Doubly-robust methods for differences in restricted mean lifetimes using pseudo-observations.

In clinical studies or trials comparing survival times between two treatment groups, the restricted mean lifetime (RML), defined as the expectation of the survival from time 0 to a prespecified time-point, is often the quantity of interest that is readily interpretable to clinicians without any modeling restrictions. It is well known that if the treatments are not randomized (as in observational studies), covariate adjustment is necessary to account for treatment imbalances due to confounding factors. In this article, we propose a simple doubly-robust pseudo-value approach to effectively estimate the difference in the RML between two groups (akin to a metric for estimating average causal effects), while accounting for confounders. The proposed method combines two general approaches: (a) group-specific regression models for the time-to-event and covariate information, and (b) inverse probability of treatment assignment weights, where the RMLs are replaced by the corresponding pseudo-observations for survival outcomes, thereby mitigating the estimation complexities in presence of censoring. The proposed estimator is double-robust, in the sense that it is consistent if at least one of the two working models remains correct. In addition, we explore the potential of available machine learning algorithms in causal inference to reduce possible bias of the causal estimates in presence of a complex association between the survival outcome and covariates. We conduct extensive simulation studies to assess the finite-sample performance of the pseudo-value causal effect estimators. Furthermore, we illustrate our methodology via application to a dataset from a breast cancer cohort study. The proposed method is implementable using the R package drRML, available in GitHub.

Women with multiple gestations have an increased risk of development of hypertension in the future.

BACKGROUND: Multiple gestations are associated with an increased incidence of preeclampsia. However, there exists no evidence for an association between multiple gestations and development of hypertension(HTN) later in life. This study aimed to determine whether multiple gestations are associated with HTN beyond the peripartum period. METHODS: In this retrospective nationwide population-based study, women who delivered a baby between January 1, 2007, and December 31, 2008, and underwent a national health screening examination within one year prior to their pregnancy were included. Subsequently, we tracked the occurrence of HTN during follow-up until December 31, 2015, using International Classification of Diseases-10th Revision codes. RESULTS: Among 362,821 women who gave birth during the study period, 4,944 (1.36%) women had multiple gestations. The cumulative incidence of HTN was higher in multiple gestations group compared with singleton group (5.95% vs. 3.78%, p < 0.01, respectively). On the Cox proportional hazards models, the risk of HTN was increased in women with multiple gestations (HR 1.35, 95% CI 1.19, 1.54) compared with those with singleton after adjustment for age, primiparity, preeclampsia, atrial fibrillation, body mass index, blood pressure, diabetes mellitus, high total cholesterol, abnormal liver function test, regular exercise, and smoking status. CONCLUSIONS: Multiple gestations are associated with an increased risk of HTN later in life. Therefore, guidelines for the management of high-risk patients after delivery should be established.

Twelve-month post-treatment parameters are superior in predicting hepatocellular carcinoma in patients with chronic hepatitis B.

BACKGROUND & AIMS: There are currently several prediction models for hepatocellular carcinoma (HCC) in chronic hepatitis B (CHB) receiving oral antiviral therapy. However, most models are based on pre-treatment clinical parameters. The current study aimed to develop a novel and practical prediction model for HCC by using both pre- and post-treatment parameters in this population. METHODS: We included two treatment-naïve CHB cohorts who were initiated on oral antiviral therapies: the derivation cohort (n = 1480, Korea prospective SAINT cohort) and the validation cohort (n = 426, the US retrospective Stanford Bay cohort). We employed logistic regression, decision tree, lasso regression, support vector machine and random forest algorithms to develop the HCC prediction model and selected the most optimal method. RESULTS: We evaluated both pre-treatment and the 12-month clinical parameters on-treatment and found the 12-month on-treatment values to have superior HCC prediction performance. The lasso logistic regression algorithm using the presence of cirrhosis at baseline and alpha-foetoprotein and platelet at 12 months showed the best performance (AUROC = 0.843 in the derivation cohort. The model performed well in the external validation cohort (AUROC = 0.844) and better than other existing prediction models including the APA, PAGE-B and GAG models (AUROC = 0.769 to 0.818). CONCLUSIONS: We provided a simple-to-use HCC prediction model based on presence of cirrhosis at baseline and two objective laboratory markers (AFP and platelets) measured 12 months after antiviral initiation. The model is highly accurate with excellent validation in an external cohort from a different country (AUROC 0.844) (Clinical trial number: KCT0003487).

