We sought to determine the mechanism of an exceptional response in a patient diagnosed with a SMARCB1/INI1-negative chordoma treated with tazemetostat, an EZH2 inhibitor, and followed by radiotherapy.
Patient and Methods: In an attempt to investigate the mechanism behind this apparent abscopal effect, we interrogated tumor tissues obtained over the clinical course. We utilized next-generation sequencing, standard IHC, and employed a novel methodology of multiplex immunofluorescence analysis.
We report an exceptional and durable response (2+ years) in a patient with SMARCB1-deleted, metastatic, poorly differentiated chordoma, a lethal disease with an overall survival of 6 months. The patient was treated for 4 weeks with tazemetostat, an EZH2 inhibitor, in a phase II clinical trial. At the time of progression she underwent radiation to the primary site and unexpectedly had a complete response at distant metastatic sites. We evaluated baseline and on-treatment tumor biopsies and demonstrate that tazemetostat resulted in pharmacodynamic inhibition of EZH2 as seen by decrease in histone trimethylation at H3K27. Tazemetostat resulted in a significant increase in intratumoral and stromal infiltration by proliferative (high Ki-67), CD8+ T cells, FoxP3+ regulatory T cells, and immune cells expressing checkpoint regulators PD-1 and LAG-3. These changes were pronounced in the stroma.
These observations are the first demonstration in patient samples confirming that EZH2 inhibition can promote a sustained antitumor response that ultimately leads to T-cell exhaustion and checkpoint activation. This suggests that targeted alteration of the epigenetic landscape may sensitize some tumors to checkpoint inhibitors.
Chordoma is a rare cancer of the spine with an annual incidence of approximately 300 new cases in the United States. It can occur at any age and presents as a localized mass causing pain and neurologic symptoms. For conventional chordoma (>95%), surgery and/or radiotherapy are considered standard of care; however, 10-year disease-specific survival is approximately 48% (1). In contrast, poorly differentiated chordoma (PDC; ref. 2) is highly aggressive, often metastatic, and uniformly fatal with a median overall survival of 9 months (3–5). Unlike conventional chordoma, PDC can be sensitive to cytotoxic chemotherapies; however, the responses are short-lived.
SMARCB1 (INI1) is a critical component of the mammalian SWItch/Sucrose Non-Fermentable (mSWI/SNF) protein complex that is essential for chromatin remodeling and in some cellular contexts can function as a tumor suppressor or oncogene. TCGA analysis shows that approximately 20% of all cancers harbor inactivating mutations in the mSWI/SNF complex (6). Loss of tumor suppression through SMARCB1/INI1 inactivation has been described as a key genetic event in various tumor types (7), including malignant rhabdoid tumors and epithelioid sarcoma. Therefore, targeting cancers with dysfunctional mSWI/SNF complexes is an area of active interest.
Loss of the SMARCB1/INI1 subunit destabilizes the mSWI/SNF complex, resulting in unopposed oncogenic activity of Enhancer of Zeste homolog 2 (EZH2; refs. 8, 9), a histone methyltransferase. EZH2 functions as an epigenetic regulator of gene expression by catalyzing the generation of trimethylated H3K27 (H3K27me3), which results in the repression of broad transcriptional gene sets thereby regulating cellular differentiation programs. Inhibition of EZH2 confers synthetic lethality in SMARCB1/INI1-deficient tumors. In a phase I study of tazemetostat, an oral inhibitor of EZH2, complete and partial responses (RECIST 1.1) were seen in patients with SMARCB1/INI1-negative tumors (10).
In preclinical studies, EZH2 has been implicated in affecting the tumor microenvironment through several mechanisms involving multiple immune cell types (11, 12). In this article, we sought to determine the mechanism of an exceptional and durable response in a patient diagnosed with a SMARCB1/INI1-negative PDC, treated with tazemetostat and sequential radiotherapy. Here, we demonstrate the first comprehensive in-patient evidence for a potential link between inhibition of EZH2 and its impact on tumor immunology.
A 25-year-old female was presented with lower back pain, bowel and bladder dysfunction, and lower extremity paresthesia. Spinal MRI scan showed a 6.4 cm sacral mass and staging studies showed locoregional spread and lung metastasis (Fig. 1A). A biopsy of the primary sacral mass confirmed PDC with SMARCB1/INI1 loss by IHC (labeled: “baseline, sacrum”; Fig. 2A–C). Next-generation sequencing of the tumor confirmed an intragenic deletion of exons 5–7, loss of heterozygosity of the SMARCB1 on chromosome 22, and very low tumor mutation burden.
