How we measure success matters when evaluating HIV therapies

From Dr. Dan Reeves, Vaccine and Infectious Disease Division

The human immunodeficiency virus (HIV) is a retrovirus that integrates into the human genome. This intact, integrated genome is referred to as a provirus and represents a latent or dormant form of the virus that can reactivate to make infectious virus. Antiretroviral therapy (ART) is extremely successful at blocking HIV replication and results in undetectable levels and restricted transmission of HIV. Yet, even after decades of ART, proviruses still persist. This latent reservoir poses a challenge when ART is interrupted, being that even a few cells with provirus can now make infectious virus that grows and eventually can be transmitted from person-to-person to cause disease. While researchers are developing alternative therapies to target HIV provirus, another important aspect of this challenge is accurately measuring the burden of intact provirus in patient samples to evaluate the success of provirus-targeting therapies. With collaborators from Fred Hutch, UCSF, and Rockefeller University, Dr. Daniel Reeves, a staff scientist in the Schiffer Group of the Vaccine & Infectious Disease Division at Fred Hutch and an Affiliate Assistant Professor in the Department of Global Health at UW, developed a mathematical model to estimate misclassifications of current PCR-based assays for enhanced accuracy of quantifying intact HIV provirus. Their results were recently published in Nature Communications.

There are several methods used to measure HIV provirus abundance in cells. One key aspect is that only intact, complete HIV genomes reactivate to make progeny virus; partial, defective HIV DNA cannot make progeny virus. In order to accurately determine the level of active provirus, a balance between assay sensitivity and specificity must be met. The gold standard for quantifying proviruses that can reactivate is to use a quantitative viral outgrowth assay (QVOA), which measures the number of cells that produce virus when cultured outside the individual in the absence of ART. Though less specific, a more sensitive approach is to quantify HIV DNA within patient cells using PCR. The intact proviral DNA assay (IPDA) uses digital droplet PCR (ddPCR) with two probes targeted to very conserved regions of the HIV genome. Yet, “a small percentage of HIV proviruses may be intact at PCR probe locations, but defective elsewhere in the genome,” stated Dr. Reeves. Therefore, this method could in theory, occasionally misclassify incomplete HIV DNA as intact since only a short area of intact sequence is needed for probe binding and amplification (depicted below). A more arduous method can also be used. This other approach combines four PCR probes and near-full length genome sequencing (Q4PCR) for enhanced specificity for intact provirus, but due to the sequencing arm of this assay, there is a loss of sensitivity for the gain in specificity due to the increase requirement of template sequence for sequencing as compared to PCR amplification. In addition to finding the balance between specificity and sensitivity of these assays, longitudinal studies have suggested the presence of longer-lived HIV provirus pools in individuals with long-term ART usage, introducing a challenge for projecting the decay of these intact provirus reservoirs over time.

The intact proviral DNA assay (IPDA) uses a PCR-based approach with two probes specific for the HIV provirus. This method may misclassify defective or partial HIV DNA as intact provirus due to limited coverage of the provirus sequence.
The intact proviral DNA assay (IPDA) uses a PCR-based approach with two probes specific for the HIV provirus. This method may misclassify defective or partial HIV DNA as intact provirus due to limited coverage of the provirus sequence. Image provided by Dr. Reeves

To understand the sensitivity and accuracy limitations of the PCR alone and PCR combined with near-full length genome sequencing techniques—IPDA and Q4PCR—researchers performed these assays on the same samples from a 10-year, longitudinal cohort of ten men living with HIV who were taking ART. While the IPDA assay detected ~230 intact provirus and ~2,500 defective viral genomes per million T cells, the Q4PCR detected 6 intact and 87 defective. These findings aligned with previous findings that the IPDA method is more sensitive. The researchers next compared the ratio of intact versus total HIV provirus DNA for both methods. Intriguingly, this ratio was similar between the two methods for early time points but diverged for later time points. While one could argue that only the more sensitive IPDA assay is able to detect very small populations of stable long-lived intact provirus while the Q4PCR assay cannot, another possibility exists. The researchers predicted that IPDA was misclassifying defective genomes as intact and generated a mathematical model to help evaluate this possibility and how misclassification might impact the half-life calculations of intact provirus. This model “showed that ‘observed intact sequences’ could appear to decay more slowly over time if truly intact sequences continue to decay and slower decaying misclassified defective proviruses increasingly predominate,” explained Dr. Reeves. This model predicted that ~5% of intact provirus detected by IPDA are misclassified and this finding is consistent with sequence data of intact versus defective provirus.

The researchers wanted to determine the effect of misclassification on modeled therapeutic outcomes to reduce or eliminate the intact, dormant HIV provirus reservoirs. By modeling different scenarios, they observed continual underestimation of therapy efficacy when misclassification of intact provirus was not included in the model. In one modeled scenario, when 8% of intact provirus were misclassified, the therapy had an estimated 10-fold reduction of intact provirus while the true efficacy was a 100-fold reduction when misclassified intact provirus were included in the model. From additional modeling scenarios, Dr. Reeves commented that “the presence of a stable long-lived population does not eliminate the impact of misclassification in therapeutic settings.”

“This work increases the ability of HIV mathematical models to properly describe experimental data and thus enhances their utility for any future projections,” stated Dr. Reeves. “It also furthers the complicated story of how best to study HIV during long-term suppressive ART and/or analyze results from an HIV cure trial that is at least partially successful in reducing reservoirs. We really want to make sure we document HIV reservoir reductions precisely and specifically such that any modality that is partially working could be improved or combined with other therapies. Assays are in continual development, including from our paper collaborators (Dr Gaebler and Dr Cohn) and I also want to highlight the beautiful work from the labs of Hutch colleagues Dr Florian Hladik, Dr Dara Lehman, and Dr Keith Jerome. Their work all pushes forward the HIV science that we need to get closer to cures.” 

The spotlighted research was funded by the National Institutes of Health, the UW/ Fred Hutch Center for AIDS Research, the Einstein-Rockefeller-CUNY Center for AIDS research, the Robertson Fund, the HJH-Foundation, the Robert S. Wennett fund, the Shapiro-Silverberg Fund for the Advancement of Translational Research, the National Center for Advancing Translational Sciences, Delaney AIDS Research Enterprise (DARE) Collaboratory, REACH: Research Enterprise to Advance a Cure for HIV, and the Howard Hughes Medical Institute.


Reeves DB, Gaebler C, Oliveira TY, Peluso MJ, Schiffer JT, Cohn LB, Deeks SG, Nussenzweig MC. 2023. Impact of misclassified defective proviruses on HIV reservoir measurements. Nat Commun. 14(1):4186.