Designing a longitudinal clinical trial based on a composite endpoint: Sample size, monitoring, and adaptation.

Longitudinal clinical trials are often designed to compare treatments on the basis of multiple outcomes. For example in the case of cardiac trials, the outcomes of interest include mortality as well as cardiac events and hospitalization. For a COVID-19 trial, the outcomes of interest include mortality, time on ventilator, and time in hospital. Earlier work by these authors proposed a non-parametric test based on a composite of multiple endpoints referred to as the Finkelstein-Schoenfeld (FS) test (Finkelstein and Schoenfeld. Stat Med. 1999;18(11):1341-1354.). More recently, an estimate of the treatment comparison based on multiple endpoints (related to the FS test) was proposed (Pocock et al. Eur Heart J. 2011;33(2):176-182.). This estimate, which summarized the ratio of the number of patients who fared better vs worse on the experimental arm was coined the win ratio. The aim of this article is to provide guidance in the design of a trial that will use the FS test or the win ratio. The issues that will be considered are the sample size, sequential monitoring, and adaptive designs.

Determining the Incidence of Asymptomatic SARS-CoV-2 Among Early Recipients of COVID-19 Vaccines (DISCOVER-COVID-19): A Prospective Cohort Study of Healthcare Workers Before, During and After Vaccination.

The impact of coronavirus disease 2019 vaccination on viral characteristics of breakthrough infections is unknown. In this prospective cohort study, incidence of severe acute respiratory syndrome coronavirus 2 infection decreased following vaccination. Although asymptomatic positive tests were observed following vaccination, the higher cycle thresholds, repeat negative tests, and inability to culture virus raise questions about their clinical significance.

Estimation for an accelerated failure time model with intermediate states as auxiliary information.

The accelerated failure time (AFT) model is a common method for estimating the effect of a covariate directly on a patient's survival time. In some cases, death is the final (absorbing) state of a progressive multi-state process, however when the survival time for a subject is censored, traditional AFT models ignore the intermediate information from the subject's most recent disease state despite its relevance to the mortality process. We propose a method to estimate an AFT model for survival time to the absorbing state that uses the additional data on intermediate state transition times as auxiliary information when a patient is right censored. The method extends the Gehan AFT estimating equation by conditioning on each patient's censoring time and their disease state at their censoring time. With simulation studies, we demonstrate that the estimator is empirically unbiased, and can improve efficiency over commonly used estimators that ignore the intermediate states.

Adjuvant denosumab in early breast cancer (D-CARE): an international, multicentre, randomised, controlled, phase 3 trial.

BACKGROUND: Denosumab is a fully human monoclonal antibody that binds to, and inhibits, the receptor activator of RANKL (TNFSF11) and might affect breast cancer biology, as shown by preclinical evidence. We aimed to assess whether denosumab combined with standard-of-care adjuvant or neoadjuvant systemic therapy and locoregional treatments would increase bone metastasis-free survival in women with breast cancer. METHOD: In this international, double-blind, randomised, placebo-controlled, phase 3 study (D-CARE), patients were recruited from 389 centres in 39 countries. We enrolled women (aged ≥ 18 years) with histologically confirmed stage II or III breast cancer and an Eastern Cooperative Oncology Group performance status of 0 or 1. On eligibility confirmation, investigators at each site telephoned an interactive voice response system to centrally randomly assign patients (1:1) based on a fixed stratified permuted block randomisation list (block size 4) to receive either denosumab (120 mg) or matching placebo subcutaneously every 3-4 weeks, starting with neoadjuvant or adjuvant chemotherapy, for about 6 months and then every 12 weeks for a total duration of 5 years. Stratification factors were breast cancer therapy, lymph node status, hormone receptor and HER2 status, age, and geographical region. The primary endpoint was the composite endpoint of bone metastasis-free survival. This trial is registered with ClinicalTrials.gov, NCT01077154. FINDINGS: Between June 2, 2010, and Aug 24, 2012, 4509 women were randomly assigned to receive denosumab (n=2256) or placebo (n=2253) and included in the intention-to-treat analysis. The primary analysis of the study was done when all patients had the opportunity to complete 5 years of follow-up with an analysis data cutoff date of Aug 31, 2017. The primary endpoint of bone metastasis-free survival was not significantly different between the groups (median not reached in either group; hazard ratio 0·97, 95% CI 0·82-1·14; p=0·70). The most common grade 3 or worse treatment-emergent adverse events, reported in patients who had at least one dose of the investigational product (2241 patients with denosumab vs 2218 patients with placebo), were neutropenia (340 [15%] vs 328 [15%]), febrile neutropenia (112 [5%] vs 142 [6%]), and leucopenia (62 [3%] vs 61 [3%]). Positively adjudicated osteonecrosis of the jaw occurred in 122 (5%) of 2241 patients treated with denosumab versus four (<1%) of 2218 patients treated with placebo; treatment-emergent hypocalcaemia occurred in 152 (7%) versus 82 (4%). Two treatment-related deaths occurred in the placebo group due to acute myeloid leukaemia and depressed level of consciousness. INTERPRETATION: Despite preclinical evidence suggesting RANKL inhibition might delay bone metastasis or disease recurrence in patients with early-stage breast cancer, in this study, denosumab did not improve disease-related outcomes for women with high-risk early breast cancer. FUNDING: Amgen.

