Search

Attempts to predict prognosis in cancer patients using high-dimensional genomic data such as gene expression in tumor tissue can be made difficult by the large number of features and the potential complexity of the relationship between features and the outcome. Integrating prior biological knowledge into risk prediction with such data by grouping genomic features into pathways and networks reduces the dimensionality of the problem and could improve prediction accuracy. Additionally, such knowledge-based models may be more biologically grounded and interpretable. Prediction could potentially be further improved by allowing for complex nonlinear pathway effects. The kernel machine framework has been proposed as an effective approach for modeling the nonlinear and interactive effects of genes in pathways for both censored and noncensored outcomes. When multiple pathways are under consideration, one may efficiently select informative pathways and aggregate their signals via multiple kernel learning (MKL), which has been proposed for prediction of noncensored outcomes. In this paper, we propose MKL methods for censored survival outcomes. We derive our approach for a general survival modeling framework with a convex objective function and illustrate its application under the Cox proportional hazards and semiparametric accelerated failure time models. Numerical studies demonstrate that the proposed MKL-based prediction methods work well in finite sample and can potentially outperform models constructed assuming linear effects or ignoring the group knowledge. The methods are illustrated with an application to 2 cancer data sets., (Copyright © 2018 John Wiley & Sons, Ltd.)

When a moderate number of potential predictors are available and a survival model is fit with regularization to achieve variable selection, providing accurate inference on the predicted survival can be challenging. We investigate inference on the predicted survival estimated after fitting a Cox model under regularization guaranteeing the oracle property. We demonstrate that existing asymptotic formulas for the standard errors of the coefficients tend to underestimate the variability for some coefficients, while typical resampling such as the bootstrap tends to overestimate it; these approaches can both lead to inaccurate variance estimation for predicted survival functions. We propose a two-stage adaptation of a resampling approach that brings the estimated error in line with the truth. In stage 1, we estimate the coefficients in the observed data set and in [Formula: see text] resampled data sets, and allow the resampled coefficient estimates to vote on whether each coefficient should be 0. For those coefficients voted as zero, we set both the point and interval estimates to [Formula: see text] In stage 2, to make inference about coefficients not voted as zero in stage 1, we refit the penalized model in the observed data and in the [Formula: see text] resampled data sets with only variables corresponding to those coefficients. We demonstrate that ensemble voting-based point and interval estimators of the coefficients perform well in finite samples, and prove that the point estimator maintains the oracle property. We extend this approach to derive inference procedures for survival functions and demonstrate that our proposed interval estimation procedures substantially outperform estimators based on asymptotic inference or standard bootstrap. We further illustrate our proposed procedures to predict breast cancer survival in a gene expression study., (© The Author 2016. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)

Integrating genomic information with traditional clinical risk factors to improve the prediction of disease outcomes could profoundly change the practice of medicine. However, the large number of potential markers and possible complexity of the relationship between markers and disease make it difficult to construct accurate risk prediction models. Standard approaches for identifying important markers often rely on marginal associations or linearity assumptions and may not capture non-linear or interactive effects. In recent years, much work has been done to group genes into pathways and networks. Integrating such biological knowledge into statistical learning could potentially improve model interpretability and reliability. One effective approach is to employ a kernel machine (KM) framework, which can capture nonlinear effects if nonlinear kernels are used (Scholkopf and Smola, 2002; Liu et al., 2007, 2008). For survival outcomes, KM regression modeling and testing procedures have been derived under a proportional hazards (PH) assumption (Li and Luan, 2003; Cai, Tonini, and Lin, 2011). In this article, we derive testing and prediction methods for KM regression under the accelerated failure time (AFT) model, a useful alternative to the PH model. We approximate the null distribution of our test statistic using resampling procedures. When multiple kernels are of potential interest, it may be unclear in advance which kernel to use for testing and estimation. We propose a robust Omnibus Test that combines information across kernels, and an approach for selecting the best kernel for estimation. The methods are illustrated with an application in breast cancer., (© 2013, The International Biometric Society.)

OBJECTIVES: To examine if racial differences in cardiovascular health (CVH) are associated with cardiovascular disease (CVD) disparities among women with breast and gynecologic cancers. SAMPLE & SETTING: The sample consisted of 252 Black women and 93 White women without a self-reported history of cancer or CVD who developed a breast or gynecologic malignancy. Women who developed CVD before their cancer diagnosis were excluded. METHODS & VARIABLES: CVH was classified using metrics of the American Heart Association's Life's Simple 7 framework. Metrics were summed to create a total CVH score (0-7). Associations among race, ideal CVH (score of 5-7), and CVD incidence following cancer diagnosis were estimated with Cox proportional hazards models. RESULTS: Ideal CVH was similar between Black women (33%) and White women (37%). Race and CVH were not associated with CVD incidence. IMPLICATIONS FOR NURSING: In a small sample of women diagnosed with breast and gynecologic cancers, racial disparities in CVH and CVD incidence were not observed. Additional investigation of potential confounders relating to social determinants of health tied to the construct of race is warranted. [ABSTRACT FROM AUTHOR]

