This study was conducted at the IPEC/FIOCRUZ where care has been provided to HIV/AIDS patients since 1986. A longitudinal observational clinical database has been maintained on patients receiving HIV care at IPEC. Cohort procedures have been described and results published elsewhere.29–31 Briefly, data are updated regularly using outpatient and inpatient clinical documentation and laboratory testing results. Prescription of ARV therapy (drug, dates of use, and dose) is documented by the medical provider and support staff in the clinical records. Trained abstractors record the information onto standardized forms for processing.

For this study, we included data from 1558 antiretroviral (ART)-naïve patients who first received cART between January 1996 and December 2010, with follow-up through August 2011. The IPEC Institutional Review Board has reviewed and approved the study.

Age at HAART initiation was the variable of interest across all analyses. Patients were stratified as 18–29 years and 30–39 years (“younger”), 40–49 years (“older”); 50–59 years and ≥60 years (“elderly”). “Elderly” was defined according to CDC definition for HIV/AIDS patients.32 Other variables used to describe our cohort included demographic, clinical and treatment related characteristics.

HIV exposure categories were presented as: heterosexual (women and men separately); men who have sex with men (MSM); injecting drug users (IDU), and others (not specified). Race was grouped as white and non-white. Schooling was stratified in ≤4 years; 5–8 years; 9–11 years; >11 years. Starting cART while participating in an ART naive clinical trial, baseline CD4+T lymphocyte count (cells/μL), baseline HIV viral load (log10copies/mL) and AIDS-defining disease were also accessed.

cART was defined as two NRTIs in combination with at least one PI or one NNRTI. Patients were grouped according to the year of cART initiation before and after year 2004, when new, less toxic and friendlier ARV options became available. First cART regimens were defined as PI-based regimen (with or without booster), NNRTI-based regimen and others. PI-based regimen that used ritonavir (RTV) as booster and the most frequent first cART regimens were also recorded.

For this study, we have only assessed cART modifications or discontinuations related to toxicity (TOX-MOD) that occurred during the first year after treatment initiation. cART discontinuation related to toxicity was defined as treatment interruption caused by any ARV-related toxicity. cART modification due to toxicity was defined as a toxicity driven substitution of at least one ARV in the regimen. ARV dosage adjustments were not considered as modifications.

The type and date of TOX-MOD were defined as given in the medical chart, and were grouped as follows: hematologic (anemia, thrombocytopenia, leukopenia, pancytopenia), central nervous system (CNS) (neuropsychiatric manifestation, e.g. hallucinations, vertigo, insomnia, nightmares, depression, phobia), peripheral neuropathy (PN), rash, GI (nausea, vomiting, diarrhea), liver (liver enzymes increase, hyperbilirubinemia, jaundice), renal (creatinine clearance decrease, serum creatinine increase, proteinuria, lithiasis, acute renal failure) and metabolic (dyslipidemia and lipodystrophy).

The outcomes of interest were overall toxicity and the most frequent toxicities during the first year after treatment initiation that had led to cART MOD grouped as described above. We estimated the incidence rate and confidence interval (95% CI) of each event stratified by age groups (18–29, 30–39, 40–49, 50–59, ≥60 years) using quasipoisson model and reported it as the number of occurrences per 100 persons-years (PY). Deaths and MOD related to other reasons during the first year of ART were censored at the time of their occurrence. Patients who did not MOD their first ART were censored at one year after cART initiation.

Cox's proportional hazards regression was applied to estimate the HR of overall TOX-MOD during the first year of cART according to the clinical/demographic characteristics. The model was also adjusted for toxicity risk factors previously identified (sex, type of regimen, and year of treatment initiation).5 The proportional hazard assumption was tested by Schoenfeld residuals analysis.

We used the statistical software R, version 2.14.1 (www.r-project.org) for all statistical analyses.

Selected demographic values and treatment characteristics distributed according to the decade of life are summarized in Table 1. A total of 1558 ART naïve patients were included in this analysis. At cART initiation, 957 (61.4%) were younger (<40 years), 420 (27.0%) were older (40–49 years), and 181 (11.6%) were elderly (≥50 years). The median age was 36 years [interquartile range (IQR): 29–43].

