Biernacka, Joanna M.

Abstract Objective Social determinants of health (SDoH) are nonclinical dispositions that impact patient health risks and clinical outcomes. Leveraging SDoH in clinical decision-making can potentially improve diagnosis, treatment planning, and patient outcomes. Despite increased interest in capturing SDoH in electronic health records (EHRs), such information is typically locked in unstructured clinical notes. Natural language processing (NLP) is the key technology to extract SDoH information from clinical text and expand its utility in patient care and research. This article presents a systematic review of the state-of-the-art NLP approaches and tools that focus on identifying and extracting SDoH data from unstructured clinical text in EHRs. Materials and Methods A broad literature search was conducted in February 2021 using 3 scholarly databases (ACL Anthology, PubMed, and Scopus) following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. A total of 6402 publications were initially identified, and after applying the study inclusion criteria, 82 publications were selected for the final review. Results Smoking status (n = 27), substance use (n = 21), homelessness (n = 20), and alcohol use (n = 15) are the most frequently studied SDoH categories. Homelessness (n = 7) and other less-studied SDoH (eg, education, financial problems, social isolation and support, family problems) are mostly identified using rule-based approaches. In contrast, machine learning approaches are popular for identifying smoking status (n = 13), substance use (n = 9), and alcohol use (n = 9). Conclusion NLP offers significant potential to extract SDoH data from narrative clinical notes, which in turn can aid in the development of screening tools, risk prediction models, and clinical decision support systems.

AbstractInterest in analyzing X chromosome single nucleotide polymorphisms (SNPs) is growing and several approaches have been proposed. Prior studies have compared power of different approaches, but bias and interpretation of coefficients have received less attention. We performed simulations to demonstrate the impact of X chromosome model assumptions on effect estimates. We investigated the coefficient biases of SNP and sex effects with commonly used models for X chromosome SNPs, including models with and without assumptions of X chromosome inactivation (XCI), and with and without SNP–sex interaction terms. Sex and SNP coefficient biases were observed when assumptions made about XCI and sex differences in SNP effect in the analysis model were inconsistent with the data‐generating model. However, including a SNP–sex interaction term often eliminated these biases. To illustrate these findings, estimates under different genetic model assumptions are compared and interpreted in a real data example. Models to analyze X chromosome SNPs make assumptions beyond those made in autosomal variant analysis. Assumptions made about X chromosome SNP effects should be stated clearly when reporting and interpreting X chromosome associations. Fitting models with SNP × Sex interaction terms can avoid reliance on assumptions, eliminating coefficient bias even in the absence of sex differences in SNP effect.

Abstract Introduction Sleep disruption is common in patients with alcohol use disorders (AUD) and major depressive disorders (MDD). Our understanding of the differences in the rates of sleep disturbance and overall sleep duration in patients with AUD, MDD, and comorbid AUD and MDD is limited. Furthermore, it is unknown whether there is variation in demographic and clinical characteristics associated with sleep disturbance and duration in these diagnostic groups. Methods This study utilized data from the UK Biobank (UKB). Depression status was determined based on review of International Classification of Diseases (ICD) codes and health records. AUD status was based on AUDIT scores (score ≥8 was defined as AUD) and sleep disturbance was evaluated utilizing a self-reported questionnaire. The sample was categorized into those with MDD alone (MDD+/AUD-)(n=18,154), AUD alone (MDD-/AUD+)(n=6123), both (MDD+/AUD+)(n=9027), and controls with neither (MDD-/AUD-)(n=27,573). We used generalized linear models (GLMs) to compare rates of sleep disruption and duration among the groups and determine the clinical predictors of sleep disturbance/duration in the four groups as well as test whether these factors differed among the groups. Results The prevalence of sleep disturbance in the control sample (MDD-/AUD-) was 26.4% and the self-reported sleep duration in this sample was 7.209±0.919. Subjects with AUD and/or MDD had greater rates of sleep disturbance and shorter sleep duration. Among the different diagnostic categories, the prevalence of sleep disturbance was highest in subjects with MDD+/AUD+ (36.5%) followed by those with MDD+/AUD- (35.6%) and MDD-/AUD+ (27.9%)(all p<0.0001). Similarly, the sleep duration was shortest in subjects with MDD+/AUD+ (7.143±1.016), followed by MDD+/AUD- (7.158±1.050) and by MDD-/AUD+ (7.202±0.891)(all p<0.0001). Subjects with sleep disturbance were more likely to be older, female, and with higher body mass index, Townsend deprivation index, and neuroticism scores across all four groups (all p<0.05). Conclusion In a large population-based cohort, MDD with and without comorbid AUD was associated with greater rates of sleep disturbance and shorter sleep duration than AUD alone. The clinical and demographic factors associated with sleep disturbance did not differ in these diagnostic categories, indicating possible similar underlying risk factors. Support (if any) NA