The patient signed the informed consent form (MSKCC IRB#15-328) to a phase II study of tazemetostat in SMARCB1/INI1-negative tumors (Clinicaltrials.gov NCT02601937) and additionally consented to optional tumor biopsies. The trial was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines, and federal and local policy on bioethics and human biologic specimens. At enrollment, imaging scans showed rapid growth in the sacral primary and metastatic pulmonary nodules (Fig. 1B). After 4 weeks of tazemetostat treatment (800 mg, orally, twice daily) the patient noted worsening symptoms and withdrew from study, although imaging revealed stable disease by RECIST 1.1 (+13%; Fig. 1C). An on-study biopsy of the sacral mass (labeled: “on-tazemetostat, sacrum”) was performed. The patient underwent radiation to the primary sacral mass (7,000 cGy over 35 fractions), which resulted in a partial response in the radiated sacrum and a complete response in the distant, metastatic lung nodules (nonradiated; Fig. 1D). A durable complete response was observed for 4 months when follow-up imaging revealed bilateral recurrent disease in the lung (Fig. 1E). The largest metastatic nodule was incompletely removed (labeled: “tazemetostat → sacral RT, lung). Given the initial abscopal effect observed on tazemetostat and sequential radiotherapy, the patient was treated with off-label nivolumab, a PD-1 inhibitor. Following six cycles of nivolumab, all lung nodules disappeared; however, the primary sacral (postradiated) chordoma remained resistant (Fig. 1F). The patient subsequently received additional radiation to the sacrum (3,000 cGy, 5 fractions) with tumor shrinkage. Following a second radiation, the CTLA-4 inhibitor ipilimumab (1 mg/kg every 21 days × 4 doses) was added to nivolumab; however, the primary sacral site remained unresponsive to immunotherapy while the metastatic lung lesions are in complete remission for 2+ years.
In an attempt to investigate the mechanism behind this apparent abscopal effect, we interrogated tumor tissues obtained over the clinical course. We utilized standard IHC and employed a novel methodology of multiplex immunofluorescence analysis.
Modulation of the tumor immune microenvironment by sequential EZH2 inhibition and radiotherapy
The treatment-naïve sacral PDC tumor displayed strong, diffuse H3K27me3 IHC staining (Fig. 2D), and treatment with tazemetostat demonstrated a dramatic decrease in H3K27me3 staining, pharmacodynamic evidence of EZH2 inhibition confirming the molecular action of tazemetostat (Fig. 2E). While there was minimal immune cell infiltration of the primary untreated tumor, tazemetostat-treated tumor showed an increase in intratumoral and stromal infiltration by CD8+ CTLs, FoxP3+ regulatory T cells (Tregs), and immune cells expressing checkpoint regulators PD-1 and LAG-3 (Figs. 2F and G and 3A and B). Tazemetostat induced increases in CD8+ T-cell subtypes with singular or combined expression of T-cell exhaustion markers PD-1 and LAG-3, changes that were more pronounced in the stroma as might be expected for T cells showing expression of so-called “exhaustion” markers with some impairment of infiltration into tumor (Fig. 3B).
Prior to tazemetostat treatment, virtually all of tumor-infiltrating CD8+ CTLs were nonproliferative (>98% Ki67-negative). The number and percentage of Ki-67–positive CD8+ CTLs increased significantly in both the intratumoral and stromal compartments with tazemetostat therapy (Figs. 2F and G and 3C). A significant percentage of selected CD8+ CTL subtypes demonstrated staining of the proliferation marker Ki-67, most notably LAG-3+ stromal CTLs (Fig. 3C). The increased number of T cells and shift toward more proliferative T lymphocytes with concomitant expression of checkpoint markers indicate a shift in the immune status of the tumor upon tazemetostat treatment consistent with activation of an antitumor immune response and emergence of CD8+ T-cell exhaustion. These observations are the first demonstration in patient samples confirming that EZH2 inhibition can promote a sustained antitumor response that ultimately leads to T-cell exhaustion defined by expression of inhibitory checkpoint expression.
