least two measurements of glycemia were considered in the current analysis. Patients who were diagnosed as diabetic at enrollment were excluded from the analysis. The Ethics Committee of Cochin Hospital (Paris) approved this study and informed consent was obtained from all participants.

Back to Top | Article Outline

Definition of diabetes

Diabetes was diagnosed according to the criteria of the Expert Committee on the Diagnosis and Classification of Diabetes mellitus [13], that is confirmed fasting glycemia at least 7.0 mmol/l or 2-h OGGT glycemia at least 11.1 mmol/l, and/or if initiation with antidiabetic drugs (including metformin but not glitazones) during follow-up. An expert physician (J.C.) checked all cases. This definition of diabetes was consistent with that used by the SHCS [10] except that nonfasting instead of post-OGTT glycemia at least 11.1 mmol/l was considered in the SHCS.

In a sensitivity analysis, we also considered definitions used in the D:A:D study [9] (two consecutive fasting glycemias over the threshold, or treatment, or report by the physician) or in the INITIO trial [14] (one fasting glycemia >7 mmol/l or nonfasting glycemia >11.1 mmol/l).

Finally, we classified patients according to their glycemic status during follow-up into the following categories: normoglycemia (fasting and postcharge glycemia always <5.6 and <7.8 mmol/l) (n = 374); impaired fasting glycemia or impaired glucose tolerance [at least one fasting or post-OGTT glycemia in the range (5.6–7) or (7.8–11.1) mmol/l, respectively] (n = 450); one fasting or postcharge glycemia in the diabetic range (n = 111), and diabetic patients as defined above (n = 111).

Back to Top | Article Outline

Analysis of inflammation and metabolic markers

In a subgroup of 366 patients, we measured the levels of insulin (ARCHITECT system, Abbott, Park, Illinois, USA), high-sensitivity C-reactive protein (hs-CRP) (by immunonephelometry; IMMAGE, Beckman-Coulter, Villepinte, France), soluble TNF-α receptor 1 (sTNFR1) and adiponectin (by ELISA; R&D Systems, Lille, France). We also evaluated insulin resistance by the homeostasis model assessment (HOMA) method: fasting serum insulin (mU/l) × glycemia (mmol/l)/22.5.

Back to Top | Article Outline

Statistical analysis

Incidence rates of diabetes were estimated as the number of cases divided by the number of PYFU, and their 95% confidence intervals (95% CI) were computed using the exact Poisson method. Follow-up duration was calculated as the difference in days from initiation of cART to the date of diabetes diagnosis, death or last visit, whichever occurred first. Follow-up of diabetic patients was censored after the date of diabetes diagnosis.

We used a Cox proportional hazards model to identify characteristics associated with diabetes including sex, age, family history of diabetes, HCV status, BMI, cART history, history of an AIDS-defining morbidity, CD4 cell count, CD4/CD8 ratio, HIV-1 RNA level (all evaluated at the start of the study), waist circumference, and waist-to-hip ratio (WHR) (at the M12/M20 examination). Thresholds of waist circumference were adapted for sex according to the definition of metabolic syndrome for Caucasians (≥94 or ≥102 cm in men or ≥80 or ≥88 cm in women) [15].

We also considered time-updated lipohypertrophy, lipoatrophy, tobacco use and blood pressure over 130/85 mmHg (categorized as never exposed vs. exposed at least once or twice for BP). Cumulative time-updated cART exposure (including treatment with NRTIs prior to inclusion) was defined in six categories: never exposed, currently exposed for less than 1 year, for 1–2 years, for 2–3 years, for at least 3 years, or exposed but currently not being treated.

The final results of the multivariate analysis were derived by three consecutive steps: first, selection of variables in a univariate analysis (P ≤ 0.25); second, multivariate model without considering cART exposure (P ≤ 0.05); and third, from this multivariate model with a forward selection of cART exposure (P ≤ 0.05). To deal with high colinearity between BMI, waist circumference and WHR in step 2, we kept only variables that minimized the Akaïke criterion of the multivariate model.

Evolution of clinical and biological parameters during follow-up was assessed with a univariate linear or nonlinear random-effect model, as appropriate (Proc MIXED or NLMIXED in SAS). Fisher's exact test, Wilcoxon rank-sum or Kruskal–Wallis tests were used to compare patient characteristics, as appropriate. Evolutions and comparisons were computed from available data.

