Regulatory Issues of Platform Trials: Learnings from EU-PEARL.

Although platform trials have many benefits, the complexity of these designs may result not only in increased methodological but also regulatory and ethical challenges. These aspects were addressed as part of the IMI project EU Patient-Centric Clinical Trial Platforms (EU-PEARL). We reviewed the available guidelines on platform trials in the European Union and the United States. This is supported and complemented by feedback received from regulatory interactions with the European Medicines Agency and the US Food and Drug Administration. Throughout the project we collected the needs of all relevant stakeholders including ethics committees, regulators, and health technology assessment bodies through active dialog and dedicated stakeholder workshops. Furthermore, we focused on methodological aspects and where applicable identified the corresponding guidance. Learnings from the guideline review, regulatory interactions, and workshops are provided. Based on these, a master protocol template was developed. Issues that still need harmonization or clarification in guidelines or where further methodological research is needed are also presented. These include questions around clinical trial submissions in Europe, the need for multiplicity control across the whole master protocol, the use of non-concurrent controls, and the impact of different randomization schemes. Master protocols are an efficient and patient-centered clinical trial design that can expedite drug development. However, they can also introduce additional operational and regulatory complexities. It is important to understand the different requirements of stakeholders upfront and address them in the trial. While relevant guidance is increasing, early dialog with relevant stakeholders can help to further support such designs.

Adaptive platform trials: the impact of common controls on type one error and power.

Platform trials offer a framework to study multiple interventions in one trial with the opportunity of opening and closing arms. The use of common controls can increase efficiency as compared to individual controls. The need for multiplicity adjustment because of common controls is currently a debate among researchers, pharmaceutical companies, and regulators. The impact of common controls on the type one error in a fixed platform trial, i.e. when all treatments start and end recruitment at the same time, has been discussed in the literature before. We complement these findings by investigating the impact of a common control on the type one error and power in a flexible platform trial, i.e. when one arm joins the platform later. We derived the correlation of test statistics to assess the impact of the overlap and compared the results to a trial with individual controls. Furthermore, we evaluate the power, and the impact of multiplicity adjustment on the power in fixed and flexible platform trials. These methodological considerations are complemented by a regulatory guideline review. With multiple arms, the FWER is inflated when no multiplicity adjustment is applied. However, the FWER inflation is smaller with common controls than with individual controls. Even after multiplicity adjustment, a trial with common controls is often beneficial in terms of sample size and power. However, in some cases, the trial with common controls loses the efficiency gain and it might be advisable to run a separate trial rather than joining a platform trial.

Stochastic Dispersal Rather Than Deterministic Selection Explains the Spatio-Temporal Distribution of Soil Bacteria in a Temperate Grassland.

Spatial and temporal processes shaping microbial communities are inseparably linked but rarely studied together. By Illumina 16S rRNA sequencing, we monitored soil bacteria in 360 stations on a 100 square meter plot distributed across six intra-annual samplings in a rarely managed, temperate grassland. Using a multi-tiered approach, we tested the extent to which stochastic or deterministic processes influenced the composition of local communities. A combination of phylogenetic turnover analysis and null modeling demonstrated that either homogenization by unlimited stochastic dispersal or scenarios, in which neither stochastic processes nor deterministic forces dominated, explained local assembly processes. Thus, the majority of all sampled communities (82%) was rather homogeneous with no significant changes in abundance-weighted composition. However, we detected strong and uniform taxonomic shifts within just nine samples in early summer. Thus, community snapshots sampled from single points in time or space do not necessarily reflect a representative community state. The potential for change despite the overall homogeneity was further demonstrated when the focus shifted to the rare biosphere. Rare OTU turnover, rather than nestedness, characterized abundance-independent ß-diversity. Accordingly, boosted generalized additive models encompassing spatial, temporal and environmental variables revealed strong and highly diverse effects of space on OTU abundance, even within the same genus. This pure spatial effect increased with decreasing OTU abundance and frequency, whereas soil moisture - the most important environmental variable - had an opposite effect by impacting abundant OTUs more than the rare ones. These results indicate that - despite considerable oscillation in space and time - the abundant and resident OTUs provide a community backbone that supports much higher ß-diversity of a dynamic rare biosphere. Our findings reveal complex interactions among space, time, and environmental filters within bacterial communities in a long-established temperate grassland.

