P442: SPINK2 PROTEIN EXPRESSION QUANTIFIED BY IMMUNOHISTOCHEMISTRY REFINES RISK STRATIFICATION AND PREDICTS THERAPY OUTCOMES IN AML

Abstract

Background: Suboptimal prognostication, therapy refractoriness and relapse contribute to the poor outcome of Acute Myeloid Leukemia (AML) patients. Leukemic stem cells (LSC) are considered crucial drivers of relapse and therapy resistance. Clinicopathological characterization of LSC-associated genes with convenient assessment tools such as immunohistochemistry (IHC) is needed. Our initial screening of AML datasets for LSC-associated genes identified Serine Protease Inhibitor Kazal type 2 (SPINK2) with high expression in AML, particularly in LSC fractions. However, in-depth analysis of its clinicopathological associations and prognostic utility in AML are lacking. Aims: To assess SPINK2 expression in adult AML with IHC, and determine its clinicopathological associations, prognostic impact and association with therapy response Methods: We studied the expression of SPINK2 by IHC in 172 non-M3 adult AML patients (median age:52yrs, range: 18-86yrs) treated at the Prince of Wales Hospital. The majority (90.8%) were de novo type and 72.3% had intermediate-risk(IR) cytogenetics, with 88.5% receiving Daunorubicin + Ara-C (DA 7 + 3) induction. SPINK2 expression was assessed by 3 hematopathologists in a double-blinded manner and quantified by a composite score (range: 0-16). Clinical data was collected for all cases. Results:SPINK2 staining in leukemic blasts was consistently cytoplasmic, and its protein expression strongly correlated with mRNA levels by qPCR in a subset of 128 patients (r=0.716, P<0.0001). The cohort was dichotomized at the median score 3, since this cut-off showed strongest association with poor event-free survival (EFS) and overall survival (OS). High SPINK2 (SPINK2hi) patients accounted for 77/172 (44.8%) cases, and low SPINK2 (SPINK2lo) patients for 95/172(55.2%). SPINK2hi associated with IR cytogenetics (P=0.014), normal karyotype (P=0.019) and NPM1 mutation (P<0.001), while SPINK2lo with t(8;21) and CEBPA double-mutation (both P<0.001). Survival and treatment-response analyses were performed on a subgroup of 137 patients who received DA 7 + 3 based induction. SPINK2hi patients had lower CR rates (73.8% vs 88.3%,P=0.028), higher 6-month relapse rates (31.8% vs 9.1%,P=0.004), lower 5-yr RFS (25.8% vs 46.8%,P=0.004), EFS (16.6% vs 37.2%,P<0.001) and OS (25.3% vs 51.2%,P<0.001). SPINK2 status also identified high risk patients of the IR cytogenetic subgroup. Of these, SPINK2hi patients had lower CR rates (68.6% vs 90.0%,P=0.013), higher 6-month relapse rates (31.4% vs 6.9%,P=0.018), lower 5-yr RFS(27.0% vs 44.6%,P=0.018), EFS(15.4% vs 38.3%,P<0.001) and OS(22.9% vs 51.5%,P=0.002). In this subgroup, median SPINK2 score was higher in patients requiring ≥2 inductions to achieve CR vs. patients requiring 1 course (7 vs 2,P=0.009). Multivariate analyses in whole and IR cytogenetics cohorts showed the poor prognostic effect of SPINK2hi on EFS and OS independent of age, cytogenetic risk, mutations and achievement of CR1, including stem cell transplantation given in CR (Table 1). In our NPM1mut subgroup (N=41), SPINK2hi status associated with poor EFS (HR:3.3,P=0.04) and OS (HR: 4.0,P=0.007) independent of age and FLT3-ITD with high allelic ratio. Incorporation of SPINK2 with FLT3 status refined ELN2017 risk definition, whereby 15/22 favorable-risk patients could be reclassified as intermediate-risk. Image:Summary/Conclusion: High SPINK2 protein expression by IHC predicted increased chemoresistance and early relapse risk in AML, and was an independent adverse prognostic biomarker, particularly in patients with intermediate-risk cytogenetics and NPM1 mutations.