Back to Top | Article Outline


Incidence of diabetes

From a total of 1281 patients enrolled in the cohort, we excluded one patient who was diabetic prior to protease inhibitor initiation and 234 patients who had no available glycemia measurement on two occasions. When compared to those included, the excluded patients were more frequently women (28.5 vs. 21.5%, P = 0.03), were younger (median age 35 vs. 37 years, P = 0.003), had lower BMI (median 21.6 vs. 22.1 kg/m2, P = 0.03) and baseline CD4 cell counts (median 223 vs. 280 cells/μl, P = 0.03). They did not differ in their history of AIDS-defining event, antiretroviral history, and exposure to indinavir or stavudine. Of the remaining 1046 patients, 44% started cART while ART-naive and 22% were coinfected with HCV (Table 1). At the first annual visit (% among available data), 372 (62%) were diagnosed with clinical lipodystrophy (men 62%; women 61%) and the waist circumference was at least 94 cm in 14% of men and at least 80 cm in 49% of women (P < 10−4).

Table 1

Table 1

Among the 1046 patients, 111 cases of new-onset diabetes were identified in the period up to 2009, during 7846 PYFU [median follow-up 9.6 years, interquartile range (IQR) 5.1–10.7] yielding an overall incidence of 14.1 per 1000 PYFU (95% CI 11.6–17.0). This incidence was similar in both sexes: 14.6 (11.7–17.9) in men and 12.6 (7.8–19.3) in women (Table 1). Since metformin can be given to some nondiabetic patients to decrease insulin resistance, we performed a sensitivity analysis excluding three patients treated with metformin and with glycemia under the diabetic threshold. Our incidence estimations were not affected when these three patients were withdrawn (data not shown). In another sensitivity analysis we used the definitions of the D:A:D and INITIO studies: the diabetes definition used by the D:A:D study yielded a total of 94 cases and an incidence of 11.7 per 1000 PYFU (95% CI 9.5–14.4) (12.5 in men, 8.8 in women). With the INITIO trial definition, 210 cases were identified and the incidence was 27.1 per 1000 PYFU (95% CI 23.6–31.0) (29.3 in men, 19.3 in women). Therefore, the incidence in our cohort was higher than those reported in other European or international cohorts.

Back to Top | Article Outline

Evolution during follow-up

The incidence of diabetes among our cohort peaked in 1999–2000 (Fig. 1a), and then gradually decreased (P < 10−4). In patients who were ART-naive at entry, the peak of incidence was in 2001–2002 (Fig. 1b). Otherwise, over the 10-year follow-up, median BMI increased from 22.5 to 22.7 kg/m2 (P = 0.01), waist circumference from 83 to 87 cm (P < 10−4), and WHR from 0.92 to 0.95 (P < 10−4). Median HDL-cholesterol increased (from 1.13 to 1.28 mmol/l, P < 10−4) and LDL-cholesterol decreased (from 3.8 to 3.2 mmol/l, P < 10−4), whereas triglyceride levels remained stable (1.6–1.5 mmol/l, P = 0.84). Fasting glycemia increased from 5.0 to 5.2 mmol/l (P < 10−4), whereas 2-h OGTT glycemia remained stable (5.9–5.8 mmol/l, P = 0.11). The prevalence of lipodystrophy and lipoatrophy increased with cumulative exposure to stavudine, reaching 72 and 64%, respectively, after 3 years. In 26% of cases, stavudine treatment was withdrawn due to the presence of lipodystrophy.

Fig. 1

Fig. 1

Back to Top | Article Outline

Risk factors for diabetes

Older age and all adiposity parameters (BMI, waist circumference and WHR) were strongly associated with diabetes. Using time-dependent measures for BMI did not modify the findings.

