Extended Cox Regression Model Research Articles

Diabetes Is Associated With Reduced Stress Hyperlactatemia in Cardiac Surgery.

Hyperglycemia and hyperlactatemia are associated with increased morbidity and mortality in critical illness. We evaluated the relationship among hyperlactatemia, glycemic control, and diabetes mellitus (DM) after cardiac surgery. This was a retrospective cohort study of 4,098 cardiac surgery patients treated between 2011 and 2015. Patients were stratified by DM and glucose-lowering medication history. Hyperglycemia (glucose >180 mg/dL), hypoglycemia (<70 mg/dL), and the hyperglycemic index were assessed postoperatively (48 h). The relationship between lactate and glucose levels was modeled using generalized linear regression. Mortality was analyzed using an extended Cox regression model. Hyperglycemia occurred in 26.0% of patients without DM (NODM), 46.5% with DM without prior drug treatment (DMNT), 62.8% on oral medication (DMOM), and 73.8% on insulin therapy (DMIT) (P < 0.0001). Hypoglycemia occurred in 6.3%, 9.1%, 8.8%, and 10.8% of NODM, DMNT, DMOM, and DMIT, respectively (P = 0.0012). The lactate levels of all patients were temporarily increased with surgery. This increase was greater in patients who also had hyperglycemia or hypoglycemia and was markedly attenuated in patients with DM. Peak lactate was 5.8 mmol/L (95% CI 5.6, 6.0) in NODM with hyperglycemia vs. 3.3 (95% CI 3.2, 3.4) without hyperglycemia; in DMNT: 4.8 (95% CI 4.4, 5.2) vs. 3.4 (95% CI 3.1, 3.6); in DMOM: 3.8 (95% CI 3.5, 4.1) vs. 2.9 (95% CI 2.7, 3.1); and in DMIT: 3.3 (95% CI 3.0, 3.5) vs. 2.7 (95% CI 2.3, 3.0). Increasing lactate levels were associated with increasing mortality; increasing glucose reduced this effect in DM but not in NODM (P = 0.0069 for three-way interaction). Stress hyperlactatemia is markedly attenuated in patients with DM. There is a three-way interaction among DM, stress hyperlactatemia, and stress hyperglycemia associated with mortality after cardiac surgery.

Association of Multiple Primary Skin Cancers With Human Immunodeficiency Virus Infection, CD4 Count, and Viral Load

Persons with human immunodeficiency virus (HIV) have a 2.8-fold higher risk than HIV-uninfected persons of nonmelanoma skin cancer (NMSC), defined as basal cell carcinoma (BCC) and squamous cell carcinoma (SCC). Individuals with a prior NMSC history are at increased risk for subsequent NMSC, but the magnitude of risk and its relation to HIV disease-related factors, including CD4 count and viral load (VL), are unknown. To better understand how laboratory markers currently used to evaluate HIV disease progression may be associated with subsequent NMSC risk. This cohort study analyzed 455 HIV-infected and 1945 HIV-uninfected patients, all of them members of the Kaiser Permanente Northern California (KPNC) health care plan, diagnosed with at least 1 NMSC from 1996-2008 to determine risk of subsequent NMSCs in relation to CD4 count and VL. All participants were white, non-Hispanic persons 18 years or older who had had at least 1 NMSC during the 1996-2008 period. Participants entered the cohort at their first NMSC diagnosis and were observed through 2008. Incidence rates were calculated and adjusted hazard ratios were estimated using extended Cox regression models with recent CD4 count and VL analyzed as time-changing covariates. Measured CD4 count, VL, and subsequent NMSC (BCC and SCC). The cohort comprised 455 HIV-infected participants (13 [3%] women) and 1952 HIV-uninfected participants (154 [8%] women). Median duration of observation was 4.6 years, and 16.5% (n = 390) either died (n = 35) or lost KPNC membership status (n = 355) without having a subsequent primary NMSC. Compared with HIV-uninfected persons, HIV-infected individuals were slightly younger (mean age, 52.5 vs 55.5 years), more likely men (97% vs 92%), more likely to have smoked (57% vs 45%), and less likely to be overweight/obese (50% vs 61%). The small observed differences by HIV status in matching characteristics (ie, age and sex) resulted from the restriction of the original cohort to those with at least 1 NMSC. Compared with uninfected individuals, those with HIV infection with a recent biomarker of more severe immune deficiency (CD4 count <200 cells/mL) had a 44% increased risk of subsequent NMSC overall and a 222% increase risk of SCC in particular, suggesting that subsequent SCC risk is associated with immune dysfunction. HIV-infected persons compared with HIV-uninfected persons were are at higher risk for subsequent new SCC but not BCC, with a dose-response relationship between risk and lower CD4 counts and higher VLs. Subsequent new primary SCCs had a strong association with lower CD4 and higher VL among HIV-infected persons, suggesting that immune dysfunction might contribute to increased SCC risk. Clinical implications include targeted monitoring for SCC among HIV-infected individuals, particularly those with low CD4 counts or high VLs.

