Home About us Contact
|
Home About us Contact | ||
|
|
||
Cox Models (cox + models)
Selected AbstractsHigh-Dimensional Cox Models: The Choice of Penalty as Part of the Model Building ProcessBIOMETRICAL JOURNAL, Issue 1 2010Axel Benner Abstract The Cox proportional hazards regression model is the most popular approach to model covariate information for survival times. In this context, the development of high-dimensional models where the number of covariates is much larger than the number of observations ( ) is an ongoing challenge. A practicable approach is to use ridge penalized Cox regression in such situations. Beside focussing on finding the best prediction rule, one is often interested in determining a subset of covariates that are the most important ones for prognosis. This could be a gene set in the biostatistical analysis of microarray data. Covariate selection can then, for example, be done by L1 -penalized Cox regression using the lasso (Tibshirani (1997). Statistics in Medicine16, 385,395). Several approaches beyond the lasso, that incorporate covariate selection, have been developed in recent years. This includes modifications of the lasso as well as nonconvex variants such as smoothly clipped absolute deviation (SCAD) (Fan and Li (2001). Journal of the American Statistical Association96, 1348,1360; Fan and Li (2002). The Annals of Statistics30, 74,99). The purpose of this article is to implement them practically into the model building process when analyzing high-dimensional data with the Cox proportional hazards model. To evaluate penalized regression models beyond the lasso, we included SCAD variants and the adaptive lasso (Zou (2006). Journal of the American Statistical Association101, 1418,1429). We compare them with "standard" applications such as ridge regression, the lasso, and the elastic net. Predictive accuracy, features of variable selection, and estimation bias will be studied to assess the practical use of these methods. We observed that the performance of SCAD and adaptive lasso is highly dependent on nontrivial preselection procedures. A practical solution to this problem does not yet exist. Since there is high risk of missing relevant covariates when using SCAD or adaptive lasso applied after an inappropriate initial selection step, we recommend to stay with lasso or the elastic net in actual data applications. But with respect to the promising results for truly sparse models, we see some advantage of SCAD and adaptive lasso, if better preselection procedures would be available. This requires further methodological research. [source] TGFB1 and TGFBR1 polymorphic variants in relationship to bladder cancer risk and prognosisINTERNATIONAL JOURNAL OF CANCER, Issue 3 2009Adela Castillejo Abstract The transforming growth factor-beta (TGF-,) signalling pathway plays an important role in tumor development and progression. We aimed at analyzing whether 7 different common variants in genes coding for 2 key members of the TGF-, signalling pathway (TGFB1 and TGFBR1) are associated with bladder cancer risk and prognosis. A total of 1,157 cases with urothelial cell carcinoma of the bladder and 1,157 matched controls where genotyped for 3 single nucleotide polymorphisms (SNPs) in TGFB1 (rs1982073, rs1800472, rs1800471) and an additional 3 SNPs and 1 indel polymorphism in TGFBR1 (rs868, rs928180, rs334358 and rs11466445, respectively). In the case-control study, we estimated odds ratios and 95% confidence intervals for each individual genetic variant using unconditional logistic regression adjusting for age, gender, study area and smoking status. Survival analysis was performed using the Kaplan-Meier method and Cox models. The endpoints of interest were tumor relapse, progression and death from bladder cancer. All the SNPs analyzed showed a similar distribution among cases and controls. The distribution of the TGFBR1*6A allele (rs11466445) was also similar among cases and controls, indicating no association with bladder cancer risk. Similarly, none of the haplotypes was significantly associated with bladder cancer risk. Among patients with muscle-invasive tumors, we found a significant association between TGFBR1- rs868 and disease-specific mortality with an allele dosage effect (p -trend = 0.003). In conclusion, the genetic variants analyzed were not associated with an increased risk of bladder cancer. The association of TGFBR1- rs868 with outcome should be validated in independent patient series. © 2008 Wiley-Liss, Inc. [source] Time to clearance of human papillomavirus infection by type and human immunodeficiency virus serostatusINTERNATIONAL JOURNAL OF CANCER, Issue 7 2006Jill E. Koshiol Abstract Persistent infection with high-risk human papillomavirus (HPV) is central to cervical carcinogenesis. Certain high-risk types, such as HPV16, may be more persistent than other HPV types, and type-specific HPV persistence may differ by HIV serostatus. This study evaluated the association between HPV type and clearance of HPV infections in 522 HIV-seropositive and 279 HIV-seronegative participants in the HIV Epidemiology