Peripartum Blood Transfusions are Associated with Increased Risk of Cancer: A National Retrospective Cohort Study.

BACKGROUND: The effect of blood transfusions on the risk of developing primary cancer remains unclear, especially when administered in the peripartum period. MATERIALS AND METHODS: We conducted a retrospective cohort study of 270,529 pregnant women who delivered between January 1, 2007 and December 31, 2009, with data obtained from three national databases in South Korea. From this cohort, we identified 4569 patients who received peripartum blood transfusions. We calculated hazard ratios (HRs) for new diagnoses of cancer and adjusted them for relevant clinical factors using a Cox proportional hazards model. RESULTS: During follow-up, patients who received peripartum transfusions had an increased risk of developing cancer, with an adjusted HR of 1.16 (95% confidence interval [CI], 1.01-1.34). In a subgroup analysis, this risk was significant only among patients who received 3 or more units of blood, with an adjusted HR of 1.40 (95% CI, 1.10-1.79). Increased risk after transfusions were seen with brain, lung, ovarian, and gallbladder cancers. The difference in cancer risk between the transfusion and no-transfusion groups remained significant during both the first (1.29% vs 1.07%, p < 0.01) and second year (0.74% vs 0.56%, p < 0.01) after delivery. CONCLUSION: Receipt of 3 or more blood transfusions in the peripartum period was associated with a significantly increased risk of developing cancer. Prospective studies should be pursued to further understand the link between blood transfusions and long-term oncologic risks.

Gene expression network analysis of lymph node involvement in colon cancer identifies AHSA2, CDK10, and CWC22 as possible prognostic markers.

Colon cancer has been well studied using a variety of molecular techniques, including whole genome sequencing. However, genetic markers that could be used to predict lymph node (LN) involvement, which is the most important prognostic factor for colon cancer, have not been identified. In the present study, we compared LN(+) and LN(-) colon cancer patients using differential gene expression and network analysis. Colon cancer gene expression data were obtained from the Cancer Genome Atlas and divided into two groups, LN(+) and LN(-). Gene expression networks were constructed using LASSO (Least Absolute Shrinkage and Selection Operator) regression. We identified hub genes, such as APBB1, AHSA2, ZNF767, and JAK2, that were highly differentially expressed. Survival analysis using selected hub genes, such as AHSA2, CDK10, and CWC22, showed that their expression levels were significantly associated with the survival rate of colon cancer patients, which indicates their possible use as prognostic markers. In addition, protein-protein interaction network, GO enrichment, and KEGG pathway analysis were performed with selected hub genes from each group to investigate the regulatory relationships between hub genes and LN involvement in colon cancer; these analyses revealed differences between the LN(-) and LN(+) groups. Our network analysis may help narrow down the search for novel candidate genes for the treatment of colon cancer, in addition to improving our understanding of the biological processes underlying LN involvement. All R implementation codes are available at journal website as Supplementary Materials.

Peripheral blood transcriptome analysis and development of classification model for diagnosing antibody-mediated rejection vs accommodation in ABO-incompatible kidney transplant.

The major obstacle to successful ABO blood group-incompatible kidney transplantation (ABOi KT) is antibody-mediated rejection (AMR). This study aimed to investigate transcriptional profiles through RNA sequencing and develop a minimally invasive diagnostic tool for discrimination between accommodation and early acute AMR in ABOi KT. Twenty-eight ABOi KT patients were selected: 18 with accommodation and 10 with acute AMR at the 10th day posttransplant protocol biopsy. Complete transcriptomes of their peripheral blood were analyzed by RNA sequencing. Candidate genes were selected by bioinformatics analysis, validated with quantitative polymerase chain reaction, and used to develop a classification model to diagnose accommodation. A total of 1385 genes were differentially expressed in accommodation compared with in AMR with P-adjusted < .05. Functional annotation and gene set enrichment analysis identified several immune-related and immunometabolic pathways. A 5-gene classification model including COX7A2L, CD69, CD14, CFD, and FOXJ3 was developed by logistic regression analysis. The model was further validated with an independent cohort and discriminated between accommodation and AMR with 92.7% sensitivity, 85.7% specificity, and 91.7% accuracy. Our study suggests that a classification model based on peripheral blood transcriptomics may allow minimally invasive diagnosis of acute AMR vs accommodation and subsequent patient-tailored immunosuppression in ABOi KT.

