ASXL1 and SETBP1 Mutations and Their Prognostic Contribution in Chronic Myelomonocytic Leukemia: a Two-Center Study of 466 Patients

ORIGINAL ARTICLE ASXL1 and SETBP1 mutations and their prognostic contribution in chronic myelomonocytic leukemia: a two-center study of 466 patients

MM Patnaik1, R Itzykson2,3,4, TL Lasho1, O Kosmider2,5, CM Finke1, CA Hanson6, RA Knudson7, RP Ketterling7, A Tefferi1 and E Solary3,4,8

In a cohort of 466 patients, we sought to clarify the prognostic relevance of ASXL1 and SETBP1 mutations, among others, in World Health Organization-defined chronic myelomonocytic leukemia (CMML) and its added value to the Mayo prognostic model. In univariate analysis, survival was adversely affected by ASXL1 (nonsense and frameshift) but not SETBP1 mutations. In multivariable analysis, ASXL1 mutations, absolute monocyte count 410 Â 10(9)/l, hemoglobin o10 g/dl, platelets o100 Â 10(9)/l and circulating immature myeloid cells were independently predictive of shortened survival: hazard ratio (95% confidence interval (CI)) values were 1.5 (1.1–2.0), 2.2 (1.6–3.1), 2.0 (1.6–2.6), 1.5 (1.2–1.9) and 2.0 (1.4–2.7), respectively. A regression coefficient-based prognostic model based on these five risk factors delineated high (Z3 risk factors; HR 6.2, 95% CI 3.7–10.4) intermediate-2 (2 risk factors; HR 3.4, 95% CI 2.0–5.6) intermediate-1 (one risk factor; HR 1.9, 95% CI 1.1–3.3) and low (no risk factors) risk categories with median survivals of 16, 31, 59 and 97 months, respectively. Neither ASXL1 nor SETBP1 mutations predicted leukemic transformation. The current study confirms the independent prognostic value of nonsense/frameshift ASXL1 mutations in CMML and signifies its added value to the Mayo prognostic model, as had been shown previously in the French consortium model.

Leukemia (2014) 28, 2206–2212; doi:10.1038/leu.2014.125

INTRODUCTION Notably, a Mayo Clinic study analyzed several clinical and Chronic myelomonocytic leukemia (CMML) is a clonal, hematopoietic laboratory parameters, including ASXL1 mutations, in 226 10 stem cell disorder, with overlapping features of myelodysplastic patients with CMML; on multivariable analysis, risk factors syndromes (MDS) and myeloproliferative neoplasms.1 The World for survival included hemoglobin o10 g/dl, platelet count Health Organization (WHO) diagnostic criteria for CMML include o100 Â 10(9)/l, AMC 410 Â 10(9)/l and the presence of 10 persistent peripheral blood (PB) absolute monocyte count (AMC) circulating immature myeloid cells (IMCs). In this study, ASXL1 of 41 Â 10(9)/l, absence of BCR-ABL1 or PDGFRA/B mutations, mutations were detected in 49% of patients and did not affect o20% myeloblasts or promonocytes in PB or bone marrow (BM) either overall (P ¼ 0.08) or leukemia-free (P ¼ 0.4) survival. The and presence of dysplasia in one or more myeloid lineages.1 study resulted in the development of the Mayo prognostic model, In addition, the WHO system recognizes CMML-1 (o5% PB blasts with three risk categories, low (0 risk factor), intermediate (1 risk and o10% BM blasts) and CMML-2 (5–19% PB blasts, 10–19% BM factor) and high (Z2 risk factors), with median survivals of 32, 18.5 blasts, or presence of Auer rods). and 10 months, respectively.10 Somatic mutations are detected in 490% of patients with In contrast to the ﬁndings from the above-discussed CMML and include involvement of the following genes: TET2 Mayo Clinic study, a GFM (Groupe Francais des Myelodysplasies) (50–60%), ASXL1 (40–50%), SRSF2 (40–50%), RUNX1 (10–15%), study demonstrated an adverse prognostic effect for ASXL1 KRAS (10%), NRAS (10%), SETBP1 (5–10%), CBL (5–10%), SF3B1 mutations in 312 patients with CMML;5 additional risk factors on (5–10%), ZRSF2 (5–10%), U2AF1 (5–10%), EZH2 (5–10%), IDH2 multivariable analysis included age 465 years, white blood (5–10%), JAK2 (5–10%), FLT3 (o5%), NPM1 (o5%), DNMT3A count (WBC) 415 Â 10(9)/l, platelet count o100 Â 10(9)/l and 5 (o5%), IDH1 (o5%) and TP53 (1%).2 Among these, only ASXL1 hemoglobin level o10 g/dl in females and o11 g/dl in males. and SETBP1 mutations have been prognostically implicated.3–7 The GFM prognostic model assigns three adverse points for ASXL1 (additional sex combs like 1) maps to chromosome 20q11 WBC 415 Â 10(9)/ l and two adverse points for each one and regulates chromatin by interacting with the polycomb-group of the remaining risk factors, resulting in a three-tiered risk repressive complex proteins (PRC1 and PRC2).8 SETBP1 is located stratiﬁcation, low (0–4 points), intermediate (5–7 points) and on chromosome 18q21.1 and encodes the SET binding protein 1. high (8–12 points), with respective median survivals of 56, 27.4 Recurrent somatic SETBP1 mutations are also seen in patients with and 9.2 months. It should be noted that all nucleotide variations atypical chronic myeloid leukemia (B25%)3,6,9 and are identical to (missense, nonsense and frameshift) were regarded as ASXL1 changes seen in patients with the Schinzel–Giedion syndrome.9 mutations in the Mayo study,10 whereas only nonsense and Some,4,5,7 but not all,10 studies have demonstrated a negative frameshift ASXL1 mutations were considered in the French prognostic impact for ASXL1 mutations in patients with CMML. study.5

