Predictors of glycaemic control among HIV-positive veterans with diabetes

Abstract

Antiretroviral therapy has decreased HIV-related mortality. However, the incidence of diabetes as a co-morbidity is increasing as HIV-positive patients age. The purpose of this study was to assess the correlation between markers of HIV-infection and diabetes and to determine the proportion of patients achieving an haemoglobin A1c (HbA1c) goal <7% according to specific antiretroviral therapy regimens and adherence. In this retrospective study, HIV-positive veterans with diabetes from 2007 to 2012 were identified. Patients were required to be on the same antiretroviral therapy and diabetes regimen for ≥3 months. In 56 patients, it was identified that for each unit increase in log₁₀ viral load, HbA1c increased 0.67 units (p = 0.0085). Only 38% of patients prescribed a protease inhibitor–based regimen vs. 56% of patients not on a protease inhibitor–based regimen achieved an HbA1c goal (p = 0.1864). Additionally, patients on an insulin-based regimen and patients that were less adherent were less likely to be at HbA1c goal (p = 0.018 and p = 0.0378, respectively). Patients with higher viral loads and patients that were less adherent to antiretroviral therapy were more likely to have a higher HbA1c demonstrating that poor adherence to antiretroviral therapy leads to poor control of both disease states.

Keywords

HIV AIDS haemoglobin A1c HbA1c diabetes glycaemic control insulin resistance cardiovascular disease antiretroviral therapy medication adherence

Introduction

The treatment of HIV infection with antiretroviral therapy (ART) has been very effective in decreasing HIV RNA viral load, increasing CD4 + T cell counts, and decreasing the incidence of acquired immune deficiency syndrome (AIDS) and AIDS-related conditions.^1,2 Consequently, patients infected with HIV and treated with ART are living longer. With extended life-expectancy, HIV-positive patients have demonstrated an increased risk of cardiovascular-related complications such as diabetes. Accordingly, the Guidelines for the Use of Antiretroviral Agents in HIV-1-Infected Adults and Adolescents include a prominent focus on these non-AIDs-related complications.¹ In general, untreated HIV-positive patients have an increased risk of cardiovascular complications due to increased inflammation and endothelial dysfunction, but regardless of HIV serostatus, the risk of cardiovascular disease and associated conditions also increases with age.^1,3,4 Additionally, some antiretroviral classes have been linked to the development of metabolic syndrome due to adverse effects of lipodystrophy, hypertriglyceridaemia, and insulin resistance.¹ Therefore, with disease progression, aging, and adverse effects of ART, co-morbidities such as diabetes mellitus are more commonly diagnosed in HIV patients.

According to the Data Collection on Adverse Events of Anti-HIV Drugs Cohort Study, (D:A:D Study), the rate of new onset diabetes in HIV-positive patients has been reported to be 5.72 per 1000 person-years follow-up. Additionally, in this prospective observational study of 33,389 HIV-infected patients, the incidence of diabetes increased with cumulative ART exposure.⁵ Data from the Multicenter AIDS Cohort Study (MACS) included HIV-infected and seronegative men. A subset evaluation included 680 men from this sample and found the incidence of diabetes for HIV-negative men, HIV-infected men not receiving ART, and HIV-infected men receiving ART was 1.4, 1.7, and 4.7 per 100 person-years, respectively.⁶ Insulin resistance alone has been noted to occur in 25–30% of HIV-infected patients and has been observed in as many as 60–80% of patients treated with protease inhibitors (PIs).^7,8 Studies have shown that PI-based regimens and specific nucleoside reverse transcriptase inhibitors (NRTIs), didanosine, stavudine, zidovudine, and lamivudine, have been associated with metabolic syndrome, decreased insulin sensitivity, and diabetes.^5,7–9 Substantial amounts of data identifying insulin resistance and associations with specific ART regimens have been published; however, there are fewer data on the correlation between markers of HIV infection and markers of poor glycaemic control. In addition, there are limited data assessing the proportion of this population that is able to reach the American Diabetes Association (ADA) defined goal of a haemoglobin A1c (HbA1c) of <7%.¹⁰ Furthermore, the Primary Care Guidelines for the Management of Persons Infected with HIV: 2013 Update includes a section on metabolic complications and recommends following ADA goals of care for patients with diabetes.¹¹

The primary objective of this study was to identify if there is a correlation between a marker of HIV-infection (HIV RNA viral load) and diabetes control (HbA1c) in a veteran population with both disease states. Additional secondary objectives were to assess if there is a correlation between CD4 count and HbA1c determine the proportion of all subjects achieving an HbA1c goal of <7% according to type of ART and diabetes medications received, and lastly to identify the association between ART adherence and HbA1c outcomes.

