The HIV epidemic in Eastern Europe and Russia is large and not well-controlled. To describe the more recent molecular epidemiology of HIV-1, transmitted drug resistance, and the relationship between the epidemics in this region, we sequenced the protease and reverse transcriptase genes of HIV-1 from 812 people living with HIV from Ukraine (n = 191), Georgia (n = 201), and Russia (n = 420) before the initiation of antiretroviral therapy. In 190 Ukrainian patients, the integrase gene sequence was also determined. The most reported route of transmission was heterosexual contact, followed by intravenous drug use, and men having sex with men (MSM). Several pre-existing drug resistance mutations were found against non-nucleoside reverse transcriptase inhibitors (RTIs) (n = 103), protease inhibitors (n = 11), and nucleoside analogue RTIs (n = 12), mostly polymorphic mutations or revertants. In the integrase gene, four strains with accessory integrase strand transfer inhibitor mutations were identified. Sub-subtype A6 caused most of the infections (713/812; 87.8%) in all three countries, including in MSM. In contrast to earlier studies, no clear clusters related to the route of transmission were identified, indicating that, within the region, the exchange of viruses among the different risk groups may occur more often than earlier reported.

Predicting the outcome of antiretroviral therapies (ART) for HIV-1 is a pressing clinical challenge, especially when the ART includes drugs with limited effectiveness data. This scarcity of data can arise either due to the introduction of a new drug to the market or due to limited use in clinical settings, resulting in clinical dataset with highly unbalanced therapy representation. To tackle this issue, we introduce a novel joint fusion model, which combines features from a Fully Connected (FC) Neural Network and a Graph Neural Network (GNN) in a multi-modality fashion. Our model uses both tabular data about genetic sequences and a knowledge base derived from Stanford drug-resistance mutation tables, which serve as benchmark references for deducing in-vivo treatment efficacy based on the viral genetic sequence. By leveraging this knowledge base structured as a graph, the GNN component enables our model to adapt to imbalanced data distributions and account for Out-of-Distribution (OoD) drugs. We evaluated these models' robustness against OoD drugs in the test set. Our comprehensive analysis demonstrates that the proposed model consistently outperforms the FC model. These results underscore the advantage of integrating Stanford scores in the model, thereby enhancing its generalizability and robustness, but also extending its utility in contributing in more informed clinical decisions with limited data availability. The source code is available at https://github.com/federicosiciliano/graph-ood-hiv.

Ritonavir-boosted atazanavir is an option for second-line therapy in low- and middle-income countries (LMICs). We analyzed publicly available HIV-1 protease sequences from previously PI-naïve patients with virological failure (VF) following treatment with atazanavir. Overall, 1497 patient sequences were identified, including 740 reported in 27 published studies and 757 from datasets assembled for this analysis. A total of 63% of patients received boosted atazanavir. A total of 38% had non-subtype B viruses. A total of 264 (18%) sequences had a PI drug-resistance mutation (DRM) defined as having a Stanford HIV Drug Resistance Database mutation penalty score. Among sequences with a DRM, nine major DRMs had a prevalence >5%: I50L (34%), M46I (33%), V82A (22%), L90M (19%), I54V (16%), N88S (10%), M46L (8%), V32I (6%), and I84V (6%). Common accessory DRMs were L33F (21%), Q58E (16%), K20T (14%), G73S (12%), L10F (10%), F53L (10%), K43T (9%), and L24I (6%). A novel nonpolymorphic mutation, L89T occurred in 8.4% of non-subtype B, but in only 0.4% of subtype B sequences. The 264 sequences included 3 (1.1%) interpreted as causing high-level, 14 (5.3%) as causing intermediate, and 27 (10.2%) as causing low-level darunavir resistance. Atazanavir selects for nine major and eight accessory DRMs, and one novel nonpolymorphic mutation occurring primarily in non-B sequences. Atazanavir-selected mutations confer low-levels of darunavir cross resistance. Clinical studies, however, are required to determine the optimal boosted PI to use for second-line and potentially later line therapy in LMICs.

The EuResist cohort was established in 2006 with the purpose of developing a clinical decision-support tool predicting the most effective antiretroviral therapy (ART) for persons living with HIV (PLWH), based on their clinical and virological data. Further to continuous extensive data collection from several European countries, the EuResist cohort later widened its activity to the more general area of antiretroviral treatment resistance with a focus on virus evolution. The EuResist cohort has retrospectively enrolled PLWH, both treatment-naïve and treatment-experienced, under clinical follow-up from 1998, in nine national cohorts across Europe and beyond, and this article is an overview of its achievement. A clinically oriented treatment-response prediction system was released and made available online in 2008. Clinical and virological data have been collected from more than one hundred thousand PLWH, allowing for a number of studies on the response to treatment, selection and spread of resistance-associated mutations and the circulation of viral subtypes. Drawing from its interdisciplinary vocation, EuResist will continue to investigate clinical response to antiretroviral treatment against HIV and monitor the development and circulation of HIV drug resistance in clinical settings, along with the development of novel drugs and the introduction of new treatment strategies. The support of artificial intelligence in these activities is essential.

Motivation: In predicting HIV therapy outcomes, a critical clinical question is whether using historical information can enhance predictive capabilities compared with current or latest available data analysis. This study analyses whether historical knowledge, which includes viral mutations detected in all genotypic tests before therapy, their temporal occurrence, and concomitant viral load measurements, can bring improvements. We introduce a method to weigh mutations, considering the previously enumerated factors and the reference mutation-drug Stanford resistance tables. We compare a model encompassing history (H) with one not using this information (NH). Results: The H-model demonstrates superior discriminative ability, with a higher ROC-AUC score (76.34%) than the NH-model (74.98%). Wilcoxon test results confirm significant improvement of predictive accuracy for treatment outcomes through incorporating historical information. The increased performance of the H-model might be attributed to its consideration of latent HIV reservoirs, probably obtained when leveraging historical information. The findings emphasize the importance of temporal dynamics in acquiring mutations. However, our result also shows that prediction accuracy remains relatively high even when no historical information is available. Availability and implementation: This analysis was conducted using the Euresist Integrated DataBase (EIDB). For further validation, we encourage reproducing this study with the latest release of the EIDB, which can be accessed upon request through the Euresist Network.