1Division of Hematology, Mayo Clinic, Rochester, MN, USA; 2Universite´ Paris Descartes, Paris, France; 3Institut Gustave Roussy, Villejuif, France; 4Universite´ Paris-Sud 11, Orsay, France; 5Institut Cochin, Paris, France; 6Division of Hematopathology, Mayo Clinic, Rochester, MN, USA; 7Division of Cytogenetics, Mayo Clinic, Rochester, MN, USA and 8INSERM U1009, Villejuif, France. Correspondence: Professor A Tefferi, Division of Hematology, Department of Medicine, Mayo Clinic, Rochester 55905, MN, USA or Professor E Solary, Inserm UMR 1009, Institut Gustave Roussy, 114, Rue Edouard Vaillant, Villejuif 94805, France. Received 10 March 2014; revised 25 March 2014; accepted 28 March 2014; accepted article preview online 3 April 2014; advance online publication, 9 May 2014 Molecular prognostication in CMML MM Patnaik et al 2207 More recently, SETBP1 mutations were seen in 12 (6.2%) of 195 French consortium CMML registry.5 The median age of the entire French patients with CMML and were reported to affect both patient cohort was 73 years (range, 20–93 years), with 66% of the overall and leukemia-free survival.3 Similar observations were patients being males (Table 1). A total of 395 (85%) patients had made by a Mayo Clinic study where SETBP1 mutations were seen CMML-1 with a median OS of 38 months, whereas 71 (15%) in 8 (4.5%) of 179 patients with CMML.6 Meggendorfer et al.11 patients had CMML-2 with a median OS of 24 months. The French identified SETBP1 mutations in 52 (9.4%) of 551 patients with patient cohort displayed significantly higher hemoglobin level MDS/myeloproliferative neoplasm overlap features, and specifically (Po0.0001), platelet counts (P ¼ 0.0009) and lower BM blast % in 21 (7.1%) of 294 patients with CMML, and demonstrated no (P ¼ 0.04) (Supplementary Table 1). In addition, a higher propor- impact of these mutations on survival. The SETBP1 mutations were tion of the French patients presented with lower risk disease found to strongly overlap with ASXL1 and CBL mutations and were based on the Spanish cytogenetic stratification system (P ¼ 0.01), mutually exclusive with JAK2 and TET2 mutations. The current MDAPS (P ¼ 0.04), GFM model (Po0.002) and the Mayo prognostic study sought to further clarify the prognostic relevance of ASXL1 model (P ¼ 0.001). The distribution of ASXL1 (P ¼ 0.05), SETBP1 and and SETBP1 mutations in a larger group of CMML patients and spliceosome component mutations was similar between the two examine the additional value of incorporating molecular patient cohorts (Supplementary Table 1). information to the Mayo prognostic model, as has been demonstrated previously by the GFM, in their specific prognostic ASXL1 mutations model.5 Archived DNA was available for ASXL1 mutation screening in 420 of the 466 study patients; 164 (39%) patients harbored ASXL1 MATERIALS AND METHODS mutations, excluding missense variations. In the Mayo Clinic cohort, 96 (43%) of 225 patients harbored nonsense/frameshift The current study was a collaborative effort between the Mayo Clinic in ASXL1 mutations and 20 (7%) had ASXL1 missense mutations. Minnesota, USA, and the French consortium for CMML.5,10 Patient data and outcomes from the French consortium for CMML had been used to help ASXL1 mutations in the French cohort did not include missense develop the GFM CMML prognostic model.5 The study was approved by mutations. Table 1 outlines presenting clinical and laboratory the institutional review boards of the Mayo Clinic and the Cochin hospital features and subsequent events stratified by ASXL1 mutational in Paris, France, respectively. Study eligibility criteria included availability of status. ASXL1-mutated patients displayed lower hemoglobin level PB smear, BM histology and cytogenetic information at the time of referral. (P ¼ 0.0007), higher WBC (Po0.0001), higher AMC (Po0.0001) and The diagnoses of CMML, including subclassification into CMML-1 or were more likely to display circulating IMCs (P ¼ 0.0006), U2AF1 CMML-2, and documentation regarding the presence or absence of ring mutations (P ¼ 0.0009) and higher risk categories based on the sideroblasts and leukemic transformation were according to the 2008 1 Spanish cytogenetic risk stratification system (P ¼ 0.01), MDAPS WHO criteria. All complete blood count differentials and PB smears were (P 0.0003), GFM model (P 0.0001) and the Mayo prognostic evaluated for presence of circulating IMCs, defined by the presence of any ¼ o of the following cells in the PB: myeloblasts, promyelocytes, myelocytes or model (P ¼ 0.009). The most common ASXL1 mutation was the metamyelocytes. Karyotype risk designation and CMML risk stratification c.1934dupG; p.G646WfsX12 variant, seen in 65 (40%) patients. In that were considered included the Spanish cytogenetic risk stratification the Mayo cohort, univariate survival analysis revealed that the system,12 the MD Anderson prognostic scoring system (MDAPS),13 Mayo exclusion of missense mutations changed the prognostic impact prognostic model10 and the GFM model.5 of ASXL1 mutations from nonsignificant (P ¼ 0.08) to significant All patients underwent bone marrow examination and cytogenetic (P ¼ 0.04). Accordingly, all subsequent analyses excluded missense evaluation at diagnosis. DNA analysis for spliceosome component ASXL1 mutations. In univariate analysis, the presence of ASXL1 mutations (SRSF2, SF3B1 and U2AF1) and ASXL1 and SETBP1 mutations mutations was associated with inferior survival (Po0.0001; were carried out on BM specimens obtained at diagnosis by previously Figure 1) but did not affect LFS (P 0.2). described techniques.6,10,14–16 DNA was available for sequencing ASXL1 ¼ and SETBP1 genes in 420 patients, whereas it was available for spliceosome component mutation analysis in 414 patients only. In order to address the SETBP1 mutations aforementioned discrepancy regarding the prognostic impact of ASXL1 mutations, relevant analyses in the Mayo cohort were first performed with SETBP1 mutations were seen in 21 (5%) patients and their and without inclusion of missense ASXL1 mutations. In the French distribution was similar in ASXL1-mutated and unmutated cases consortium patients, ASXL1 frameshift and nonsense variations (but not (Table 1). Seven (33%) patients had concomitant SETBP1 and missense), including the common c.1934dupG; p.G646WfsX12 variant, ASXL1 mutations, whereas 9 (43%) had concomitant SETBP1 and were considered as mutations. spliceosome component mutations (6 SRSF2,2U2AF1 and 1 with All statistical analyses considered clinical and laboratory parameters both SF3B1 and SRSF2 mutations). Table 2 outlines the spectrum of obtained at time of referral to the Mayo Clinic and the French consortium SETBP1 mutations seen in 21 patients with WHO-defined CMML. centers that, in most instances, coincided with time of bone marrow biopsy The mutations in order of frequency of occurrence were: D868N and study sample collection. Differences in the distribution of continuous (48%), G870S (14%), G870D (10%), D868Y (5%), D908N (5%), I871T variables between categories were analyzed by either Mann–Whitney (for comparison of two groups) or Kruskal–Wallis (comparison of three or more (5%), I871S (5%), S869R (5%) and E858K (5%). Thirteen (68%) of 19 groups) test. Patient groups with nominal variables were compared by evaluable patients with SETBP1 mutations had a normal karyotype, w2 test. Overall survival (OS) was calculated from the date of first referral to whereas 3 (16%) had abnormalities of chromosome 7 (monosomy date of death (uncensored) or last contact (censored). Leukemia-free 7-2, del(p11.2p15-1) and one each had isochromosome 17q and survival (LFS) was calculated from the date of first referral to date of trisomy 8 respectively. The distribution of SETBP1-mutated leukemic transformation (uncensored) or death/last contact (censored). patients across the Mayo prognostic risk categories was 4 (19%) Overall and leukemia-free survival curves were prepared by the Kaplan– low risk, 9 (43%) intermediate risk and 8 (38%) high risk. At last Meier method and compared by the log-rank test. Cox proportional hazard follow-up, 16 (76%) deaths and 3 (14%) leukemic transformations regression model was used for multivariable analysis. The P-values of were documented. In univariate analysis, the presence of SETBP1 o0.05 were considered significant. The Stat View (SAS Institute, Cary, NC, USA) statistical package was used for all calculations. mutations did not significantly influence overall (P ¼ 0.07) or leukemia-free (P ¼ 0.87) survival.

