March 15, 2015
Epizyme Inc. is focused on developing drugs for histone methyltransferases (HMT – also called ‘Chromatin Writers,’ written about previously here). Their area of focus is cancer, where many HMTs have a defined genetic linkage — somatic mutation, amplification, synthetic lethality, etc — to the disease. The company has two clinical candidates – EPZ-5676 and EPZ-6438. Epizyme has inked therapeutic collaborations with Celgene, Eisai, and GlaxoSmithKline (GSK), which have provided attractive up-front economic benefits ($184 million in non-dilutive funding to date), while retaining significant ownership of drug candidates. These pharma deals have the potential to provide additional liquidity with another $1 billion in potential future development milestones. Epizyme sees itself as a fully-integrated platform company, and there are early signs to believe. At the annual 2014 ASH meeting Epizyme showcased the potential of its platform engine by the nomination of another clinical candidate – a PRMT5 inhibitor (see EPZM ASH presentation here), to be developed in collaboration with GSK for the treatment of Mantle Cell Lymphoma (MCL). Most recently, Epizyme reacquired the worldwide rights (ex-Japan) to EPZ-6438 from Eisai, signaling its confidence in this asset and commitment to a go-it-alone/fully-integrated strategy. Considering that human genome encodes 96 HMTs, and pre-clinical evidence implicates approximately 20 of them in cancer, this could be the tip of the drugging-HMTs-in-cancer iceberg. At this moment, Epizyme is strategically positioned as the leader in this niche of cancer epigenetics, with leading technical know-how in the discovery process. Epizyme patients and shareholders are poised to benefit if drugging HMTs proves to be a sound strategy in oncology.
EPZ-6438 (E7438) – A First-In-Class EZH2 Inhibitor
I touched briefly on the science behind drugging EZH2 in a previous post (get up to speed here). When this program began, the clinical line-of-sight was aimed squarely at EZH2-mutated Non-hodgkin Lymphoma (NHL). The pre-clinical understanding had demonstrated somatic EZH2 mutations in this disease could give rise to an oncogenic (and neomorphic) enzyme function. Drugging the mutant enzyme appeared to be the perfectly rational, targeted approach to bring into the clinic. Fast forward to today, and the dose-finding study for EPZ-6483 has yielded objective responses (including a CR) in 4/10 evaluable NHL patients (see Figure 1. and the EPZM EORTC-NCI-AACR meeting slides here). The overall safety and tolerability of this agent appear quite good with only one treatment-related grade 3/4 event in 24 patients. Here’s the shocker: none of the responders had tumors bearing a somatic EZH2 mutation, and there are responders in both GCB and non-GCB subtypes. So far Epizyme hasn’t recruited any EZH2-mutant NHL patients to the trial. Based on pre-clinical data, it’s possible that EZH2-mutated tumors will respond better than their wildtype counterparts. Interestingly, most analyst models assume the market for this drug to be a patient population of approximately 12,000 EZH2-mutant relapsed/refractory patients. But this phase I dataset demonstrates efficacy EZH2 wildtype setting – if we account for this when modeling the market potential of this drug, the addressable relapsed/refractory NHL population grows to approximately 70,000 patients – a population nearly six times bigger than the original line-of-sight.
What’s more, Epizyme has already demonstrated initial signs of EPZ-6438 activity in solid tumors. Notably, in the phase I study, one patient with a rare cancer called Malignant Rhabdoid Tumor (MRT) exhibited a partial response and remains on drug (Figure 2). This particular tumor type is characterized by loss-of-function germline INI1 (SMARCB1) mutations, occurring in approximately one third of patients. In this setting, SMARCB1 acts as a tumor suppressor and SMARCB1 loss is thought to be a driving oncogenic event in these patients, resulting in deregulation of canonical cancer-related pathways (see a recent review by Dr. Charles Roberts here).
It would be easy to overlook the Epizyme MRT data. This is a rare tumor, and it may not add significantly to the market potential of EPZ-6438. However, by digging into the mechanism of action for EPZ-6438 in MRT we can uncover additional upside for this agent in much larger solid tumor populations.