Predicting peripartum blood transfusion: focusing on pre-pregnancy characteristics.

BACKGROUND: Obstetric hemorrhage is one of the most common causes of obstetrical morbidity and mortality, and transfusion is the most important management for hemorrhage. The aim of our study was to investigate the pre-pregnancy and pregnancy risk factors for peripartum transfusion. METHODS: Women who delivered a baby from 2010 to 2014 in Korea and participated in the Korean National Health Screening Program for Infants and Children were included. To analyze pre-pregnant risk factors for peripartum transfusion, an additional analysis was done for women who underwent a National Health Screening Examination within 1 year before pregnancy, including maternal waist circumference, body mass index, blood pressure, laboratory tests and history of smoking. Multivariable logistic regression analysis was used to estimate the risk factors for peripartum transfusion. RESULTS: Of the total 1,980,126 women who met the inclusion criteria, 36,868 (1.86%) were transfused at peripartum. In a multivariable regression model, the pregnancy risk factors for peripartum transfusion included maternal age above 35 years [odds ratio (OR): 1.41; 95% confidence interval (CI): 1.32-1.50], preterm birth (OR: 2.39; 95% CI: 2.15-2.65), and maternal hypertension (OR: 2.49; 95% CI: 2.24-2.77). Pre-pregnancy risk factors including fasting glucose level of more than 126 mg/dL (OR: 1.11; 95% CI: 1.02-1.20), current-smoker status (OR: 1.20; 95% CI: 1.06-1.37), and waist-circumference less than 80 cm (OR: 1.18; 95% CI: 1.06-1.30) were independently associated with peripartum blood transfusion. CONCLUSIONS: Several pre-pregnancy and pregnancy risk factors were associated with peripartum blood transfusion. Some identified factors are modifiable before conception, and our study validated peripartum blood transfusion as a form of triage.

Low Alanine Aminotransferase Cut-Off for Predicting Liver Outcomes; A Nationwide Population-Based Longitudinal Cohort Study.

BACKGROUND AND AIM: Recent practice guidelines suggest healthy normal alanine aminotransferase (ALT) levels should be less than 30 U/L for males and 19 U/L for females. We tried to validate the prediction power of the "low cut off" for liver related outcomes in the general population. METHODS: A total of 426,013 subjects were followed up for 10 years using the National Health Screening Cohort database. Prediction ability of long term mortality and liver related outcomes between conventional (<40 U/L in men and women) and low (<30 U/L in men and <19 U/L in women) ALT cut-off values were compared. RESULTS: Both conventional and low ALT cut-offs predicted liver related unfavorable outcomes in Kaplan-Meier analysis. Following adjustment for age, body mass index, smoking, exercise, alcohol consumption, fasting blood glucose, and cholesterol via multivariate Cox regression, abnormal ALT using new 'low ALT cut off' was a significant independent predictor for liver-related mortality, HCC, and decompensated liver events. When the low cut-off criteria were added to the prediction model, the ability to predetect liver-related hard outcomes significantly increased in both men and women (p-values < 0.0001). The C-index values for predicting liver-related adverse events were the same in both ALT cut-offs, after adjusting confounding factors (C index value: 0.73~0.88). CONCLUSIONS: New low ALT cut-off showed good prediction power for liver related unfavorable outcomes.

Integrative Proteo-genomic Analysis to Construct CNA-protein Regulatory Map in Breast and Ovarian Tumors.