Design and analysis of a clinical trial using previous trials as historical control.

BACKGROUND/AIMS: For single arm trials, a treatment is evaluated by comparing an outcome estimate to historically reported outcome estimates. Such a historically controlled trial is often analyzed as if the estimates from previous trials were known without variation and there is no trial-to-trial variation in their estimands. We develop a test of treatment efficacy and sample size calculation for historically controlled trials that considers these sources of variation. METHODS: We fit a Bayesian hierarchical model, providing a sample from the posterior predictive distribution of the outcome estimand of a new trial, which, along with the standard error of the estimate, can be used to calculate the probability that the estimate exceeds a threshold. We then calculate criteria for statistical significance as a function of the standard error of the new trial and calculate sample size as a function of difference to be detected. We apply these methods to clinical trials for amyotrophic lateral sclerosis using data from the placebo groups of 16 trials. RESULTS: We find that when attempting to detect the small to moderate effect sizes usually assumed in amyotrophic lateral sclerosis clinical trials, historically controlled trials would require a greater total number of patients than concurrently controlled trials, and only when an effect size is extraordinarily large is a historically controlled trial a reasonable alternative. We also show that utilizing patient level data for the prognostic covariates can reduce the sample size required for a historically controlled trial. CONCLUSION: This article quantifies when historically controlled trials would not provide any sample size advantage, despite dispensing with a control group.

Are rare cancer survivors at elevated risk of subsequent new cancers?

BACKGROUND: Although rare cancers account for 27% of cancer diagnoses in the US, there is insufficient research on survivorship issues in these patients. An important issue cancer survivors face is an elevated risk of being diagnosed with new primary cancers. The primary aim of this analysis was to assess whether a history of rare cancer increases the risk of subsequent cancer compared to survivors of common cancers. METHODS: This was a prospective cohort study of 16,630 adults with personal and/or family history of cancer who were recruited from cancer clinics at 14 geographically dispersed US academic centers of the NIH-sponsored Cancer Genetics Network (CGN). Participants' self-reported cancer histories were collected at registration to the CGN and updated annually during follow-up. At enrollment, 14% of participants reported a prior rare cancer. Elevated risk was assessed via the cause-specific hazard ratio on the time to a subsequent cancer diagnosis. RESULTS: After a median follow-up of 7.9 years, relative to the participants who were unaffected at enrollment, those with a prior rare cancer had a 23% higher risk of subsequent cancer (95% CI: -1 to 52%), while those with a prior common cancer had no excess risk. Patients having two or more prior cancers were at a 53% elevated risk over those with fewer than two (95% CI: 21 to 94%) and if the multiple prior cancers were rare cancers, risk was further elevated by 47% (95% CI: 1 to 114%). CONCLUSION: There is evidence suggesting that survivors of rare cancers, especially those with multiple cancer diagnoses, are at an increased risk of a subsequent cancer. There is a need to study this population more closely to better understand cancer pathogenesis.