Examination of the largest, recurrent lung lesion (tazemetostat → sacral RT, lung) that emerged after the abscopal effect seen with sequential tazemetostat and radiation therapy revealed the presence of infiltrating immune cells, including CD8+ CTLs, FoxP3+ Tregs, and PD-1+, PD-L1+, and LAG-3+ immune cells (Fig. 2H and I). Increased intratumoral density of CD8+ T-cell subtypes compared with the primary sacral PDC tumor was detected but stromal infiltration was much more robust (Fig. 3A and B). As seen with tazemetostat treatment, elevated numbers of proliferative stromal CTLs were observed compared with the primary PDC (Fig. 3C). Multiple intratumoral and stromal CD8+ CTL subpopulations were Ki-67+ following radiotherapy, and stromal PD-1+ CTL subtypes with proliferative potential were generally the most abundant (Fig. 3C). It is not possible to ascertain from these data what contribution each therapy made to the systemic immune response that resulted in T-cell infiltration of the metastatic lung tumors. More studies are needed to determine the individual and combined effect of tazemetostat and radiotherapy to this phenomenon. Interestingly, the presence of PD-L1+ immune cells (Fig. 2I) in the resected lung nodule indicates immune checkpoint engagement and dampening of the antitumor immune response, which may have contributed to recurrence of some lung lesions following radiotherapy. Furthermore, the presence of proliferative (Ki-67+) CD8+ CTL subtypes at the metastatic tumor site may explain the impressive responses of the lung lesions to single-agent nivolumab.
In this article, we provide evidence that EZH2 blockade with tazemetostat can induce an antitumor immune response by promoting tumor infiltration by various T-cell populations, further demonstrating the potential impact of epigenetic regulation of immunologic responses in cancer. The patient described in this report is now alive at 27+ months following diagnosis and continues to have a complete response in the metastatic sites. This represents an exceptional response in this otherwise lethal disease with a median overall survival of 9 months.
EZH2 and its role in the tumor microenvironment
EZH2 function impairs antigen presentation through inhibiting MHC class I and II gene expression (13, 14), interferes with expression of cytokines, their receptors, and the IFNγ transcriptional program (15), promotes maintenance of Treg cell identity and immunosuppressive function (11, 12, 16), attenuates trafficking of effector CD8+ T cells through repression of chemokines CXCL9 and CXCL10, and impairs maturation of natural killer cells (17). Preclinically, increased EZH2 is implicated in resistance to anti-CTLA4 immunotherapy through diminished antigen presentation, which is subsequently overcome by EZH2 inhibition leading to PD-L1 downregulation and infiltration of IFNγ-producing PD-1low CD8+ T cells (18).
Tazemetostat treatment appears to have induced an antitumor response in the primary PDC based on modulation of the tumor immune microenvironment in multiple ways. In our study, we observed substantial tumor infiltration by various lymphocytes, including proliferative CD8+ T cells, following EZH2 inhibition with tazemetostat alone. On the basis of many published observations that increased intratumoral infiltrating lymphocyte density, most notably CD8+ cells, correlates with improved survival in many tumors, we quantitatively reported intratumoral and stromal densities for the multiplex panel marker expression providing spatial contexture to CD8 cell infiltration and coexpression. Tazemetostat may have induced lymphocyte infiltration of the PCD through reported effects of EZH2 disruption on reprogramming of tumor infiltrating Tregs to stimulate T-cell infiltration (11, 12, 17) or by upregulating chemokine and cytokine expression (15, 19). Further evidence of an antitumor immune response in the primary PDC induced by tazemetostat is apparent from the detection of T-cell exhaustion, as evidenced by the expression of inhibitory immune checkpoint receptors PD-1 and LAG-3 either singularly or in combination on CD8+ T cells found both intratumorally and within the stroma (Fig. 3B). Induced expression of these exhaustion markers is indicative of prolonged exposure of T cells to their cognate antigen that is associated with decreased proliferation and loss of T-cell effector functions after an initial immune response (20). Generally coupregulation of multiple inhibitory receptors indicates a higher degree of T-cell dysfunction (21). Nonetheless, further studies are needed to investigate the involvement of these mechanisms in the phenotypic effects of EZH2 inhibition in the immune microenvironment of human tumors.
