Biomarkers and Strategy to Detect Preinvasive and Early Pancreatic Cancer: State of the Field and the Impact of the EDRN

Flexible variable selection in the presence of missing data

In many applications, it is of interest to identify a parsimonious set of features, or panel, from multiple candidates that achieves a desired level of performance in predicting a response. This task is often complicated in practice by missing data arising from the sampling design or other random mechanisms. Most recent work on variable selection in missing data contexts relies in some part on a finite-dimensional statistical model, e.g., a generalized or penalized linear model. In cases where this model is misspecified, the selected variables may not all be truly scientifically relevant and can result in panels with suboptimal classification performance. To address this limitation, we propose a nonparametric variable selection algorithm combined with multiple imputation to develop flexible panels in the presence of missing-at-random data. We outline strategies based on the proposed algorithm that achieve control of commonly used error rates. Through simulations, we show that our proposal has good operating characteristics and results in panels with higher classification and variable selection performance compared to several existing penalized regression approaches in cases where a generalized linear model is misspecified. Finally, we use the proposed method to develop biomarker panels for separating pancreatic cysts with differing malignancy potential in a setting where complicated missingness in the biomarkers arose due to limited specimen volumes.

Pancreatic cancer: Advances and challenges

Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest cancers. Significant efforts have largely defined major genetic factors driving PDAC pathogenesis and progression. Pancreatic tumors are characterized by a complex microenvironment that orchestrates metabolic alterations and supports a milieu of interactions among various cell types within this niche. In this review, we highlight the foundational studies that have driven our understanding of these processes. We further discuss the recent technological advances that continue to expand our understanding of PDAC complexity. We posit that the clinical translation of these research endeavors will enhance the currently dismal survival rate of this recalcitrant disease.

A Flexible Method for Diagnostic Accuracy with Biomarker Measurement Error

Diagnostic biomarkers are often measured with errors due to imperfect lab conditions or analytic variability of the assay. The ability of a diagnostic biomarker to discriminate between cases and controls is often measured by the area under the receiver operating characteristic curve (AUC), sensitivity, specificity, among others. Ignoring measurement error can cause biased estimation of a diagnostic accuracy measure, which results in misleading interpretation of the efficacy of a diagnostic biomarker. Existing assays available are either research grade or clinical grade. Research assays are cost effective, often multiplex, but they may be associated with moderate measurement errors leading to poorer diagnostic performance. In comparison, clinical assays may provide better diagnostic ability, but with higher cost since they are usually developed by industry. Correction for attenuation methods are often valid when biomarkers are from a normal distribution, but may be biased with skewed biomarkers. In this paper, we develop a flexible method based on skew-normal biomarker distributions to correct for bias in estimating diagnostic performance measures including AUC, sensitivity, and specificity. Finite sample performance of the proposed method is examined via extensive simulation studies. The methods are applied to a pancreatic cancer biomarker study.

A pilot study to troubleshoot quality control metrics when assessing circulating miRNA expression data reproducibility across study sites

BACKGROUND: Given the growing interest in using microRNAs (miRNAs) as biomarkers of early disease, establishment of robust protocols and platforms for miRNA quantification in biological fluids is critical. OBJECTIVE: The goal of this multi-center pilot study was to evaluate the reproducibility of NanoString nCounter™ technology when analyzing the abundance of miRNAs in plasma and cystic fluid from patients with pancreatic lesions. METHODS: Using sample triplicates analyzed across three study sites, we assessed potential sources of variability (RNA isolation, sample processing/ligation, hybridization, and lot-to-lot variability) that may contribute to suboptimal reproducibility of miRNA abundance when using nCounter™, and evaluated expression of positive and negative controls, housekeeping genes, spike-in genes, and miRNAs. RESULTS: Positive controls showed a high correlation across samples from each site (median correlation coefficient, r> 0.9). Most negative control probes had expression levels below background. Housekeeping and spike-in genes each showed a similar distribution of expression and comparable pairwise correlation coefficients of replicate samples across sites. A total of 804 miRNAs showed a similar distribution of pairwise correlation coefficients between replicate samples (p= 0.93). After normalization and selecting miRNAs with expression levels above zero in 80% of samples, 55 miRNAs were identified; heatmap and principal component analysis revealed similar expression patterns and clustering in replicate samples. CONCLUSIONS: Findings from this pilot investigation suggest the nCounter platform can yield reproducible results across study sites. This study underscores the importance of implementing quality control procedures when designing multi-center evaluations of miRNA abundance.