RESULTS Spliceosome pathway component mutations A total of 466 patients with WHO-deﬁned CMML were included in Of the 414 patients who underwent spliceosome pathway the study. Of these, 271 (58%) were seen at the Mayo Clinic from component mutation analysis, 223 (56%) were mutated: 177 1997 through 2012, and 195 (42%) represented a subset of the (43%) SRSF2, 26 (6%) SF3B1 and 29 (7%) U2AF1. There was no

& 2014 Macmillan Publishers Limited Leukemia (2014) 2206 – 2212 Molecular prognostication in CMML MM Patnaik et al 2208 Table 1. Clinical and laboratory features and subsequent events in 466 patients with World Health Organization-deﬁned chronic myelomonocytic leukemia, stratiﬁed by ASXL1 mutational status

Variable All patients CMML patients CMML patients P-value; CMML patients with with CMML with ASXL1 without ASXL1 ASXL1 mutations vs those (n ¼ 466) mutations mutations without ASXL1 mutations (n ¼ 164) (n ¼ 256)

Age in years, median (range) 73 (20–93) 73 (27–91) 73 (20–93) 0.59 Males, n (%) 309 (66) 118 (72) 161 (63) 0.07 Hemoglobin g/dl, median (range) 11 (4.3–16.9) 10.5 (6.4–15.5) 11.6 (4.3–16.9) 0.0007 WBC Â 109/l, median (range) 12.3 (1.3–302) 15.7 (2–302) 10.7 (1.3–126) o0.0001 ANC Â 109/l, median (range) 5.9 (0.1–151) 7.6 (0.3–151) 5.2 (0–92) 0.0001 AMC Â 109/l, median (range) 2.6 (1–110) 3 (1–110) 2.3 (1–38.9) o0.0001 ALC Â 109/l, median (range) 1.7 (0–22) 1.8 (0–22) 1.6 (0–12) 0.09 Platelets Â 109/l, median (range) 106 (8–1100) 115 (8–1100) 110 (9–876) 0.46 Presence of circulating immature myeloid cells, n (%) 235 (50) 97 (59) 105 (41) 0.0006 PB blast %, median (range) 0 (0–19) 0 (0–19) 0 (0–17) 0.06 BM blast %, median (range) 4 (0–19) 4 (0–19) 4 (0–19) 0.57 WHO morphological subtype, n (%) CMML-1 395 (85) 137 (84) 221 (86) 0.37 CMML-2 71 (15) 27 (16) 35 (14) SETBP1 mutations, n (%)a 21 (5) 7 (4) 14 (5) 0.09 Spliceosome component mutational analysis, n (%)b SF3B1 26 (6) 5 (3) 18 (7) 0.06 SRSF2 177 (43) 71 (43) 100 (39) 0.48 U2AF1 29 (7) 20 (12) 9 (3) 0.0009 Spanish cytogenetic risk stratiﬁcation, n (%)c Low 344 (78) 110 (70) 199 (83) 0.01 Intermediate 46 (10) 23 (14) 19 (8) High 54 (12) 25 (16) 23 (9) MD Anderson prognostic risk categories, n (%) Low 227 (49) 61 (37) 152 (59) 0.0003 Intermediate-1 138 (30) 56 (35) 67 (26) Intermediate-2 85 (18) 40 (24) 30 (12) High 16 (3) 7 (4) 7 (3) Mayo Model prognostic risk categories, n (%) Low 99 (21) 79 (48) 66 (26) 0.009 Intermediate 172 (37) 58 (35) 105 (41) High 195 (42) 27 (16) 85 (33) GFM prognostic risk categories, n (%)d Low 188 (45) 25 (15) 163 (64) o0.0001 Intermediate 154 (37) 76 (46) 78 (30) High 78 (18) 63 (39) 15 (6) Leukemic transformations, n (%) 75 (16) 26 (16) 44 (17) 0.46 Deaths, n (%) 279 (60) 117 (71) 137 (54) 0.003 Abbreviations: ALC, absolute lymphocyte count; AMC, absolute monocyte count; ANC, absolute neutrophil count; ASXL1, additional sex combs 1 gene; BM, bone marrow; CMML, chronic myelomonocytic leukemia; GFM, Groupe Francais des Myelodsyplasies; PB, peripheral blood; SF3B1, splicing factor 3B, subunit 1; SRSF2, serine/arginine-rich splicing factor 2; U2AF1, U2 small nuclear RNA auxiliary factor 1; WBC, white blood cell count; WHO, World Health Organization. aSETBP1 mutation analysis was performed in 420 patients because of limited DNA availability. bSpliceosome component mutation analysis was performed in 414 patients because of limited DNA availability. cThe Spanish cytogenetic risk stratiﬁcation was computed in 444 patients (158 patients with ASXL1 mutations and 241 patients without ASXL1 mutations). Twenty-two patients had inadequate chromosomal studies at diagnosis. dASXL1 mutation analysis was performed in 420 patients because of limited DNA availability. Hence, the GFM score was computed in 420 patients only.