Moreover, lipohypertrophy, lipoatrophy and elevated BP were associated with diabetes (Table 2). The associations remained in the multivariate analysis, except for elevated BP and lipohypertrophy (Table 2). There was no association between increased diabetes incidence and a family history of diabetes, smoking, HCV coinfection, HIV-related characteristics, or ethnic origin (78.9% Caucasian, 9.0% African, 12.1% missing data). When considering patients’ exposition to NRTIs prior to initiating treatment with protease inhibitors, overall, 823 (79%) patients received stavudine, 945 (90%) received zidovudine, and 711 (68%) received didanosine (median total duration of exposure 39, 41 and 28 months, respectively). During the follow-up period, 562 (54%) patients were exposed to indinavir (median duration 22 months) and 591 (56%) to nelfinavir (21 months). In the multivariate analyses, an increased incidence of diabetes was associated with exposure to indinavir during the first year or stavudine during the first 2 years of treatment; a decreased incidence was associated with exposure to nelfinavir, consistent with its use exclusive of indinavir. Consistently, 53% (59/111) of diabetic patients had been exposed to both stavudine and indinavir before the occurrence of diabetes compared to 41% of nondiabetic patients (P = 0.04). A high incidence of diabetes was also associated with longer use of didanosine (between 2 and 3 years), suggesting a lower level of toxicity than stavudine and indinavir. When these antiretrovirals were stopped, the risk of diabetes dropped to the risk of unexposed individuals (Table 2). Overall, a large proportion of patients receiving indinavir (24/35, 69%) or stavudine 38/62, 61%) stopped these drugs after diagnosis of diabetes. Exposure to the protease inhibitors lopinavir, saquinavir or atazanavir was not associated with diabetes (data not shown).

Table 2

Table 2

The multivariate analysis was then performed in the subgroup of naive patients (n = 459, 41 cases of diabetes). Though this analysis lacked power, hazard ratios were in the same range as those reported in the whole cohort, including exposure to stavudine and indinavir. These results are in favor of a true additional contribution of these specific antiretrovirals.

Back to Top | Article Outline

Systemic inflammation and metabolic markers and diabetes incidence

To analyze whether systemic inflammation and/or insulin resistance were associated with development of diabetes in patients initiating cART, we evaluated the levels of hs-CRP, sTNFR1 adiponectin and insulin and calculated the HOMA value at the M12/M20 examination in a subgroup of 356 patients included in a metabolic substudy. Compared to the remaining 658 patients, these patients were older (median age at the start of the study 37.4 years, IQR 33.0–43.0, vs. 36.1 years, IQR 31.5–41.7, P = 0.008), and less frequently ART-naïve (39.3 vs. 47.3%, P = 0.017). Importantly their BMIs, waist circumferences, WHRs and diabetes incidence were similar to other patients. Among the 356 patients, 38 developed diabetes after the M12/M20 examination. At that examination, when compared to those never diagnosed with diabetes (Table 3A), they had increased hs-CRP, insulin and HOMA values, whereas their adiponectin levels were lower. After adjusting for age at inclusion, glycemia, BMI, WHR, lipoatrophy and exposure to stavudine, didanosine and indinavir, and comparing the values separated into quartiles, there was a trend for higher levels of sTNFR1, but not hs-CRP and adiponectin, to be associated with incident diabetes. A higher level of HOMA values was strongly associated with incident diabetes (Table 3B).

Table 3

Table 3

Back to Top | Article Outline

Patients’ characteristics and exposure to antiretroviral according to glycemic status during follow-up

Age at inclusion, BMI, waist circumference, elevated BP, being ART-experienced, presence of lipoatrophy and of lipohypertrophy gradually increased when comparing the four categories of glycemic status from normoglycemia to diabetes, during follow-up (Table 4). Patients who had presented at least one diabetic hyperglycemia were more likely to have been exposed to indinavir than patients without diabetic hyperglycemia.

Table 4

Table 4

Back to Top | Article Outline


In this cohort of 1046 HIV-infected patients initiated with first-generation protease inhibitors and then followed for 10 years, the incidence of new-onset diabetes was as high as 14.1 per 1000 PYFU, markedly higher than that reported for European uninfected or other HIV-infected populations (4–6/1000 PYFU), probably due to exposure to some antiretroviral drugs and to lipodystrophy.

Indeed, when we compared our data with those from other large series of HIV-infected patients after adjustment for BMI distribution, we find more than double the incidence of diabetes than in any of the large series, even when applying their specific definitions [14]. (Note that the high incidences of diabetes in North American studies [8,16] can probably be explained by higher BMIs than in European patients.)

The discrepancy between our findings and those of the other three studies might be explained by several factors: the earlier period of inclusion in our APROCO-COPILOTE cohort (1997–1999 compared to 1999–2002 in the INITIO trial, 1999–2006 in the D:A:D study and 2000–2006 in the SHCS); exposure to high levels of first-generation antiretrovirals such as stavudine and indinavir, which were used commonly at the time we initiated our study; the systematic longitudinal follow-up of our study; the fact that our cohort was closed while the D:A:D study and the SHCS continued to recruit throughout, or our use of OGTT, which may detect more diabetes cases than does analysis of fasting glycemia.

