ASXL1 and CBL mutations are independently predictive of inferior survival in advanced systemic mastocytosis

KIT D816V is considered the driver mutation in systemic mastocytosis (SM) however this mutation alone does not explain the diverse clinical manifestations of this disorder (Pardanani, 2015). Recent studies have reported pathogenic mutations in genes other than KIT, such as TET2, that co-segregate with KIT D816V, and thereby enhance its oncogenic potential (Tefferi et al, 2009; Soucie et al, 2012; Schwaab et al, 2013). While non-KIT mutations occur in SM, their pattern(s) of co-segregation and prognostic value, independent of age and SM subtype, have not been fully defined in a mature dataset. In this pilot study of advanced SM patients, we sequenced a panel of myeloid-relevant genes for known pathogenic mutations, to ascertain: (i) the frequency of individual mutations; (ii) the pattern(s) of mutation co-segregation; and (iii) the prognostic value of mutations, both individually and in combination, on overall- (OS) and leukaemia-free survival (LFS). The current study was approved by the Mayo Clinic institutional review board. All patients consented to review of their medical records for research. SM was classified as per World Health Organization criteria (Horny et al, 2008). DNA was isolated from archived bone marrow (BM) cell pellets. Next-generation deep sequencing by HiSeq sequencing by synthesis SBS chemistry (Illumina, San Diego, CA, USA) was performed to investigate the coding regions of the following 22 genes as previously described (Lasho et al, 2015): KIT, JAK2, CALR, SRSF2, U2AF1, SF3B1, ASXL1, TET2, EZH2, SETBP1, DNMT3A, IDH1, IGH2, SUZ12, CBL, NRAS, FLT3, PTPN11, RUNX1, CEBPA, TP53 and NPM1. Base-calling was performed using Illumina's Real Time Analyser (RTA) version 1.17.21.3. Genesifter® software was utilized (PerkinElmer, Danvers, MA, USA) to analyse sequencing data. Nucleotide variants were called using the Genome Analysis Toolkit (GATK -Broad Institute, Cambridge, MA, USA). Genetic variants were manually curated to identify known mutations. Double mutations targeting a single gene were counted as one mutation because the current analysis did not distinguish between mono-allelic versus bi-allelic mutations. Clinical statistical analyses considered patient parameters obtained at time of referral. Differences in the distribution of continuous variables between categories were analysed by either Mann–Whitney or Kruskal–Wallis tests. Patient groups with nominal variables were compared by χ2-test. Overall survival analysis was considered from the date of diagnosis to date of death (uncensored) or last contact (censored). Overall survival curves were prepared by the Kaplan–Meier method and compared by the log-rank test. A Cox proportional hazards regression model was used for multivariate analysis. P values <0·05 were considered significant. The Stat View statistical package (SAS Institute, Cary, NC, USA) was used for all calculations. We studied 19 patients with advanced SM; 6 had aggressive SM (ASM) and 13 had SM with an associated haematological disease (SM-AHD). Clinical and laboratory characteristics are presented in Table 1. Of the patients with SM-AHD, 5 (38%) had myeloproliferative neoplasm-unclassified (MPN-u), 3 (23%) chronic myelomonocytic leukaemia (CMML), 2 (15%) acute myeloid leukaemia (AML), and 1 patient (8%) each had polycythaemia vera, myelodysplastic syndrome and monoclonal gammopathy. The median follow-up from diagnosis was 20 months (range 2–84). Over this period, 5 patients (26%) developed leukaemic transformation and 12 patients (63%) have died. 175 (9·4–1050) n = 17 463 (34–1050) n = 5 136 (9·4–674) n = 12 KIT D816V was identified in 9 (47%) patients; 9 of 10 patients who tested negative by deep sequencing were screened by allele-specific polymerase chain reaction (sensitivity ~0·01%), with 8 additional patients testing positive. Thus, the overall KIT D816V frequency was 89%. Fifteen (79%) patients harboured at least one additional non-KIT mutation; of these, seven patients had one additional mutation (ASXL1/TET2/SF3B1 = 2 patients each and JAK2 = 1 patient), four patients had two additional mutations, three patients had three additional mutations and one patient had four additional mutations (Table SI). The most frequently mutated genes (other than KIT) were ASXL1/TET2 (five patients each, 26%), CBL (four patients, 21%), SETBP1/SF3B1/DNMT3A (two patients each, 11%) and JAK2/CALR/FLT3/IDH1/RUNX1/TP53 (one patient each, 5%) (Table SI). There was no overlap between patients harbouring TET2 and ASXL1 mutations; however, 2 and 1 CBL-mutated patients concurrently harboured TET2 and ASXL1 mutations, respectively. SF3B1 but not SETBP1 or DNMT3A mutations occurred independently of TET2 and ASXL1 mutations. There was no significant difference in the distribution of individual mutations or total number of mutations between ASM and SM-AHD subgroups (Table 1). When considered individually, mutations in the non-KIT genes ASXL1, CBL and SETBP1 (present in at least two patients) were associated with a significantly inferior OS (P < 0·05). When the three mutated genes were considered together, only ASXL1 and CBL maintained their significance, independent of age, SM subtype, and presence of anaemia for inferior OS [Hazard ratio (95% confidence interval) = 17·5 (2·2–142·3) and 25·3 (2·9–223·7), respectively]. Patients harbouring ASXL1 and/or CBL mutations (n = 8, 8 deaths, median OS = 11 months) had a significantly worse OS as compared to those without either mutation (n = 11, 4 deaths, median OS = 84 months) (P = 0·0002) (Fig 1a). When considering the number of non-KIT mutations (zero, one, two, three and four mutations in 5, 7, 3, 3 and 1 patients, respectively), the group with ≥3 mutations (n = 4, 4 deaths, median OS = 7 months) had a significantly inferior OS as compared to those with <3 mutations (n = 15, 8 deaths, median OS = 48 months) (P = 0·01) (Fig 1b). Presence of mutations was not correlated with LFS in this analysis. Despite the small sample size, our study was highly informative for molecular predictors of clinical outcome in advanced SM given the long patient follow-up (63% and 26% of patients followed to death and leukaemic transformation, respectively). Non-KIT mutations were frequent (overall prevalence 79%), particularly TET2, ASXL1 and CBL mutations. Of these, TET2 and ASXL1 mutations were mutually exclusive. Presence of ASXL1 and/or CBL mutations, or occurrence of ≥3 non-KIT mutations, was independently associated with significantly inferior OS (but not LFS). Currently, risk stratification of advanced SM is suboptimal and based primarily on conventional prognostic factors (Lim et al, 2009; Pardanani et al, 2009). More recently however, molecular risk assignment has proven increasingly useful in other chronic myeloid neoplasms (Bejar et al, 2011; Tefferi & Pardanani, 2015). Consequently, we believe the current molecular data will be clinically useful, if confirmed in a larger cohort, in identifying those patients at highest risk who may be candidates for aggressive treatment approaches, including allogeneic stem cell transplantation. The authors have no competing financial interests to declare. TL, AT and AP designed the project and analysed the data. TL analysed the sequencing data. CF extracted DNA for sequencing. DZ contributed patient samples for the study. RAA, EAW and NG abstracted data from medical charts. CAH reviewed bone marrow histology. RPK reviewed cytogenetic and sequencing data. AP wrote the first draft of the paper. All the authors reviewed the manuscript draft, provided critical input and approved the final version of the manuscript. TL and AP had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.