24 August 2020 in PLOS ONE
Microbiome data consists of operational taxonomic unit (OTU) counts characterized by zero-inflation, over-dispersion, and grouping structure among samples. Currently, statistical testing methods are commonly performed to identify OTUs that are associated with a phenotype. The limitations of statistical testing methods include that the validity of p-values/q-values depend sensitively on the correctness of models and that the statistical significance does not necessarily imply predictivity. Predictive analysis using methods such as LASSO is an alternative approach for identifying associated OTUs and for measuring the predictability of the phenotype variable with OTUs and other covariate variables. We investigate three strategies of performing predictive analysis: (1) LASSO: fitting a LASSO multinomial logistic regression model to all OTU counts with specific transformation; (2) screening+GLM: screening OTUs with q-values returned by fitting a GLMM to each OTU, then fitting a GLM model using a subset of selected OTUs; (3) screening+LASSO: fitting a LASSO to a subset of OTUs selected with GLMM. We have conducted empirical studies using three simulation datasets generated using Dirichlet-multinomial models and a real gut microbiome data related to Parkinson’s disease to investigate the performance of the three strategies for predictive analysis. Our simulation studies show that the predictive performance of LASSO with appropriate variable transformation works remarkably well on zero-inflated data. Our results of real data analysis show that Parkinson’s disease can be predicted based on selected OTUs after the binary transformation, age, and sex with high accuracy (Error Rate = 0.199, AUC = 0.872, AUPRC = 0.912). These results provide strong evidences of the relationship between Parkinson’s disease and the gut microbiome.
1 March 2016 in Anti-Cancer Drugs
To observe drug-induced hepatotoxicity by long-term gefitinib administration in the treatment of non-small-cell lung cancer. The data of 101 patients with locally advanced or metastatic non-small-cell lung cancer, for which gefitinib had been used orally for 3 months or longer, were retrospectively analyzed. The median duration of gefitinib administration was 14 months (3-60 months). Forty patients (39.6%) developed abnormal hepatic function, among whom 30 patients (29.7%) had grade I hepatotoxicity, six patients (5.9%) had grade II, and four patients (4.0%) had grade III, respectively. The median time from starting gefitinib oral therapy to developing liver dysfunction was 4 months (1-23 months) for the entire cohort. The incidence of hepatotoxicity in the group with a duration of more than 14 months was much higher than that in the group with a duration of less than 14 months (52.0 vs. 27.5%, P=0.012). In thirty-two patients (32/40), abnormal liver function resolved with hepatoprotective treatment, whereas eight patients (8/40) had persistent grade I hepatotoxicity until the last follow-up. Our study showed that long-term gefitinib-induced hepatotoxicity was a common adverse event, especially for the cohort with a duration of longer than 14 months. In most patients with hepatotoxicity, normal liver function was restored and discontinuation of gefitinib was not necessary.
12 March 2021 in PLOS Computational Biology
KSKonstantin Shestopaloff
Current advances in next-generation sequencing techniques have allowed researchers to conduct comprehensive research on the microbiome and human diseases, with recent studies identifying associations between the human microbiome and health outcomes for a number of chronic conditions. However, microbiome data structure, characterized by sparsity and skewness, presents challenges to building effective classifiers. To address this, we present an innovative approach for distance-based classification using mixture distributions (DCMD). The method aims to improve classification performance using microbiome community data, where the predictors are composed of sparse and heterogeneous count data. This approach models the inherent uncertainty in sparse counts by estimating a mixture distribution for the sample data and representing each observation as a distribution, conditional on observed counts and the estimated mixture, which are then used as inputs for distance-based classification. The method is implemented into ak-means classification andk-nearest neighbours framework. We develop two distance metrics that produce optimal results. The performance of the model is assessed using simulated and human microbiome study data, with results compared against a number of existing machine learning and distance-based classification approaches. The proposed method is competitive when compared to the other machine learning approaches, and shows a clear improvement over commonly used distance-based classifiers, underscoring the importance of modelling sparsity for achieving optimal results. The range of applicability and robustness make the proposed method a viable alternative for classification using sparse microbiome count data. The source code is available athttps://github.com/kshestop/DCMDfor academic use.
