PD-L1 and Tumor Mutational Burden Are Not Predictive of Thrombosis in a Cohort of 1,221 Patients with Solid Organ Malignancies

Introduction: Malignancy is a well-known risk factor for thrombosis. While many clinical risk factors for cancer-associated thrombosis have been described, it remains unknown how certain tumor specific observations (such as programmed death ligand 1 expression (PD-L1) and tumor mutational burden (TMB)) correspond to thrombotic risk. To determine the relationship between PD-L1 and TMB quantification and thrombosis in patients with solid tumors, we evaluated a large cohort of clinical samples from an NCI designated cancer center, correlating tissue sample pathology with the development of venous or arterial thromboembolism.

Methods: We performed a retrospective cohort study of patients between the ages of 18-89 who underwent histopathologic examination and next generation sequencing for a diagnosis of a solid organ malignancy at the Knight Cancer Institute between June 2019 and Feb 2021. Medical records were reviewed to document clinical and demographic information as well as for the development of thrombosis after cancer diagnosis. Multivariable logistic regression was performed to assess if PD-L1 expression and TMB are independent predictors of thrombosis. Subgroup analysis was then performed to evaluate whether this relationship differed by the organ of the primary malignancy. All analyses were conducted in R (R Core Team 2019).

Results: We identified 1,221 patients with solid organ malignancies (mean age 62, 53.6% male). The most common malignancies in the cohort were lung cancer (13.8%), pancreatic cancer (12.3%) and colon cancer (12.3%). Thrombotic events occurred in 206 patients (16.8%) after diagnosis of their malignancy (100 deep vein thrombosis, 42 pulmonary embolism, 46 visceral vein thromboses, 6 superficial vein thromboses, and 11 arterial events). The mean time from initial biopsy to thrombosis was 224 days. TMB (mean 8.5 mutations/Mb) was evaluated in all patients in the cohort, while PD-L1 expression testing (mean 23.0%) was available for 255 patients. On multivariable logistic regression adjusting for age and patient sex, neither TMB (ORadj: 0.97, CI: 0.92-1.03) nor PD-L1 expression (ORadj 1.00, CI: 1.00-1.02) were significant predictors of thrombotic events. Subgroup analysis by primary malignancy type did not demonstrate any specific primary sites for which TMB or PD-L1 were predictive of thrombosis.

Discussion: Our analysis found no predictive relationship between TMB and PD-L1 expression with thrombotic events in patients with solid organ malignancy. Further analysis is needed to determine if specific treatment protocols, such as the use of immunotherapy in patients with varying levels of PD-L1 and TMB, alter thrombosis risk.