A pilot investigation of an iOS-based app for toilet training children with autism spectrum disorder.

We developed an iOS-based app with a transmitter/disposable sensor and corresponding manualized intervention for children with autism spectrum disorder. The app signaled the onset of urination, time-stamped accidents for analysis, reminded parents to reinforce intervals of continence, provided a visual outlet for parents to communicate reinforcement, and afforded opportunity for timely feedback from clinicians. We compared this intervention with an intervention that uses standard behavioral treatment in a pilot randomized controlled trial of 33 children with autism spectrum disorder aged 3-6 years with urinary incontinence. Parents in both groups received initial training and four booster consultations over 3 months. Results support the feasibility of parent-mediated toilet training studies (e.g., 84% retention rate, 92% fidelity of parent-implemented intervention). Parents used the app and related technology with few difficulties or malfunctions. There were no statistically significant group differences for rate of urine accidents, toilet usage, or satisfaction at close of intervention or 3-month follow-up; however, the alarm group trended toward greater rate of skill acquisition with significantly less day-to-day intervention. Further development of alarm and related technology and future comparative studies with a greater number of participants are warranted.

Graphing the Win Ratio and its components over time.

Clinical trials are often designed to compare treatments on the basis of multiple outcomes. For the analysis of the treatment comparison from such a trial, in 1999, the Finkelstein-Schoenfeld test was proposed, which was a generalization of the Gehan-Wilcoxon test based on pairwise comparison of patients on a primary outcome when possible but otherwise on a secondary outcome. In 2012, Pocock and colleagues suggested an estimate based on this concept, the Win Ratio, which summarized the ratio of the number of patients who fared better versus worse on the experimental arm. However, in 2016, Oakes noted that the Win Ratio could be a function of the distribution of follow-up times of the trial. The aim of this paper is to propose an approach to representing the Win Ratio graphically in such a way that the effect of time on the estimate would be apparent. In addition, the methods are used to display the contribution of each endpoint to the composite. We apply the methods to clinical trials in cancer, cardiology, and neurology. Software is available named winRatioAnalysis in CRAN.

Physical symptoms in long-term survivors of rare cancer.

PURPOSE: Advances in cancer detection and treatment have resulted in a growing population of long-term survivors, but even years after treatment has concluded, many survivors report physical symptoms that interfere with daily living. While there are studies of late effects following common cancers, less is known about these complications in rare cancers. This study focuses on the physical symptoms reported by long-term survivors enrolled in the NIH-sponsored Rare Cancer Genetics Registry. METHODS: The Rotterdam Symptom Checklist-Modified was administered to evaluate the severity of physical symptoms commonly reported by long-term cancer survivors. Logistic regression was used to assess association between symptoms and demographic and clinical factors. RESULTS: In 309 subjects with a median time of 7.6 years from a diagnosis of one or more rare cancers, the median number of symptoms present per participant was 7. The most prevalent symptom reported was tiredness/lack of energy, which was present/very bothersome in 70%/25% of registrants. Women, non-whites, current smokers, and upper GI cancer survivors are particularly affected. Overall, symptom prevalence was similar across rare cancer types, time since diagnosis, and type of treatment. CONCLUSIONS: Rare cancer survivors continue to experience troublesome symptoms many years after diagnosis, regardless of cancer type or treatment modality. IMPLICATIONS FOR CANCER SURVIVORS: There is a need for continued emphasis on smoking cessation in cancer survivors as well as enhanced monitoring of long-term complications in female, non-white, and upper GI cancer survivors.

Efficacy of Anti-HER2 Agents in Combination With Adjuvant or Neoadjuvant Chemotherapy for Early and Locally Advanced HER2-Positive Breast Cancer Patients: A Network Meta-Analysis.

