Prediction of HIV sensitivity to monoclonal antibodies using aminoacid sequences and deep learning

Susceptibility Screening of HIV-1 Viruses to Broadly Neutralizing Antibodies, Teropavimab and Zinlirvimab, in People With HIV-1 Suppressed by Antiretroviral Therapy

BACKGROUND

HIV envelope (env) diversity may result in resistance to broadly neutralizing antibodies (bNAbs). Assessment of genotypic or phenotypic susceptibility to antiretroviral treatment is often performed in people with HIV-1 (PWH) and used for clinical trial screening for HIV-1 bNAb susceptibility. Optimal bNAb susceptibility screening methods are not yet clear.

METHODS

Phenotypic and genotypic analyses were conducted on 124 screening samples from a phase 1b study of bNAbs teropavimab (3BNC117-LS) and zinlirvimab (10-1074-LS) administered with lenacapavir in virally suppressed PWH. Phenotypic analysis was conducted on integrated HIV-1 provirus and stimulated outgrowth virus, with susceptibility to bNAbs defined as 90% inhibitory concentration ≤2 μg/mL. The proviral DNA HIV env gene was genotyped using deep sequencing, and bNAb susceptibility predicted using published env amino acid signatures.

RESULTS

Proviral phenotypic results were reported for 109 of 124 samples; 75% (82/109) were susceptible to teropavimab, 65% (71/109) to zinlirvimab, and 50% (55/109) to both bNAbs. Phenotypic susceptibility of outgrowth viruses was available for 39 samples; 56% (22/39) were susceptible to teropavimab, and 64% (25/39) to zinlirvimab. Phenotypic susceptibilities correlated between these methods: teropavimab r = 0.82 ( P < 0.0001); zinlirvimab r = 0.77 ( P < 0.0001). Sixty-seven samples had genotypic and phenotypic data. Proviral genotypic signatures predicted proviral phenotypic susceptibility with high positive predictive value (68%-86% teropavimab; 63%-90% zinlirvimab).

CONCLUSIONS

bNAb susceptibility was correlated among all 3 in vitro assays used to determine teropavimab and zinlirvimab susceptibility in virally suppressed PWH. These findings may help refine PWH selection criteria for eligibility for future studies.

Learning patterns of HIV-1 resistance to broadly neutralizing antibodies with reduced subtype bias using multi-task learning

The ability to predict HIV-1 resistance to broadly neutralizing antibodies (bnAbs) will increase bnAb therapeutic benefits. Machine learning is a powerful approach for such prediction. One challenge is that some HIV-1 subtypes in currently available training datasets are underrepresented, which likely affects models' generalizability across subtypes. A second challenge is that combinations of bnAbs are required to avoid the inevitable resistance to a single bnAb, and computationally determining optimal combinations of bnAbs is an unsolved problem. Recently, machine learning models trained using resistance outcomes for multiple antibodies at once, a strategy called multi-task learning (MTL), have been shown to improve predictions. We develop a new model and show that, beyond the boost in performance, MTL also helps address the previous two challenges. Specifically, we demonstrate empirically that MTL can mitigate bias from underrepresented subtypes, and that MTL allows the model to learn patterns of co-resistance to combinations of antibodies, thus providing tools to predict antibodies' epitopes and to potentially select optimal bnAb combinations. Our analyses, publicly available at https://github.com/iaime/LBUM, can be adapted to other infectious diseases that are treated with antibody therapy.

Predicting neutralization susceptibility to combination HIV-1 monoclonal broadly neutralizing antibody regimens

Combination monoclonal broadly neutralizing antibodies (bnAbs) are currently being developed for preventing HIV-1 acquisition. Recent work has focused on predicting in vitro neutralization potency of both individual bnAbs and combination regimens against HIV-1 pseudoviruses using Env sequence features. To predict in vitro combination regimen neutralization potency against a given HIV-1 pseudovirus, previous approaches have applied mathematical models to combine individual-bnAb neutralization and have predicted this combined neutralization value; we call this the combine-then-predict (CP) approach. However, prediction performance for some individual bnAbs has exceeded that for the combination, leading to another possibility: combining the individual-bnAb predicted values and using these to predict combination regimen neutralization; we call this the predict-then-combine (PC) approach. We explore both approaches in both simulated data and data from the Los Alamos National Laboratory's Compile, Neutralize, and Tally NAb Panels repository. The CP approach is superior to the PC approach when the neutralization outcome of interest is binary (e.g., neutralization susceptibility, defined as inhibitory 80% concentration < 1 μg/mL). For continuous outcomes, the CP approach performs nearly as well as the PC approach when the individual-bnAb prediction algorithms have strong performance, and is superior to the PC approach when the individual-bnAb prediction algorithms have poor performance. This knowledge may be used when building prediction models for novel antibody combinations in the absence of in vitro neutralization data for the antibody combination; this, in turn, will aid in the evaluation and down-selection of these antibody combinations into prevention efficacy trials.

