Making Sense of Diverse Diabetes Rate Reports in HIV Cohorts
|Abstract: Diabetes prevalence continues to climb at an alarming pace across the world. Although HIV populations are no exception to this trend, diabetes prevalence varies 10-fold across studies of HIV cohorts because of differences in their composition and in diabetes definitions applied. Two of three studies that compared diabetes prevalence in people with and without HIV found higher rates in the HIV group. Among people with HIV infection, diabetes incidence is higher in U.S. cohorts than in other groups, probably because of higher risk factor rates in the U.S. groups. Though diabetes incidence was higher with than without HIV in earlier studies -- when people were taking more toxic antiretroviral regimens -- the largest and latest comparisons found lower diabetes incidence with HIV than without HIV.|
Diabetes history reads like a roster of euphoniously named heroes. Everyone knows Banting and Best, who found that insulin regulates blood sugar. But who remembers Mering and Minkowski, or Sharpey-Schafer, or Aretaeus of Cappodocia? Not to mention the team of Langerhans and Languesse. Many an endocrinologist can tell you about Paul Langerhans, father of the eponymous pancreatic islets. But how many know G.E. Languesse named those islets (Figure 1)? And we can thank Sharpey-Schafer (Sir Edward Albert) for proposing the term insulin (after the Latin for island)1 rather than Langerhansin (and the consequent Langerhansinemia). If quizzed on who first described diabetes -- or nearly any classical disease -- you're safe to credit the ancient Egyptians. That's probably not because primeval Nile valley denizens had such an acute appreciation of diverse debilities, but because they kept heaps of hieroglyphic charts. Pritzker School of Medicine dean Kenneth Polonsky notes that disease-conscious Egyptians first recorded diabetes around 1500 BCE, pinpointing two key traits -- prodigious urination and weight loss.1 In the second century CE Greek healer Aretaeus minted the term diabetes mellitus,* acknowledging the sweet taste of a diabetic's urine. Mellitus means "honey sweet," but we do not know if Aretaeus personally gathered evidence of this saccharine signal. Diabetes comes from the Greek word for siphon, a possible nod to the diabetic's unquenchable thirst.3
More than 1600 years elapsed after these Aretaean observations before someone -- an English physician called Matthew Dobson -- got the idea to measure glucose in urine of diabetics and confirmed lofty values.1 In 1889 Mering and Minkowski (Austrians Joseph von and Oskar) ablated the pancreas from dogs, who promptly died of diabetes, and thereby showed that the organ plays a critical role in ruling glucose.1
But how? Sharpey-Schafer, an English physiologist, hypothesized (and named) the pancreatic chemical insulin in 1910.1 Only 11 years later Frederick Banting and Charles Best reversed diabetes induced in dogs by infusing pancreatic islet cells from healthy dogs, and their colleague James Collip purified the hormone insulin. The Nobel committee bestowed its esteemed prize on Banting and his boss John Macleod but passed up Collip and Best.4 Insulin became the first hormone ever cloned then massively produced for treatment via recombinant DNA technology,1 the first chapter in a biotech revolution that has saved countless lives.
Global and U.S. Diabetes Numbers Grow
Diabetes mellitus shares three features with HIV infection. (1) Before research identified effective therapies, both diseases remained almost universally -- and painfully -- fatal. (2) Once solid benchwork nailed down the pathogenic basics of each disease, brisk development of practical treatment turned both into manageable chronic diseases. (3) Neither can be cured. Despite diverse effective remedies for diabetes, Kenneth Polonsky argues that, "if one views diabetes from a public health and overall societal standpoint, little progress has been made toward conquering the disease during the past 200 years," largely because about 90% of cases today result from a global obesity epidemic.1
Polonsky concludes that diabetes has become "one of the most common and most serious medical conditions humankind has had to face."1 To be sure, the global burden of 35 million cases of HIV infection pales beside the 387 million with diabetes.5 The International Diabetes Federation projects that diabetes prevalence will climb to 592 million by 2035 and notes that every country in the world records a rising tally of residents with type 2 diabetes.5
The American Diabetes Association figures that 29.1 million people in the United States had diabetes in 2012, or 9.3% of the population.6 Those numbers rose from 25.8 million and 8.3% in 2010. Centers for Disease Control and Prevention (CDC) number crunchers reckon that diabetes prevalence in the country will double from 2005 to 2050.7 Parsing recent trends, Polonsky figures that 1 in 3 U.S. adults could have diabetes by 2050.1 The ADA estimates that 86 million U.S. residents 20 or older had prediabetes (blood glucose 100 to 125 mg/dL) in 2012, up from 79 million 2 years earlier.6
Diabetes Prevalence 3% to 14% With Treated HIV
Research pegs diabetes prevalence in Western HIV populations at anywhere from an elfin 2.6% to a flush 14%,8-15 compared with the U.S. general population estimate of 9%.6 This wide range should spark no surprise since the HIV studies varied in diabetes definitions, populations assessed, their diabetes risk factors, and their antiretroviral treatment status. The U.S. estimate of 9% includes about 8 million people with undiagnosed diabetes, while the HIV calculations include only diagnosed people.
