Pharmacogenetics of weight gain following switch from efavirenz- to integrase inhibitor-containing regimens.

BACKGROUND: Excessive weight gain affects some persons with HIV after switching to integrase strand transfer inhibitor (INSTI)-containing antiretroviral therapy (ART). We studied associations between CYP2B6 genotype and weight gain after ART switch among ACTG A5001 and A5322 participants. METHODS: Eligible participants switched from efavirenz- to INSTI-containing ART, had genotype data, and had weight data at least once from 4 weeks to 2 years post-switch. Multivariable linear mixed effects models adjusted for race/ethnicity, CD4, age, BMI and INSTI type assessed relationships between CYP2B6 genotype and estimated differences in weight change. RESULTS: A total of 159 eligible participants switched ART from 2007 to 2019, of whom 138 had plasma HIV-1 RNA < 200 copies/mL (65 CYP2B6 normal, 56 intermediate, 17 poor metabolizers). Among participants with switch HIV-1 RNA < 200 copies/mL, weight increased in all 3 CYP2B6 groups. The rate of weight gain was greater in CYP2B6 poor than in CYP2B6 normal metabolizers overall, and within 9 subgroups (male, female, White, Black, Hispanic, dolutegravir, elvitegravir, raltegravir, and TDF in the pre-switch regimen); only in Hispanic and elvitegravir subgroups were these associations statistically significant ( P  < 0.05). Compared to normal metabolizers, CYP2B6 intermediate status was not consistently associated with weight gain. CONCLUSION: CYP2B6 poor metabolizer genotype was associated with greater weight gain after switch from efavirenz- to INSTI-containing ART, but results were inconsistent. Weight gain in this setting is likely complex and multifactorial.

Mitochondrial DNA mutation pathogenicity score and neurocognitive performance in persons with HIV.

BACKGROUND: Mitochondrial DNA (mtDNA) genetic variation is associated with neurocognitive (NC) impairment (NCI) in people with HIV (PWH). Other approaches use sequence conservation and protein structure to predict the impact of mtDNA variants on protein function. We examined predicted mtDNA variant pathogenicity in the CHARTER study using MutPred scores, hypothesizing that persons with higher scores (greater predicted pathogenicity) have more NCI. METHODS: CHARTER included NC testing in PWH from 2003 to 2007. MutPred scores were assigned to CHARTER participants with mtDNA sequence; any score > 0.5 was considered potentially deleterious. Outcomes at cohort entry were NCI, defined by global and seven NC domain deficit scores, and by mean global and domain NC performance T-scores. Univariate and multivariable regression analyses assessed associations between having a deleterious variant and NCI. Additional models included estimated peripheral blood cell mtDNA copy number. RESULTS: Data were available for 744 PWH (357 African ancestry; 317 European; 70 Hispanic). In the overall cohort, PWH having any potentially deleterious variant were less likely to have motor impairment (16 vs. 25 %, p = 0.001). In multivariable analysis, having a deleterious variant remained associated with lower likelihood of motor impairment (adjusted odds ratio 0.59 [95 % CI 0.41-0.88]; p = 0.009), and better motor performance by T-score (ß 1.71 [0.31-3.10], p = 0.02). Associations persisted after adjustment for estimated mtDNA quantity. CONCLUSIONS: In these PWH, having a potentially deleterious mtDNA variant was associated with less motor impairment. These unexpected findings suggest that potentially deleterious mtDNA variations may confer protection against impaired motor function by as yet unknown mechanisms.

Detection of distant relatedness in biobanks for identification of undiagnosed carriers of a Mendelian disease variant: application to Long QT Syndrome.

Rare genetic diseases are typically studied in referral populations, resulting in underdiagnosis and biased assessment of penetrance and phenotype. To address this, we developed a generalizable method of genotype inference based on distant relatedness and deployed this to identify undiagnosed Type 5 Long QT Syndrome (LQT5) rare variant carriers in a non-referral population. We identified 9 LQT5 families referred to a single specialty clinic, each carrying p.Asp76Asn, the most common LQT5 variant. We uncovered recent common ancestry and a single shared haplotype among probands. Application to a non-referral population of 69,819 BioVU biobank subjects identified 22 additional subjects sharing this haplotype, subsequently confirmed to carry p.Asp76Asn. Referral and non-referral carriers had prolonged QTc compared to controls, and, among carriers, QTc polygenic score additively associated with QTc prolongation. Thus, our novel analysis of shared chromosomal segments identified undiagnosed cases of genetic disease and refined the understanding of LQT5 penetrance and phenotype.