Materials and methods

This study was a retrospective chart review conducted at a single-centre, tertiary-care Veterans Affairs Medical Centre and was approved by the associated institutional review board. Subjects were identified by querying hospital databases from 1 January 2007 to 1 January 2012. Eligible subjects were included if they had a diagnosis of HIV-infection or AIDS and diabetes mellitus. Diabetes mellitus was identified based on ICD-9 diagnosis codes or if the patient had received at least one diabetes medication during the study period. HIV-infection or AIDS was similarly identified by ICD-9 diagnosis codes or if the patient had received at least one antiretroviral medication during the study period. Patients were also included if they were at least 18 years of age, receiving primary care at the facility, and receiving stable ART and diabetes medication regimens for at least 3 months. Patients were excluded if they were receiving medications for diabetes or HIV-infection from an outside facility or if they were missing laboratory values for HbA1c or HIV RNA viral load after at least 3 months of therapy for HIV-infection and diabetes.

For subjects meeting the eligibility criteria, demographic, clinical, and laboratory data were collected on standardised case report forms from electronic medical records. The most recent ART and diabetes medications with the start date of the medication regimen were also recorded. The start date was used to determine if a patient had been on the same therapy for both their ART and diabetes medication regimens for at least 3 months. Laboratory values for HbA1c, CD4 count, and HIV RNA viral load were recorded after the patient had received at least 3 months of the same ART and anti-diabetes medication regimens.

Medication adherence was calculated as a percent (%) using a medication possession ratio (MPR) equation. MPR was defined as the sum of the days the medication was supplied to the patient divided by the corresponding number of days within the eligible refill interval.^12,13 MPR was calculated based on refill history of the PI, non-nucleoside reverse transcriptase inhibitor (NNRTI), or integrase inhibitor component of ART. If patients were receiving two or all three of these classes, the order of selection for which medication was used to calculate the MPR was the PI, then the NNRTI, then the integrase inhibitor. The NRTI backbone was not used to calculate adherence, as some patients were on a nucleoside-sparing regimen. Likewise diabetes medications were not used to calculate adherence due to the difficulty in calculating days supply with insulin titration. The number of days medication was supplied was summed over the prior 6 months from the date of HbA1c lab draw and therefore the denominator was 180 days. If a patient was receiving the medication for ≥3 months but <6 months then a 90-day medication supply was used.

Descriptive statistics were used to report baseline characteristics and the percent of HIV-positive patients with concomitant diabetes achieving ADA-defined goal of HbA1c of <7%. For the primary objective, linear regression was used to assess the correlation between HIV RNA viral load and HbA1c. HIV RNA viral load was log-transformed because of non-normal distribution. Further analysis included Chi square test of association to assess HIV RNA viral load (as detectable vs. non-detectable) and HbA1c control (as HbA1c at goal <7% vs. not at goal ≥7%). Linear regression was also used to assess the correlation between CD4 + T cell count and HbA1c. A Chi square test of association was used to assess ability to reach HbA1c goal (<7%) based on which ART and diabetes medication regimens subjects received. For this comparison, ART regimens were categorised as either a PI-based or a non-PI-containing regimen (NNRTI-based or integrase inhibitor-based regimen) and diabetes medication regimens were categorised as insulin-based vs. non-insulin-based. A Chi square test was used to assess the association between medication adherence (defined by MPR ≥95%) and ability to reach HbA1c goal of <7%. JMP (Version 10.0, SAS Corporation, Cary, NC) was used for all statistical analyses.

Results

A total of 97 patients were initially screened; of these, 56 patients met the inclusion criteria for the study. The most common reasons for exclusion from the study were initiation of ART for another disease state such as hepatitis B (17 patients), unstable HIV or diabetes medication regimens (11 patients), receiving medications from an outside facility (8 patients), missing lab results (4 patients), or receiving ART as post-exposure prophylaxis (1 patient). Baseline characteristics of the patients are shown in Table 1. Patients were all men, with a mean age of 59 years, and mostly African American. The average HbA1c was 7.36% (6.85–7.87%) and the median viral load was undetectable (<48 copies/mL) with a range of undetectable to 116,000 copies/mL. Median CD4 + T cell count was 479.5 cells/mm³ with a range of 25–3031 cells/mm³.

Table 1.

Baseline demographics.

