Characterization of Molecular Cluster Detection and Evaluation of Cluster Investigation Criteria Using Machine Learning Methods and Statewide Surveillance Data in Washington State

Molecular Epidemiology of HIV-1 Subtype B Infection across Florida Reveals Few Large Superclusters with Metropolitan Origin

Florida is considered an epicenter of HIV in the United States. The U.S. federal plan for Ending the HIV Epidemic (EHE) within 10 years prioritizes seven of Florida's 67 counties for intervention. We applied molecular epidemiology methods to characterize the HIV infection networks in the state and infer whether the results support the EHE. HIV sequences (N = 34,446) and associated clinical/demographic metadata of diagnosed people with HIV (PWH), during 2007 to 2017, were retrieved from the Florida Department of Health. HIV genetic networks were investigated using MicrobeTrace. Associates of clustering were identified through boosted logistic regression. Assortative trait mixing was also assessed. Bayesian phylogeographic methods were applied to evaluate evidence of imported HIV-1 lineages and illustrate spatiotemporal flows within Florida. We identified nine large clusters spanning all seven EHE counties but little evidence of external introductions, suggesting-in the absence of undersampling-an epidemic that evolved independently from the rest of the country or other external influences. Clusters were highly assortative by geography. Most of the sampled infections (82%) did not cluster with others in the state using standard molecular surveillance methods despite satisfactory sequence sampling in the state. The odds of being unclustered were higher among PWH in rural regions, and depending on demographics. A significant number of unclustered sequences were observed in counties omitted from EHE. The large number of missing sequence links may impact timely detection of emerging transmission clusters and ultimately hinder the success of EHE in Florida. Molecular epidemiology may help better understand infection dynamics at the population level and underlying disparities in disease transmission among subpopulations; however, there is also a continuous need to conduct ethical discussions to avoid possible harm of advanced methodologies to vulnerable groups, especially in the context of HIV stigmatization. IMPORTANCE The large number of missing phylogenetic linkages in rural Florida counties and among women and Black persons with HIV may impact timely detection of ongoing and emerging transmission clusters and ultimately hinder the success of epidemic elimination goals in Florida.

Optimized phylogenetic clustering of HIV-1 sequence data for public health applications

Clusters of genetically similar infections suggest rapid transmission and may indicate priorities for public health action or reveal underlying epidemiological processes. However, clusters often require user-defined thresholds and are sensitive to non-epidemiological factors, such as non-random sampling. Consequently the ideal threshold for public health applications varies substantially across settings. Here, we show a method which selects optimal thresholds for phylogenetic (subset tree) clustering based on population. We evaluated this method on HIV-1 pol datasets (n = 14, 221 sequences) from four sites in USA (Tennessee, Washington), Canada (Northern Alberta) and China (Beijing). Clusters were defined by tips descending from an ancestral node (with a minimum bootstrap support of 95%) through a series of branches, each with a length below a given threshold. Next, we used pplacer to graft new cases to the fixed tree by maximum likelihood. We evaluated the effect of varying branch-length thresholds on cluster growth as a count outcome by fitting two Poisson regression models: a null model that predicts growth from cluster size, and an alternative model that includes mean collection date as an additional covariate. The alternative model was favoured by AIC across most thresholds, with optimal (greatest difference in AIC) thresholds ranging 0.007-0.013 across sites. The range of optimal thresholds was more variable when re-sampling 80% of the data by location (IQR 0.008 - 0.016, n = 100 replicates). Our results use prospective phylogenetic cluster growth and suggest that there is more variation in effective thresholds for public health than those typically used in clustering studies.

HIV Cluster and Outbreak Detection and Response: The Science and Experience

The Respond pillar of the Ending the HIV Epidemic in the U.S. initiative, which consists of activities also known as cluster and outbreak detection and response, offers a framework to guide tailored implementation of proven HIV prevention strategies where transmission is occurring most rapidly. Cluster and outbreak response involves understanding the networks in which rapid transmission is occurring; linking people in the network to essential services; and identifying and addressing gaps in programs and services such as testing, HIV and other medical care, pre-exposure prophylaxis, and syringe services programs. This article reviews the experience gained through 30 HIV cluster and outbreak responses in North America during 2000-2020 to describe approaches for implementing these core response strategies. Numerous jurisdictions that have implemented these response strategies have demonstrated success in improving outcomes related to HIV care and viral suppression, testing, use of prevention services, and reductions in transmission or new diagnoses. Efforts to address important gaps in service delivery revealed by cluster and outbreak detection and response can strengthen prevention efforts broadly through multidisciplinary, multisector collaboration. In this way, the Respond pillar embodies the collaborative, data-guided approach that is critical to the overall success of the Ending the HIV Epidemic in the U.S. initiative.