1 February 2025 in Carcinogenesis
Lung cancer is the leading cause of cancer mortality. To investigate genetic determinants for prognosis among patients diagnosed with early-stage non-small cell lung cancer (NSCLC), we conducted the first large-scale genome-wide association prognostic study using data from the International Lung Cancer Consortium (ILCCO) through a two-phase analysis. Phase 1 includes the discovery of genome-wide association studies analysis using a multivariable Cox PH model on 3428 NSCLC patients of European ancestry from 10 ILCCO participating studies to identify genetic variants associated with overall survival and validation analysis for genome-wide significant variants (P-value ≤5 × 10−8) using the Cancer Genome Atlas (TCGA). Phase 2 aims to identify causal variants using functional analyses of genome-wide significant and suggestive variants (P-value ≤1 × 10−5), including variant-epigenetic functional annotation (FAVOR), CHIP-seq data, variant-gene expression association, and colocalization analysis. We identified two significant variants; of those, a locus at 9q21.31 (rs117979484) was significant at the genome-wide level (P = 3.67 × 10−8) and validated in TCGA (P = 0.03). Three suggestive variants were found to have a putative epigenetic function: intronic variants rs149281784 (BCL7B gene) and rs148031766 (POM121 gene) both located at 7q11.23 and in moderate linkage disequilibrium with each other; and variant rs2471630 (SRCIN1 gene; 17q12). Specifically, variants rs149281784 and rs148031766 have potential regulatory roles in the transcriptional activation of the BCL7B gene and POM121 gene. Exploratory survival analyses in the squamous cell carcinomas subgroup also identified a significant variant, rs138467404 (GRHL-2 gene; 8q22.3) at a genome-wide level (P = 4.75 × 10−8) and validated by TCGA (P = 0.02). These new findings indicate potential novel pathways associated with early-stage NSCLC prognosis. Future research may validate additional genome-wide suggestive variants as being relevant for lung cancer outcomes.
1 March 2022 in Cancer Epidemiology, Biomarkers & Prevention
Background: Somatic EGFR mutations define a subset of non–small cell lung cancers (NSCLC) that have clinical impact on NSCLC risk and outcome. However, EGFR-mutation-status is often missing in epidemiologic datasets. We developed and tested pragmatic approaches to account for EGFR-mutation-status based on variables commonly included in epidemiologic datasets and evaluated the clinical utility of these approaches. Methods: Through analysis of the International Lung Cancer Consortium (ILCCO) epidemiologic datasets, we developed a regression model for EGFR-status; we then applied a clinical-restriction approach using the optimal cut-point, and a second epidemiologic, multiple imputation approach to ILCCO survival analyses that did and did not account for EGFR-status. Results: Of 35,356 ILCCO patients with NSCLC, EGFR-mutation-status was available in 4,231 patients. A model regressing known EGFR-mutation-status on clinical and demographic variables achieved a concordance index of 0.75 (95% CI, 0.74–0.77) in the training and 0.77 (95% CI, 0.74–0.79) in the testing dataset. At an optimal cut-point of probability-score = 0.335, sensitivity = 69% and specificity = 72.5% for determining EGFR-wildtype status. In both restriction-based and imputation-based regression analyses of the individual roles of BMI on overall survival of patients with NSCLC, similar results were observed between overall and EGFR-mutation-negative cohort analyses of patients of all ancestries. However, our approach identified some differences: EGFR-mutated Asian patients did not incur a survival benefit from being obese, as observed in EGFR-wildtype Asian patients. Conclusions: We introduce a pragmatic method to evaluate the potential impact of EGFR-status on epidemiological analyses of NSCLC. Impact: The proposed method is generalizable in the common occurrence in which EGFR-status data are missing.
2 September 2026 in SSRN Electronic Journal
Background: Polygenic risk scores (PRS) can provide a proxy for genetic liability to smoking- and alcohol-related behaviors. If these behavior-associated PRS can also predict the risk or survival of head and neck squamous cell carcinoma (HNSCC), then such PRS can provide motivation for behavioral modification during counseling by associating behavior with its consequences.Methods: Based on the VOYAGER Consortium of 3,254 controls and 5,538 HNSCC cases, where 3,194 were HPV-positive, we constructed PRS for multiple smoking and alcoholic metrics, using summary statistics from GSCAN. We evaluated the HPV-stratified HNSCC risk by logistic regression, overall survival by Cox models comparing different percentile strata of PRS and compared these results to observed behaviors.Findings: Each higher PRS predicted greater corresponding smoking and alcohol use behaviors. For HPV-positive diseases, higher PRS for smoking initiation and alcoholic drinks-per-week predicted increased HNSCC risk. Similarly, higher PRS for smoking initiation, cigarettes-per-day, smoking cessation, and alcoholic drinks-per-week each predicted higher HPV-negative HNSCC risk. For overall survival, higher PRS of smoking initiation (HR = 1.08, P=0.017) and alcoholic drinks-per-week (HR=1.14, P=2.66×10-4) predicted poorer prognosis in HPV-positive cases. Illustrative models showed similar results for a genetic counseling purpose. Each PRS predicted its corresponding observed behavior which mediated part of the associations (7.30%-54.6%) between PRSs and HNSCC risk or survival.Interpretation: Smoking- and alcohol-related PRS predicted observed behaviors, greater HNSCC risk and poorer overall survival. These findings support integrating genetic liability with clinical factors to refine counseling strategies for behavior modification.