Research Study (HERS, United States, 1993,2000). Type-specific HPV infections were detected using MY09/MY11/HMB01-based PCR and 26 HPV type-specific probes. The estimated duration of type-specific infections was measured from the first HPV-positive visit to the first of two consecutive negative visits. Hazard ratios (HRs) and 95% confidence intervals (CIs) for HPV clearance were calculated using Cox models adjusted for study site and risk behavior (sexual or injection drugs). A total of 1,800 HPV infections were detected in 801 women with 4.4 years median follow-up. HRs for clearance of HPV16 and related types versus low-risk HPV types were 0.79 (95% CI: 0.64,0.97) in HIV-positive women and 0.86 (95% CI: 0.59,1.27) in HIV-negative women. HRs for HPV18 versus low-risk types were 0.80 (95% CI: 0.56,1.16) and 0.57 (95% CI: 0.22,1.45) for HIV-positive and -negative women, respectively. HPV types within the high-risk category had low estimated clearance rates relative to low-risk types, but HRs were not substantially modified by HIV serostatus. © 2006 Wiley-Liss, Inc. [source] Recycled carbon fiber filled polyethylene compositesJOURNAL OF APPLIED POLYMER SCIENCE, Issue 3 2008Tony McNally Abstract Composites of recycled carbon fiber (CF) with up to 30 wt % loading with polyethylene (PE) were prepared via melt compounding. The morphology of the composites and the degree of dispersion of the CF in the PE matrix was examined using scanning electron microscopy, and revealed the CF to be highly dispersed at all loadings and strong interfacial adhesion to exist between the CF and PE. Raman and FTIR spectroscopy were used to characterize the surface chemistry and potential bonding sites of recycled CF. Both the Young's modulus and ultimate tensile stress increased with increasing CF loading, but the percentage stress at break was unchanged up to 5 wt % loading, then decreased with further successive addition of CF. The effect of CF on the elastic modulus of PE was examined using the Halpin-Tsai and modified Cox models, the former giving a better fit with the values determined experimentally. The electrical conductivity of the PE matrix was enhanced by about 11 orders of magnitude on addition of recycled CF with a percolation threshold of 7 and 15 wt % for 500-,m and 3-mm thick samples. © 2007 Wiley Periodicals, Inc. J Appl Polym Sci, 2008 [source] Quantitation of cytomegalovirus (CMV) DNA by real-time PCR for occurrence of CMV disease in HIV-infected patients receiving highly active antiretroviral therapyJOURNAL OF MEDICAL VIROLOGY, Issue 3 2003Karine Gourlain Abstract In HIV-infected patients treated with highly active antiretroviral therapy (HAART) included in the Predivir cohort, we have evaluated the usefulness of CMV DNA quantitation by a TaqMan® PCR assay from peripheral blood leukocytes (PBLs) to predict CMV disease occurrence. In parallel with the immune restoration after treatment by HAART, the percentage of positive samples decreased progressively from 7.3% at Day 0 to 3.5% at Month 12. Among the CMV markers, the smallest concordance with PBL CMV TaqMan® PCR, as evaluated by kappa, was observed with pp65 antigenemia, whereas concordance with all other CMV markers was high. Among the 16 patients with CMV DNA copies at least once >100/150,000 cells, CMV disease occurred in six during follow-up, whereas among the 159 patients with CMV DNA copies always <10/150,000 cells, CMV disease occurred in three and among the seven patients with CMV DNA copies >10 and <100 occurred in only one. In univariate Cox models, all the CMV markers including PBL CMV TaqMan® PCR >10/150,000 cells (RR: 27.6, IC95: 7.1,107.2), the CD4 cell count <75 cells/mm3 and the HIV viral load >100,000 copies/ml were predictive for CMV disease. In a stepwise multivariate analysis, which should be interpreted with caution due to the small number of events (n = 10), three covariates were associated independently with CMV disease: pp65 antigenemia >100 nuclei/200,000, PBL CMV TaqMan® PCR >10 copies/150,000 cells and HIV viral load remaining or increasing >100,000 copies/ml. J. Med. Virol. 69:401,407, 2003. © 2003 Wiley-Liss, Inc. [source] Ascites improves upon plus serum sodium model for end-stage liver disease (MELD) for predicting mortality in patients with advanced liver diseaseALIMENTARY PHARMACOLOGY & THERAPEUTICS, Issue 7 2009M. SOMSOUK Summary Background, The clinical impact of ascites has historically been well recognized; however, its value is unclear in the context of current prognostic models. Aim, To determine whether ascites can improve risk discrimination beyond model for end-stage liver disease (MELD) and serum sodium (MELDNa). Methods, Consecutive cirrhotic patients were evaluated for ascites on the basis of an outpatient CT along with concurrent MELD and Na values. Cox models were used to determine the added value of ascites for predicting 1-year mortality. Increases in the C-index, integrated discrimination improvement (IDI) and the net reclassification index (NRI) were used to assess improvements in discrimination after the addition of ascites. Results, A