Early Detection of Pancreatic Cancer: Applying Artificial Intelligence to Electronic Health Records.

ABSTRACT: The potential of artificial intelligence (AI) applied to clinical data from electronic health records (EHRs) to improve early detection for pancreatic and other cancers remains underexplored. The Kenner Family Research Fund, in collaboration with the Cancer Biomarker Research Group at the National Cancer Institute, organized the workshop entitled: "Early Detection of Pancreatic Cancer: Opportunities and Challenges in Utilizing Electronic Health Records (EHR)" in March 2021. The workshop included a select group of panelists with expertise in pancreatic cancer, EHR data mining, and AI-based modeling. This review article reflects the findings from the workshop and assesses the feasibility of AI-based data extraction and modeling applied to EHRs. It highlights the increasing role of data sharing networks and common data models in improving the secondary use of EHR data. Current efforts using EHR data for AI-based modeling to enhance early detection of pancreatic cancer show promise. Specific challenges (biology, limited data, standards, compatibility, legal, quality, AI chasm, incentives) are identified, with mitigation strategies summarized and next steps identified.

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.

Is Cancer Solvable? Towards Efficient and Ethical Biomedical Science.

Global Cumulative Treatment Analysis (GCTA) is a novel clinical research model combining expert knowledge, and treatment coordination based upon global information-gain, to treat every patient optimally while efficiently searching the vast space that is the realm of cancer research.

Frequency of Appropriate Use of Pancreatic Enzyme Replacement Therapy and Symptomatic Response in Pancreatic Cancer Patients.

OBJECTIVES: Pancreatic cancer (PC) and its treatments can result in pancreatic exocrine insufficiency that requires pancreatic enzyme replacement therapy (PERT). Appropriate PERT usage is during meals and snacks. The aim was to determine the frequency of appropriate use of PERT and its impact on symptom alleviation in PC through a patient-reported outcomes online platform. METHODS: Users in the Pancreatic Cancer Action Network's Patient Registry were prompted to answer a standalone questionnaire about their experience with PERT. RESULTS: Two hundred sixty-two users completed the PERT questionnaire (January 2016-January 2018). Patients who reported taking PERT with meals had higher alleviation of symptoms compared with those taking PERT prior to or after meals. Specifically, "feeling of indigestion," "light-colored or orange stools," and "visible food particles in stool" were significantly decreased. Patients taking PERT with meals reported weight gain and less weight loss. CONCLUSIONS: Of the 89% of PC patients prescribed PERT, 65% were prescribed PERT appropriately with all meals and snacks. Overall compliance with PERT administration guidelines was low (50% [105/208]). Improvement in symptoms significantly correlated with appropriate use of PERT. Increase in PC patient and provider education about appropriate PERT usage and administration is warranted.

Low-Dose Continuous 5-Fluorouracil Combined with Leucovorin, nab-Paclitaxel, Oxaliplatin, and Bevacizumab for Patients with Advanced Pancreatic Cancer: A Retrospective Analysis.

