The HIV patient profile in 2013 and 2003: Results from the Greek AMACS cohort

Combined Antiretroviral therapy (cART) has improved life-expectancy of people living with HIV (PLHIV) but as they age, prevalence of chronic non-AIDS related comorbidities may increase. We study the evolution of HIV-disease markers and comorbidities’ prevalence in PLHIV in Greece. Two cross-sectional analyses (2003 and 2013) on data from the AMACS cohort were performed. Comparisons were based on population average models and were repeated for subjects under follow-up at both 2003 and 2013. 2,403 PLHIV were identified in 2003 and 4,910 in 2013 (1,730 contributing for both cross-sections). Individuals in 2013 were on average older, diagnosed/treated for HIV for longer, more likely to be on cART, virologically suppressed, and with higher CD4 counts. Chronic kidney disease, dyslipidemia and hypertension prevalence increased over time. There was an increase in prescription of lipid-lowering treatment (3.5% in 2003 vs. 7.7% 2013, p<0.001). Among 220 and 879 individuals eligible for Framingham 10-year Event Risk calculation, the proportion of patients in the high-risk group (>20%) increased from 18.2% to 22.2% (p = 0.002). Increase in the prevalence of comorbidities was more pronounced in the subset of patients who were followed in both 2003 and 2013. The increased availability and uptake of cART led to significant improvements in the immuno-virological status of PLHIV in Greece, but they aged alongside an increase in prevalence of non-AIDS related comorbidities. These results highlight the need for appropriate monitoring, optimal cART selection and long-term management and prevention strategies for such comorbidities.


Introduction
The introduction of combined antiretroviral therapy (cART) has produced significant improvements in the outcomes of people living with HIV (PLHIV), with significant gains in individuals' survival and reduced HIV related morbidity [1][2][3][4]. Mainly due to cART efficacy, the number of PLHIV as well as their age has increased during the last decades [5][6][7]. Moreover, a substantial proportion of individuals acquire HIV at older ages [8], further contributing to the growing number of PLHIV living into their 50s and 60s [9].
Patients' aging brings new challenges to the infected individuals themselves, physicians and national health providers. Several studies indicate that the duration of HIV infection is associated with various comorbidities, including cardiovascular diseases (CVD), hypertension, type-II diabetes mellitus, chronic kidney disease (CKD), osteopenia/osteoporosis, and non-AIDS malignancies (e.g. lung, anal, liver cancer, Hodgkin lymphoma). The majority of such comorbidities are more prevalent and tend to appear at an earlier age when compared to the general population [10][11][12][13][14][15][16]. Besides the infection itself, cART has been related to increased prevalence of dyslipidemia, diabetes mellitus, and insulin resistance [17,18]. Other potential contributing factors may be the increased prevalence of known risk factors, immune dysfunction and chronic HIV-related immune activation [19][20][21].
In light of these findings, it is essential to study the HIV epidemic's evolution, in terms of both risk factors and subsequent comorbidity changes. The present study aims to assess how the evolution of the HIV epidemic during a decade impacts the prevalence of comorbidities and risk factors for those comorbidities by comparing results from two cross-sectional analyses of patients enrolled and followed-up in Athens Multicenter AIDS Cohort Study (AMACS) in 2003 and 2013. The study also focuses on the aging cohort of individuals who were alive and under follow-up in both 2003 and 2013. By analysing changes over 10 years in this subgroup, not only the effect of ageing but also the long-term consequences of HIV infection and of treatment can be investigated. Focus is given to the most common comorbidities (such as CKD, CVD, type-II diabetes mellitus, dyslipidemia and hypertension), related risk factors but also demographics, treatment and disease markers (i.e. CD4 cell count and HIV-RNA viral load). The results of the study may prove useful in the description of the current needs in HIV care with the specificities of the national HIV care landscape and those of people aging with the HIV infection.

