Between-patient Heterogeneity Research Articles

Beyond pain: Using Unsupervised Machine Learning to Identify Phenotypic Clusters of Small Fiber Neuropathy.

Small fiber neuropathy (SFN) is characterized by dysfunction and loss of peripheral unmyelinated and thinly myelinated nerve fibers, resulting in a phenotype that includes varying combinations of somatosensory and dysautonomia symptoms, which can be profoundly disabling and lead to decreased quality of life. Treatment aimed mainly at pain reduction, which may not target the underlying pathophysiology, is frequently ineffective. Another impediment to the effective management of SFN may be the significant between-patient heterogeneity. Accordingly, we launched this study to gain insights into the symptomatic variability of SFN and determine if SFN patients can be sub-grouped based on clinical characteristics. To characterize the phenotype and investigate how patients with SFN differ from those with large fiber involvement, 105 patients with skin-biopsy proven SFN and 45 with mixed fiber neuropathy (MFN) were recruited. Using unsupervised machine learning, SFN patients were clustered based upon symptom concurrence and severity. Demographics, clinical data, symptoms, and skin biopsy- and laboratory findings were compared between the groups. MFN- as compared to SFN patients, were more likely to be male, older, had a lower intraepidermal nerve fiber density at the ankle and more frequent abnormal immunofixation. Beyond these differences, symptom prevalence and intensities were similar in the two cohorts. SFN patients comprised three distinct phenotypic clusters, which differed significantly in symptom severity, co-occurrence, localization, and skin biopsy findings. Only one subgroup, constituting about 20% of the patient population, was characterized by intense neuropathic pain, which was always associated with several other SFN symptoms of similarly high intensities. A pauci-symptomatic cluster comprised patients who experienced few SFN symptoms, generally of low to moderate intensity. The largest cluster was characterized by intense fatigue, myalgias and subjective weakness, but lower intensities of burning pain and paresthesia. This data-driven study introduces a new approach to subgrouping patients with SFN. Considering both neuropathic pain and pernicious symptoms beyond pain, we identified three clusters, which may be related to distinct pathophysiological mechanisms. Although additional validation will be required, our findings represent a step towards stratified treatment approaches and, ultimately, personalized treatment.

Airway "Resistotypes" and Clinical Outcomes in Bronchiectasis.

Rationale: Chronic infection and inflammation shapes the airway microbiome in bronchiectasis. Utilizing whole-genome shotgun metagenomics to analyze the airway resistome provides insight into interplay between microbes, resistance genes, and clinical outcomes. Objectives: To apply whole-genome shotgun metagenomics to the airway microbiome in bronchiectasis to highlight a diverse pool of antimicrobial resistance genes: the "resistome," the clinical significance of which remains unclear. Methods: Individuals with bronchiectasis were prospectively recruited into cross-sectional and longitudinal cohorts (n = 280), including the international multicenter cross-sectional Cohort of Asian and Matched European Bronchiectasis 2 (CAMEB 2) study (n = 251) and two independent cohorts, one describing patients experiencing acute exacerbation and a further cohort of patients undergoing Pseudomonas aeruginosa eradication treatment. Sputum was subjected to metagenomic sequencing, and the bronchiectasis resistome was evaluated in association with clinical outcomes and underlying host microbiomes. Measurements and Main Results: The bronchiectasis resistome features a unique resistance gene profile and increased counts of aminoglycoside, bicyclomycin, phenicol, triclosan, and multidrug resistance genes. Longitudinally, it exhibits within-patient stability over time and during exacerbations despite between-patient heterogeneity. Proportional differences in baseline resistome profiles, including increased macrolide and multidrug resistance genes, associate with shorter intervals to the next exacerbation, whereas distinct resistome archetypes associate with frequent exacerbations, poorer lung function, geographic origin, and the host microbiome. Unsupervised analysis of resistome profiles identified two clinically relevant "resistotypes," RT1 and RT2, the latter characterized by poor clinical outcomes, increased multidrug resistance, and P. aeruginosa. Successful targeted eradication in P. aeruginosa-colonized individuals mediated reversion from RT2 to RT1, a more clinically favorable resistome profile demonstrating reduced resistance gene diversity. Conclusions: The bronchiectasis resistome associates with clinical outcomes, geographic origin, and the underlying host microbiome. Bronchiectasis resistotypes link to clinical disease and are modifiable through targeted antimicrobial therapy.