total of 1003 patients had Na and MELD scores available within 30 days of the CT scan. A total of 60 deaths occurred within 1 year, with mortality higher in patients with ascites (21.4% vs. 4.0%, HR 6.08, 95% CI 3.62,10.19, P < 0.0005). In the presence of ascites, the MELD and MELDNa scores underestimated mortality risk when the scores were less than 21. The addition of ascites to the MELDNa model substantially improved discrimination by the C-index (0.804 vs. 0.770, increase of 3.4%, 95% CI 0.2,9.9%), IDI (1.8%, P = 0.016) and NRI (15.8%, P = 0.0006). Conclusion, The incorporation of radiographic ascites significantly improves upon MELDNa for predicting 1-year mortality. The presence of ascites may help identify patients at increased risk for mortality, not otherwise captured by either MELD or MELDNa. [source] Mortality among sheet metal workers participating in a medical screening programAMERICAN JOURNAL OF INDUSTRIAL MEDICINE, Issue 8 2009John Dement PhD Abstract Background The Sheet Metal Occupational Health Institute Trust (SMOHIT) was formed in 1985 to examine the health hazards of the sheet metal industry in the U.S. and Canada through an asbestos disease screening program. A study of mortality patterns among screening program participants was undertaken. Methods A cohort of 17,345 individuals with 20 or more years in the trade and who participated in the asbestos disease screening program were followed for vital status and causes of death between 1986 and 2004. Data from the screening program included chest X-ray results by International Labour Office (ILO) criteria and smoking history. Standardized mortality ratios (SMRs) by cause were generated using U.S. death rates and Cox proportional hazards models were used to investigate lung cancer risk relative to chest X-ray changes while controlling for smoking. Results A significantly reduced SMR of 0.83 (95% CI,=,0.80,0.85) was observed for all causes combined. Statistically significant excess mortality was observed for pleural cancers, mesothelioma, and asbestosis in the SMR analyses. Both lung cancer and COPD SMRs increased consistently and strongly with increasing ILO profusion score. In Cox models, which controlled for smoking, increased lung cancer risk was observed among workers with ILO scores of 0/1 (RR,=,1.17, 95% CI,=,0.89,1.54), with a strong trend for increasing lung cancer risk with increasing ILO profusion score >0/0. Conclusions Sheet metal workers are at increased risk for asbestos-related diseases. This study contributes to the literature demonstrating asbestos-related diseases among workers with largely indirect exposures and supports an increased lung cancer risk among workers with low ILO profusion scores. Am. J. Ind. Med. 52:603,613, 2009. © 2009 Wiley-Liss, Inc. [source] Carbon black and lung cancer: Testing a new exposure metric in a German cohortAMERICAN JOURNAL OF INDUSTRIAL MEDICINE, Issue 8 2007Peter Morfeld PhD Abstract Background A cohort mortality study of a large carbon black production plant in Germany showed an elevation in lung cancer mortality (SMR 1.81 (95% CI: 1.34,2.39)), although the elevation could not be linked to exposure to carbon black. Methods In follow up to a British study of carbon black production workers [Sorahan et al., 2007] in which risk of lung cancer progressively declined after cessation of employment,in contrast to an expected upward trend,we evaluated the German cohort with a similar methodology, that is, by focusing on the first 15 years after leaving employment in terms of lung cancer SMR. Results In our SMR analysis of the German cohort of 1,528 men and an inception cohort of 1,271 men, distinctly different results from the British cohort were observed. We observed a rising trend in lung cancer SMR, in contrast to the declining SMR trend noted in the British cohort. In fact, Cox models for lung cancer mortality with attained age as the basic time variable showed negative coefficients for the full and inception cohorts. Conclusions Our analysis of a large German cohort of carbon black workers does not support the concept of a declining risk of lung cancer following cessation of employment. Am. J. Ind. Med. 50:565,567, 2007. © 2007 Wiley-Liss, Inc. [source] Racial and Ethnic Differences in Mortality in Children Awaiting Heart Transplant in the United StatesAMERICAN JOURNAL OF TRANSPLANTATION, Issue 12 2009T. P. Singh Racial differences in outcomes are well known in children after heart transplant (HT) but not in children awaiting HT. We assessed racial and ethnic differences in wait-list mortality in children <18 years old listed for primary HT in the United States during 1999,2006 using multivariable Cox models. Of 3299 listed children, 58% were listed as white, 20% as black, 16% as Hispanic, 3% as Asian and 3% were defined as ,Other'. Mortality on the wait-list was 14%, 19%, 21%, 17% and 27% for white, black, Hispanic, Asian and Other children, respectively. Black (hazard ratio [HR] 1.6, 95% confidence interval [CI] 1.3, 1.9), Hispanic (HR 1.5, CI 1.2, 1.9), Asian (HR, 2.0, CI 1.3, 3.3) and Other children (HR 2.3, CI 