Diabetes and HIV and HCV; aging/HIV and diabetes

Judith A Aberg and Jules Levin, NATAP

This article reports on studies presented at the 11th Retrovirus Conference held in February 2004 and highlights the concern that persons with HIV can be at greater risk for developing diabetes. This report also highlights research findings that diabetes may be more prevalent in older HIV-positive individuals and may lead to cognitive impairment, thus identifying a new risk factor associated with aging and HIV.

Reports include:

  • What is insulin resistance and diabetes and why is it important?
  • HIV-positive men were three times more likely to have diabetes than HIV-negative men in MACS
  • HIV-positive women: risk for developing diabetes is associated with over 50 years of age, smoking cigarettes, being Hispanic, body mass index (weight)
  • Atazanavir did not impair sugar metabolism and raise triglycerides in healthy volunteers
  • Comparative effects of nelfinavir and efavirenz on lipids and glucose in ACTG 384 Study
  • Lipids and sugar metabolism improve after switch from protease inhibitor to nevirapine, efavirenz or abacavir regimens in NEFA Study
  • Aging and HIV: diabetes found to be associated with cognitive impairment in older HIV-positive individuals; perhaps, a new risk factor for older HIV-positive individuals

What are insulin resistance and diabetes and why is it important?

There were several interesting abstracts presented at Retrovirus suggesting that insulin resistance and diabetes are associated with HIV and/or its therapies, particularly some of the protease inhibitors. First, I think it would be beneficial to briefly discuss what insulin resistance is and why one should be concerned about developing it. When one consumes carbohydrate, the body breaks it down into sugars, also called glucose. The body also produces insulin, which carries the sugar out of the bloodstream and into the tissue and cells.

The body does this so that it tries to maintain our blood sugar in what we label as a normal range of blood sugar. Some people require higher amounts of insulin to maintain glucose in a normal range and this is called insulin resistance. Insulin resistance by itself is associated with vascular disease and over time may progress to diabetes where the body can no longer keep the blood sugar in the normal range.

Mild diabetes can sometimes be managed by diet and exercise but frequently patients need to take pills or even shots of insulin to control the blood sugar. Uncontrolled diabetes may lead to kidney disease, blindness, neuropathies, vascular disease and even death. But even insulin resistance without diabetes can have major unhealthy effects including elevated blood pressure, abnormal lipids and coronary heart disease, which also can lead to significant illness and death.

HIV-positive men were three times more likely to have diabetes than HIV-negative men in MACS

Brown and colleagues presented Prevalence and Incidence of Pre-diabetes and Diabetes (DM) in the Multicenter AIDS Cohort Study [1]. They examined the prevalence of hyperglycaemia (elevated blood glucose) in 1,107 men enrolled in the Multicenter AIDS Cohort Study (MACS), using data from April 1999 to September 2002.

Hyperglycaemia (pre-diabetes and DM) was defined as a fasting plasma glucose (FPG) >110 mg/dL, use of anti-diabetic medication, or self-reported diagnosis of DM. DM was defined as a FPG >126 mg/dL, use of anti-diabetic medication, or self-reported diagnosis of DM. Of the 1,107 men, 563 were HIV-negative and 544 were HIV-positive (423 on HAART).

Of HIV-positive men on HAART, 14% had prevalent DM at baseline compared with 5% in the HIV-negative group (odds ratio = 4.4; 95% confidence interval [CI]: 2.6, 7.4, after adjustment for age and body mass index [BMI]). For the 618 men with a FPG <105 mg/dL, no history of DM or use of anti-diabetic medication at baseline, 79 (13%) incurred incident hyperglycaemia in 1,054 person-years yielding an overall rate of 7.5 cases per 100 person-years (95% CI: 6.0, 9.4), and 38 incurred incident DM in 1,088 person-years, yielding an overall rate of 3.5 cases per 100 person-years (95% CI: 1.7, 3.7).