Time-dependent effects of prognostic biomarkers of systemic inflammation in patients with metastatic renal cell carcinoma.

The goal of this study was to examine time-dependent effects of prognostic biomarkers of systemic inflammation in patients with metastatic renal cell carcinoma. Retrospective chart reviews were conducted at the Winship Cancer Institute of Emory University and the Atlanta Veterans Administration Medical Center with authorization from the Emory University Institutional Review Board and the Veterans Administration Research and Development Committee. Inclusion criteria included age ⩾18 years, treatment with targeted therapy for clear cell or non-clear cell metastatic renal cell carcinoma and concomitant assessment of C-reactive protein and albumin levels on ⩾3 occasions that were ⩾10 days apart. Discovery, expansion, and external validation cohorts were identified. Established prognostic variables were evaluated by univariate and multivariate analyses. Intensity of systemic inflammation was assessed at all time points with C-reactive protein and albumin as prognostic covariates for overall survival in an extended Cox regression model. Intensity of systemic inflammation was assessed on 3186 occasions in 181 patients. Risk status changed in 131 patients (72%). The hazard ratio for overall survival was 21.41 (95% confidence interval = 8.26-55.50) with a type 3 p value of <0.001 for the reference cohort and 9.68 (2.07-45.31) with a type-3 p value of 0.006 for the external validation cohort when time points associated with severe systemic inflammation were compared to all other time points. The bias-corrected c-statistic was 0.839 (0.773-0.905) and 0.818 (0.691-0.946), respectively. Terminal disease progression with severe systemic inflammation was detected in 87% of the 90 patients who died. In conclusion, time-dependent effects are a prominent feature of intensity of systemic inflammation, a powerful prognostic biomarker for metastatic renal cell carcinoma.

Impact of salvage regimens on response and overall survival in acute myeloid leukemia with induction failure.

We evaluated the impact of salvage regimens and allogeneic hematopoietic cell transplantation (allo-HCT) in acute myeloid leukemia (AML) with induction failure. Between 1993 and 2009, 3324 patients with newly diagnosed AML were enrolled in 5 prospective treatment trials of the German-Austrian AML Study Group. After first induction therapy with idarubicin, cytarabine and etoposide (ICE), 845 patients had refractory disease. In addition, 180 patients, although responding to first induction, relapsed after second induction therapy. Of the 1025 patients with induction failure, 875 (median age 55 years) received intensive salvage therapy: 7+3-based (n=59), high-dose cytarabine combined with mitoxantrone (HAM; n=150), with all-trans retinoic acid (A; A-HAM) (n=247), with gemtuzumab ozogamicin and A (GO; GO-A-HAM) (n=140), other intensive regimens (n=165), experimental treatment (n=27) and direct allo-HCT (n=87). In patients receiving intensive salvage chemotherapy (n=761), response (complete remission/complete remission with incomplete hematological recovery (CR/CRi)) was associated with GO-A-HAM treatment (odds ratio (OR), 1.93; P=0.002), high-risk cytogenetics (OR, 0.62; P=0.006) and age (OR for a 10-year difference, 0.75; P<0.0001). Better survival probabilities were seen in an extended Cox regression model with time-dependent covariables in patients responding to salvage therapy (P<0.0001) and having the possibility to perform an allo-HCT (P<0.0001). FLT3 internal tandem duplication, mutated IDH1 and adverse cytogenetics were unfavorable factors for survival.

Joint modeling of longitudinal cholesterol measurements and time to onset of dementia in an elderly African American Cohort

ABSTRACTThis paper presents a statistical method for analyzing the association between longitudinal cholesterol measurements and the timing of onset of dementia. The proposed approach jointly models the longitudinal and survival processes for each individual on the basis of a shared random effect, where a linear mixed effects model is assumed for the longitudinal component and an extended Cox regression model is employed for the survival component. A dynamic prediction model is built based on the joint model, which provides prediction of the conditional survival probabilities at different time points using available longitudinal measurements as well as baseline characteristics. We apply our method to the Indianapolis-Ibadan Dementia project, a 20-year study of dementia in elderly African Americans living in Indianapolis, Indiana. We find that with baseline covariates and comorbidities adjusted, the risk of dementia decreases by 1% per one mg/dl increase in total cholesterol. Therefore we conclude that, in a healthy cohort of African Americans aged 65 years or more, high late-life cholesterol level is associated with lower incidence of dementia.

