Recurrent STAT3-JAK2 fusions in indolent T-cell lymphoproliferative disorder of the gastrointestinal tract

TO THE EDITOR:

Indolent T-cell lymphoproliferative disorder of the gastrointestinal tract (GI TLPD) is a newly recognized entity in the World Health Organization (WHO) classification of lymphoid neoplasms.¹  GI TLPD is defined as a clonal T-cell proliferation occurring in the GI tract, most commonly in the colon and small bowel, with bland cytologic appearance, a generally noninvasive architectural pattern, and indolent clinical behavior. Patients typically are adults (with a predominance of males) who present with GI symptoms, including abdominal pain, vomiting, diarrhea, and weight loss. Unlike enteropathy-associated T-cell lymphoma (EATL), from which it must be distinguished, GI TLPD is not associated with underlying celiac disease, and its etiology and molecular pathogenesis remain unknown. GI TLPD tends not to respond to chemotherapy and, despite its indolent nature, symptoms are often debilitating and challenging to treat.² 

We examined the pathological and genetic features of GI TLPD to better understand this new entity (see supplemental Data, available on the Blood Web site). By reviewing clinical records and pathological material, we identified and confirmed 10 patients with GI TLPD and 1 patient with natural killer cell enteropathy, which is considered a separate entity by WHO criteria.^1-3  There were 6 males and 5 females (mean age, 56 years; range, 39-74 years; Table 1). Histologic features of GI TLPD were similar to those previously reported (Figure 1A-B).^1,2  Of the patients with GI TLPD, 5 had a CD4⁺ T-cell phenotype, 4 had a CD8⁺ phenotype, and 1 had a double-positive phenotype (Table 1). Clonal T-cell receptor gene rearrangements were identified in all 10 patients with GI TLPD but not in the patient with natural killer cell enteropathy. Samples from four patients underwent testing by FoundationOne Heme (Foundation Medicine, Cambridge, MA; supplemental Table 1). Notably, 1 CD4⁺ GI TLPD sample was found to have a STAT3-JAK2 fusion. JAK2 fusions have been identified in a variety of myeloid and precursor lymphoid neoplasms^4,5  but are rare in mature T-cell lymphomas^6-9 ; among 200 Mayo Clinic patients with a variety of T-cell lymphoma subtypes previously studied by fluorescence in situ hybridization (FISH) using a JAK2 break-apart probe, all were negative for rearrangement.⁷ 

To confirm the JAK2 rearrangement and determine whether it was recurrent in GI TLPD, we performed FISH on formalin-fixed, paraffin-embedded tissue sections using a previously published break-apart probe.^7,10  The initial sample identified by FoundationOne Heme and samples from 3 additional patients showed rearrangements of the JAK2 gene region on 9p24.1; they represented 4 of 5 patients with a CD4⁺ phenotype, whereas patients with other phenotypes lacked rearrangements (Table 1; Figure 1C-F). We tested 23 EATLs (in addition to 4 EATLs previously studied⁷ ) using a JAK2 break-apart FISH probe, including 9 type II EATLs, now designated as monomorphic epitheliotropic intestinal T-cell lymphomas in the WHO classification¹¹ ; all 27 samples were negative. We then designed a STAT3-JAK2 dual-fusion FISH probe and demonstrated that all 4 GI TLPDs demonstrated fusion signals (Figure 1G), confirming a recurrent t(9;17)(p24.1;q21.2) rearrangement.

To evaluate the precise breakpoints of the fusion transcripts, we performed RNA sequencing from formalin-fixed, paraffin-embedded tissue and obtained interpretable results in 3 of the 4 patients with t(9;17)(p24.1;q21.2) rearrangements. All 3 showed STAT3-JAK2 fusion with the identical messenger RNA breakpoint (Figure 1J), confirmed by Sanger sequencing (Figure 1I). The predicted fusion protein retained the C-terminal JH1 kinase domain of JAK2 and a portion of the JH2 pseudokinase domain but lacked the upstream SH2 and FERM domains (Figure 1J), similar to other JAK2 fusions.^4,5,7  Reciprocal JAK2-STAT3 fusion transcripts were also detected in all 3 sequenced samples; however, the breakpoint was different in each sample and likely nonfunctional (1 sample out of frame and 2 with predicted proteins lacking the JAK2 tyrosine kinase domain and virtually all of the STAT3 coding region). These data demonstrate recurrent STAT3-JAK2 fusions in GI TLPDs. The exclusivity for CD4⁺ samples and identical breakpoints strongly suggest that STAT3-JAK2 fusions contribute to the pathogenesis of this disorder.