The first report of high blood glucose in people with HIV charted proportions of patients with extremely high glucose -- at or above 250 mg/dL, that is, twice the diabetes diagnosis cutoff of 126 mg/dL -- in the 8 years before combination antiretroviral therapy (cART) arrived.8 This review from the University of Alabama at Birmingham does not say how many people were taking antiretrovirals, though most probably were since the study period began in 1988. But the study period ended in 1995, just before the cART era. Twenty-five people, almost 2%, had sky-high glucose (Table 1). Eleven of those 25 were black, and 10 were taking megestrol acetate (Megace), a synthetic progesterone then prescribed to stimulate appetite in people with HIV but now rarely used by HIV-positive people. The researchers did not report how many people met the 126-mg/dL cutoff.
In a 419-person antiretroviral-naive U.S. cohort studied from 1999 through 2002, diabetes prevalence measured only 2.6%,9 not much more than the ultrahigh glucose prevalence in the Alabama study.8 This 1999-2002 study group had two notable high-glucose risk factors -- 60% were black and 10% Hispanic, and average body mass index approached the overweight threshold at 24.2 kg/m2. But this study probably underestimated diabetes prevalence because it used a loose definition -- diabetes reported in patients' medical records.
Defined more rigorously (see footnote to Table 1), diabetes affected 3% to 6% of HIV cohorts in the Bronx, New York,11 in France,12 across Europe, the U.S., Australia, and Argentina,13 and in the Bologna14 and Milan15 regions of Italy. Almost all of these people were taking cART, high proportions were men, most were white, and most were not overweight or obese (Table 1). The Bronx group is an exception to these generalizations.11 More than three quarters in these people were taking cART, but 88% were black or Hispanic, 53% were women, 61% were overweight or obese, and 58% had HCV infection, an established diabetes risk factor.
Another U.S. analysis involved 568 HIV-positive men and 718 HIV-negative but at-risk men in the Multicenter AIDS Cohort Study (MACS) seen from 1999 through 2003 in four cities.10 Only 15% were nonwhite, and almost three quarters were taking cART (Table 1). Median body mass index in HIV-positive men measured 25 kg/m2 -- at the overweight threshold -- regardless of cART status. Prevalence of diabetes (defined as fasting glucose ≥126 mg/dL, self-reported diabetes diagnosis, or self-reported use of antidiabetic medication) stood at 5% in HIV-negative men, 7% in HIV-positive men not on cART, and 14% in HIV-positive men on cART.
