TFH lymphomas: the times they aza-changin’?

In this issue of Blood, O’Connor and colleagues¹  report a phase 1 study of romidepsin combined with oral azacytidine in patients with relapsed/refractory (R/R) lymphomas, including complete remissions among 3 patients with angioimmunoblastic T-cell lymphoma (AITL) and single patients with multiple other lymphoma subsets.

Three laments begin nearly every manuscript, grant application, and lecture on peripheral T-cell lymphomas (PTCLs): (1) these diseases are poorly understood, (2) they confer a dismal prognosis, and (3) targeted therapies are desperately needed. To wit, the PTCL subtype with highest incidence (among >30 subtypes) is NOS, an acronym for “Not Otherwise Specified” that may as well stand for “No Obvious Solution.” Between 75% and 90% of patients with common PTCL subtypes historically succumbed to their disease.²  Cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) remains the first-line treatment of most PTCL because there are no randomized trials showing an advantage for anything else.³  Agents approved in the United States for the treatment of R/R PTCL include histone deacetylase inhibitors and the antifolate pralatrexate. These drugs induce responses in 25% to 30% of patients with median duration of only 4 months.

One approach for improving outcomes in patients with PTCL is to develop subset-specific therapies that either target druggable vulnerabilities or build on empiric observations. For example, the anti-CD30 antibody-drug conjugate (ADC) brentuximab is highly active in patients with anaplastic large-cell lymphoma (ALCL), a subset of PTCL that expresses CD30 on the surface of all lymphoma cells. Approximately two-thirds of patients with untreated ALCL who received CHP (CHOP without the vincristine) plus brentuximab in the ECHELON-II study had >2-year disease-free survival.⁴  The large majority of those patients are likely to be cured. A subset of patients with other subtypes of PTCL who highly express CD30 may also benefit from CHP-brentuximab.⁴ 

An important lesson of ECHELON-II is that regimens of cytotoxic agents (including the ADCs that deliver them) tailored to disease biology can cure more patients with PTCL. Unfortunately, most patients with PTCL do not have the ALCL subtype and are still more likely than not to die of their lymphomas. Thus, other subtype-specific therapies are desperately needed. A previous study reported an overall response rate of 75% and a complete response rate of 41% among 12 patients with AITL treated with the DNA methyltransferase inhibitor azacytidine.⁵  AITL is a subset of PTCL with features of follicular helper-T (TFH) cells. TFH-like lymphomas frequently harbor mutations in TET2, DNMT3A, and/or IDH2 that can affect DNA methylation.⁶ 

Building on the single-agent experience with azacytidine, O’Connor and colleagues performed a phase 1 study of romidepsin added to oral azacytidine among patients with R/R lymphomas. They escalated using a standard 3+3 design until they reached a maximum tolerated dose (MTD) associated with significant nausea, vomiting, and cytopenias. Eight (73%) of 11 patients with T-cell lymphomas responded, with complete responses in all 3 patients with AITL as well as single patients with adult T-cell leukemia/lymphoma, T-lymphoblastic lymphoma, and extranodal NK/T-cell lymphoma. Of note, only 1 patient with PTCL-NOS was enrolled.

Is the addition of romidepsin to azacytidine an advance for patients with AITL and other TFH-like lymphomas? Only time will tell. Other doublets for R/R PTCL have also achieved responses in 60% to 80% of patients with PTCL but, like the report by O’Connor and colleagues, the studies were small, typically phase 1, and enrolled highly selected patients.^7,8  Caveat emptor applies; the real-world activity of these doublets is likely to be less impressive.

The data from phase 2 studies testing doublets in patients with R/R PTCL, including a trial testing oral azacytidine plus romidepsin, will likely be available in the next 1 to 2 years. Unfortunately, PTCL studies are often too small and lack companion biomarkers to guide patient selection. As a field, we need bigger, smarter, and more comprehensive studies but funding, specimen acquisition, and cat herding remain formidable challenges. Patients with multiple lymphoma subtypes had complete responses to azacytidine plus romidepsin, but there are no biomarkers to identify those patients. Those responses, which spanned a broad diversity of lymphoma genetics and clinical behavior, also challenge any dogma relating mutations in genes affecting DNA methylation (eg, TET2) to response. O’Connor et al attempted methylation profiling of nonmalignant blood cells and sequenced a handful of patient tumors, but those studies were highly underpowered to give definitive results.

One final and possibly picayune point remains. There is no evidence that romidepsin and azacytidine (or any other combinations) act synergistically in patients with PTCL. Synergy is a mathematical term denoted by “the effect of drug A combined with drug B is greater than a mathematical prediction of additivity.” A quantitative analysis of hundreds of clinical and murine therapeutic studies demonstrated that almost no drug combinations given for any kind of cancer, including targeted agents, chemotherapies, and immunotherapies, achieve synergy in vivo.⁹  Instead, treating patients with multiple drugs simply gives those patients multiple opportunities to respond to one of the drugs. In other words, patients get multiple shots on goal. This does not mean that in vitro studies demonstrating synergy at very specific doses, as shown in cell lines for azacytidine and romidepsin,¹⁰  have incorrect data. Instead, it means that the doses given to patients (typically the MTD) should not be assumed to retain those synergistic effects in vivo. Combining at MTD may still offer the greatest benefit to patients, but it compels us as a field to stop claiming that the drugs achieve synergy. In so doing, we can rescue that word from a trash heap of meaningless modifiers (see “marked,” “notable,” “robust,” and “profound”) that lessen our clinical literature.