AbstractIntroductionA key mechanism of lithium is the inhibition of glycogen synthase kinase‐3β (GSK3β) and activation of mammalian target of rapamycin (mTOR), two contributors to insulin signaling. We explored the relationship between these markers and clinical response to lithium in bipolar disorder (BD).MethodsThirty‐four subjects with BD who had been taking lithium for ≥2 years and had a maintenance lithium Alda score defined as either high (≥7; n = 20) or low (≤2; n = 14) were included in the study. Baseline protein expression of GSK3β and mTOR (total and phosphorylated (p)) was obtained from a buffy coat. Peripheral blood mononuclear cells (PBMCs) from a subset of each group (n = 11) were stimulated with insulin (10 µg) and change in protein expression was determined using Western blot.ResultsIn buffy coat samples, significantly higher levels of pmTOR were present in subjects with an Alda score ≤2 (lithium non‐responsive), relative to those with scores ≥7 (lithium‐responsive). No differences were observed for pGSK3β. In contrast, functional PBMC responses to 5 min of insulin stimulation demonstrated robust increases in pGSK3β (87.05 ± 43.41%) and pmTOR (105.7 ± 66.48%) in the lithium responsive group only. This contrasted observed decreases in pGSK3β (34.08 ± 16.12%) and pmTOR (37.84 ± 14.39%) 5 mins post‐insulin in non‐responders.ConclusionsDynamic increases in pmTOR and pGSK3β post‐insulin stimulation may reflect an immunometabolic state that facilitates lithium response. Further prospective analyses are needed to replicate and extend these preliminary findings and further investigate the role of insulin signaling in lithium response in BD.

AbstractBackgroundLithium is regarded as a first‐line treatment for bipolar disorder (BD), but partial response and non‐response commonly occurs. There exists a need to identify lithium non‐responders prior to initiating treatment. The Pharmacogenomics of Bipolar Disorder (PGBD) Study was designed to identify predictors of lithium response.MethodsThe PGBD Study was an eleven site prospective trial of lithium treatment in bipolar I disorder. Subjects were stabilized on lithium monotherapy over 4 months and gradually discontinued from all other psychotropic medications. After ensuring a sustained clinical remission (defined by a score of ≤3 on the CGI for 4 weeks) had been achieved, subjects were followed for up to 2 years to monitor clinical response. Cox proportional hazard models were used to examine the relationship between clinical measures and time until failure to remit or relapse.ResultsA total of 345 individuals were enrolled into the study and included in the analysis. Of these, 101 subjects failed to remit or relapsed, 88 achieved remission and continued to study completion, and 156 were terminated from the study for other reasons. Significant clinical predictors of treatment failure (p < 0.05) included baseline anxiety symptoms, functional impairments, negative life events and lifetime clinical features such as a history of migraine, suicidal ideation/attempts, and mixed episodes, as well as a chronic course of illness.ConclusionsIn this PGBD Study of lithium response, several clinical features were found to be associated with failure to respond to lithium. Future validation is needed to confirm these clinical predictors of treatment failure and their use clinically to distinguish who will do well on lithium before starting pharmacotherapy.

