An Interview With Todd T. Brown, M.D., Ph.D.
Dr. Brown is Associate Professor of Medicine and Epidemiology at Johns Hopkins University in Baltimore and the primary endocrine consultant to the Johns Hopkins HIV Clinic. He holds an M.D. from the Columbia College of Physicians & Surgeons and a Ph.D. from the Johns Hopkins Bloomberg School of Public Health. Over the past decade, Dr. Brown has become a leading authority on metabolic, endocrine, and skeletal abnormalities in people with HIV infection, particularly as those conditions relate to aging. A coinvestigator in the Multicenter AIDS Cohort Study (MACS), he chairs that cohort's Metabolic Working Group. In the AIDS Clinical Trial Group (ACTG), Dr. Brown is a member of the Inflammation/End Organ Disease Transformative Science Group and serves in leadership positions on multiple studies. Since 1995 he has coauthored over 70 reports of original research in peer-reviewed publications and has contributed eight book chapters.
Is HIV a Diabetes Risk Factor?
Mascolini: Many HIV populations have high frequencies of classic diabetes risk factors, but is HIV itself a risk factor?
Brown: This is a controversial question: different cohorts have different results. Some cohorts like the MACS found a clear increase in the prevalence and incidence of diabetes in HIV-infected people compared with uninfected controls.1 This has also been seen in the Women's Interagency HIV Study (WIHS)2 and in several other cohorts. But other cohorts such as the Veterans Aging Cohort Study (VACS) have seen quite the opposite -- that HIV-infected patients in the VA system have a lower incidence of diabetes than veterans without HIV.3 So I think the impact of HIV on diabetes risk is probably population-dependent.
What we're seeing with HIV and diabetes is similar to what we see for other comorbidities: HIV disease itself may have some impact, antiretroviral therapy may have an impact, and traditional risk factors for the comorbidity are also critically important. We can look at each of those factors in turn.
For HIV disease itself, we know that HIV is associated with chronic inflammation. Even patients who have undetectable viral loads have residual inflammation that's thought to lead to comorbidities, and diabetes is one of them. We showed in the AIDS Clinical Trials Group that TNF receptor levels 48 weeks after antiretroviral initiation, when people had a suppressed viral load, were associated with incident diabetes.4 From that perspective, HIV is probably an independent risk factor for diabetes.
The next category is the effect of antiretrovirals. Antiretrovirals have changed quite a bit since the introduction of highly active antiretroviral therapy (HAART): In general the medications have become a lot more metabolically friendly, and this improvement definitely includes effects on glucose metabolism. In the late 1990s, soon after widespread uptake of HAART, research showed that early protease inhibitors (PIs) had a marked effect on glucose metabolism. PIs have gotten better over the course of the past 15-plus years, and currently used PIs have relatively modest effects on glycemia. Thymidine nucleoside analogs, particularly stavudine, were also associated with abnormal glucose metabolism, not only through direct effects, but also indirectly by inducing lipoatrophy, which in turn is associated with insulin resistance and diabetes onset.
Probably the biggest drivers of diabetes in HIV populations are traditional risk factors -- mainly increased adiposity and obesity. We're seeing high rates of obesity in HIV populations, just as we are in the general population, and this is the biggest driver of the diabetes epidemic in the United States. The other major factor of course is age. The HIV population is getting older, and age is a major diabetes risk factor.
Other reasons why HIV patients may have an increased risk of diabetes are other infections: Hepatitis C infection is associated with abnormal glucose metabolism, and this may contribute to diabetes in HIV/HCV-coinfected patients. Concomitant medications may also play a role. Steroids have a huge effect on glucose metabolism, and for patients taking medications that interfere with steroid metabolism -- such as ritonavir and cobicistat -- this may be a particular problem. Also, some HIV patients take atypical antipsychotics, which can increase diabetes risk by increasing weight and also through an independent effect on glucose metabolism.
The bottom line is that HIV patients can have several diabetes risk factors that HIV-uninfected patients don't have, so they may be at higher risk for diabetes.
Mascolini: What's going on in your own patient population in terms of diabetes incidence? Have you seen any change over the past decade or since the introduction of combination antiretroviral therapy?
Brown: It's a little tricky to tease this apart because we saw overall exponentially increased rates of diabetes in the general population during this time. I think diabetes incidence may have decreased slightly in HIV patients during the HAART era as the medications got better, and a cohort study in France showed this quite nicely.7 But because as a population we're getting fatter, overall diabetes rates are very high, not only in HIV populations but in the general population. Perhaps the specific impact of antiretroviral therapy on diabetes incidence has decreased over this time, but the overall rates of diabetes have actually increased, probably because of increased adiposity.
|Tips From This Interview on Diabetes Management in People With HIV|
- The impact of HIV itself on diabetes risk remains controversial, but there is no question that HIV populations carry a heavy burden of traditional diabetes risk factors.
- The hemoglobin A1c test underestimates blood glucose in people with HIV. A hemoglobin A1c of 6.5% in an HIV patient may be equivalent to 7.0% in an HIV-uninfected patient, so this test should probably not be used to screen for diabetes in people with HIV.
- Hemoglobin A1c targets for diabetes treatment should probably be individualized, with more stringent targets (6.0% to 6.5%) for younger people without comorbidities, and looser targets (7.0% to 8.0%) for older people with comorbidities and thus a higher risk of hypoglycemia.
- High mean corpuscular volume (MCV) is a strong predictor of hemoglobin A1c/glycemia discordance. If I see an MCV in the high 90s or over 100 in an HIV-infected patient, I know not to trust the HbA1c too much.
