The study series involved 101 treatment naïve patients from the Serbian cohort of HIV infected patients treated at the HIV/AIDS Centre at the Institute for Infectious and Tropical Diseases in Belgrade, who presented between January 1998 and December 2007 with most advanced HIV-induced immunodeficiency. Inclusion criteria consisted of adult HIV infection (stages A-C according to the 1993 Centers for Disease Control case definition criteria) [17], CD4 cell count below 50/µL (“late presenters”), and initiation on HAART after diagnosis. Follow-up was continued until the last visit, death or 31 December 2007. Patients who did not survive the index AIDS defining condition were excluded from the study. Consent for participation was obtained from all, and the study was approved by the Clinical Centre of Serbia Ethics Committee.

Demographic characteristics, AIDS diagnosis prior to HAART initiation, co-infection with hepatitis C virus (HCV), along with parameters associated with response to HAART including plasma viral load and CD4+ cell count as well as the duration of treatment, were all assessed as variables with possible influence on treatment efficacy and survival. Influence of antiretroviral drug classes used as HAART components was also analyzed.

The immunological and virological responses to HAART were evaluated by measuring plasma viral load (pVL) and CD4+ T cell counts. Criteria for the type of response to HAART were as follows: the response was considered favorable in case of achievement of a sustained pVL reduction to undetectable values, along with CD4 cell count increases to above 200/µL. In contrast, treatment failure was defined as a pVL over 2.9 log₁₀ copies of HIV RNA/ml of plasma regardless of immunological improvement. In addition to these two clear-cut types of response to HAART, dissociation between immunological and virological responses to HAART occurred at times, defined as achievement of undetectable viremia during treatment but without a rise in CD4 cell counts to above 200/µL [18, 19]. The threshold of 200 CD4 cells/µL was taken as one decreasing the risk of serious opportunistic infections. However, the results were additionnally analyzed according to a threshold of CD4 > 400 cells/µL, which is one approaching normal values, and has been used in previous analyses of our cohort [19].

CD4 cell counts and viral loads were measured in peripheral blood samples drawn at presentation and every 4-6 months during follow-up. CD4 cells were quantified by flow cytometry. Plasma HIV-1 RNA loads were measured by a quantitative reverse transcriptase polymerase chain reaction (Roche Molecular Systems, Branchburg, NY, USA), which has a lower limit of detection of 50 copies/mL (1.7 log₁₀).

All analyses were performed using an electronic database organized in the SPSS (version 11.5) statistical package. The association between the type of response to HAART, and variables including baseline and end-point CD4 cell counts and pVL, type of HAART regimen, and age above 40, were assessed using the chi-square test and univariate and multivariate logistic regression. Odds ratios (ORs) and the corresponding 95% confidence intervals (CIs) were used as estimators of relative risk for HAART failure, and/or lethal outcome. The Kaplan-Meier product limit method was used to determine the probability of reaching immunological and virological improvement, and survival, as well as to assess the probability of developing drug toxicities over time. One-way analysis of variance (ANOVA) was used to compare means. The level of statistical significance was 0.05.

Baseline characteristics of all patients are presented in Table 1.

Late presenters were more likely to be male (72.3%). Regarding transmission risk factors, the frequency of men having sex with men (MSM), heterosexuals and IV drug users was similar. There was only one patient who got infected vertically.

Treatment characteristics are presented in Table 2. The overall duration of HAART was 2.86 ± 1.80 years (range 1-9), but varied according to regimen. HAART regimens included combinations of two NRTIs as a treatment backbone, with (I) one or two PI (taken by 34.6% of all patients), (II) one NNRTI (35.6%), and (III) combinations of drugs from all three classes or triple NRTI regimens during some period of their treatment (29.7% switched multiple regimens during follow-up). Regimen sequencing was performed in case of virological and/or immunological failure, and also in some instances to simplify treatment by lowering pill burden. Antiretrovirals available in Serbia included zidovudine (AZT), lamivudine (3TC), didanosine (ddI), stavudine (d4T), abacavir (ABV), nevirapine, efavirenz, indinavir, saquinavir, nelfinavir, lopinavir/r, ritonavir(r), and enfuvirtide. Ritonavir was used for PI boosting only. Genotypic resistance testing was not routinely performed; thus, subsequent regimens were composed on the basis of standards of care [20]. However, three extensively pre-treated patients, after being diagnosed with multi drug-classes’ resistant virus, were switched to newer drugs, such as enfuvirtide, darunavir/r, and atazanavir/r (the latter two as a compassionate treatment).