difference in Spliceosome pathway component mutational presence of SRSF2, SF3B1 or U2AF1 mutations had no impact on frequencies between the Mayo and French patient cohorts OS or LFS (Table 3). (Supplementary Table 1). U2AF1 mutations were more common in CMML patients with concomitant ASXL1 mutations (P ¼ 0.009). OS and LFS data All 13 Mayo Clinic patients with SF3B1 mutations had 415% BM At a median follow-up of 23 months, considering all 466 study ring sideroblasts (ring sideroblast analysis was not available in the patients, 279 (60%) deaths and 75 (16%) leukemic transformations French cohort). The common spliceosome component mutations were documented. In the Mayo clinic cohort, there were 200 (74%) included SRSF2 P95H-68 (38%), P95L-57 (32%) and P95R-29 (16%), deaths and 42 (15%) leukemic transformations; the corresponding SF3B1 K700E-12 (46%), H662Q-2 (8%), K666N-2 (8%), and U2AF1, ﬁgures in the French cohort were 79 (41%) and 33 (17%). Median S34F-10 (34%), Q157P-8 (28%), Q157R-6 (21%) and Q157G-2 (7%), survivals were 38 months for CMML-1 and 24 months for CMML-2 respectively. There was no statistically signiﬁcant difference (P ¼ 0.11) and, according to the Mayo prognostic model, among the three mutation groups (SRSF2 vs SF3B1 vs U2AF1) 19 months for high-risk, 48 months for intermediate-risk and in prognostically relevant parameters. In univariate analysis, the 67 months for low-risk categories (Po0.0001; Figure 2a).

Leukemia (2014) 2206 – 2212 & 2014 Macmillan Publishers Limited Molecular prognostication in CMML MM Patnaik et al 2209 In univariate analysis, the factors adversely influencing OS (P ¼ 0.0015), lower hemoglobin (P ¼ 0.0002), higher WBC included high WBC (Po0.0001), high absolute neutrophil count (P ¼ 0.0001), higher absolute neutrophil count (P ¼ 0.0035), (Po0.0001), high absolute lymphocyte count (P ¼ 0.0002), high higher absolute lymphocyte count (P ¼ 0.0001), higher AMC AMC (Po0.0001), low platelet count (P ¼ 0.0027), presence of (Po0.0001), lower platelet count (P ¼ 0.0002), presence of circulating IMCs (Po0.0001), PB blasts (P ¼ 0.002), WHO morpho- circulating IMCs (Po0.0001), PB blasts (Po0.0001), BM blasts logical categories (P ¼ 0.01), ASXL1mutations (Po0.0001), MDAPS (Po0.0001), WHO CMML subcategories (Po0.0001), SRSFR2 (Po0.0001), Mayo prognostic model (Po0.0001), GFM model mutations (Po0.0001), MDAPS (Po0.0001), Mayo prognostic (Po0.0001) and the Spanish cytogenetic risk stratification system model (Po0.0001), GFM model (Po0.0001) and the Spanish (Po0.0001) (Table 3). On multivariable analysis, increased AMC cytogenetic risk stratification system (P ¼ 0.0002). Notably, (410 Â 10(9)/l, hazard ratio (HR) 2.2, 95% confidence interval (CI) ASXL1 mutations (P ¼ 0.2) and SETBP1 mutations (P ¼ 0.87) did 1.6–3.1), presence of circulating IMCs (HR 2.0, 95% CI 1.4–2.7), not affect LFS. On multivariable analysis, only increased AMC decreased hemoglobin (o10 g/dl; HR 2.0, 95% CI 1.6–2.6), (P ¼ 0.0001), presence of circulating blasts (P ¼ 0.0016), presence decreased platelet count (o100 Â 10(9)/l; HR 1.5, 95% CI 1.2–1.9) of circulating IMCs (P ¼ 0.01), WHO CMML subcategories and ASXL1 mutations (HR 1.5, 95% CI 1.1–2.0) retained significance (P ¼ 0.005) and high risk stratification by the Spanish cytoge- (Table 3). ASXL1 mutations were able to distinguish prognostically netic risk stratification system (P ¼ 0.04) retained prognostic different subgroups within high-, intermediate- and low-risk significance (Table 3). categories, according to the Mayo prognostic model (Figures 2b–d). Table 3 also provides results of analyses for LFS. In univariate analysis, factors adversely affecting LFS included increased age Incorporation of ASXL1 mutation status into the Mayo prognostic model Using regression coefficients, we developed an ASXL1-inclusive 1 prognostic survival model for CMML, referred to as the Mayo CMML patients without ASXL1 mutations, n = 256, median survival 44 months. Molecular Model (MMM). In this model, 2 points each were CMML patients with ASXL1 assigned for AMC 41010(9)/l, presence of circulating IMCs and 0.8 mutations, n = 164, median survival 24 months. hemoglobin level o10 g/dl, whereas 1.5 points each were assigned for ASXL1 mutations and platelet count 0.6 o100 Â 10 Â (9)/l. MMM effectively stratified patients into four risk categories, low risk (0 points), intermediate-1 risk (r2), intermediate-2 risk (2.5–4.5) and high risk (Z5), with median Survival 0.4 P=<0.0001 survivals of 97, 59, 31 and 16 months, respectively (Figure 3a). MMM risk assignment can be further simplified into low (no risk