After adjustment for age and BMI, the hazard of pre-diabetes or DM among the HIV-positive HAART group was 1.8 times (95% CI: 1.1, 3.0) that of the HIV-negative group, and the hazard of DM among the HIV-positive HAART group was 3.1 times (95% CI: 1.3, 7.1) that of the HIV-negative group. Exposure to a HAART regimen including a PI (hazard ratio [HR] = 1.9; 95% CI: 1.1, 3.3), d4T (HR = 2.1; 95% CI: 1.1, 3.9) or efavirenz (HR = 3.9; 95% CI: 1.6, 9.5) were each significantly associated with a higher rate of incident pre-diabetes or DM compared to the HIV-negative group.

The study concluded that HIV-positive men with HAART exposure had an increased prevalence and incidence of pre-diabetes and DM. Exposure to a HAART regimen — including PIs, d4T, or efavirenz — was associated with an apparent increased risk of hyperglycaemia.

HIV-positive women and glucose metabolism: associated with diabetes – over 50 years of age, smoking cigarettes, being Hispanic, body mass index (weight)

Howard and colleagues presented Impaired Glucose Metabolism and Antiretroviral Use Among HIV-infected Women [2]. As with the study discussed above, many have focused on men and there is limited knowledge of the effects of HIV and its therapies on women.

They performed a 75-g oral glucose tolerance test in 125 HIV-infected and 90 at-risk HIV-uninfected women without a history of diabetes, and assessed the association of antiretroviral use and non-medication related factors with impaired glucose tolerance, diabetes mellitus, and insulin resistance (HOMA).

The median age was 45 years (range 35 to 70); 51% were black, 38% Hispanic, 10% white; 38% had a family history of diabetes mellitus and 13% reported giving birth to a baby >9 lbs; median body mass index was 28.8 kg/m2 and mean waist-to-hip ratio was 0.89; 90% had ever smoked cigarettes (median 15.0 pack-years); 68% were current smokers; 41% had a history of injection drug use with no difference by HIV status. Among HIV-infected women, 25% were HAART-naïve, 23% were on HAART but protease inhibitor (PI)-naïve, and 52% were on HAART with PI. Median duration of PI use was 43 months. Median CD4 count was 481 cells/mm3.

The prevalence of diabetes (fasting glucose ≥126 mg/dL or 2-hour glucose ≥200 mg/dL) among all women was 6% (n = 14) and of impaired glucose tolerance (IGT, 2-hour glucose ≥140 and <200) was 11% (n = 23), with no difference by HIV status, HAART, or PI use.

Mean log insulin resistance (HOMA) (U/mL·mM) was lower among HAART-naïve HIV-infected women (0.40) compared with those on PI-HAART (0.45), non-PI HAART (0.48), or HIV-uninfected women (0.47), but this difference was not significant.

In a logistic regression model, factors independently associated with an abnormal oral glucose tolerance test (impaired glucose tolerance or diabetes mellitus) included age ≥50years (ORadj 4.5, 95%CI 1.5, 13.4) and smoking (ORadj 1.7 per 10 pack-years, 95%CI 1.2, 2.4), after controlling for HIV, HAART use, PI use, race, family history of diabetes, and waist-to-hip ratio.

In a linear regression model, factors independently associated with log insulin resistance (HOMA) among HIV-infected women included body mass index (p <0.0005), Hispanic race (p = 0.047), and non-PI HAART (p = 0.04), after controlling for PI use and CD4 count.

They concluded that impaired glucose tolerance and diabetes mellitus were detected by oral glucose tolerance tests in a substantial minority of women, and were associated with traditional diabetes risk factors rather than HIV infection, PI or HAART use.

However, among HIV-infected women, non-PI HAART use was independently associated with greater insulin resistance. One has to be careful in interpreting this study especially given the high background of traditional risk factors and one may need a much larger sample size to reduce the impact of these confounding factors. For example, the majority of the subjects were from ethnic backgrounds that clearly have a higher rate of diabetes plus 38% had a family history of diabetes. All in all, this is a first step and further studies among women and minorities are warranted.

Nevertheless, both these studies do demonstrate that significant numbers of persons have either diabetes or insulin resistance. There were a few studies exploring the effects of various ART. This can be quite complicated and some studies examined the effects of ART among HIV sero-negative subjects while others examined the effects among those subjects infected with HIV.

