Amrita Cheema

PhD

Georgetown University

Associate Professor & Co-Director-PMSR Experimental Therapeutics

PLCO (Learn more about this study)

PLCO-119

Nov 25, 2014

Metabolomic profiling for pancreatic cancer risk in the PLCO Cancer Screening Trial

Although the survival outcomes of Pancreatic Ductal Adenocarcinoma (PDAC) continue to be poor, clinical and basic research in this field is disproportionately underrepresented as compared to other cancer sites. Currently, there are no blood or fluid screening tests in routine use for early detection of PDAC. The disease typically does not cause symptoms in the early stages and is often only diagnosed in the advanced or metastatic setting when curative treatment options are not possible. Although known environmental risks factors are associated with PDAC, the absolute risk for the development of PDAC is low in the general population even when multiple risk factors are present. Thus developing a specific and sensitive panel of biomarkers in conjunction with population data offers a pragmatic approach for PDAC risk prediction and early detection. This study will use a metabolomics-based molecular phenotyping approach to identify and characterize markers with potential clinical relevance, using the longitudinally collected PLCO biospecimens. In Specific Aim 1 we seek to validate predictive metabolite biomarkers we previously discovered in a clinical population, by comparing metabolomics profiles of plasma samples from normal controls to peri-diagnostic PDAC patients in the PLCO (case-control design). We previously tested the underlying hypothesis that targeted, quantitative evaluation of molecular fingerprints of PDAC in plasma (originally evaluated via matched plasma and tissue metabolomic profiles), may have potential clinical applicability. The biomarker classifier consisting of glycerophospholipids, sphingolipids, acylcarnitines, biogenic amines and amino acids showed a predictive accuracy of >95%. In Specific Aim 2 of the proposed study we will perform comparative untargeted metabolomics profiling of the PLCO’s longitudinally collected plasma samples for discovering new markers for early detection (prospective cohort design). We have recently demonstrated the utility of the metabolomics approach for identification of pre-clinical disease, which also provides a window of opportunity for leveraging disease modifying therapeutic interventions [1]. In this new study, we will test the hypothesis that novel plasma biomarkers assayed in a pre-diagnostic cohort can anticipate phenoconversion to PDAC at a later stage. Using appropriate controls (including those who never developed PDAC, and a separate “disease control” group of those who developed colorectal cancer), the biomarker panel will be tested for predictive specificity for identifying asymptomatic individuals who later developed PDAC. In Specific Aim 3 we propose to add value to the above aims by incorporating the PLCO’s unique longitudinal data on lifestyle, demographic, and clinical variables. Factors such as gender, race, family history of pancreatic cancer, age at diagnosis, obesity, history of diabetes, and tobacco smoking history will be investigated for subjects in the prospective cohort design (Aim 2) with two main objectives: (a) to study the influence of these variables on PDAC risk and outcomes, and (b) to study their effects on the predictiveness of the biomarker panels discovered in those aims.

Specific Aim 1. Validation of a panel of metabolite biomarkers for PDAC.
We propose to validate a novel set of peripheral bio-signatures of PDAC that were discovered in our previous work by comparing tissue profiles of patients diagnosed with pancreatitis to those with PDAC using samples from our local clinical population. Using stable isotope dilution multiple reaction monitoring mass spectrometry (SID-MRM-MS), this panel was validated in plasma samples. We hypothesize that such predictive bio-signatures will be absent in normal controls and hence propose to validate this biomarker panel in an independent set of plasma samples from the PLCO population that could be potentially used for PDAC diagnosis. To do this, we will conduct a case-control study using SID-MRM-MS to quantify the levels of the metabolites in plasma samples from normal controls (n= 63) to the peri-diagnostic cases of PDAC (n= 63), with blood samples collected within 24 months of diagnosis.

Specific Aim 2. Identification and validation of metabolites predictive of PDAC in a longitudinal cohort.

We hypothesize that metabolite signatures of PDAC will be present in pre-diagnostic samples. Therefore, in Specific Aim 2 we will perform untargeted metabolomic/lipidomic plasma profiling (see methods) of longitudinally collected samples from 307 PDAC patients collected ≤ 12 months prior to diagnosis (n = 38), 13- 24 months prior to diagnosis (n = 63), 25- 36 months prior to diagnosis (n=49) and 36 – 60 months prior to diagnosis (n = 157) in the PLCO. Based on the effect size of .73 in the previous study, we will have 97% power to detect differential metabolites at an overall significance level of 5%. SID-MRM-MS will be used to determine differential abundance of markers in longitudinally collected plasma samples over time. The PDAC biomarker sensitivity and specificity will be assessed by comparing the metabolites detected in the pre-diagnostic PDAC samples to samples from normal controls (n=150) and pre-diagnostic colorectal cancer patients (n=150).

Specific Aim 3. Delineation of epigenetic factors predictive of risk to PDAC.

Leveraging the richness of the PLCO data base, we proposed to investigate the roles of specific demographic and clinical factors and lifestyle exposures on PDAC risk. Male sex, nonwhite race, family history of pancreatic cancer, age at diagnosis, obesity, history of diabetes, tobacco smoking history, diet and alcohol drinking history will be investigated for subjects in the prospective cohort design (Aim 2, N=307) with two main objectives: (a) to study the influence of these variables on PDAC risk and outcomes, and (b) to study their effects on the predictiveness of the biomarker panels discovered in those aims. For example, in the latter analysis we would ascertain whether the biomarker panel achieves the same levels of sensitivity and specificity for important subgroups, e.g. males vs. females, and smokers vs. non-smokers. Patient demographics, medical history, survival, response to treatment and metobolomic profiling data would be integrated to build predictive models for pathway mitigation for the PLCO PDAC cohort.

Kathryn L. Taylor, Ph.D
Georgetown University

Christopher Loffredo, PhD
Georgetown University

Keith Unger, MD
Georgetown University