The importance of exposure to first-generation antiretroviral drugs in the development of new-onset diabetes is consistent with the drop in incidence of new-onset diabetes in recent years in our cohort, which correlates with the use of new antiretrovirals; the incidence rates over the course of our study have become similar to those observed in the D:A:D study and the SHCS.

In this study, we found that diabetes incidence was high both in men (14.6) and women (12.6), in contrast to the other European studies – the SHCS and the D:A:D study [9,10] – and with studies of the general population [11] reporting higher incidences in men than in women.

As observed in the general population [11,17] (and also in HIV-infected adults [7,9,10,18]), we report a major role for age and adiposity (as evaluated by BMI, waist circumference, and WHR) in the development of new-onset diabetes, even stronger than the effect of antiretroviral drugs. We also found that both lipohypertrophy (as recorded by increased WHR) and lipoatrophy were risk factors for diabetes. The involvement of lipohypertrophy is consistent with several previous studies [9,10,14,19]; the involvement of lipoatrophy, independently of abdominal obesity, however, is alluded to only in the D:A:D study [9]. Elevated blood pressure and a family history of diabetes were not associated with incident diabetes in our cohort, in contrast to the general population, suggesting that the form of diabetes in our cohort is particular to these patients.

HIV-related parameters were not related to diabetes incidence, consistent with the D:A:D and INITIO studies [9,14] but differing from the SHCS [10]. Our data also confirm the absence of association between diabetes and HCV infection, in accordance with the SHCS but not with the Women's Interagency HIV Study [10,18,20] and the general population.

We evaluated whether inflammation and/or metabolic markers could be predictive of diabetes incidence. We found that systemic sTNFR1 levels at the M12/M20 examination were marginally predictive of subsequent diabetes in patients initiating cART, in accordance with the study by Brown et al. [21] that found that sTNFR1 levels 48 weeks after ART initiation were associated with increased risk of diabetes. This is consistent with data on the general population indicating that chronic subclinical inflammation might be involved in the pathogenesis of type 2 diabetes due to systemic causative factors such as central obesity and insulin resistance [22]. In addition, we observed a strong relationship with insulin resistance, in accordance with data in the general population [23].

The relation between cART and the incidence of diabetes is a major issue for the care of patients. We found a strong association of new-onset diabetes with short-term exposure to stavudine, indinavir or didanosine in accordance with the demonstrated ability of indinavir to induce insulin resistance, decrease insulin secretion [1,24] and alter glucose tolerance as previously shown in APROCO [4], and of stavudine and didanosine to induce insulin resistance in cohort studies [1]. Regarding protease inhibitors, some studies failed to find a relation between protease inhibitor treatment and the incidence of diabetes [8,14,18], whereas diabetes was independently associated with current exposure to indinavir in some [10,25] but not all [9,16] studies. In the D:A:D study, exposure to ritonavir was protective, whereas in the Multicenter AIDS Cohort Study it was a risk factor [9,16]. Regarding NRTIs, combination treatments with lamivudine–stavudine, didanosine–stavudine or didanosine–tenofovir have previously been found associated with a higher risk of diabetes in the SHCS [10]. This was the case also in the D:A:D study for treatments with zidovudine, stavudine or didanosine [9,16]. In other studies, zidovudine but not stavudine [25] or lamivudine but not stavudine [8] was also found associated with diabetes.

The association we found between short-term exposure to indinavir, stavudine and didanosine and diabetes was no longer seen in patients exposed for a longer term to these drugs. This might be explained if patients prone to develop diabetes are affected in the first years and that those who remain treated with these drugs longer-term are those who do not experience major side effects including elevated glycemia or diabetes.

The elevated risk of diabetes due to exposure to drugs in our cohort might explain why we did not identify risks associated with personal and familial/genetic [7,9,10,18] characteristics that are traditionally associated with late-onset diabetes. This elevated risk might be reversed by modification of the cART. Indeed, we find that patients who were previously exposed to stavudine, didanosine or indinavir reduced their risk of diabetes once they stopped these drugs. This might also be explained if patients who are at risk of developing diabetes do so within the first 2 years of exposure, or if there is no persistent risk after withdrawal.

