Diagnostic performance and practical feasibility of the core outcome set for cardio-oncology : a retrospective observational analysis
B Von Kemp, E Wirix, B Roosens, A Motoc, B CosynsAbstract
Background & Aims
Cardio-oncology research faces particular practical challenges, many originating from its multidisciplinarity, tracing back to data generalizability, comparability and granularity. A Cardio-Oncology Core Outcome Set (COS-CO) has been developed to facilitate standardized data collection and comparison, but has never formally undergone validation.
Methods
This retrospective observational study assessed construct validity and diagnostic performance of COS-CO for cancer therapy-related cardiovascular toxicity (CTR-CVT) detection in a real-world cancer patient population who underwent standard-of-care surveillance at a cardio-oncology clinic. We identified Red Flag parameters to highlight high-risk patients and compared these to CTR-CVT diagnoses, and calculated sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). Registration completeness of COS-CO parameters was evaluated to assess feasibility.
Results
We included 144 patients (54.2% male, mean age 61.2(±15.2)y). Most common cancers were breast (20.8%), melanoma (18.1%) and prostate cancer (13.2%). Pre-existing cardiovascular disease was present in 32.6%. We diagnosed 66 CTR-CVT cases and 99 patients were flagged as high-risk in COS-CO (≥1 Red Flag parameter) of whom 57 were CTR-CVT cases. Sensitivity of this 'Red Flag' approach was 86.4%, specificity 46.2%, PPV 57.6% and NPV 80.0%. We observed a positive likelihood ratio of 1.61 and a negative likelihood ratio of 0.29. For standalone COS-CO parameters, high specificity was observed but sensitivity was low. In standard-of-care, 82.8% of COS-CO parameters were registered (average 12.4, median 13/15).
Conclusion
COS-CO appears to be a feasible clinical tool, fit for universal application at cardio-oncology clinics with acceptable sensitivity and NPV for CTR-CVT detection using a multiparametric approach. Further research is required to identify the most prognostically relevant COS-CO parameters and the subpopulation where COS-CO is most predictive. Clinical feasibility of COS-CO registration is high, allowing incorporation in standard care without significant additional time investment.Graphic Abstract