Recent development in high throughput proteomics and genomics profiling enable one to study regulations of genome alterations on protein activities in a systematic manner. In this article, we propose a new statistical method, ProMAP, to systematically characterize the regulatory relationships between proteins and DNA copy number alterations (CNA) in breast and ovarian tumors based on proteogenomic data from the CPTAC-TCGA studies. Because of the dynamic nature of mass spectrometry instruments, proteomics data from labeled mass spectrometry experiments usually have non-ignorable batch effects. Moreover, mass spectrometry based proteomic data often possesses high percentages of missing values and non-ignorable missing-data patterns. Thus, we use a linear mixed effects model to account for the batch structure and explicitly incorporate the abundance-dependent-missing-data mechanism of proteomic data in ProMAP. In addition, we employ a multivariate regression framework to characterize the multiple-to-multiple regulatory relationships between CNA and proteins. Further, we use proper statistical regularization to facilitate the detection of master genetic regulators, which affect the activities of many proteins and often play important roles in genetic regulatory networks. Improved performance of ProMAP over existing methods were illustrated through extensive simulation studies and real data examples. Applying ProMAP to the CPTAC-TCGA breast and ovarian cancer data sets, we identified many genome regions, including a few novel ones, whose CNA were associated with protein and or phosphoprotein abundances. For example, in breast tumors, a small region in 8p11.21 was recognized as the second biggest hub in the CNA-phosphoprotein regulatory map, and further investigation of the regulatory targets suggests the potential role of 8p11.21 CNA in perturbing oxygen binding and transport activities in tumor cells. This and other findings from our analyses help to characterize the impacts of CNAs on protein activity landscapes and cast light on the genetic regulation mechanisms underlying these tumors.

Prediction of survival risks with adjusted gene expression through risk-gene networks.

MOTIVATION: Network-based analysis of biomedical data has been extensively studied over the last decades. As a successful application, gene networks have been used to illustrate interactions among genes and explain the associated phenotypes. However, the gene network approaches have not been actively applied for survival analysis, which is one of the main interests of biomedical research. In addition, a few previous studies using gene networks for survival analysis construct networks mainly from prior knowledge, such as pathways, regulations and gene sets, while the performance considerably depends on the selection of prior knowledge. RESULTS: In this paper, we propose a data-driven construction method for survival risk-gene networks as well as a survival risk prediction method using the network structure. The proposed method constructs risk-gene networks with survival-associated genes using penalized regression. Then, gene expression indices are hierarchically adjusted through the networks to reduce the variance intrinsic in datasets. By illustrating risk-gene structure, the proposed method is expected to provide an intuition for the relationship between genes and survival risks. The risk-gene network is applied to a low grade glioma dataset, and produces a hypothesis of the relationship between genetic biomarkers of low and high grade glioma. Moreover, with multiple datasets, we demonstrate that the proposed method shows superior prediction performance compared to other conventional methods. AVAILABILITY AND IMPLEMENTATION: The R package of risk-gene networks is freely available in the web at http://cdal.korea.ac.kr/NetDA/. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Risk of Adverse Obstetric Outcomes and the Abnormal Growth of Offspring in Women with a History of Thyroid Cancer.

Background: It is unclear whether a history of thyroid cancer is associated with an increased risk of adverse pregnancy outcomes in subsequent pregnancies. This study aimed to evaluate the risk of adverse obstetric outcomes and the abnormal growth of offspring in women with a history of thyroid cancer. Methods: This retrospective observational study used nationwide data from between 2006 and 2014 to compare pregnancy outcomes of women with a history of thyroid cancer and those with no such history. Cases of thyroid cancer were identified using ICD-10 codes. Results: During the study period, 7232 women with a history of thyroid cancer and 2,269,051 women without a history of thyroid cancer gave birth. The risks of cesarean section, preterm birth, low birth weight, large for gestational age, preeclampsia, placental abruption, placenta previa, and stillbirth were not different between the groups. Women with a history of thyroid cancer had a statistically higher risk of postpartum hemorrhage (odds ratio [OR] = 1.23 [confidence interval (CI) 1.15-1.32], p < 0.05, corrected with the false discovery rate). Additionally, generalized estimating equations analysis showed that there was no difference in the risk of underweight (OR = 1.05 [CI 0.93-1.19]) and obese (OR = 0.94 [CI 0.84-1.05]) offspring assessed over a period of 80 months after adjusting for confounding factors. Conclusions: Women with a history of thyroid cancer have similar pregnancy outcomes and offspring growth to those with no such history.