BACKGROUND: Several (neo)adjuvant treatments for patients with HER2-positive breast cancer have been compared in different randomized clinical trials. Since it is not feasible to conduct adequate pairwise comparative trials of all these therapeutic options, network meta-analysis offers an opportunity for more detailed inference for evidence-based therapy. METHODS: Phase II/III randomized clinical trials comparing two or more different (neo)adjuvant treatments for HER2-positive breast cancer patients were included. Relative treatment effects were pooled in two separate network meta-analyses for overall survival (OS) and disease-free survival (DFS). RESULTS: 17 clinical trials met our eligibility criteria. Two different networks of trials were created based on the availability of the outcomes: OS network (15 trials: 37,837 patients); and DFS network (17 trials: 40,992 patients). Two studies-the ExteNET and the NeoSphere trials-were included only in this DFS network because OS data have not yet been reported. The concept of the dual anti-HER2 blockade proved to be the best option in terms of OS and DFS. Chemotherapy (CT) plus trastuzumab (T) and lapatinib (L) and CT + T + Pertuzumab (P) are probably the best treatment options in terms of OS, with 62.47% and 22.06%, respectively. In the DFS network, CT + T + Neratinib (N) was the best treatment option with 50.55%, followed by CT + T + P (26.59%) and CT + T + L (20.62%). CONCLUSION: This network meta-analysis suggests that dual anti-HER2 blockade with trastuzumab plus either lapatinib or pertuzumab are probably the best treatment options in the (neo)adjuvant setting for HER2-positive breast cancer patients in terms of OS gain. Mature OS results are still expected for the Aphinity trial and for the sequential use of trastuzumab followed by neratinib, the treatment that showed the best performance in terms of DFS in our analysis.

Expressed Gene Fusions as Frequent Drivers of Poor Outcomes in Hormone Receptor-Positive Breast Cancer.

We sought to uncover genetic drivers of hormone receptor-positive (HR+) breast cancer, using a targeted next-generation sequencing approach for detecting expressed gene rearrangements without prior knowledge of the fusion partners. We identified intergenic fusions involving driver genes, including PIK3CA, AKT3, RAF1, and ESR1, in 14% (24/173) of unselected patients with advanced HR+ breast cancer. FISH confirmed the corresponding chromosomal rearrangements in both primary and metastatic tumors. Expression of novel kinase fusions in nontransformed cells deregulates phosphoprotein signaling, cell proliferation, and survival in three-dimensional culture, whereas expression in HR+ breast cancer models modulates estrogen-dependent growth and confers hormonal therapy resistance in vitro and in vivo Strikingly, shorter overall survival was observed in patients with rearrangement-positive versus rearrangement-negative tumors. Correspondingly, fusions were uncommon (<5%) among 300 patients presenting with primary HR+ breast cancer. Collectively, our findings identify expressed gene fusions as frequent and potentially actionable drivers in HR+ breast cancer.Significance: By using a powerful clinical molecular diagnostic assay, we identified expressed intergenic fusions as frequent contributors to treatment resistance and poor survival in advanced HR+ breast cancer. The prevalence and biological and prognostic significance of these alterations suggests that their detection may alter clinical management and bring to light new therapeutic opportunities. Cancer Discov; 8(3); 336-53. ©2017 AACR.See related commentary by Natrajan et al., p. 272See related article by Liu et al., p. 354This article is highlighted in the In This Issue feature, p. 253.

Informatively clustering longitudinal microarrays using binary or survival outcome data.

The goal of this research is to discover what groups of genes are associated with the disease process. We use binary and failure time outcomes to inform the clustering of longitudinally-collected microarray data. We propose a linear model with normally distributed cluster-specific random effects for the longitudinal gene expression trajectory. The random effects are linearly related to a latent continuous representation of the outcome, where the probability or hazard of the outcome depends on these latent variables. We apply our method to microarray data collected from trauma patients in the Inflammation and Host Response to Injury project.

Early Detection of Ovarian Cancer using the Risk of Ovarian Cancer Algorithm with Frequent CA125 Testing in Women at Increased Familial Risk - Combined Results from Two Screening Trials.