Broadly neutralizing antibodies for HIV prevention: a comprehensive review and future perspectives

SUMMARYThe human immunodeficiency virus (HIV) epidemic remains a formidable global health concern, with 39 million people living with the virus and 1.3 million new infections reported in 2022. Despite anti-retroviral therapy's effectiveness in pre-exposure prophylaxis, its global adoption is limited. Broadly neutralizing antibodies (bNAbs) offer an alternative strategy for HIV prevention through passive immunization. Historically, passive immunization has been efficacious in the treatment of various diseases ranging from oncology to infectious diseases. Early clinical trials suggest bNAbs are safe, tolerable, and capable of reducing HIV RNA levels. Although challenges such as bNAb resistance have been noted in phase I trials, ongoing research aims to assess the additive or synergistic benefits of combining multiple bNAbs. Researchers are exploring bispecific and trispecific antibodies, and fragment crystallizable region modifications to augment antibody efficacy and half-life. Moreover, the potential of other antibody isotypes like IgG3 and IgA is under investigation. While promising, the application of bNAbs faces economic and logistical barriers. High manufacturing costs, particularly in resource-limited settings, and logistical challenges like cold-chain requirements pose obstacles. Preliminary studies suggest cost-effectiveness, although this is contingent on various factors like efficacy and distribution. Technological advancements and strategic partnerships may mitigate some challenges, but issues like molecular aggregation remain. The World Health Organization has provided preferred product characteristics for bNAbs, focusing on optimizing their efficacy, safety, and accessibility. The integration of bNAbs in HIV prophylaxis necessitates a multi-faceted approach, considering economic, logistical, and scientific variables. This review comprehensively covers the historical context, current advancements, and future avenues of bNAbs in HIV prevention.

Predicting neutralization susceptibility to combination HIV-1 monoclonal broadly neutralizing antibody regimens

Combination monoclonal broadly neutralizing antibodies (bnAbs) are currently being developed for preventing HIV-1 infection. Recent work has focused on predicting in vitro neutralization potency of both individual bnAbs and combination regimens against HIV-1 pseudoviruses using Env sequence features. To predict in vitro combination regimen neutralization potency against a given HIV-1 pseudovirus, previous approaches have applied mathematical models to combine individual-bnAb neutralization and have predicted this combined neutralization value; we call this the combine-then-predict (CP) approach. However, prediction performance for some individual bnAbs has exceeded that for the combination, leading to another possibility: combining the individual-bnAb predicted values and using these to predict combination regimen neutralization; we call this the predict-then-combine (PC) approach. We explore both approaches in both simulated data and data from the Los Alamos National Laboratory's Compile, Neutralize, and Tally NAb Panels repository. The CP approach is superior to the PC approach when the neutralization outcome of interest is binary (e.g., neutralization susceptibility, defined as inhibitory concentration < 1 μg/mL. For continuous outcomes, the CP approach performs at least as well as the PC approach, and is superior to the PC approach when the individual-bnAb prediction algorithms have poor performance. This knowledge may be used when building prediction models for novel antibody combinations in the absence of in vitro neutralization data for the antibody combination; this, in turn, will aid in the evaluation and down-selection of these antibody combinations into prevention efficacy trials.

Predicted resistance to broadly neutralizing antibodies (bnAbs) and associated HIV-1 envelope characteristics among seroconverting adults in Botswana

We used HIV-1C sequences to predict (in silico) resistance to 33 known broadly neutralizing antibodies (bnAbs) and evaluate the different HIV-1 Env characteristics that may affect virus neutralization. We analyzed proviral sequences from adults with documented HIV-1 seroconversion (N = 140) in Botswana (2013-2018). HIV-1 env sequences were used to predict bnAb resistance using bNAb-ReP, to determine the number of potential N-linked glycosylation sites (PNGS) and evaluate Env variable region characteristics (VC). We also assessed the presence of signature mutations that may affect bnAb sensitivity in vitro. We observe varied results for predicted bnAb resistance among our cohort. 3BNC117 showed high predicted resistance (72%) compared to intermediate levels of resistance to VRC01 (57%). We predict low resistance to PGDM100 and 10-1074 and no resistance to 4E10. No difference was observed in the frequency of PNGS by bNAb susceptibility patterns except for higher number of PNGs in V3 bnAb resistant strains. Associations of VC were observed for V1, V4 and V5 loop length and net charge. We also observed few mutations that have been reported to confer bnAb resistance in vitro. Our results support use of sequence data and machine learning tools to predict the best bnAbs to use within populations.