Comparative Evaluation of the Biodegradability and Wrinkle Reduction Efficacy of Human-Derived Collagen Filler and Hyaluronic Acid Filler.

BACKGROUND: The development of fillers for wrinkle prevention is growing to meet rising demands to reduce the aging of skin. OBJECTIVE: In this experiment, we confirmed the effects of human collagen and hyaluronic acid filler biodegradation for wrinkle reduction using a photo-aging mouse model. MATERIALS AND METHODS: A total of 10 hairless mice (SKH1-Hrhr) were randomly divided into two groups and injected with hyaluronic acid and human-derived collagen filler. At 0, 2, 4, 8, and 12 weeks, PRIMOSlite®, folliscope, and MRI were used to evaluate the biodegradability of the fillers after the injections. We also studied the photo-aging mouse model for skin roughness and histological evaluation and confirmed that the filler injection had excellent anti-wrinkle effects. RESULTS: Human-derived collagen fillers had excellent biodegradability compared to that of hyaluronic acid fillers. The skin surface roughness in the photo-aging mouse models was significantly reduced after injections of human-derived collagen filler. CONCLUSION: Our results showed that the human-derived collagen filler had excellent biodegradability and effectively reduced wrinkle formation in a photo-aging mouse model. NO LEVEL ASSIGNED: This journal requires that authors assign a level of evidence to each article. For a full description of these Evidence-Based Medicine ratings, please refer to the Table of Contents or the online Instructions to Authors www.springer.com/00266 .

Correction: Long-term effect of pregnancy-related factors on the development of endometrial neoplasia: A nationwide retrospective cohort study.

[This corrects the article DOI: 10.1371/journal.pone.0214600.].

Long-term effect of pregnancy-related factors on the development of endometrial neoplasia: A nationwide retrospective cohort study.

OBJECTIVE: By identifying pregnancy-related risk factors for endometrial neoplasia, women's risk of developing this disease after childbirth can be predicted and high-risk women can be screened for early detection. METHODS: Study data from women who gave birth in Korea in 2007 were collected from the Korea National Health Insurance (KNHI) claims database between 2007 and 2015. The adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for the development of endometrial neoplasia were estimated by multivariate Cox proportional hazards models. RESULTS: Data from 386,614 women were collected for this study. By 2015, 3,370 women from the initial cohort had been diagnosed with endometrial neoplasia secondary to delivery. Multivariate Cox proportional hazards regression revealed that preeclampsia (HR 1.55, 95% CI 1.29, 1.86), advanced maternal age (≥ 35; HR 1.52, 95% CI 1.39, 1.66), multifetal pregnancy (HR 1.81, 95% CI 1.46, 2.23), multiparity (HR 1.16, 95% CI 1.08, 1.24), cesarean section (HR 1.15, 95% CI 1.07, 1.23) and delivery of a large-for-gestational-age infant (HR 1.19, 95% CI 1.02, 1.39) were independent risk factors for future endometrial neoplasia. The risk for endometrial neoplasia increased as the number of risk factors increased (risk factors ≥3: HR 2.11, 95% CI 1.86-2.40). CONCLUSION: This study showed that six pregnancy-related factors-advanced maternal age, multiparity, multifetal pregnancy, cesarean section, delivery of a large-for-gestational-age infant, and preeclampsia-are positively correlated with future development of endometrial neoplasia, including endometrial hyperplasia or cancer. Close observation and surveillance are warranted to enable early diagnosis of endometrial diseases, including endometrial cancer after pregnancy in high-risk women. However, due to unavailability of clinical information, many clinical/epidemiological factors can become confounders. Further research is needed on factors associated with the risk of endometrial neoplasia.

BTNET : boosted tree based gene regulatory network inference algorithm using time-course measurement data.