1.5, 3.4) were all at higher risk of wait-list death compared to white children after controlling for age, listing status, cardiac diagnosis, hemodyamic support, renal function and blood group. After adjusting additionally for medical insurance and area household income, the risk remained higher for all minorities. We conclude that minority children listed for HT have significantly higher wait-list mortality compared to white children. Socioeconomic variables appear to explain a small fraction of this increased risk. [source] Joint Modelling of Repeated Transitions in Follow-up Data , A Case Study on Breast Cancer DataBIOMETRICAL JOURNAL, Issue 3 2005B. Genser Abstract In longitudinal studies where time to a final event is the ultimate outcome often information is available about intermediate events the individuals may experience during the observation period. Even though many extensions of the Cox proportional hazards model have been proposed to model such multivariate time-to-event data these approaches are still very rarely applied to real datasets. The aim of this paper is to illustrate the application of extended Cox models for multiple time-to-event data and to show their implementation in popular statistical software packages. We demonstrate a systematic way of jointly modelling similar or repeated transitions in follow-up data by analysing an event-history dataset consisting of 270 breast cancer patients, that were followed-up for different clinical events during treatment in metastatic disease. First, we show how this methodology can also be applied to non Markovian stochastic processes by representing these processes as "conditional" Markov processes. Secondly, we compare the application of different Cox-related approaches to the breast cancer data by varying their key model components (i.e. analysis time scale, risk set and baseline hazard function). Our study showed that extended Cox models are a powerful tool for analysing complex event history datasets since the approach can address many dynamic data features such as multiple time scales, dynamic risk sets, time-varying covariates, transition by covariate interactions, autoregressive dependence or intra-subject correlation. (© 2005 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim) [source] Does perineural invasion on prostate biopsy predict adverse prostatectomy outcomes?BJU INTERNATIONAL, Issue 11 2010Stacy Loeb Study Type , Prognostic (case series) Level of Evidence 4 OBJECTIVE To determine the relationship between perineural invasion (PNI) on prostate biopsy and radical prostatectomy (RP) outcomes in a contemporary RP series, as there is conflicting evidence on the prognostic significance of PNI in prostate needle biopsy specimens. PATIENTS AND METHODS From 2002 to 2007, 1256 men had RP by one surgeon. Multivariable logistic regression and Cox proportional hazards models were used to examine the relationship of PNI with pathological tumour features and biochemical progression, respectively, after adjusting for prostate-specific antigen level, clinical stage and biopsy Gleason score. Additional Cox models were used to examine the relationship between nerve-sparing and biochemical progression among men with PNI. RESULTS PNI was found in 188 (15%) patients, and was significantly associated with aggressive pathology and biochemical progression. On multivariate analysis, PNI was significantly associated with extraprostatic extension and seminal vesicle invasion (P < 0.001). Biochemical progression occurred in 10.5% of patients with PNI, vs 3.5% of those without PNI (unadjusted hazard ratio 3.12, 95% confidence interval 1.77,5.52, P < 0.001). However, PNI was not a significant independent predictor of biochemical progression on multivariate analysis. Finally, nerve-sparing did not adversely affect biochemical progression even among men with PNI. CONCLUSION PNI is an independent risk factor for aggressive pathology features and a non-independent risk factor for biochemical progression after RP. However, bilateral nerve-sparing surgery did not compromise the oncological outcomes for patients with PNI on biopsy. [source] Residual serum monoclonal protein predicts progression-free survival in patients with previously untreated multiple myelomaCANCER, Issue 3 2010Eric W. Schaefer MS Abstract BACKGROUND: Currently used treatment response criteria in multiple myeloma (MM) are based in part on serum monoclonal protein (M-protein) measurements. A drawback of these criteria is that response is determined solely by the best level of M-protein reduction, without considering the serial trend. The authors hypothesized that metrics incorporating the serial trend of M-protein would be better predictors of progression-free survival (PFS). METHODS: Fifty-five patients with measurable disease at baseline (M-protein ,1 g/dL) who received ,4 cycles of treatment from 2 clinical trials in previously untreated MM were included. Three metrics based on the percentage of M-protein remaining relative to baseline (residual M-protein) were considered: metrics based on the number of times