Prognostic models based on survival data frequently make use of the Cox proportional hazards model. Developing reliable Cox models with few events relative to the number of predictors can be challenging, even in low-dimensional datasets, with a much larger number of observations than variables. In such a setting we examined the performance of methods used to estimate a Cox model, including (i) full model using all available predictors and estimated by standard tech-niques, (ii) backward elimination (BE), (iii) ridge regression, (iv) least absolute shrinkage and selec-tion operator (lasso), and (v) elastic net. Based on a prospective cohort of patients with manifest coronary artery disease (CAD), we performed a simulation study to compare the predictive accu-racy, calibration, and discrimination of these approaches. Candidate predictors for incident cardio-vascular events we used included clinical variables, biomarkers, and a selection of genetic variants associated with CAD. The penalized methods, i.e., ridge, lasso, and elastic net, showed a compara-ble performance, in terms of predictive accuracy, calibration, and discrimination, and outperformed BE and the full model. Excessive shrinkage was observed in some cases for the penalized methods, mostly on the simulation scenarios having the lowest ratio of a number of events to the number of variables. We conclude that in similar settings, these three penalized methods can be used interchangeably. The full model and backward elimination are not recommended in rare event scenarios.

BACKGROUNDIn an earlier study, we identified a locus for Moyamoya disease (MMD) on 17q25.3.

METHODSLinkage analysis and fine mapping were conducted for two new families in additional to the previously studied 15 families. Three genes, CARD14, Raptor, and AATK, were selected based on key words, namely, "inflammation", "apoptosis", "proliferation", and "vascular system", for further sequencing. A segregation analysis of 34 pedigrees was performed, followed by a case-control study in Japanese (90 cases vs. 384 controls), Korean (41 cases vs. 223 controls), Chinese (23 cases and 100 controls), and Caucasian (25 cases and 164 controls) populations.

RESULTSLinkage analysis increased the LOD score from 8.07 to 9.67 on 17q25.3. Fine mapping narrowed the linkage signal to a 2.1-Mb region. Sequencing revealed that only one newly identified polymorphism, ss161110142, which was located at position -1480 from the transcription site of the Raptor gene, was common to all four unrelated sequenced familial affected individuals. ss161110142 was then shown to segregate in the 34 pedigrees studied, resulting in a two-point LOD score of 14.2 (P = 3.89 × 10(-8)). Its penetrance was estimated to be 74.0%. Among the Asian populations tested (Japanese, Korean, and Chinese), the rare allele was much more frequent in cases (26, 33, and 4%, respectively) than in controls (1, 1, and 0%, respectively) and was associated with an increased odds ratio of 52.2 (95% confidence interval 27.2-100.2) (P = 2.5 × 10(-49)). This allele was, however, not detected in the Caucasian samples. Its population attributable risk was estimated to be 49% in the Japanese population, 66% in the Korean population, and 9% in the Chinese population.

CONCLUSIONss161110142 may confer susceptibility to MMD among East Asian populations.

ELECTRONIC SUPPLEMENTARY MATERIALThe online version of this article (doi:10.1007/s12199-009-0116-7) contains supplementary material, which is available to authorized users.