Allogeneic Stem Cell Transplantation Improves Survival in Patients with FLT3-ITD-Negative and NPM1-Wild Type Intermediate Risk (ELN) Acute Myeloid Leukemia

Background: So called triple-negative acute myeloid leukemias (AML) form a heterogeneous subgroup of intermediate-risk AML according to ELN criteria. Molecularly this group can be defined by the absence of internal tandem duplications (ITDs) in the fms related tyrosine kinase 3 gene (FLT3), wild type (wt) nucleophosmin gene (NPM1) and wt or single (homozygously or heterozygously) mutated CCAAT/enhancer binding protein α gene (CEBPA). To date, it remains unclear whether this group of patients benefits from allogeneic stem cell transplantation (alloSCT) as consolidation strategy in first complete remission (CR1).Aims: To evaluate the impact of alloSCT on the overall (OS) and relapse free survival (RFS) in patients with triple negative AML in first remission in comparison to post remission chemotherapy (PRT).Methods: We performed a subgroup analysis of 3041 AML patients aged 16-60 years who were enrolled into the AML 96 and the AML 2003 trials of the Study Alliance Leukemia (SAL). Selection criteria for this subgroup were NPM1 wt, negativity for FLT3-ITD and CEPBA double mutations, a karyotype that does not define the AML as favorable or adverse according to ELN criteria and the accomplishment of CR1. Status of molecular markers was evaluated with standard PCR techniques. Within the AML2003 trial, donor status was evaluated at study entry, making these data eligible for a donor-versus-no-donor analysis. Kaplan-Meier estimates were used to report on point estimates for survival probabilities. Multivariate Cox models were fitted to analyze the impact of alloSCT as time-dependent covariate. Age, gender, white blood cell count, lactate dehydrogenase, AML type (de novo, secondary AML following MDS or MPN, or therapy-related myeloid neoplasms) and ECOG performance status at diagnosis were selected as adjusting covariables. As-treated analyses used data from both trials, AML96 and AML2003. For these analyses alloSCT or PRT were entered as time-dependent covariates into extended Cox regression models. Survival outcomes were displayed with Simon-Makuch-plots.Results: In total, 497 patients (AML96: 217, AML2003: 280) with a median age of 47 years were evaluable for the analysis of OS from diagnosis. A total of 302 patients had reached CR1 and could be evaluated for RFS. In a multivariate donor-versus-no-donor analysis, OS of patients with a sibling donor was not significantly different to patients without a donor (HR 0.79, 95%CI 0.53 to 1.16, p=.2). Irrespective of whether the patient actually received alloSCT in CR1, the probability of OS at 5 years from study enrollment was 55% (95%CI, 45% to 67%) for patients with a sibling donor and 47% (95%CI, 40% to 54%) for patients without a donor. For RFS, the hazard ratio was 0.72 (95%CI, 0.5 to 1.05, p=0.09), with a trend in favor of better remission-control for patients with a sibling donor. At five years from CR1, RFS of patients with sibling donor was 48% (95%CI, 38% to 61%) compared to 36% (95%CI, 30% to 44%). However, the transplantation rate in the donor group was only 53% and 15% of patients in the no-donor group actually received alloSCT. Therefore, 'cross-over' effects lowered the power of donor-versus-no-donor analysis.In the multivariable as-treated analysis including patients from AML96 and AML 2003, OS and RFS of patients with alloSCT were significantly longer (OS: HR 0.58, 95%CI, 0.37 to 0.9, p=.02, RFS: 0.51, 95%CI, 0.34 to 0.76, p=0.001) compared to the PRT group. The probability of OS at 5 years from initiation of consolidation treatment (alloSCT vs. PRT) was 66% (95%CI, 57% to 76%) for patients who received alloSCT compared to 46% (95%CI, 38% to 55%) for PRT patients. The probability of RFS at 5 years from initiation of consolidation treatment was 55% (95%CI, 46% to 67%) for alloSCT patients and 31% (95%CI, 24% to 39%) for PRT patients.Conclusions: Due to cross-over effects which limit the power of the donor-versus-no-donor analysis we give more weight to the results of the as-treated analysis. This analysis suggests that eligible intermediate-risk AML patients with NPM1 wt and absent FLT3-ITD benefit from alloSCT in CR1. However, bias introduced by selection and confounding factors cannot be excluded for this type of analysis and could only be circumvented in randomized controlled trials. DisclosuresThiede:AgenDix: Employment, Other: Ownership. Rösler:Janssen: Consultancy, Other: Travel/Accommodation/Expenses. Middeke:Sanofi: Honoraria. Schetelig:Sanofi: Honoraria.