The high incidence of diabetes in treatment-experienced patients in 1997–1998, which peaked in 1999–2000 and decreased thereafter, is likely related to exposure to indinavir or stavudine shortly preceding the onset of diabetes. This peak was delayed to 2000–2001 in ART-naive patients, who were exposed less and later to stavudine. Consistent with our observations, in the D:A:D study a similar higher incidence was reported in patients included in the study in 1999–2000 when compared to those included in 2005–2006. By contrast, in the SHCS the incidence did not differ according to the date of inclusion [9,10].

Our comparison of persistently normoglycemic patients with those with transient hyperglycemia or established diabetes pointed to a higher frequency of lipoatrophy and lipohypertrophy and a higher exposure to indinavir in association or not with stavudine in the two latter groups, further arguing for a role for lipodystrophy and indinavir in glycemic alterations.

We acknowledge some limitations to our analysis. The number of patients in the cohort limits its power. The number of cases is substantial, however, to estimate the incidence and associations in the long-term, 10-year follow-up. It is important to have a consensual definition of diabetes to be used in cohorts. We chose a similar definition to SHCS based on two abnormal glycemias. Finally, since our cohort was initiated in 1997–1999, most patients received drugs highly associated with lipodystrophy that might play an indirect role in diabetes incidence in addition to the direct role shown for indinavir and stavudine. These drugs are no longer used in most Western countries; however, in developing countries stavudine is still used widely, despite WHO recommendations, leading to lipodystrophy and diabetes.

In conclusion, we found that diabetes occurred more frequently in HIV-infected patients previously exposed to indinavir, stavudine and didanosine and persisted in most cases after drug withdrawal. Diabetes incidence peaked in 1999–2000 but markedly decreased thereafter. The diabetes developed by these patients differs to some extent from that of the general population since incidence is not linked to traditional risk factors (i.e. sex, family history, hypertension and HCV coinfection) except for age and markers of adiposity. In addition to increased adiposity and lipodystrophy, early and frequent exposure to indinavir and/or stavudine might account for the high incidence of permanent diabetes observed over time in the majority of cases. Most HIV-infected patients in developed countries are now treated with new-generation cART associated with a lower risk of diabetes; however, they are reaching older ages than before and often continue to gain weight, thus their case management should include measure of adiposity markers (waist circumference, BMI) and fasting glycemia at least yearly to identify at-risk patients. It is of utmost importance that diabetes is detected and adequately treated in order to favorably influence long-term cardiovascular risk.

Back to Top | Article Outline


We thank all patients, nurses and physicians at the clinical sites that participated in this study.

Part of the study was presented at the 18th Conference on Retroviruses and Opportunistic Infections meeting 2011 in Boston, USA.

Authors’ contribution: J.C. planned the data analyses and wrote the manuscript; V.B. analyzed the data and wrote the manuscript; C.K. contributed to patients’ recruitment and investigation and revised the manuscript; J.-P.B. contributed to patients evaluation and revised the manuscript; V.G. contributed to patients’ investigation and monitoring; D.S.-C. contributed to patients’ recruitment and investigation; C.P. contributed to patients’ investigation and monitoring; C.L. was responsible for cohort initiation and coordination, contributed to patients’ recruitment and investigation and revised the manuscript; F.R. was responsible for cohort initiation and coordination, contributed to patients’ recruitment and investigation and revised the manuscript; G.C. contributed to cohort initiation, planned the data analysis and wrote the manuscript.