Incorporating metadata in HIV transmission network reconstruction: A machine learning feasibility assessment

HIV molecular epidemiology estimates the transmission patterns from clustering genetically similar viruses. The process involves connecting genetically similar genotyped viral sequences in the network implying epidemiological transmissions. This technique relies on genotype data which is collected only from HIV diagnosed and in-care populations and leaves many persons with HIV (PWH) who have no access to consistent care out of the tracking process. We use machine learning algorithms to learn the non-linear correlation patterns between patient metadata and transmissions between HIV-positive cases. This enables us to expand the transmission network reconstruction beyond the molecular network. We employed multiple commonly used supervised classification algorithms to analyze the San Diego Primary Infection Resource Consortium (PIRC) cohort dataset, consisting of genotypes and nearly 80 additional non-genetic features. First, we trained classification models to determine genetically unrelated individuals from related ones. Our results show that random forest and decision tree achieved over 80% in accuracy, precision, recall, and F1-score by only using a subset of meta-features including age, birth sex, sexual orientation, race, transmission category, estimated date of infection, and first viral load date besides genetic data. Additionally, both algorithms achieved approximately 80% sensitivity and specificity. The Area Under Curve (AUC) is reported 97% and 94% for random forest and decision tree classifiers respectively. Next, we extended the models to identify clusters of similar viral sequences. Support vector machine demonstrated one order of magnitude improvement in accuracy of assigning the sequences to the correct cluster compared to dummy uniform random classifier. These results confirm that metadata carries important information about the dynamics of HIV transmission as embedded in transmission clusters. Hence, novel computational approaches are needed to apply the non-trivial knowledge collected from inter-individual genetic information to metadata from PWH in order to expand the estimated transmissions. We note that feature extraction alone will not be effective in identifying patterns of transmission and will result in random clustering of the data, but its utilization in conjunction with genetic data and the right algorithm can contribute to the expansion of the reconstructed network beyond individuals with genetic data.

The Role of Phylogenetics in Unravelling Patterns of HIV Transmission towards Epidemic Control: The Quebec Experience (2002-2020)

Phylogenetics has been advanced as a structural framework to infer evolving trends in the regional spread of HIV-1 and guide public health interventions. In Quebec, molecular network analyses tracked HIV transmission dynamics from 2002-2020 using MEGA10-Neighbour-joining, HIV-TRACE, and MicrobeTrace methodologies. Phylogenetics revealed three patterns of viral spread among Men having Sex with Men (MSM, n = 5024) and heterosexuals (HET, n = 1345) harbouring subtype B epidemics as well as B and non-B subtype epidemics (n = 1848) introduced through migration. Notably, half of new subtype B infections amongst MSM and HET segregating as solitary transmissions or small cluster networks (2-5 members) declined by 70% from 2006-2020, concomitant to advances in treatment-as-prevention. Nonetheless, subtype B epidemic control amongst MSM was thwarted by the ongoing genesis and expansion of super-spreader large cluster variants leading to micro-epidemics, averaging 49 members/cluster at the end of 2020. The growth of large clusters was related to forward transmission cascades of untreated early-stage infections, younger at-risk populations, more transmissible/replicative-competent strains, and changing demographics. Subtype B and non-B subtype infections introduced through recent migration now surpass the domestic epidemic amongst MSM. Phylodynamics can assist in predicting and responding to active, recurrent, and newly emergent large cluster networks, as well as the cryptic spread of HIV introduced through migration.

Predictive Value of Time-Space Clusters for HIV Transmission in Washington State, 2017-2019

BACKGROUND

Pillar 4 of the United States' End the HIV Epidemic plan is to respond quickly to HIV outbreaks, but the utility of CDC's tool for identifying HIV outbreaks through time-space cluster detection has not been evaluated. The objective of this evaluation is to quantify the ability of the CDC time-space cluster criterion to predict future HIV diagnoses and to compare it to a space-time permutation statistic implemented in SaTScan software.

SETTING

Washington State from 2017 to 2019.

METHODS

We applied both cluster criteria to incident HIV cases in Washington State to identify clusters. Using a repeated-measures Poisson model, we calculated a rate ratio comparing the 6 months after cluster detection with a baseline rate from 24 to 12 months before the cluster was detected. We also compared the demographics of cases within clusters with all other incident cases.

RESULTS

The CDC criteria identified 17 clusters containing 192 cases in the 6 months after cluster detection, corresponding to a rate ratio of 1.25 (95% confidence interval: 0.95 to 1.65) relative to baseline. The time-space permutation statistic identified 5 clusters containing 25 cases with a rate ratio of 2.27 (95% confidence interval: 1.28 to 4.03). Individuals in clusters identified by the new criteria were more likely to be of Hispanic origin (61% vs 20%) and in rural areas (51% vs 12%).

CONCLUSIONS

The space-time permutation cluster analysis is a promising tool for identification of clusters with the largest growth potential for whom interruption may prove most beneficial.

Increasing Capacity to Detect Clusters of Rapid HIV Transmission in Varied Populations-United States

Molecular cluster detection analyzes HIV sequences to identify rapid HIV transmission and inform public health responses. We describe changes in the capability to detect molecular clusters and in geographic variation in transmission dynamics. We examined the reporting completeness of HIV-1 polymerase sequences in quarterly National HIV Surveillance System datasets from December 2015 to December 2019. Priority clusters were identified quarterly. To understand populations recently affected by rapid transmission, we described the transmission risk and race/ethnicity of people in clusters first detected in 2018–2019. During December 2015 to December 2019, national sequence completeness increased from 26% to 45%. Of the 1212 people in the 136 clusters first detected in 2018–2019, 69% were men who have sex with men (MSM) and 11% were people who inject drugs (PWID). State-by-state analysis showed substantial variation in transmission risk and racial/ethnic groups in clusters of rapid transmission. HIV sequence reporting has increased nationwide. Molecular cluster analysis identifies rapid transmission in varied populations and identifies emerging patterns of rapid transmission in specific population groups, such as PWID, who, in 2015–2016, comprised only 1% of people in such molecular clusters. These data can guide efforts to focus, tailor, and scale up prevention and care services for these populations.