Reyataz: did not impair sugar metabolism and raise triglycerides in healthy volunteers

Investigators from Bristol-Myers Squibb presented The Effect of Atazanavir vs Lopinavir/ritonavir on Insulin-stimulated Glucose Disposal Rate in Healthy Subjects [3]. A proposed mechanism for why protease inhibitors may be associated with the development of diabetes is via blockade of the glucose transporters that take the glucose from the bloodstream into the tissues.

Atazanavir (ATV), unlike indinavir, lopinavir, and ritonavir, appears not to block glucose transport through the glucose transporter-4 insulin-sensitive transporter in vitro.

This study compared the effects of ATV and lopinavir/ritonavir (Kaletra, LPV/r) to placebo on insulin-stimulated glucose disposal rates. This was a randomised, double-blind, cross-over study of the effect of five days of treatment with ATV, LPV/r, or placebo on insulin-stimulated glucose disposal in healthy HIV-negative volunteers. Each subject was studied on two of three possible treatments using the hyperinsulinaemic euglycaemic clamp technique (180 minutes) with ≥14 days of wash-out.

Difference among groups in insulin-stimulated glucose disposal per unit of insulin and glycogen storage rate (proportion of total glucose disposal taken up by the tissue but not oxidised) was analysed by ANOVA. They studied 30 healthy HIV seronegative adult men with median age of 35 years (range 19 to 49), mean weight of 76.4 kg (SD = 9.9), mean body mass index of 24.0 kg/m2 (SD = 2.4).

During steady-state euglycaemia (60 to 180 minutes), insulin levels were raised comparably (65.4, 63.0, 63.9 µU/mL) and glucose was clamped at ~75 mg/dL under all conditions.

LPV/r decreased the mean insulin-stimulated glucose disposal per unit of insulin (M/I) by 24% compared to placebo and by 23% compared to ATV. LPV/r decreased glycogen storage rate (GSR) by 35% compared to placebo and by 38% compared to ATV.

[NOTE – please see PDF file for TABLE]

In conclusion, ATV did not reduce insulin sensitivity and had no effect on insulin-stimulated glucose disposal or GSR. In contrast, LPV/r induced insulin resistance and reduced the glucose disposal per unit of insulin and glycogen storage rate. These data are consistent with in vitro studies showing that ATV does not interfere with glucose transporter-4 activity and does not induce fasting hyperinsulinaemia, substantiating the findings of large clinical trials. In addition, fasting triglycerides were not affected by ATV, but increased a mean of 43% on LPV/r. This is welcomed and supporting evidence that ATV is not associated with the metabolic complications as many of the other antiretroviral drugs are.

Comparative effects of nelfinavir and efavirenz on lipids and glucose in ACTG 384 study

The ACTG 5005s team presented partial results of a metabolic substudy of a large, randomised study (ACTG 384) which compared the nucleoside (NRTI) backbones of either AZT/3TC or d4T/ddI with evafirenz (EFV), nelfinavir (NFV) or combined EFV/NFV regimens [4].

The primary objective of A5005s was to determine whether NFV- and EFV-based therapies differ with respect to changes in fasting lipids and insulin resistance. Secondary objectives included comparisons among NRTI regimens.

Antiretroviral-naïve subjects (n=334) received NFV (99), EFV (110), or both (125) plus zidovudine (ZDV) + lamivudine (3TC) (154) or didanosine (ddI) + stavudine (d4T) (180) in a substudy of a 3×2 randomised factorial trial. Fasting samples were collected at entry, 8, 16, 32, 48, and 64 weeks. Primary analyses (Wilcoxon tests) are intent-to-treat; changes from entry are reported as median [IQR] at week 32. The EFV+NFV group was excluded from NFV vs EFV comparisons, but included in NRTI analyses.

