Comparison of the effectiveness of risk models used in acute promyelocytic leukemia

Abstract

Newly diagnosed acute promyelocytic leukemia (APL) patients were evaluated, and the effectiveness of the Sanz risk model was compared with other risk models developed for early mortality. To determine a simple, reliable, and highly effective risk model used in clinical practice for earlier recognition of high-risk patients at high risk of mortality. This is a retrospective analysis of 57 patients diagnosed with APL in our clinic between January 2002 and June 2019. Patients were grouped under the risk models of Sanz score, modified We found higher white blood count (WBC) is independently associated with 30-day mortality [Odds ratio (OR): 1.030, 95% confidence interval (CI): 1.005 - 1.055, p= 0.017]. Albumin, another variable included in the multivariable model, was found non-significant (p= 0.055). The modified Sanz risk score had a sensitivity of 77.78% and specificity of 66.67% to predict 30-day mortality for high and ultra-high-risk patients [Area under the curve (AUC): 0.727, 95% CI: 0.514 - 0.939, p= 0.032]. Additionally, the AUC of the modified Sanz risk score was significantly higher than the Sanz risk score (p= 0.028). We found no significant difference between the AUC of the Sanz risk score and & Ouml;sterroos et al.'s, Cai et al.'s, and ISTH risk scores. Timely recognition of high-risk patients, taking appropriate protective measures, and administering more aggressive supportive care can help reduce the early mortality of APL patients. Sanz risk score and other scoring systems have been guiding the identification of high-risk APL patients. However, the most effective scoring system could not be determined at the end of the study. There is still a need for standardized scoring systems to identify high-risk patients more effectively, including comorbidities.