- The dose of saxagliptin, an oral hypoglycemic agent, may need to be lowered to 2.5 mg when given with a potent CYP3A4 inhibitor.5
- The HIV integrase inhibitor dolutegravir increases metformin plasma exposure and may necessitate downward dose adjustment of metformin.6
- Whether an HIV patient with diabetes should be referred to an endocrinologist depends on the provider's comfort level in managing diabetes. Bear in mind that specialists have an array of resources that can improve diabetes management, like certified diabetes educators and nutritionists.
Impact of Newer Antiretrovirals
Mascolini: Older antiretrovirals like d4T and indinavir were linked to glucose abnormalities. Among antiretrovirals often prescribed today, do any pose a risk of dysregulated glucose?
Brown: Glucose metabolism and adiposity generally get worse after antiretroviral initiation. This effect is probably driven by the fat increase many people have after starting antiretroviral therapy. One outstanding question is whether this is a return-to-health phenomenon that is part and parcel of getting better from HIV, or whether it is a maladaptive response and whether the proportion of fat versus lean mass is abnormal in people who start antiretroviral therapy. But I don't think there are big differences between individual current antiretrovirals in terms of their effects on glucose metabolism.
ACTG A5257 was a study of HIV-infected treatment-naive patients initiating either ritonavir-boosted atazanavir, boosted darunavir, or raltegravir with tenofovir/emtricitabine.8 In the A5260 substudy of A5257, we looked at the effects of these three regimens on glucose metabolism and adiposity,9 and these results will be detailed at CROI in an oral presentation.
Mascolini: When you talk with primary HIV providers, do they seem to have a good understanding of how current antiretrovirals affect glucose or insulin?
Brown: I think there's an overemphasis on the effects of specific antiretrovirals on glucose metabolism. As we already discussed, some first-generation antiretrovirals had an important effect on glucose metabolism, but the agents used today generally don't. That's not to say there aren't some patients whose blood sugars become quite deranged while they're taking certain current antiretrovirals, probably for genetic reasons. But I think there's an overemphasis in the HIV community on how much currently prescribed antiretrovirals affect glucose metabolism.
Advice on Screening HIV Patients for Diabetes
Mascolini: In your group at Hopkins, what's the protocol for glucose screening and follow-up for a person with newly diagnosed HIV?
Brown: The HIV guidelines in place right now suggest that we should get some measure of glucose metabolism and screen for diabetes at the time of HIV diagnosis, after 3 to 6 months of antiretroviral therapy, and yearly thereafter. These guidelines were written earlier in the HAART era, when we were seeing a large effect of antiretroviral therapy on glucose metabolism. Today this recommendation might be a little bit of overkill. However, the tests to screen for diabetes are relatively benign and relatively low in cost, so it probably does make sense to get these fasting tests when you're getting fasting lipids, for example.
The American Diabetes Association recommends diagnosing diabetes with one of four tests: (1) a fasting glucose, with 126 mg/dL or higher indicating diabetes, (2) a 2-hour 75-g oral glucose tolerance test, with 200 mg/dL or greater considered diabetes, (3) a random glucose with symptoms of diabetes -- polyuria and polydipsia -- and a diagnostic cutoff of 200 mg/dL or higher, or (4) hemoglobin A1c (HbA1c), which came online in 2010.
Hemoglobin A1c has some advantages over the other tests in that it needs no special preparation -- patients don't have to fast before the test or swallow a glucose solution. Also, instead of giving you a one-point-in-time look at glucose, which can bounce around a little bit, it gives you an integrated view over 3 months, which is the life of the red blood cell: The test calculates glucose by measuring its attachment to a hemoglobin moiety in red blood cells. Because of these advantages, hemoglobin A1c is quite attractive.
The caveat is that the hemoglobin A1c may underestimate glycemia in HIV patients.10 The effect of this underestimation is somewhere between 0.2 to 0.5 hemoglobin A1c percentage points. A hemoglobin A1c of 6.5% in an HIV patient may be equivalent to an HbA1c of about 7.0% in an HIV-uninfected patient. This makes the use of hemoglobin A1c for the diagnosis of diabetes a little bit tricky in HIV-positive patients. It could potentially lead to the underdiagnosis of diabetes in people with HIV. For that reason I generally recommend that we stick to the fasting glucose tests for screening.
The reasons behind this hemoglobin A1c discordance in people with HIV aren't entirely clear. HIV patients may have a low-grade hemolysis that causes increased red blood cell turnover and an artificial lowering of hemoglobin A1c -- that is, a disconnect between glycosylation of the hemoglobin molecule and what the actual blood glucose is.
One very strong risk factor for hemoglobin A1c/ glycemia discordance that has been seen in multiple studies, including our work in the MACS,10 is a high mean corpuscular volume (MCV). The higher the MCV is over 90 or so, the greater the hemoglobin A1c/glycemia discordance. This is useful clinically since MCV is routinely measured on a complete blood count (CBC). In my practice, if I see an MCV in the high 90s or over 100 in an HIV-infected patient, I know not to trust the HbA1c too much. The reasons underlying this association are unclear, but we hypothesize that these larger red cells are taken out of circulation more quickly than normal-size cells.
Mascolini: Does hemoglobin A1c have a role in monitoring patients once they start treatment for high blood glucose?
Brown: I do follow the A1c in HIV patients because it is a critical measure. One of the important developments in the diabetes world over the past few years is the individualization of the hemoglobin A1c goal. We used to shoot for a hemoglobin A1c less than 7.0% in everyone.
Now the recommendation is to take a more individualized approach. For people at potentially higher risk for hypoglycemia or with comorbidities, the recommendation is to relax the hemoglobin A1c target to between 7 and 7.5, and in some cases up to 8. For younger people who don't have comorbidities, you might push them a little bit harder to reach a lower A1c target. For an HIV patient, I take that individualized approach and might push them a little bit harder, knowing that their hemoglobin A1c may be an underestimate of their real glycemia.