The SWI/SNF Chromatin Remodeling Complex
It turns out SMARCB1 is a subunit of a large multi-protein complex, known as SWItch/Sucrose NonFermentable, or SWI/SNF. SWI/SNF is an evolutionarily conserved ATP-dependent nucleosome remodeling complex that contains 12-15 individual subunits. SWI/SNF complexes can modulate gene expression through interaction with transcription factors, co-activators/repressors, and functions in nucleosome remodeling at promoter/enhancer regions. SWI/SNF has been shown to be critical for orchestrating critical cell-identity gene programs in development and disease. Specificity of these gene expression programs can be achieved through combinatorial assembly of SWI/SNF variant subunits, and their relative affinities for various transcription factors.
While mutations in SMARCB1 were first identified in MRT patients in the late 1990’s, more recent cancer genome sequencing efforts have uncovered the cross-cancer prevalence of SWI/SNF subunit mutations. To date, nine of the SWI/SNF subunits have been found to be recurrently mutated across almost all cancer types. Astoundingly, these mutations occur with a frequency of approximately 20% (a number on par with p53 mutations) across all cancers. A quick query of cBioPortal puts this kind of mutational frequency in perspective (Figure 4).
SWI/SNF-Mutated Cancer as an Opportunity for EPZ-6438 – Beyond Malignant Rhabdoid Tumor
With a 20% cross-cancer mutation frequency, SWI/SNF-mutated tumors represent an enormous potential market opportunity for EPZ-6438. Indulging in a quick back-of-the-napkin calculation using SEER NCI Cancer Statistics gives us 1.48 million SWI/SNF-mutated malignant cancer patients in the US (7.4 million malignant US patients x 20% mutation rate). The economic implications are enormous if even a fraction of these patients stand to benefit from EPZ-6438. Besides the rare Malignant Rhabdoid Tumor, there are no clinical data demonstrating efficacy using EZH2 inhibitors for SWI/SNF-mutated cancers. However, some compelling pre-clinical work has been done that demonstrates this could be a viable clinical approach for additional SWI/SNF-mutated tumors.
For example, a recently published manuscript in Nature Medicine demonstrates synthetic lethality using EZH2 inhibitors in ARID1A-mutated cancers (see here). The authors use Ovarian Clear Cell Carcinoma (OCC) as a model since 50% of these cancers contain somatic ARID1A mutations. In this tumor setting, response to EZH2 inhibitors appears to be driven by the mutational status of ARID1A. Mechanistically, ARID1A-mutant OCC cells upregulate PIK3IP1 in response to EZH2 inhibition, resulting in loss of PI3K-AKT signaling and subsequent cell death. In the US there are approximately 20,000 Ovarian Clear Cell/Endometroid ARID1A-mutant patients. This patient population is in need of new therapeutic options and could stand to benefit from EZH2 inhibitors.
A similar antagonism between EZH2 and SWI/SNF may exist in other SWI/SNF-mutated cancer types (see the simplified model in Figure 5), and thus may represent additional therapeutic opportunities for EZH2 inhibitors. If borne out with eventual clinical data this key concept could unlock significant value for Epizyme patients and shareholders.
The phase I dataset for EPZ-6438 demonstrates a clear efficacy signal in EZH2 wildtype NHL patients, thus providing a six-fold increase in the addressable B-cell Lymphoma patient population. Outside of B-cell malignancies, EPZ-6438 appears to have expanded utility in solid tumors, particularly in tumors bearing SWI/SNF mutations. An objective response in 1/1 SMARCB1-deficient Malignant Rhabdoid Tumor (MRT) patients in the phase I study provides clinical evidence in support of this hypothesis. Frequent cross-cancer SWI/SNF alterations represent attractive, currently unrecognized opportunies for EPZ-6438. Recent pre-clinical evidence expands the potential utility of EZH2-targeted compounds to ARID1A-mutant ovarian cancer. In our view, the ARID1A-mutant example could represent one of many additional patient populations amenable to EZH2 inhibition. Given the current evidence, we believe this drug will have both single-agent and combination utility across additional indications beyond B-cell malignancies.
Disclaimer: All opinions expressed on Oncology Discovery are my own and do not necessarily represent the position of my employer. The information presented within this article is not a solicitation for investment.
Copyright © 2015 Oncology Discovery. All Rights Reserved. Unauthorized use and/or duplication of this material without permission is strictly prohibited.