Restricted Mean Survival Research Articles

A metabolism-related gene signature for predicting the prognosis in thyroid carcinoma.

Metabolic reprogramming is one of the cancer hallmarks, important for the survival of malignant cells. We investigated the prognostic value of genes associated with metabolism in thyroid carcinoma (THCA). A prognostic risk model of metabolism-related genes (MRGs) was built and tested based on datasets in The Cancer Genome Atlas (TCGA), with univariate Cox regression analysis, LASSO, and multivariate Cox regression analysis. We used Kaplan-Meier (KM) curves, time-dependent receiver operating characteristic curves (ROC), a nomogram, concordance index (C-index) and restricted mean survival (RMS) to assess the performance of the risk model, indicating the splendid predictive performance. We established a three-gene risk model related to metabolism, consisting of PAPSS2, ITPKA, and CYP1A1. The correlation analysis in patients with different risk statuses involved immune infiltration, mutation and therapeutic reaction. We also performed pan-cancer analyses of model genes to predict the mutational value in various cancers. Our metabolism-related risk model had a powerful predictive capability in the prognosis of THCA. This research will provide the fundamental data for further development of prognostic markers and individualized therapy in THCA.

Analysis of survival data with nonproportional hazards: A comparison of propensity-score-weighted methods.

One of the most common ways researchers compare cancer survival outcomes across treatments from observational data is using Cox regression. This model depends on its underlying assumption of proportional hazards, but in some real-world cases, such as when comparing different classes of cancer therapies, substantial violations may occur. In this situation, researchers have several alternative methods to choose from, including Cox models with time-varying hazard ratios; parametric accelerated failure time models; Kaplan-Meier curves; and pseudo-observations. It is unclear which of these models are likely to perform best in practice. To fill this gap in the literature, we perform a neutral comparison study of candidate approaches. We examine clinically meaningful outcome measures that can be computed and directly compared across each method, namely, survival probability at time T, median survival, and restricted mean survival. To adjust for differences between treatment groups, we use inverse probability of treatment weighting based on the propensity score. We conduct simulation studies under a range of scenarios, and determine the biases, coverages, and standard errors of the average treatment effects for each method. We then demonstrate the use of these approaches using two published observational studies of survival after cancer treatment. The first examines chemotherapy in sarcoma, which has a late treatment effect (i.e., similar survival initially, but after 2 years the chemotherapy group shows a benefit). The other study is a comparison of surgical techniques for kidney cancer, where survival differences are attenuated overtime.

Perinatal intimate partner violence and postpartum contraception timing among currently married women in Southern Ethiopia: A multilevel Weibull regression modeling.

Adopting contraception on time is a critical intervention for postpartum women, but violence exposure around pregnancy may interfere with postpartum contraceptive use behaviors. Hence, this study aimed to investigate the time duration of the first modern contraceptive adoption and its individual-and community-level predictors among postpartum women in the Wolaita zone, South Ethiopia. A community-based prospective follow-up study was conducted among 1,292 postpartum women nested in 38 "Kebles" (clusters) using multistage-clustered sampling techniques. A multilevel Weibull regression model was employed to investigate predictors of time-to-method initiation after childbirth using STATA Version 14. Kaplan-Meier curve and Wilcoxon log-rank test were used to estimate time-to-modern contraceptive use across different variables. All variables with p-values <0.05 were considered for multivariate analysis. Adjusted time ratios (ATR) with 95 % CI were computed using Weibull accelerated failure time models. Of the respondents, 62% (95% CI: 59.1-64.5) had started the first modern contraception within a year after childbirth. The restricted mean survival time-to-postpartum modern contraceptive use was 6.28 months. Being a rural dweller (aTR: 1.44; 95% CI: 1.06-1.99) and living in the middle household wealth quintiles (aTR: 1.10; 95% CI: 1.02-1.19) predicted longer time duration to adopt first modern contraception by 44 and 10%, respectively. The women from the community with a high early marriage (aTR: 1.14; 95% CI: 1.01-1.28) took longer time to initiate modern postpartum methods. Furthermore, women who had no history of perinatal abuse took less time than those who had a history of abuse to start postpartum contraception (aTR: 0.71; 95% CI: 0.66-0.78). Rural residence, poor household wealth status, history of perinatal abuse, and a high rate of early marriage in the community are predicted to lengthen the time duration to start modern postpartum contraception. Thus, community-level women's empowerment, particularly among rural women and integration of intimate partner violence screening into family planning counseling throughout the continuum of care will likely to improve postpartum contraception timing.