Data source: The Athens Multicenter AIDS Cohort Study (AMACS)
Data were derived from the AMACS electronic database. AMACS is a collaborative, population-based cohort study initiated in 1996. Currently, 14 out of 16 clinics that follow PLHIV in Greece, participate in the study. All HIV infected patients seen in one of the collaborating clinics who were either diagnosed before and alive on 1/1/1996 or diagnosed after 1/1/1996 are eligible for inclusion in AMACS, provided they have been followed for at least one year, or are still under follow-up. Patients who died within the first year of prospective follow-up are not excluded from the cohort. In accordance with data protection policy, data are provided by the clinics after de-identification. The study has been approved by the Athens University IRB (http://en.uoa.gr/), the HCIDC IRB (http://www.keelpno.gr/en-us/home.aspx) and the National Organization of Medicines (http://www.eof.gr/web/guest/home).
A standardized protocol is used for data collection. A variety of data are recorded at study entry including demographic characteristics, clinical events and deaths, antiretroviral therapy (including reasons for change and adverse events) and laboratory tests. Data are updated yearly and thoroughly checked for errors and inconsistencies. To date (last data merger January, 2015), data on 7575 subjects (demographic characteristics, clinical events and deaths, antiretroviral therapy, laboratory tests and resistance tests) have been recorded and checked for errors.

Definitions and formulas
The 10-year risk of CVD was estimated using the Framingham Risk Score Calculator for Coronary Heart Disease [22] whereas the estimated glomerular filtration rate (eGFR), used as a renal function index, was evaluated through the CKD-EPI formula [23]. CKD was defined as having eGFR<60 ml/min/1.73m 2 whereas eGFR values between 60 and 89 were considered as indicative of mildly reduced kidney function. Diabetes was defined, as fasting glucose ! 126 mg/dL, non-fasting glucose > 200 mg/dL, or taking antidiabetic drugs or insulin. Dyslipidemia was defined as elevated total cholesterol !240 mg/dl, and/or decreased HDL-cholesterol 35mg/dl, and/or elevated triglycerides !150 mg/dl, and/or increased LDL-cholesterol!150 mg/dl, and or total cholesterol/HDL ratio>4, and/or use of a lipid-lowering medication. Hypertension was defined as systolic blood pressure !140 mmHg and/or diastolic blood pressure !90 mmHg or taking antihypertensive drugs.

Statistical analysis
Two cross sectional descriptive analyses were performed at two points in time: 2003 and 2013. Individuals, aged !18 years at HIV diagnosis, alive and under active follow-up (having at least one clinical visit or treatment initiation/change or any laboratory test/measurement) during 2003 and/or 2013 were eligible for this study. Distributions of categorical variables were summarized through absolute and relative (%) frequencies or graphically through bar plots. For continuous variables, medians and interquartile ranges (IQR) were used. Box-plots were used to graphically summarize the distributions of continuous variables. Comparisons between 2003 and 2013 were performed mainly through population average (marginal) models for repeated measurements, fitted through generalized estimating equations (GEE), in order to take into account the potential correlations between measurements on the same individuals taken in both 2003 and 2013. For unordered categorical variables with more than two levels, ordered categorical variables and non-normally distributed continuous variables multinomial logistic, ordinal logistic and median regression models (with adjustments for repeated measurements on the same individual) were used, respectively. All analyses were repeated on the subset of individuals contributing data to both 2003 and 2013 cross-sections. We use the terms a) "open cohort" and b) "closed cohort" to refer to a) the full sample of all eligible patients contributing data to either one or both cross sections and b) the subset of patients who contributed data to both cross-sections, respectively. Analyses of the "open cohort" data were repeated after excluding individuals infected through intravenous drug use. P-values lower than 0.05 were considered statistically significant. All analyses were performed using Stata 14.2 (Stata Corp., TX USA).

Results
Out of 2,908 AMACS participants who were diagnosed in or before 2003, 2,403 (82.6%) were eligible for inclusion in the 2003 cross-section. The respective number for pre-2013 diagnoses was 7,360, of whom 4,910 (66.7%) were included in the 2013 cross-section. The two cross-sections included in total 5,583 individuals (open cohort) of whom 1,730 (31.0%) contributed data for both years (closed cohort). Within this 10-year period there were 3,180 new patients entering the cohort, whereas 673 of those included in the 2003 sample were not included in the 2013 cross-section (267 died before 2013 and 406 were lost to follow up). Compared to those who remained on follow-up, patients who were lost to follow-up were less likely to be Caucasians, slightly younger and were treated for shorter periods whereas those who died were significantly older, more likely to be heterosexually infected or IDUs, with higher prevalence of AIDS or comorbidities. The most common causes of death were related to cardiovascular diseases (26%), AIDS related conditions (15.7%), non-AIDS malignancies (12.8%) and other non-AIDS related infections (9.8%). Both groups had worse CD4 and viral load levels. The selection process for the two cross-sectional samples is presented graphically in Fig 1.