Publisher Correction: Optimizing the dynamics of protein expression.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

EMA Review of Axicabtagene Ciloleucel (Yescarta) for the Treatment of Diffuse Large B-Cell Lymphoma.

On June 28, 2018, the Committee for Advanced Therapies and the Committee for Medicinal Products for Human Use adopted a positive opinion, recommending the granting of a marketing authorization for the medicinal product Yescarta for the treatment of adult patients with relapsed or refractory diffuse large B-cell lymphoma and primary mediastinal large B-cell lymphoma, after two or more lines of systemic therapy. Yescarta, which was designated as an orphan medicinal product and included in the European Medicines Agency's Priority Medicines scheme, was granted an accelerated review timetable. The active substance of Yescarta is axicabtagene ciloleucel, an engineered autologous T-cell immunotherapy product whereby a patient's own T cells are harvested and genetically modified ex vivo by retroviral transduction using a retroviral vector to express a chimeric antigen receptor (CAR) comprising an anti-CD19 single chain variable fragment linked to CD28 costimulatory domain and CD3-zeta signaling domain. The transduced anti-CD19 CAR T cells are expanded ex vivo and infused back into the patient, where they can recognize and eliminate CD19-expressing cells. The benefits of Yescarta as studied in ZUMA-1 phase II (NCT02348216) were an overall response rate per central review of 66% (95% confidence interval, 56%-75%) at a median follow-up of 15.1 months in the intention to treat population and a complete response rate of 47% with a significant duration. The most common adverse events were cytokine release syndrome, neurological adverse events, infections, pyrexia, diarrhea, nausea, hypotension, and fatigue. IMPLICATIONS FOR PRACTICE: Yescarta (axicabtagene ciloleucel) was the first chimeric antigen receptor T-cell therapy to be submitted for evaluation to the European Medicines Agency and admitted into the "priority medicine" scheme; it was granted accelerated assessment on the basis of anticipated clinical benefit in relapsed/refractory diffuse large B-cell lymphoma, a condition of unmet medical need. Indeed, Yescarta showed an overall response rate of 66% and a complete response rate of 47% with a significant duration and a manageable toxicity that compared very favorably with historical controls. Here the analysis of benefits and risks is presented, and specific challenges with this important novel product are highlighted, providing further insights and reflections for future medical research.

Randomized phase-III-trial of concurrent chemoradiation for locally advanced head and neck cancer comparing dose reduced radiotherapy with paclitaxel/cisplatin to standard radiotherapy with fluorouracil/cisplatin: The PacCis-trial.