Purpose: Women at familial/genetic ovarian cancer risk often undergo screening despite unproven efficacy. Research suggests each woman has her own CA125 baseline; significant increases above this level may identify cancers earlier than standard 6- to 12-monthly CA125 > 35 U/mL.Experimental Design: Data from prospective Cancer Genetics Network and Gynecologic Oncology Group trials, which screened 3,692 women (13,080 woman-screening years) with a strong breast/ovarian cancer family history or BRCA1/2 mutations, were combined to assess a novel screening strategy. Specifically, serum CA125 q3 months, evaluated using a risk of ovarian cancer algorithm (ROCA), detected significant increases above each subject's baseline, which triggered transvaginal ultrasound. Specificity and positive predictive value (PPV) were compared with levels derived from general population screening (specificity 90%, PPV 10%), and stage-at-detection was compared with historical high-risk controls.Results: Specificity for ultrasound referral was 92% versus 90% (P = 0.0001), and PPV was 4.6% versus 10% (P > 0.10). Eighteen of 19 malignant ovarian neoplasms [prevalent = 4, incident = 6, risk-reducing salpingo-oophorectomy (RRSO) = 9] were detected via screening or RRSO. Among incident cases (which best reflect long-term screening performance), three of six invasive cancers were early-stage (I/II; 50% vs. 10% historical BRCA1 controls; P = 0.016). Six of nine RRSO-related cases were stage I. ROCA flagged three of six (50%) incident cases before CA125 exceeded 35 U/mL. Eight of nine patients with stages 0/I/II ovarian cancer were alive at last follow-up (median 6 years).Conclusions: For screened women at familial/genetic ovarian cancer risk, ROCA q3 months had better early-stage sensitivity at high specificity, and low yet possibly acceptable PPV compared with CA125 > 35 U/mL q6/q12 months, warranting further larger cohort evaluation. Clin Cancer Res; 23(14); 3628-37. ©2017 AACR.

High-Content Biopsies Facilitate Molecular Analyses and Do Not Increase Complication Rates in Patients With Advanced Solid Tumors.

PURPOSE: Precision oncology relies on frequent pathologic, molecular, and genomic assessments of tumor tissue to guide treatment selection, evaluate pharmacodynamic effects of novel agents, and determine drug resistance mechanisms. Newer forms of analyses such as drug screens in cell lines and patient-derived xenografts demand increasing amounts of tissue material. It remains unknown how the need for serial biopsies with large numbers of tumor cores relates to tissue yields and biopsy complication rates. MATERIALS AND METHODS: In this study, we performed a retrospective analysis of 199 focal liver biopsies performed in 143 patients in the setting of oncologic research protocols (research biopsy group) over a 4-year period at a single-intervention oncology service. Practice patterns and complication rates were compared with those related to 1,522 consecutive biopsies performed in 1,154 patients in whom two cores were obtained for standard clinical management of patients (standard biopsy). RESULTS: In the research biopsy group, 1,100 tissue cores (average, 5.5 cores per procedure) were harvested and distributed to trial sponsors, internal research laboratories, and pathology services. The complication rate in this cohort was 0.5% for major complications (one of 199) and 1.0% for minor complications managed conservatively (two of 199). In the standard biopsy control group, major complications were observed in 1.4% of procedures (22 of 1,522) and minor complications in 0.2% (three of 1,522). These complication rates were not statistically different. CONCLUSION: Harvesting extra tissue cores through coaxial needles during focal liver biopsies does not increase complication rates and yields valuable tissue for additional experimental testing.

Physical and psychological health in rare cancer survivors.

PURPOSE: Registries provide a unique tool for tracking quality of life in rare cancer survivors, whose survivorship experience is less known than for common cancers. This paper reports on these outcomes in 321 patients enrolled in the Rare Cancer Genetics Registry diagnosed with rare gastrointestinal, genitourinary, gynecologic, sarcoma, head/neck, or hematologic cancers. METHODS: Four outcomes were assessed, reflecting registrants' self-reported physical and mental health, psychological distress, and loneliness. Combining all patients into a single analysis, regression was used to evaluate the association between outcomes and socio-demographic and clinical factors. RESULTS: Median time since diagnosis was 3 years (range 0-9); 69 % were no longer in treatment. Poorer physical health was reported in registrants who were older at diagnosis, unmarried, and still in treatment. Poorer mental status was associated with younger diagnosis age and unmarried status. Psychological distress varied by cancer type and was higher among currently treated and unmarried registrants. Greater loneliness was reported in registrants with gynecological cancers, and those who were less educated or unmarried. The physical and mental health profile of rare cancer survivors is similar to what is reported for common cancers. CONCLUSIONS: Unmarried participants reported poorer outcomes on all measures of quality of life. Furthermore, physical and mental health were not significantly different by cancer type after adjustment for diagnosis age, whether currently in treatment and marital status. Thus, the combined analysis performed here is a useful way to analyze outcomes in less common diseases. Our findings could be valuable in guiding evaluation and intervention for issues impacting quality of life. IMPLICATIONS FOR CANCER SURVIVORS: Rare cancer survivors, particularly those without spousal support, should be monitored for challenges to the physical as well as psychological aspects of quality of life.

