Frequent STAT5B mutations in γδ hepatosplenic T-cell lymphomas

Hepatosplenic T-cell lymphomas (HSTCL) account for o1% of all non-Hodgkin’s lymphomas. These rare T-cell lymphomas primarily affect young males, display aggressive behavior and have a dismal prognosis.¹ Patients present with B-symptoms, hepatosplenome-galy, cytopenias and bone marrow infiltration in the absence of both lymphadenopathy and peripheral blood involvement. Most cases are believed to arise from γδ T cells resident in the spleen and diffusely involve splenic and hepatic sinuses. Accordingly, HSTCL are most commonly TCRγδ⁺, TCRβ⁻, CD4⁻/CD8⁻, CD5⁻, CD56⁺, TIA1⁺, perforin⁻ and granzyme B⁻, the latter markers being indicators of an immature cytotoxic T-cell phenotype.

Little is known about the etiology and pathobiology of this lymphoma, in part due to its rarity. Its prevalence in patients with autoimmune disease treated with anti-TNFα and in immunosuppressed transplant recipients suggests that chronic antigenic stimulation and immune system impairment might play some role in its genesis.¹ Cytogenetic studies indicate a high frequency of nonrandom chromosomal abnormalities, including isochromo-some 7q [i(7)(q10)], trisomy 8 (8+) or loss of the Y chromosome,² but the genes involved in neoplastic transformation are as yet unknown. Gene expression profiling studies report an NK/T cell lymphoma-like signature, and overexpression of several genes including multidrug resistance genes (MDR1), cell trafficking genes (S1PR5) and growth and survival genes (SYK). Interestingly, significant enrichment of genes involved in JAK/STAT signaling was also observed.³

Signal transducer and activator of transcription factors STAT3 and STAT5 have been implicated in lymphocyte development, survival and growth, playing prominent roles in transducing environmental signals.^4,5 In particular, STAT5 activation has been linked to T-cell development and homeostasis, as well as to the initiation of γδ T-cell differentiation.⁴ Deregulation of the STAT3 pathway has been reported in anaplastic large-cell lymphoma (ALCL) and Sezary syndrome.⁶ More recently, activating mutations of STAT3 were described in both T and NK large granular lymphocytic leukemia (LGL),^7,8 and in rare cases of ALCL, ALK negative, and peripheral T-cell lymphoma-NOS, CD30 positive.⁹ Similarly, constitutive activation of STAT5B has been described in myeloid neoplasms and in B-acute lymphoblastic leukemia (B-ALL),⁴ and very recently acquired mutations of STAT5B have been reported in T-cell prolymphocytic leukemia (T-PLL),¹⁰ rare aggressive cases of LGL¹¹ and T-ALL.¹²

Given the central role of STAT3 and STAT5B in the control of T-cell proliferative responses and in γδ T-cell development, and accumulating evidence for their involvement in T-cell tumorigenesis, we reasoned that these genes could potentially play a role in hepatosplenic γδ T-cell lymphomas as well.

To carry out this study, 21 cases of HSTCL with available formalin-fixed paraffin-embedded splenic tissue were retrieved from the consultation files of the Hematopathology Section of the National Cancer Institute under an IRB approval protocol. All cases were reviewed by four of the co-authors (ESJ, SDP, MR, AN) and the diagnoses were confirmed according to the current World Health Organization classification criteria.1 Immunohistochemical studies were performed using the following panel of antibodies: CD20, CD2, CD3, CD4, CD5, CD7, CD8, CD56, CD57, βF1, TCRγ, T-cell restricted intracellular antigen-1 (TIA-1), granzyme B and perforin. The panel of antibodies, clone designations, dilutions and sources are listed in Supplementary Table 1. FISH analysis for evidence of chromosome 7q imbalances was also performed; details are provided in the Supplementary Methods.