BACKGROUND AND PURPOSE: This multicenter, phase 3 trial investigates whether the incorporation of concurrent paclitaxel and cisplatin together with a reduced total dose of radiotherapy is superior to standard fluorouracil-cisplatin based CRT. MATERIALS AND METHODS: Patients with SCCHN, stage III-IVB, were randomized to receive paclitaxel/cisplatin (PacCis)-CRT (arm A; paclitaxel 20 mg/m2 on days 2, 5, 8, 11 and 25, 30, 33, 36; cisplatin 20 mg/m2, days 1-4 and 29-32; RT to a total dose of 63.6 Gy) or fluorouracil/cisplatin (CisFU)-CRT (arm B; fluorouracil 600 mg/m2; cisplatin 20 mg/m2, days 1-5 and 29-33; RT: 70.6 Gy). Endpoint was 3-year-disease free survival (3y-DFS). RESULTS: A total of 221 patients were enrolled between 2010 and 2015. With a median follow-up of 3.7 years, 3y-DFS in the CisFU arm and PacCis arm was 58.2% and 48.4%, respectively (HR 0.82, 95% CI 0.56-1.21, p = 0.52). The 3y-OS amounted to 64.6% in the CisFU arm, and to 59.2% in the PacCis arm (HR 0.82, 95% CI 0.54-1.24, p = 0.43). In the subgroup of p16-positive oropharyngeal carcinomas, 3y-DFS and 3y-OS was 84.6% vs 83.9% (p = 0.653), and 92.3% vs. 83.5% (p = 0.76) in arm A and B, respectively. Grade 3-4 hematological toxicities were significantly reduced in arm A (anemia, p = 0.01; leukocytopenia, p = 0.003), whereas grade 3 infections were reduced in arm B (p = 0.01). CONCLUSION: Paclitaxel/cisplatin-CRT with a reduced RT-dose is not superior to standard fluorouracil/cisplatin-CRT. Subgroup analyses indicate that a reduced radiation dose seems to be sufficient for p16+ oropharyngeal cancer or non-smokers. CLINICAL TRIAL INFORMATION: NCT01126216; EudraCT Number 2005-003484-23.

Current Statistical Considerations and Regulatory Perspectives on the Planning of Confirmatory Basket, Umbrella, and Platform Trials.

Master protocols have received a growing interest during the last years. By assigning patients to specific substudies, they aim at targeting and accelerating clinical development. Given their complexity, basket, umbrella, and platform designs have raised challenging regulatory and statistical questions, especially the control of multiplicity in confirmatory trials. In basket trials, regulatory assessment of the benefit/risk in pooled populations and choice of the treatment indication is challenging. We provide here our perspectives on these topics. In master protocols, as long as the statistical hypotheses tested between the different substudies are independent, no supplementary adjustment for multiplicity over the different substudies should be required. Moreover, sharing a control arm within an umbrella or a platform trial investigating different drugs would not require a correction for the type I error rate, whereas the chance of multiple false positive regulatory decisions should be recognized. In basket trials, pooling across substudies requires a rationale supporting the intended indication and should be preplanned. Assessment of the benefit/risk in pooled target populations can be complicated by differences in design or in efficacy/safety signals between the substudies. While trials governed by a master protocol can offer logistic and financial advantages, more experience is needed to gain a deeper insight into this novel framework.

Long-term Results of Conservative Therapy of Adolescent Idiopathic Scoliosis Using the Cheneau Brace.

Adolescent idiopathic scoliosis is a disease of the growing skeleton. The goal in therapy is to decelerate or to prevent progression of the spinal curve. As part of a retrospective study the patient group of the scoliosis surgery from 1995-2016 was analysed according to the inclusion criteria of the Scoliosis research Society (SRS). 159 of 643 of the patients fullfilled the specific criteria. The assessment of effectiveness was based on the progression of the angle of curvature. If it was ≤5° it was judged positively. The therapy in accordance to Cheneau is a full-time bracing orthosis. Data from 159 patients - 136 fem. (85.5%), 23 male (14.5%) was analysed. The average age at the beginning of treatment was 13.3±1.7 years. The average duration was 3.47±1.2 years. The average Cobb-angle before treatment was 28.39°±9.44°. At the completion is was 27.7°±12.34°. Stabilisation of scoliosis (≤5°) was achieved for 136 of the patients (85.5%). The final follow-up showed a progression of the Cobb angel>5° in 23 cases, of which 19 had to undergo secondary surgery. The length of therapy had a positive influence (p=0.057) on the result. Brace treatment constitutes an effective method of therapy at curvatures between 20-40°. Short duration of therapy correlates with a expressively increased risk towards progression (p=0.057). The Cheneau brace treatment constitutes an effective treatment at curvature angles between 20-40°. The risk of progression can be reduced by a timely and correct identification.

[Current challenges in the assessment of ethical proposals-aspects of digitalization and personalization in the healthcare system]. / Aktuelle Herausforderungen bei der Bewertung von Ethikanträgen ­ Aspekte der Digitalisierung und Personalisierung im Gesundheitswesen.

The global aim of medical ethics committees is to judge the scientific quality and the integrity of the content of medical research projects (studies), thereby assessing the benefit-risk profile. Apart from judging content-related aspects and the legal correctness, the study design and the analysis strategy must also be assessed from a biostatistical point of view. This very sophisticated task is further complicated by the fact that medical research constantly faces new challenges.Within this work, current developments in medical research that directly impact the assessability of ethical proposals will be identified and discussed. The aim is to sensitize researchers to the opportunities and challenges of new developments.The work focusses on the topics of digitalization in the healthcare system and individualized medicine. The authors illustrate some problems resulting from these developments that affect the ethical justification of medical research projects. Problems related to medical as well as biostatistical aspects are presented and their direct implications on the legal justification and ethical and moral conceptual integrity are highlighted.New developments in medical research such as digitalization and individualized medicine offer new perspectives for optimized therapies. These promising developments must be further advanced. A critical view on the so far only poorly investigated consequences of embedding new data sources and study designs must urgently accompany this process. Transparency and clarity in formulating ethical proposals is thereby of utmost importance.

Optimizing the dynamics of protein expression.

Heterologously expressed genes require adaptation to the host organism to ensure adequate levels of protein synthesis, which is typically approached by replacing codons by the target organism's preferred codons. In view of frequently encountered suboptimal outcomes we introduce the codon-specific elongation model (COSEM) as an alternative concept. COSEM simulates ribosome dynamics during mRNA translation and informs about protein synthesis rates per mRNA in an organism- and context-dependent way. Protein synthesis rates from COSEM are integrated with further relevant covariates such as translation accuracy into a protein expression score that we use for codon optimization. The scoring algorithm further enables fine-tuning of protein expression including deoptimization and is implemented in the software OCTOPOS. The protein expression score produces competitive predictions on proteomic data from prokaryotic, eukaryotic, and human expression systems. In addition, we optimized and tested heterologous expression of manA and ova genes in Salmonella enterica serovar Typhimurium. Superiority over standard methodology was demonstrated by a threefold increase in protein yield compared to wildtype and commercially optimized sequences.

Natalizumab-associated progressive multifocal leukoencephalopathy in Germany.

OBJECTIVE: To evaluate characteristics relevant to diagnosis of JC polyomavirus-associated progressive multifocal leukoencephalopathy (PML), and PML risk stratification in a large national cohort of patients with multiple sclerosis during therapy with natalizumab. METHODS: Analysis of 292 adverse drug reaction forms on suspected cases of PML reported to the German national competent authority until July 2017. Patients not fulfilling PML diagnostic criteria or with insufficient information available were excluded. RESULTS: Of the 142 confirmed patients with PML, 72.3% (95% confidence interval [CI] 64.4%-79.1%) were women, and the median age was 43 years (range 19-69). Of these patients, 7.7% (95% CI 4.3%-13.5%) were clinically asymptomatic at time of PML diagnosis. PML was fatal in 9.1% (95% CI 5.3%-15.1%) of the patients. Infratentorial lesions on imaging were reported in 40% (95% CI 32.0%-48.6%) of the patients. JC polyomavirus DNA in CSF was undetectable at time of first analysis in 23.8% (95% CI 17.3%-31.9%) of the patients. Three patients tested negative for anti-JC polyomavirus antibodies within 6 to 18 months before PML diagnosis, with seroconversion confirmed 5.5 months, 7 months (in a post hoc analysis only), or at time of PML diagnosis. CONCLUSIONS: JC polyomavirus DNA detection in CSF has limited sensitivity in early PML, and clinical and imaging presentation may be atypical. Thus, critical revision of current PML diagnostic criteria is warranted. Negative anti-JC polyomavirus antibodies in sera do not preclude the later development of PML. This emphasizes the need for close and regular serologic, imaging, and clinical monitoring in patients treated with natalizumab.

