We consider variable selection for high-dimensional multivariate regression using penalized likelihoods when the number of outcomes and the number of covariates might be large. To account for within-subject correlation, we consider variable selection when a working precision matrix is used and when the precision matrix is jointly estimated using a two-stage procedure. We show that under suitable regularity conditions, penalized regression coefficient estimators are consistent for model selection for an arbitrary working precision matrix, and have the oracle properties and are efficient when the true precision matrix is used or when it is consistently estimated using sparse regression. We develop an efficient computation procedure for estimating regression coefficients using the coordinate descent algorithm in conjunction with sparse precision matrix estimation using the graphical LASSO (GLASSO) algorithm. We develop the Bayesian Information Criterion (BIC) for estimating the tuning parameter and show that BIC is consistent for model selection. We evaluate finite sample performance for the proposed method using simulation studies and illustrate its application using the type II diabetes gene expression pathway data.
Publications by Year: 2014
2014
OBJECTIVE: Successful pulmonary wedge resection for early-stage non-small cell lung cancer requires a pathologically confirmed negative margin. To date, however, no clear evidence is available regarding whether an optimal margin distance, defined as the distance from the primary tumor to the closest resection margin, exists. Toward addressing this gap, we investigated the relationship between the margin distance and local recurrence risk.
METHODS: We reviewed all adult patients who had undergone wedge resection for small (≤2 cm) non-small cell lung cancer from January 2001 to August 2011, with follow-up through to December 31, 2011. The exclusion criteria included other active noncutaneous malignancies, bronchoalveolar carcinomas, lymph node or distant metastases at diagnosis, large cell cancer, adenosquamous cancer, multiple, multifocal, and/or metastatic disease, and previous chemotherapy or radiotherapy. Using Cox regression analysis, we examined the relationship between the margin distance and interval to local recurrence, adjusting for chronic obstructive pulmonary disease, forced expiratory volume in 1 second, smoking, diabetes, tumor size, tumor location, surgeon, open versus video-assisted thoracoscopic surgery, and whether the lymph nodes were sampled.
RESULTS: Of 557 consecutive adult patients, 479 met our inclusion criteria. The overall, unadjusted 1- and 2-year local recurrences rate was 5.7% and 11.0%, respectively. From the adjusted analyses, an increased margin distance was significantly associated with a lower risk of local recurrence (P = .033). Patients with a 10-mm margin distance had a 45% lower local recurrence risk than those with a 5-mm distance (hazard ratio, 0.55; 95% confidence interval, 0.35-0.86). Beyond 15 mm, no evidence of additional benefit was associated with an increased margin distance.
CONCLUSIONS: In wedge resection for small non-small cell lung cancer, increasing the margin distance ≤15 mm significantly decreased the local recurrence risk, with no evidence of additional benefit beyond 15 mm.
BACKGROUND: Exposure to air particulate matter is known to elevate blood biomarkers of inflammation and to increase cardiopulmonary morbidity and mortality. Major components of airborne particulate matter typically include black carbon from traffic and sulfates from coal-burning power plants. DNA methylation is thought to be sensitive to these environmental toxins and possibly mediate environmental effects on clinical outcomes via regulation of gene networks. The underlying mechanisms may include epigenetic modulation of major inflammatory pathways, yet the details remain unclear.
METHODS: We sought to elucidate how short-term exposure to air pollution components, singly and/or in combination, alter blood DNA methylation in certain inflammation-associated gene networks, MAPK and NF-κB, which may transmit the environmental signal(s) and influence the inflammatory pathway in vivo. To this end, we utilized a custom-integrated workflow-molecular processing, pollution surveillance, biostatical analysis, and bioinformatic visualization-to map novel human (epi)gene pathway-environment interactions.
RESULTS: Specifically, out of 84 MAPK pathway genes considered, we identified 11 whose DNA methylation status was highly associated with black carbon exposure, after adjusting for potential confounders-age, sulfate exposure, smoking, blood cell composition, and blood pressure. Moreover, after adjusting for these confounders, multi-pollutant analysis of synergistic DNA methylations significantly associated with sulfate and BC exposures yielded 14 MAPK genes. No associations were found with the NF-κB pathway.
CONCLUSION: Exposure to short-term air pollution components thus resulted in quantifiable epigenetic changes in the promoter areas of MAPK pathway genes. Bioinformatic mapping of single- vs. multi-exposure-associated epigenetic changes suggests that these alterations might affect biological pathways in nuanced ways that are not simply additive or fully predictable via individual-level exposure assessments.