CNS Status at Diagnosis Can Identify Acute Myeloid Leukemia Patients with Poor Induction Outcomes

Background: The incidence of central nervous system (CNS) involvement in acute myeloid leukemia (AML) is not well known, and the use of routine lumbar punctures (LPs) in these patients (pts) has been widely debated. Clear guidelines for diagnostic LP in newly diagnosed AML pts are lacking, practices vary among experts and institutions, and LPs are not routinely performed in AML pts. Estimates of CNS involvement vary due to differing detection methods, and current AML management may have reduced the frequency of leukemic CNS involvement. Our institution guidelines recommend performing LPs for pts with symptomatic disease, monocytic differentiation, or a presenting white blood cell count (WBC) >100 × 10^9/L. We aimed to determine the incidence of CNS involvement in pts presenting to our inpatient hematology service with AML, identify clinical characteristics associated with CNS involvement, and assess overall survival (OS), event-free survival (EFS), and relapse rates.

Methods: The AML database at our institution was queried for pts diagnosed with AML from January 1, 2010, to December 18, 2022, who underwent an LP at diagnosis. Patients with acute promyelocytic leukemia (APL) and those younger than 18 years were excluded. Data was collected on patient demographics, AML morphology, cytogenetics, molecular markers, laboratory parameters, cerebrospinal fluid (CSF) flow cytometry/morphology, CNS symptoms, and induction treatment regimen/intensity. CNS grading was determined utilizing the Steinherz/Bleyer algorithm, and AML risk stratification was determined utilizing the 2022 European Leukemia Network (ELN) criteria.

CNS involvement was defined as: CNS1: no blasts in CSF; CNS2a: blasts present with WBC < 5/μL and red blood cells (RBC) <10/μL; CNS2b: blasts present with WBC < 5/μL and RBC ≥ 10/μL; CNS3a: blasts present with WBC ≥ 5/μL and RBC < 10/μL; CNS3b: blasts present with WBC ≥ 5/μL and RBC ≥ 10/μL; CNS4 = unknown.

Pearson’s Chi-squared test, Fisher’s exact test, Wilcoxon rank-sum test (Mann-Whitney U-test), and Kruskal-Wallis rank-sum test were used for statistical analysis. Cox proportional hazards regression, hazard ratios, and confidence intervals (CIs) were calculated for OS and EFS. Analyses were performed using R version 4.3.1.

Results: From 2010-2022, 880 pts presented with AML at our institution. A total of 126 AML pts underwent LP between initial presentation and day 35, the majority had no CNS symptoms at the time of LP. A total of 56/126 pts (44%) were female, and 108 (86%) were white. The median age at diagnosis was 50 years (range: 18-82). Based on the 2022 ELN risk category, 31 (25%) pts were favorable risk, 55 (44%) were intermediate risk, and 40 (32%) were adverse risk. CNS disease (CNS2, 3) was present in 10 (34%) of favorable-risk pts, 22 (42%) of intermediate-risk pts, and 10 (26%) of adverse-risk pts. CNS risk category grading was CNS1, CNS2a, CNS2b, CNS3a, CNS3b, and CNS4 in 77, 17, 16, 4, 5, and 5 pts, respectively. Of the 80 pts without CNS symptoms, 38% had evidence of CSF involvement (CNS2, 3). Patients with a WBC > 100 × 10^9/L at diagnosis were more likely to have CNS2 or 3 disease (65%) compared to those with a WBC < 100 × 10^9/L (29%).

The 5-year OS probability for the group was 48% with a median of 4.2 years. The 5-year OS for CNS1, CNS2, and CNS3 was 50%, 45%, and 22%, respectively (p = 0.033). Multivariate analysis showed that ELN adverse risk category (HR 4.34, 95% CI: 2.01-9.37, p < 0.001), age at diagnosis (HR 1.03, 95% CI: 1.01-1.05, p = 0.004), and CNS3 disease (HR 3.73, 95% CI: 1.53-9.05, p = 0.004) were associated with worse OS. The 5-year incidence of CNS relapse in CNS1, CNS2, and CNS3 disease was 6.9%, 9.1%, and 22%, respectively. The 5-year EFS for the group was 41% with a median of 1.5 years. The 5-year EFS for CNS1, CNS2, and CNS3 was 43%, 42%, and 22%, respectively (p = 0.11). Multivariate analysis showed that ELN adverse risk category (HR 3.68, 95% CI: 1.78-7.63, p < 0.001), age at diagnosis (HR 1.03, 95% CI: 1.01-1.05, p = 0.006), and CNS3 disease (HR 2.98, 95% CI: 1.24-7.15, p = 0.014) were associated with worse EFS.

Conclusion: Our data suggest that diagnostic LP with intrathecal chemotherapy is warranted for pts with a high WBC count or monocytic differentiation. Using these criteria, 38% of pts had blasts in the CSF. Patients with CNS involvement have poor outcomes regardless of risk category. Further studies are needed to establish a standard of care for CNS prophylaxis and therapy in AML pts.