Individualised profiling of white matter organisation in moderate-to-severe traumatic brain injury patients

Background and purposeApproximately 65% of moderate-to-severe traumatic brain injury (m-sTBI) patients present with poor long-term behavioural outcomes, which can significantly impair activities of daily living. Numerous diffusion-weighted MRI studies have linked these poor outcomes to decreased white matter integrity of several commissural tracts, association fibres and projection fibres in the brain. However, most studies have focused on group-based analyses, which are unable to deal with the substantial between-patient heterogeneity in m-sTBI. As a result, there is increasing interest and need in conducting individualised neuroimaging analyses. Materials and methodsHere, we generated a detailed subject-specific characterisation of microstructural organisation of white matter tracts in 5 chronic patients with m-sTBI (29 – 49y, 2 females), presented as a proof-of-concept. We developed an imaging analysis framework using fixel-based analysis and TractLearn to determine whether the values of fibre density of white matter tracts at the individual patient level deviate from the healthy control group (n = 12, 8F, Mage = 35.7y, age range 25 – 64y). ResultsOur individualised analysis revealed unique white matter profiles, confirming the heterogenous nature of m-sTBI and the need of individualised profiles to properly characterise the extent of injury. Future studies incorporating clinical data, as well as utilising larger reference samples and examining the test–retest reliability of the fixel-wise metrics are warranted. ConclusionsIndividualised profiles may assist clinicians in tracking recovery and planning personalised training programs for chronic m-sTBI patients, which is necessary to achieve optimal behavioural outcomes and improved quality of life.

Integrating biology into clinical trial design.

Critical care medicine revolves around syndromes, such as acute respiratory distress syndrome (ARDS), sepsis and acute kidney injury. Few interventions have shown to be effective in large clinical trials, likely because of between-patient heterogeneity. Translational evidence suggests that more homogeneous biological subgroups can be identified and that differential treatment effects exist. Integrating biological considerations into clinical trial design is therefore an important frontier of critical care research. The pathophysiology of critical care syndromes involves a multiplicity of processes, which emphasizes the difficulty of integrating biology into clinical trial design. Biological assessment can be integrated into clinical trials using predictive enrichment at trial inclusion, time-dependent variation to better understand treatment effects and biological markers as surrogate outcomes. Integrating our knowledge on biological heterogeneity into clinical trial design, which has revolutionized other medical fields, could serve as a solution to implement personalized treatment in critical care syndromes. Changing the trial design by using predictive enrichment, incorporation of the evaluation of time-dependent changes and biological markers as surrogate outcomes may improve the likelihood of detecting a beneficial effect from targeted therapeutic interventions and the opportunity to test multiple lines of treatment per patient.

Precision Care in Cardiac Arrest: ICECAP (PRECICECAP) Study Protocol and Informatics Approach.

Most trials in critical care have been neutral, in part because between-patient heterogeneity means not all patients respond identically to the same treatment. The Precision Care in Cardiac Arrest: Influence of Cooling duration on Efficacy in Cardiac Arrest Patients (PRECICECAP) study will apply machine learning to high-resolution, multimodality data collected from patients resuscitated from out-of-hospital cardiac arrest. We aim to discover novel biomarker signatures to predict the optimal duration of therapeutic hypothermia and 90-day functional outcomes. In parallel, we are developing a freely available software platform for standardized curation of intensive care unit-acquired data for machine learning applications. The Influence of Cooling duration on Efficacy in Cardiac Arrest Patients (ICECAP) study is a response-adaptive, dose-finding trial testing different durations of therapeutic hypothermia. Twelve ICECAP sites will collect data for PRECICECAP from multiple modalities routinely used after out-of-hospital cardiac arrest, including ICECAP case report forms, detailed medication data, cardiopulmonary and electroencephalographic waveforms, and digital imaging and communications in medicine files (DICOMs). We partnered with Moberg Analytics to develop a freely available software platform to allow high-resolution critical care data to be used efficiently and effectively. We will use an autoencoder neural network to create low-dimensional representations of all raw waveforms and derivative features, censored at rewarming to ensure clinical usability to guide optimal duration of hypothermia. We will also consider simple features that are historically considered to be important. Finally, we will create a supervised deep learning neural network algorithm to directly predict 90-day functional outcome from large sets of novel features. PRECICECAP is currently enrolling and will be completed in late 2025. Cardiac arrest is a heterogeneous disease that causes substantial morbidity and mortality. PRECICECAP will advance the overarching goal of titrating personalized neurocritical care on the basis of robust measures of individual need and treatment responsiveness. The software platform we develop will be broadly applicable to hospital-based research after acute illness or injury.