AbstractAimsLong‐term lithium therapy (LTLT) has been associated with kidney insufficiency in bipolar disorder (BD). We aimed to investigate the risk factors of chronic kidney disease (CKD) development and progression among BD patients receiving LTLT.MethodsWe included adult patients with BD on LTLT (≥1 year) who were enrolled in the Mayo Clinic Bipolar Biobank, Rochester, Minnesota. We reviewed electronic medical records to extract information related to lithium therapy and kidney‐related data to assess changes in the estimated glomerular filtration rate (eGFR). CKD severity was assessed based on eGFR.ResultsAmong 154 patients who received LTLT, 41 patients (27%) developed CKD, of whom 20 (49%) patients continued lithium (continuers) and 19 (46%) discontinued it (discontinuers). The median time to stage 3 CKD development was 21.7 years from the start of Li treatment. Type‐2 diabetes mellitus and benzodiazepine use were independent predictors for CKD development in the survival analysis, after controlling for age. The subsequent CKD progression rate did not differ between continuers and discontinuers (mean GFR 48.6 vs. 44.1, p = 0.13) at the end of follow‐up duration (mean duration: 3.5 ± 4.4 years for continuers and 4.9 ± 5.3 years for discontinuers).ConclusionCKD was observed in one fourth of patients with BD receiving LTLT. There was no significant difference in the progression of CKD among Li continuers versus discontinuers, at the mean follow‐up duration of 4.2 years, after the CKD diagnosis. Progression of CKD could be influenced by existing comorbidities and may not necessarily be due to lithium alone.

AbstractBackgroundIt is important to determine the natural history of sporadic and familial frontotemporal lobar degeneration (FTLD) and generate clinical, neuropsychological, neuroimaging and biofluid data for planning disease‐modifying trials.MethodAs part of the ARTFL LEFFTDS Longitudinal Frontotemporal Lobar Degeneration (ALLFTD; U19 AG063911) protocol, investigators at 19 centers in North America will enroll 2100 participants with FTLD over the next 5 years beginning in early 2020.ResultAs of 1/20/20, the ARTFL/LEFFTDS (A/L) Consortium had enrolled 1832 participants, including 850 in kindreds with familial FTLD (239 associated with mutations in MAPT, 192 in GRN, 370 in C9orf72, 4 with mutations in both GRN and C9orf72, and 45 with a mutation in a different gene or no mutation in any known FTLD‐associated gene). Over 500 participants have undergone 2 or more annual visits to date. MRI has been performed in 1105. Biofluid samples have also been collected, with blood (DNA, plasma, serum, mRNA, PBMC) in 1413 and CSF in 303. Over 60 manuscripts using A/L data or samples have been published to date. Five clinical trials involving A/L and ALLFTD participants are in progress or planned. The longitudinal arm in ALLFTD will enroll 500 of existing A/L and 600 future participants for annual assessments with similar methodology to A/L. An additional 1000 FTLD patients will undergo focused one‐time clinical evaluations and biofluid collection.ConclusionThe data/samples from already‐enrolled and planned participants in ALLFTD and findings published to date underscore the utility of evaluating FTLD subjects. The absence of identifiable mutations in some with familial FTLD suggests that other genes are yet to be discovered. ALLFTD data will inform clinical trial design, and many participants will be eligible for future trials. The key data and samples in ALLFTD are available to interested investigators worldwide. Supported by: AG063911, AG045390, NS092089, AG016976, AG21886.

