Evolving use of blast percentage thresholds and classification systems in AML clinical trials (2022–2025): A data-driven trend analysis

Abstract Background: The diagnostic threshold for acute myeloid leukemia (AML) has evolved significantly across classification systems over time. The FAB classification originally required ≥30% blasts for diagnosis. This was revised to ≥20% in the 2001 WHO criteria and maintained in subsequent versions. However, recent updates, particularly WHO 2022 and International Consensus Classification (ICC, launched in 2022), have expanded molecular designation of AML and introduced the MDS/AML overlap category, recognizing patients with 10–19% blasts in specific contexts, signaling a paradigm shift. The real-world adoption of these evolving classifications and blast thresholds in clinical trial design remains unclear. We aimed to evaluate trends in blast percentage usage and classification references in registered AML clinical trials over the past four years. Methods: We systematically reviewed AML studies registered on ClinicalTrials.gov with start dates from January 1, 2022, to June 1, 2025. Study characteristics were extracted using R (v4.4.2) via API, including trial start year, blast percentage thresholds explicitly defined in the Eligibility Criteria, classification systems referenced (e.g., WHO editions, ICC, NCCN), study phase, enrollment size, and regional distribution. Results: A total of 712 AML studies were identified between 2022–2025, of which 241 (33.8%) explicitly reported a blast percentage threshold in eligibility criteria. Across all years, thresholds of ≥20%, ≥10%, and ≥5% were observed, with a notable increase in the use of 10% blast thresholds over time: from 4.0% in 2022 to 25.8% in 2025, while use of 20% declined from 68.0% to 41.9%. The use of a 5% threshold was found almost exclusively in trials targeting relapsed/refractory AML. Parallel to this trend, adoption of newer classification systems also evolved over time. While WHO 2016 was the most frequently referenced system in 2022 and 2023, more recent classifications gained traction. In particular, WHO 2022 and WHO 2022/ICC were cited in 6.49%, 13.64%, 46.8%, and 59.57% of trials initiated in 2022 through 2025, respectively. Meanwhile, the ICC system, introduced in 2022, appeared in studies starting in 2023 and showed continued growth. A subset of trials referenced multiple classification systems simultaneously, reflecting the ongoing evolution and lack of uniform adoption across study protocols. From 2022 to mid-2025, the proportion of AML clinical trials involving CAR T/NK therapies remained relatively stable, ranging from 6.0% to 8.3% of total studies initiated each year. Geographic participation was global, with United States, China, and France consistently contributing the largest share of trials, though the proportion of U.S.-led studies modestly increased over time. Trial volume was highest in 2024 (n = 216), with observed numbers in 2025 lower to date, due to partial-year data collection through June 1, 2025. Study phase data were missing in some trials; among those with available data, Phase 1 and Phase 2 trials comprised the majority each year, accounting for 68.87%, 61.87%, 72.22%, and 70.88% of trials from 2022 through 2025, respectively. Most studies enrolled relatively small cohorts: across all years, >50% of trials enrolled ≤50 patients, while fewer than 5% included >1000 participants. Conclusion: From 2022–2025, AML trials demonstrated a clear trend toward lower blast percentage thresholds, paralleling the adoption of WHO 2022 and MDS/AML concepts. These findings suggest growing alignment of trial design with evolving disease biology and classification updates. However, the continued heterogeneity in blast cutoffs and classification usage across regions and studies emphasizes the need for greater standardization in eligibility criteria. Further evaluation is warranted to assess how these shifts impact trial outcomes, enrollment efficiency, and generalizability of findings.