A favorable response to HAART was achieved in 55 (54.5%), a virological / immunological dissociation occurred in 32 (31.7%), while treatment failure was recorded in 14 (13.9%) patients (Table 3A). The mean number of drugs needed to achieve a favorable response was 4.8 ± 2.1 (range 3-10), similar (P > 0.05) to the number of drugs taken by those with HAART failure (5.0 ± 1.1) and a dissociative response (5.0 ± 1.5).

Gender, age over 40, and HIV transmission risk factor were not associated with HAART failure, nor was clinical AIDS at presentation (P > 0.05). Moreover, no specific AIDS-defining condition affected the type of response to HAART (P = 0.33), nor did any specific treatment regimen, whether PI- or NNRTI-based.

Ten (9.9%) patients died during the follow-up period. The most frequent causes of death were AIDS-related conditions including PML, TB, ADC, and disseminated CMV disease in two cases each, while non-Hodgkin lymphoma and end-stage HCV-related liver disease were the causes of death in one patient each. Of those with a fatal outcome, eight patients (80%) experienced HAART failure, and two (20%) had a dissociative response to HAART.

The overall estimated survival was eight years (Table 3B). Analysis of factors associated with survival showed that gender, age over 40, HIV disease stage, and HCV co-infection did not affect survival (P > 0.05), nor did specific AIDS-defining conditions. Survival was also unaffected by treatment regimen, whether PI- or NNRTI-based (P = 0.66). However, the achievement of undetectable viremia regardless of the degree of immune recovery was a predictor of longer survival (Figs. 1,2). For patients with a mean viral load of 1.7 ± 0.23 (1.7 – 1.9), the estimated survival was 8.7 years, significantly longer (P < 0.001) than in those with a mean viral load of 3.6 ± 0.82 (2.9 – 5.2), whose estimated survival was 4.5 years. Achievement of undetectable viremia during treatment appeared life saving (OR = 42.5, 95% CI 7.1 – 251.9, P = 0.000), as was as a CD4 cell count of above 200/μL recorded in 57 (56.4%) patients (OR = 6.4, 95 CI 1.2-31.8, P = 0.023). The estimated survival in late presenters who achieved a rise in the CD4 cell count to above 200/μL was 8.7 years, as opposed to 5.8 years in those whose CD cell count did not recover to this level (P = 0.006) (Fig. 2). Interestingly, a CD4 count rise to over 400 cells/μL provided no extra benefit (P = 0.31, log rank). However, as revealed by multivariate logistic regression, achievement of undetectable viremia was the single independent predictor of survival (OR = 42.5, 95% CI 7.1 – 251.9, P = 0.000).

Fig. (1). Estimated probability of survival of late presenters according to achievement of undetectable viremia. pVL ≤ 2 log10 ___ , pVL ≥ 3log10 ----- , P = 0.009.

Fig. (2). Estimated probability of survival of late presenters according to achievement of CD4 cell coubt above 200/µL during HAART. Achieved ___ , not achieved ----- , P = 0.00.

Immune restoration disease (IRD) developed in 11 (10.9%) patients, at a mean CD4 cell count of 270.4 ± 131.0. Dermatomal herpes zoster was the leading IRD manifestation, appearing in six patients. Other IRD presentations included CMV vitritis, genital herpes, TB, cryptococcal meningitis and a hepatitis flare in an HCV co-infected patient. The metabolic syndrome occurred frequently during treatment of late presenters. Hyper-lipidemia and lipodystrophy were recorded in 58 (7.4%) and 31 (30.6%) patients, respectively. Type 2 diabetes mellitus was recorded in 8 (7.9%). The median time to develop hyperlipidemia and lipodystrophy was 5 and 6 years, respectively. The 9-year probability of developing hyperlipidemia and lipodystrophy was 100%, while it was 30% for type 2 diabetes mellitus.