Knowledge and Uptake of Genetic Counseling and Colonoscopic Screening Among Individuals at Increased Risk for Lynch Syndrome and their Endoscopists from the Family Health Promotion Project.

OBJECTIVES: Individuals whose families meet the Amsterdam II clinical criteria for hereditary non-polyposis colorectal cancer are recommended to be referred for genetic counseling and to have colonoscopic screening every 1-2 years. To assess the uptake and knowledge of guideline-based genetic counseling and colonoscopic screening in unaffected members of families who meet Amsterdam II criteria and their treating endoscopists. METHODS: Participants in the Family Health Promotion Project who met the Amsterdam II criteria were surveyed regarding their knowledge of risk-appropriate guidelines for genetic counseling and colonoscopy screening. Endoscopy/pathology reports were obtained from patients screened during the study to determine the follow-up recommendations made by their endoscopists. Survey responses were compared using Fisher's Exact and the χ(2) test. Concordance in participant/provider-reported surveillance interval was assessed using the kappa statistic. RESULTS: Of the 165 participants, the majority (98%) agreed that genetics and family history are important predictors of CRC, and 63% had heard of genetic testing for CRC, although only 31% reported being advised to undergo genetic counseling by their doctor, and only 7% had undergone genetic testing. Only 26% of participants reported that they thought they should have colonoscopy every 1-2 years and 30% of endoscopists for these participants recommended 1-2-year follow-up colonoscopy. There was a 65% concordance (weighted kappa 0.42, 95% CI 0.24-0.61) between endoscopist recommendations and participant reports regarding screening intervals. CONCLUSIONS: A minority of individuals meeting Amsterdam II criteria in this series have had genetic testing and reported accurate knowledge of risk-appropriate screening, and only a small percentage of their endoscopists provided them with the appropriate screening recommendations. There was moderate concordance between endoscopist recommendations and participant knowledge suggesting that future educational interventions need to target both health-care providers and their patients.

Identification of early breast cancer patient cohorts who may benefit from lapatinib therapy.

In resource-constrained environments many patients with human epidermal growth factor receptor 2 (HER2)+ early breast cancer are currently not offered adjuvant anti-HER2 therapy. For patients who might be able to receive the tyrosine kinase inhibitor (TKI) lapatinib (e.g. after patent expiration), it is important to identify subgroups of patients for whom anti-HER2 TKI therapy could be beneficial. To do this, we used data from 2489 patients with centrally confirmed HER2+ disease enrolled in the adjuvant Tykerb Evaluation After Chemotherapy (TEACH) trial, investigating the effect of lapatinib in patients with HER2+ early breast cancer not treated with trastuzumab. We performed subgroup analyses and number-needed-to-treat (NNT) calculations using patient and tumour associated predictors. Hormone receptor negative (HR-) patients on lapatinib had a significantly prolonged disease-free survival (DFS) compared to HR- patients on placebo (hazard ratio 0.64, P=0.003). For patients with HR- disease, starting treatment with lapatinib ≤1 year from diagnosis improved DFS by 12.1% [2.1-22.1] at 2 years and 15.7% [4.1-27.2] at 5 years. Depending on lymph node status and time since diagnosis the NNT for recurrence (at 5 years) was between 5.9 (node positive patients <1 year from diagnosis) and 15.9. These numbers are in range with numbers reported for up-front adjuvant trastuzumab for HR unselected patients (e.g. 15.6 for DFS at 4 years in HERA). In a subgroup analysis of the adjuvant TEACH trial, we show that anti-HER2 monotherapy with a TKI is beneficial as adjuvant therapy in a subgroup of patients. NNT in HER2+ HR- patients are in range with those reported from up-front adjuvant trastuzumab trials.