BACKGROUND: Identifying gene regulatory networks is an important task for understanding biological systems. Time-course measurement data became a valuable resource for inferring gene regulatory networks. Various methods have been presented for reconstructing the networks from time-course measurement data. However, existing methods have been validated on only a limited number of benchmark datasets, and rarely verified on real biological systems. RESULTS: We first integrated benchmark time-course gene expression datasets from previous studies and reassessed the baseline methods. We observed that GENIE3-time, a tree-based ensemble method, achieved the best performance among the baselines. In this study, we introduce BTNET, a boosted tree based gene regulatory network inference algorithm which improves the state-of-the-art. We quantitatively validated BTNET on the integrated benchmark dataset. The AUROC and AUPR scores of BTNET were higher than those of the baselines. We also qualitatively validated the results of BTNET through an experiment on neuroblastoma cells treated with an antidepressant. The inferred regulatory network from BTNET showed that brachyury, a transcription factor, was regulated by fluoxetine, an antidepressant, which was verified by the expression of its downstream genes. CONCLUSIONS: We present BTENT that infers a GRN from time-course measurement data using boosting algorithms. Our model achieved the highest AUROC and AUPR scores on the integrated benchmark dataset. We further validated BTNET qualitatively through a wet-lab experiment and showed that BTNET can produce biologically meaningful results.

Gene Regulatory Network Analysis for Triple-Negative Breast Neoplasms by Using Gene Expression Data.

PURPOSE: To better identify the physiology of triple-negative breast neoplasm (TNBN), we analyzed the TNBN gene regulatory network using gene expression data. METHODS: We collected TNBN gene expression data from The Cancer Genome Atlas to construct a TNBN gene regulatory network using least absolute shrinkage and selection operator regression. In addition, we constructed a triple-positive breast neoplasm (TPBN) network for comparison. Furthermore, survival analysis based on gene expression levels and differentially expressed gene (DEG) analysis were carried out to support and compare the network analysis results, respectively. RESULTS: The TNBN gene regulatory network, which followed a power-law distribution, had 10,237 vertices and 17,773 edges, with an average vertex-to-vertex distance of 8.6. The genes ZDHHC20 and RAPGEF6 were identified by centrality analysis to be important vertices. However, in the DEG analysis, we could not find meaningful fold changes in ZDHHC20 and RAPGEF6 between the TPBN and TNBN gene expression data. In the multivariate survival analysis, the hazard ratio for ZDHHC20 and RAPGEF6 was 1.677 (1.192-2.357) and 1.676 (1.222-2.299), respectively. CONCLUSION: Our TNBN gene regulatory network was a scale-free one, which means that the network would be easily destroyed if the hub vertices were attacked. Thus, it is important to identify the hub vertices in the network analysis. In the TNBN gene regulatory network, ZDHHC20 and RAPGEF6 were found to be oncogenes. Further study of these genes could help to reveal a novel method for treating TNBN in the future.

Cross-activating c-Met/ß1 integrin complex drives metastasis and invasive resistance in cancer.

The molecular underpinnings of invasion, a hallmark of cancer, have been defined in terms of individual mediators but crucial interactions between these mediators remain undefined. In xenograft models and patient specimens, we identified a c-Met/ß1 integrin complex that formed during significant invasive oncologic processes: breast cancer metastases and glioblastoma invasive resistance to antiangiogenic VEGF neutralizing antibody, bevacizumab. Inducing c-Met/ß1 complex formation through an engineered inducible heterodimerization system promoted features crucial to overcoming stressors during metastases or antiangiogenic therapy: migration in the primary site, survival under hypoxia, and extravasation out of circulation. c-Met/ß1 complex formation was up-regulated by hypoxia, while VEGF binding VEGFR2 sequestered c-Met and ß1 integrin, preventing their binding. Complex formation promoted ligand-independent receptor activation, with integrin-linked kinase phosphorylating c-Met and crystallography revealing the c-Met/ß1 complex to maintain the high-affinity ß1 integrin conformation. Site-directed mutagenesis verified the necessity for c-Met/ß1 binding of amino acids predicted by crystallography to mediate their extracellular interaction. Far-Western blotting and sequential immunoprecipitation revealed that c-Met displaced α5 integrin from ß1 integrin, creating a complex with much greater affinity for fibronectin (FN) than α5ß1. Thus, tumor cells adapt to microenvironmental stressors induced by metastases or bevacizumab by coopting receptors, which normally promote both cell migration modes: chemotaxis, movement toward concentrations of environmental chemoattractants, and haptotaxis, movement controlled by the relative strengths of peripheral adhesions. Tumor cells then redirect these receptors away from their conventional binding partners, forming a powerful structural c-Met/ß1 complex whose ligand-independent cross-activation and robust affinity for FN drive invasive oncologic processes.