Bayesian spatial models for voxel-wise prostate cancer classification using multi-parametric magnetic resonance imaging data.

Multi-parametric magnetic resonance imaging (mpMRI) has been playing an increasingly important role in the detection of prostate cancer (PCa). Various computer-aided detection algorithms were proposed for automated PCa detection by combining information in multiple mpMRI parameters. However, there are specific features of mpMRI, including between-voxel correlation within each prostate and heterogeneity across patients, that have not been fully explored but could potentially improve PCa detection if leveraged appropriately. This article proposes novel Bayesian approaches for voxel-wise PCa classification that accounts for spatial correlation and between-patient heterogeneity in the mpMRI data. Modeling the spatial correlation is challenging due to the extreme high dimensionality of the data, and we propose three scalable approaches based on Nearest Neighbor Gaussian Process (NNGP), reduced-rank approximation, and a conditional autoregressive (CAR) model that approximates a Gaussian Process with the Matérn covariance, respectively. Our simulation study shows that properly modeling the spatial correlation and between-patient heterogeneity can substantially improve PCa classification. Application to in vivo data illustrates that classification is improved by all three spatial modeling approaches considered, while modeling the between-patient heterogeneity does not further improve our classifiers. Among the proposed models, the NNGP-based model is recommended given its high classification accuracy and computational efficiency.

Making Patient-Specific Treatment Decisions Using Prognostic Variables and Utilities of Clinical Outcomes.

Simple SummaryClinicians often erroneously discount prognostic information as unlikely to change patient management. This is fueled by the mistaken belief that only “predictive” subgroups or biomarkers can modify the differences in clinical benefit between treatment choices. We use the treatment of metastatic clear cell carcinoma as an example to illustrate how clinical decisions can be informed by prognostic variables. Diametrically opposite decisions can be made depending on individual patient prognosis and on the clinical outcome of interest that clinicians choose to focus on. We also demonstrate why such patient-specific treatment decisions inevitably should be guided by each patient’s goals and values, which can be explicitly represented by utility functions.We argue that well-informed patient-specific decision-making may be carried out as three consecutive tasks: (1) estimating key parameters of a statistical model, (2) using prognostic information to convert these parameters into clinically interpretable values, and (3) specifying joint utility functions to quantify risk–benefit trade-offs between clinical outcomes. Using the management of metastatic clear cell renal cell carcinoma as our motivating example, we explain the role of prognostic covariates that characterize between-patient heterogeneity in clinical outcomes. We show that explicitly specifying the joint utility of clinical outcomes provides a coherent basis for patient-specific decision-making.

Costs of medication use among patients with juvenile idiopathic arthritis in the Dutch healthcare system

ABSTRACT Background: This study aims to quantify medication costs in juvenile idiopathic arthritis (JIA), based on subtype. Research design and methods: This study is a single-center, retrospective analysis of prospective data from electronic medical records of JIA patients, aged 0–18 years between 1 April 2011 and 31 March 2019. Patient characteristics (age, gender, subtype) and medication use were extracted. Medication use and costs were reported as: 1) mean total annual costs; 2) between-patient heterogeneity in these costs; 3) duration of medication use; and, 4) costs over the treatment course. Results: The analysis included 691 patients. Mean total medication costs were €2,103/patient/year, including €1,930/patient/year (91.8%) spent on biologicals. Costs varied considerably between subtypes, with polyarticular rheumatoid-factor positive and systemic JIA patients having the highest mean costs (€5,020/patient/year and €4,790/patient/year, respectively). Mean annual medication costs over the patient’s treatment course ranged from <€1,000/year (71.1% of patients) to >€11,000/year (2.5% of patients). Etanercept and adalimumab were the most commonly used biologicals. Cost fluctuations over the treatment course were primarily attributable to biological use. Conclusions: Polyarticular rheumatoid-factor positive and systemic JIA patients had the highest mean total annual medication costs, primarily attributable to biologicals. Costs varied considerably between subtypes, individuals, and over the treatment course.

Abstract A10: Phenotypic and genomic characterization of intratumoral heterogeneity in high-grade serous ovarian cancer

Abstract High-grade serous ovarian cancer (HGSOC) is the most common subtype of ovarian cancer, accounting for more than 70% of all epithelial ovarian cancers. It is characterized by high degrees of genomic instability and heterogeneity, with the majority of patients eventually acquiring resistance to platinum chemotherapy. The diversity in platinum-resistance mechanisms and limited effective predictive biomarkers mean delivering the best treatment options for patient tumor remains challenging. The purpose of this study is to understand the extent of intratumoral heterogeneity (ITH) in advanced-stage HGSOC and how this changes over time at relapse, to describe the molecular mechanisms behind peritoneal dissemination, and to delineate the link between ITH at the molecular and phenotypic levels. Patients undergoing radical upfront debulking for advanced HGS ovarian cancer underwent tumor mapping of their tumor dissemination patterns (n=50). Biopsies were collected from disseminated tumors (range 4-15, median=9), snap frozen, and placed in short-term cultures. Tumor cultures were treated with cisplatin, apoptosis/viability assayed and IC50 for cisplatin determined. DNA was extracted from frozen tumors (5 tumors per patient plus germline) and Illumina Human OmniExpress genotyping performed. Allele-specific copy number (CN) was quantified using ASCAT. Genomic heterogeneity was quantified as the estimated number of CN aberration events distinct between each pair of deposits. Clonal diversity within a patient’s deposits was calculated using the difference between within-patient and between-patient heterogeneity. When relapsed, patients had paired biopsies collected for genomic and phenotypic analysis. Broad heterogeneity was observed in response to platinum treatment in vitro across cases at the phenotypic level (n=50), with higher variances in apoptosis induction observed in patients with platinum-resistant disease. Genomic analysis of copy number data revealed widespread variations in patterns of evolution for different patients’ tumors, including the relationship between primary deposits and relapsed disease. Variations in CCNE1 CN were observed across multiple tumors in the same patients, and overall higher CCNE1 CN associated with poorer patient outcome (p=0.041). Extensive heterogeneity is observed at the phenotypic and genomic levels in HGSOC patients, which correlates with the subsequent development of platinum-resistant disease. CCNE1 copy number variations across multiple intra-abdominal samples within patients indicate that single-site biopsies do not truly represent overall disseminated HGSOC biology and may have implications for overinterpretation of studies relating to outcome and platinum resistance. Citation Format: Paula Cunnea, Ed Curry, Katherine Nixon, Ratri Wulandari, Kerstin Thol, Chun Hei Kwok, Jennifer Ploski, Iain McNeish, Elizabeth Christie, David Bowtell, Christina Fotopoulou. Phenotypic and genomic characterization of intratumoral heterogeneity in high-grade serous ovarian cancer [abstract]. In: Proceedings of the AACR Special Conference on Advances in Ovarian Cancer Research; 2019 Sep 13-16, 2019; Atlanta, GA. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(13_Suppl):Abstract nr A10.

