Proteomic Profiling Based Classification of CLL Provides Prognostication for Modern Therapy and Identifies Novel Therapeutic Targets

The availability of targeted therapy and improved molecular characterization of Chronic Lymphocytic Leukemia (CLL) require a re-evaluation of treatment paradigms. As CLL heterogeneity is dependent on molecular and environmental factors, there is a need to create a new classification based on the integration of several factors. Here, we accomplish this goal by identifying CLL signatures using Reverse Phase Protein Array. Protein expression for 384 total and post translationally modified proteins was assessed in 871 CLL and Mature Small B Cell Leukemia (MSBL: HCL, HCLV, LGL-T, MCL, MZL, PLL, Richter's, T-Cell PLL) patients and was integrated with clinical data to identify strategies for improving diagnostics and therapy, making this the largest CLL proteomics study to date. Proteins were categorized into 40 protein functional groups (PFGs) based on literature and intra-dataset protein correlations and patients clustered based on PFG expression patterns into 6 recurrent protein expression signatures (PES) (Figure 1A). Individual protein expression (58/384 proteins), PFG expression (32/40) and overall PES were all highly prognostic of survival (OS) and time to first or second treatment (TTFT, TTST) (Figures 1B-C). The adhesion, apoptosis-occurring, apoptosis-regulating, heat shock, histone1 (marks), histone 2 (modifiers) and the STP-regulation PFGs were prognostic for all 3 outcome measures. Notably SG-A contained most of the MSBL and 15/16 cases of hairy cell leukemia, but the CLL cases within this SG fared very poorly. For OS, groups A and C had markedly inferior survival (P<0.0001) (10.3 and 20.3 median years) relative to the other 4 groups, which were statistically similar to each other. First treatment occurred sooner for Groups A and C (5.8 and 5.23 median years). Additionally, the TTST was also inferior for Group A (median 3.5 years). There were significant differences in age, hemoglobin, platelets, % BM and PB lymphocytes and β2M between the SG, but not for race (p = 0.84) or gender (p = 0.72). , Historically adverse cytogenetic aberrations del 11q and del17p events (23% overall) were less common in SG A, B, D and E (15, 14, 16, 17%) and overrepresented in SG C (32%), while historically favorable 13q changes were seen across all groups as was Trisomy 12 (14% overall), although SGs A and E were enriched (25%, 22%) while SG-F was low (5%) for Trisomy 12. SG membership superseded other traditional prognostic factors (Rai Staging, IGHV Status) and were prognostic for modern (BTK inhibition) and older CLL therapies. SGs A and C responded poorly to chemotherapy regimens compared to the other groups, whereas all groups responded well to BTK inhibitors except for SG-A. SGs and PFGs membership provided novel drug targets (see our other abstracts on TP53BP1 and ASNS) and defined optimal candidates for Watch and Wait (WaW) vs. early intervention. A model based on the accumulation of irregularities in ANXA1, TFRC, and SMAD2.p245 expression, optimally predicted TTFT overall and in early stage CLL patients. Patients with < 1 negative level of either of the 3 proteins, have a median TTFT of 14.59 years, whereas patients having 2-3 have a median of 5-6.27 years (P<0.0001). CHEK1.pS345, GAB2, IGFBP2, S100A4, WEE1.pS642, and ZAP70 were universally overexpressed by all SGs, suggesting them as ideal targets for inhibition. Collectively proteomics demonstrates promise for improving classification, therapy strategy determination, and identifying novel therapeutic targets.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal