Treatment with 5-azacytidine induces a sustained response in patients with angioimmunoblastic T-cell lymphoma

TO THE EDITOR:

Peripheral T-cell lymphomas (PTCLs) are heterogeneous diseases resulting from the malignant transformation of mature T or natural killer cells. Their epidemiology varies widely, but PTCLs that derive from T follicular helper (TFH) cells, which include angioimmunoblastic T-cell lymphoma (AITL), follicular PTCL, and other nodal PTCLs with a TFH phenotype,¹  appear to be frequent among PTCLs.^2,3  Patients have a poor prognosis, especially after relapse, with a median overall survival (OS) of ∼6 months.⁴  The US Food and Drug Administration has approved pralatrexate, romidepsin, and belinostat for relapsing/refractory PTCL, but these 3 drugs still show limited activity in PTCL, with an overall response rate of ∼30%.^5-7  Thus, PTCL therapy is still an unmet medical need.

Recurrent mutations in TET2,^8-10 DNMT3A,^9-11  and/or IDH2^9,10,12  are detected in ∼80%, 25%, and 25% of patients with TFH-derived PTCL, respectively. These 3 genes directly or indirectly regulate cytosine methylation and hydroxymethylation, and mutations of these genes result in changes in DNA methylation levels.¹³ TET2, DNMT3A, and IDH1/2 are also mutated in myeloid neoplasms, especially acute myeloid leukemia and myelodysplastic syndromes.¹³  Treatment with the hypomethylating agents (HMAs) 5-azacytidine and decitabine shows efficacy in these diseases, and the response rate to HMAs appears to correlate with TET2, IDH1/2, and/or DNMT3A mutations.^14-16  This suggests that HMAs could have activity against TFH-derived PTCL. We previously reported 2 patients with AITL and chronic myelomonocytic leukemia (CMML) who experienced sustained complete remission of both diseases after treatment with 5-azacytidine.^17,18  Here, we expand these observations and describe the response and outcome of a retrospective series of 12 AITL patients who received 5-azacytidinefor concomitant myeloid neoplasm or used as compassionate therapy in relapsing/refractory AITL patients in the absence of available therapy or when such therapy was contraindicated, at the discretion of the physician.

Patients received 5-azacytidine 75 mg/m² daily, subcutaneously, for 7 consecutive days, every 28 days, until progression or unacceptable toxicity. Evaluations were performed by computed tomography scan, and responses were assessed by investigators following the 1999 Cheson criteria.¹⁹ 

Two patients (PTCL1¹⁷  and PTCL7¹⁸ ) were previously reported. AITL diagnoses were all confirmed by expert pathologists in the framework of the national program “lymphopath,”³  based on the criteria of the World Health Organization 2008 classification, and tumor samples were collected in the frame of the Tenomic network from the Lymphoma Study Association. TET2, IDH2, DNMT3A, and RHOA status was centrally determined by targeted deep sequencing of DNA extracted from formalin-fixed, paraffin-embedded samples using PGM technology (ThermoFisher) (see supplemental Methods, available at the Blood Web site). The median depth of sequencing was 1432×.

Patient characteristics are given in Table 1. The median age was 70.5 years (interquartile range [IQR], 67.5-73.5 years). All but one had relapsed-refractory disease, with a median of 2 lines of therapy (IQR, 1.75-3), including cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP)–like therapy in 10 cases and mini-CHOP in one. PTCL1 was the only treatment-naive patient. She received 5-azacytidine as first-line therapy for concomitant CMML with excess blasts and thrombocytopenia. Three other patients had asymptomatic CMML, and 1 patient had refractory cytopenia with multilineage dysplasia. In total, 5 out of 12 patients (41%) had an associated myeloid neoplasm. All patients had disseminated (stage III-IV) disease, and 4 out of 12 patients (33%) had a poor performance status (Eastern Cooperative Oncology Group [ECOG] >2) at 5-azacytidine initiation.