Renal Diuretic Delivery is Highly Variable in Hospitalized Decompensated Heart Failure Patients and Associated with Diuretic Resistance

Introduction It is often assumed that diuretic delivery to the site of action is adequate if the route of administration is intravenous (IV). Although IV administration circumvents gastrointestinal bioavailability, the diuretic must be delivered and secreted into the tubular lumen induce diuresis. In ADHF, perturbations such as reduced renal perfusion and renal dysfunction may impact delivery to the tubular site of action. Hypothesis There will be substantial between-patient heterogeneity in renal diuretic delivery across hospitalized AHDF patients receiving IV diuretics. Methods ADHF patients receiving IV diuretics at Yale New Haven Hospital underwent intensively supervised timed urine collection. Renal diuretic delivery was defined as the percentage of loop diuretic recovered unchanged in the urine in the 6h after IV diuretic administration. Results 105 patients were studied after a median 80 mg (IQR 40-160mg) IV furosemide equivalents, with a median of 30 mg (12-59 mg) of unchanged loop diuretic recovered in the urine. Thus, 37% (21-54%) reached the site of action, substantially less than that reported in normal subjects (∼50-70%, p 0.1). A modest correlation between better renal filtration (eGFR r=0.23, p=0.02; creatinine clearance r=0.33, p=0.01; Cystatin C r=-0.30, p=0.003) and higher renal diuretic delivery was noted. Renal urea handling had the strongest correlation with renal diuretic delivery (fractional excretion of urea r=0.49, p= Conclusion Large variations IV loop diuretic delivered to the tubular site of action exists between hospitalized ADHF patients, and these differences relate to the ultimate diuretic response. The strongest associated factor is renal urea handling, which captures elements of renal perfusion, filtration, and solute handling. Given that patients with poor diuretic delivery may preferentially respond to loop diuretic titration, additional research is warranted to better identify patients with poor diuretic delivery.

Bayesian hierarchical vector autoregressive models for patient-level predictive modeling.

Predicting health outcomes from longitudinal health histories is of central importance to healthcare. Observational healthcare databases such as patient diary databases provide a rich resource for patient-level predictive modeling. In this paper, we propose a Bayesian hierarchical vector autoregressive (VAR) model to predict medical and psychological conditions using multivariate time series data. Compared to the existing patient-specific predictive VAR models, our model demonstrated higher accuracy in predicting future observations in terms of both point and interval estimates due to the pooling effect of the hierarchical model specification. In addition, by adopting an elastic-net prior, our model offers greater interpretability about the associations between variables of interest on both the population level and the patient level, as well as between-patient heterogeneity. We apply the model to two examples: 1) predicting substance use craving, negative affect and tobacco use among college students, and 2) predicting functional somatic symptoms and psychological discomforts.

Group sequential designs with robust semiparametric recurrent event models.

Robust semiparametric models for recurrent events have received increasing attention in the analysis of clinical trials in a variety of diseases including chronic heart failure. In comparison to parametric recurrent event models, robust semiparametric models are more flexible in that neither the baseline event rate nor the process inducing between-patient heterogeneity needs to be specified in terms of a specific parametric statistical model. However, implementing group sequential designs in the robust semiparametric model is complicated by the fact that the sequence of Wald statistics does not follow asymptotically the canonical joint distribution. In this manuscript, we propose two types of group sequential procedures for a robust semiparametric analysis of recurrent events. The first group sequential procedure is based on the asymptotic covariance of the sequence of Wald statistics and it guarantees asymptotic control of the type I error rate. The second procedure is based on the canonical joint distribution and does not guarantee asymptotic type I error rate control but is easy to implement and corresponds to the well-known standard approach for group sequential designs. Moreover, we describe how to determine the maximum information when planning a clinical trial with a group sequential design and a robust semiparametric analysis of recurrent events. We contrast the operating characteristics of the proposed group sequential procedures in a simulation study motivated by the ongoing phase 3 PARAGON-HF trial (ClinicalTrials.gov identifier: NCT01920711) in more than 4600 patients with chronic heart failure and a preserved ejection fraction. We found that both group sequential procedures have similar operating characteristics and that for some practically relevant scenarios, the group sequential procedure based on the canonical joint distribution has advantages with respect to the control of the type I error rate. The proposed method for calculating the maximum information results in appropriately powered trials for both procedures.

Phenotyping Cardiac Arrest: Bench and Bedside Characterization of Brain and Heart Injury Based on Etiology.