Initiation into prescription opioid injection and associated trends in heroin use among people who use illicit drugs

ObjectivesPrescription opioid (PO) injection among people who use illicit drugs (PWUD) is an ongoing concern, yet little is known about drug use trajectories associated with initiating PO injection, including potential associations with heroin use. This study aimed to identify predictors of PO injection initiation among PWUD, and examine trends in heroin use before and after initiating PO injection. MethodsData were merged from three cohorts of PWUD recruited between September 2005 and November 2015. Predictors of PO injection initiation were identified using extended Cox regression models. Trends in heroin use pre- and post-initiation were examined with McNemar’s test and compared to matched controls with linear growth curve models. ResultsAmong 1580 participants, 247 initiated PO injection yielding an incidence density of 3.9 (95% Confidence Interval [CI]: 3.4–4.4) per 100 person-years. In a multivariable analysis, independent predictors of PO injection initiation included heroin injection (Adjusted Hazard Ratio [AHR]=4.39, 95% CI: 3.24–5.95) and non-injection PO use (AHR=1.99, 95% CI: 1.25–3.17). In a sub-analysis, compared to matched controls, PO injection corresponded with elevated heroin use post-initiation (p≤0.05). DiscussionIn this study, heroin use and non-injection PO use strongly predicted PO injection initiation. Those who initiated PO injecting had elevated heroin use patterns post-initiation compared to controls. These findings suggest that transitioning to PO injection does not appear to be a substitute for heroin use among PWUD. These findings highlight the importance of addressing PO injection in harm reduction and treatment settings.

Usefulness of overnight pulse oximeter as the sleep assessment tool to assess the 6-year risk of road traffic collision: evidence from the Taiwan Bus Driver Cohort Study.

In order to support health service organizations in arranging a system for prevention of road traffic collisions (RTC), it is important to study the usefulness of sleep assessment tools. A cohort study was used to evaluate the effectiveness of subjective and objective sleep assessment tools to assess for the 6-year risk of both first RTC event only and recurrent RTC events. The Taiwan Bus Driver Cohort Study (TBDCS) recruited 1650 professional drivers from a large bus company in Taiwan in 2005. The subjects were interviewed in person, completed the sleep assessment questionnaires and had an overnight pulse oximeter survey. Moreover, this cohort of drivers was linked to the National Traffic Accident Database (NTAD) and researchers found 139 new RTC events from 2005 to 2010. Primary outcomes were traffic collisions from NTAD, nocturnal oxygen desaturation index (ODI) from pulse oximeter, Pittsburg sleeping quality score, Epworth daytime sleepiness score, Snore Outcomes Survey score and working patterns from questionnaires. A Cox proportional hazards model and an extended Cox regression model for repeated events were performed to estimate the hazard ratio for RTC. The RTC drivers had increased ODI4 levels (5.77 ± 4.72 vs 4.99 ± 6.68 events/h; P = 0.008) and ODI3 levels (8.68 ± 6.79 vs 7.42 ± 7.94 events/h; P = 0.007) in comparison with non-RTC drivers. These results were consistent regardless of whether ODI was evaluated as a continuous or a categorical variable. ODI4 and ODI3 levels increased the 6-year RTC risks among professional drivers even after adjusting for age, education, history of cardiovascular disease, caffeine intake, sleeping pills used, bus driving experience and shift modes. Moreover, there was an increased trend for ODI between the stratification of the number of RTCs in comparison with the non-RTC group. In the extended Cox regression models for repeated RTC events with the Anderson and Gill intensity model and Prentice-Williams-Petersen model, measurement of ODI increased hazards of the subsequent RTC events. This study showed that an increase in the 6-year risk of RTC was associated with objective measurement of ODI for a sign of sleep-disordered breathing (SDB), but was not associated with self-reported sleeping quality or daytime sleepiness. Therefore, the overnight pulse oximeter is an effective sleep assessment tool for assessing the risk of RTC. Further research should be conducted regarding measures to prevent against SDB among professional drivers.

Time-to-Cessation of Postoperative Opioids: A Population-Level Analysis of the Veterans Affairs Health Care System.

