Early mortality in patients with acute promyelocytic leukemia: a systematic review and meta-analysis

In patients with acute promyelocytic leukemia (APL), early mortality, defined as death within 30 days following diagnosis, is the main driver of overall survival. Intensivists play a pivotal role in the management of patients with APL.

The PubMed and Embase databases were searched on December 1, 2023. All studies focusing on early mortality in patients with newly diagnosed APL were included. Two reviewers independently selected the studies and assessed bias using the risk-of-bias 2 scale for randomized controlled trials (RCT) and the Newcastle-Ottawa scale for other types of studies. A meta-analysis using a random effects model was performed. The primary outcome was early mortality. A subgroup analysis was performed to assess the differences between interventional and observational studies. Secondary outcomes included reasons for early mortality and the occurrence of severe hemorrhage or differentiation syndrome (DS).

Of the 950 studies screened, 140 met the inclusion criteria. A total of 54,923 patients newly diagnosed with APL were included in the meta-analysis. A meta-analytical early mortality rate of 12% was calculated across 122 studies. Early mortality was lower in interventional studies (8%) than in observational studies (15%). Hemorrhage was the main cause of early mortality (57%), followed by sepsis (18%), DS (12%), and thrombotic events (2%). In a meta-regression, early mortality was influenced by patients' age, white blood cell count at diagnosis, time, study type, and country of origin according to income classification.

There is an encouraging decrease over time of early mortality in patients with APL. Early mortality is lower in interventional studies. This result reflects the difficulty of translating the successes of RCT and interventional research into real-world clinical practice. This study puts forward the need to improve and standardize early management, notably in the intensive care unit.

This study's protocol was preregistered on PROSPERO (CRD42023488480).