ANRS CO8 APROCO-COPILOTE Cohort Study Group. Scientific Committee: Steering Committee: Principal investigators: C. Leport, F. Raffi. Methodology: G. Chêne, R. Salamon. Social sciences: J.-P. Moatti, J. Pierret, B. Spire. Virology: F. Brun-Vézinet, H. Fleury, B. Masquelier. Pharmacology: G. Peytavin, R. Garraffo. Other members: D. Costagliola, P. Dellamonica, C. Katlama, L. Meyer, D. Salmon, A. Sobel. Events Validation Committee: L. Cuzin, M. Dupon, X. Duval, V. Le Moing, B. Marchou, T. May, C. Rabaud, A. Waldner-Combernoux. Project coordination: P. Reboud. ANRS representatives: Sandrine Couffin-Cadiergues, Lucie Marchand. Data monitoring and statistical analysis: V. Bouteloup, A.D. Bouhnik, C. Brunet-François, V. Caron, M.P. Carrieri, M. Courcoul, F. Couturier, L. Hardel, P. Kurkdji, S. Martiren, M. Préau, C. Protopopescu, J. Surzyn, A.Taieb, V. Villes. Promotion: Agence Nationale de Recherches sur le Sida et les hépatites virales (ANRS, Action Coordonnée n°7). Other support: Collège des Universitaires de Maladies Infectieuses et Tropicales (CMIT ex APPIT), Sidaction Ensemble contre le Sida and associated pharmaceutical companies: Abbott, Boehringer-Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Gilead Sciences, Pfizer and Roche. Clinical Centres (investigators): Amiens (J.L. Schmit), Angers (JM. Chennebault), Belfort (J.P. Faller), Besançon (N. Mgy-Bertrand, B. Hoen, D. Drobachef), Bobigny (O. Bouchaud), Bordeaux (M. Dupon, Longy-Boursier, P. Morlat, JM. Ragnaud), Bourg-en-Bresse (P. Granier), Brest (M. Garré), Caen (R. Verdon), Compiègne (D. Merrien), Corbeil Essonnes (A. Devidas), Créteil (A. Sobel), Dijon (L. Piroth), Garches (C. Perronne), Lagny (E. Froguel), Libourne (J. Ceccaldi), Lyon (D. Peyramond), Meaux (C. Allard), Montpellier (J. Reynes), Nancy (T. May), Nantes (F. Raffi), Nice (J.G. Fuzibet, P. Dellamonica), Orléans (P. Arsac), Paris (E. Bouvet, F. Bricaire, J.F. Bergmann, J. Cabane, J. Monsonego, P.M. Girard, L. Guillevin, S. Herson, C. Leport, M.C. Meyohas, J.M. Molina, G. Pialoux, D. Salmon), Poitiers (P. Roblot), Reims (R. Jaussaud), Rennes (C. Michelet), Saint-Etienne (F. Lucht), Saint-Mandé (T. Debord), Strasbourg (D. Rey), Toulon (J.P. De Jaureguiberry), Toulouse (B.Marchou), Tours (L. Bernard).

This study was supported by the Agence Nationale de Recherches sur le Sida et les Hépatites Virales (ANRS), Inserm, Collège des Universitaires de Maladies Infectieuses et Tropicales (CMIT ex APPIT), Sidaction Ensemble contre le Sida and associated pharmaceutical companies: Abbott, Boehringer-Ingelheim, Bristol-Myers Squibb, GlaxoSmithKline, Gilead Sciences, Pfizer and Roche.

Back to Top | Article Outline

Conflicts of interest

None of the authors have any relevant conflicts of interest to disclose.