Modeling spatiotemporal abundance of mobile wildlife in highly variable environments using boosted GAMLSS hurdle models.

Modeling organism distributions from survey data involves numerous statistical challenges, including accounting for zero-inflation, overdispersion, and selection and incorporation of environmental covariates. In environments with high spatial and temporal variability, addressing these challenges often requires numerous assumptions regarding organism distributions and their relationships to biophysical features. These assumptions may limit the resolution or accuracy of predictions resulting from survey-based distribution models. We propose an iterative modeling approach that incorporates a negative binomial hurdle, followed by modeling of the relationship of organism distribution and abundance to environmental covariates using generalized additive models (GAM) and generalized additive models for location, scale, and shape (GAMLSS). Our approach accounts for key features of survey data by separating binary (presence-absence) from count (abundance) data, separately modeling the mean and dispersion of count data, and incorporating selection of appropriate covariates and response functions from a suite of potential covariates while avoiding overfitting. We apply our modeling approach to surveys of sea duck abundance and distribution in Nantucket Sound (Massachusetts, USA), which has been proposed as a location for offshore wind energy development. Our model results highlight the importance of spatiotemporal variation in this system, as well as identifying key habitat features including distance to shore, sediment grain size, and seafloor topographic variation. Our work provides a powerful, flexible, and highly repeatable modeling framework with minimal assumptions that can be broadly applied to the modeling of survey data with high spatiotemporal variability. Applying GAMLSS models to the count portion of survey data allows us to incorporate potential overdispersion, which can dramatically affect model results in highly dynamic systems. Our approach is particularly relevant to systems in which little a priori knowledge is available regarding relationships between organism distributions and biophysical features, since it incorporates simultaneous selection of covariates and their functional relationships with organism responses.

The betaboost package-a software tool for modelling bounded outcome variables in potentially high-dimensional epidemiological data.

Motivation: To provide an integrated software environment for model fitting and variable selection in regression models with a bounded outcome variable. Implementation: The proposed modelling framework is implemented in the add-on package betaboost of the statistical software environment R. General features: The betaboost methodology is based on beta-regression, which is a state-of-the-art method for modelling bounded outcome variables. By combining traditional model fitting techniques with recent advances in statistical learning and distributional regression, betaboost allows users to carry out data-driven variable and/or confounder selection in potentially high-dimensional epidemiological data. The software package implements a flexible routine to incorporate linear and non-linear predictor effects in both the mean and the precision parameter (relating inversely to the variance) of a beta-regression model. Availability: The software is hosted publicly at [http://github.com/boost-R/betaboost] and has been published under General Public License (GPL) version 3 or newer.

Bouldering psychotherapy reduces depressive symptoms even when general physical activity is controlled for: A randomized controlled trial.

BACKGROUND: Bouldering psychotherapy (BPT) combines psychotherapeutic elements with physical activity (PA). It might be effective for reducing symptoms of depression, but so far, no study has assessed individuals' levels of PA to control for whether positive effects on depression can also be found when adjusting for participants' levels of PA. This is important because PA itself has been proven effective in reducing depression and therefore might be an important variable to account for - especially in therapies using sport as one therapeutic mechanism. METHODS: Using a waitlist control group design, outpatients with depression were assessed at baseline and after eight, 16, and 24 weeks. The intervention group took part in an eight-week bouldering psychotherapy which met once a week for three hours. Self-report measures before and after the intervention included the Symptom Checklist-90-R (SCL-90-R), the Beck Depression Inventory (BDI-II), and the questionnaire on resources and self-management skills (FERUS). PA was assessed during the first 16-week period via FitBit Zip accelerometers. RESULTS: Altogether, 47 complete cases (20 men and 27 women) were included in the final analyses. Depression scores dropped by up to 6.74 (CI 2.80-10.67) points on the SCL-90-R depression scale and by up to 8.26 (CI 4.21-12.31) points on the BDI-II during the BPT intervention, the control group remained stable (SCL-90-R Cohen's d = 0.60; BDI-II: Cohen's d = .50). All Participants accrued an average of 6,515 steps per day, which is considered "low-active." Participants of the BPT intervention were significantly more likely to reduce their depressive symptoms (p = .025) than participants of the control group, even when PA was controlled for in a regression analysis. LIMITATIONS: Limitations of the study are the relatively small number of patients and the assessment of outcome scores via self-report. CONCLUSIONS: This study provides evidence that short-term BPT can be effective for reducing symptoms of depression even if controlled for other therapeutically active confounders including antidepressant medication, psychotherapy and general level of PA.