This study aims to determine 1) the epidemiology of perioperative opioid use; and 2) the association between patterns of preoperative opioid use and time-to-cessation of postoperative opioids. Retrospective, cohort study. National, population-level study of Veterans Healthcare Administration (VHA) electronic clinical data. All VHA patients (n = 64,391) who underwent surgery in 2011, discharged after stays of ≥1 day, and receiving ≥1 opioid prescription within 90 days of discharge. Patients' preoperative opioid use were categorized as 1) no opioids, 2) tramadol only, 3) short-acting (SA) acute/intermittent (≤ 90 days fill), 4) SA chronic (> 90 days fill), or 5) any long-acting (LA). After defining cessation as 90 consecutive, opioid-free days, the authors calculated time-to-opioid-cessation (in days), from day 1 to day 365, after hospital discharge. The authors developed extended Cox regression models with a priori identified predictors. Sensitivity analyses used alternative cessation definitions (30 or 180 consecutive days). Almost 60% of the patients received preoperative opioids: tramadol (7.5%), SA acute/intermittent (24.1%), SA chronic (17.5%), and LA (5.2%). For patients opioid-free preoperatively, median time-to-cessation of opioids postoperatively was 15 days. The SA acute/intermittent cohort (HR =1.96; 95% CI =1.92-2.00) had greater risk for prolonged time-to-cessation than those opioid-free (reference), but lower risk than those taking tramadol only, SA chronic (HR = 9.09; 95% CI = 8.33-9.09), or LA opioids (HR = 9.09; 95% CI = 8.33-10.00). Diagnoses of chronic pain, substance-use, or affective disorders were weaker positive predictors. Sensitivity analyses maintained findings. Greater preoperative levels of opioid use were associated with progressively longer time-to-cessation postoperatively.

Survival of breast cancer patients with synchronous or metachronous central nervous system metastases

BackgroundCentral nervous system (CNS) metastases represent a devastating complication for advanced breast cancer patients. This observational study examines the influence of patient, tumour and treatment characteristics on overall survival after synchronous or metachronous CNS metastases. MethodsInformation on 992 breast cancer patients with CNS metastases (whose primary tumour was diagnosed between 2004 and 2010) was retrieved from the Netherlands Cancer Registry (NCR). Overall survival was calculated from the date of CNS metastatic diagnosis, and the impact of prognostic factors on survival was assessed using univariate and multivariate extended Cox-regression models. ResultsWe identified 165 patients with synchronous and 827 patients with metachronous CNS metastases. The majority of patients (88%) presented with brain metastases only, 12% had leptomeningeal metastases. Overall median survival was 5.0months. Non-triple-negative breast cancer and systemic therapy were associated with improved survival in both groups. In patients with synchronous CNS metastases, surgery for the primary tumour and the metastases also improved survival. In patients with metachronous metastases, younger age (<50years), lower initial tumour stage (I), ductal carcinoma, a prolonged time interval until diagnosis of CNS metastasis (>1year), and absence of extracranial metastases were associated with improved survival. Metastasectomy and radiation therapy did not provide benefit beyond the first six months. ConclusionsNo difference in survival was established between synchronous and metachronous CNS metastases. Triple-negative disease is prognostically unfavourable in both groups, while those receiving treatment have a better outcome. Metastasectomy and radiotherapy improve survival within the first six months, and additional benefit may be derived from systemic therapy.

Time to unsafe sexual practice among cross-border female sex workers in Metemma Yohannes, North West Ethiopia.

BackgroundBecause of the nature of their work, female sex workers are at risk of sexually transmitted diseases. Cross-border areas are places where this situation becomes worse. In Ethiopia, there has been a serious scarcity of studies on the time at which unsafe sexual practice starts and on factors which determine the practice among female sex workers there. Therefore, this study aimed to fill this identified gap.MethodA total of 467 women who had been sex workers at least for three months prior to the resumption of the study were included. A structured and pre-tested questionnaire was used to collect data from July-August, 2010. Descriptive statistics was used to explore the data, and the Extended Cox-Regression model was employed to identify the predictors of time-to-unsafe sexual practice.ResultThe study participants were followed for 6, 643 person-months. The overall incidence density of unsafe sexual practice was 44.71 persons per 1000 persons-months. The hazard of unsafe sexual practice increased by 3.0 % every month (p-value =0.040) due to problem-drinking. Those female sex workers with familiarized clients had a two-fold hazard of practicing unsafe sex compared to their counterparts (AHR = 1.94 95 % CI 1.49, 2.53). The predominant sexual client type and the work place of sex workers were the other significant predictors of unsafe sexual practice.ConclusionsThe incidence of unsafe sexual practice was found to be high among sex workers in the cross-border area. Time-to-unsafe sexual practice was significantly associated with female sex workers’ status of familiarity with their clients, predominant sexual client type, their work place, and the interaction term of time and problem-drinking. Interventions need to be made on these controllable social and behavioral characteristics to help sex workers extend the duration of their safe sexual practice beyond the time they will quit sex work.

