Early circulating tumor DNA dynamics as a pan-tumor biomarker for long-term clinical outcome in patients treated with durvalumab and tremelimumab.

There is an urgent need to identify biomarkers of early response that can accurately predict the benefit of immune checkpoint inhibitors (ICI). Patients receiving durvalumab/tremelimumab had tumor samples sequenced before treatment (baseline) to identify variants for the design of a personalized circulating tumor (ctDNA) assay. ctDNA was assessed at baseline and at 4 and/or 8 weeks into treatment. Correlations between ctDNA changes to radiographic response and overall survival (OS) were made to assess potential clinical benefit. 35/40 patients (87.5%) had personalized ctDNA assays designed, and 29/35 (82.9%) had plasma available for baseline analysis, representing 16 unique solid tumor histologies. As early as 4 weeks after treatment, decline in ctDNA from baseline predicted improved OS (P = 0.0144; HR = 9.98) and ctDNA changes on treatment-supported and refined radiographic response calls. ctDNA clearance at any time through week 8 identified complete responders by a median lead time of 11.5 months ahead of radiographic imaging. ctDNA response monitoring is emerging as a dynamic, personalized biomarker method that may predict survival outcomes in patients with diverse solid tumor histologies, complementing and sometimes preceding standard-of-care imaging assessments.

Adaptive Fisher method detects dense and sparse signals in association analysis of SNV sets.

BACKGROUND: With the development of next generation sequencing (NGS) technology and genotype imputation methods, statistical methods have been proposed to test a set of genomic variants together to detect if any of them is associated with the phenotype or disease. In practice, within the set, there is an unknown proportion of variants truly causal or associated with the disease. There is a demand for statistical methods with high power in both dense and sparse scenarios, where the proportion of causal or associated variants is large or small respectively. RESULTS: We propose a new association test - weighted Adaptive Fisher (wAF) that can adapt to both dense and sparse scenarios by adding weights to the Adaptive Fisher (AF) method we developed before. Using simulation, we show that wAF enjoys comparable or better power to popular methods such as sequence kernel association tests (SKAT and SKAT-O) and adaptive SPU (aSPU) test. We apply wAF to a publicly available schizophrenia dataset, and successfully detect thirteen genes. Among them, three genes are supported by existing literature; six are plausible as they either relate to other neurological diseases or have relevant biological functions. CONCLUSIONS: The proposed wAF method is a powerful disease-variants association test in both dense and sparse scenarios. Both simulation studies and real data analysis indicate the potential of wAF for new biological findings.

Invariance in the recurrence of large returns and the validation of models of price dynamics.

Starting from a robust, nonparametric definition of large returns ("excursions"), we study the statistics of their occurrences, focusing on the recurrence process. The empirical waiting-time distribution between excursions is remarkably invariant to year, stock, and scale (return interval). This invariance is related to self-similarity of the marginal distributions of returns, but the excursion waiting-time distribution is a function of the entire return process and not just its univariate probabilities. Generalized autoregressive conditional heteroskedasticity (GARCH) models, market-time transformations based on volume or trades, and generalized (Lévy) random-walk models all fail to fit the statistical structure of excursions.