Quantification of Renal Function and Cardiovascular Mortality in Patients Admitted to the Emergency Department with Suspected Acute Coronary Syndromes.

BACKGROUND: Increases in the novel serum marker cystatin C are detectable much earlier in the course of chronic kidney disease (CKD) even when levels of serum creatinine are still in the normal range. A major factor causing a decrease in serum creatinine is increasing age. Patients with CKD are more likely to develop cardiovascular disease (CVD) than a healthy population and to suffer premature deaths from CVD related to CKD. The aim of this study was to investigate whether cystatin C, serum creatinine, and estimated glomerular filtration rate (eGFR) predict cardiovascular mortality in patients admitted to the emergency department (ED) with suspected acute coronary syndromes (ACS). METHODS: In 1,282 patients (mean age 62 ± 15 years, 477 women, 805 men) with suspected ACS, baseline cystatin C concentrations, serum creatinine, and estimated glomerular filtration rate (eGFR) were measured at the ED. Clinical assessment and serial high sensitivity cardiac troponin T (hs-cTnT) measurements were used for the diagnosis of ACS. Seventeen cardiovascular deaths were registered during a median follow-up of 365 days. RESULTS: HRs from univariate Cox regression models for each of the potential biomarkers were 12.02 (95% CI 5.10 - 28.34) for cystatin C, 4.53 (1.75 - 11.70) for serum creatinine, and 0.97 (0.96 - 0.99) for eGFR. All three biomarkers showed a significant association with cardiovascular mortality in univariate analyses. The HRs from a model with all three potential biomarkers were 59.21 (95% CI 9.69 - 361.76) for cystatin C, 0.08 (0.01 - 0.58) for serum creatinine, and 0.98 (0.96 - 1.01) for eGFR. The risk association was significant for ln (cystatin C) and ln (serum creatinine). CONCLUSIONS: Results of this prospective study show that the quantification of renal function using cystatin C is useful for predicting cardiovascular mortality in patients with suspected ACS at the ED.

Pathway-Based Kernel Boosting for the Analysis of Genome-Wide Association Studies.

The analysis of genome-wide association studies (GWAS) benefits from the investigation of biologically meaningful gene sets, such as gene-interaction networks (pathways). We propose an extension to a successful kernel-based pathway analysis approach by integrating kernel functions into a powerful algorithmic framework for variable selection, to enable investigation of multiple pathways simultaneously. We employ genetic similarity kernels from the logistic kernel machine test (LKMT) as base-learners in a boosting algorithm. A model to explain case-control status is created iteratively by selecting pathways that improve its prediction ability. We evaluated our method in simulation studies adopting 50 pathways for different sample sizes and genetic effect strengths. Additionally, we included an exemplary application of kernel boosting to a rheumatoid arthritis and a lung cancer dataset. Simulations indicate that kernel boosting outperforms the LKMT in certain genetic scenarios. Applications to GWAS data on rheumatoid arthritis and lung cancer resulted in sparse models which were based on pathways interpretable in a clinical sense. Kernel boosting is highly flexible in terms of considered variables and overcomes the problem of multiple testing. Additionally, it enables the prediction of clinical outcomes. Thus, kernel boosting constitutes a new, powerful tool in the analysis of GWAS data and towards the understanding of biological processes involved in disease susceptibility.

An Update on Statistical Boosting in Biomedicine.