	N = 56
Mean age (95% CI), years	59 (57–62)
Men, n (%)	56 (100)
Race
African American, n (%)	45 (80.4)
Caucasian, n (%)	11 (19.6)
Mean HbA1c (95% CI), %	7.36 (6.85–7.87)
Median viral load (25th–75th percentile), copies/mL	<48 (<48–197) Range: <48–116,000
Mean viral load (95% CI), copies/mL	5879 (663–11,095)
Median CD4 + T cell count (25th–75th percentile), cells/mm³	479.5 (287.3–748.5) Range: 25–3031
Mean CD4 + T cell count (95% CI), cells/mm³	560 (442–677)

CI: confidence interval.

Linear regression demonstrated a positive correlation between HIV viral load and HbA1c, whereby for each unit increase in log₁₀ viral load there was an HbA1c increase of 0.67 units (p = 0.0085) as shown in Figure 1. When dichotomising viral load to detectable and undetectable vs. HbA1c <7% or ≥7%, patients with an undetectable viral load were 2.92 times more likely to be at HbA1c goal compared to patients with a detectable viral load (p = 0.0662) as shown in Table 2.

Figure 1.

Log transformed HIV RNA viral load vs. HbA1c.

Table 2.

Predictors in achievement of HbA1c goal.

	% at HbA1c goal	OR (95% CI), p value
Undetectable viral load	20/36 (56%)	2.92 (0.91–9.3), p = 0.0662
Detectable viral load	6/20 (30%)	2.92 (0.91–9.3), p = 0.0662
Non-PI-based ART	15/27 (56%)	1.54 (0.53–4.35), p = 0.1864
PI-based ART	11/29 (38%)	1.54 (0.53–4.35), p = 0.1864
Non-insulin based diabetes therapy	16/25 (64%)	3.70 (1.23–11.11), p = 0.018
Insulin-based diabetes therapy	10/31 (32%)	3.70 (1.23–11.11), p = 0.018
≥95% adherent	14/22 (64%)	3.2 (1.049–9.81), p = 0.0378
<95% adherent	12/34 (35%)	3.2 (1.049–9.81), p = 0.0378

When assessing CD4 + T cell count as a predictor of HbA1c, no correlation was identified (p = 0.3059) as shown in Figure 2. Overall, 46% of patients (26/56) were at the ADA-defined HbA1c goal of <7% (Table 2). In addition, 38% of patients prescribed a PI-based regimen compared to 56% of patients not on a PI-based regimen were able to achieve an HbA1c goal of <7% (p = 0.1864). Similarly, more patients with a non-insulin-based diabetes medication regimen were able to achieve an HbA1c goal of <7% (64%) compared to patients on an insulin-based regimen (32%), OR 3.70 (95% CI 1.23–11.11). Antiretroviral adherence was also associated with achievement of HbA1c goals, as 64% of patients in the ≥95% adherent group were able to achieve an HbA1c goal of <7% compared to 35% in the less adherent group (<95%) (OR 3.20, 95% CI 1.05–9.81).

Figure 2.

CD4 count vs. HbA1c.

Discussion

This study identified important findings related to HIV-infection and diabetes control. We identified a significant correlation between HIV RNA viral load and HbA1c in patients concomitantly being treated for HIV-infection and diabetes. We found that for each unit increase in log transformed HIV RNA viral, there was an HbA1c increase of 0.67 units. Use of log transformed HIV RNA viral load for such an analysis is consistent with previous literature.¹⁴ Further analysis indicated that patients with an undetectable viral load were more likely to be at HbA1c goal, although this was not statistically significant, likely due to the small number of patients included in this study. Together these findings suggest that poor control of HIV-infection may be associated with poor control of diabetes. These findings are similar to a study by Monroe et al.,¹⁴ which utilised the Johns Hopkins HIV Clinical Cohort and included 70 patients with HIV-infection and diabetes. In their study, they demonstrated that for each log₁₀ increase in HIV RNA there was an HbA1c increase of 0.47 units (p < 0.001).

In contrast, literature suggests that patients who are adherent to ART regimens have low viral loads and high CD4 counts may have higher HbA1c due to the metabolic side effects of some ART medications. This effect has been shown previously when using CD4 + T cell counts as a correlate to diabetes in HIV-positive patients. Analysis of data from 315 HIV-positive women demonstrated that increasing CD4 + T cell count was associated with a 0.23 unit increase in log-transformed HbA1c (95% CI 0.05–0.40, p = 0.011). Additionally, a detectable HIV RNA viral load was associated with a trend toward higher glucose-adjusted log HbA1c values in these patients.⁴ This suggests that better control of HIV, both from a virologic and immune function standpoint, could be associated with worsening of diabetes control, with an increase in HbA1c. Likewise, another study investigating nucleoside analogue-sparing regimens (raltegravir, etravirine, and maraviroc or darunavir/ritonavir) demonstrated that a positive change in CD4 count was associated with a non-significant increase in fasting glucose.¹⁵ In the present study, we did not find a correlation between CD4 count and HbA1c. Echoing this, the recent guidelines for primary care of HIV patients state that there is a lower frequency of adverse effects such as dyslipidaemia and diabetes with newer ART agents.¹¹