Each age category had more male (overall average of 67.2%), and the number of white individuals decreased with age (54.1% for 18–29 years and 30.6% for ≥60 years; p<0.0043). HIV exposure categories significantly fluctuated over age groups (p<0.001). Considering only men, the proportion of MSM was: 48.4%, 39.7%, 41.2%, 19.3% and 18.3% for 18–29 years, 30–39 years, 40–49 years, 50–59 years and ≥60 years, respectively (p<0.0001). Elderly patients had less years of education (≤4 years) than younger and older patients (14.5% for 18–29 years and 51.4% for ≥60 years; p<0.0001). There was no statistically significant difference on ART clinical trial participation among the age groups.

At the time of cART initiation, 430 (27.6%) patients had already presented at least one AIDS-defining disease, and the frequency increased with age, except for patients ≥60 years (p=0.0247). Baseline CD4+T lymphocyte count significant decreased with age (248cells/μL for 18–29 years and 150cells/μL for ≥60 years; p=0.016), while there was no significant difference on baseline HIV viral load among the age groups.

The majority of patients started cART after 2004 (overall average 68.2%); as well as the majority of patients started the first cART with a NNTRI-based regimen (1088; 69.8%) and there were no statistically significant differences among age groups neither for the calendar year nor for the regimen. RTV-boosted PI-based regimen was used, by 56.4% (243/431) of patients and it significantly increased with age (17.0% for 18–29 years and 30.0% for ≥60 years; p<0.0001).

The most frequent first cART regimens stratified by age are depicted in Table 2. A combination of zidovudine (ZDV)+lamivudine (3TC)+efavirenz (EFV) was used by two-fifths of the study population (628, 40.3%). Nevirapine (NVP) was used in 3.8% of cART regimens. Comparing NRTI use along the calendar year, TDF use increased after 2004, while the use of d4T, ddI, and ABC decreased. A continuous increase on ZDV and 3TC use was observed from 1996 to 2010.

Patients were followed for a total of 1369 PY, from ART initiation up to one year of treatment or up to any MOD that occurred within the first year of treatment. A total of 239 (15.3%) events that led to any MOD within the first year of treatment were observed; 228 (95.4%) of these were related to toxicity (TOX-MOD), corresponding to an incidence rate of 16.6 per 100 PY (95% CI: 14.6–18.9). The median time from ART initiation to TOX-MOD during the first year of ART was 1.46 months (IQR: 0.5–4.0). The overall probability of TOX-MOD in the first year of ART was 14.6% (228/1558). Almost half of the patients who presented TOX-MOD were on ZDV+3TC+EFV.

The most frequent toxicity events associated with MOD during the first cART regimen were hematologic (59/228; 26.3%), CNS (47/228; 20.9%), rash (42/228; 19.1%), and GI (38/228; 16.7%). The great majority of the hematologic events were anemia (48; 81.4%), followed by leucopenia (6; 10.3%). For the GI events, the most common were nausea and vomiting (25; 65.8%), gastrointestinal intolerance (7; 18.4%) and diarrhea (4; 10.5%).

Frequency and incidence of TOX-MOD increased with age (Table 3). For younger patients the overall frequency of TOX-MOD was 12.0% and 14.4% (18–29 and 30–39 years, respectively) while for older and elderly patients increases of 2% and 3% per decade of age were observed, respectively. The incidence rate of TOX-MOD for patients aged 18–29 years was 13.4 per 100 PY (95% CI: 10.1–17.6) while for patients aged 50–59 years, and ≥60 years was 19.4 per 100 PY (95% CI: 13.0–29.0) and 22.8 per 100 PY (95% CI: 10.9–47.8), respectively. Stratifying by age groups, both frequency and incidence increased with age for most of toxicities, and this increase was more pronounced from 40 to 49 years and above. In contrast, the incidence rate of TOX-MOD by gastrointestinal events was much higher among patients aged 30–39 years (44.0 per 100 PY (95% CI: 28.4–68.2). Frequency of liver, renal, PN and metabolic toxicities was low in this study.

The results of the multivariate model (Cox's proportional hazards regression) showed that the incidence ratio of TOX-MOD during the first year of cART progressively increased with age, albeit not reaching statistical significance. This profile was maintained after adjusting for sex, cART regimen and year of cART initiation: HR 1.18 (95% CI: 0.82–1.68) for 30–39 years; HR 1.41 (95% CI: 0.97–2.03) for 40–49 years; HR 1.42 (95% CI: 0.87–2.30) for 50–59 years; HR 1.61 (95% CI: 0.73–3.55) for ≥60 years (Table 4). No violation of Schoenfeld proportional hazard assumption was found.