HIV-1 Treatment-as-Prevention

The Chinese national observational cohort study suggests that the treatment-as-prevention (TasP) approach can be an effective public health HIV-1 prevention strategy. However, results from that study may have been biased because the follow-up time of index patients prior to their initiation of antiretroviral therapy (ART) was excluded. In this study, we correct for such bias by using an extended time-dependent Cox regression model to conduct a cohort study analysis of serodiscordant couples in Guangxi of China, inclusive of all follow-up time.During the follow-up of this observational cohort study of HIV-1 sero-discordant couples, the positive index partners may have never be treated with ART, or enter untreated but subsequently began treatment, or may have been treated immediately upon entry into the public health system. The treatment effectiveness of ART in HIV-1 acquisition among HIV-negative partners is assessed by the extended Cox regression model with treatment status as a time-varying covariate.A total of 6548 sero-discordant couples were included in the cohort study analysis. Among them, 348 negative partners sero-converted. HIV seroincidence was significantly higher among the nontreated (4.3 per 100 person-years, 3.7–4.9) compared with those receiving ART (1.8 per 100 person-years, 1.5–2.0). An overall 35% reduction in risk of HIV transmission was associated with receiving ART (adjusted hazard ratio [AHR] 0.65, 95% confidence interval [CI] 0.51–0.83), and the yearly risk reduction was also significant in the first 3 consecutive years of follow-up. Moreover, ART was found to be significantly inversely associated with multiple baseline characteristics of index partners.TasP may be feasible on a national or regional scale. In addition to other proven preventive strategies such as the use of condoms, ART adherence to maintain viral suppression would then be the key challenge for successful TasP implementation.

Fitting Time to First Birth Using Extended Cox Regression Model in Presence of Nonproportional Hazard

The Cox regression model, which is widely used for the analysis of factor effects with censored survival data, makes the assumption of constant hazard ratio. Different methods should be used to deal with non-proportionality of hazards when this assumption is violated. In this study, we use the Extended Cox regression model where time dependent covariate terms with fixed functions of time are considered. Time to first birth for the ever married women after marriage, taken from BDHS 2011 women data is fitted using Extended Cox regression model due to the failure of existence of proportionality assumption. This model performs well as expected compared to Cox regression model. DOI: http://dx.doi.org/10.3329/dujs.v63i1.21764 Dhaka Univ. J. Sci. 63(1): 25-30, 2015 (January)

Does Breastfeeding Have an Effect on Infant Mortality in India? An Analysis of National Family Health Survey Data

Introduction: Sub-optimal infant and early childhood feeding practices in India contribute to the high burden of child under-nutrition and infant death. Objective: To study the effect of breast-feeding duration on the survival of infants along with other demographic, socioeconomic and service related factors and to compare the decadal changes in the association. Methods: A total of 36,754 and 26,782 births less than 3 years of age from two rounds of the National Family Health Survey (NFHS) carried out in India during 1992-1993 (NFHS-1), and 2005-2006 (NFHS-3) respectively comprised the sample. Infant mortality was examined against a set of variables and breastfeeding as a time dependent covariate using multivariate extended Cox regression model. Results: Hazard of infant death observed in NFHS-3 (2005-2006) was 18 percent less [HR = 0.82; 95% CI = 0.759 - 0.879] as compared to that observed in NFHS-1 (1992-1993). After adjustment of other factors, the risk of infant death was 97% less amongst children who were breastfed [HR = 0.03; 95% CI = 0.029 - 0.033] as compared to those who were not breastfed. In contrast to rural area, babies belonging to urban area had an 18 percent less hazard of death during the infancy period [HR = 0.82]. Hazard of infant death was less among births to mothers with higher maternal education levels [HR = 0.93] and in those who were in gainful occupation [HR = 0.91]. The risk of infant death was higher in babies perceived to be small at birth [HR = 1.40] and belonging to states with medium post neonatal mortality rate (PNMR) [HR = 1.23; 95% CI = 1.112 - 1.371] and high PNMR [HR = 1.46; 95% CI = 1.327 - 1.609] in contrast to those born to women belonging to states with low post neonatal mortality. Conclusion: Promotion and implementation of evidenced based strategies are needed in India to combat the high rates of child under-nutrition and infant and child mortality.

Analysis of Survival Time and Other Event Times - What Does the Surgeon Need to Know?

Survival time is an important parameter to investigate therapeutic measures. It plays a crucial role in study concepts, data analyses as well as publications. The aim of this study was to emphasise essential points, which need to be taken into account to (i) gain resilient results of survival time analysis and (ii) appropriately evaluate scientific reports. Corner Points/Main Statements: (i) The main analytical methods are Kaplan-Meier procedure to estimate survival time curves, the log rank test to compare two or more survival curves from independent samples and Cox regression for comparisons under simultaneous consideration of several influencing factors. (ii) Dependent relationships between survival and censoring probabilities may falsify these statistical procedures. (iii) For several end points, which need to be differentiated (such as death, progression etc.), and for interdependent sample elements, extended statistical procedures such as competing risk analyses or extended Cox regression models are available. Survival time analysis can be considered as being extraordinarily important for evaluation of data obtained in therapeutic studies. For the academic and publishing physician, in particular, for the clinical surgeon, a basic understanding of these methodological aspects in statistics is indispensable.

