RR:C19 Evidence Scale rating by reviewer:
***************************************
Review: Reddy, et al investigated the effect of ART initiation in hyperacute infection on the total HIV DNA load, the genetic landscape of proviruses that persist on ART for up to 1 year, and on the dynamics and size of the HIV reservoir (the intact HIV proviruses that persist on ART). I commend the authors for focusing their study on HIV subtype C as most cure-related studies are focused on HIV subtype B, and for investigating the reservoir in women, as most studies are centered on HIV in men. The manuscript is well written, which makes it easy to follow, and the findings could be an important contribution to the literature describing the HIV reservoir and the impact of very early intervention. However, there are some major limitations that require more experiments to address the questions presented. The biggest shortcoming is the extremely limited sampling of proviruses in the acute ART-treated donors, with what appears to be a total of only 6 proviruses sequenced across all the donors in this arm at the 1-year timepoint. The authors conclude that the HIV reservoir is 0% after 1 year on ART when treatment is initiated in hyperacute infection, when in fact it is <1 in 6 proviruses or <16%. They also claim that this fraction is less than in people who initiate ART in chronic infection at 14%, which it is not at their level of sampling. The most worrisome issue is that, despite the total HIV DNA load in the early treated donors after 1 year on ART being a median of about 50 HIV DNA copies/million PBMC, and they obtained a median of 1.4M PBMC per donor, they only obtained 6 proviral sequences. They should have easily sequenced >10 proviruses per donor sample, yet they sequenced a total of only 6 across all the donor samples. This outcome suggests that either the primers used for the PCR did not match most of the proviruses in the donor samples and/or that the PCR reaction conditions were extremely inefficient. It is well known that patient-specific primers must be used for PCR-based assays on HIV subtype C due to the high genetic diversity of variants within this subtype. However, the methods show that the same primers were used on all donors, likely leading to the difficulty in obtaining FLIPseq data in the samples on ART. While they did obtain more sequences from the pre-ART samples using the same primers (a total of 108 sequences from all early-treated donors combined), this is still only a small fraction of the total proviruses in the median of 1.4M PBMC sampled from each of the donors (the total would be >1000), indicating that primer mismatch was likely an issue for the pre-ART samples as well. The same limited sampling issue applies to the analysis of the decay dynamics, along with differential treatment regimens in the chronic vs. acute group (with 0% including an INSTI in the chronic group and 67% including an INSTI in the acute group). Due to the apparent technical issues with the application of the FLIPseq assay and the very shallow sampling that resulted, the overall conclusions and main claims are not substantiated by the evidence presented.
Comments:
Abstract- the 0% intact proviruses reported for the early treated group should be changed to <16% (1/the total number of proviruses sequenced which appears to be 6).
Abstract – “no longer detectable” should be changed to “not detectable at our level of sampling”
Results- It is stated that the 1-LTR proviruses are detected with the primers and probe used for total HIV-1 DNA measurements, but it appears that the reverse primer is in gag, not LTR. If so, solo LTRs will not be detected with this method. Please make this clarification.
Figure 1A– It is not clear if the measurements were take prior to ART or on ART for the “Acute Tx” group.
Change, “However, at time of peak viremia, the untreated and the early treated groups did not differ in total proviral DNA load (Figure 1A, bottom panel)” to “However, at time of peak viremia, the untreated and the early treated groups did not significantly differ in total proviral DNA load (Figure 1A, bottom panel)”. The P value is 0.1 which is a trend towards lower DNA levels in the “Acute Tx” group.
“In contrast, treatment initiated during acute infection resulted in both a rapid decrease of plasma viremia so that all participants had undetectable viremia at one-year post-ART (p<0.0001, Figure 1C top panel) and steady decrease of total proviral load over the same time period (p=0.0004) (Figure 1C, bottom panel).” I would change the word “steady” to “slight”. The decrease is modest but still significant overall.
Please include the statistical tests used in the figure legends.
Change “These results indicate that early treatment leads to a measurable decline in proviral DNA during the first year of treatment that is not seen when therapy is initiated during chronic infection” to “These results indicate that early treatment leads to a measurable decline in proviral DNA during the first year of treatment that is not seen at our level of sampling when therapy is initiated during chronic infection.” Previous large studies have shown that HIV DNA declines by ~1-log when ART is initiated in chronic infection.
As in the results, change 0% to <16% for the fraction of intact genomes in the acute treated group.
The following sentence is misleading: “In contrast, there was a more rapid decrease in the proportion of genome-intact viruses following ART initiation in acute infection such that these viruses were no longer detectable at our sampling depth after 1 year of treatment (57% to 0%) (Figure 3C).” The decline is actually not significantly more rapid according to the data in Figure 3. The decline is 34% to <10% at some timepoints on ART in the chronic treated group vs. 57% to <16% in the acute treated group. The text in the results section and throughout should be corrected to indicate that while fewer total proviruses persist in the acute treated group compared to the chronic treated group, the fraction of those that are intact do not appear to be different at your level of sampling. It is curious that only 6 proviruses were sequenced at the 1 year timepoint in the individuals treated in acute infection since it appears that the number of proviruses at this timepoint had a median of about 50 per million PBMC. This outcome suggests that the primers used for the FLIPseq do not match the proviruses that persist on ART. For subtype C sequencing, it is important to use patient-specific primers to avoid this major barrier to this study.
The pipeline used to determine HIV provirus intactness appears to account for only a single defect per genome while it is known that many defective proviruses have more than a single defect. The pie charts should be modified to reflect the fraction of proviruses with each type of defect, even if some proviruses are included in multiple slices.
In figure 2D, it is not clear which timepoints were used to generate the pie charts.
What is the difference between “inferred intact” and “intact” in Figures 2 and 3?
In terms of determining clonal expansion based on proviral sequence identity, when the HIV genetic diversity is very high and the individual has viremia suppressed on ART, it is true that most intact identical proviruses likely result from clonal expansion. However, this statement is not accurate either during untreated infection or when treatment is initiated in acute infection. I would restrict this analysis only to the donors who have viremia suppressed on ART and initiated treatment with high genetic diversity.
Creative Commons Attribution 4.0 International License (CC-BY 4.0)