Pancreatic Cancer Surveillance and Novel Strategies for Screening

Individuals with a genetic susceptibility to pancreatic ductal adenocarcinoma (PDAC) may benefit from surveillance to increase the likelihood of early detection. Currently, candidates for surveillance are identified based on genetic test results and family history of PDAC, and surveillance is accomplished through imaging of the pancreas (endoscopic ultrasound or MRI). Novel methods that incorporate personalized risk, biomarkers, and radiomics are being investigated in an attempt to improve identification of at-risk individuals and to increase detection of precursor and early-stage lesions.

Biomarker Risk Score Algorithm and Preoperative Stratification of Patients with Pancreatic Cystic Lesions

BACKGROUND

Pancreatic cysts are incidentally detected in up to 13% of patients undergoing radiographic imaging. Of the most frequently encountered types, mucin-producing (mucinous) pancreatic cystic lesions may develop into pancreatic cancer, while nonmucinous ones have little or no malignant potential. Accurate preoperative diagnosis is critical for optimal management, but has been difficult to achieve, resulting in unnecessary major surgery. Here, we aim to develop an algorithm based on biomarker risk scores to improve risk stratification.

STUDY DESIGN

Patients undergoing surgery and/or surveillance for a pancreatic cystic lesion, with diagnostic imaging and banked pancreatic cyst fluid, were enrolled in the study after informed consent (n = 163 surgical, 67 surveillance). Cyst fluid biomarkers with high specificity for distinguishing nonmucinous from mucinous pancreatic cysts (vascular endothelial growth factor [VEGF], glucose, carcinoembryonic antigen [CEA], amylase, cytology, and DNA mutation) were selected. Biomarker risk scores were used to design an algorithm to predict preoperative diagnosis. Performance was tested using surgical (retrospective) and surveillance (prospective) cohorts.

RESULTS

In the surgical cohort, the biomarker algorithm outperformed the preoperative clinical diagnosis in correctly predicting the final pathologic diagnosis (91% vs 73%; p < 0.000001). Specifically, nonmucinous serous cystic neoplasms (SCN) and mucinous cystic neoplasms (MCN) were correctly classified more frequently by the algorithm than clinical diagnosis (96% vs 30%; p < 0.000008 and 92% vs 69%; p = 0.04, respectively). In the surveillance cohort, the algorithm predicted a preoperative diagnosis with high confidence based on a high biomarker score and/or consistency with imaging from ≥1 follow-up visits.

CONCLUSIONS

A biomarker risk score-based algorithm was able to correctly classify pancreatic cysts preoperatively. Importantly, this tool may improve initial and dynamic risk stratification, reducing overdiagnosis and underdiagnosis.

Artificial Intelligence and Early Detection of Pancreatic Cancer: 2020 Summative Review

ABSTRACT

Despite considerable research efforts, pancreatic cancer is associated with a dire prognosis and a 5-year survival rate of only 10%. Early symptoms of the disease are mostly nonspecific. The premise of improved survival through early detection is that more individuals will benefit from potentially curative treatment. Artificial intelligence (AI) methodology has emerged as a successful tool for risk stratification and identification in general health care. In response to the maturity of AI, Kenner Family Research Fund conducted the 2020 AI and Early Detection of Pancreatic Cancer Virtual Summit (www.pdac-virtualsummit.org) in conjunction with the American Pancreatic Association, with a focus on the potential of AI to advance early detection efforts in this disease. This comprehensive presummit article was prepared based on information provided by each of the interdisciplinary participants on one of the 5 following topics: Progress, Problems, and Prospects for Early Detection; AI and Machine Learning; AI and Pancreatic Cancer-Current Efforts; Collaborative Opportunities; and Moving Forward-Reflections from Government, Industry, and Advocacy. The outcome from the robust Summit conversations, to be presented in a future white paper, indicate that significant progress must be the result of strategic collaboration among investigators and institutions from multidisciplinary backgrounds, supported by committed funders.