Why did HIV-positive MACS men shoulder a higher diabetes burden than other U.S. and European populations -- a burden more in line with estimated diabetes prevalence in the general U.S. population? The MACS analysis cannot be compared with the early University of Alabama study, which recorded prevalence of blood glucose at or above 250 mg/dL rather than the diabetes cutoff of 126 mg/dL.8 Diabetes prevalence may be higher in MACS than in the antiretroviral-naive U.S. group9 because that study did not actually measure fasting blood glucose and so probably underestimated the true diabetes rate. Even so, blacks and Hispanics -- with their higher diabetes risk than whites -- made up a much greater proportion of the naive cohort (70%) than the MACS cohort (15%). And even the naive HIV-positive MACS contingent10 had more than a twice higher diabetes rate than the naive cohort.9
Substantially higher BMI could contribute to greater diabetes prevalence in MACS men10 (25 kg/m2) than in the largely European DAD Study (23.0 kg/m2)13 and the three other European groups (23.1 and 23.2 kg/m2 in the Italian groups,14,15; in France ranging from 22.5 kg/m2 in people with no sign of lipodystrophy to 24.1 kg/m2 in people with isolated fat accumulation12). Diabetes definitions and antiretroviral treatment rates were similar in MACS and the DAD and European studies. MACS included a 5-fold higher proportion of nonwhites (15%)10 than the Bologna study (3%)14 but a similar proportion to the DAD study (13%);13 the French and Milan studies did not report race. MACS men were almost a decade older than the DAD13 or Bologna14 groups but similar in age to the Milan group.15 One third in each of the Italian groups had HCV infection. But the MACS investigators and the other studies did not report HCV status.
Why diabetes prevalence stood so much higher in MACS men10 than in Bronx men and women11 is tougher to construe. The groups had similar proportions taking antiretrovirals, but the Bronx cohort had loftier rates of other risk factor that might have led to a higher diabetes prevalence -- somewhat older age, higher body mass index, a much higher proportion of blacks and Hispanics, plus a lofty HCV prevalence (Table 1). The different diabetes rates could reflect a different diabetes definition in the Bronx study (see Table 1 footnote) and the much smaller cohort.
Regardless of variables that may explain differing diabetes rates in these study group,8-15 how do these HIV cohorts compare with selected control groups in diabetes prevalence? Three of these eight diabetes studies compared prevalence in HIV-positive people with the rate in an HIV-negative control group. Two of them found a higher diabetes rate in the HIV group10,15 and one did not.11
The MACS study of 568 HIV-positive men and 710 HIV-negative men seen at the same four U.S. centers recorded lower diabetes prevalence in HIV-negative men (5%) than in HIV-positive men not taking antiretrovirals (7%) or HIV-positive men taking antiretrovirals (14%).10 After adjustment for age and body mass index, HIV-positive men not on treatment had more than a twice higher prevalence than HIV-negative men (prevalence ratio 2.21, 95% confidence interval [CI] 1.12 to 4.38), and HIV-positive men on treatment had more than a 4-fold higher diabetes rate (prevalence ratio 4.64, 95% CI 3.30 to 7.10) (Figure 2).
The study of 4249 HIV-positive people in Milan compared them with 9148 healthy controls from 15 regions around Italy.15 Prevalence in the HIV group stood at 4.1%, compared with 2.5% in healthy controls, a highly significant difference (P < 0.0001). Statistical analysis adjusted for age, gender, BMI, triglycerides, total cholesterol, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol determined that people with HIV had more than 50% higher odds of diabetes (adjusted odds ratio 1.55, 95% CI 1.02 to 2.31, P = 0.0035) (Figure 2).
A comparison of 222 HIV-positive people and 155 HIV-negative people in two Bronx cohorts charted diabetes prevalence of 5% in the HIV group and 8% in the no-HIV group.11 At the same time, the combined rate of impaired fasting glucose plus impaired glucose tolerance was almost twice higher in the HIV group than in the comparison group: 26% versus 14%. High rates of diabetes risk factors in both the HIV-positive group and the HIV-negative group (nonwhite race, overweight and obesity, HCV positivity) could explain why diabetes prevalence did not differ by HIV status in this study.
None of the three studies comparing diabetes prevalence by HIV status took place within the past 6 years.10,11,15 Diabetes incidence comparisons by HIV status, reviewed below, indicate that diabetes incidence has become lower in HIV-positive than negative people in recent years.
Incidence Above 1 per 100 Person-Years in U.S. HIV+
The International Diabetes Federation,5 the American Diabetes Association (ADA),6 and the CDC7 agree that diabetes prevalence keeps climbing at a heady pace around the world, and that can mean only one thing: Incidence -- the new diagnosis rate -- far outstrips mortality of people already afflicted with diabetes.