Treatment was introduced between January 2013 and July 2016. Patients received a median of 5.5 cycles (IQR, 3.75-17 cycles). In addition to 5-azacytidine, 6 out of 12 patients (50%) received rituximab because of the presence of Epstein-Barr virus replication or numerous Epstein-Barr virus B-blasts in the lymph node biopsy.

Treatment was well tolerated. Three patients required transfusion, and none developed febrile neutropenia. One patient experienced grade 2 neuropathy that was considered to be paraneoplastic syndrome unrelated to treatment, and another experienced unexpected digestive toxicity (grade 3 diarrhea). There were no treatment-related deaths.

Nine patients showed a response, including 6 complete responses (CRs) and 3 partial responses (PRs), leading to an overall response rate of 75%, including 50% CRs and 25% PRs (Figure 1A). After a median follow-up of 27 months, the median progression-free survival was 15 months and median OS of 21 months (Figure 1 B-C). No patient died of myeloid neoplasm during the follow-up period. It is noteworthy that some elderly patients with poor performance status (ECOG 3-4) had a sustained AITL response after 5-azacytidine treatment, with an acceptable tolerance.

Three patients stopped treatment after 5, 7, and 6 cycles: 1 because of digestive toxicity, 1 at his request, and 1 at the request of the physician. The first 2 patients remained in complete remission >18 months after discontinuing 5-azacytidine, whereas the last (PTCL4) relapsed 9 months after. Treatment was resumed, rapidly resulting in a sustained CR. He is currently still on therapy.

Five patients have shown a sustained response, as they are still in complete remission >23 months after treatment initiation. Three are still receiving treatment, whereas it was discontinued in the other 2 patients, raising the question of the optimal duration of HMA therapy and the possibility of discontinuing treatment.

Molecular studies were centrally performed using targeted deep sequencing. We detected TET2 mutations in 12 out of 12 patients, and 7 out of 12 patients (58%) had 2 mutations. In addition, 4 out of 12 patients (33%) had DNMT3A mutations, 5 out of 12 patients (41%) had RHOA mutations, 4 out of 5 patients had a p.G17V substitution, and only 1 patient had an IDH2^R172 mutation (supplemental Table 1). We were unable to assess the impact of TET2 mutation on treatment response because the samples from all patients were TET2 mutated. Within the limit of this cohort, we observed no relationship among number of TET2 mutations; mutations in DNM3A, IDH2, or RHOA; and response to treatment. An association with rituximab was not associated with a higher response rate (66% vs 83%; P = 1), likely reflecting the limited efficacy of this drug in AITL.²⁰  Finally, the 5 patients with an associated myeloid neoplasm responded (4 CRs and 1 PR), while 4 out of 7 patients without associated myeloid anomalies responded (2 CRs and 2 PRs; P = .2), indicating that the effect of 5-azacytidine on AITL is not restricted to patients with associated myeloid disease.

We highlight here an association of AITL with CMML in 4 patients. This association has rarely been reported^17,18,21  but could be more frequent than previously thought. Indeed, these 2 neoplasms share common oncogenic events, such as TET2 or DNMT3A mutations, which usually occur in hematopoietic progenitor cells and could lead to the development of both diseases. Clinical and molecular studies are required for a better understanding of this association. Furthermore, the mechanism of action of 5-azacytidine in AITL has not been elucidated yet. Indeed, it is unknown whether 5-azacytidine has a direct effect on neoplastic T cells or whether it acts on abnormal TET2-mutated myeloid cells. Indeed, it could be hypothesized that abnormal TET2-mutated myeloid cells could provide signals promoting survival and expansion of neoplastic T cells, which would be reversed by 5-azacytidine.

Several other questions remain unanswered, especially whether TFH-derived PTCLs other than AITL, which share similar mutational anomalies with AITL (especially recurrent TET2, DNMT3A, and RHOA mutations),²²  respond similarly to 5-azacytidine or whether genetic or epigenetic response markers could be determined to predict which patients could benefit from this treatment. These questions warrant a prospective study, which is planned to start this year (EudraCT #2017-003909-17).