RPPA-Profiling in Pediatric and Adult T-Cell Acute Lymphoblastic Leukemia Identifies Protein Patterns Associated with Outcome

Background: T-cell acute lymphoblastic leukemia (T-ALL) is an aggressive disease that arises from a malignant transformation and clonal expansion of poorly differentiated T-cell precursor cells. While pediatric T-ALL is among the most successfully treated cancers, outcome for adult T-ALL is more guarded. With the exception of the poor prognostic subgroup in adults, early T-precursor (ETP)-ALL, risk stratification is primarily based on measurable residual disease. Because proteins are the central effectors in the leukemia cell and as most chemotherapeutic agents target proteins, we set out to perform RPPA-profiling in primary T-ALL.

Methods: RPPA analysis was used to measure protein expression in 361 newly diagnosed T-ALL samples (age < 18, n=268; age ≥ 18, n=93) and 29 normal CD34+ samples using 321 validated antibodies. Out of these, 292 patients were enrolled on the COG AALL1231 phase 3 clinical trial; 139 were treated with modified augmented AFBM; intrathecal cytarabine, vincristine, dexamethasone, daunorubicin, PEG asparaginase, intrathecal methotrexate, and 150 with AFBM plus bortezomib (BTZ) (AFBM+B). The remaining 69 adult T-ALL patients were treated under a variety of protocols, and 13 were treated off protocol. Expression levels were normalized against CD34+ cells. Proteins were allocated into 32 functionally related subgroups (Protein Functional Group (PFG)). A progeny clustering algorithm was applied to each PFG to search for strong correlations between proteins and to identify an optimal number of Protein Clusters (PC). Block clustering identified PC that recurrently co-occurred together (Protein Constellation (CON)) and patients that expressed similar combination of CON were defined as Protein Signature (SIG).

Results: Cluster analysis in the context of 32 PFG resulted in 125 PC. Co-clustering of the 125 PC identified 13 CON that formed 10 SIG. SIGs were significantly associated with age, with pediatric samples (age 2-18yr) predominantly clustered in SIG-1-3 and SIG-6 and 7, and AYAs and adults age ≥30 yr. tended to fall more frequently in SIG-5 and 10 (p<0.001). ETP positive patients clustered mainly in SIG-4, 9 and 10, while none of them grouped in SIG-6 (p<0.001). Due to the successful OS-rate and EFS-rate, SIGs were not prognostic among the AALL1231 treated patients. However, when survival analysis was restricted to AALL121 study arms (i.e., ABFM, ABFM+B), two SIGs were identified that did either significantly better with (SIG-1, 3-yr. EFS 88% vs 60%, p=0.05) or without (SIG-10, 3-yr. EFS 86% vs 40%, p=0.01) BTZ. Ribosome/translation (e.g., EEF2, EEF2K, EIF2S1, EIF4EBP1), heat shock (HSF1, HSP90) and histone modification proteins (SUZ12, EZH2, H3-histone marks, HDAC2/6, KMT2A/B), were lower in SIG-1 (BTZ-sensitive) vs SIG-10 (BTZ resistant). Signal transduction pathway proteins were higher in SIG-1 (e.g., AXL, EGFR, ERBB2, IGF1R). In addition, survival analysis was performed for each individual protein within the ETP and near-ETP pediatric/ AALL1231 treated patients combined. This identified 22 prognostic proteins (p<0.05), which clustered into 2 groups, associated with significant different outcome (n=63, OS 87% vs 44%, p<0.001) (Figure).

Conclusion: We performed RPPA-profiling in a large cohort of pediatric and adult T-ALL. Step-wise clustering analysis classified patients into 10 SIGs based on recurrent patterns in protein expression. Two SIGs were identified that did better either with or without BTZ addition, pointing towards a potential utility for protein data to identify patients that benefit from BTZ. While more research is needed to understand why SIG-1 responds well to BTZ addition, the up-regulation of ribosomal proteins, histone modification proteins and stress proteins may suggest an association with the degree of protein synthesis/ cell turnover. In addition, poor vs favorable prognostic group were identified within pediatric ETP patients, indicating that protein data could add to conventional diagnostic tools in the process of risk stratification. Further studies are needed to identify protein targets that could potentially offer clinical utilization.

Figure. A. Heatmap showing 22 proteins that are individually significantly prognostic within the 63 pediatric ETP cases. Cluster analysis identified 2 protein clusters. B. Overall survival (B-left) and event-free survival (B-right) analysis for the two identified protein clusters.

Teachey:BEAM Therapeutics: Consultancy; Sobi: Consultancy.