Predicted broadly neutralizing antibody (bnAb) resistance and associated envelope characteristics of adults with HIV-1 seroconversion in Botswana

We used HIV-1C sequences to predict (in silico) resistance to 33 known broadly neutralizing antibodies (bNAbs) and evaluate the different HIV-1 env characteristics that may affect virus neutralization. We analyzed proviral sequences from adults with documented HIV-1 seroconversion (N=140) in Botswana (2013-2018). HIV-1 env sequences were used to predict bnAb resistance using bNAb-ReP, to determine the number of potential N-linked glycosylation sites (PNGS) and evaluate env variable region characteristics (VC). We also assessed the presence of signature mutations that may affect bnAb sensitivity in vitro. We observe varied results for predicted bnAb resistance among our cohort. 3BNC117 showed high predicted resistance (72%) compared to intermediate levels of resistance to VRC01 (57%). We predict low resistance to PGDM100 and 10-1074 and no resistance to 4E10. No difference was observed in the frequency of PNGS by bNAb susceptibility patterns except for higher number of PNGs in V3 bnAb resistant strains. Associations of VC were observed for V1, V4 and V5 loop length and net charge. We also observed few mutations that have been reported to confer bnAb resistance in vitro. Our results support use of sequence data and machine learning tools to predict the best bnAbs to use within populations.

Development of screening assays for use of broadly neutralizing antibodies in people with HIV

PURPOSE OF REVIEW

Treatment with combinations of complementary broadly neutralizing antibodies (bnAbs) has increased the proportion of participants for whom bnAbs can maintain virus suppression upon cessation of antiretroviral therapy (ART). There remains, however, a population of trial participants who experience virus rebound despite high plasma concentrations of bnAbs. Thus, baseline resistance remains a critical barrier to the efficacy of bnAbs for use in the treatment and cure of HIV, and the development of a screening assay to guide bnAb selection is a high priority.

RECENT FINDINGS

There are two conceptual approaches to assess the putative rebound-competent HIV-1 reservoir for bnAb sensitivity: to assess neutralization sensitivity of reactivated virus in outgrowth assays and sequence-based approaches that include a selection for intact genomes and assessment of known resistance mutations within the env gene. Currently, the only phenotypic assay for bnAb screening that is clinical laboratory improvement amendments certified (CLIA certified) and available for clinical trial use is Monogram Biosciences' PhenoSense HIV Neutralizing Antibody Assay.

SUMMARY

Several new approaches for screening are currently under development and future screening methods must address three issues. First, complete sampling of the reservoir may be impossible, and determination of the relevance of partial sampling is needed. Second, multiple lines of evidence indicate that in vitro neutralization measures are at least one correlate of in vivo bnAb activity that should be included in screening, but more research is needed on how to use in vitro neutralization assays and other measures of antibody functions and measures of other antibody features. Third, the feasibility of screening assays must be a priority. A feasible, predictive bnAb screening assay will remain relevant until a time when bnAb combinations are substantially more broad and potent.

The applications of machine learning in HIV neutralizing antibodies research-A systematic review

Machine learning algorithms play an essential role in bioinformatics and allow exploring the vast and noisy biological data in unrivaled ways. This paper is a systematic review of the applications of machine learning in the study of HIV neutralizing antibodies. This significant and vast research domain can pave the way to novel treatments and to a vaccine. We selected the relevant papers by investigating the available literature from the Web of Science and PubMed databases in the last decade. The computational methods are applied in neutralization potency prediction, neutralization span prediction against multiple viral strains, antibody-virus binding sites detection, enhanced antibodies design, and the study of the antibody-induced immune response. These methods are viewed from multiple angles spanning data processing, model description, feature selection, evaluation, and sometimes paper comparisons. The algorithms are diverse and include supervised, unsupervised, and generative types. Both classical machine learning and modern deep learning were taken into account. The review ends with our ideas regarding future research directions and challenges.