All cases showed diffuse involvement of the splenic red pulp with tumor localized to the splenic sinuses and cords, and all were positive for TCRγ and negative for βF1 protein expression. With one exception, all expressed CD3 and were CD5 negative. Seventeen cases were double negative for CD4 and CD8; four additional cases showed partial expression of CD8 only. Ten were CD56 positive, and none of the 11 cases analyzed expressed CD57. All but one case expressed TIA-1 and most were negative for granzyme B and perforin (Table 1). The neoplastic nature of the T-cell proliferation was further confirmed by clonal T-cell receptor gamma-chain gene rearrangement studies, which were positive in 17/18 cases with adequate quality of DNA to allow the analysis (Table 1). Seventeen of the 21 cases showed chromosome 7q abnormalities, consistent with the presence of isochromosome 7q, by FISH analysis (see Table 1 and Supplementary Figure S1). In addition, cases 6 and 7 had amplifications involving 7q. Classical cytogenetic reports from submitting institutions were available on three patients, and were concordant with our own FISH results.

Table 1.

Summary of clinical, immunophenotypical, TRG clonality, 7q imbalance, STAT3 and STAT5B mutational status

Abbreviation: TRG, T-cell receptor gamma-chain gene rearrangement. Red—positive; pink—positive focal or weak; green—negative; gray—not available; blue—wild type; ochre—mutated.

The clinical features of the patient cohort are summarized in Table 1. Briefly, γδ HSTCL were more frequent in males (M:F ratio = 2.5) with a median age of 35 years (range 12–79). Ten of 13 patients for whom clinical data were available showed various degrees of cytopenia and splenomegaly, and 7 of 10 patients (70%) with available bone marrow biopsy or flow-cytometry data showed involvement. Two of these also had liver biopsies positive for lymphoma.

Targeted analysis for STAT3 mutations located between codons 614 and 661, and for STAT5B mutation located between codons 642 and 665 was performed using pyrosequencing on a PyroMark Q24 instrument (Qiagen, Alameda, CA, USA) on involved spleens from all cases. The pyrosequencing assay was designed to target all hotspot mutations identified in recent publications,^7,8,11,12 using PyroMark Assay Design v2.0 (Qiagen) (see Supplementary Methods). Seven cases (33.3%) were found to have a STAT5B somatic missense mutation. All shared the same hot spot mutation, c.1924A>C; p.N642H. Six of the seven cases had imbalances of chromosome 7q. Two additional cases (9.5%) had a STAT3 somatic missense mutation (c.1981G>T; p.D661Y, and c.1919A>T; p.Y640F), one of which had an abnormality of chromosome 7q. All remaining cases were wild type for STAT3 at the targeted sites. STAT3 and STAT5B missense mutations were mutually exclusive; therefore 9/21 HTSCL cases (42.8%) contained STAT mutations (Table 1). Representative cases are shown in Figure 1.

Figure 1.

Representative γδ HSTCL cases with STAT5B and STAT3 mutation. Case 1 γδ HSTCL with N642H STAT5B mutation (a–c). (a) Monotonous proliferation of medium-sized lymphocytes with pale cytoplasm infiltrating dilated splenic sinuses (H&E × 200). (b) The cells are strongly positive for TCRγ (×200). (c) The pyrogram reveals a c.1924A>C (N642H) mutation in STAT5B. Case 20 γδ HSTCL with D661Y STAT3 mutation (d–f). (d) Dilated splenic sinuses filled with atypical medium-sized lymphocytes (H&E, × 400). (e) The cells are strongly positive for TCRγ (×400). (f) The pyrogram shows a c.1981 G>T (D661Y) mutation in STAT3.

