It is thought that unlike acute lymphoblastic leukemia, adult acute myeloid leukemia (AML) rarely disseminates to the central nervous system (CNS). Therefore, routine lumbar puncture (LP) is rarely performed in patients with newly diagnosed AML, and the incidence of CNS leukemic infiltrate in AML patients is unknown. Because induction chemotherapy for AML includes high-dose cytarabine that penetrates the blood-brain barrier, cytarabine may eliminate any CNS disease present at diagnosis. However, high-dose cytarabine has not been included in recent induction chemotherapy protocols designed primarily for elderly patients with AML. To determine the incidence of CNS leukemia in these patients and to identify risk factors for CNS disease, we reviewed the medical records of 1,412 patients who received induction therapy at The University of Texas MD Anderson Cancer Center between January 2000 and December 2012 (144 months). Of these patients, 779 (55.2%) were male, and the median age at diagnosis was 62 years (range, 14–89 years). In the majority of the patients (1370/1412; 97.0%), LP was performed only if clinically indicated. Of these patients, 45 (3.4%) had CNS disease (12 at diagnosis and 33 during the course of the disease). The incidence of CNS leukemia among 42 AML patients (median age, 41 years; range, 14–49 years) who underwent LP at diagnosis as part of an investigational treatment regimen (N=8; 19%) was significantly higher (P<0.0001). The median time from diagnosis to CNS disease detection was 8 months (range, 0–120 months). Compared with that of Caucasian patients, CNS leukemia rates were higher among African American patients (odds ratio [OR], 4.1; 95% confidence interval [CI], 1.9–8.9) and patients of Asian (OR, 4.8; CI, 1.3–15.8) or Hispanic (OR, 2.5; CI, 1.0–5.9) descent. Induction chemotherapy included high- or medium-dose cytarabine in 826 patients (52.8%), low-dose cytarabine in 187 patients (11.9%), and no cytarabine in 552 patients (35.3%). The CNS disease incidence among patients who received cytarabine (3.2%) was similar to that of patients who did not receive high dose cytarabine (3.3%).

Compared with patients who had no CNS disease, patients who had CNS dissemination had more bone marrow blasts at diagnosis (P = 0.003) and higher lactate dehydrogenase levels (P = 0.001). The CNS disease incidence among 132 AML patients with acute monoblastic leukemia (French-American-British classification M5; 7.6%) was significantly higher than that of patients with other AML subtypes (2.8%; P = 0.004). Conversely, the incidence of CNS disease among patients with prior myelodysplastic syndrome was only 0.6% (P = 0.05). Age; sex; cytogenetic abnormalities including complex karyotype, t(8:21), and inv16; and molecular aberrations including NPM1, FLT3, and RAS mutations were not associated with an increased incidence of CNS dissemination. In a multivariate logarithmic model, only ethnicity (African American or Asian descent), high bone marrow blast count, and high LDH levels predicted CNS involvement. Patients with or without CNS disease had similar overall survival rates. However, the median disease-free survival (DFS) duration of patients without CNS disease (29 months; 95% CI, 19 - 39), was significantly longer than that of patients with CNS disease (7 months; 95% CI, 2 - 12; P<0.0001). Taken together, our data suggest that, contrary to the current dogma, AML spreads into the CNS in nearly 20% of newly diagnosed patients. Patients who received high-, low-, or no cytarabine had similar incidences of post-treatment CNS infiltration. Risk factors for post-treatment CNS infiltration were a high number of bone marrow blasts, acute monoblastic leukemia, and African American or Asian descent. The DFS duration of patients with CNS dissemination was significantly shorter than that of patients with no CNS dissemination, suggesting that the CSF is a sanctuary for residual AML. Therefore, further studies that determine whether, in whom, and at what treatment stage AML patients should undergo LP or receive intrathecal chemotherapy are warranted.

Disclosures:

No relevant conflicts of interest to declare.

Author notes

*

Asterisk with author names denotes non-ASH members.

This icon denotes a clinically relevant abstract

Sign in via your Institution