Headcounters reporting for the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) say clinicians diagnosed 1.7 million new diabetes cases in the United States in 2012.16 In contrast, about 50,000 U.S. residents get infected with HIV every year -- 3% of the diabetes rate. Overall U.S. diabetes incidence stood at 0.78 per 100 people but was higher in people 45 to 64 (1.20 per 100) and people 65 or older (1.15 per 100). And those rates do not reflect untold cases of undiagnosed diabetes.
In 11 studies of people with HIV infection,10,11,13,17-24 diabetes incidence ranged from 0.37 per 100 person-years (p-y) in Denmark20 and 0.42 per 100 p-y in the largely European DAD study19 to 4.7 per 100 p-y in a 10-year-old MACS survey10 (Table 2).** This 10-fold range reflects cohort differences in age, diabetes risk factors (especially race and weight), and diabetes definition. The study with the longest follow-up, a 10-year audit from France, charted an overall diabetes prevalence of 1.41 per 100 person-years.22 But that rate plunged from a high of about 2.5 per 100 p-y in 1999-2000, when many people were taking toxic antiretrovirals like indinavir, stavudine, and didanosine, to a low of about 0.5 per 100 p-y in 2005-2006 and 2007-2009.
Except for the nationwide French study,22 diabetes incidence lay below 1 per 100 p-y in European studies13,19-21 and above 1 per 100 p-y in U.S. studies (Table 2).10,17,18,23,24 (In another 1149-person U.S. study that did not lend itself well to the comparisons made here, diabetes incidence measured 3.5 per 100-person-years.25 More than two thirds in this Baltimore cohort were black.) The European cohorts included more than 42,000 HIV-positive people,13,19-21 compared with about 13,000 in the cited U.S. studies. From 7% of 15% of the three big European cohorts were nonwhite,13,19-21 as in the U.S. MACS study.10 But in the other U.S. studies more than half were black and many others Hispanic -- groups with a higher diabetes risk than whites. Body mass index centered around a healthy 23 kg/m2 in the European studies,13,19-21 compared with median BMIs usually in the overweight range in the U.S. studies.
Recent Diabetes Incidence Lower With HIV
Seven of the 11 diabetes incidence studies -- 6 in the United States10,11,17,18,23,24 and 1 in Denmark20 -- compared incidence between the HIV group and an HIV-negative comparison set (Table 3). Four studies found a significant difference in incident diabetes risk between the HIV group and the non-HIV group,10,20,23,24 and in three of those studies having HIV predicted a lower risk of new diabetes.20,23,24 In Denmark the largest of these comparisons found a higher diabetes risk with HIV only in the early combination antiretroviral years (1996-1998),20 when HIV-positive people still took antiretrovirals that warped glucose metabolism. In more recent years (1999-2010), diabetes risk was similar in people with and without HIV (Table 3). When the researchers limited the 1999-2010 analysis to the time before people started antiretroviral therapy, the HIV group had a 55% lower diabetes risk (adjusted IRR 0.45, 95% CI: 0.21 to 0.96).