Detecting Problematic Opioid Use in the Electronic Health Record: Automation of the Addiction Behaviors Checklist in a Chronic Pain Population.

Importance: Individuals whose chronic pain is managed with opioids are at high risk of developing an opioid use disorder. Large data sets, such as electronic health records, are required for conducting studies that assist with identification and management of problematic opioid use. Objective: Determine whether regular expressions, a highly interpretable natural language processing technique, could automate a validated clinical tool (Addiction Behaviors Checklist1) to expedite the identification of problematic opioid use in the electronic health record. Design: This cross-sectional study reports on a retrospective cohort with data analyzed from 2021 through 2023. The approach was evaluated against a blinded, manually reviewed holdout test set of 100 patients. Setting: The study used data from Vanderbilt University Medical Center's Synthetic Derivative, a de-identified version of the electronic health record for research purposes. Participants: This cohort comprised 8,063 individuals with chronic pain. Chronic pain was defined by International Classification of Disease codes occurring on at least two different days.18 We collected demographic, billing code, and free-text notes from patients' electronic health records. Main Outcomes and Measures: The primary outcome was the evaluation of the automated method in identifying patients demonstrating problematic opioid use and its comparison to opioid use disorder diagnostic codes. We evaluated the methods with F1 scores and areas under the curve - indicators of sensitivity, specificity, and positive and negative predictive value. Results: The cohort comprised 8,063 individuals with chronic pain (mean [SD] age at earliest chronic pain diagnosis, 56.2 [16.3] years; 5081 [63.0%] females; 2982 [37.0%] male patients; 76 [1.0%] Asian, 1336 [16.6%] Black, 56 [1.0%] other, 30 [0.4%] unknown race patients, and 6499 [80.6%] White; 135 [1.7%] Hispanic/Latino, 7898 [98.0%] Non-Hispanic/Latino, and 30 [0.4%] unknown ethnicity patients). The automated approach identified individuals with problematic opioid use that were missed by diagnostic codes and outperformed diagnostic codes in F1 scores (0.74 vs. 0.08) and areas under the curve (0.82 vs 0.52). Conclusions and Relevance: This automated data extraction technique can facilitate earlier identification of people at-risk for, and suffering from, problematic opioid use, and create new opportunities for studying long-term sequelae of opioid pain management.

MitoTrace: A Computational Framework for Analyzing Mitochondrial Variation in Single-Cell RNA Sequencing Data.

Genetic variation in the mitochondrial genome is linked to important biological functions and various human diseases. Recent progress in single-cell genomics has established single-cell RNA sequencing (scRNAseq) as a popular and powerful technique to profile transcriptomics at the cellular level. While most studies focus on deciphering gene expression, polymorphisms including mitochondrial variants can also be readily inferred from scRNAseq. However, limited attention has been paid to investigate the single-cell landscape of mitochondrial variants, despite the rapid accumulation of scRNAseq data in the community. In addition, a diploid context is assumed for most variant calling tools, which is not appropriate for mitochondrial heteroplasmies. Here, we introduce MitoTrace, an R package for the analysis of mitochondrial genetic variation in bulk and scRNAseq data. We applied MitoTrace to several publicly accessible data sets and demonstrated its ability to robustly recover genetic variants from scRNAseq data. We also validated the applicability of MitoTrace to scRNAseq data from diverse platforms. Overall, MitoTrace is a powerful and user-friendly tool to investigate mitochondrial variants from scRNAseq data.

Detection of distant familial relatedness in biobanks for identification of undiagnosed carriers of a Mendelian disease variant: application to Long QT syndrome.