Cardiac arrest etiology may be an important source of between-patient heterogeneity, but the impact of etiology on organ injury is unknown. We tested the hypothesis that asphyxial cardiac arrest results in greater neurologic injury than cardiac etiology cardiac arrest (ventricular fibrillation cardiac arrest), whereas ventricular fibrillation cardiac arrest results in greater cardiovascular dysfunction after return of spontaneous circulation. Prospective observational human and randomized animal study. University laboratory and ICUs. Five-hundred forty-three cardiac arrest patients admitted to ICU. Seventy-five male Sprague-Dawley rats. We examined neurologic and cardiovascular injury in Isoflurane-anesthetized rat cardiac arrest models matched by ischemic time. Hemodynamic and neurologic outcomes were assessed after 5 minutes no flow asphyxial cardiac arrest or ventricular fibrillation cardiac arrest. Comparison was made to injury patterns observed after human asphyxial cardiac arrest or ventricular fibrillation cardiac arrest. In rats, cardiac output (20 ± 10 vs 45 ± 9 mL/min) and pH were lower and lactate higher (9.5 ± 1.0 vs 6.4 ± 1.3 mmol/L) after return of spontaneous circulation from ventricular fibrillation cardiac arrest versus asphyxial cardiac arrest (all p < 0.01). Asphyxial cardiac arrest resulted in greater early neurologic deficits, 7-day neuronal loss, and reduced freezing time (memory) after conditioned fear (all p < 0.05). Brain antioxidant reserves were more depleted following asphyxial cardiac arrest. In adjusted analyses, human ventricular fibrillation cardiac arrest was associated with greater cardiovascular injury based on peak troponin (7.8 ng/mL [0.8-57 ng/mL] vs 0.3 ng/mL [0.0-1.5 ng/mL]) and ejection fraction by echocardiography (20% vs 55%; all p < 0.0001), whereas asphyxial cardiac arrest was associated with worse early neurologic injury and poor functional outcome at hospital discharge (n = 46 [18%] vs 102 [44%]; p < 0.0001). Most ventricular fibrillation cardiac arrest deaths (54%) were the result of cardiovascular instability, whereas most asphyxial cardiac arrest deaths (75%) resulted from neurologic injury (p < 0.0001). In transcending rat and human studies, we find a consistent phenotype of heart and brain injury after cardiac arrest based on etiology: ventricular fibrillation cardiac arrest produces worse cardiovascular dysfunction, whereas asphyxial cardiac arrest produces worsened neurologic injury associated with greater oxidative stress.

Precision oncology for acute myeloid leukemia using a knowledge bank approach.

Underpinning the vision of precision medicine is the concept that causative mutations in a patient's cancer drive its biology and, by extension, its clinical features and treatment response. However, considerable between-patient heterogeneity in driver mutations complicates evidence-based personalization of cancer care. Here, by reanalyzing data from 1,540 patients with acute myeloid leukemia (AML), we explore how large knowledge banks of matched genomic-clinical data can support clinical decision-making. Inclusive, multistage statistical models accurately predicted likelihoods of remission, relapse and mortality, which were validated using data from independent patients in The Cancer Genome Atlas. Comparison of long-term survival probabilities under different treatments enables therapeutic decision support, which is available in exploratory form online. Personally tailored management decisions could reduce the number of hematopoietic cell transplants in patients with AML by 20-25% while maintaining overall survival rates. Power calculations show that databases require information from thousands of patients for accurate decision support. Knowledge banks facilitate personally tailored therapeutic decisions but require sustainable updating, inclusive cohorts and large sample sizes.

Abstract LB-340: A novel method for estimating somatic mutations cellular prevalence in cancer using whole genome haplotype phasing

