A phase 2 biomarker-driven study of ruxolitinib demonstrates effectiveness of JAK/STAT targeting in T-cell lymphomas

Abstract

Signaling through JAK1 and/or JAK2 is common among tumor and nontumor cells within peripheral T-cell lymphoma (PTCL). No oral therapies are approved for PTCL, and better treatments for relapsed/refractory disease are urgently needed. We conducted a phase 2 study of the JAK1/2 inhibitor ruxolitinib for patients with relapsed/refractory PTCL (n = 45) or mycosis fungoides (MF) (n = 7). Patients enrolled onto 1 of 3 biomarker-defined cohorts: (1) activating JAK and/or STAT mutations, (2) ≥30% pSTAT3 expression among tumor cells by immunohistochemistry, or (3) neither or insufficient tissue to assess. Patients received ruxolitinib 20 mg PO twice daily until progression and were assessed for response after cycles 2 and 5 and every 3 cycles thereafter. The primary endpoint was clinical benefit rate (CBR), defined as the combination of complete response, partial response (PR), and stable disease lasting at least 6 months. Only 1 of 7 patients with MF had CBR (ongoing PR > 18 months). CBR among the PTCL cases (n = 45) in cohorts 1, 2, and 3 were 53%, 45%, and 13% (cohorts 1 & 2 vs 3, P = .02), respectively. Eight patients had CBR > 12 months (5 ongoing), including 4 of 5 patients with T-cell large granular lymphocytic leukemia. In an exploratory analysis using multiplex immunofluorescence, expression of phosphorylated S6, a marker of PI3 kinase or mitogen-activated protein kinase activation, in <25% of tumor cells was associated with response to ruxolitinib (P = .05). Our findings indicate that ruxolitinib is active across various PTCL subtypes and support a precision therapy approach to JAK/STAT inhibition in patients with PTCL. This trial was registered at www.clincialtrials.gov as #NCT02974647.

Graphical abstract

Figure 1.

CONSORT diagram showing distribution of patients among cohorts upon initial enrollment, based on prior NGS performed on patient biopsies, and at final assignment after NGS of preenrollment biopsy and/or pSTAT3 IHC. Note that the final cohort for evaluation was 52 patients, as 1 patient in cohort 2 withdrew consent during the first week of treatment.

Figure 2.

Swimmer’s plot for all evaluable patients treated with ruxolitinib. Each bar represents 1 subject in the study. Right arrow cap indicates ongoing treatment. ALCL included systemic ALK⁻ ALCL (n = 3) and primary cutaneous ALCL (n = 1). G/D TCL included hepatosplenic T-cell lymphoma (n = 2), monomorphic epitheliotropic intestinal T-cell lymphoma (n = 1), and primary cutaneous γδ-TCL (n = 1). JAK/STAT Mutations: only JAK or STAT mutations listed. (−) indicates no JAK or STAT mutation present. Additional mutations identified by next generation sequencing (NGS) are provided in supplemental Table 1. pSTAT3: (+) is defined as ≥30% expression in tumor cells by immunohistochemical staining; (−) defined as <30% expression. ND indicates that NGS or IHC for pSTAT3 was not done. Cohorts: cohort 1 (JAK/STAT mutations present); cohort 2 (pSTAT3 ≥30% by IHC); cohort 3 (neither or ND). AITL/TFH, angioimmunoblastic T-cell lymphoma and other T-follicular helper lymphomas; ALK⁻ ALCL, ALK⁻ anaplastic large cell lymphoma; BOR, best overall response; POD, progression of disease.

Figure 3.

Multispectral immunofluorescence with 6 markers plus 4′,6-diamidino-2-phenylindole (DAPI) using the Vectra platform in pretreatment biopsies from 9 patients. (A) Spatial definition of the tumor involvement in a biopsy of MF using the HALO software. (B) Representative mIF image from a biopsy of PTCL stained as indicated. (C) Phospho-S6 expression in tumor cells distinguished patients with CBR or no CBR from ruxolitinib (P = .05). Phospho-STAT3/5 expression in tumor cells did not differ between patients with CBR or no CBR (P = .9).