BACKGROUND: Continuous-infusion 5-fluorouracil (5FU) and calcium leucovorin plus nab-paclitaxel and oxaliplatin have been shown to be active in patients with pancreatic cancer. As a protracted low-dose infusion, 5FU is antiangiogenic, and has synergy with bevacizumab. As shown in the treatment of breast cancer, bevacizumab and nab-paclitaxel are also synergetic. OBJECTIVE: In this paper we retrospectively analyze the survival of 65 patients with advanced pancreatic cancer who were treated with low-dose continuous (metronomic) chemotherapy given in conjunction with conventional anti-VEGF therapy. PATIENTS AND METHODS: Since July of 2008, we have treated 65 patients with 5FU (180 mg/m2/day × 14 days) via an ambulatory pump. Calcium leucovorin (20 mg/m2 IV), nab-paclitaxel (60 mg/m2) IV as a 30-min infusion, and oxaliplatin (50 mg/m2) IV as a 60-min infusion were given on days 1, 8, and 15. Bevacizumab (5 mg/kg) IV over 30 min was administered on days 1 and 15. Cycles were repeated every 28-35 days. There were 42 women and 23 men, and the median age was 59 years. Forty-six patients had stage IV disease. RESULTS: The median survival was 19 months, with 82% of patients surviving 12 months or longer. The overall response rate was 49%. There were 28 patients who had received prior treatment, 15 of whom responded to therapy. Fifty-two patients had elevated CA 19-9 prior to treatment. Of these, 21 patients had 90% or greater reduction in CA 19-9 levels. This cohort had an objective response rate of 71% and a median survival of 27 months. Thirty patients stopped treatment due to disease progression, and an additional 22 stopped because of toxicity. One patient died while on therapy. CONCLUSIONS: This non-gemcitabine-based regimen resulted in higher response rates and better survival than what is commonly observed with therapy given at conventional dosing schedules. Low-dose continuous (metronomic therapy) cytotoxic chemotherapy combined with antiangiogenic therapy is safe and effective.

Model to Determine Risk of Pancreatic Cancer in Patients With New-Onset Diabetes.

BACKGROUND & AIMS: Of patients with new-onset diabetes (NOD; based on glycemic status) older than 50 years, approximately 1% are diagnosed with pancreatic cancer (PC) within 3 years. We aimed to develop and validate a model to determine risk of PC in patients with NOD. METHODS: We retrospectively collected data from 4 independent and nonoverlapping cohorts of patients (N = 1,561) with NOD (based on glycemic status; data collected at date of diagnosis and 12 months previously) in the Rochester Epidemiology Project from January 1, 2000 through December 31, 2015 to create our model. The model weighed scores for 3 factors identified in the discovery cohort to be most strongly associated with PC (64 patients with PC and 192 with type 2 diabetes): change in weight, change in blood glucose, and age at onset of diabetes. We called our model Enriching New-Onset Diabetes for Pancreatic Cancer (ENDPAC). We validated the locked-down model and cutoff score in an independent population-based cohort of 1,096 patients with diabetes; of these, 9 patients (82%) had PC within 3 years of meeting the criteria for NOD. RESULTS: In the discovery cohort, the END-PAC model identified patients who developed PC within 3 years of diabetes onset (area under receiver operating characteristic curve 0.87); a score of at least 3 identified patients who developed PC with 80% sensitivity and specificity. In the validation cohort, a score of at least 3 identified 7 of 9 patients with PC (78%) with 85% specificity; the prevalence of PC in patients with a score of at least 3 (3.6%) was 4.4-fold greater than in patients with NOD. A high END-PAC score in patients who did not have PC (false positives) was often due to such factors as recent steroid use or different malignancy. An ENDPAC score no higher than 0 (in 49% of patients) meant that patients had an extremely low risk for PC. An END-PAC score of at least 3 identified 75% of patients in the discovery cohort more than 6 months before a diagnosis of PC. CONCLUSIONS: Based on change in weight, change in blood glucose, and age at onset of diabetes, we developed and validated a model to determine risk of PC in patients with NOD based on glycemic status (END-PAC model). An independent prospective study is needed to further validate this model, which could contribute to early detection of PC.

A pilot study evaluating concordance between blood-based and patient-matched tumor molecular testing within pancreatic cancer patients participating in the Know Your Tumor (KYT) initiative.

