Ann Oberg

PhD

Mayo Clinic

Professor of Biostatistics

PLCO (Learn more about this study)

2016-0041

Feb 2, 2017

Phase 3 validation of early detection biomarker panel for pancreatic cancer

Biomarkers are needed to detect pancreatic ductal adenocarcinoma (PDAC), usually diagnosed too late for surgical or other therapeutic treatments. Starting with a PDAC cell reprogramming model (iPS line) that recapitulates human PDAC progression, and from which secreted and released proteins had been identified, we tested a subset of the proteins for their potential as robust plasma biomarkers for detection of PDAC using rigorous criteria for validation. We have optimized ELISA to perform the biomarker validation phases (Pepe M et al. JNCI 2001; 93:1054-61) with independent sets of plasma samples from patients with various stages of PDAC, from individuals with benign pancreatic disease, and from healthy controls. Our samples from Mayo Clinic consisted of: Phase 1 discovery (N=20), Phase 2a validation (N=189), and Phase 2b validation (N=537) design. We found that THBS2 can discriminate all stages of PDAC consistently over the three investigations (receiver operating characteristic (ROC) c-statistic=0.76 in Phase 1, 0.842 in Phase 2a, and 0.875 in Phase 2b), performing better in resectable Stage I cancer (c-statistic 0.839 in Phase 2a, and 0.896 in Phase 2b) than in Stage III/IV cancer (c-statistic 0.866 in Phase 2b). THBS2 levels combined with those for CA19-9, a previously identified PDAC biomarker, yield ROC c-statistics of 0.977 in Phase 2a and 0.952 in Phase 2b of Stage I PDAC and 0.940 in Phase 2a and 0.961 in Phase 2b of Stage II PDAC. Across all stages of PDAC, the combination of THBS2 and CA19-9 yielded c-statistics of 0.956 in the Phase 2a study and 0.970 in the Phase 2b study. We propose to validate the CA19-9 and THBS2 panel in the PLCO cohort incident cases using PRoBE design principles (Pepe M et al. JNCI 2008 100: 1432-8). The biomarker panel consisting of CA19-9 and THBS2 will require small quantities of serum and plasma for a validation study. Specific aims are: (1) To obtain plasma samples and covariate data on incident pancreatic cancer cases which have occurred in PLCO and 1:1 non cancer controls (matched by age and date of study entry); and (2) a. Test the performance of CA19-9 and THBS2 in a subset of participants who were diagnosed with PDAC within one year of the drawn sample; b. Assess the longitudinal trajectory of CA19-9 and THBS2 panel in the patients with multiple samples before PDAC diagnosis; and c. Within- patient trajectories of CA19-9 and THBS2 will be assessed graphically, and use these data to understand the window of time when PDAC diagnosis would be possible. This research will be supported by a U01 grant (1U01CA210138) to Drs. Gloria Petersen at Mayo Clinic and Kenneth Zaret at University of Pennsylvania; project start date was August 15, 2016. The results found to date were submitted in a manuscript that is currently in review (at Science Translational Medicine).

Aim 1. To obtain plasma samples and covariate data on available incident pancreatic cancer cases which have occurred in PLCO and 1:1 non cancer controls (matched by age and date of study entry). We will request 500 ul serum and 200 ul EDTA plasma. Based upon information provided by Claire Zhu, Ph.D. Coordinator of the PLCO biospecimen program at NCI, Table 1 lists the plasma samples that were known as of 11/19/2015 and potentially available.

Table 1. PLCO samples available for incident pancreatic cancer cases in cohort. Note that numbers are cumulative, i.e., 52 is a subset of 104, etc.
----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Number in samples drawn relative to diagnosis, within: Total pancreatic cancer cases
------------------------------------------------------------------------ ----------------------------------------
1 year 2 years 3 years 4 years
Plasma samples 52 104 148 176 349
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
These counts have been updated in feasibility discussions with PLCO personnel and we anticipate that at least 210 will be the usable number of subjects.

Aim 2. a) Test the performance of CA19-9 and THBS2 in a subset of participants were diagnosed with PDAC within one year of the drawn sample.
b) Assess the longitudinal trajectory of CA19-9 and THBS2 panel in the patients with multiple samples before PDAC diagnosis.
c) Within- patient trajectories of CA19-9 and THBS2 will be assessed graphically, and use these data to understand the window of time when PDAC diagnosis would be possible.