  • Lipid values (mg/dL) increased in all groups. The proportion with total-cholesterol >200 increased from 13% to 45% at week 32 (cholesterol went up);
  • Those with HDL (good)-cholesterol <40 fell from 75% to 48% (each p <0.001), so there was a negative effect on good cholesterol.
  • HDL-C (positive effect on good cholesterol) increases correlated with higher HIV RNA and lower CD4 at entry (each p <0.001).
  • Similar increases occurred with both NFV and EFV in total-cholesterol (NFV 28 [9, 56], EFV 25 [9, 52], non-HDL-Cholesterol (22 [7, 52] versus 19 [0, 42]), and triglycerides (TG) (13 [-13, 56] versus 32 [-19, 76]).
  • Only 6% with EFV and 5% with NFV had TG >400 at week 32.
  • HDL-C increases tended to be greater with EFV (7 [2, 12]) than NFV (5 [0, 10], p = 0.11), with a more favourable change in total:HDL-C ratio with EFV (-0.4 versus 0.4, p = 0.03).
  • There was some evidence of greater increases in total-cholesterol with ddI+d4T (45 [12, 71]) versus ZDV+3TC (29 [10, 55], p = 0.07) and non-HDL-C (35 [10, 60] versus 26 [0, 52], p = 0.12). HDL-C (8 [2, 14] versus 7 [0, 12]) and TG (30 [-13, 84] versus 25 [-14, 60]) increases were similar.
  • Insulin resistance (by HOMA-insulin resistance) increased over time for the whole group. From a baseline of 1.39 [0.90, 2.05], median change at week 8 was 0.20 [-0.37, 0.72] (p = 0.03); no changes were significant within any group.
  • At week 32, the overall median increase was 0.38 [-0.22,0.97] (p = 0.002), but differences in the change in HOMA-insulin resistance between groups were minimal: NFV 0.41 [-0.37, 1.43], EFV 0.39 [-0.17, 0.82] (p = 0.9); ddI+d4T 0.32 [-0.29, 0.85], ZDV+3TC 0.40 [-0.19, 1.30] (p = 0.4).


Overall, the PI NFV and the NNRTI EFV appear to have comparable effects on fasting lipids, with a better total:HDL-C ratio with EFV.

  • Lipids tended to be slightly more favorable with ZDV+3TC than ddI+d4T.
  • HDL-cholesterol changes correlated with entry HIV disease status.
  • Insulin resistance worsened in the group as a whole, but did not differ between regimens.
  • The lack of an early increase in insulin resistance with NFV suggests that acute insulin resistance is not a PI drug class effect.

This supports earlier studies that suggested that NFV does not significantly inhibit the gluc-4 transporter system as mentioned above. Nevertheless, insulin resistance did worsen over time in a subset of subjects and further study is warranted to explore these findings.

Lipids and sugar metabolism improve after switch from protease inhibitor to nevirapine, efavirenz or abacavir regimens in NEFA Study

In addition, Dr Fisac presented further data on the metabolic complications evaluated in the NEFA study [5].

NEFA was an open-label randomised study comparing three different protease inhibitor-sparing regimens (ABC-abacavir, EFV-efavirenz, and NVP-nevirapine) in HIV-positive individuals who had been previously exposed to protease inhibitor-containing regimens. A sub-study in 92 patients was conducted to evaluate the switching effect on metabolic and body composition parameters. The metabolic outcomes in 69 patients who maintained the initially allocated treatment for 24 months (ABC: n = 22; EFV: n = 21; NVP: n = 26) were presented.

Fasting serum total cholesterol, low-density lipoprotein cholesterol (LDLc), high-density lipoprotein cholesterol (HDLc), triglycerides, glucose, and insulin were determined. Insulin resistance by the homeostasis model assessment and total cholesterol:HDLc ratio were also calculated.

In an overall analysis, insulin, insulin resistance, total cholesterol, LDLc, HDLc, and total cholesterol:HDLc ratio improved (baseline vs 24-month data). EFV and NVP arms showed similar metabolic benefits after two years of therapy.

The sample size is too small to compare individual PI-containing regimens at baseline. However, these results remain encouraging that metabolic complications that may be associated with certain PIs may improve after switching to a NNRTI.

Aging and HIV: diabetes found to be associated with cognitive impairment in older HIV-positive individuals; perhaps, a new risk factor for older HIV-positive individuals

Finally, we would like to mention a study that did not receive that much attention but one that brings up the issue of aging and how age plays into the role of HIV and its associated complications. A group of investigators from Hawaii has been following an aging population with HIV, “The Hawaii Aging with HIV Cohort” [6]. It is estimated that over 10% of newly diagnosed HIV infections occur in the population over the age of 50. Age plays a major role in the development of CHD, diabetes, hypertension and many diseases.

