Identification of high‐risk DUSP22‐rearranged ALK‐negative anaplastic large cell lymphoma

Recent studies have identified two mutually exclusive recurrent rearrangements in anaplastic lymphoma kinase (ALK)negative anaplastic large cell lymphoma (ALCL) that have important clinical significance (Feldman et al, 2011) (Vasmatzis et al, 2012). The DUSP22 rearrangement, which involves the DUSP22-IRF4 locus on 6p25.3, most commonly occurs as a t(6;7)(p25.3;q32.3)(2) and the TP63 rearrangement, which results from a TP63-TBL1XR1 inversion (Vasmatzis et al, 2012). In the first clinical report, DUSP22 rearrangements occurred in ~30% of all ALK-negative ALCL and were associated with a very favourable prognosis [5-year overall survival (OS) 90%] whereas TP63 rearrangements occurred in 8% and were associated with a dismal prognosis (5-year OS 17%) (Parrilla Castellar et al, 2014). The majority of ALK-negative ALCL were ‘triple negative’, lacking any known rearrangements, and had an intermediate prognosis (5-year OS 42%) (Parrilla Castellar et al, 2014). More recently, five cases of DUSP22-rearranged ALK-negative ALCL from a Danish series (Pedersen et al, 2017a) and eight cases (including one PTCL-not otherwise specified) from an upfront transplant Phase 2 study (Pedersen et al, 2017b), were evaluated, with similar favourable outcomes (5 year OS >80%). These data have led to treatment guideline modifications, but represent a limited number of cases (NCCN 2018). Herein, we evaluated the frequency, clinical features and outcome of previously defined ALCL genetic subgroups in an independent series of systemic ALCL. In addition, the prognostic significance of immunohistochemical (IHC) markers was explored. All cases of newly diagnosed ALCL were identified in the British Columbia Cancer Lymphoid Cancer database and confirmed by expert haematopathologists based on the World Health Organization classification (GWS, PF). A tissue microarray was constructed and IHC and fluorescence in situ hybridisation (FISH) was performed as previously described using in-house bacterial artificial chromosome break-apart probes for DUSP22 and TP63 loci (Figs S1 and S2) (Scott et al, 2012). Of 62 ALK-negative ALCL cases evaluated, 12 (19%) harboured a DUSP22 rearrangement, one (2%) had a TP63 rearrangement and the remainder were triple negative (n = 49, 79%). All DUSP22 rearrangements were verified on whole sections and at an independent laboratory (AF). Most patients (78%) received CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone)/CHOP-like chemotherapy (92% in DUSP22) (Table I, Table SI). Some high-risk clinical features were noted in the DUSP22-rearranged cases: Median age 61 5 years; extranodal involvement (67%); bone/bone marrow involvement (42%); high lactate dehydrogenase (42%) (Table I). The median follow-up for all living patients was 8 6 years (range 1 8–34 years). Consistent with prior studies, outcomes in ALK-negative ALCL were inferior to those with ALK-positive ALCL [5-year progression-free survival (PFS) 23% vs. 62%, P < 0 002; 5-year OS 32% vs. 69%, P < 0 01] (Fig S3A,B). Of note, survival estimates for ALCL in our study are lower than some, but not all other series (Hapgood & Savage, 2015), possibly reflecting the population-based nature of this analysis. Surprisingly, the outcome of DUSP22-rearranged ALKnegative ALCL cases was lower than that observed in published series, with a 5-year PFS and OS of 40% (Fig 1A,B) and 5-year disease-specific survival of 45%, with similar findings when only those treated with curative intent chemotherapy are evaluated (DUSP22-rearranged, n = 11), 5-year PFS and OS 44%. One patient had a central nervous system (CNS) parenchymal relapse (Table SI). Interestingly, five of the relapses occurred over 1 year from diagnosis, including two ≥4 years. Further details on the clinical course are provided in the Supplementary Material. Of note, the 5-year PFS and OS estimates were poor for triple negative ALKnegative ALCL (19% and 28%, respectively) but comparable to the Danish series (n = 20, 5-year OS 33%) (Fig 1A,B). The sole case with a TP63 rearrangement died within 6 months of diagnosis. Excluding the case with a TP63 rearrangement, multivariable Cox proportional hazard models were used to estimate hazard ratios (HR) using ALK-positive ALCL as the reference group (Table SII). There was no statistical difference in OS and PFS in both crude and adjusted [for international prognostic index (IPI) and age] analyses. Similarly, no differences were observed using triple negative cases as the reference group (results not shown). This may reflect the challenge of analysing small datasets with limited power. Regardless, the outcome observed in DUSP22-rearranged cases remains of clinical significance. Despite the aggressive clinical course in some patients, the IHC features of DUSP22-rearranged cases were in keeping with prior reports and highlight that it is a defined entity. CD2 and CD3 expression was frequent and all cases were