The primary, heavily radiated, recurrent sacral mass was refractory to immune checkpoint blockade (ICB) administered well after tazemetostat therapy. We hypothesize that increased levels of exhausted intratumoral and stromal CD8+ CTLs, as evidenced by PD-1 and LAG-3 expression on these cells in the primary PDC (Fig. 3B), may explain the attenuated response of the primary tumor to tazemetostat despite the presence of proliferation-competent CD8+ CTLs in the tumor. Concurrent treatment of the primary PDC with tazemetostat and an immune checkpoint inhibitor such as nivolumab may have been required to reverse T-cell exhaustion and reinvigorate infiltrating T cells to effect tumor shrinkage. However, the dual expression of PD-1 and LAG-3 on exhausted T cells in the sacral PDC may have precluded an objective response to combination tazemetostat/ICB treatment. In addition, gene expression profiling of PDC tumors from patients treated with tazemetostat, radiotherapy, or ICBs to investigate the molecular profiles of T-cell exhaustion may provide future insights into sensitivity and resistance to these agents. Finally, we hypothesize that radiotherapy of the sacrum likely depleted the TILs induced by tazemetostat at the primary site, thus making the sacral mass refractory to immunotherapy.
The complete response of the metastatic lung tumors following radiotherapy of the primary PDC conforms to the classical definition of the abscopal effect of radiotherapy that results in tumor regression at nonirradiated, distant sites. This phenomenon is extremely rare. The precise mechanisms underlying abscopal responses remain unclear, but emerging evidence suggests systemic activation of the immune system is the basis for this response (22). It is thought that radiotherapy induces immunogenic cell death, characterized by release of danger-associated molecular pattern antigens that then trigger antigen uptake and activation of antigen-presenting cells leading to subsequent priming of CTLs and an adaptive immune response.
Tazemetostat clearly stimulated an antitumor immune response within the primary PDC, and robust immune infiltration of the distant lung tumor was detected following radiotherapy. Nonetheless, it is difficult to discern the exact contribution of EZH2 inhibition, radiotherapy, and PD-1 blockade on the immunologic and overall responses of the patient. To further discern the contribution of each of these therapies, it will be important to study sequential biopsies from both primary and recurrent sacral and metastatic tumors, as well as peripheral blood samples to evaluate systemic immunologic changes with each intervention in future clinical studies. These additional samples will also help elucidate the role of tumor mutation burden/neoantigen load, T-cell receptor repertoire, and expansion of specific T-cell clones in immunologic and therapeutic responses to EZH2 inhibition and radiotherapy.
In conclusion, we convincingly demonstrate for the first time in a patient that EZH2 inhibition with tazemetostat induces an immune response within the tumor microenvironment. Following tazemetostat treatment, sequential radiotherapy and checkpoint blockade led to an exceptional response in this patient. Further studies are warranted to elucidate the contribution of each therapy to TME remodeling and therapeutic response. Importantly, the favorable tolerability and safety of tazemetostat makes it a promising drug to conduct investigational combination studies with radiation and/or checkpoint inhibitors.
Disclosure of Potential Conflicts of Interest
M.M. Gounder reports receiving speakers bureau honoraria from and is a consultant/advisory board member for Epizyme, Daiichi Sankyo Inc., Springworks, Bayer, Amgen, Karyopharm, and TRACON. E. Lis holds ownership interest (including patents) in Medtronic. N.R. Michaud holds ownership interest (including patents) in AstraZeneca. No potential conflicts of interest were disclosed by the other authors.
Conception and design: M.M. Gounder, S.J. Blakemore, M. Hameed, T.J. Hollmann
Development of methodology: T.J. Hollmann
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.M. Gounder, G. Zhu, E. Lis, T.J. Hollmann
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.M. Gounder, G. Zhu, L. Roshal, E. Lis, S.J. Blakemore, N.R. Michaud, T.J. Hollmann
Writing, review, and/or revision of the manuscript: M.M. Gounder, G. Zhu, E. Lis, S.R. Daigle, S.J. Blakemore, N.R. Michaud, M. Hameed, T.J. Hollmann
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.M. Gounder, G. Zhu, T.J. Hollmann
Study supervision: M.M. Gounder, T.J. Hollmann
The clinical trial was funded by Epizyme. The work was supported by a Memorial Sloan Kettering Cancer Center core grant (P30 CA008748) and The Draper Family Fund.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.