Dasatinib in relapsed T follicular helper cell lymphoma: a phase 2 answer

A single-arm test of a TCR-signaling inhibitor

TFHL is a peripheral T cell lymphoma subtype in which recurrent mutations activate T cell receptor (TCR) signaling, most notably RHOA p.G17V, and often co-occur with epigenetic alterations such as TET2 and DNMT3A. Dasatinib is a tyrosine kinase inhibitor that preclinical work in this paper's framing suppresses VAV1 phosphorylation, a step downstream of TCR signaling. The trial asked whether that mechanism could produce enough tumor shrinkage in patients whose disease had returned after chemotherapy.

The design was a multi-center, single-arm, open-label, investigator-initiated phase 2 trial run at 13 sites in Japan, registered as jRCT2031190079. Eligible patients had TFHL diagnosed by histopathology and relapse after at least one prior chemotherapy regimen, or stable or progressive disease on their latest regimen. Dasatinib was given orally at 100 mg once daily for 30 consecutive days, continued for up to 90 days unless disease progression, unacceptable toxicity, withdrawal, or investigator decision intervened.

Nineteen patients were enrolled, 18 received at least one dose, and 17 were evaluable for efficacy after central pathology review excluded one patient reclassified as PTCL not otherwise specified. Median age was 73 years, half were female, and angioimmunoblastic T cell lymphoma was the predominant subtype (13 patients). The median number of prior therapies among the 17 evaluable patients was 2, ranging from 1 to 9.

The 23.5% response rate fell short of 30%

The primary endpoint was overall response rate (ORR) by CT using Lugano criteria 2014, tested against a null hypothesis of 30% or less. Among 17 evaluable patients, ORR was 23.5% (4/17; 95% CI 3.4–43.7%), comprising three complete responses (17.6%) and one partial response (5.9%). The trial therefore did not meet its predefined primary endpoint.

A second measure gives a slightly different picture and should not be substituted for the primary result. ORR by 18F-fluorodeoxyglucose PET-CT was 29.4% (5/17; 95% CI 7.8–51.1%), with four complete metabolic responses and one partial metabolic response. One patient stopped treatment on day 33 after an investigator assessment of progressive metabolic disease that central review later adjudicated as a complete metabolic response.

Durability was limited. Median dasatinib treatment duration was 32.0 days, with a maximum of 193 days. After a median follow-up of 277 days, median progression-free survival was 59 days (95% CI 17–74) and median overall survival was 284 days (95% CI 214 to not estimable). Treatment-related adverse events occurred in 11 of 18 treated patients (61.1%), grade 3 or higher events in four (22.2%), and no treatment-related deaths were reported.

Responses concentrated in RHOA p.G17V carriers

Mutation profiling combined whole-exome sequencing of tumor tissue with droplet digital PCR and targeted sequencing of plasma cell-free DNA. Across these assays, 76.5% (13/17) of patients carried TCR signaling-related mutations and 64.7% (11/17) carried RHOA-G17V. This combined approach matters because cell-free DNA detected RHOA-G17V in two patients who were negative on tumor whole-exome sequencing and in two without available tumor DNA, which the authors attribute to low tumor burden in the sequencing-negative cases.

Response differed sharply by genotype. In the TCR signaling mutation-positive subgroup, ORR by CT was 30.8% (4/13) and by PET-CT 38.5% (5/13); in the mutation-negative subgroup both were 0% (0/4). Patients with RHOA-G17V had ORR by CT of 36.4% (4/11) and by PET-CT of 45.5% (5/11), and all patients showing clinical benefit carried RHOA-G17V. No responses occurred among the six RHOA-G17V-negative patients.

Progression-free survival was longer in RHOA-G17V-positive patients (median 65 versus 18 days; P = 0.00019) and in TCR signaling mutation-positive patients (median 64 versus 18 days; P = 0.0051). Overall survival did not differ significantly between these groups (RHOA-G17V-positive versus negative: median 340 versus 284 days; P = 0.82). In multivariate analysis, RHOA-G17V was associated with longer PFS (P = 0.0044; HR 0.020; 95% CI 0.0013–0.29). These are associations in a small cohort, not proof that the mutation causes the response, and RHOA-G17V co-occurred with other TCR signaling mutations in 11 of 13 patients, making independent effects hard to separate.

Responders showed cytotoxic and inflammatory immune signatures

Exploratory single-cell RNA and TCR sequencing of peripheral blood and tumor tissue examined why some patients responded. In responders, the paper reports interferon-γ and tumor necrosis factor-enriched tumor microenvironments and peripheral expansion of cytotoxic T cell clones. In other words, the tumors of responding patients were already surrounded by, and their blood contained expanded populations of, immune cells with an inflammatory, cell-killing profile.

The authors interpret this as dasatinib acting partly as an immunomodulatory agent rather than only as a direct tumor-cell kinase inhibitor. That interpretation is exploratory: it rests on a limited number of patients with available specimens, and the trial was not designed to test immune activation as a causal mediator of response.

TET2-mutant immune clones tracked with early progression

The resistance side of the story concerns clonal hematopoiesis, in which mutations accumulate in blood-forming cells and appear in both malignant and non-malignant immune cells. The authors sorted peripheral blood immune subsets and measured the variant allele frequencies of 12 clonal hematopoiesis-related mutations (10 in TET2, two in DNMT3A) plus four tumor-specific mutations in four responders and three non-responders.

TET2 mutations appeared in immune cells lacking tumor-specific mutations, confirming their non-malignant origin. In two non-responders, TET2 mutations were present at high variant allele frequencies in both myeloid and lymphoid lineages, indicating an early origin near hematopoietic stem cells; responders generally showed low TET2 mutation frequencies across immune subsets (median below 2.3%). In CD8+ T cells, TET2-mutant frequencies were significantly lower in naive T cells of responders than non-responders (P = 0.048), and after treatment non-responders showed a decrease in naive CD8+ T cells alongside an increase in central memory CD8+ T cells, suggesting differentiation of TET2-mutated naive cells.

The paper's proposed mechanism is that expanded TET2-mutated CD8+ naive and central memory subsets fail to differentiate into effector populations, limiting dasatinib-induced interferon-γ and tumor necrosis factor signaling in the tumor microenvironment. This is a hypothesis the authors advance to explain their observations, not a demonstrated causal pathway; the analysis covered only seven patients with available frozen samples.

A biomarker hypothesis, not a treatment recommendation

The authors describe the biomarker analyses as hypothesis-generating, intended to understand biological heterogeneity and resistance rather than to establish clinically actionable predictors. The limitations they list are a small sample size, no post-treatment tumor tissue analysis, and too few DNMT3A-positive cases to assess that alteration. The single-arm design also means the 23.5% response rate has no randomized comparator.

Within those bounds, the most decision-relevant split in this paper is by mutation status rather than by the overall response rate alone. Patients whose tumors carried RHOA-G17V accounted for every observed clinical benefit, with a 36.4% CT-based ORR and a median PFS of 65 days, while RHOA-G17V-negative and TCR signaling mutation-negative patients showed no responses. The authors suggest that dasatinib might serve as an immunomodulatory option for patients without clonally expanded TET2 mutations, and that combining it with epigenetic therapies such as histone deacetylase or DNA methylation inhibitors might restore T cell differentiation, but they explicitly frame both as future possibilities requiring larger, independent validation.

Sources used in this answer