Importance: The diagnosis and study of rare genetic disease is often limited to referral populations, leading to underdiagnosis and a biased assessment of penetrance and phenotype. Objective: To develop a generalizable method of genotype inference based on distant relatedness and to deploy this to identify undiagnosed Type 5 Long QT Syndrome (LQT5) rare variant carriers in a non-referral population. Participants: We identified 9 LQT5 probands and 3 first-degree relatives referred to a single Genetic Arrhythmia clinic, each carrying D76N (p.Asp76Asn), the most common variant implicated in LQT5. The non-referral population consisted of 69,879 ancestry-matched subjects in BioVU, a large biobank that links electronic health records to dense array data. Participants were enrolled from 2007-2022. Data analysis was performed in 2022. Exposures: We developed and applied a novel approach to genotype inference (Distant Relatedness for Identification and Variant Evaluation, or DRIVE) to identify shared, identical-by-descent (IBD) large chromosomal segments in array data. Main Outcomes and Measures: We sought to establish genetic relatedness among the probands and to use genomic segments underlying D76N to identify other potential carriers in BioVU. We then further studied the role of D76N in LQT5 pathogenesis. Results: Genetic reconstruction of pedigrees and distant relatedness detection among clinic probands using DRIVE revealed shared recent common ancestry and identified a single long shared haplotype. Interrogation of the non-referral population in BioVU identified a further 23 subjects sharing this haplotype, and sequencing confirmed D76N carrier status in 22, all previously undiagnosed with LQT5. The QTc was prolonged in D76N carriers compared to BioVU controls, with 40% penetrance of QTc ≥ 480 msec. Among D76N carriers, a QTc polygenic score was additively associated with QTc prolongation. Conclusions and Relevance: Detection of IBD shared chromosomal segments around D76N enabled identification of distantly related and previously undiagnosed rare-variant carriers, demonstrated the contribution of polygenic risk to monogenic disease penetrance, and further established LQT5 as a primary arrhythmia disorder. Analysis of shared chromosomal regions spanning disease-causing mutations can identify undiagnosed cases of genetic diseases.

Polygenic Contributions to Chronic Overlapping Pain Conditions in a Large Electronic Health Record Sample.

Chronic overlapping pain conditions (COPCs) are believed to share common etiological mechanisms involving central sensitization. Genetic and environmental factors putatively combine to influence susceptibility to central sensitization and COPCs. This study employed a genome-wide polygenic risk score approach to evaluate genetic influences on 8 common COPCs. COPCs were identified by International Classification of Disease codes in Vanderbilt's deidentified clinical biorepository (BioVU), with each COPC condition empirically weighted for the level of central sensitization based on prior work. A centralized pain score (CPS) was calculated for 55,340 individuals by summing the weighted number of COPCs. Overall, 12,502 individuals (22.6%) were diagnosed with at least 1 COPC, with females exhibiting nearly twice the mean CPS as males. To assess the genetic influence on centralized pain in COPCs, 6 pain polygenic risk scores (PRSs) were developed using UK Biobank data to predict 6 pain criteria (no pain, neck/shoulder, abdomen, hip, knee, low back pain). These PRSs were then deployed in the BioVU cohort to test for association with CPS. In regression models adjusted for age, sex, and BMI, all pain PRSs except hip pain were significantly associated with CPS. Our findings support a shared polygenic influence across COPCs potentially involving central sensitization mechanisms. PERSPECTIVE: This study used a polygenic risk score approach to investigate genetic influences on chronic overlapping pain conditions. Significant findings in this study provide evidence supporting previous hypotheses that a shared polygenic influence involving central sensitization may underly chronic overlapping pain conditions and can guide future biomarker and risk assessment research.

Identifying problematic opioid use in electronic health record data: Are we looking in the right place?

OBJECTIVE: To examine the value of data obtained outside of regular healthcare visits (clinical communications) to detect problematic opioid use in electronic health records (EHRs). DESIGN: A retrospective cohort study. PARTICIPANTS: Chronic pain patient records in a large academic medical center. INTERVENTIONS: We compared evidence for problematic opioid use in (1) clinic notes, (2) clinical communications, and (3) full EHR data. We analyzed keyword counts and calculated concordance and Cohen's κ between data sources. MAIN OUTCOME MEASURE: Evidence of problematic opioid use in EHR defined as none, some, or high. RESULTS: Twenty-six percent of records were discordant in determination of problematic opioid use between clinical communications and clinic notes. Of these, 54 percent detected more evidence in clinical communications, and 46 percent in clinic notes. Compared to full EHR review, clinic notes exhibited higher concordance (78 percent; κ = 0.619) than clinical communications (60 percent; κ = 0.290). CONCLUSION: Clinical communications are a valuable addition to opioid EHR research.

How T118M peripheral myelin protein 22 predisposes humans to Charcot-Marie-Tooth disease.