Recent improvements in next-generation sequencing (NGS) technology have enabled detection of biomarkers in cell-free DNA in blood and may ultimately replace invasive tissue biopsies. However, a better understanding of the performance of blood-based NGS assays is needed prior to routine clinical use. As part of an IRB-approved molecular profiling registry trial of pancreatic ductal adenocarcinoma (PDA) patients, we facilitated blood-based NGS testing of 34 patients from multiple community-based and high-volume academic oncology practices. 23 of these patients also underwent traditional tumor tissue-based NGS testing. cfDNA was not detected in 9/34 (26%) patients. Overall concordance between blood and tumor tissue NGS assays was low, with only 25% sensitivity of blood-based NGS for tumor tissue NGS. Mutations in KRAS, the major PDA oncogene, were only detected in 10/34 (29%) blood samples, compared to 20/23 (87%) tumor tissue biopsies. The presence of mutations in circulating DNA was associated with reduced overall survival (54% in mutation-positive versus 90% in mutation-negative). Our results suggest that in the setting of previously treated, advanced PDA, liquid biopsies are not yet an adequate substitute for tissue biopsies. Further refinement in defining the optimal patient population and timing of blood sampling may improve the value of a blood-based test.

High HER2 protein levels correlate with increased survival in breast cancer patients treated with anti-HER2 therapy.

INTRODUCTION: Current methods to determine HER2 (human epidermal growth factor receptor 2) status are affected by reproducibility issues and do not reliably predict benefit from anti-HER2 therapy. Quantitative measurement of HER2 may more accurately identify breast cancer (BC) patients who will respond to anti-HER2 treatments. METHODS: Using selected reaction monitoring mass spectrometry (SRM-MS), we quantified HER2 protein levels in formalin-fixed, paraffin-embedded (FFPE) tissue samples that had been classified as HER2 0, 1+, 2+ or 3+ by immunohistochemistry (IHC). Receiver operator curve (ROC) analysis was conducted to obtain optimal HER2 protein expression thresholds predictive of HER2 status (by standard IHC or in situ hybridization [ISH]) and of survival benefit after anti-HER2 therapy. RESULTS: Absolute HER2 amol/µg levels were significantly correlated with both HER2 IHC and amplification status by ISH (p < 0.0001). A HER2 threshold of 740 amol/µg showed an agreement rate of 94% with IHC and ISH standard HER2 testing (p < 0.0001). Discordant cases (SRM-MS-negative/ISH-positive) showed a characteristic amplification pattern known as double minutes. HER2 levels >2200 amol/µg were significantly associated with longer disease-free survival (DFS) and overall survival (OS) in an adjuvant setting and with longer OS in a metastatic setting. CONCLUSION: Quantitative HER2 measurement by SRM-MS is superior to IHC and ISH in predicting outcome after treatment with anti-HER2 therapy.

Pancreatic cancer clinical trials and accrual in the United States.

PURPOSE: Pancreatic cancer clinical trials open in the United States and their accrual were examined to identify opportunities to accelerate progress in the treatment of pancreatic cancer. METHODS: Pancreatic cancer-specific clinical trials open in the United States in the years 2011 and 2012 were obtained from the Pancreatic Cancer Action Network database. Accrual information was obtained from trial sponsors. RESULTS: The portfolio of pancreatic cancer clinical trials identified by type (adenocarcinoma or neuroendocrine), phase, disease stage, and treatment approach is reported. More than half of trials for patients with pancreatic ductal adenocarcinoma applied biologic insights to new therapeutic approaches, and 38% focused on optimization of radiation or chemotherapy delivery or regimens. In 2011, pancreatic cancer trials required total enrollment of 11,786 patients. Actual accrual to 93.2% of trials was 1,804 patients, an estimated 4.57% of the patients with pancreatic cancer alive in that year. The greatest need was for patients with resectable cancer. Trials open in 2011 enrolled an average of 15% of their total target accrual. Physician recommendations greatly influenced patients' decision to enroll or not enroll onto a clinical trial. Matching to a clinical trial within a 50-mile radius and identifying trials for recurrent/refractory disease were documented as challenges for patient accrual. CONCLUSION: Overall trial enrollment indicates that pancreatic cancer trials open in 2011 would require 6.7 years on average to complete accrual. These results suggest that harmonizing patient supply and demand for clinical trials is required to accelerate progress toward improving survival in pancreatic cancer.