Introduction: High-risk (HR) acute promyelocytic leukemia (APL) remains a challenge due to early death (ED) and relapse. All-trans retinoic acid (ATRA) and arsenic trioxide (ATO) have revolutionized the APL treatment. Although the first results from the phase 3 APOLLO trial suggested that ATRA + ATO induction with idarubicin could further improve the prognosis of HR APL compared with AIDA regimen, the standard of care for HR APL has not been conclusively established. As China has pioneered in integrating arsenic agent into the APL treatment paradigm, ATRA and ATO dual induction has become the mainstream regimen for most centers in China. Therefore, this study aimed to explore the real-world prognosis HR APL in the dual induction era and develop a prognostic model based on the real-world data from a single center using Least Absolute Shrinkage and Selection Operator (LASSO) -Cox regression model.

Methods:

HR APL patients defined as white blood cell (WBC) count over 10×109/L from a tertiary hospital in China was included. All patients received ATRA and ATO regimen together with cytoreductive anthracycline chemotherapy in the induction phase. Demographic features, clinical characteristics, and survival data were retrieved from electronic medical records. Risk factors of ED were analyzed by logistic regression model. Overall survival (OS) was analyzed by Kaplan-Meier (KM) method and a visualized prognostic model was established by LASSO-Cox regression.

Results: From January 2009 to December 2023, a total of 109 HR APL patients were included. The median age was 39 years (range, 14~71 years). The male to female ratio was 1.3: 1. After a median follow-up of 65 months (95% confidence interval [CI]: 56.1~73.9), OS was not reached and the estimated 5-year OS rate was 86% in this retrospective cohort. The ED rate was 7.3%. Logistic regression revealed that peak WBC count during induction (MaxWBC)>100×109/L (odds ratio [OR]= 92.12, 95% CI: 3.607 ~ 689.38; p=0.004) and prothrombin time (PT)>18 seconds (s) (OR=21.78, 95%CI=1.016~466.735; p=0.049) was significantly associated with ED. Using cox proportional hazard model, male gender (hazard ratio[HR]=0.122; 95%CI: 0.021~0.073; p=0.019), MaxWBC﹥100×109/L (HR=5.675, 95%CI: 1.773~18.165; p=0.003), hemoglobin (Hb)<60g/L (HR=0.053, 95%CI: 0.007~0.386; p=0.004), PT>18s (HR=6.572, 95%CI: 1.656~26.078; p=0.006), platelet count (Plt)<40 ×109/L (HR= 4.045, 95%CI: 1.133~14.439; p=0.031) were independent risk factors of adverse OS. Considering the small sample number of the entire cohort, LASSO regression was used for variable selection, followed by multivariate Cox regression analysis to identify independent prognostic factors. This LASSO-Cox model confirmed that four variables were independent prognostic factors for OS, including MaxWBC (p=0.001), Plt (p=0.005), Hb (p=0.005), and PT (p=0.015). A nomogram was subsequently constructed based on theses selected predictors to provide a visual tool for prognosis assessment. The predictive performance of the nomogram was evaluated using receiver operating characteristic (ROC) curves and calibration curves. The resulting nomogram showed excellent discrimination with the area under the curve (AUC) of 0.892 and 0.766 for 1- and 3-year survival, respectively. The 1- year and 3- year calibration curves also demonstrated the good forecast precision for this nomogram.

Conclusion: Our findings showed that the real-world outcomes of HR APL have been dramatically improved with ATRA+ATO induction in the past decade. A nomogram including four clinical factors has been developed and could help to predict the prognosis of HR APL in the dual induction era. This nomogram model can serve as an efficient individualized quantitative tool for the physicians to identify the “true” or “extremely” high risk APL patients besides the Sanz risk model. This model should be validated by multi-center cohorts in the future.

This content is only available as a PDF.
2025
Sign in via your Institution