P08.14 * SURVIVAL OF BREAST CANCER PATIENTS WITH SYNCHRONOUS OR METACHRONOUS CNS METASTASES

BACKGROUND: Central nervous system (CNS) metastases represent a serious complication for breast cancer patients. This retrospective survey examines overall survival of unselected patients diagnosed with synchronous or metachronous CNS metastases to establish patient and tumour characteristics as well as treatment modalities affecting patients' prognosis. PATIENTS AND METHODS: Information on 732 breast cancer patients with CNS metastases was retrieved from the Netherlands Cancer Registry (NCR) database. Over the period 2004-2010, we identified 163 patients who presented with synchronous CNS metastases, 142 of which involved brain parenchymal metastases. 569 Patients who were initially disease-free developed metachronous CNS metastases during follow-up, and 489 of them were diagnosed with brain metastases. We measured patients' overall survival from the date of metastatic diagnosis, and the impact of prognostic factors on survival was assessed both univariately and multivariately using extended Cox-regression models. RESULTS: For all patients, triple-negative breast cancer and abstinence of systemic therapy proved prognostically unfavourable. In addition, surgery for the primary tumour and metastasectomy appear to improve survival in patients with synchronous CNS metastasis. In patients with metachronous metastases, higher initial stage (III) and the occurrence of other metastases prior to and simultaneously with CNS metastasis were associated with poorer survival whereas younger age ( 1 year) are prognostically favourable. Although metastasectomy and radiation therapy also showed an association with improved survival, their impact varied time-dependently and they were therefore included in our model as time-varying covariates. CONCLUSIONS: Triple-negative disease and abstinence of systemic therapy constitute prognostically unfavourable factors for all patients with CNS metastases. Other factors affecting survival vary between synchronous and metachronous metastasis. With respect to therapeutic factors, the impact of selection bias and hence overestimation of treatment effects should be kept in mind.

Long-term survival of women with locally advanced breast cancer with ≥ 10 involved lymph nodes at diagnosis.

Axillary lymph node status at diagnosis remains the strongest predictor of long-term survival in breast cancer. Patients with more than ten axillary lymph nodes at diagnosis have a poor long-term survival. In this single institutional study, we set out to evaluate the prognosis of this high-risk group in the era of multimodality therapy. In this retrospective study, we looked at all breast cancer patients with greater than ten axillary lymph nodes diagnosed at Mount Sinai Medical Center (MSMC) from January 1st 1990 to December 31st 2007 (n=161). In the univariate analysis, descriptive frequencies, median survival, and 5- and 10-year survival rates were estimated for common prognostic factors. A multivariate prognostic analysis for time-to-event data, using the extended Cox regression model was carried out. With a median and mean follow-up of 70 and 89.9 months, respectively, the overall median survival was estimated to be 99 months. The five-year disease-free survival (DFS) was 59.3% and the ten-year DFS was 37.9%, whereas the five- and ten-year overall survival (OS) was 66.6% and 43.9%, respectively. Multivariate analysis revealed a significant improvement in DFS among black patients compared to whites (p=0.05), improved DFS and OS among young patients (ages 21-45) compared to elderly patients (age greater than 70) (p=0.00176, p=0.0034, respectively), and improved DFS and OS among patients whose tumors were ER positive (p=0.049, p=0.0034). In this single institution study of patients with greater than 10 positive axillary nodes, black patients had a significantly improved DFS compared with white patients. Young age and ER tumor positivity was associated with improved outcomes. Using multivariate analysis, there were no other variables associated with statistically significant improvements in DFS or OS including date of diagnosis. Further work is needed to improve breast cancer survival in this subgroup of patients.

Event time analysis of longitudinal neuroimage data

This paper presents a method for the statistical analysis of the associations between longitudinal neuroimaging measurements, e.g., of cortical thickness, and the timing of a clinical event of interest, e.g., disease onset. The proposed approach consists of two steps, the first of which employs a linear mixed effects (LME) model to capture temporal variation in serial imaging data. The second step utilizes the extended Cox regression model to examine the relationship between time-dependent imaging measurements and the timing of the event of interest. We demonstrate the proposed method both for the univariate analysis of image-derived biomarkers, e.g., the volume of a structure of interest, and the exploratory mass-univariate analysis of measurements contained in maps, such as cortical thickness and gray matter density. The mass-univariate method employs a recently developed spatial extension of the LME model. We applied our method to analyze structural measurements computed using FreeSurfer, a widely used brain Magnetic Resonance Image (MRI) analysis software package. We provide a quantitative and objective empirical evaluation of the statistical performance of the proposed method on longitudinal data from subjects suffering from Mild Cognitive Impairment (MCI) at baseline.