residual M-protein fell within prespecified thresholds, metrics based on area under the residual M-protein curve, and metrics based on the average residual M-protein reduction between Cycles 1 and 4. The predictive value of these metrics was assessed in Cox models using landmark analysis. RESULTS: The average residual M-protein reduction was found to be significantly predictive of PFS (P = .02; hazard ratio, 0.37), in which a patient with a 10% lower average residual M-protein reduction from Cycle 1 to 4 was estimated to be at least 2.7× more likely to develop disease progression or die early. None of the other metrics was predictive of PFS. The concordance index for the average residual M-protein reduction was 0.63, compared with 0.56 for best response. CONCLUSIONS: The average residual M-protein reduction metric is promising and needs further validation. This exploratory analysis is the first step in the search for treatment-based trend metrics predictive of outcomes in MM. Cancer 2010. © 2009 American Cancer Society. [source] Prospective investigation of the cigarette smoking,head and neck cancer association by sex,,CANCER, Issue 7 2007Neal D. Freedman PhD Abstract BACKGROUND. Men are approximately 3 times more likely to develop squamous cancers of the head and neck (oral cavity, pharynx, and larynx) than women. Very few prospective studies have examined the association between cigarette smoking and cancers of the head and neck in women, even though the rates of smoking in women are increasing rapidly worldwide. METHODS. The association between cigarette smoking and head and neck cancer was investigated in 476,211 participants, aged 50,71 years, of the National Institutes of Health/American Association of Retired Persons (NIH-AARP) diet and health study by using age-standardized incidence rates and hazard ratios from Cox models adjusted for other risk factors for these cancers. RESULTS. Over the course of follow-up (1995 through 2000), 584 men and 175 women were diagnosed with head and neck cancer. Nonsmoking (24.4), former smoking (36.9), and current smoking (147.3) men had higher rates of incident head and neck cancer per 100,000 person-years of follow-up than women did in each equivalent category of cigarette use (non: 4.8; former: 17.2; current: 75.7). The hazard ratios associated with smoking were significantly larger in women (12.96; 95% confidence interval [CI], 7.81,21.52) than in men (5.45, 95% CI, 4.22,7.05; P for interaction: <.001) for head and neck cancer overall and also for the 3 subsites (oral cavity, oro-hypopharynx, and larynx) examined in stratified analyses. Ever-smoking accounted for 45% of head and neck cancers in men and 75% in women, assuming causality. CONCLUSIONS. Cigarette smoking is a strong risk factor for head and neck cancer in both men and women. Incidence rates of head and neck cancer were higher in male smokers than female smokers, but smoking may explain a higher proportion of head and neck cancer in women than in men. Cancer 2007. Published 2007 by the American Cancer Society. [source] Racial differences in tumor stage and survival for colorectal cancer in an insured population,CANCER, Issue 3 2007Chyke A. Doubeni MD Abstract BACKGROUND. Despite declining death rates from colorectal cancer (CRC), racial disparities have continued to increase. In this study, the authors examined disparities in a racially diverse group of insured patients. METHODS. This study was conducted among patients who were diagnosed with CRC from 1993 to 1998, when they were enrolled in integrated healthcare systems. Patients were identified from tumor registries and were linked to information in administrative databases. The sample was restricted to non-Hispanic whites (n = 10,585), non-Hispanic blacks (n = 1479), Hispanics (n = 985), and Asians/Pacific Islanders (n = 909). Differences in tumor stage and survival were analyzed by using polytomous and Cox regression models, respectively. RESULTS. In multivariable regression analyses, blacks were more likely than whites to have distant or unstaged tumors. In Cox models that were adjusted for nonmutable factors, blacks had a higher risk of death from CRC (hazard ratio [HR], 1.17; 95% confidence interval [95% CI], 1.06,1.30). Hispanics had a risk of death similar to whites (HR, 1.04; 95% CI, 0.92,1.18), whereas Asians/Pacific Islanders had a lower risk of death from CRC (HR, 0.89; 95% CI, 0.78,1.02). Adjustment for tumor stage decreased the HR to 1.11 for blacks, and the addition of receipt of surgical therapy to the model decreased the HR further to 1.06. The HR among Hispanics and Asians/Pacific Islanders was stable to adjustment for tumor stage and surgical therapy. CONCLUSIONS. The relation between race and survival from CRC was complex and appeared to be related to differences in tumor stage and therapy received, even in insured populations. Targeted interventions to improve the use of effective screening and treatment among vulnerable populations may be needed to eliminate disparities in CRC. Cancer 2007;109:612,620. © 2006 American Cancer Society. [source] | ||||||||||