Dynamic Alterations in Gene Expression in Ibrutinib Treated CLL Reveal Profound Impact on Multiple Signaling Pathways

Ibrutinib covalently binds Bruton tyrosine kinase (BTK) leading to sustained target inhibition. On-target effects include reductions in BCR and NF-κB signaling and inhibition of tumor proliferation and survival (Herman, Blood 2014). In randomized trials, ibrutinib induced superior responses and extended the survival of patients with chronic lymphocytic leukemia (CLL) in both the relapsed/refractory and frontline setting (Byrd, NEJM 2014; Burger, NEJM 2015). However, deep remissions are uncommon and most patients have readily detectable residual disease in the blood. To better understand the biology of persistent disease on ibrutinib, we performed RNA-sequencing on CD19+ selected tumor cells from 14 of 86 patients treated with single-agent ibrutinib (ClinicalTrials.gov NCT01500733); 8/14 (57%) had an unmutated IGVHsequence, 8/14 (57%) had del(17p), 6/14 (43%) were previously untreated and 4/14 (29%) had relapsed on ibrutinib at the time of analysis (range: 8-47 months). At best response, 3/14 (21%) achieved a CR, 10/14 (71%) a PR or PRL and 1/14 (7%) had SD. Three samples per patient were analyzed: T1 (baseline), T2 (day 28) and T3 (6 months), for a total of 42 samples. Principal component analysis (PCA) grouped samples by patient, as observed in a similar study (Herishanu, Blood 2011). After adjusting for patient-specific effects, PCA clearly revealed the effect of treatment, separating on-treatment samples (T2, T3) from baseline (T1). To identify differentially expressed (DE) genes, we selected transcripts that showed at least a 2-fold change between baseline and treatment samples (FDR<0.1). Of the ~14,000 measured genes, there were 653 DE genes on-treatment compared to baseline. Notably, T3 samples demonstrated stronger changes in gene expression than T2 samples, in line with a more profound treatment response with continued drug exposure. Compared to T1, there were 80 DE genes expressed at T2 only, 330 at T3 only, and 243 at both T2 and T3. Among all DE genes, 498 decreased and 155 increased at either T2 or T3 compared to T1.

To identify the biologic impact of treatment, we dissected the gene expression changes using pathway-specific gene signatures represented in the KEGG pathways with the molecular signatures database tool (MSigDB; Broad Institute, Boston, MA). Strikingly, no pathways were significantly enriched on treatment. In contrast, multiple gene signatures (n=46) were significantly turned down on ibrutinib (FDR<0.01). Using more stringent cutoffs (p<0.0005; FDR q<0.005; ≥5 significant genes per pathway), nine KEGG pathways were significantly inhibited by ibrutinib, including cytokine signaling, cell adhesion, systemic lupus, p53 response, MAPK signaling, cell cycle, focal adhesion, calcium signaling, and Wnt signaling. We also applied the same analysis to experimentally-derived B cell-specific gene signatures (Shaffer, Immunity 2001; Herishanu, Blood 2011), demonstrating significant inhibition of both BCR and NF-κB signaling pathways. We derived signature scores by averaging the expression of the signature-specific genes and estimated the degree of pathway inhibition from the ratio of signature averages in treated versus baseline samples. Of the 11 significantly altered pathways, all demonstrated >50% reduction in signature scores, ranging from a 71% reduction for BCR to 54% for p53. The degree of inhibition of the signaling pathways was independent of biologic and clinical features, including IGVH mutation status, presence or absence of del(17p), and prior treatment status.

In summary, ibrutinib induces widespread changes in gene expression in CLL cells in vivo reflecting strong inhibitory effects across multiple pathways that suggest a broader impact on tumor biology than expected from pure loss of BTK function. The surprising extent of changes across multiple diverse pathways could reflect inhibition of tumor-microenvironment crosstalk and consequent changes in the tumor milieu. This notion is consistent with the more profound inhibition observed with longer treatment duration. Equally notable, we found no distinct biologic signature(s) upregulated in response to ibrutinib which may indicate the absence of a cellular stress response and is consistent with the apparent resting state of the residual tumor and the observed slow attrition of cells over time.

This research was supported by the Intramural Research Program of the NHLBI.