Data from gnomAD indicate that a missense mutation encoding the T118M variation in human peripheral myelin protein 22 (PMP22) is found in roughly one of every 75 genomes of western European lineage (1:120 in the overall human population). It is unusual among PMP22 variants that cause Charcot-Marie-Tooth (CMT) disease in that it is not 100% penetrant. Here, we conducted cellular and biophysical studies to determine why T118M PMP22 predisposes humans to CMT, but with only incomplete penetrance. We found that T118M PMP22 is prone to mistraffic but differs even from the WT protein in that increased expression levels do not result in a reduction in trafficking efficiency. Moreover, the T118M mutant exhibits a reduced tendency to form large intracellular aggregates relative to other disease mutants and even WT PMP22. NMR spectroscopy revealed that the structure and dynamics of T118M PMP22 resembled those of WT. These results show that the main consequence of T118M PMP22 in WT/T118M heterozygous individuals is a reduction in surface-trafficked PMP22, unaccompanied by formation of toxic intracellular aggregates. This explains the incomplete disease penetrance and the mild neuropathy observed for WT/T118M CMT cases. We also analyzed BioVU, a biobank linked to deidentified electronic medical records, and found a statistically robust association of the T118M mutation with the occurrence of long and/or repeated episodes of carpal tunnel syndrome. Collectively, our results illuminate the cellular effects of the T118M PMP22 variation leading to CMT disease and indicate a second disorder for which it is a risk factor.

Compendium of proteins containing segments that exhibit zero-tolerance to amino acid variation in humans.

Genetic missense tolerance ratio (MTR) analysis systematically evaluates all possible segments in a given protein-encoding transcript found in the human population. This method scores each segment for the number of observed missense variants versus the number of silent mutations in that same segment. An MTR score of 0 indicates that no missense mutations are observed within a given segment. This is indicative of evolutionary purifying selection, which excludes mutations in that segment from the general human population. Here, we conducted MTR analysis on each of the roughly 20,000 protein-encoding human genes. It was seen that there are 257 genes with at least one 31-residue encoding segment with MTR = 0 (1.3% of all human genes). The proteins encoded by these 257 genes were tabulated along with information regarding the sequence location of each intolerant segment, the likely function of the protein, and so forth. The most functionally-enriched family among these proteins is a collection of several dozen proteins that are directly involved in RNA splicing. Some of the other proteins with zero-tolerance segments have thus far escaped significant characterization. Indeed, while a number of these proteins have previously been genetically linked to human disorders, many have not. We hypothesize that this compendium of human proteins with zero-tolerance segments can be used to complement disease mutation data as a pointer to genes and proteins that are associated with interesting and underexplored human biology.

Is it time to reassess variant annotation?

Variant annotation is one of the most essential steps in selecting candidates for further investigation. With the advancement in functional genomics, new variant annotation tools focus on annotation based on empirically generated data instead of theoretically based predictions. This is a direct result of the large national and international consortia that generated enormous experiment-based or validated data at multiple omics levels. Here, we highlight the recent empirically based annotation methods and discuss their strengths and weaknesses.

Plasma Arginine and Citrulline are Elevated in Diabetic Retinopathy.

PURPOSE: To determine if plasma levels of six arginine-related and citrulline-related metabolites (arginine, citrulline, asymmetric dimethylarginine [ADMA], ornithine, proline, and argininosuccinate) differ between patients with type 2 diabetes and diabetic retinopathy (DR) and type 2 diabetic controls or between patients with proliferative DR (PDR) and non-proliferative DR (NPDR). DESIGN: Cross-sectional study. METHODS: Adults with type 2 diabetes were recruited from the Vanderbilt Eye Institute. Exclusion criteria included non-diabetic retinal disease. Plasma metabolite levels were quantified in 159 diabetic controls and 156 DR patients (92 NPDR, 64 PDR) using isotope dilution liquid chromatography-tandem mass spectrometry (LC-MS/MS). Metabolite levels were compared using Wilcoxon Rank Sum test and logistic regressions adjusting for age, sex, hemoglobin A1c, diabetes duration, statin use, and angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker use. A secondary analysis that included creatinine in the regression model was performed for the subset of patients with available creatinine values (135 diabetic controls, 100 DR patients [58 NPDR, 42 PDR]). RESULTS: Multivariable logistic regression analyses determined that arginine (OR = 1.20, [1.06-1.38], P = .0067) and citrulline (OR = 1.53, [1.20-1.98], P = .0025) were significantly elevated in DR patients compared to diabetic controls. While ADMA differed between NPDR and PDR patients in the primary analysis (OR = 1.56, [1.15-2.16], P = .0051), it was not significantly different when adjusting for creatinine (OR = 1.30, [0.90-1.91], P = .15). CONCLUSIONS: Plasma arginine and citrulline were significantly elevated in type 2 diabetic patients with DR compared to diabetic controls. None of the tested metabolites significantly differed between NPDR and PDR patients in the adjusted analysis.

Two data-driven approaches to identifying the spectrum of problematic opioid use: A pilot study within a chronic pain cohort.

BACKGROUND: Although electronic health records (EHR) have significant potential for the study of opioid use disorders (OUD), detecting OUD in clinical data is challenging. Models using EHR data to predict OUD often rely on case/control classifications focused on extreme opioid use. There is a need to expand this work to characterize the spectrum of problematic opioid use. METHODS: Using a large academic medical center database, we developed 2 data-driven methods of OUD detection: (1) a Comorbidity Score developed from a Phenome-Wide Association Study of phenotypes associated with OUD and (2) a Text-based Score using natural language processing to identify OUD-related concepts in clinical notes. We evaluated the performance of both scores against a manual review with correlation coefficients, Wilcoxon rank sum tests, and area-under the receiver operating characteristic curves. Records with the highest Comorbidity and Text-based scores were re-evaluated by manual review to explore discrepancies. RESULTS: Both the Comorbidity and Text-based OUD risk scores were significantly elevated in the patients judged as High Evidence for OUD in the manual review compared to those with No Evidence (p = 1.3E-5 and 1.3E-6, respectively). The risk scores were positively correlated with each other (rho = 0.52, p < 0.001). AUCs for the Comorbidity and Text-based scores were high (0.79 and 0.76, respectively). Follow-up manual review of discrepant findings revealed strengths of data-driven methods over manual review, and opportunities for improvement in risk assessment. CONCLUSION: Risk scores comprising comorbidities and text offer differing but synergistic insights into characterizing problematic opioid use. This pilot project establishes a foundation for more robust work in the future.

EditPredict: Prediction of RNA editable sites with convolutional neural network.

RNA editing exerts critical impacts on numerous biological processes. While millions of RNA editings have been identified in humans, much more are expected to be discovered. In this work, we constructed Convolutional Neural Network (CNN) models to predict human RNA editing events in both Alu regions and non-Alu regions. With a validation dataset resulting from CRISPR/Cas9 knockout of the ADAR1 enzyme, the validation accuracies reached 99.5% and 93.6% for Alu and non-Alu regions, respectively. We ported our CNN models in a web service named EditPredict. EditPredict not only works on reference genome sequences but can also take into consideration single nucleotide variants in personal genomes. In addition to the human genome, EditPredict tackles other model organisms including bumblebee, fruitfly, mouse, and squid genomes. EditPredict can be used stand-alone to predict novel RNA editing and it can be used to assist in filtering for candidate RNA editing detected from RNA-Seq data.

Targeting diacylglycerol lipase reduces alcohol consumption in preclinical models.

Alcohol use disorder (AUD) is associated with substantial morbidity, mortality, and societal cost, and pharmacological treatment options for AUD are limited. The endogenous cannabinoid (eCB) signaling system is critically involved in reward processing and alcohol intake is positively correlated with release of the eCB ligand 2-Arachidonoylglycerol (2-AG) within reward neurocircuitry. Here we show that genetic and pharmacological inhibition of diacylglycerol lipase (DAGL), the rate limiting enzyme in the synthesis of 2-AG, reduces alcohol consumption in a variety of preclinical models ranging from a voluntary free-access model to aversion resistant-drinking and dependence-like drinking induced via chronic intermittent ethanol vapor exposure in mice. DAGL inhibition during either chronic alcohol consumption or protracted withdrawal was devoid of anxiogenic and depressive-like behavioral effects. Lastly, DAGL inhibition also prevented ethanol-induced suppression of GABAergic transmission onto midbrain dopamine neurons, providing mechanistic insight into how DAGL inhibition could affect alcohol reward. These data suggest reducing 2-AG signaling via inhibition of DAGL could represent an effective approach to reduce alcohol consumption across the spectrum of AUD severity.

LYSMD3: A mammalian pattern recognition receptor for chitin.

Chitin, a major component of fungal cell walls, has been associated with allergic disorders such as asthma. However, it is unclear how mammals recognize chitin and the principal receptor(s) on epithelial cells that sense chitin remain to be determined. In this study, we show that LYSMD3 is expressed on the surface of human airway epithelial cells and demonstrate that LYSMD3 is able to bind chitin, as well as ß-glucan, on the cell walls of fungi. Knockdown or knockout of LYSMD3 also sharply blunts the production of inflammatory cytokines by epithelial cells in response to chitin and fungal spores. Competitive inhibition of the LYSMD3 ectodomain by soluble LYSMD3 protein, multiple ligands, or antibody against LYSMD3 also blocks chitin signaling. Our study reveals LYSMD3 as a mammalian pattern recognition receptor (PRR) for chitin and establishes its role in epithelial cell inflammatory responses to chitin and fungi.

Higher CSF Ferritin Heavy-Chain (Fth1) and Transferrin Predict Better Neurocognitive Performance in People with HIV.

HIV-associated neurocognitive disorder (HAND) remains prevalent despite antiretroviral therapy and involves white matter damage in the brain. Although iron is essential for myelination and myelin maintenance/repair, its role in HAND is largely unexplored. We tested the hypotheses that cerebrospinal fluid (CSF) heavy-chain ferritin (Fth1) and transferrin, proteins integral to iron delivery and myelination, are associated with neurocognitive performance in people with HIV (PWH). Fth1, transferrin, and the pro-inflammatory cytokines TNF-α and IL-6 were quantified in CSF at baseline (entry) in 403 PWH from a prospective observational study who underwent serial, comprehensive neurocognitive assessments. Associations of Fth1 and transferrin with Global Deficit Score (GDS)-defined neurocognitive performance at baseline and 30-42 months of follow-up were evaluated by multivariable regression. While not associated with neurocognitive performance at baseline, higher baseline CSF Fth1 predicted significantly better neurocognitive performance over 30 months in all PWH (p < 0.05), in PWH aged < 50 at 30, 36, and 42 months (all p < 0.05), and in virally suppressed PWH at all three visit time-points (all p < 0.01). Higher CSF transferrin was associated with superior neurocognitive performance at all visits, primarily in viremic individuals (all p < 0.05). All associations persisted after adjustment for neuro-inflammation. In summary, higher CSF Fth1 is neuroprotective over prolonged follow-up in all and virally suppressed PWH, while higher CSF transferrin may be most neuroprotective during viremia. We speculate that higher CSF levels of these critical iron-delivery proteins support improved myelination and consequently, neurocognitive performance in PWH, providing a rationale for investigating their role in interventions to prevent and/or treat HAND.

Mitochondrial DNA haplogroups and weight gain following switch to integrase strand transfer inhibitor-based antiretroviral therapy.

BACKGROUND: Integrase strand transfer inhibitor (INSTI)-based antiretroviral therapy (ART) has been associated with excess weight gain in some adults, which may be influenced by genetic factors. We assessed mitochondrial DNA (mtDNA) haplogroups and weight gain following switch to INSTI-based ART. METHODS: All AIDS Clinical Trials Group A5001 and A5322 participants with mtDNA genotyping who switched to INSTI were included. mtDNA haplogroups were derived from prior genotyping algorithms. Race/ethnicity-stratified piecewise linear mixed effects models assessed the relationship between mtDNA haplogroup and weight change slope differences before and after switch to INSTI. RESULTS: A total of 291 adults switched to INSTI: 78% male, 50% non-Hispanic White, 28% non-Hispanic Black, and 22% Hispanic. The most common European haplogroups were H [n = 66 (45%)] and UK [32 (22%)]. Non-H European haplogroups had a significant increase in weight slope after the switch. This difference was greatest among non-H clade UK on INSTI-based regimens that included tenofovir alafenamide (TAF) [3.67 (95% confidence interval 1.12, 6.21) kg/year; P = 0.005]. Although small sample size limited analyses among non-Hispanic Black and Hispanic persons, similarly significant weight gain was seen among the most common African haplogroup, L3 [n = 29 (39%); slope difference 4.93 (1.54, 8.32) kg/year, P = 0.005], after switching to TAF-containing INSTI-based ART. CONCLUSION: Those in European mtDNA haplogroup clade UK and African haplogroup L3 had significantly greater weight gain after switching to INSTI-based ART, especially those receiving TAF. Additional studies in large and diverse populations are needed to clarify the mechanisms and host risk factors for weight gain after switching to INSTI-based ART, with and without TAF.