OC 46.2 - Plasma protein signature by targeted mass spectrometry allows prediction of venous thromboembolism in colorectal cancer patients with 90% sensitivity

Monday, June 26, 2023

15:00 – 15:15 ET

Room: 510

Session Available As: On-Demand

Co-Author(s)

JB

Jeroen Buijs

Post Doc
Leiden University Medical Center
Leiden, Zuid-Holland, Netherlands
Nick van Es, MD PhD (he/him/his)

MD
Department of Vascular Medicine, Amsterdam University Medical Center, Amsterdam, The Netherlands
Amsterdam, Noord-Holland, Netherlands
Rayna Anijs, MSc

Post Doc
Leiden University Medical Center
Leiden, Zuid-Holland, Netherlands
FB

Floris T.M. Bosch

MD
Amsterdam University Medical Centers
Amsterdam, Noord-Holland, Netherlands
Bv

Bart J.M. van Vlijmen

Associate professor
Department of Internal Medicine, Division of Thrombosis and Hemostasis, Einthoven Laboratory for Vascular and Regenerative Medicine, Leiden University Medical Center, Leiden, The Netherlands.
Leiden, Zuid-Holland, Netherlands
FM

Frits I. Mulder

PhD student
Dept. of Vascular Medicine, Amsterdam UMC location AMC, University of Amsterdam, Amsterdam, The Netherlands
Amsterdam, Noord-Holland, Netherlands
HV

Henri Versteeg

Professor
Leiden University Medical Center
Leiden, Zuid-Holland, Netherlands

Presenting Author(s)

Yassene Mohammed, PhD (he/him/his)

Assistant professor
Center for Proteomics and Metabolomics, Leiden University Medical Center, Leiden, The Netherlands
Leiden, Zuid-Holland, Netherlands

Background: Current prediction models for cancer-associated thrombosis (CAT) perform suboptimal. We sought to assess the use of absolute quantitative proteomics for CAT.

Aims: identify plasma biomarker candidates to develop a prediction model for VTE in cancer patients.

Methods: Plasma samples were used from 142 stage III/IV colorectal cancer patients from a multinational, prospective cohort study. The primary outcome was objectively confirmed symptomatic or incidental deep vein thrombosis or pulmonary embolism during a 6-month follow-up period. In total 11 (7.7%) patients developed CAT. Mass-spectrometry-based quantitative targeted proteomics assays for a panel of 269 proteins were used, of which 64 are related to coagulation and complement, and 61 to cancer. The Mann-Whitney U test was used to compare groups.

Results: 156 proteins passed quantifiability criteria, of which eight proteins were differentially abundant with a fold change >15% and a P-value < 0.05 in patients who developed VTE compared to patients who did not (Table 1, Figure 1A). A progressive logistic regression with repeated cross-validation using 80/20% training/testing sets was performed using the discriminating proteins, c-statistics are reported in Figure 1B. Best model using top 5 proteins achieved a sensitivity of 90.9%% and a specificity of 62.6%, outperforming the dichotomized Khorana, Vienna-CATS, PROTECHT and CONKO risk scores, which had sensitivities of 9.1%, 45.5%, 27.3%, 36.4%, and a specificities of 79.4%, 55.7%, 64.1%, 77.9%, respectively.

Conclusion(s): Here we present our initial results on using protein abundance determined by absolute targeted proteomics to predict CAT in advanced CRC patients. Internal cross-validation showed promising results and upon external validation in an independent cohort, these biomarker candidates may be incorporated in risk models to improve CAT prediction that might help identifying cancer patients who benefit most from thromboprophylaxis.

Learning Objectives:

Participants will learn about targeted quantitative proteomics for plasma protein characterization
Participant will learn about deciphering plasma protein signature of cancer associated thrombosis measured in a prospective cohort
We will show how to use quantitative proteomics results to model and develop a classifier to discriminate between patient groups