Our study analysed adherence to ART by calculating an MPR based on refill history of the patient’s PI, NNRTI, or integrase inhibitor as these are typically the “base” of an ART regimen. We found the odds of having an HbA1c at goal in the ≥95% adherent group is 3.2 times that of the <95% adherent group. Therefore, patients who were more adherent to their ART regimen were more likely to be at an HbA1c goal of <7%. This finding suggests patients who have poor adherence to ART likely have poor adherence to their diabetes medication regimen as well and, therefore, have associated lack of control of both disease states.

Overall, only 46% of patients were at the ADA HbA1c goal of <7%. This is similar to other studies that have shown approximately 50–54% of patients with HIV-infection and diabetes are able to meet this HbA1c goal.^16,17 When assessing specific medication classes, we were unable to detect differences between individual agents due to the small sample size. We did identify that patients on a PI-based regimen were less likely to be at HbA1c goal vs. patients not on a PI, however this finding was not significant. Other studies have correlated PI-based regimens with poor glycaemic control.^18,19 Mechanistically, studies have shown PIs, specifically indinavir, blocks glucose transport in insulin-secreting cells causing peripheral insulin resistance.^7,20 A study by Han et al.²¹ showed that HIV-infected patients initiating diabetes therapy had smaller absolute mean reductions in HbA1c as compared to a HIV-negative patient cohort. In a subgroup analysis, they showed that patients on a PI-based regimen, compared to a non-PI-based regimen, had significantly smaller benefit in HbA1c lowering, attributed to insulin resistance with PIs.

An unexpected finding of our study was that patients on an insulin-based regimen were significantly less likely to be at HbA1c goal. Patients not receiving insulin were 3.7 times more likely to be at HbA1c goal. There are several possible explanations for this finding. Extended ART use has been associated with insulin resistance specifically with PI-based regimens; however, appropriate dose titration of insulin therapy would be expected to overcome this insulin resistance.^7,8 Although insulin doses were not collected in our study, these individuals had higher HbA1c values and may not have had adequate follow-up for insulin titration. Additionally, this finding could be the result of failing oral hypoglycaemic agents with increasing blood glucose and HbA1c and therefore the need for these individuals to be switched to an insulin-based regimen.

There are several potential limitations to our study. This study was a small retrospective analysis of 56 patients and was underpowered to detect statistically significant results for some of the secondary endpoints. Given our study design, we were unable to control for other possible confounding variables. With glycaemic control, it is possible that other factors such as close follow-up with primary care physicians, diet, exercise, other medications, time since diabetes diagnosis, and other disease states could impact the results. Since fasting blood glucose or HbA1c are typically measured only annually in non-diabetic HIV-positive patients, it was not possible for us to include a non-diabetic HIV-infected control group. This could possibly be done in a future prospective study. We also excluded some patients that were not on stable therapy for 3 months, which decreased the sample size and potentially limits generalisability of the study population. Since we did not collect data on excluded patients (per our IRB-approved protocol) we could not compare them to the patients that were included in the study. However, requiring at least 3 months of stable therapy was to more accurately assess if specific ART or diabetes medication regimens and adherence correlated with glycaemic control. As HbA1c measures an average blood glucose over a 3-month period, we felt that to assess ART and diabetes medications for achievement of HbA1c goal, patients should be on the same therapy for a minimum of 3 months. Regarding medication adherence, we only assessed adherence of the PI, NNRTI, or integrase inhibitor. This required us to make the assumption that if a patient is adherent to one medication in their ART regimen, then the patient was adherent to all medications, which may not always be objectively supported. We did not assess adherence to diabetes medications due to the variable nature of pharmacotherapy regimens. Our findings, however, do support that patients adherent to ART have better control of their diabetes, which correlated to better control of their viral load. We also utilised ≥95% MPR to define adherence, which is consistent with ART adherence in other studies.⁶ Lastly, we utilised <7% as the definition of glycaemic control as suggested by the ADA. The ADA does suggest less intensive goals in certain populations (i.e. limited life expectancy or patients with severe hypoglycaemia), however at this time the ADA does not differentiate goals between HIV-positive and HIV-negative individuals with Type II diabetes.¹⁰ Other studies that have identified HIV-positive patients with diabetes used a similar goal of <7%.^16,17

