Risk factors for new-onset diabetes mellitus in patients receiving protease inhibitor therapy

230 1Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, and the Capital Health Region, Edmonton, Alberta; 2Saskatoon Health Region, Saskatoon, Saskatchewan; 3Institute of Health Economics and the Department of Public Health Sciences, University of Alberta; 4Division of Infectious Diseases, Department of Medicine, University of Alberta, Edmonton, Alberta Correspondence: Christine A Hughes, Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, 3126 Dent Pharm, Edmonton, Alberta T6G 2N8. Telephone 780-492-5903, fax 780-492-1217, e-mail christine.hughes@ualberta.ca Received for publication October 6, 2004. Accepted May 5, 2005 CA Hughes, RP Cashin, DT Eurich, S Houston. Risk factors for new-onset diabetes mellitus in patients receiving protease inhibitor therapy. Can J Infect Dis Med Microbiol 2005;16(4):230-232.


METHODS
Using the database, the authors identified 496 subjects initiated on PI therapy who met the inclusion criteria. From this population, subjects with new-onset DM were identified using the database records and confirmed by the medical chart. Diabetes cases were defined as physician documentation of DM in the outpatient medical chart and/or those subjects receiving an antidiabetic agent.
Demographic and clinical characteristics were collected for all subjects using the database records and were described as mean and standard deviation or proportions, as appropriate. The characteristics of the two groups under study were compared with a Student's t test, χ 2 test or Fisher's exact test depending on the nature of the variable.
Logistic regression models were used to examine the relationship between new-onset DM and potential risk factors, including age, sex, ethnicity, weight, alcohol use, hepatitis C virus (HCV) infection, total treatment days with PI therapy, and concurrent use of other agents known to potentially cause hyperglycemia (eg, atypical antipsychotic agents, beta-blockers, corticosteroids, diuretics, pentamidine). Body mass index (BMI) could not be entered into the model because height measurements were not systematically recorded in the database. Crude and adjusted odds ratios (ORs) and 95% CIs were calculated using logistic regression for each variable. In addition, all potentially important clinical interactions were assessed in the logistic regression models; no important or statistically significant interactions (P<0.10) were identified. All statistical analyses were performed using SPSS software (version 11.0, SPSS Inc, USA) for Windows (Microsoft Corporation, USA) using an a priori alpha of 0.05. This study was approved by the University of Alberta Health Sciences Faculties, the Capital Health Authority and the Caritas Health Group Health Research Ethics Board.

RESULTS
The mean age of the PI-treated subjects was 43.4±9.4 years (range 19 to 77) and the mean duration of PI therapy was 3.0±1.9 years (range 0.17 to 7.9). New-onset DM was identified in 18 of the 496 subjects (3.6%) who were initiated on PI therapy. The onset of DM was variable, with a mean onset of 475±301 days (range within two months up to 48 months after PI initiation). The mean age of subjects in the diabetes group was 51.6±9.6 years versus 42.1±9.2 years in the nondiabetes group (Table 1). When compared with the nondiabetes group, subjects in the diabetes group were heavier and had been on PI therapy longer (Table 1). There were no significant differences between the groups in terms of ethnicity; however, there was a higher proportion of Aboriginal subjects in the nondiabetes group, whereas subjects in the diabetes group were more likely to be members of a non-Aboriginal minority. There was a significantly higher number of subjects coinfected with HCV in the nondiabetes group.
A comparison of the demographic characteristics of individuals from the diabetes group and patients excluded from the study due to a prior history of diabetes (n=30) revealed a higher proportion of Aboriginal subjects with a prior history of diabetes (27%) and a smaller proportion of non-Aboriginal minority subjects with a prior diabetes diagnosis (7%). Thirtythree per cent of patients with diabetes before PI therapy initiation were coinfected with HCV.
In the univariate analyses, weight, age, ethnicity and a higher number of total treatment days of PI therapy were significant risk factors for developing DM (Table 2). In the multivariate model, subjects who were older were more likely to develop DM (OR 1.12, 95% CI 1.05 to 1.19; P=0.001). This corresponds to a 12% increased risk of developing DM for each one-year increase in age. In addition, subjects that were heavier had an increased risk of developing DM (OR 1.06, 95% CI 1.03 to 1.10; P=0.001), as did those belonging to a non-Aboriginal minority group (eg, African American, Hispanic, Asian) when compared with Caucasians (OR 6.67, 95% CI 1.56 to 28.41; P=0.01). Of the non-Aboriginal minority patients treated with a PI, 7.7% developed diabetes. A longer duration of PI therapy was also significantly associated with developing DM (OR 1.52, 95% CI 1.07 to 2.17; P=0.02).