Event History Regression with Pseudo-Observations: Computational Approaches and an Implementation in R

Due to tradition and ease of estimation, the vast majority of clinical and epidemiological papers with time-to-event data report hazard ratios from Cox proportional hazards regression models. Although hazard ratios are well known, they can be difficult to interpret, particularly as causal contrasts, in many settings. Nonparametric or fully parametric estimators allow for the direct estimation of more easily causally interpretable estimands such as the cumulative incidence and restricted mean survival. However, modeling these quantities as functions of covariates is limited to a few categorical covariates with nonparametric estimators, and often requires simulation or numeric integration with parametric estimators. Combining pseudo-observations based on non-parametric estimands with parametric regression on the pseudo-observations allows for the best of these two approaches and has many nice properties. In this paper, we develop a user friendly, easy to understand way of doing event history regression for the cumulative incidence and the restricted mean survival, using the pseudo-observation framework for estimation. The interface uses the well known formulation of a generalized linear model and allows for features including plotting of residuals, the use of sampling weights, and correct variance estimation.

Model-free estimates that complement information obtained from the hazard ratio.

The online version contains supplementary material available at 10.1007/s12055-021-01167-4.

Blood natural killer cells during treatment in recurrent ovarian cancer

Objective Recent research indicated favorable prognostic impact of intratumoral natural killer (NK) cells in ovarian carcinoma (OC). The role of NK cells during chemotherapy in OC is unknown. We investigated impact of NK cells in OC patients treated with palliative chemotherapy. Methods Participants receiving palliative chemotherapy for recurrent OC (N = 72) had prospectively blood samples at baseline and before cycle 2. NK cell counts were quantified by flow cytometry. NK cell activity was measured by the NK Vue® assay, estimating interferon-gamma production. Overall survival (OS) was the primary endpoint. Cutoffs were predefined, NK numbers (≥184 × 106 cells/L vs. <184 × 106 cells/L) and NK activity (<200 pg/mL vs. ≥200 pg/mL). Results Median OS in patients with low vs. high NK cell count at baseline was 7.1 months vs. 15.6 months (p = .028), respectively, and before cycle 2 was 5.7 vs. 17.3 months, p < .001, respectively. The difference in restricted mean survival (ΔRMST) was 5.7 months (95% CI: 3.3–8.0) at cycle 2 vs. 2.5 months (95% CI: −0.6 to 5.6) at baseline, showing a significant difference with no overlap of confidence intervals. In multivariate analyses, low NK cell count remained significant with a hazard ratio (HR)=2.83, 95% CI: 1.53–5.22, p = .001 (baseline) and HR = 3.34, 95% CI: 1.67–6.71, p = .001 (before cycle 2). Patients with both low NK count and NK activity at baseline (N = 20) had median OS 6.5 months vs. 11.5 months in patients with either high activity, high count or both (p = .007). In parallel, patients with both low NK activity and count at cycle 2 (N = 18) had a median survival of 4.0 months vs. 15.4 months (p < .001). Conclusions A low blood NK cell count in recurrent metastatic ovarian cancer during chemotherapy is associated with unfavorable prognostic impact. Early increase in survival difference based on NK cell status suggests an association between NK cell count and treatment benefit.

Multi-state analysis of hypertension and mortality: application of semi-Markov model in a longitudinal cohort study

BackgroundMost previous research has studied the association of hypertension with cardiovascular disease (CVD) and all-cause mortality by focusing on the transition from the initial state to a single outcome. We investigated the impact of hypertension, defined according to the 2017 American College of Cardiology/American Heart Association (ACC/AHA) (new) and the Seventh Report of the Joint National Committee (JNC7) (old), on CVD death and all-cause mortality considering non-fatal CVD as an intermediate event between two CVD-free and mortality states.MethodsA total of 3002 Iranian population (47.4% men), aged ≥50 years were followed from 1999 to 2014. Two multi-state semi-Markov models with three transitions were defined for CVD death and all-cause mortality as two outcomes. The multivariable Cox model was used to estimate the effect of hypertension on transition hazards. The mean of 15-year life expectancy of participants in each transition was estimated using the restricted mean survival time.ResultsThe ACC/AHA guideline increased the prevalence of hypertension from 43.3 to 68.6%. Among CVD-free individuals, hypertension was significantly associated with increased risk of non-fatal CVD [Hazard Ratio, 1.52 (1.28–1.81) and 1.48 (1.21–1.80)], CVD death [2.96 (2.06–4.25) and 1.98 (1.30–3.04)] and all-cause mortality [1.64 (1.32–2.05) and 1.31 (1.01–1.69)] according the old and new guidelines, respectively. However, after incident non-fatal CVD, the association between hypertension and mortality events was not significant according to both definitions. Hypertensive participants experienced a first non-fatal CVD about 0.9 and 0.6 years earlier than normotensive population according to JNC7 and the 2017 ACC/AHA guidelines, respectively.ConclusionHypertension, according to JNC7 and the ACC/AHA guidelines, significantly increased the risk of mortality events among CVD-free population although the risk was attenuated using ACC/AHA guideline. Hypertension also decreased the number of years lived without CVD and early onset of CVD, and consequently, an increase in the time spent with these diseases. After non-fatal CVD, hypertension had no significant impact on mortality risk according to both guidelines.