0.2 factors), intermediate-1 (one risk factor), intermediate-2 (2 risk factors) and high (Z3 risk factors) risk categories. The patient cohort was then risk stratiﬁed according to the GFM model 0 (Figure 3b), with median OS being 65 months (low risk), 28 0 25 50 75 100 125 150 175 200 months (intermediate risk) and 17 months (high risk), respectively. Months Notably, the MMM was superior to the GFM model in identifying Figure 1. Survival data for 420 patients with CMML stratiﬁed by low-risk CMML patients (median OS 97 months for MMM low risk ASXL1 mutation status. vs 65 months for GFM low risk).

Table 2. Spectrum of SETBP1 mutations in 21 patients with WHO-deﬁned CMML

Gender Age WHO SETBP1 ASXL1 Spliceosome Cytogenetics MDACC risk Mayo risk Leukemic Outcome (years) CMML mutation mutation component category category transformation subtype subtype mutation

M 76 CMML-1 D868N Y N 46, XY [20] Intermediate-2 Intermediate No Dead M 80 CMML-1 D868N Y SRSF2 46, XY[20] Low Intermediate No Alive SF3B1 M 82 CMML-1 D868N Y SRSF2 45,X-Y,i(17)(q10)[20] Intermediate-2 High No Dead M 54 CMML-1 D868N Y SRSF2 NA Intermediate-1 Intermediate No Dead M 84 CMML-1 I871T Y U2AF1 46, XY [20] Low Low No Dead F 69 CMML-2 G870S Y Y 46, XX[20] Intermediate-2 High No Dead M 84 CMML-1 D868N Y U2AF1 46, XY, Intermediate-2 High No Dead del(12)(p11.2p12)[5]/45,idem,-7[5] F 66 CMML-1 D908N N Y NA Intermediate-1 Intermediate No Dead M 58 CMML-1 D868N N N 46,XY[20] Intermediate-1 Intermediate No Alive M 81 CMML-1 G870S N N 46,XY[20] Intermediate-1 Intermediate No Dead M 84 CMML-2 D868N N N 46, XY[20] Intermediate-1 Intermediate No Alive F 75 CMML-1 G870D N SRSF2 NA Low Low No Dead M 59 CMML-1 D868N N SRSF2 46, XY Intermediate-1 High Yes Dead F 80 CMML-2 G870S N N 46, XX [20] Intermediate-2 High Yes Dead M 73 CMML-1 D868N N SRSF2 45, X-Y, del(7)(p11.2p15) [20] Intermediate-1 High No Dead M 74 CMML-1 G870D N SRSF2 47, XY, þ 8[20] Intermediate-1 Intermediate No Dead M 79 CMML-2 S869R N N 46, XY[20] Intermediate-1 High Yes Dead M 54 CMML-1 E858K N N 46, XY[20] Low Low No Alive F 78 CMML-1 D868Y N N 45,XX,-7[20] Intermediate-1 High No Dead M 59 CMML-1 I871S N N 46, XY 20] Low Intermediate No Dead M 60 CMML-1 D868N N N 46, XY[20] Intermediate-1 Low No Alive Abbreviations: ASXL1, additional sex combs 1 gene; CMML, chronic myelomonocytic leukemia; F, female; M, male; MDACC, MD Anderson Cancer Center; N, no; NA, not available; WHO, World Health Organization; Y, yes.

& 2014 Macmillan Publishers Limited Leukemia (2014) 2206 – 2212 Molecular prognostication in CMML MM Patnaik et al 2210 Table 3. Univariate and multivariable overall and leukemia-free survival analysis for 420 patients with CMML

Variables Overall survival Leukemia-free survival

Univariate analysis Multivariable analysis Univariate analysis Multivariable analysis P-value P-value P-value P-value

Age 0.89 0.0015 NS Sex 0.32 0.15 Hemoglobin level o0.0001 0.0002 NS Hemoglobin o10 g/dl o0.0001 o0.0001 White blood cell count o0.0001 NS 0.0001 NS Absolute neutrophil count o0.0001 NS 0.0035 NS Absolute lymphocyte count 0.0002 NS 0.0001 NS Absolute monocyte count (AMC) o0.0001 o0.0001 0.0001 AMC 410 Â 10(9)/l o0.0001 0.0004 Platelet count 0.0027 0.0002 NS Platelet count o100 Â 10(9)/ l o0.0001 0.0006 Circulating immature myeloid cells o0.0001 0.04 o0.0001 0.01 Peripheral blood blast % 0.002 NS o0.0001 0.0016 Bone marrow blast % 0.13 NS o0.0001 NS ASXL1 mutational status (%) o0.00001 o0.0001 0.2 SETBP1 mutational status (%) 0.07 NS 0.87 SRSF2 mutational status (%) 0.48 0.04 NS SF3B1 mutation status (%) 0.4 0.7 U2AF1 mutation status (%) 0.13 0.5 MD Anderson prognostic scoring system o0.0001 NS o0.0001 NS Mayo model prognostic score o0.0001 NS o0.0001 NS GFM model o0.0001 NS o0.0001 NS Spanish cytogenetic risk categories o0.0001 NS 0.0002 0.04 for high risk WHO morphological categories (CMML-1 vs CMML-2) 0.01 NS o0.0001 0.0005 Abbreviations: ASXL1, additional sex combs 1 gene; CMML, chronic myelomonocytic leukemia; GFM, Groupe Francais des Myelodsyplasies; NS, not signiﬁcant; SF3B1, splicing factor 3B, subunit 1; SRSF2, serine/arginine-rich splicing factor 2; U2AF1, U2 small nuclear RNA auxiliary factor 1; WHO, World Health Organization.