Computer-assisted measurement of primary tumor area is prognostic of recurrence-free survival in stage IB melanoma patients.

Current staging guidelines are insufficient to predict which patients with thin primary melanoma are at high risk of recurrence. Computer-assisted image analysis may allow for more practical and objective histopathological analysis of primary tumors than traditional light microscopy. We studied a prospective cohort of stage IB melanoma patients treated at NYU Langone Medical Center from 2002 to 2014. Primary tumor width, manual area, digital area, and conformation were evaluated in a patient subset via computer-assisted image analysis. The associations between histologic variables and survival were evaluated using Cox proportional hazards model. Logistic regressions were used to build a classifier with clinicopathological characteristics to predict recurrence status. Of the 655 patients with stage IB melanoma studied, a subset of 149 patient tumors (63 recurred, 86 did not recur) underwent computer-assisted histopathological analysis. Increasing tumor width (hazard ratios (HR): 1.17, P=0.01) and digital area (HR: 1.08, P<0.01) were significantly associated with worse recurrence-free survival, whereas non-contiguous conformation (HR: 0.57, P=0.05) was significantly associated with better recurrence-free survival. The novel histopathological classifier composed of digital area, conformation, and baseline variables effectively distinguished recurrent cases from non-recurrent cases (AUC: 0.733, 95% confidence interval (CI): 0.647-0.818), compared to the baseline classifier alone (AUC: 0.635, 95% CI: 0.545-0.724). Primary tumor cross-sectional area, width, and conformation measured via computer-assisted analysis may help identify high-risk patients with stage IB melanoma.

Outcomes in Melanoma Patients Treated with BRAF/MEK-Directed Therapy or Immune Checkpoint Inhibition Stratified by Clinical Trial versus Standard of Care.

OBJECTIVES: Since 2011, metastatic melanoma treatment has evolved with commercial approval of BRAF- and MEK-targeted therapy and CTLA-4- and PD-1-blocking antibodies (immune checkpoint inhibitors, ICI). While novel therapies have demonstrated improved prognosis in clinical trials, few studies have examined the evolution of prognosis and toxicity of these drugs among an unselected population. We assess whether survival and toxicity reported in trials, which typically exclude most patients with brain metastases and poor performance status, are recapitulated within a commercial access population. METHODS: 182 patients diagnosed with stage IV melanoma from July 2006 to December 2013 and treated with BRAF- and/or MEK-targeted therapy or ICI were prospectively studied. Outcomes and clinicopathologic differences between trial and commercial cohorts were assessed. RESULTS: Patients receiving commercial therapy (vs. on trial) had poorer prognostic features (i.e., brain metastases) and lower median overall survival (mOS) when assessed across all treatments (9.2 vs. 17.5 months, p = 0.0027). While toxicity within trial and commercial cohorts did not differ, patients who experienced toxicity had increased mOS (p < 0.001), irrespective of stratification by trial status or therapy. CONCLUSION: Metastatic melanoma patients receiving commercial treatment may represent a different clinical population with poor prognostic features compared to trial patients. Toxicity may prognosticate treatment benefit.

A Comparison of Two-Stage Approaches Based on Penalized Regression for Estimating Gene Networks.

Graphical models are commonly used for illustrating gene networks. However, estimating directed networks are generally challenging because of the limited sample size compared with the dimensionality of an experiment. Many previous studies have provided insight into the problem, and recently, two-stage approaches have shown significant improvements for estimating directed acyclic graphs. These two-stage approaches find neighborhoods in the first stage and determine the directions of the edges in the second stage. However, although numerous methods to find neighborhoods and determine directions exist, the most appropriate method to use with two-stage approaches has not been evaluated. Therefore, we compared such methods through extensive simulations to select effective methods for the first and second stages. Results show that adaptive lasso is the most effective for both stages in most cases. In addition, we compared methods to handle asymmetric entries to estimate an undirected network. Some previous studies indicate that the method used to handle asymmetric entries does not affect performance significantly; however, we found that the selection of the handling method for such edges is a significant factor for finding neighborhoods when using adaptive lasso.