RNA processing genes characterize RNA splicing and further stratify colorectal cancer.

ObjectivesDue to the limited evaluation of the prognostic value of RNA processing genes (RPGs), which are regulators of alternative splicing events (ASEs) that have been shown to be associated with tumour progression, this study sought to determine whether colorectal cancer (CRC) could be further stratified based on the expression pattern of RPGs.Materials and MethodsThe gene expression profiles of CRCs were collected from TCGA (training set) and three external validation cohorts, representing 1060 cases totally. Cox regression with least absolute shrinkage and selection operator (LASSO) penalty was used to develop an RNA processing gene index (RPGI) risk score. Kaplan‐Meier curves, multivariate Cox regression and restricted mean survival (RMS) analyses were harnessed to evaluate the prognostic value of the RPGI.ResultsA 22‐gene RPGI signature was developed, and its risk score served as a strong independent prognostic factor across all data sets when adjusted for major clinical variables. Moreover, ASEs for certain genes, such as FGFR1 and the RAS oncogene family, were significantly correlated with RPGI. Expression levels of genes involved in splicing‐ and tumour‐associated pathways were significantly correlated with RPGI score. Furthermore, a combination of RPGI with age and tumour stage resulted in significantly improved prognostic accuracy.ConclusionsOur findings highlighted the prognostic value of RPGs for risk stratification of CRC patients and provide insights into specific ASEs associated with the development of CRC.

Effectiveness and Healthcare Cost of Adding Trastuzumab to Standard Chemotherapy for First-Line Treatment of Metastatic Gastric Cancer: A Population-Based Cohort Study.

A randomized clinical trial showed that trastuzumab, added to traditional chemotherapy, significantly improved overall survival in human epidermal growth factor receptor 2 (HER2)-overexpressing metastatic gastric cancer patients. This population-based study aimed at evaluating both the clinical and economic impact of trastuzumab in a real-world setting. By using the healthcare utilization databases of Lombardy, Italy, a cohort of patients newly diagnosed with metastatic gastric cancer during the period 2011–2016 was selected. Among these, patients initially treated with either trastuzumab-based chemotherapy or standard chemotherapy alone were followed up until death, migration in other regions or June 2018. Overall survival and average cumulative costs were estimated and compared between the two treatment arms. Among the 1198 metastatic gastric cancer patients who started therapy within six months after metastasis detection, 87 were initially treated with trastuzumab-based chemotherapy and 1111 with standard chemotherapy. Median overall survival and restricted mean survival were 10.2 and 7.4 months, and 14.9 and 11.4 months, respectively, in the two treatment arms. The adjusted hazard ratio of death was 0.73 (95% CI 0.57–0.93). The average per capita cumulative healthcare costs were, respectively, EUR 39,337 and 26,504, corresponding to an incremental cost-effectiveness ratio of EUR 43,998 for each year of survival gained. Our study shows that adding trastuzumab to conventional chemotherapy is effective and cost-effective.

Deep learning for survival outcomes.

Deep learning is a class of machine learning algorithms that are popular for building risk prediction models. When observations are censored, the outcomes are only partially observed and standard deep learning algorithms cannot be directly applied. We develop a new class of deep learning algorithms for outcomes that are potentially censored. To account for censoring, the unobservable loss function used in the absence of censoring is replaced by a censoring unbiased transformation. The resulting class of algorithms can be used to estimate both survival probabilities and restricted mean survival. We show how the deep learning algorithms can be implemented by adapting software for uncensored data by using a form of response transformation. We provide comparisons of the proposed deep learning algorithms to existing risk prediction algorithms for predicting survival probabilities and restricted mean survival through both simulated datasets and analysis of data from breast cancer patients.

Do Stable Patients Supported with Long-Term Mechanical Circulatory Support Device Have Equitable Access to Donor Hearts in the New French Allocation System?