Back to Top | Article Outline


1. Florescu D, Kotler DP. Insulin resistance, glucose intolerance and diabetes mellitus in HIV-infected patients. Antivir Ther 2007; 12:149–162.
2. Samaras K. Prevalence and pathogenesis of diabetes mellitus in HIV-1 infection treated with combined antiretroviral therapy. J Acquir Immune Defic Syndr 2009; 50:499–505.
3. Hadigan C, Meigs JB, Corcoran C, Rietschel P, Piecuch S, Basgoz N, et al. Metabolic abnormalities and cardiovascular disease risk factors in adults with human immunodeficiency virus infection and lipodystrophy. Clin Infect Dis 2001; 32:130–139.
4. Saves M, Raffi F, Capeau J, Rozenbaum W, Ragnaud JM, Perronne C, et al. Factors related to lipodystrophy and metabolic alterations in patients with human immunodeficiency virus infection receiving highly active antiretroviral therapy. Clin Infect Dis 2002; 34:1396–1405.
5. Rotger M, Gsponer T, Martinez R, Taffe P, Elzi L, Vernazza P, et al. Impact of single nucleotide polymorphisms and of clinical risk factors on new-onset diabetes mellitus in HIV-infected individuals. Clin Infect Dis 2010; 51:1090–1098.
6. Carr A, Samaras K, Thorisdottir A, Kaufmann GR, Chisholm DJ, Cooper DA. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: a cohort study. Lancet 1999; 353:2093–2099.
7. Justman JE, Benning L, Danoff A, Minkoff H, Levine A, Greenblatt RM, et al. Protease inhibitor use and the incidence of diabetes mellitus in a large cohort of HIV-infected women. J Acquir Immune Defic Syndr 2003; 32:298–302.
8. Tien PC, Schneider MF, Cole SR, Levine AM, Cohen M, DeHovitz J, et al. Antiretroviral therapy exposure and incidence of diabetes mellitus in the Women's Interagency HIV Study. AIDS 2007; 21:1739–1745.
9. De Wit S, Sabin CA, Weber R, Worm SW, Reiss P, Cazanave C, et al. Incidence and risk factors for new-onset diabetes in HIV-infected patients: the Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) study. Diabetes Care 2008; 31:1224–1229.
10. Ledergerber B, Furrer H, Rickenbach M, Lehmann R, Elzi L, Hirschel B, et al. Factors associated with the incidence of type 2 diabetes mellitus in HIV-infected participants in the Swiss HIV Cohort Study. Clin Infect Dis 2007; 45:111–119.
11. Meisinger C, Thorand B, Schneider A, Stieber J, Doring A, Lowel H. Sex differences in risk factors for incident type 2 diabetes mellitus: the MONICA Augsburg cohort study. Arch Intern Med 2002; 162:82–89.
12. Fagot-Campagna A, Romon I, Fosse S, Roudier C. ‘Prevalence and incidence of diabetes and diabetes-related mortality in France’ Report InVS. In. Saint-Maurice, France: InVS (Institut de Veille Sanitaire); 2010. pp. 1–12.
13. Genuth S, Alberti KG, Bennett P, Buse J, Defronzo R, Kahn R, et al. Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 2003; 26:3160–3167.
14. Wand H, Calmy A, Carey DL, Samaras K, Carr A, Law MG, et al. Metabolic syndrome, cardiovascular disease and type 2 diabetes mellitus after initiation of antiretroviral therapy in HIV infection. AIDS 2007; 21:2445–2453.
15. Alberti KG, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, et al. Harmonizing the metabolic syndrome: a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation 2009; 120:1640–1645.
16. Brown TT, Cole SR, Li X, Kingsley LA, Palella FJ, Riddler SA, et al. Antiretroviral therapy and the prevalence and incidence of diabetes mellitus in the multicenter AIDS cohort study. Arch Intern Med 2005; 165:1179–1184.
17. Meisinger C, Doring A, Thorand B, Heier M, Lowel H. Body fat distribution and risk of type 2 diabetes in the general population: are there differences between men and women? The MONICA/KORA Augsburg cohort study. Am J Clin Nutr 2006; 84:483–489.
18. Jain MK, Aragaki C, Fischbach L, Gibson S, Arora R, May L, et al. Hepatitis C is associated with type 2 diabetes mellitus in HIV-infected persons without traditional risk factors. HIV Med 2007; 8:491–497.
19. Howard AA, Floris-Moore M, Lo Y, Arnsten JH, Fleischer N, Klein RS. Abnormal glucose metabolism among older men with or at risk of HIV infection. HIV Med 2006; 7:389–396.
20. Howard AA, Hoover DR, Anastos K, Wu X, Shi Q, Strickler HD, et al. The effects of opiate use and hepatitis C virus infection on risk of diabetes mellitus in the Women's Interagency HIV Study. J Acquir Immune Defic Syndr 2010; 54:152–159.
21. Brown TT, Tassiopoulos K, Bosch RJ, Shikuma C, McComsey GA. Association between systemic inflammation and incident diabetes in HIV-infected patients after initiation of antiretroviral therapy. Diabetes Care 2010; 33:2244–2249.
22. Goldberg RB. Cytokine and cytokine-like inflammation markers, endothelial dysfunction, and imbalanced coagulation in development of diabetes and its complications. J Clin Endocrinol Metab 2009; 94:3171–3182.
23. Bonora E, Kiechl S, Willeit J, Oberhollenzer F, Egger G, Meigs JB, et al. Population-based incidence rates and risk factors for type 2 diabetes in white individuals: the Bruneck study. Diabetes 2004; 53:1782–1789.
24. Lee GA, Rao MN, Grunfeld C. The effects of HIV protease inhibitors on carbohydrate and lipid metabolism. Curr HIV/AIDS Rep 2005; 2:39–50.
25. Lo YC, Chen MY, Sheng WH, Hsieh SM, Sun HY, Liu WC, et al. Risk factors for incident diabetes mellitus among HIV-infected patients receiving combination antiretroviral therapy in Taiwan: a case-control study. HIV Med 2009; 10:302–309.

diabetes mellitus; HIV infections; indinavir; lipodystrophy; stavudine

© 2012 Lippincott Williams & Wilkins, Inc.