In contrast to these results, the MACS study found a 4-fold higher diabetes rate in HIV-positive men taking antiretroviral therapy than in HIV-negative men (Table 3).10 But this 1999-2003 study included many men taking outmoded unboosted PIs (including ritonavir and indinavir) and toxic nucleosides (stavudine, zidovudine, and didanosine) that may affect glucose metabolism. The MACS team later determined that every year taking nucleosides of this vintage boosted odds of high insulin 8%.26
|Table 3. Diabetes Incidence in HIV Cohorts Versus Control Cohorts|
|With HIV||Without HIV|
|Author||n, risk factors||Incidence/100 p-y||n, risk factors||Incidence/100 p-y|
|Justman17 1995-1998, WIHS||1435, age 37, 55% black, 26% Hispanic, BMI 25.5*||2.8 on PI|
1.2 on NRTIs only
1.2 not on ART
|350, age 36, 52% black, 30% Hispanic, BMI 26.4*||1.4 (P = 0.06 vs HIV group on PI; no other differences)|
Trend to higher diabetes rate in HIV-positives on PI
|Tien18 2000-2006, WIHS||1524, age 39.2,* 56% black, 28% Hispanic, BMI 26.8,* HCV 30%*||2.5 on PI|
2.89 non-PI ART
3.4 NRTIs only
1.53 no ART
|564, age 34.3,* 58% black, 27% Hispanic, BMI 28.0,* HCV 16%*||1.96 (not significantly different from any HIV group)|
|Brown10 1999-2003, MACS||319†||4.7 on ART|
1.7 not on ART
|361†||1.4 (relative rate 4.11 for HIV group on ART vs non-HIV group)|
Higher diabetes rate in HIV-positives on ART
|Polsky11 2002-2006, Bronx, New York||222, age 50, 61% black, 27% Hispanic, 61% overweight or obese, HCV 42%||5.4% (12 of 222)||155, age 50, 46% black, 34% Hispanic, 69% overweight or obese, HCV 43%||8.4% (8 of 155)|
Lower diabetes rate in HIV-positives
|Rasmussen20 1996-2010, Denmark||3540, age 38.7, 2.5% nonwhite, 18.1% overweight or obese||0.37||14,160, age 38.7||0.387|
For HIV+ vs HIV-:
IRR 1.02 overall
IRR 2.83 in 1996-1998
IRR 0.90 1999-2010
IRR 0.45 1999-2010 before ART begins
Similar or lower diabetes rates in HIV-positives vs negatives after early combination ART years
|Herrin23 1999-2013, Veterans Aging Cohort Study||2891, age 48, 55.4% black, 38% overweight or obese||1.7||7567, age 50, 47.7% black, 72.5% overweight or obese||3.1 (hazard ratio 0.74, P < 0.001, for HIV group vs non-HIV group)|
Lower diabetes rate in HIV-positives
|Tripathi24 1994-2011, South Carolina||6816‡||1.135||6816‡||1.360 (adjusted hazard ratio 0.55 for HIV group on ART vs non-HIV group)|
Lower diabetes rate in HIV-positives
For details on participant demographics and follow-up, see Table 2. ART, antiretroviral therapy; BMI, body mass index (kg/m2); HCV, hepatitis C virus; IRR, incidence rate ratio; MACS, Multicenter AIDS Cohort Study; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; WIHS; Women's Interagency HIV Study.
* Asterisk indicates significant difference by HIV status.
† Risk factors not reported separately for HIV-positive and negative men in the incidence population. See Table 1 for risk factors in prevalence population.
‡ Matched groups.
A Women's Interagency HIV Study (WIHS) analysis discerned a trend toward higher diabetes risk in women on PI therapy than in HIV-negative women (incidence 2.8 versus 1.4 per 100 p-y, P = 0.06).17 But this study dates to the dawn of the combination therapy era (1995 to 1998) when most people took unboosted indinavir, ritonavir, or saquinavir and first-generation nucleosides. In a small Bronx cohort a lower proportion of people with than without HIV had newly diagnosed diabetes (5.4% versus 8.4%), but the authors did not report significance calculations.11 The three studies that calculated lower incident-diabetes risk in people with HIV, or equivalent risk in people with and without HIV, are also the largest and most recent studies that compare incidence by HIV status.20,23,24
If we put these pieces together, it appears that HIV-positive people may have run a greater risk of diabetes in the early days of combination antiretroviral therapy, but that today better care of people with HIV with safer antiretrovirals has reversed that risk balance. HIV metabolic experts like Jacqueline Capeau (University Pierre and Marie Curie, Paris) argue that certain HIV groups may sidestep some common chronic diseases more often than HIV-negatives because they see their clinician much more regularly than people without HIV, because they get screened for comorbidities more often, and because awareness of these mortal diseases runs high in the HIV community.
* Aretaeus was no lightweight. He published two four-volume tomes, one called On the Causes and Indications of Acute and Chronic Diseases, the other On the Treatment of Acute and Chronic Diseases, wherein he described not only diabetes mellitus, but also pleurisy, diphtheria, tetanus, pneumonia, asthma, epilepsy, and how to tell spinal from cerebral paralysis.3
** This analysis includes two DAD cohort studies, which primarily include Europeans. The later DAD analysis19 differs from the earlier analysis13 in excluding people with prior cardiovascular disease and including only those with a complete diabetes risk factor profile. Thus the study population in the later analysis is about half as big as in the first analysis.