Open cohort characteristics and outcomes
Demographic characteristics, disease markers and outcomes of the open cohort are shown in Table 1A. In 2013, patients were older by approximately 3 years on average, compared to those in the 2003 sample. The proportion of those aged over 50 years increased from 18.2% to 26.9% (p<0.001). The proportion of men was slightly higher in 2013, 85.4% vs. 82.1% (p<0.001). The increase in the proportion of those infected through intravenous drug use (IVDU) was substantial (from 2.7% to 10.1%; p<0.001). Individuals in the 2013 sample were diagnosed and treated for HIV for a longer duration compared to those in 2003 (6.7 vs. 6.0 and 4.5 vs. 3.8 years on median, respectively; p<0.001 for both), but the proportion of those who had already progressed to clinical AIDS was lower (13.0% vs. 16 (Fig 2A). The proportion of individuals with a previous cardiovascular event (myocardial infarction or stroke) was 1.7% in the 2003 sample and increased to 2.1% 2013 (p<0.001; Table 1A).
The combined effect of these changes on the risk of CVD resulted in a non-statistically significant reduction in the median Framingham risk score (FRS) from 9.    Table 1A and Fig 3A). It should be noted though, that data for FRS calculation in 2003 and 2013 were available for 220 and 879 individuals, respectively. Serum creatinine levels seemed stable [Median-IQR 0.9 (0.8, 1.0) mg/dL in both years; Fig  2A], eGFR were only slightly higher in 2013 [Median-IQR 102.2 (87.9, 112.3) vs. 101.9 (87.8, 113.4) ml/min/1.73m 2 in 2013 and 2003, respectively; Fig 3A] but the prevalence of CKD (defined as eGFR<60 ml/min/1.73m 2 ) increased significantly (p = 0.006) from 2.4% in 2003 to 3.4% in 2013 (Table 1A and Fig 4A).
Finally, prevalence of all comorbidities was increasing with age and in most cases (with the exception of diabetes), it was slightly higher in all age groups in 2013 compared to the same age groups in 2003 (Fig 5A) Excluding IVDUs from both samples resulted in an increase of 2 years in the median age, slightly higher prevalence of dyslipidemia and hypertension (72.0% and 36.2%, respectively) and better immunologic and virologic outcomes (CD4>500 cells/μL: 68.6%; viral load <50 copies/mL: 77.1%) in the 2013 sample with all the other measured parameters having negligible changes compared to those of the main analysis.

Closed cohort characteristics and outcomes
Most of the changes observed in the open cohort and described above were of the same direction, yet of increased magnitude in the aging cohort of patients who contributed data for both cross sections (alive and under follow-up in both 2003 and 2013).
As shown in Table 1B (Table 1B and Fig 3B). Significant (p<0.001) increases were also observed in the proportion of patients with mildly reduced kidney function (i.e. with eGFR between 60 and 89 ml/min/1.73m 2 ).