The impact on aging and its association with development of dementia in HIV is unknown. The investigators took a step further and asked whether the metabolic complications associated with HIV have even more of an impact on the aging HIV-infected population.

Participants were from one of two groups (under 40 or 50+ years old) within the HIV-positive arm of the Hawaii Aging with HIV Cohort. Evaluations included comprehensive neuropsychological testing. Three measures of cognitive functioning were constructed from combinations of scores on neuropsychological test results standardised within our sample: an overall measure of cognitive functioning, NPZ8; a measure of memory, NPZ3-memory; and a measure of psychomotor functioning, NPZ3-psychomotor.

Trained personnel obtained medical histories including established diagnoses for diabetes (DM) using a structured interview. Data from 169 participants (73 younger and 96 older) were available for these analyses.

  • Frequency of DM was 8.9% (15.6% among older and 0% among younger).
  • DM was negatively associated with overall cognitive functioning (F = 19.15, p<0.01), accounting for 11% of the variance in NPZ8 scores. DM was also negatively associated with psychomotor functioning (F = 14.16,pp < 0.01) accounting for 8% of the variance in NPZ3-psychomotor scores.
  • There was no association between DM and NPZ3-memory scores.
  • Controlling for age, ARV, current hypertension, current
    hypercholesterolaemia, pack-years of smoking, ethnicity, and duration of HIV infection, did not substantially alter these results.

These data suggest that diabetes is associated with decreased overall cognitive performance and specifically psychomotor performance in patients with HIV.

Our findings are driven exclusively by diabetes in older patients and thus, if confirmed, may represent a newly identified risk factor for older HIV-positive patients. This risk is independent of other vascular risk factors. The underlying mechanism is not clear. While speculative, this could be associated with metabolic dysfunction and abnormalities in glucose regulation. Further studies are certainly needed that explore the effects of HIV and its therapies on aging HIV-positive individuals, as many have other traditional risk factors for metabolic and cardiac diseases and we will need to know how best to manage them.

In summary, a quote from a colleague, Dr Donald Kotler: “Insulin resistance kills. Why would anyone think having HIV would be protective?” Dr Kotler’s point is well taken. Although we do not know what the risk of progressing from insulin resistance to diabetes is among those infected with HIV, we do know insulin resistance without HIV is bad. Just as we cannot be passive and let individuals sit with high lipids and ignore markers for cardiac disease, we should be managing our HIV-infected patients with insulin resistance or diabetes as we would the general population. The question remains whether conventional therapies for diabetes will work similarly in the HIV infected population and further studies are warranted exploring the pathogenesis and management of insulin resistance in this population.


  1. Brown TT, Cole SR, Li X et al. Prevalence and incidence of pre-diabetes and diabetes in the Multicenter AIDS Cohort Study. Abstract 73.
  2. Howard A, Floris-Moore M, Arnsten J et al. Impaired glucose metabolism and antiretroviral use among HIV-infected women. Abstract 701.
  3. Noor M, Grasela D, Parker R et al. The effect of atazanavir vs lopinavir/ritonavir on insulin-stimulated glucose disposal rate in healthy subjects. Abstract 702.
  4. Dube M, Zackin R, Parker R et al. Prospective study of glucose and lipid metabolism in antiretroviral-naive subjects randomised to receive nelfinavir, efavirenz, or both combined with zidovudine+lamivudine (ZDV+3TC) or didanosine+stavudine: A5005s, a substudy of ACTG 384. Abstract 74.
  5. Fisac C, Fumero E, Crespo M et al. Metabolic changes in patients switching from a protease inhibitor-containing regimen to abacavir, efavirenz, or nevirapine: 24-month results of a randomised study. Abstract 78.
  6. Valcour V, Shikuma C, Shiramizu B et al. Diabetes and cognitive functioning among HIV seropositive patients. The Hawaii Aging with HIV Cohort. Abstract 502.

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