DISCUSSION The CALR/ASXL1 mutation-based prognostic model was DIPSS- The ASXL1 gene contains 13 exons and maps to chromosome plus independent (Po0.0001) and on a multivariable analysis was 20q11, regulating epigenetic functions (histone and chromatin found to be the strongest risk factor for survival (HR 3.7 vs 2.8 for 26 modification) and transcription.17 In CMML, B40% of age 465 years, vs 2.7 for unfavorable karyotype). In systemic patients carry ASXL1 mutations, with the most frequent being mastocytosis, ASXL1 mutations were seen in 9 (14%) of 62 patients the c.1934dupG; p.G646WfsX12 (B 50%).5,10 Although some and predicted for a shorted OS, independent of ‘C’ findings, investigators considered c.1934dupG; p.G646WfsX12 to be a associated hematological nonmast cell subtypes, AMC and 27 PCR artifact,18 subsequent studies have demonstrated its absolute neutrophil count. Similarly, molecular aberrations absence in germline DNA and control DNA.19,20 Furthermore, were identified in 24 (89%) of 27 patients with aggressive patients with c.1934dupG; p.G646WfsX12 were found to be systemic mastocytosis, with abnormalities of TET2, ASXL1, SRSF2 phenotypically similar to those with other ASXL1 mutations; and CBL being frequent occurrences.28 In acute myeloid leukemia, accordingly, c.1934dupG; p.G646WfsX12 is now considered to ASXL1 mutations have been found to be mutually exclusive with be a bona fide ASXL1 mutation.5,19 the favorable NPM1 mutations, with some,29,30 but not all,31 ASXL1 mutations are common in myeloid neoplasms, including studies demonstrating an independent prognostic impact. MDS,8,21 CMML,4,5,10 primary myelofibrosis,8,22 and acute myeloid Metzeler et al.,32 in a study involving 423 cytogenetically normal leukemia,21,23 with respective mutational frequencies in the range acute myeloid leukemia patients, demonstrated an adverse of 15–20%, 40–50%, 20–35% and 5–10% (30% in patients with prognostic impact of ASXL1 mutations in only the ELN secondary acute myeloid leukemia).19 In general, ASXL1 mutations (European Leukemia Net) favorable subgroup of patients. are associated with an aggressive phenotype.4,22,23 In MDS, Thol Similar to the above-discussed scenario with other myeloid et al.20 identified ASXL1 mutations in 40 (20%) of 193 patients, with neoplasms, there has been an effort to integrate molecular these mutations predicting for a shortened OS, independent of abnormalities into clinical prognostic models in CMML. In IPSS (International Prognostic Scoring System) risk stratification, particular, two recent studies produced conflicting results with karyotype, transfusion dependance and IDH1 mutations. Similarly, regard to the prognostic role of ASXL1 mutations in CMML.5,10 The Bejar et al.24 identified ASXL1 mutations in 63 (14.4%) of 439 MDS key difference between the two studies was the inclusion of ASXL1 patients and found these to be IPSS-independent predictors for missense mutations in the Mayo Clinic study but not in the French shortened OS. In a large (879 patients) primary myelofibrosis consortium study. In the current collaborative study between the collaborative study, ASXL1 mutations were identified in 20% of two groups, the exclusion of missense mutations unmasked the patients and were associated with older age, presence of significantly adverse impact of ASXL1 mutations in both the Mayo constitutional symptoms, leukocytosis and circulating blasts.25 Clinic and French consortium patient cohorts. This information is ASXL1 mutations were found to be DIPSS-plus (Dynamic important for interpretation of future and previous studies looking International Prognostic Scoring System) independent into the prognostic impact of ASXL1 mutations in myeloid prognosticators for shortened OS.25 Similarly, in a prognostic neoplasms in general and CMML in particular. study involving 570 primary myelofibrosis patients, survival was In the current study, 164 (39%) patients had nonsense/ best in the presence of CALR and absence of ASXL1 mutations frameshift ASXL1 mutations, with 65 (40%) harboring the (CALRmut,ASXL1wt) and worst otherwise (CALRwt,ASXL1mut).26 c.1934dupG; p.G646WfsX12 variant. ASXL1 mutations were associated

Leukemia (2014) 2206 – 2212 & 2014 Macmillan Publishers Limited Molecular prognostication in CMML MM Patnaik et al 2211 Survival data for 466 CMML patients Mayo Low risk CMML patients stratified stratified by Mayo Prognostic Model. by ASXL1 mutational status.

Low risk CMML patients, n = 101, 1 median survival 67 months. Intermediate risk CMML patients, n= 171, Low risk patients with out ASXL1 mutations, median survival 48 months. 0.8 n= 66, median survival 99 months High risk CMML patients, n= 194, Low risk patients with ASXL1 mutations, median survival 19 months. n= 26, median survival 44 months 0.6 Survival P=<0.0001 0.4 Survival P=0.001 0.2

0 0 25 50 75 100 125 150 175 200 0 20 40 60 80 100 120 140 160 Months Months

Mayo Intermediate risk CMML patients stratified Mayo High risk CMML patients stratified by ASXL1 mutational status by ASXL1 mutational status

1 1 High risk patients with out ASXL1 mutations, Intermediate risk patients with out ASXL1 mutations, n= 85, median survival 25 months n= 105, median survival 59 months 0.8 0.8 High risk patients with ASXL1 mutations, n= 80, median survival 17 months Intermediate risk patients with ASXL1 mutations, n= 58, median survival 30 months 0.6 0.6

Survival P=0.02 Survival 0.4 0.4 P=0.03

0.2 0.2

0 0 0 25 50 75 100 125 150 175 200 0 20 40 60 80 100 120 140 160 Months Months Figure 2. (a) Survival data for 466 patients with CMML stratified by the Mayo prognostic model. (b) Mayo prognostic model, low-risk CMML patients further stratified by ASXL1 mutational status (frameshift and nonsense mutations only). (c) Mayo prognostic model, intermediate-risk CMML patients further stratified by ASXL1 mutational status (frameshift and nonsense mutations only). (d) Mayo prognostic model, high-risk CMML patients further stratified by ASXL1 mutational status (frameshift and nonsense mutations only).