Discussion
In this study we used data from a large multicenter observational database in order to investigate changes, within the decade 2003 to 2013, in the profile of PLHIV who were diagnosed and linked to care in Greece.
Our results confirmed the increase in the average age of PLHIV seen in clinics as approximately one over four out of the 2013 sample was more than 50 years old and the proportion was one over two in the subset of those who were under follow-up in 2003 and kept on being under follow-up in 2013. Sex between men was the most frequent mode of transmission in both years but the proportion of those infected through intravenous drug use was almost quadrupled, most probably due to the 2011-2012 HIV outbreak in IVDUs in Athens [24]. It is noteworthy though that increased The AMACS population aged alongside an increase in prevalence of non-HIV related chronic comorbidities, such as hypertension, dyslipidemia and CKD despite the high proportion of new patients who entered the cohort over time. These increases were particularly evident for the aging cohort of patients who were captured in the data cohorts in both years (2003, 2013), a sample which reflects better the HIV population moving forward as it not impacted by the entry of new patients into the cohort. Thus, in the closed cohort over the 10 year period an increase in all studied comorbidities was observed, with younger age groups (those under 40 in 2003) particularly contributing to the increase in dyslipidemia and hypertension and older age groups to increase in diabetes and CKD. This outcome was observed despite the potentially improved disease prognosis a potential survivor bias might be expected to produce. The open cohort patients managed to keep lipid and blood pressure levels relatively stable through medical management. It should be noted though, that the proportions of patients who were on anti-lipid or anti-hypertensive treatment seem low. A degree of underreporting in non-ART drugs cannot be excluded but given the established benefits of such treatments we expect to observe increased administration in post-2013 data. Smoking prevalence had not decreased, which in combination with the increased prevalence of type-II diabetes mellitus and aging itself led to an increase in the proportion of patients with high (!20%) 10-year risk for coronary heart disease as estimated by the Framingham risk score. This was notable in the subgroup of the aging subgroup, where this proportion reached approximately 40% in 2013. Additionally, a non-negligible proportion of the study participants, especially among those in the closed cohort, had already experienced a myocardial infarction or a stroke. It is worth mentioning that the corresponding data reported in 2013 are cumulated with the events recorded in 2003 thus the differences in the closed cohort reflect events which occurred between 2003 and 2013. Finally, we also found increases in the proportions of those with at least mild to moderate loss of kidney function indicating a possible increase in the prevalence of chronic kidney disease. However, the drop in median eGFR levels observed in the closed cohort is not far from what would be expected due to the aging of the participants alone [25].
These findings are in general consistent with those reported by other researchers in similar settings but comparative studies between different time points are rare. for European participants but participants in the START trial were younger, with much shorter durations of HIV infection compared to the participants of our study and ART-naive at the time point considered in the specific study.
Our study is characterized by a rich data source: AMACS includes approximately 70% of the population of diagnosed PLHIV in Greece and is, to a large degree, representative of that population [35]. The long follow-up of AMACS allowed us to compare the profile of the average HIV patient in two time points ten years apart, but also to assess the changes within an aging subcohort of individuals seen at the collaborating clinics continuously from at least 2003 to 2013. The data we used come directly from the patients' records at each collaborating HIV clinic or the respective hospital databases and are recorded and merged in a central electronic database.
One of the strong assets our study is the ability to highlight both the current needs of the cohort in total but also the individual needs of those aging with HIV who are additionally affected by longer duration of the infection and prolonged ART exposure.
There are some limitations in our study, stemming mainly from the lack of some additional types of data or the presence of missing values. For example, FRS calculation was only possible for a subset of the study participants due to the lack of blood pressure measurements or HDLcholesterol measurements especially in the 2003 sample. The use of FRS, instead of other HIV-specific formulas [36] to quantify CVD risk has been a subject of debate but some recent studies argue that the Framingham equations may still be preferable [37,38]. Bone mass density measurements were very rare in practice; thus, we were not able to assess rates of osteopenia/ osteoporosis and their respective changes over time. Due to the cross-sectional nature of the study, distinction between the effects of aging and duration of the infection was also not possible. However, we also analysed the changes over a 10 years period in patients who were in care during 2003 and 2013. Interestingly, not only the major findings from the open (cross-sectional) cohort were confirmed in this subgroup, but also diabetes prevalence and Framingham risk score increased significantly, reflecting not only the effect of aging, but also the long-term consequences of HIV infection and treatment. The improvement in the efficacy and tolerability of antiretroviral treatment has clearly changed the prognosis of PLHIV thus the focus of HIV clinicians has partly shifted towards other, non-AIDS but probably HIV related conditions. Although continuous efforts are being made in AMACS to collect complete and reliable data, this shift may have some effect on our results as it is probably associated with a higher attention to test for and record such conditions during the more recent years.
Our results highlight the impressive immunological and virological improvements induced by the modern antiretroviral treatment but at the same time underline the increased frequency of serious and potentially life threatening comorbidities and risks. As some of the comorbidities observed in our cohort may be associated with treatment, cART choices become very important as PLHIV live longer, and thus accumulate greater long-term exposure.
The shift in HIV epidemic paradigm should be addressed with appropriate monitoring and holistic management of HIV care, in terms of optimal cART selection, and long-term management and prevention of comorbidities potentially leading to a continuous improvement of the overall health status of PLHIV. In this context, effective cART options that balance HIV outcomes with less long-term impact on cardiovascular and renal diseases would be beneficial to patient's care and treatment.