Endometrial cancer stage is a strong prognostic factor; however, the current stage classification does not incorporate transtubal spread as determined by intraluminal tumor cells (ILTCs). We examined relationships between ILTCs and survival outcomes according to histological subtype and stage and examined whether identification of ILTCs improves prognostic accuracy of endometrial cancer staging. We conducted a retrospective cohort study of women diagnosed with endometrial cancer at five academic hospitals between 2007 and 2012. Pathologists determined ILTC presence (no vs. yes) and location (free in lumen vs. attached to epithelial surface) based on pathology review of hematoxylin and eosin-stained sections of fallopian tubes. Associations between ILTCs with time to recurrence (TTR) and overall survival (OS) were examined with Cox proportional hazards models adjusted for other prognostic factors. Model discrimination metrics were used to assess the addition of ILTCs to stage for prediction of 5-year TTR and OS. In the overall study population (N = 1303), ILTCs were not independently associated with TTR (HR = 0.95, 95% CI = 0.69–1.32) or OS (HR = 0.97, 95% CI = 0.72–1.31). Among 805 women with stage I disease, ILTCs were independently associated with worse TTR (HR = 2.31, 95% CI = 1.06–5.05) and OS (HR = 2.16, 95% CI = 1.14–4.11). Upstaging early-stage cases with ILTCs present did not increase model discrimination. While our data do not suggest that endometrial cancer staging guidelines should be revised to include ILTCs, associations between ILTCs and reduced survival observed among stage I cases suggest this tumor feature holds clinical relevance for subgroups of endometrial cancer patients. • In endometrial cancer, the role of transtubal spread as determined by intraluminal tumor cell presence is controversial. • Presence of intraluminal tumor cells was not related to recurrence or overall survival in the overall study population. • Among women with stage 1 disease, intraluminal tumor cells were related to worse recurrence and overall survival. • Upstaging women with early-stage endometrial cancer with intraluminal tumor cells did not affect model discrimination. [ABSTRACT FROM AUTHOR]

Objective: Neuromyelitis optica spectrum disorders (NMOSD) represent rare autoimmune diseases of the central nervous system largely targeting optic nerve(s) and spinal cord. The present analysis used real-world data to identify clinical and epidemiological correlates of treatment change in patients with NMOSD. Methods: CIRCLES is a longitudinal, observational study of NMOSD conducted at 15 centers across North America. Patients with ≥ 60 days of follow-up and receiving on-study maintenance treatment were evaluated. The mean annual relapse rate (ARR) was estimated using negative binomial models; the likelihood of treatment change was estimated using Cox proportional hazards models. Relapses were included as time-varying covariates to estimate the relationship to treatment change. Results: Of 542 patients included, 171 (31.5%) experienced ≥ 1 relapse on the study and 133 patients (24.5%) had ≥ 1 change in the treatment regimen. Two categories of variables significantly correlated with the likelihood of treatment change: (1) relapse: any on-study relapse (hazard ratio [HR] = 2.91; p < 0.001), relapse phenotypes (HR range = 2.15–5.49; p < 0.001), and pre-study ARR > 0.75 (HR 2.28; p < 0.001); 2) disease phenotype: brain syndrome only vs transverse myelitis involvement at onset (HR 2.44; p = 0.008), disease duration < 1 vs > 5 years (HR 1.66; p = 0.028), or autoimmune comorbidity (HR 1.55; p = 0.015). A subset of these factors significantly correlated with shorter time to first rituximab discontinuation. Conclusions: In CIRCLES, relapse patterns and disease phenotype significantly correlated with changes in the maintenance treatment regimen. Such findings may facilitate the identification of patients with NMOSD who are likely to benefit from treatment change to reduce relapse risk or disease burden and enhance the quality of life. [ABSTRACT FROM AUTHOR]

Simple Summary: Novel hormonal therapies (such as abiraterone and enzalutamide) and docetaxel are approved treatments for metastatic prostate cancer. Upfront use of these agents has been shown to improve overall survival. However, we do not know the real-world treatment patterns of these agents or the comparative effectiveness of these agents after treatment with a prior novel hormonal therapy in patients with metastatic prostate cancer. In this large study, we found that most patients with metastatic prostate cancer received only androgen deprivation therapy as upfront therapy without novel hormonal therapies or docetaxel. In patients treated with one novel hormonal therapy, alternate novel hormonal therapy was the most common next therapy and was associated with improved overall survival over docetaxel with the caveat of this being a non-randomized comparison. The study's limitations also include its retrospective design. Background: Both novel hormonal therapies and docetaxel are approved for treatment of metastatic prostate cancer (mPC; in castration sensitive or refractory settings). Present knowledge gaps include lack of real-world data on treatment patterns in patients with newly diagnosed mPC, and comparative effectiveness of novel hormonal therapies (NHT) versus docetaxel after treatment with a prior NHT. Methods: Herein we extracted patient-level data from a large real-world database of patients with mPC in United States. Utilization of NHT or docetaxel for mPC and comparative effectiveness of an alternate NHT versus docetaxel after one prior NHT was evaluated. Comparative effectiveness was examined via Cox proportional hazards model with propensity score matching weights. Each patient's propensity for treatment was modeled via random forest based on 22 factors potentially driving treatment selection. Results: The majority of patients (54%) received only androgen deprivation therapy for mPC. In patients treated with an NHT, alternate NHT was the most common next therapy and was associated with improved median overall survival over docetaxel (abiraterone followed by docetaxel vs. enzalutamide (8.7 vs. 15.6 months; adjusted hazards ratio; aHR 1.32; p = 0.009; and enzalutamide followed by docetaxel vs. abiraterone (9.7 vs. 13.2 months aHR 1.40; p = 0.009). Limitations of the study include retrospective design. [ABSTRACT FROM AUTHOR]