Survival data for 420 patients with Survival data for 420 patients with CMML stratified by the Molecular Mayo Model. CMML stratified by the GFM Model.

MMM low risk patients, n= 64, GFM low risk patients, n= 188, 1 median survival 65 months. 1 median survival 97 months. MMM intermediate-1 risk patients, GFM intermediate risk patients, n= 128, median survival 59 months. n= 154, median survival 28 months. MMM intermediate-2 risk patients, 0.8 GFM high risk patients, n= 78, 0.8 n= 118, median survival 31 months. median survival 17 months. MMM high risk patients, n= 110, median survival 16 months. 0.6 0.6

Survival 0.4 Survival 0.4 P=<0.0001 P=<0.0001

0.2 0.2

0 0 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 Months Months Figure 3. (a) Survival data for 420 patients with CMML stratiﬁed by the Molecular Mayo Model. (b) Survival data for 420 patients with CMML stratiﬁed by the Groupe Francais des Myelodsyplasies model.

& 2014 Macmillan Publishers Limited Leukemia (2014) 2206 – 2212 Molecular prognostication in CMML MM Patnaik et al 2212 with a proliferative phenotype, including higher WBC, higher AMC and are strongly associated with atypical CML, monosomy 7, isochromosome 5,10,19 and presence of circulating IMCs. The significant clustering of i(17)(q10), ASXL1 and CBL mutations. Leukemia 2013; 27: 1852–1860. ASXL1 mutations with U2AF1 mutations was previously recognized 12 Such E, Cervera J, Costa D, Sole F, Vallespi T, Luno E et al. Cytogenetic risk strati- by Tefferi et al.33 in the setting of primary myelofibrosis. Most fication in chronic myelomonocytic leukemia. Haematologica 2011; 96: 375–383. importantly, ASXL1 mutations were shown to carry adverse 13 Onida F, Kantarjian HM, Smith TL, Ball G, Keating MJ, Estey EH et al. Prognostic prognostic effect for OS but not LFS. In multivariable analysis that factors and scoring systems in chronic myelomonocytic leukemia: a retrospective included CMML-relevant risk factors, the prognostic significance of analysis of 213 patients. Blood 2002; 99: 840–849. ASXL1 mutations was sustained, along with AMC 410 Â 10(9)/l, 14 Lasho TL, Jimma T, Finke CM, Patnaik M, Hanson CA, Ketterling RP et al. SRSF2 presence of circulating IMCs, hemoglobin o10 g/dl and platelet mutations in primary myelofibrosis: significant clustering with IDH mutations and independent association with inferior overall and leukemia-free survival. Blood count 100 Â 10(9)/l. A similar analysis for LFS identified AMC o 2012; 120: 4168–4171. 410 Â 10(9)/l, presence of circulating blasts, presence of circulating 15 Patnaik MM, Lasho TL, Hodnefield JM, Knudson RA, Ketterling RP, Garcia-Manero G et IMCs, WHO CMML subcategories and karyotype as independent risk al. SF3B1 mutations are prevalent in myelodysplastic syndromes with ring side- factors. The inclusion of ASXL1 mutational status to the Mayo roblasts but do not hold independent prognostic value. Blood 2012; 119:569–572. prognostic model produced a robust MMM that delineated risk 16 Yoshida K, Sanada M, Shiraishi Y, Nowak D, Nagata Y, Yamamoto R et al. categories with median survivals ranging from 16 to 97 months. In Frequent pathway mutations of splicing machinery in myelodysplasia. Nature comparison with the GFM model, the MMM was superior in risk 2011; 478:64–69. stratifying low-risk patients (median OS 97 months with MMM vs 65 17 Tefferi A. Novel mutations and their functional and clinical relevance in months with GFM). Finally, in the current international study of a myeloproliferative neoplasms: JAK2, MPL, TET2, ASXL1, CBL, IDH and IKZF1. large cohort of patients with CMML, SETBP1 mutations did not Leukemia 2010; 24: 1128–1138. affect either OS or LFS. 18 Abdel-Wahab O, Kilpivaara O, Patel J, Busque L, Levine RL. The most commonly reported variant in ASXL1 (c.1934dupG;p.Gly646TrpfsX12) is not a somatic alteration. Leukemia 2010; 24: 1656–1657. CONFLICT OF INTEREST 19 Gelsi-Boyer V, Brecqueville M, Devillier R, Murati A, Mozziconacci MJ, Birnbaum D. Mutations in ASXL1 are associated with poor prognosis across the spectrum of The authors declare no conflict of interest. malignant myeloid diseases. J Hematol Oncol 2012; 5:12. 20 Thol F, Friesen I, Damm F, Yun H, Weissinger EM, Krauter J et al. Prognostic significance of ASXL1 mutations in patients with myelodysplastic syndromes. ACKNOWLEDGEMENTS J Clin Oncol 2011; 29: 2499–2506. This study is supported in part by grants from the ‘Myeloproliferative Disorders 21 Boultwood J, Perry J, Pellagatti A, Fernandez-Mercado M, Fernandez-Santamaria C, Foundation, Chicago, IL, USA’ and ‘The Henry J. Predolin Foundation for Research in Calasanz MJ et al. Frequent mutation of the polycomb-associated gene ASXL1 in Leukemia, Mayo Clinic, Rochester, MN, USA’. the myelodysplastic syndromes and in acute myeloid leukemia. Leukemia 2010; 24: 1062–1065. 