Gottlieb et al.[1] suggest that the significance of the lower viral load in HIV-2-infected patients compared with HIV-1-infected patients could be biased by different distributions of CD4 cell counts and that we failed to adjust for CD4 cell count and other potential confounding factors like age, sex and the duration of infection. In fact, our analysis of DNA levels was stratified by CD4 cell counts (fig. 1 in reference [2]) and showed lower levels of DNA viral load in HIV-2 compared with HIV-1 in the higher CD4 strata (≥300 cells/μl), but not in the lower CD4 stratum (<300 cells/μl). To complete this analysis, we have performed a multivariate linear regression analysis including age, sex and HIV RNA within each CD4 cell count stratum (Table 1). After adjustment for age the difference of HIV DNA between the two types of HIV persisted in the higher strata (≥300 cells/μl), but it was only at the limit of significance in both strata after adjustment for age and sex. The relationship between viral type and DNA load was not significant any more after adjustment for HIV RNA, suggesting that the difference was mainly explained by the relation between viral replication and amount of HIV DNA. Therefore, at a given level of replication, no difference in the DNA level is found between HIV-1 and HIV-2. These results were also confirmed when we took into account the difference in sensitivity of our PCR method by imputing DNA values of 5 copies/μg for HIV-1 and 10 copies/μg for HIV-2 when the sample was below the respective detection limits.Table 1: Multivariate linear regression analysis of factors associated with HIV DNA: ANRS CO5 HIV-2 cohort.Given the low detection rate of HIV-2 RNA in plasma, Gottlieb et al.[1] also questioned whether our population of HIV-2-infected patients was representative. Studies in Africa generally overrepresent patients with advanced HIV disease and detectable plasma viral load, as in the study by Gottlieb et al.[1] in Senegal, taken as an example. We recently published viro-immunological follow-up data among a subset of incident HIV-2 infected patients from the French ANRS CO5 HIV-2 cohort compared with a matched HIV-1-infected patient incident group. The HIV RNA load was detectable in 86% and 15% of HIV-1-patients and HIV-2-patients, respectively, with 4.11 log10 copies/ml for HIV-1 and 2.09 log10 copies/ml for HIV-2, in full agreement with the values reported in this HIV DNA load comparative study involving prevalent cases [3]. Furthermore, the 29% overall rate of detectable HIV-2 RNA in prevalent cases of the French HIV-2 cohort, with the limit of detection of 100 copies/ml, ranges from 8% for a CD4 cell count greater than 500 cells/μl, to 35% when in the range 300–500 cells/μl, and 62% when below 300 cells/μl. However, the main conflict between our data and those of Gottlieb et al.[1] concerns the level of HIV-1 DNA. Gottlieb et al.[1] report median HIV DNA of 1.45 log10/μg which is lower than that reported in other studies quantifying HIV-1 DNA in antiretroviral-naïve patients, using the same quantification method (Amplicor HIV-1 Monitor, Roche Molecular Systems, Inc., Branchburg, New Jersey, USA) [4–6]. Indeed, these studies report median values between 2.0 and 2.6 log10 of total HIV-1 DNA/μg. Assuming that the amount of target DNA initially present in the PCR tube follows Poisson's law, the claimed sensitivity of the Monitor test (1 copy/μg) is controversial as it is highly unlikely to detect 1 copy/PCR in 100% of cases [7]. If a sample with 1 copy/μg is systematically detectable, then the calibration standard is underestimated. Moreover, HIV-1 strains circulating in Senegal are highly divergent [8] and sometimes difficult to quantify with primers and probes corresponding to less highly conserved genome regions than those selected for our PCR method [9]. Our work shows that significant differences in total DNA levels exist between HIV-1 and HIV-2 and that they are highly related to viral replication and CD4+ cell count as expected. These differences should be considered as one of the many factors contributing to the clinical and epidemiological differences between these two infections.
Read full abstract