NCAM1 (CD56) As a Predictor of Response to Standard Intensive Induction in Newly Diagnosed Acute Myeloid Leukemia

Background

CD56, also known as neural cell adhesion molecule 1 (NCAM1), is aberrantly expressed in 20% of AML patients (pts) (Sasca et. al. Blood 2019) and its presence on the surface of myeloblasts has been identified as an indicator of worse response to Venetoclax (VEN)-based therapy in newly diagnosed (ND) and relapsed or refractory (R/R) AML (Saultz et al. ASH 2023). Whether this negative prognostic impact of CD56 expression applies to other standard-of-care AML treatments is largely unknown. We sought to determine if CD56 expression on myeloid blasts reduces response and survival after standard intensive induction chemotherapy (IC) in a retrospective cohort of ND AML. We also compare the estimated effects in this IC cohort with a retrospective cohort of pts from the same institution treated with VEN and a hypomethylating agent (HMA).

Methods

ND AML pts treated with cytarabine and idarubicin since 2007 at Oregon Health & Science University (OHSU) on whom adequate follow-up data was available for determining clinical response were included. APL was excluded. Data were retrospectively obtained from the medical record and included clinical characteristics; cytogenetic, mutational (m), and flow cytometry-based immunophenotype results on AML-diagnostic bone marrow samples; post-induction clinical response and early death (Döhner et al. Blood 2022); and overall survival (OS).

Comparisons between categorical variables were assessed with Fisher’s test. The predictor of interest was pre-treatment CD56 blast expression (any level as noted in the pathology report summary of flow output). The rate of composite complete remission (cCR), defined as achievement of a complete remission with or without count recovery, was estimated by sample proportion and 95% exact binomial confidence interval (CI) and modeled with logistic regression. Patient follow-up was measured from date of IC start and estimated with the reverse Kaplan-Meier (KM) method. OS was estimated using KM and modeled with Cox regression, the latter yielding unadjusted or covariate-adjusted hazard ratios (HR) and Wald test p-values. Allogeneic hematopoietic cell transplant (HCT) was considered a time-dependent covariate. Multivariable models included CD56 expression (positive vs. negative) and other pt or disease features that were associated with the clinical outcome, as defined by univariable p-value <0.100.

Results

Of 282 pts meeting inclusion criteria, 53% were male, median age was 58 years (range 19-80), and 9% had an antecedent hematologic disorder (AHD). ELN 2022 prognostic risk was favorable in 19%, intermediate in 35%, adverse in 37%, and unknown in 9% of pts. Among 236 pts with diagnostic NGS, 7% had mTP53, 10% mRUNX1, 23% mRAS, 15% mIDH1/2, and 8% mPTPN11. 20% had FLT3-ITD. CD56 expression was detected in 82 pts (29%). CD56 was not associated with ELN classification, mTP53, mRAS, mIDH, mPTPN11, or FLT3-ITD (p≥0.090). CD56 positivity was far less common among mRUNX1 pts (9%) compared to wildtype (31%; p=0.028). 117 pts (41%) underwent HCT, at a median time of 4.8 months (mos).

cCR rate was 63.1% (95% CI: 57.2% - 68.8%). By ELN risk group, cCR rate was 84.9% in favorable, 69.0% in intermediate, and 49.1% in adverse risk pts. However, there was no association between CD56 positivity and cCR (odds ratio [OR]=0.95, p=0.837).

Median follow-up was 99 mos and median OS was 24 mos (95% CI: 18 - 40) with 175 deaths (62%). OS rates were 64.8% at 1 year and 44.8% at 3 years post-IC. Male sex (HR=1.30, p=0.082), age ≥60 (HR=2.09, p<0.001), AHD (HR=2.89, p<0.001), and non-favorable ELN risk (HR=2.85, p<0.001) were predictors of inferior OS. Notably, there was no correlation between CD56 expression and OS in the univariable (HR=1.08, p=0.650) or multivariable (HR=0.99, p=0.956) setting.

In comparison to this lack of association between CD56 positivity and clinical outcomes, in a previously reported cohort of 98 ND AML pts treated with VEN+HMA at OHSU between 2018 and 2022 (median age 73, 58% adverse risk), CD56 expression was an independent predictor of shorter OS (HR=2.10, p=0.012) (Saultz et al. ASH 2023).

Conclusions

NCAM1 was not predictive of worse clinical outcomes in AML patients treated with the standard 7+3 regimen. The prognostic role of CD56 may be limited to VEN-based therapy. Thus, when adding upfront VEN to IC backbone for fit patients, the chemo component may nullify the negative impact of CD56 that has been observed following VEN+HMA therapy.