- Polonsky KS. The past 200 years of diabetes. N Engl J Med. 2012;367:1332-1340.
- Jolles S. Paul Langerhans. J Clin Pathol. 2002;55:243.
- Encyclopaedia Britannica. Online edition. Aretaeus of Cappadocia.
- American Diabetes Association. History of diabetes.
- International Diabetes Federation: IDF Diabetes Atlas Sixth Edition.
- American Diabetes Association. Statistics about diabetes.
- Narayan KM, Boyle JP, Geiss LS, Saaddine JB, Thompson TJ. Impact of recent increase in incidence on future diabetes burden: U.S., 2005-2050. Diabetes Care. 2006;29:2114-2116.
- Kilby JM, Tabereaux PB. Severe hyperglycemia in an HIV clinic: preexisting versus drug-associated diabetes mellitus. J Acquir Immune Defic Syndr Hum Retrovirol. 1998;17:46-50.
- El-Sadr WM, Mullin CM, Carr A, et al. Effects of HIV disease on lipid, glucose and insulin levels: results from a large antiretroviral-naive cohort. HIV Med. 2005;6:114-121.
- Brown TT, Cole SR, Li X, 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.
- Polsky S, Floris-Moore M, Schoenbaum EE, Klein RS, Arnsten JH, Howard AA. Incident hyperglycaemia among older adults with or at-risk for HIV infection. Antivir Ther. 2011;16:181-188.
- Savès M, Raffi F, Capeau J, 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.
- De Wit S, Sabin CA, Weber R, 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.
- Calza L, Masetti G, Piergentili B, et al. Prevalence of diabetes mellitus, hyperinsulinaemia and metabolic syndrome among 755 adult patients with HIV-1 infection. Int J STD AIDS. 2011;22:43-45.
- Galli L, Salpietro S, Pellicciotta G, et al. Risk of type 2 diabetes among HIV-infected and healthy subjects in Italy. Eur J Epidemiol. 2012;27:657-665.
- National Institute of Diabetes and Digestive and Kidney Diseases. National diabetes statistics, 2014.
- Justman JE, Benning L, Danoff A, 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.
- Tien PC, Schneider MF, Cole SR, et al. Antiretroviral therapy exposure and incidence of diabetes mellitus in the Women's Interagency HIV Study. AIDS. 2007;21:1739-1745.
- Petoumenos K, Worm SW, Fontas E, et al. Predicting the short-term risk of diabetes in HIV-positive patients: the Data Collection on Adverse Events of Anti-HIV Drugs (D:A:D) study. J Int AIDS Soc. 2012;15:17426.
- Rasmussen LD, Mathiesen ER, Kronborg G, Pedersen C, Gerstoft J, Obel N. Risk of diabetes mellitus in persons with and without HIV: a Danish nationwide population-based cohort study. PLoS One. 2012;7:e44575.
- Ledergerber B, Furrer H, Rickenbach M, 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.
- Capeau J, Bouteloup V, Katlama C, et al. Ten-year diabetes incidence in 1046 HIV-infected patients started on a combination antiretroviral treatment. AIDS. 2012;26:303-314.
- Herrin M, Tate J, Freiberg M, et al. Risk of incident diabetes associated with weight gain after cART initiation. 20th Conference on Retroviruses and Opportunistic Infections. March 3-6, 2013. Atlanta. Abstract 804.
- Tripathi A, Liese AD, Jerrell JM, et al. Incidence of diabetes mellitus in a population-based cohort of HIV-infected and non-HIV-infected persons: the impact of clinical and therapeutic factors over time. Diabet Med. 2014;31:1185-1193.
- Mehta SH, Moore RD, Thomas DL, et al. The effect of HAART and HCV infection on the development of hyperglycemia among HIV-infected persons. J Acquir Immune Defic Syndr. 2003;33:577-584.
- Brown TT, Li X, Cole SR, et al. Cumulative exposure to nucleoside analogue reverse transcriptase inhibitors is associated with insulin resistance markers in the Multicenter AIDS Cohort Study. AIDS. 2005;19:1375-1383.