Abstract Background: Genetic studies have shifted to sequencing-based rare variants discovery after decades of success in identifying common disease variants by Genome-Wide Association Studies using Single Nucleotide Polymorphism chips. Sequencing-based studies require large sample sizes for statistical power but often inadvertently introduce batch effects because samples are typically collected, processed, and sequenced at multiple centers. Conventionally, batch effects are first detected and visualized using Principal Components Analysis and then controlled by including batch covariates in the disease association models. For sequencing-based genetic studies, because all variants included in the association analyses have passed quality control measures, this conventional approach treats every variant as equal and ignores the substantial differences still remaining in variant qualities and characteristics such as genotype quality scores, alternative allele fractions (fraction of reads supporting alternative allele at a variant position) and sequencing depths. Results: In the Alzheimer’s Disease Sequencing Project (ADSP) exome dataset of 9,904 cases and controls, we discovered hidden variant-level differences between sample batches of three sequencing centers and two exome capture kits. Although sequencing centers were included as a covariate in our association models, we observed differences at the variant level in genotype quality and alternative allele fraction between samples processed by different exome capture kits that significantly impacted both the confidence of variant detection and the identification of disease-associated variants. Furthermore, we found that the association signals of a subset of top disease risk variants came exclusively from samples processed by one exome capture kit that was more effective at capturing the alternative alleles compared to the other kit. Conclusions: Our findings highlight the importance of additional variant-level quality control for large sequencing-based genetic studies. More importantly, we demonstrate that automatically filtering out variants with batch differences may lead to false negatives if the batch discordance came largely from quality differences and if the variants from one batch had better quality scores.

Abstract Polygenic risk scores (PRSs) have become an increasingly popular approach for demonstrating polygenic influences on complex traits and for establishing common polygenic signals between different traits. PRSs are typically constructed using pruning and thresholding (P+T), but the best choice of parameters is uncertain; thus multiple settings are used and the best is chosen. Optimization can lead to inflated Type I error. Permutation procedures can correct this, but they can be computationally intensive. Alternatively, a single parameter setting can be chosen a priori for the PRS, but choosing suboptimal settings results in loss of power. We propose computing PRSs under a range of parameter settings, performing principal component analysis (PCA) on the resulting set of PRSs, and using the first PRS–PC in association tests. The first PC reweights the variants included in the PRS to achieve maximum variation over all PRS settings used. Using simulations and a real data application to study PRS association with bipolar disorder and psychosis in bipolar disorder, we compare the performance of the proposed PRS–PCA approach with a permutation test and an a priori selected p ‐value threshold. The PRS–PCA approach is simple to implement, outperforms the other strategies in most scenarios, and provides an unbiased estimate of prediction performance.

Several antiepileptic drugs (AEDs) have US Food and Drug Administration (FDA) approval for use as mood stabilizers in bipolar disorder (BD), but not all BD patients respond to these AED mood stabilizers (AED‐MSs). To identify genetic polymorphisms that contribute to the variability in AED‐MS response, we performed a discovery genome‐wide association study (GWAS) of 199 BD patients from the Mayo Clinic Bipolar Disorder Biobank. Most of these patients had been treated with the AED‐MS valproate/divalproex and/or lamotrigine. AED‐MS response was assessed using the Alda scale, which quantifies clinical improvement while accounting for potential confounding factors. We identified two genome‐wide significant single‐nucleotide polymorphism (SNP) signals that mapped to the THSD7A (rs78835388, P = 7.1E‐09) and SLC35F3 (rs114872993, P = 3.2E‐08) genes. We also identified two genes with statistically significant gene‐level associations: ABCC1 (P = 6.7E‐07; top SNP rs875740, P = 2.0E‐6), and DISP1 (P = 8.9E‐07; top SNP rs34701716, P = 8.9E‐07). THSD7A SNPs were previously found to be associated with risk for several psychiatric disorders, including BD. Both THSD7A and SLC35F3 are expressed in excitatory/glutamatergic and inhibitory/γ‐aminobutyric acidergic (GABAergic) neurons, which are targets of AED‐MSs. ABCC1 is involved in the transport of valproate and lamotrigine metabolites, and the SNPs in ABCC1 and DISP1 with the strongest evidence of association in our GWAS are strong splicing quantitative trait loci in the human gut, suggesting a possible influence on drug absorption. In conclusion, our pharmacogenomic study identified novel genetic loci that appear to contribute to AED‐MS treatment response, and may facilitate precision medicine in BD.