To identify synergistic drug combinations, we work with Craig Thomas at NCATS on matrix drug screens, which use acoustic dispensing robots to array two drugs in a dose titration series against one another, looking for doses at which the drugs achieve greater cell killing together than individually. We use a chemical library of 2000 drugs that are either approved or in development as cancer therapeutics plus small molecules that serve as tool compounds for important signaling and regulatory pathways, such as NF-kB. In the initial screen using ABC DLBCL lines, we searched for compounds that would synergize or antagonize the toxicity of ibrutinib. Multiple agents targeting the PI(3) kinase pathway strongly synergized with ibrutinib, including drugs targeting the PI(3) kinase catalytic subunit, Akt, or mTORC1. This result is consistent with the notion that ibrutinib primarily targets the pro-survival NF-kB pathway, while having less effect on the PI(3) kinase survival pathway, which is also engaged by BCR signaling. Ibrutinib synergized strongly with PRT-060318, an inhibitor of the SYK tyrosine kinase that is activated proximally in the BCR pathway, and with ABT-199, a BCL2 inhibitor that we have studied extensively in the context of ibrutinib resistance. Another broad class of synergistic compounds included cancer chemotherapeutic agents that elicit a DNA damage response or trigger apoptosis by interfering with microtubules. The reasons behind this synergism is likely due to the ability of NF-kB to antagonize the apoptotic effects of chemotherapy. Indeed, all components of both the CHOP and EPOCH chemotherapy regimens synergized with ibrutinib in killing ABC DLBCL cells, providing impetus to combine ibrutinib with these regimens. Based on our identification of chronic active BCR signaling as a key survival pathway in ABC DLBCL, we conducted a phase I/II clinical trial of ibrutinib using gene expression profiling to assign patients to the ABC and GCB subtypes of DLBCL. As predicted by our laboratory investigations, ibrutinib produced a 37% response rate in ABC DLBCL patients but only a 5% response rate in GCB DLBCL, demonstrating that the molecular diagnosis of DLBCL subtypes can inform precision medicine trials. This translated into improved overall survival in ABC relative to GCB DLBCL, including several patients who have remained in complete remission for more than 3-6 years, taking ibrutinib daily without discernable side effects. Based on these promising results, ibrutinib plus chemotherapy is now being evaluated in untreated non-GCB DLBCL patients in a phase 3 randomized trial being conducted by Jannsen. This trial used an immunohistochemical test we developed with our LLMPP colleagues to identify patients with non-GCB DLBCL and is the first phase 3 trial in DLBCL to use molecular profiling for enrollment. This trial has enrolled 800 patients and will read out in 2018. To understand the molecular basis for response to ibrutinib within ABC DLBCL, we resequenced the tumors for recurrent oncogenic mutations. Tumors with CD79B mutations responded more frequently than those with wild type CD79B (55% vs 30%), demonstrating the clinical validity of our observation that these mutations augment BCR signaling. Nonetheless, the majority of responding patients on this trial had wild type CD79B, which is in keeping with our demonstration that chronic active BCR signaling is driven by self-antigen reactivity of the immunoglobulin variable regions. Notably, tumors that had both a CD79B mutation and a MYD88 L265P mutation responded frequently (80%), whereas those with only a MYD88 mutation did not respond. This double mutant genotype occurred in more ABC tumors (11%) than expected by chance based on the prevalence of each mutation individually, providing genetic evidence that the MYD88 and BCR pathways cooperate in these tumors. We showed that ABC cell lines with the double mutant genotype responded to ibrutinib, whereas ABC lines with only MYD88 mutations did not, and that a MYD88 dimerization inhibitor decreased proximal BCR signaling. Together, these observations suggested that the double mutant genotype creates strong addiction to BCR signaling that is hyper-responsive to ibrutinib. Given the frequent responses to ibrutinib in ABC DLBCL tumors with both CD79B and MYD88 L265P mutations, we searched for other lymphoma types that have this double mutant genotype. Recent work has demonstrated that MYD88 L265P mutations are enriched in several types of extranodal lymphoma, including primary breast lymphoma, primary testicular lymphoma, and primary central nervous system lymphoma (PCNSL). Interestingly, the co-occurrence of CD79B and MYD88 L265P mutations in PCNSL is 2-3 times greater than in nodal ABC DLBCL, suggesting that they may be hyper-addicted to BCR signaling due to synergy between the BCR and MYD88 pathways. To explore this hypothesis, we worked with the Lymphoid Malignancies Branch clinical team led by Wyndham Wilson to devise an ibrutinib-based regimen for PCNSL. The basic design of this regimen entailed giving ibrutinib as monotherapy for 2 weeks, followed by the combination of ibrutinib with a set of brain-penetrant chemotherapy agents, given in cycles. By comparing pre-treatment MRI scans with scans immediately after ibrutinib monotherapy, we observed objective responses to ibrutinib in 17/18 treated PCNSL patients, the majority of whom had relapsed/refractory disease. The high rate of response to ibrutinib monotherapy in this trial supports our hypothesis that PCNSL is hyperaddicted to BCR signaling. Although we only had biopsy material available on 4 patients, it was notable that responses were seen in one patient with the double mutant genotype, but also in 2 patients with only a CD79B mutation and 1 with only a MYD88 L265P mutation. Hence, the enrichment in PCNSL for the double mutant genotype was a genetic clue that PCNSL as a whole is typically hyper-addicted to BCR-dependent NF-kB activation. The partial responses to ibrutinib monotherapy in PCNSL were converted into complete responses (CRs) with added chemotherapy in 86% of patients. Notably, 8 of these patients have ongoing CRs with a predicted progression-free survival of 15.5 months, including 5 patients with disease that was refractory to the last therapy. We therefore believe that this chemotherapy platform, which we term TEDDI-R, can serve as the foundation for further improvements in subsequent studies. It is clear that monotherapy with targeted agents will not typically yield durable remissions in DLBCL, so we are undertaking clinical trials to explore combinations of targeted agents. For example, we are initiating a multi-center precision medicine trial in relapsed/refractory DLBCL, which will use Nanostring-based gene expression profiling and genetic analysis by Foundation Medicine to create molecularly-defined baskets that will receive therapeutic combinations appropriate for the molecular features. This MasterLymph protocol will incorporate comprehensive genomic profiling of all tumors using high-throughput sequencing of the exome transcriptome and whole genome. We will use these data in the light of therapeutic response or resistance to iterate the molecular baskets in subsequent phase of this master protocol.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Investigator-Initiated Intramural Research Projects (ZIA)
Project #
Application #
Study Section
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Basic Sciences
Zip Code
Roschewski, Mark; Staudt, Louis M; Wilson, Wyndham H (2016) Dynamic monitoring of circulating tumor DNA in non-Hodgkin lymphoma. Blood 127:3127-32