The new French allocation system grants priority to patients on long-term mechanical circulatory support (MCS) with device complications but does not prioritize stable long-term MCS candidates. Some experts expressed concerns regarding waiting time of long-term MCS candidates without complication. The aim of this study was to evaluate the access to transplantation of long-term MCS candidates. The study included all candidates on long-term MCS without complication registered on the French heart transplant waiting list between January 2018 and March 2019 (group 1, n=65) and those ambulatory without circulatory support(group 2, n=220). The main outcomes were 6-month cumulative incidence (CI) of transplantation, and, of waitlist mortality or delisting for worsening medical condition. The CI were estimated using the competing risk analyses and compared using Gray's test. In addition, the restricted mean survival was used to estimate the difference in access to transplantation between the 2 groups. Among the 65 stable long-term MCS candidates, 37 were on support at listing and 28 were implanted while listed. The 6-month CI of transplantation was significantly lower in patients on long-term MCS than in ambulatory candidates (25% vs 49%; p<0.01). The difference of restricted mean survival between the groups was 37 days. The 6-month CI of waitlist mortality was higher in long-term MCS candidates implanted while listed (19%) than in patients on MCS at listing (8 %) and in ambulatory patients (1%). Access to transplantation was lower in stable long-term MCS patients than in ambulatory candidates. The increased waitlist mortality was only observed in candidates implanted while listed. The French allocation system is being revised with a new version granting a provision in stable long-term MCS candidates.

Comprehensive Analysis of miRNA Sequencing Profiles Identifies Novel Deregulated and Prognostic Biomarkers in Hepatocellular Carcinoma

Background: Liver cancer is the fourth most common cause of cancer-related death and rank sixth in terms of incident cases. With a 5-year survival of 18%, liver cancer is the second most lethal tumor, after pancreatic cancer. We aim to identify a set of miRNAs and a miRNA-based signature related to tumorigenesis and prognosis in patients with hepatocellular carcinoma (HCC). Methods: We analyzed the miRNA sequencing profiles of 373 HCC patients downloaded from The Cancer Genome Atlas LIHC program. The isoform quantification profiles were transformed into 5p and 3p mature miRNA names. Differentially expressed (DE) miRNAs between tumor and adjacent normal tissues were identified by Wald test based on the negative binomial distribution. Prognostic miRNAs associated with overall survival were confirmed by multivariate Cox proportional hazards models. The miRNA-based signatures were obtained from the linear predictors of cox regression, and the prognostic performance was compared by Harrel’s C-index and revealed by the restricted mean survival (RMS) curve. Findings: The selected twelve DE miRNAs showed a good performance to classify tumor tissues from normal tissues. Meanwhile, a miRNA-based prognostic signature of eight mature miRNAs was constructed, which significantly stratified patients into high- vs low-risk groups in terms of overall survival (hazard ratio, 4.11; 95% CI, 2.71-6.24; P<0.001). When integrated with clinical information, the composite miRNA-clinical signature showed improved prognostic accuracy relative to the eight-miRNA signature alone. As we set the follow-up time at 5 years, the estimated RMST difference between low- and high-risk group stratified by miRNA index was 1.39 (95% CI: 0.95-1.83) months, which is lesser than the difference between miRNA-clinical risk groups (1.63, 95%CI: 1.20-2.06). Functional enrichment analysis indicated that the target mRNAs of selected miRNAs were mainly enriched in cancer-related pathways and vital cell biological processes. Interpretation: The proposed DE miRNAs and miRNA-clinical signature are promising biomarkers for discrimination and predicting overall survival respectively in HCC patients. These biomarkers may have significant relevance for development of new drug research and targeting therapies for HCC patients. Funding Statement: This study was supported by National Science and Technology Major Project of China (2017ZX10203207-004-004), Natural Science Foundation of the Jiangsu Higher Education Institutions of China (Grant No. 17KJB330003). Declaration of Interests: The authors who have taken part in this study declared that they do not have anything to disclose regarding funding or conflict of interest with respect to this manuscript.

Radical Prostatectomy or Observation for Clinically Localized Prostate Cancer: Extended Follow-up of the Prostate Cancer Intervention Versus Observation Trial (PIVOT)

BackgroundVery long-term mortality in men with early prostate cancer treated with surgery versus observation is uncertain. ObjectiveTo determine long-term effects of surgery versus observation on all-cause mortality for men with early prostate cancer. Design, setting, and participantsThis study evaluated long-term follow-up of a randomized trial conducted at the US Department of Veterans Affairs and National Cancer Institute sites. The participants were men (n=731) ≤75yr of age with localized prostate cancer, prostate-specific antigen (PSA) <50ng/ml, life expectancy ≥10yr, and medically fit for surgery. InterventionRadical prostatectomy versus observation. Outcome measurements and statistical analysisAll-cause mortality was assessed in the entire cohort and patient and tumor subgroups. Intention-to-treat analysis was conducted using Kaplan-Meier methods with log-rank tests and Cox proportional hazard models; cumulative mortality incidence, between-group differences, and relative risks were also assessed at predefined time periods. Results and limitationsDuring 22.1yr (median follow-up for survivors=18.6yr; interquartile range: 16.6–20.0), 515 men died; 246 of 346 men (68%) were assigned to surgery versus 269 of 367 (73%) assigned to observation (hazard ratio 0.84 [95% confidence interval {CI}: 0.70–1.00]; p= 0.044 [absolute risk reduction = 5.7 percentage points, 95% CI: –0.89 to 12%]; relative risk: 0.92 [95% CI: 0.84–1.01]). The restricted mean survival in the surgical group was 13.6 yr (95% CI: 12.9–14.3) versus 12.6 yr (95% CI: 11.8–13.3) in the observation group; a mean of 1 life-year was gained with surgery. Results did not significantly vary by patient or tumor characteristics, although differences were larger favoring surgery among men aged <65 yr, of white race, and having better health status, fewer comorbidities, ≥34% positive prostate biopsy cores, and intermediate-risk disease. Results were not adjusted for multiple comparisons, and we could not assess outcomes other than all-cause mortality. ConclusionsSurgery was associated with small very long-term reductions in all-cause mortality and increases in years of life gained. Absolute effects did not vary markedly by patient characteristics. Absolute effects and mean survival were much smaller in men with low-risk disease, but were greater in men with intermediate-risk disease although not in men with high-risk disease. Patient summaryIn this randomized study, we evaluated death from any cause in men with early prostate cancer treated with either surgery or observation. Overall, surgery may provide small very long-term reductions in death from any cause and increases in years of life gained. Absolute effects were much smaller in men with low-risk disease, but were greater in men with intermediate-risk disease although not in men with high-risk disease. Strategies are needed to identify men needing and benefitting from surgery while reducing ineffective treatment and overtreatment.

PCN146 VALIDATION OF PREDICTED SURVIVAL ESTIMATES FROM PREVIOUS DATA CUTS OF THE PHASE III CHECKMATE 025 STUDY IN ADVANCED RENAL CELL CARCINOMA

CheckMate 025 (CM-025; NCT01668784) is a randomized phase 3 trial investigating nivolumab and everolimus in pre-treated metastatic renal cell carcinoma (RCC). It has the longest follow-up of any phase III study of immunotherapy in pre-treated metastatic RCC. This study investigates the accuracy of overall survival (OS) modelling from previous data cuts of the CM-025 study. Parametric models were fitted to CM-025 trial data with minimum follow-up of 14.0 months, 26.1 months, and 39.4 months, using methods described in NICE Technical Support Document 14. Estimated survival rates and restricted mean survival (RMS) durations under each of the fitted models were compared against corresponding rates and RMS durations observed within the trial to assess predictive accuracy. Observed 3-year OS rates were 39.1% and 29.5% for nivolumab and everolimus, respectively. Predicted 3-year OS rates from the chosen parametric models (log-logistic) fitted to data at minimum follow-up of 14.0 months, 26.1 months, and 39.4 months were 39.1%, 39.0% and 39.6% for nivolumab, and 29.8%, 29.3%, and 30.0% for everolimus, respectively. The exponential and Weibull functions provided poorer fits, under predicting the 3-year rate for nivolumab by up to 4.3%. Observed 3-year RMS was 23.8 months and 20.6 months for nivolumab and everolimus respectively. Three-year RMS estimated for the data cut with 39.4 months minimum follow-up were 23.6 months for nivolumab and 20.6 months for everolimus. Nivolumab has consistently demonstrated superior and sustained OS compared with everolimus for pre-treated RCC patients over the long-term. OS rate predictions and RMS from the chosen functions fitted to previous data cuts of CM-025 were consistent with the longer-term observed data. This confirms that the base case parametric models fitted to previous data cuts submitted to reimbursement authorities were highly accurate.