STAT3 activation is critical in many T-cell neoplasms, and activating mutations of the STAT3 gene are relatively common.^12-16  However, Perry et al²  found minimal phosphorylated STAT3 (pSTAT3) in GI TLPDs and did not observe any STAT3 SH2 domain hotspot mutations. To determine whether pSTAT3 might be present in GI TLPDs with STAT3-JAK2 fusions, we performed immunohistochemistry for pSTAT3^Y705 and samples from all 4 patients were negative (Figure 1K), as were samples from 5 of 7 patients who did not have STAT3-JAK2 fusion (Table 1). Although the functionally critical Y705 amino acid of STAT3 was present in the predicted STAT3-JAK2 fusion protein, the loss of the C-terminal portion of the STAT3 transcription activation domain or conformational alterations might inhibit Y705 phosphorylation. Notably, however, the N-terminal domain essential for formation of unphosphorylated STAT3 dimers¹⁷  was also retained in the predicted STAT3-JAK2 fusion protein. Interestingly, STAT3-RARA fusions recently were described in acute promyelocytic leukemia, and coimmunoprecipitation studies confirmed the ability of the resultant fusion proteins to homodimerize.¹⁸  JAK2 fusion partners such as TEL and PCM1 have been shown to contribute functionally to oncogenic signaling by facilitating fusion protein dimerization, leading to STAT5 activation.^4,5,7  Therefore, we examined pSTAT5 by immunohistochemistry and indeed 3 of 4 GI TLPD samples with STAT3-JAK2 fusions were positive (Figure 1L) compared with only 1 of 6 evaluable samples without fusions. The patient sample that was negative for pSTAT5 was also negative for pSTAT1, as were samples from 6 of 8 additional patients tested.

Although limited follow-up data were available, 2 patients developed overt T-cell lymphomas. Patient 4 had a CD4⁺ GI TLPD with STAT3-JAK2 fusion and subsequently developed a large-cell transformation that shared the same fusion (supplemental Figure 1). The WHO classification notes that GI TLPDs expressing CD4 seem to have a higher risk of progression than those expressing CD8.^1,19  Interestingly, FISH in the large-cell component from patient 4 also showed copy number gains of STAT3, present in a previously reported CD4⁺ GI TLPD with large-cell transformation.¹⁹  Patient 9 had a CD8⁺ GI TLPD that lacked JAK2 rearrangement and subsequently developed systemic anaplastic lymphoma kinase–negative anaplastic large-cell lymphoma. Polymerase chain reaction studies identified different clonal T-cell receptor gene rearrangements in the 2 specimens (not shown), suggesting that the anaplastic large-cell lymphoma arose as a second, distinct process.

Taken together, our findings conclusively demonstrate STAT3-JAK2 fusions in GI TLPD, representing the first recurrent genetic abnormality reported in this disease. Fusions were seen in 4 of 5 patients with CD4⁺ phenotypes but not in other phenotypes, suggesting that the molecular pathogenesis of GI TLPD may vary on the basis of cell of origin and that distinct molecular subentities may exist. FISH for JAK2 rearrangements might help distinguish GI TLPD from reactive processes because GI TLPDs are cytologically bland and may lack phenotypic aberrancy. Furthermore, STAT3-JAK2 fusions represent a candidate therapeutic target in GI TLPD. A variety of strategies to inhibit JAK-STAT signaling are clinically available or in preclinical development for lymphoma, other malignancies, and nonneoplastic conditions.^20,21  T-cell lymphoma cell lines with PCM1-JAK2 fusion are sensitive to the selective JAK2 inhibitor TG101348,⁷  whereas clinical responses to the JAK1/2 inhibitor ruxolitinib have been reported in myeloid neoplasms with PCM1-JAK2.^22-24  Thus, the activity of JAK2 inhibitors in GI TLPD with STAT3-JAK2 fusion merits further study.