Abstract
Introduction
Lymphoma patients are at increased risk of thromboembolic events (TE), however, thromboprophylaxis in these patients is largely under utilized. Actual guidelines recommend different models for thromboembolic risk estimation in cancer patients. Proposed models are of limited use in lymphoma patients as their development is not based on specific characteristics for this patient population. Previously, we developed and internally validated a simple model, based on individual clinical and laboratory patient characteristics that would classify lymphoma patients at risk for a TE. The variables independently associated with the risk for thromboembolism were: previous venous and/or arterial events, mediastinal involvement, BMI>30 kg/m2, reduced mobility, extranodal localization, development of neutropenia and hemoglobin level < 100g/L. For patients classified at risk in derivation cohort (n=1236), the model revealed positive predictive value of 25.1%, negative predictive value of 98.5%, sensitivity of 75.4%, and specificity of 87.5%. The diagnostic performance measures retained similar values in the internal validation cohort (n=584). The aim of this study was to perform external validation of the previously developed thrombosis lymphoma (Throly) score.
Methods
The study population included patients with a confirmed diagnosis of non-Hodgkin lymphoma (NHL), Hodgkin lymphoma (HL), and chronic lymphocytic leukemia (CLL)/ small lymphocytic lymphoma (SLL) from 8 lymphoma centers from USA, France, Spain, Croatia, Austria, Switzerland, Macedonia, and Jordan. During 2015 to 2016, data were prospectively collected for venous TE events from time of diagnosis to 3 months after the last cycle of therapy for newly diagnosed and relapsed patients who had completed a minimum of one chemotherapy cycle. The score development and validation were done according to TRIPOD suggested guidelines. Sensitivity analyses were carried out to test the model robustness to possible different settings, according to in/out patient settings and according to different countries included.
Results
External validation cohort included 1723 patients, similar to the developed group and consisted of 467 indolent NHL, 647 aggressive NHL, 235 CLL/SLL and 366 HL patients, out of which 121 (7%) patients developed venous thromboembolic events. For patients classified at risk in external validation cohort, the model resulted in positive and negative predictive values of 17% and 93%, respectively. Based on new available information from this large prospective cohort study this model was revised to include the following variables: diagnosis/clinical stage, previous VTE, reduced mobility, hemoglobin level < 100g/L and presence of vascular devices. In the new score we divided patients in two groups: low risk patients, score value ≤ 2; and high risk patients, score value > 2. For patients classified at risk by the revised model, the model produced positive predictive value of 22%, negative predictive value of 96%, sensitivity of 51%, and specificity of 72%. In sensitivity analysis, the final model proved its robustness in different settings of major importance for lymphoma patients. The final model presented good discrimination and calibration performance. Concordance C statistics was 0.794 (95% CI 0.750-0.837).
Conclusions
Revised Thrombosis Lymphoma - ThroLy score is more specific for lymphoma patients than any other available score targeting thrombosis risk in solid cancer patients. We included biological characteristic of lymphoma, indolent vs aggressive, as well as data about dissemination of disease, localized vs advanced stage, reflecting specificity of lymphomas comparing to other types of cancer. Also, we pointed out significance of central vascular devices as risk factor having considered the role of vascular damage during insertion as a potential trigger for activation of the clotting cascade. This score is user friendly for daily clinical practice and provides a very good predictive power to identify patients who are candidates for pharmacological thromboprophylaxis.
Disclosures
Cheson: AbbVie, Roche/Genentech, Pharmacyclics, Acerta, TG Therapeutics: Consultancy. Ghielmini:Roche: Consultancy, Honoraria, Research Funding, Speakers Bureau. Jaeger:Gilead: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; Celgene: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees; Novartis: Membership on an entity's Board of Directors or advisory committees, Research Funding; AOP Orphan: Membership on an entity's Board of Directors or advisory committees; Roche: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees, Research Funding; Janssen: Consultancy, Honoraria, Membership on an entity's Board of Directors or advisory committees; MSD: Research Funding; Bioverativ: Membership on an entity's Board of Directors or advisory committees; AbbVie: Consultancy, Honoraria; Mundipharma: Membership on an entity's Board of Directors or advisory committees; Takeda-Millenium: Membership on an entity's Board of Directors or advisory committees; Takeda-Millenium: Membership on an entity's Board of Directors or advisory committees; Amgen: Membership on an entity's Board of Directors or advisory committees; GSK: Membership on an entity's Board of Directors or advisory committees; Infinity: Membership on an entity's Board of Directors or advisory committees.
Deep learning detects heart failure with preserved ejection fraction using a baseline electrocardiogram