Abstract The process of malignant transformation results in the acquisition of complex chromosomal copy number alterations and somatic mutations. Moreover, cancer exhibits extensive intra-tumor heterogeneity with multiple sub-populations of tumor cells containing both common and private genomic alterations. Within- and between-patient tumor heterogeneity is now seen as one of the major obstacles for precision medicine and the development of effective treatment strategies. Revealing the proportion of cells affected by particular somatic changes is emerging as a key factor for designing targeted therapies, characterizing tumor evolution and understanding chemotherapy-resistance mechanisms. Determining the proportion of cells harboring a particular mutation (the so called cellular prevalence) is a difficult task as it is confounded by several variables. Normal cell contamination, local copy number profile and the temporal relationship between the mutation and the copy number alteration represent some of the key variables. In order to more effectively quantify the cellular prevalence of somatic mutations we developed a novel method, which uses a combination of side, and haplotype phase information obtained from long fragment read sequencing to more accurately compute mutational prevalence. The method utilizes three sources of information: the phasing information, the copy number variation and the allele counts for prevalence estimation using a simple equation in both a linear and a non-linear form. We present theoretical examples of known genomics alterations and show using simulation that our method was able to accurately estimate the cellular prevalence compared to standard methods. We also apply our method to a real data set obtained from dense long fragment read sequencing of a single ovarian cancer and demonstrate the highly accurate estimation of cellular prevalence. Finally we provide an implementation of the method in the statistical software. Citation Format: Donatien Chedom-Fotso, Ahmed Ahmed, Christopher Yau. A novel method for estimating somatic mutations cellular prevalence in cancer using whole genome haplotype phasing. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr LB-340.

Polymyalgia Rheumatica (PMR) Special Interest Group at OMERACT 11: Outcomes of Importance for Patients with PMR

We worked toward developing a core outcome set for clinical research studies in polymyalgia rheumatica (PMR) by conducting (1) patient consultations using modified nominal group technique; (2) a systematic literature review of outcome measures in PMR; (3) a pilot observational study of patients presenting with untreated PMR, and further discussion with patient research partners; and (4) a qualitative focus group study of patients with PMR on the meaning of stiffness, using thematic analysis. (1) Consultations included 104 patients at 4 centers. Symptoms of PMR included pain, stiffness, fatigue, and sleep disturbance. Function, anxiety, and depression were also often mentioned. Participants expressed concerns about diagnostic delay, adverse effects of glucocorticoids, and fear of relapse. (2) In the systematic review, outcome measures previously used for PMR include pain visual analog scores (VAS), morning stiffness, blood markers, function, and quality of life; standardized effect sizes posttreatment were large. (3) Findings from the observational study indicated that asking about symptom severity at 7 AM, or "on waking," appeared more relevant to disease activity than asking about symptom severity "now" (which depended on the time of assessment). (4) Preliminary results were presented from the focus group qualitative study, encompassing broad themes of stiffness, pain, and the effect of PMR on patients' lives. It was concluded that further validation work is required before a core outcome set in PMR can be recommended. Nevertheless, the large standardized effect sizes suggest that pain VAS is likely to be satisfactory as a primary outcome measure for assessing response to initial therapy of PMR. Dissection of between-patient heterogeneity in the subsequent treatment course may require attention to comorbidity as a potential confounding factor.

Robustness of methods for blinded sample size re‐estimation with overdispersed count data

Counts of events are increasingly common as primary endpoints in randomized clinical trials. With between-patient heterogeneity leading to variances in excess of the mean (referred to as overdispersion), statistical models reflecting this heterogeneity by mixtures of Poisson distributions are frequently employed. Sample size calculation in the planning of such trials requires knowledge on the nuisance parameters, that is, the control (or overall) event rate and the overdispersion parameter. Usually, there is only little prior knowledge regarding these parameters in the design phase resulting in considerable uncertainty regarding the sample size. In this situation internal pilot studies have been found very useful and very recently several blinded procedures for sample size re-estimation have been proposed for overdispersed count data, one of which is based on an EM-algorithm. In this paper we investigate the EM-algorithm based procedure with respect to aspects of their implementation by studying the algorithm's dependence on the choice of convergence criterion and find that the procedure is sensitive to the choice of the stopping criterion in scenarios relevant to clinical practice. We also compare the EM-based procedure to other competing procedures regarding their operating characteristics such as sample size distribution and power. Furthermore, the robustness of these procedures to deviations from the model assumptions is explored. We find that some of the procedures are robust to at least moderate deviations. The results are illustrated using data from the US National Heart, Lung and Blood Institute sponsored Asymptomatic Cardiac Ischemia Pilot study.