The prevalence of STAT5B mutations found in HSTCL is similar to the 36% frequency recently reported by Kiel et al.¹⁰ in T-PLL, a rare highly aggressive form of T-cell leukemia of αβ subtype. In the Kiel study, all of the STAT5B mutations were found in the Src-like homology (SH2) domain, with 11 occurring recurrently at N642, identical to the mutations in the present report. Similar STAT5B mutations have also been reported at much lower frequency in two other T-lymphoid neoplasms. Rajala et al.¹¹ reported a frequency of only 2% in LGL and Kontro et al.¹² found a frequency of 8% in T-ALL. Interestingly, two of the four patients with STAT5B mutations in the Rajala study also had the identical N642H mutation seen in our patients, and these two patients had an unusually aggressive, fatal form of LGL. The N642H mutation was also the most common STAT5B mutation reported in T-ALL and tended to occur in adult cases,¹² which are generally very aggressive. Furthermore, high STAT5 phosphorylation correlated with poor overall survival in BCR-ABL-positive B-ALL patients.⁴ As such, the presence of N642H STAT5B mutation may, in part, be responsible for the aggressive behavior of the T-cell neoplasms in which it is found, including the γδ HSTCL. Unfortunately, due to the consultative nature of our cases, no follow-up data were available to correlate clinical behavior with STAT5B mutation status.

STAT3 mutations were found in only 9.5% of γδ HSTCL, contrasting with the 40–70% frequency found in T-LGL.^7,8 The two cases carried the D661Y and Y640F mutations, respectively. These are also the most common mutations found in T-LGL patients.^7,8 Although we considered the possibility that the two STAT3 mutated cases might represent rare γδ T-LGL involving the spleen, both cases lacked circulating large granular lymphocytes and were negative for CD5, CD8 and CD57, which are generally positive in γδ T-LGL,¹³ and one of the two also had a chromosome 7q abnormality, consistent with ischromosome 7q (see Supplementary Figure S1–G). These data reinforce other reports indicating that STAT3 mutations may not be restricted to LGL.⁹

The high frequency of STAT5B and STAT3 mutations in γδ HSTCL strongly suggests that deregulation of the STAT pathway is an important oncogenic event in the pathogenesis of this lymphoma. There is convincing data that the STAT3 D661Y and Y640F, and the STAT5B N642H mutations found in the γδ HSTCL result in constitutive tyrosine phosphorylation and an increase in transcriptional activity of downstream genes implicated in tumor cell proliferation and survival.^7,8,10–12 The role of STAT5B in lymphomagenesis is further supported by development of CD8+ lymphoblastic lymphoma in STAT5B transgenic mice.¹⁴

No differences in immunophenotype were noted between STAT mutated cases and wild-type cases except for a higher expression of CD56 in the STAT5B mutated cases, 85% vs 28%. The same observation was made by Rajala et al.¹¹ in STAT5B mutated LGL. As STAT5B is important for NK-cell activation, and STAT5B-deficient patients have reduced numbers of NK cells, these authors suggested that STAT5B mutations might be involved in transformation toward NK-cell lineage.¹¹

In summary, we report a high frequency of STAT5B and STAT3 mutations in γδ HSTCL. This is the first report of a recurrent oncogenic mutation in this rare lymphoma subtype, and is likely to be important in the pathogenesis of this aggressive entity. The identification of frequent STAT5B mutations suggests a therapeutic approach with novel STAT5 inhibitors under development such as pimozide, or by targeting essential STAT transcriptional cofactors.^10,15,16

Supplementary Material

Supplement figure 1

Figure S1

Chromosome 7q imbalances in γ/δ hepatosplenic T-cell lymphomas.

Representative examples of chromosome 7q imbalances using dual color-FISH probes (7q31, Spectrum Orange/ CEP7 Spectrum Green) (Panels A-C and E-G). Panel A. Case 9 (STAT5B WT), negative for chr 7q imbalance; Panels B. Case 7 (STAT5B mutated), C. Case 6 (STAT5B mutated), E. Case 13 (STAT5B WT), F. Case 2 (STAT5B mutated) and G. Case 21 (STAT3 mutated), all positive for chr 7q imbalance. Arrows show characteristic pattern of isochromosome with two red signals (7q) flanking the central centromeric green signal in most cases. Note that case 6 (Panel C) also shows amplification of 7q signals. Panel D. Case 6 (STAT5B mutated). To confirm loss of chromosome 7p selective cases (Case 6 shown) were subjected to hybridization with dual color probe for 7q31 (Spectrum Orange)/7p11(Spectrum Green), showing loss of 7p (arrowheads). Note that Panels C and D represent the same Case 6 hybridized with two different probe combinations.