22 Carbuccia N, Murati A, Trouplin V, Brecqueville M, Adelaide J, Rey J et al. Mutations REFERENCES of ASXL1 gene in myeloproliferative neoplasms. Leukemia 2009; 23: 2183–2186. 1 Swederlow S, Campo E, Harris NL, Jaffe ES, Stefano PA, Stein H et al. (eds). 23 Carbuccia N, Trouplin V, Gelsi-Boyer V, Murati A, Rocquain J, Adelaide J et al. WHO Classification of Tumours of Haematopoietic and Lymphoid Tissues. IARC: Mutual exclusion of ASXL1 and NPM1 mutations in a series of acute myeloid Lyon, 2008. leukemias. Leukemia 2010; 24: 469–473. 2 Patnaik MM, Parikh SA, Hanson CA, Tefferi A. Chronic myelomonocytic leukaemia: 24 Bejar R, Stevenson K, Abdel-Wahab O, Galili N, Nilsson B, Garcia-Manero G et al. a concise clinical and pathophysiological review. Br J Haematol 2014; 165: Clinical effect of point mutations in myelodysplastic syndromes. N Engl J Med 273–286. 2011; 364: 2496–2506. 3 Damm F, Itzykson R, Kosmider O, Droin N, Renneville A, Chesnais V et al. 25 Vannucchi AM, Lasho TL, Guglielmelli P, Biamonte F, Pardanani A, Pereira A et al. SETBP1 mutations in 658 patients with myelodysplastic syndromes, chronic Mutations and prognosis in primary myelofibrosis. Leukemia 2013; 27: 1861–1869. myelomonocytic leukemia and secondary acute myeloid leukemias. Leukemia 26 Tefferi A, Guglielmelli P, Lasho TL, Rotunno G, Finke C, Mannarelli C et al. CALR and 2013; 27: 1401–1403. ASXL1 mutations-based molecular prognostication in primary myelofibrosis: an 4 Gelsi-Boyer V, Trouplin V, Adelaide J, Bonansea J, Cervera N, Carbuccia N et al. international study of 570 patients. Leukemia 2014; 28: 1494–1500. Mutations of polycomb-associated gene ASXL1 in myelodysplastic syndromes 27 Damaj G, Joris M, Chandesris O, Hanssens K, Soucie E, Canioni D et al. ASXL1 but and chronic myelomonocytic leukaemia. Br J Haematol 2009; 145: 788–800. not TET2 mutations adversely impact overall survival of patients suffering 5 Itzykson R, Kosmider O, Renneville A, Gelsi-Boyer V, Meggendorfer M, Morabito M systemic mastocytosis with associated clonal hematologic non-mast-cell diseases. et al. Prognostic score including gene mutations in chronic myelomonocytic PLoS One 2014; 9: e85362. leukemia. J Clin Oncol 2013; 31: 2428–2436. 28 Schwaab J, Schnittger S, Sotlar K, Walz C, Fabarius A, Pfirrmann M et al. 6 Laborde R, Patnaik M, Lasho T, Finke C, Hanson C, Knudson R et al. Comprehensive mutational profiling in advanced systemic mastocytosis. Blood SETBP1 mutations in 415 patients with primary myelofibrosis or chronic 2013; 122: 2460–2466. myelomonocytic leukemia: independent prognostic impact in CMML. Leukemia 29 Patel JP, Gonen M, Figueroa ME, Fernandez H, Sun Z, Racevskis J et al. Prognostic 2013; 27: 2100–2102. relevance of integrated genetic profiling in acute myeloid leukemia. N Engl J Med 7 Meggendorfer M, Roller A, Haferlach T, Eder C, Dicker F, Grossmann V et al. SRSF2 mutations in 275 cases with chronic myelomonocytic leukemia (CMML). 2012; 366: 1079–1089. Blood 2012; 120: 3080–3088. 30 Schnittger S, Eder C, Jeromin S, Alpermann T, Fasan A, Grossmann V et al. ASXL1 8 Abdel-Wahab O, Pardanani A, Patel J, Wadleigh M, Lasho T, Heguy A et al. exon 12 mutations are frequent in AML with intermediate risk karyotype and are Concomitant analysis of EZH2 and ASXL1 mutations in myelofibrosis, chronic independently associated with an adverse outcome. Leukemia 2013; 27: 82–91. myelomonocytic leukemia and blast-phase myeloproliferative neoplasms. 31 Shen Y, Zhu YM, Fan X, Shi JY, Wang QR, Yan XJ et al. Gene mutation patterns and Leukemia 2011; 25: 1200–1202. their prognostic impact in a cohort of 1185 patients with acute myeloid leukemia. 9 Piazza R, Valletta S, Winkelmann N, Redaelli S, Spinelli R, Pirola A et al. Blood 2011; 118: 5593–5603. Recurrent SETBP1 mutations in atypical chronic myeloid leukemia. Nat Genet 32 Metzeler KH, Becker H, Maharry K, Radmacher MD, Kohlschmidt J, Mrozek K et al. 2013; 45: 18–24. ASXL1 mutations identify a high-risk subgroup of older patients with primary 10 Patnaik MM, Padron E, LaBorde RR, Lasho TL, Finke CM, Hanson CA et al. cytogenetically normal AML within the ELN Favorable genetic category. Blood Mayo prognostic model for WHO-defined chronic myelomonocytic leukemia: 2011; 118: 6920–6929. ASXL1 and spliceosome component mutations and outcomes. Leukemia 2013; 27: 33 Tefferi A, Finke CM, Lasho TL, Wassie EA, Knudson R, Ketterling RP et al. U2AF1 1504–1510. mutations in primary myelofibrosis are strongly associated with anemia and 11 Meggendorfer M, Bacher U, Alpermann T, Haferlach C, Kern W, Gambacorti-Pas- thrombocytopenia despite clustering with JAK2V617F and normal karyotype. serini C et al. SETBP1 mutations occur in 9% of MDS/MPN and in 4% of MPN cases Leukemia 2014; 28: 431–433.

Supplementary Information accompanies this paper on the Leukemia website (http://www.nature.com/leu)

Leukemia (2014) 2206 – 2212 & 2014 Macmillan Publishers Limited