Between-Patient and Within-Patient (Site-to-Site) Variability in Estrogen Receptor Binding, Measured In Vivo by 18F-Fluoroestradiol PET

Heterogeneity of estrogen receptor (ER) expression may be an important predictor of breast cancer therapeutic response. (18)F-fluoroestradiol PET produces in vivo quantitative measurements of regional estrogen binding in breast cancer tumors. We describe within-patient (site-to-site) and between-patient heterogeneity of lesions in patients scheduled to receive endocrine therapy. In 91 patients with a prior ER-positive biopsy, 505 lesions were analyzed for both (18)F-fluoroestradiol and (18)F-FDG uptake and the (18)F-fluoroestradiol/(18)F-FDG uptake ratio. Standardized uptake values (SUVs) were recorded for up to 16 lesions per patient, of 1.5 cm or more and visible on (18)F-FDG PET or conventional staging. Linear mixed-effects regression models examined associations between PET parameters and patient or lesion characteristics and estimated variance components. A reader study of SUV measurements for 9 scans further examined sources of within-patient variability. Average (18)F-fluoroestradiol uptake and (18)F-fluoroestradiol/(18)F-FDG ratio varied greatly across these patients, despite a history of ER-positive disease: about 37% had low or absent (18)F-fluoroestradiol uptake even with marked (18)F-FDG uptake. (18)F-fluoroestradiol SUV and (18)F-fluoroestradiol/(18)F-FDG ratio measurements within patients with multiple lesions were clustered around the patient's average value in most cases. Summarizing these findings, the intraclass correlation coefficient (proportion of total variation that is between-patient) was 0.60 (95% confidence interval, 0.50-0.69) for (18)F-fluoroestradiol SUV and 0.65 (95% confidence interval, 0.56-0.73) for the (18)F-fluoroestradiol/(18)F-FDG ratio. Some within-patient variation in PET measures (22%-44%) was attributable to interobserver variability as measured by the reader study. A subset of patients had mixed uptake, with widely disparate (18)F-fluoroestradiol SUV or (18)F-fluoroestradiol/(18)F-FDG ratio for lesions in the same scan. (18)F-fluoroestradiol uptake and the (18)F-fluoroestradiol/(18)F-FDG ratio varied greatly between patients but were usually consistent across lesions in the same scan. The average (18)F-fluoroestradiol SUV and (18)F-fluoroestradiol/(18)F-FDG ratio for a limited sample of lesions appear to provide a reasonable summary of synchronous ER expression for most patients. However, imaging the entire disease burden remains important to identify the subset of patients with mixed uptake, who may be at a critical point in their disease evolution.

Blinded sample size reestimation with count data: Methods and applications in multiple sclerosis

Sample size estimation in clinical trials depends critically on nuisance parameters, such as variances or overall event rates, which have to be guessed or estimated from previous studies in the planning phase of a trial. Blinded sample size reestimation estimates these nuisance parameters based on blinded data from the ongoing trial, and allows to adjust the sample size based on the acquired information. In the present paper, this methodology is developed for clinical trials with count data as the primary endpoint. In multiple sclerosis such endpoints are commonly used in phase 2 trials (lesion counts in magnetic resonance imaging (MRI)) and phase 3 trials (relapse counts). Sample size adjustment formulas are presented for both Poisson-distributed data and for overdispersed Poisson-distributed data. The latter arise from sometimes considerable between-patient heterogeneity, which can be observed in particular in MRI lesion counts. The operation characteristics of the procedure are evaluated by simulations and recommendations on how to choose the size of the internal pilot study are given. The results suggest that blinded sample size reestimation for count data maintains the required power without an increase in the type I error.

Statistical methods for the analysis of relapse data in MS clinical trials

Patients with multiple sclerosis (MS) often experience unpredictable recurrent relapses with periods of remission. The modeling of MS relapse data is complicated because both within-subject serial dependence between relapses and between-patient heterogeneity may exist. We compare six statistical methods for assessing the treatment efficacy in reducing the frequency of relapses in MS clinical trials. All methods can be implemented in SAS ®, and are grouped into two classes, one based on Poisson-type regressions for count data and the other on Cox proportional hazards models for time to relapse. We apply these models to the data of a Tysabri ® (Natalizumab) MS trial and interpret the differences in results based on the underlying assumptions. Negative binomial regression is recommended for evaluating the overall treatment effect because of its simplicity and efficiency.