Matching-adjusted indirect comparison of bosutinib, dasatinib and nilotinib effect on survival and major cytogenetic response in treatment of second-line chronic phase chronic myeloid leukemia

Objective: In clinical trials of second-line therapies for chronic phase chronic myeloid leukemia (CP-CML), to date, only single-arm trials have been conducted for the available tyrosine kinase inhibitor treatments (bosutinib, dasatinib and nilotinib). These trials included heterogeneous patient populations in terms of disease and baseline characteristics. These hamper the use of standard network meta-analyses for indirect treatment comparison of relative efficacy. In this situation, a matching-adjusted indirect comparison (MAIC) in second-line CP-CML was performed. The aim was to compare the relative efficacies of bosutinib, dasatinib and nilotinib in second-line CP-CML patients.Methods: The MAIC was preceded by a systematic literature review that ensured inclusion of the underlying data for the analyses. The outcomes were measured in terms of overall survival (OS), progression-free survival (PFS) and major cytogenetic response (MCyR). The treatments were quantitatively compared based on Cox proportional hazard ratio (HR) regressions, on restricted mean survival (RMST, when the proportionality assumption showed evidence of violation) and on odds ratios (for response measures).Results: Comparing with dasatinib, bosutinib resulted in HRs for PFS and OS of 0.63 (0.44–0.90, p < .05) and 0.82 (0.54–1.26, p = .37) respectively, and resulted in an OR for MCyR of 0.78 (0.53–1.16). Although the proportionality of hazards assumption was violated for PFS, the RMST analyses confirmed the findings of the Cox regression. When compared with nilotinib, bosutinib showed a significant HR of 0.54 (0.38–0.76, p < .01) in favor of bosutinib for PFS, a non-significant HR of 0.72 (0.46–1.13, p = .16) for OS and a non-significant OR of 0.98 (0.71–1.35) for MCyR.Conclusions: Bosutinib had a significantly greater PFS than both dasatinib and nilotinib. For OS, the findings were numerically in favor of bosutinib, but not statistically significant. For MCyR, the findings were numerically in favor of dasatinib and nilotinib, but not statistically significant.

MS4 A COMPARISON OF THE PARTITIONED SURVIVAL AND MULTISTATE EXTRAPOLATION METHODS: A CASE STUDY BASED ON THE GASTRIC CANCER TRIAL ATTRACTION-2

Partitioned survival modelling (PartSA) is frequently used in cost-effectiveness analysis in oncology and is characterized by separate modelling of time to event (TTE) outcomes. The multistate modelling (MSM) approach is defined by its ability to jointly estimate TTEs. The aim of this study was to compare PartSA and MSM-obtained survival extrapolations using data from the ATTRACTION-2 (NCT02267343) trial, which compared nivolumab with best supportive care (BSC) in advanced or metastatic gastric cancer. Two data cutoffs were available for ATTRACTION-2, from August 2016 and February 2018, with median follow-up periods of 9 and 24 months, respectively. Standard parametric models were fitted to the 2016 data using the PartSA and MSM approaches. The MSM approach included a covariate to account for patients’ history of time on treatment. Extrapolations for overall survival (OS) based on 2016 data were compared with the more mature 2018 data. The observed restricted mean survival (RMS) at 3 years in ATTRACTION-2 was 0.50 years (95% CI: 0.42-0.58 years) and 0.78 years (95% CI: 0.70-0.87 years) for BSC and nivolumab, respectively (2018 data). Both modelling approaches using the 2016 data successfully predicted the observed OS in the 2018 data, as the 3-year RMS estimates were within the observed 95% CIs. OS estimates for BSC were similar for both approaches (3-year RMS, PartSA: 0.51 years; MSM: 0.52 years). OS estimates for nivolumab were lower using PartSA than MSM (PartSA: 0.77 years; MSM: 0.85 years). Both PartSA and MSM, based on the 2016 data, provided accurate predictions up to three years as estimates were within the observed 95% CI of the 2018 data. MSM predicted a higher mean survival for nivolumab compared to PartSA, as MSM predicted a long-term survival plateau whereas PartSA did not. Clinical insight could provide additional guidance on the preferred extrapolation approach.