Abstract P6-09-03: Improved clinical outcomes associated with statin use during adjuvant chemotherapy in patients with HER2+ non-metastatic breast cancer

Abstract Background Statins are established as first line therapy in patients with hypercholesterolemia and are thought to have pleiotropic effects including inhibition of signaling pathways known to drive cell proliferation and survival responses. Numerous preclinical and observational studies have demonstrated activity and improved outcomes in breast cancer patients who received statins, particularly lipophilic statins, as part of their adjuvant therapy. In the Life After Cancer Epidemiology (LACE) Study, 1,945 early-stage breast cancer survivors were shown to have a 33% decreased risk of breast cancer recurrence with post-diagnosis statin use, an effect that was magnified with increasing duration of statin use. We hypothesized that statin use concurrent with trastuzumab based chemotherapy in patients treated for non-metastatic HER2 positive breast cancer would be associated with improved outcomes. Methods We performed a retrospective review of all patients (n = 300) with HER2+ breast cancer who received trastuzumab from 2006-2012 at UM/SCCC. We identified two groups of patients for comparison - those who received statins (ST) during adjuvant chemotherapy (n = 45) or no statins during chemotherapy (NST) during adjuvant chemotherapy (n = 200). Patients with unclear documentation of concurrent statin use, men, patients with de-novo-metastatic breast cancer or prior primary cancer, bilateral breast cancers, and patients without follow-up were excluded. 5-year disease free (DFS) and overall survival (OS) were calculated. A univariate analysis was performed. Multivariate analysis for time-to-event data, using the Cox and extended Cox regression model will be performed and reported. Results The median age at diagnosis was 58 and 48, in the ST and NST groups, respectively. The most commonly used statin was lipophilic simvastatin (34.1%). The average baseline and post chemotherapy total cholesterol level in the ST group was 200.8 and 218.1mg/dL, respectively. Descriptive characteristics of the ST and NST groups respectively are as follows: postmenopausal (77.8% vs. 43.5%) tumor size &amp;lt;2cm (48.9% vs. 41.0%), node-negative disease (40.0% vs. 40.0%), HR+ status (64.4% vs. 60.5%), diabetes mellitus (26.7% vs. 9.0%), baseline BMI &amp;gt;30 (40.0% vs. 27.0%), and concomitant metformin (28.9% vs. 11.5%). The NST group was more likely to have a reduction of greater than 10% in their EF (9.0% vs. 6.7%). At a median follow-up of 32 and 26 months, the estimated 5-year DFS was significantly improved in the ST group (81.8% vs. 51.5%, p = 0.046). Meanwhile, the estimated 5-year OS was not significantly different between both groups (93.4% vs. 91.6%, p = 0.121). Metformin was not associated with a significant improvement in DFS or OS. Conclusion Our study demonstrated that statin use concurrent with trastuzumab-based chemotherapy for non-metastatic HER2 breast cancer was associated with improved estimated 5-year DFS, but not OS. The retrospective nature of our study does not allow for a definitive answer but further research needs to be done to define the role of statins in HER2+ breast cancer treatment. Citation Information: Cancer Res 2013;73(24 Suppl): Abstract nr P6-09-03.

Characterizing Ipomopsis rubra (Polemoniaceae) germination under various thermal scenarios with non-parametric and semi-parametric statistical methods

Time-to-event analysis represents a collection of relatively new, flexible, and robust statistical techniques for investigating the incidence and timing of transitions from one discrete condition to another. Plant biology is replete with examples of such transitions occurring from the cellular to population levels. However, application of these statistical methods has been rare in botanical research. Here, we demonstrate the use of non- and semi-parametric time-to-event and categorical data analyses to address questions regarding seed to seedling transitions of Ipomopsis rubra propagules exposed to various doses of constant or simulated seasonal diel temperatures. Seeds were capable of germinating rapidly to >90 % at 15-25 or 22/11-29/19 °C. Optimum temperatures for germination occurred at 25 or 29/19 °C. Germination was inhibited and seed viability decreased at temperatures ≥30 or 33/24 °C. Kaplan-Meier estimates of survivor functions indicated highly significant differences in temporal germination patterns for seeds exposed to fluctuating or constant temperatures. Extended Cox regression models specified an inverse relationship between temperature and the hazard of germination. Moreover, temperature and the temperature × day interaction had significant effects on germination response. Comparisons to reference temperatures and linear contrasts suggest that summer temperatures (33/24 °C) play a significant role in differential germination responses. Similarly, simple and complex comparisons revealed that the effects of elevated temperatures predominate in terms of components of seed viability. In summary, the application of non- and semi-parametric analyses provides appropriate, powerful data analysis procedures to address various topics in seed biology and more widespread use is encouraged.