Background
HIV replication can be measured as viral load in plasma (VL), which is the preferred approved marker for treatment success monitoring. Sustained plasma viral load below the limit of detection (<50 copies/µL) is the goal in treatment strategy of HIV patients. However, in time of successful long-term therapies and upcoming cure strategies, there is need for a further marker to analyse the effects on the proviral reservoir. In order to analyse the dynamics of proviral load (pVL) and evolution of proviruses in patients with viremia and undetectable plasma viral load we are collecting and examining longitudinal collected samples from more than 70 patients on ART. Here we present the updated results of this ongoing analysis.

Materials and methods
Sequential blood samples (at least 3) during a period of 2.5 years were collected from each patient. For pVL measurement, cellular DNA was extracted from PBMCs. Proviral HIV load and PBMC count were measured in a real-time PCR to calculate pVL (log cop/Mio cells). The data were analysed with regard to VL, pVL, CD4 cell count, HIV treatment combinations and also the period of undetectable VL.

Results
Three subgroups were defined depending on the VL: SVL (suppressed viral load, <50 copies/µL), LLV (low level viremia, 50-499 copies/µL) and viremia (>500 copies/µL). The highest pVL was in the LLV subgroup and the lowest in the SVL subgroup.
We saw a slight decline of the pVL over time in all patients on successful treatment. A regression line was calculated from all available pVL values in the observed period for each patient. The slope of the regression line indicates the extent of increasing or decreasing proviral load. The LLV group showed the strongest decline of the proviral load in median. In contrary, in patients with constant suppressed viral load the lowest decline was observed.
There were no significant differences neither in mean nor in the slope of proviral load between the patients with an INI-containing regimen and without INI.
Comparing the course of pVL and CD4 cell counts of the consequent time points we could observe three different groups of trends. One group consisted of patients with similar courses of CD4 count and pVL (concordant ascending/descending curves). Second group was characterized by disconcordant curves, e.g. descending pVL in spite of increasing CD4 counts and vice versa. Third group was built by patients, which could not fit neither of the characteristics of previous groups.

Conclusions
Our analysis showed that a sustained suppressed viral load leads in the long term to the reduction of the proviral reservoir. The decrease of the pVL can serve as a marker for success of the antiretroviral therapy. An increase of the pVL may indicate a minimal virus replication and thus progressing the damage of the immune system. Furthermore, there is need to examine in detail which factors contribute to the dynamics in proviral DNA load in spite of suppressed plasma VL. Proviral load could be an easily to perform and useful readout in routine diagnostics.

Background: Studies on virus evolution are mostly restricted to short fragments of the HIV-1 genome, often located in the fast evolving env or gag genes. Amplification and sequencing of the whole 9.2 kb long genome of the virus remains technically challenging. The infeasibility to reliably link short sequence fragments to individual variants impedes research aimed at characterization of the HIV-1 quasispecies, tracing of immune evasion or in vivo recombination. As we were particularly interested in the study of recombination, we developed and validated a method for amplification of individual viral variants over an as long as possible length. We then assessed the possibility to execute this amplification on single viral RNA molecules and study recombination in patients with dual HIV-1 infection.

Material & Methods: The near full-length HIV-1 genome is covered by 2 overlapping amplicons with a length of respectively 4,511 and 4,627 nucleotides. The efficiency and sensitivity of the amplification was assessed on EDTA blood plasma from 41 patients, sampled early after infection. Subsequently the usefulness, sensitivity and accuracy of the method was evaluated by limiting dilution using a plasma pool of 3 different HIV-1 subtypes. Either viral RNA or in-bulk generated cDNA were diluted to the limit after which multiple replicates of the end point dilution were analyzed. The distribution of the 3 subtypes and the occurrence of in vitro recombination was assessed through construction of phylogenetic trees and highlighter plots. The protocol for limiting dilution single genome sequencing was also applied on samples from 5 patients with HIV-1 dual infection.

Results: Amplification and sequencing of the full HIV-1 genome was successful for 38 of the 41 samples that comprised a broad range of HIV-1 subtypes and a range of viral load between 330 and 10^7 copies/mL. For the plasma pool of 3 HIV-1 subtypes equal sensitivity and accuracy in identifying the 3 subtypes and their distribution was obtained after RNA and cDNA limiting dilution. One evidence of in vitro recombination was demonstrated in 20 sequences obtained after cDNA dilution. No indications for in vitro recombination were found in the 25 sequences obtained after RNA limiting dilution. Analysis of the single genome sequences of samples from 5 patients with dual infection revealed a heterogeneous virus population with co-circulation of at least 3 recombinant forms in 1 patient and clear indications for the presence of multiple recombinant variants in 2 other patients.

Conclusions: This study presents a method for the sequencing of individual near full-length HIV-1 genomes using a 2 amplicon approach and limiting dilution of either viral RNA or cDNA. The possibility to dilute viral RNA to the extreme before reverse transcription was successfully explored as a way to reduce in vitro recombination. Using the method for the analysis of HIV variant composition in patients with dual infection showed co-circulation of multiple recombinant forms.