Association of Primary Care Physician Supply With Population Mortality in the United States, 2005-2015

Recent US health care reforms incentivize improved population health outcomes and primary care functions. It remains unclear how much improving primary care physician supply can improve population health, independent of other health care and socioeconomic factors. To identify primary care physician supply changes across US counties from 2005-2015 and associations between such changes and population mortality. This epidemiological study evaluated US population data and individual-level claims data linked to mortality from 2005 to 2015 against changes in primary care and specialist physician supply from 2005 to 2015. Data from 3142 US counties, 7144 primary care service areas, and 306 hospital referral regions were used to investigate the association of primary care physician supply with changes in life expectancy and cause-specific mortality after adjustment for health care, demographic, socioeconomic, and behavioral covariates. Analysis was performed from March to July 2018. Age-standardized life expectancy, cause-specific mortality, and restricted mean survival time. Primary care physician supply increased from 196 014 physicians in 2005 to 204 419 in 2015. Owing to disproportionate losses of primary care physicians in some counties and population increases, the mean (SD) density of primary care physicians relative to population size decreased from 46.6 per 100 000 population (95% CI, 0.0-114.6 per 100 000 population) to 41.4 per 100 000 population (95% CI, 0.0-108.6 per 100 000 population), with greater losses in rural areas. In adjusted mixed-effects regressions, every 10 additional primary care physicians per 100 000 population was associated with a 51.5-day increase in life expectancy (95% CI, 29.5-73.5 days; 0.2% increase), whereas an increase in 10 specialist physicians per 100 000 population corresponded to a 19.2-day increase (95% CI, 7.0-31.3 days). A total of 10 additional primary care physicians per 100 000 population was associated with reduced cardiovascular, cancer, and respiratory mortality by 0.9% to 1.4%. Analyses at different geographic levels, using instrumental variable regressions, or at the individual level found similar benefits associated with primary care supply. Greater primary care physician supply was associated with lower mortality, but per capita supply decreased between 2005 and 2015. Programs to explicitly direct more resources to primary care physician supply may be important for population health.

Incidence of autoimmune pulmonary alveolar proteinosis estimated using Poisson distribution.

The incidence and prevalence of autoimmune pulmonary alveolar proteinosis in Japan were previously estimated to be 0.49 and 6.2 per million, respectively. Thereafter, an increase in serological diagnosis forced a re-estimation of the incidence based on more contemporaneous data using more robust methods.Sera of 702 patients were positive for granulocyte-macrophage colony-stimulating factor autoantibody during the 2006–2016 period (group A). Of these patients, 43 were actively surveyed in Niigata prefecture (group B) for estimation of the incidence. To estimate the survival period, 103 patients (group C) were investigated retrospectively for the 1999–2017 period using restricted mean survival time.In group A, the number of patients diagnosed in each prefecture was closely correlated with the corresponding population, indicating no regional integration of onset. In group B, a total of 43 patients were diagnosed, the annual number followed a Poisson distribution and the incidence was thus estimated to be 1.65 per million. In group C, the retrospective cohort study revealed the mean survival period to be 16.1 years. Taken together, the prevalence was estimated to be 26.6 per million, indicating that the previous data for incidence and prevalence was an underestimation.

Abstract GS3-03: A phase III multicentre double blind randomised trial of celecoxib versus placebo in primary breast cancer patients (REACT – Randomised EuropeAn celecoxib trial)