DISCUSSION
We found that a higher number of total treatment days of PI therapy was associated with a substantial risk of developing newonset DM. In addition, other independent risk factors for DM included older age, increased body weight and ethnicity (non-Aboriginal minority). Traditional risk factors for the development of type 2 DM include family history, being overweight, an age of at least 40 years, and being a member of a high-risk population (people of Aboriginal, Hispanic, South Asian, Asian or African American descent) (10). Thus, these results support a multifactorial etiology of DM in patients receiving PI therapy, where traditional risk factors play an important role.
The results from the present study are consistent with those from other studies. Palacios et al (11) conducted a casecontrol study of HIV-infected patients who developed DM after starting antiretroviral therapy between January 1997 and December 2001. These authors found that obesity at the start of the follow-up for HIV infection, the duration of treatment with PI and the presence of lipodystrophy were all associated with the development of DM. In a prospective study of approximately 2000 women (both HIV-infected and HIVuninfected), Justman et al (12) found that PI use, older age and being overweight were independent risk factors for DM.
Aboriginal ethnicity was not found to be associated with the development of new-onset DM in the present study. Further analyses revealed that Aboriginal patients were, on average, five years younger than the rest of the population (both Caucasians and other minorities; P<0.001), which may partially explain these results. A higher proportion of Aboriginal patients had diabetes before starting PI therapy. This is not surprising, however, because Aboriginal patients are at a significantly increased risk of developing diabetes when compared with Caucasians (10). In addition, HCV infection was also not found to be associated with DM after adjusting for other risk factors. This is contrary to studies (13,14) that have found that HCV infection is associated with hyperglycemia and DM. However, on average, patients with HCV infection were younger (P<0.01) and had a shorter duration of PI therapy (P<0.01) in the present study. As well, patients with HCV infection had a lower mean body weight, although this did not reach statistical significance (P=0.25).
Although not a specific outcome of the present study, we also examined whether specific PI agents were associated with the development of new-onset DM. After controlling for other risk factors in the multivariate model, there was no significant difference found among the individual PIs. Importantly, however, the majority of subjects (79.3%) received indinavir or nelfinavir as their first PI therapy.
An important limitation of the present study was its retrospective design; data could not be collected for other diabetes risk factors (eg, family history), and the presence or absence of lipodystrophy could not be determined. The calculation of BMI was not possible due to missing height data for some patients; however, weight has also been associated with the development of diabetes. In addition, the case definition of diabetes may have resulted in an underestimate of DM. However, despite this, a longer duration of PI use was associated with an increased risk of developing DM.
The present study suggests that traditional risk factors play an important role in the development of new-onset DM in patients starting PI therapy. Before starting antiretroviral therapy, patients should be screened for diabetes risk factors, including family history. Given the serious, long-term health consequences of diabetes, consideration may be given to starting a non-PI regimen (or possibly atazanavir) for patients at higher risk (15). For patients initiated on PI therapy, a fasting blood glucose should be measured at baseline and regularly during PI treatment. In addition, BMI should be monitored regularly and weight loss recommended for patients who are overweight and at higher risk of developing diabetes.