Clinical Implications of Central Nervous System Involvement in Acute Myeloid Leukemia

Introduction: Acute myeloid leukemia (AML) is an aggressive malignancy with variable clinical presentations and outcomes. Although central nervous system (CNS) involvement is uncommon, it can occur in both pediatric and adult patients. CNS localization is not routinely assessed in adults, leading to potential underdiagnosis of asymptomatic CNS involvement. Our study aimed to determine the incidence of CNS involvement in AML patients, using multiparametric flow cytometry (MFC) of cerebrospinal fluid (CSF) at diagnosis. We also explored risk factors associated with CNS involvement and its impact on prognosis.

Methods: We retrospectively analyzed data from 245 adult patients consecutively diagnosed with non-promyelocytic AML and treated with intensive chemotherapy (IC) at Fondazione Policlinico Tor Vergata between January 2010 and December 2023. Genetic/cytogenetic risk was assessed following the 2017 European Leukemia Net guidelines. Lumbar puncture (LP) was routinely performed at AML presentation, even in the absence of neurologic symptoms. Positivity by conventional cytology (CC) was defined as unequivocal morphologic evidence of leukemic blasts in CSF and/or white blood cell count (WBCc) ≥ 5/μl with fewer than 10 erythrocytes/μl. MFC analysis was performed by applying the same leukemic associated immunophenotype, identified in bone marrow/peripheral blood at diagnosis for each patient, via a multiple-color assay with a cocktail of 8-12 monoclonal antibodies. On average, 15074 events were acquired (range 0-201790). The presence of >10 clonally restricted or phenotypically abnormal events was regarded as a MFC positivity. CC positive patients received biweekly intrathecal injections of methotrexate and cytarabine, until CSF blast clearance, followed by weekly injections for 3-4 weeks. Patients with CC-MFC+ underwent additional LP evaluation at the end of induction chemotherapy. Intrathecal chemotherapy was administered only if CNS positivity persisted.

Results: Among the 245 patients, 133 underwent LP, while 118 did not due to various reasons (e.g., consent refusal, low platelet count unresponsive to platelets transfusion, coagulation disorders, urgent chemotherapy). Patients not undergoing LP were older (median age 59 vs. 56 years, p=0.001) and predominantly male (54% vs. 38%, p=0.012) but had similar cytogenetic risk (favorable 16% vs 30%; intermediate 48% vs 47%; adverse 25% vs 20%, p=ns). CSF+ was confirmed in 40/133 (30%) patients, 14(10%) CC+ and MFC+ and 26 (20%) CC- and MFC+. Five (4%) patients showed neurological symptoms at AML presentation, with 2 testing CC+MFC+, and 3 CC-MFC+. In the CSF+ group, the median of the total events acquired, and the leukemic events were 5846 (110-191142) and 34 (11-4949), respectively. Patients with CSF+ had a significantly higher frequency of FLT3 mutations (p=0.016) and a WBCc >50x10⁹/L (p=0.018). Overall, median follow-up was 16 months (range, 1-156 months) for patients undergoing LP. No significant difference was observed in complete remission rates between CSF+ and CSF- patients (78% vs. 87%, p=ns) or in relapse rates (55% vs. 50%, p=ns). However, overall survival was significantly shorter in CNS+ patients (median 13 months vs. 29 months; log-rank 4.94, p=0.028). Isolated CNS relapse occurred in 2 CC-MFC+ patients (8%) compared to 0% of those treated with intrathecal chemotherapy and in 5% (6/118) of patients not tested at AML presentation.

Conclusions: Our study highlights that CNS involvement in newly diagnosed AML patients may be more frequent than expected and significantly impacts outcomes in those undergoing IC. Despite the absence of neurological symptoms, systematic MFC examination of CSF is crucial, since relying solely on CC may result in un-detecting a substantial proportion of CNS localizations. In the era of precision medicine, a more careful assessment of the CNS could be particularly useful given the non-intensive therapies and targeted drugs whose pharmacological properties to overcome the blood-brain barrier is less well known than IC.