Abstract Background Inhibition of COX-2 has been shown to attenuate the metastatic process in pre-clinical models of human breast cancer (BC). The primary aim of this study was to assess the effect of 2 years adjuvant therapy with the COX-2 inhibitor celecoxib compared with placebo in HER2-ve primary BC patients. Patients &amp; Methods Patients were randomised in a 2:1 ratio to receive celecoxib 400mg once daily or placebo for 2 years. Patients had to have completely resected BC with prior local and systemic adjuvant treatment according to local practice. Concurrent radiotherapy was permitted and hormone receptor +ve patients received endocrine therapy according to local practice. Patients with HER2+ or node negative, T1 and grade 1 disease were excluded. Median age of patients was 55 years (IQR: 49-63). 50% of patients had tumours &amp;gt;2cm; 42% were grade 3; 48% had node +ve disease. According to local assessment 73% were ER/PgR +ve. Primary endpoint was Disease Free Survival (DFS); defined as time from randomisation to date of first event, with events contributing to analysis defined as recurrence (distant/local), new primary BC (ipsilateral/contralateral) and death. Secondary endpoints included Overall Survival (OS), toxicity, cardiovascular mortality and incidence of second primaries. Subgroup analysis by hormone receptor status was pre-planned. Survival endpoints are analysed using Cox-proportional hazards and log-rank tests; restricted mean survival is used where proportional hazards do not hold. Results Between January 2007 and November 2012, 2639 patients were randomised (1763 celecoxib; 876 placebo) from 181 centres across the UK and Germany. At 13th April 2017, median follow up was 60 months (IQR: 48-72) with 428 DFS events reported. Unadjusted survival analysis results are presented below, with hazard ratio&amp;lt;1 favouring celecoxib: 5 year survival estimate (95% CI)Hazard ratio (95% CI)p-valueDFS (all patients) Celecoxib83% (81, 85)1.02 (0.83 – 1.24)0.88Placebo83% (80, 86)1- DFS within ER+ Celecoxib87% (85, 89)0.89 (0.69 – 1.16)0.40Placebo86% (83, 89)1- DFS within ER- Celecoxib72% (68, 76)1.17 (0.85 – 1.61)0.33Placebo75% (69, 80)1- OS (all patients) Celecoxib90% (88, 91)0.97 (0.75 – 1.25)0.81Placebo90% (88, 92)1- The interaction between ER status and treatment was not significant; p=0.36. In the celecoxib and placebo groups there were 17 and 8 deaths respectively in patients who had not relapsed. These were due to cardiac (n=3; 2) and other (n=14; 6) in the celecoxib and placebo groups respectively; none were GI related. In total 304 serious adverse events were observed in 265 patients (186/1763 celecoxib; 79/876 placebo). In the celecoxib and placebo groups respectively these were related to cardiac (n=12; 7), GI (n=9; 2) and other (n=193; 81). Work is ongoing to determine whether a subset of ER+ patients whose primary tumours show the characteristics of a COX-2 signature receive greater benefit from celecoxib. Conclusions There is no benefit of celecoxib in the ITT population. Further exploratory studies focussing on the ER+ subpopulation are ongoing. Celecoxib treatment is not associated with significant toxicity when compared to placebo in this population of BC patients. Citation Format: Coombes RC, Tovey H, Kilburn L, Mansi J, Palmieri C, Bartlett J, Hicks J, Makris A, Evans A, Loibl S, Denkert C, Murray E, Grieve R, Coleman R, Schmidt M, Klare P, Rezai M, Rautenberg B, Klutinus N, Rhein U, Mousa K, Ricardo-Vitorino S, von Minckwitz G, Bliss J. A phase III multicentre double blind randomised trial of celecoxib versus placebo in primary breast cancer patients (REACT – Randomised EuropeAn celecoxib trial) [abstract]. In: Proceedings of the 2017 San Antonio Breast Cancer Symposium; 2017 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2018;78(4 Suppl):Abstract nr GS3-03.

Association of Lymph Node Count and Overall Survival in Node-Negative Endometrial Cancers.

To estimate whether the number of lymph nodes removed during surgery is associated with overall survival among women with endometrial cancer. We performed a retrospective cohort study of women with node-negative, stage I to IIIB endometrial cancer (n = 152,702) identified from the 1998-2011 National Cancer Database. Multivariable Cox proportional hazards regression tested for an association of lymph node count with survival. Restricted mean survival and relative hazard curves were plotted for survival as a function of number of removed lymph nodes. Among women with node-negative endometrioid endometrial cancer, for each additional five lymph nodes removed, the hazard for death decreased: stage I, the hazard ratio (HR) was 0.95 (95% CI, 0.93 to 0.97; P < .001), stage II, HR was 0.90 (95% CI, 0.87 to 0.94; P < .001); and stage IIIA-B, HR was 0.92 (95% CI, 0.88 to 0.96; P < .001). When grouped by grade, each additional five lymph nodes removed was also associated with decreased hazard for death: grade 1, HR was 0.96 (95% CI, 0.93 to 0.99; P = .009); grade 2,HR was0.91 (95%CI, 0.89 to0.94; P <.001);and grade 3,HR was 0.95 (95%CI, 0.92 to 0.97; P <.001). Increased lymph node dissection was also associated with increased survival among women with node-negative stage II (HR, 0.92; 95% CI, 0.86 to 0.98; P = .01) or stage IIIA-B (HR, 0.94; 95% CI, 0.89 to 0.99; P = .025) uterine serous carcinoma, but not among women with carcinosarcoma or clear cell adeno-carcinoma. Five-year survival for women with one to four nodes removed and endometrioid or serous histology was 85% (95% CI, 84% to 85%) and 54% (95% CI, 50% to 59%), respectively. Five-year survival was significantly higher for women with ≥ 20 removed nodes and endometrioid (91%; 95% CI, 90% to 91%) or serous (72%; 95% CI, 68% to 76%) histology (P < .001). Increased lymph node count is associated with a 1% to 14% decreased hazard of death per each additional five lymph nodes removed and a 5% to 20% increased 5-year survival among women with pathologically node-negative endometrioid and serous endometrial cancers.