Conclusions

Overall, we found that patients with higher viral loads were more likely to have a higher HbA1c. Additionally, patients that were less adherent to ART were more likely to have a higher HbA1c. This leads us to believe that poor adherence to ART could indicate patients had decreased adherence to all medications including diabetes medications, and therefore poor control of both disease states. This finding furthermore supports that ART does not necessarily lead to poor glycaemic control as long as patients are adherent to medications for both disease states. Continued education should be given to providers on HbA1c goals as we are only achieving this in 46% of HIV-infected patients with diabetes.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

AIDSinfo. Guidelines for the use of antiretroviral agents in HIV-1-infected adults and adolescents. February 2013, http://aidsinfo.nih.gov/guidelines (accessed 15 May 2013).

Boyd

. Management of HIV-infection in treatment-naive patients: a review of the most current recommendations. Am J Health Syst Pharm 2011; 68: 991–1001.

Samaras

. The burden of diabetes and hyperlipidemia in treated HIV-infection and approaches for cardiometabolic care. Curr HIV/AIDS Rep 2012; 9: 206–217.

Marshall

Hoover

Shi

. Glycated hemoglobin in diabetic women with and without HIV-infection: data from the women’s interagency HIV study. Antivir Ther 2012; 15: 571–577.

De Wit

Sabin

Worm

. Incidence and risk factors for new-onset diabetes in HIV-infected patients. Diabetes Care 2008; 31: 1224–1229.

Brown

Cole

. Antiretroviral therapy and the prevalence and incidence of diabetes mellitus in the multicenter AIDS cohort study. Arch Int Med 2005; 165: 1179–1184.

Walli

Herfort

Michl

. Treatment with protease inhibitors associated with peripheral insulin resistance and impaired oral glucose tolerance in HIV-1-infected patients. AIDS 1998; 12: F167–F173.

Kim

Woods

Georgoff

. A1c underestimated glycemia in HIV-infection. Diabetes Care 2009; 32: 1591–1593.

Hardy

Esch

Morse

. Glucose disorders associated with HIV and its drug therapy. Ann Pharmacother 2001; 35: 343–351.

10.

American Diabetes Association. Standards of Medical Care in Diabetes-2014. Diabetes Care 2014; 37: S14–S80.

11.

Aberg

Gallant

Ghanem

. Primary care guidelines for the management of persons infected with HIV: 2013 update by the HIV medicine association of the infectious diseases society of America. Clin Infect Dis 2014; 58: e1–e34.

12.

Chu L, Kawatkar A and Gi A. A SAS® Macro program to calculate medication adherence rate for single and multiple medication use. Presented at: October 12–14, WUSS 2011 in San Francisco, CA. Cary, NC: SAS Institute Inc.

13.

Leslie

Gwadry-Sridhar

Thiebaud

. Calculating medication compliance, adherence and persistence in administrative pharmacy claims databases. Pharma Program 2008; 1: 13–19.

14.

Monroe

Chander

Moore

. Control of medical comorbidities in individuals with HIV. J Acquir Immune Defic Syndr 2011; 58: 458–462.

15.

Bigoloni

Gianotti

Spagnuolo

. Long-term glucose tolerance in highly experienced HIV-infected patients receiving nucleoside analogue sparing regimens. AIDS 2012; 26: 1837–1840.

16.

Bury

Stroup

Stephens

. Achieving American Diabetes Association goals in HIV-seropositive patients with diabetes mellitus. Proc (Baylor University. Medical Center) 2007; 20: 118–123.

17.

Adeyemi

Vibhaker

Max

. Are we meeting the American Diabetes Association goals for HIV-infected patients with diabetes mellitus? Clin Infect Dis 2009; 49: 799–802.

18.

Koster

Remedi

Qiu

. HIV protease inhibitors acutely impair glucose-stimulated insulin release. Diabetes 2003; 52: 1687–1700.

19.

Samaras

Wand

Law

. Prevalence of metabolic syndrome in HIV-infected patients receiving highly active antiretroviral therapy using International Diabetes Foundation and Adult Treatment Panel III criteria: associations with insulin resistance, disturbed body fat compartmentalization, elevated C-reactive protein, and [corrected] hypoadiponectinemia. Diabetes Care 2007; 30: 113–119.

20.

Murata

Hruz

Mueckler

. Indinavir inhibits the glucose transporter isoform Glut4 at physiologic concentrations. AIDS 2002; 16: 859–863.

21.

Han

Crane

Bellamy

. HIV-infection and glycemic response to newly initiated diabetic medical therapy. AIDS 2012; 26: 2089–2095.