Statistical boosting algorithms have triggered a lot of research during the last decade. They combine a powerful machine learning approach with classical statistical modelling, offering various practical advantages like automated variable selection and implicit regularization of effect estimates. They are extremely flexible, as the underlying base-learners (regression functions defining the type of effect for the explanatory variables) can be combined with any kind of loss function (target function to be optimized, defining the type of regression setting). In this review article, we highlight the most recent methodological developments on statistical boosting regarding variable selection, functional regression, and advanced time-to-event modelling. Additionally, we provide a short overview on relevant applications of statistical boosting in biomedicine.

Perioperative complications of scoliosis surgery in patients with Duchenne muscular dystrophy and spinal muscular atrophy, focussing on wound healing disorders.

PURPOSE: Patients with Duchenne muscular dystrophy (DMD) or spinal muscular atrophy (SMA), both neuromuscular diseases, sustain spinal scoliosis in the course of their disease. To reduce the concomitant major morbidity and to improve their quality of life, patients require surgical spine stabilization. This can lead to complications like respiratory, cardiac or neurological complications or wound healing disorders (WHD). To find out the different complexities and risk factors increasing the chance to develop a WHD, the inpatient database was analyzed. METHODS: We performed a retrospective statistical study. Therefore, we analyzed the inpatient database of 180 patients (142 DMD and 38 SMA patients). The focus was on WHD. To figure out the risk factors leading to WHD, we conducted a logistic regression. RESULTS: Cardiac complications occurred most frequently, followed by pulmonary complications and neurological lesions. Fifty-seven out of 180 patients developed a WHD. In 23 cases the WHD was aseptic, in the other 34 cases dermal organisms, Pseudomonas species and intestinal organisms were responsible. By means of the logistic regression, we were able to identify two more risk factors, in addition to diagnosis and gender, for developing a WHD in our patients: the year of surgery and the direction of pelvic tilt. CONCLUSIONS: Most common complications following scoliosis surgery are respiratory and cardiac complications. WHD is a severe complication that implies a prolonged therapy. Some risk factors for developing WHD could be identified in this analysis. Specifically, these were the date of surgery and the direction of pelvic tilt.

Boosting the discriminatory power of sparse survival models via optimization of the concordance index and stability selection.

BACKGROUND: When constructing new biomarker or gene signature scores for time-to-event outcomes, the underlying aims are to develop a discrimination model that helps to predict whether patients have a poor or good prognosis and to identify the most influential variables for this task. In practice, this is often done fitting Cox models. Those are, however, not necessarily optimal with respect to the resulting discriminatory power and are based on restrictive assumptions. We present a combined approach to automatically select and fit sparse discrimination models for potentially high-dimensional survival data based on boosting a smooth version of the concordance index (C-index). Due to this objective function, the resulting prediction models are optimal with respect to their ability to discriminate between patients with longer and shorter survival times. The gradient boosting algorithm is combined with the stability selection approach to enhance and control its variable selection properties. RESULTS: The resulting algorithm fits prediction models based on the rankings of the survival times and automatically selects only the most stable predictors. The performance of the approach, which works best for small numbers of informative predictors, is demonstrated in a large scale simulation study: C-index boosting in combination with stability selection is able to identify a small subset of informative predictors from a much larger set of non-informative ones while controlling the per-family error rate. In an application to discover biomarkers for breast cancer patients based on gene expression data, stability selection yielded sparser models and the resulting discriminatory power was higher than with lasso penalized Cox regression models. CONCLUSION: The combination of stability selection and C-index boosting can be used to select small numbers of informative biomarkers and to derive new prediction rules that are optimal with respect to their discriminatory power. Stability selection controls the per-family error rate which makes the new approach also appealing from an inferential point of view, as it provides an alternative to classical hypothesis tests for single predictor effects. Due to the shrinkage and variable selection properties of statistical boosting algorithms, the latter tests are typically unfeasible for prediction models fitted by boosting.