Assessing survival benefit when treatment delays disease progression.

BACKGROUND: For a potentially lethal chronic disease like cancer, it is often infeasible to compare treatments on the basis of overall survival, so a combined outcome such as progression-free survival (which is the time from randomization to progression or death) has become an acceptable primary endpoint. The rationale of using an efficacy measure that is dominated by the time to progression is that an effective treatment will delay progression and when treatment is stopped at progression, the effect of treatment after this time is small. However, often trials that show a significant benefit for delaying progression but not on overall survival are not universally viewed as providing convincing evidence that the drug should become the standard of care. METHODS: We propose that when there is a significant treatment effect of delaying progression, a Bayesian analysis of overall survival should be undertaken. We suggest using a joint piecewise exponential model, where the treatment effect on the hazard for progression and for death after progression is captured through two distinct parameters. We develop a plot of the overall survival advantage of the new therapy versus the prior distribution of the relative hazard for death after progression. This plot can augment the discussion about whether the new treatment is beneficial on survival. RESULTS: In the example of an early breast cancer trial for which a new treatment significantly delayed disease recurrence, our Bayesian analysis showed that with very reasonable assumptions on the effects of treatment after recurrence, there is a high probability that the new treatment improves overall survival. CONCLUSION: For a clinical trial for which treatment delays progression, the proposed method can improve the interpretability of the survival comparison using data from the study.

Global rank tests for multiple, possibly censored, outcomes.

Clinical trials often collect multiple outcomes on each patient, as the treatment may be expected to affect the patient on many dimensions. For example, a treatment for a neurological disease such as ALS is intended to impact several dimensions of neurological function as well as survival. The assessment of treatment on the basis of multiple outcomes is challenging, both in terms of selecting a test and interpreting the results. Several global tests have been proposed, and we provide a general approach to selecting and executing a global test. The tests require minimal parametric assumptions, are flexible about weighting of the various outcomes, and are appropriate even when some or all of the outcomes are censored. The test we propose is based on a simple scoring mechanism applied to each pair of subjects for each endpoint. The pairwise scores are then reduced to a summary score, and a rank-sum test is applied to the summary scores. This can be seen as a generalization of previously proposed nonparametric global tests (e.g., O'Brien, 1984). We discuss the choice of optimal weighting schemes based on power and relative importance of the outcomes. As the optimal weights are generally unknown in practice, we also propose an adaptive weighting scheme and evaluate its performance in simulations. We apply the methods to analyze the impact of a treatment on neurological function and death in an ALS trial.

Outcome-Driven Cluster Analysis with Application to Microarray Data.

One goal of cluster analysis is to sort characteristics into groups (clusters) so that those in the same group are more highly correlated to each other than they are to those in other groups. An example is the search for groups of genes whose expression of RNA is correlated in a population of patients. These genes would be of greater interest if their common level of RNA expression were additionally predictive of the clinical outcome. This issue arose in the context of a study of trauma patients on whom RNA samples were available. The question of interest was whether there were groups of genes that were behaving similarly, and whether each gene in the cluster would have a similar effect on who would recover. For this, we develop an algorithm to simultaneously assign characteristics (genes) into groups of highly correlated genes that have the same effect on the outcome (recovery). We propose a random effects model where the genes within each group (cluster) equal the sum of a random effect, specific to the observation and cluster, and an independent error term. The outcome variable is a linear combination of the random effects of each cluster. To fit the model, we implement a Markov chain Monte Carlo algorithm based on the likelihood of the observed data. We evaluate the effect of including outcome in the model through simulation studies and describe a strategy for prediction. These methods are applied to trauma data from the Inflammation and Host Response to Injury research program, revealing a clustering of the genes that are informed by the recovery outcome.