FAB subtypes of NPM1mut AML
In this study, we picked out 238 patients with NPM1mut AML for our purpose analysis. The study cohort consisted of 105 males and 133 females, with a median age of 49 (range 15–81) years. The most common FAB subtypes of NPM1mut AML were AML-M5 (59.7%), followed by M2 (17.6%) and M4 (15.5%), similar to other findings.(8, 9) According to FLT3-ITD, M2 is more common in NPM1mut/FLT3-ITD(−) group [34/143 (23.8%) vs. 8/95 (8.4%); P = 0.002], while M5 is slightly more common in NPM1mutFLT3-ITD(+) group [64/95 (67.4%) vs. 78/143 (54.5%); P = 0.048], as shown in Fig. 1a.
The expression incidence of CD34 and CD7 in NPM1mut/FLT3-ITD(+) group were higher than in NPM1mut/FLT3-ITD(−) group.
As per literature(33), The leukemic blasts at the initial diagnosis could be divided into leukemic myeloid blasts and leukemic immature monocyte populations, with the latter detected in about 50% cases and mostly in M4 or M5 morphologic subtypes. The leukemic myeloid blasts recurred when AML relapsed, while leukemic immature monocyte populations often disappeared, indicating that the leukemic myeloid blasts may enrich more LSCs which serve as source of disease relapse. Consequently, in the description of baseline FCM characteristic, we only analyzed the antigen expression aspects of leukemic myeloid cells. In entire NPM1mut cohort, the antigens positively expressed at an incidence of 80% or more were CD117 (211/235, 89.8%), CD13 (207/233, 88.8%), CD33 (233/233, 100%), CD123 (230/232, 99.1%) and CD38 (209/233, 89.7%). The positive incidence of CD34 and TDT were 31.5% (74/235) and 6.9% (16/231), respectively. The positive incidence of HLA-DR was 67.4% (157/233) and MPO was 74.9% (57/227). CD7 was positively expressed in 43.1% (94/218), CD19 in 3.5% (8/228) and CD79a in 0.4% (1/227) cases, as shown in Fig. 1b. According to FLT3-ITD, the positive incidence of CD34 and CD7 in NPM1mut/FLT3-ITD(+) were significantly higher than in NPM1mut/FLT3-ITD(−) group (CD34: 47.9% vs. 20.6%, P < 0.001; CD7: 61.5% vs. 29.9%, P < 0.001), while the positive incidence of other antigens were not different between the two genotypic groups.
Chromosomal karyotypes in NPM1mut AML
Of all 238 patients with NPM1mut, 234 patients had evaluable metaphases, in whom 208 (88.9%) were normal karyotypes (NK) and 26 (11.1%) were abnormal karyotypes. Among 143 cases with NPM1mut/FLT3-ITD(−), 140 had evaluable metaphases, with 131 cases in the intermediate-risk group (including 121 cases NK; 10 cases intermediate-risk abnormal karyotype, Fig. 1c), 5 cases in the favorable-risk group [including 4 cases t(8;21)(q22;q22); 1 case inv (16) (p13q22)] and 4 cases in the adverse-risk group (including 1 case complex karyotype, monosomy karyotype, t(6;9)(p23;q34), t(8;9;22)(q24;q34;q11.2) for each]. Among 95 cases with NPM1mut/FLT3-ITD(−), 94 had evaluable metaphases, with all of them in the intermediate-risk group (including 87 cases NK; 7 cases intermediate-risk abnormal karyotype, Fig. 1c) and none in the favorable- or adverse-risk group. There was no difference in the distribution of both NK and intermediate-risk abnormal karyotype between NPM1mut/FLT3-ITD(−) and NPM1mut/FLT3-ITD(+) group (P = 0.144 and 0.930, respectively), while the favorable- and adverse-risk karyotypes were only enriched in the NPM1mut/FLT3-ITD(−) group and not in the NPM1mut/FLT3-ITD(+) group (6.4% vs. 0%; P = 0.031). No correlation was found between other coexisting gene mutations and abnormal karyotypes (all P > 0.05, data not shown). No KMT2A (MLL) or TP53 rearrangement was identified in 206 NPM1mut patients with available FISH data.
NPM1 mut loci, types and co-mutation patterns
In entire NPM1mut cohort, totaling 240 NPM1 mutant events were identified, among whom 230 (230/240, 95.8%) were out-of-frame indels. Ten cases (10/240, 4.2%) were detected to habour NPM1mut missense events, i.e., 3 cases with c.578A > G→p.K193R, 2 with c.676G > A→p.E226K and 5 with c.733G > C→p.E245Q. All these missense codons did not disrupt any of the tryptophan residues W288 and W290 which are indispensible for the nucleolar localization signal (NoLS). Furthermore, all but one of these missense mutations (9/10, 90.0%) were accompanied by an AML subtype-defining recurrent cytogenetic or molecular abnormality, with 7 cases in the favorable risk and 2 in adverse risk (Table 1).
Table 1
AML subtype-defining cytogenetic or molecular abnormalities accompanied by NPM1mut missense mutations
NPM1mut missense | AML subtype-defining cytogenetic or molecular abnormalities |
c.578A > G→p.K193R | t(8;21)(q22;q22),-Y |
c.578A > G→p.K193R | inv(16)(p13q22) |
c.578A > G→p.K193R | t(6;9)(p23;q34) |
c.676G > A→p.E226K | NPM1mut [p.W288Cfs*12] |
c.676G > A→p.E226K | CEBPAdm [p.E309_D322indelsGQTQQKVLELTS; p.D80Gfs] |
c.733G > C→p.E245Q | t(8;21)(q22;q22) |
c.733G > C→p.E245Q | Complex: +8,-14,i(17)(q10),+2 ~ 3mar |
c.733G > C→p.E245Q | NPM1mut [p.W288Cfs*12] |
c.733G > C→p.E245Q | CEBPAdm [p.K313dupK; p.Q87X] |
c.733G > C→p.E245Q | — |
Abbreviations: NPM1mut, NPM1 mutation; CEBPAdm, double allele mutation of CEBPA |
At least one co-mutation was detected in all 238 NPM1mut cases. Including NPM1mut, the median number of mutated genes per individual was 4.5 (2–14), with 4.0 (2–14) in the NPM1mut/FLT3-ITD(−), insignificantly different from 5.0 (2–10) in the NPM1mut/FLT3-ITD(+) group (P = 0.378, Fig. 2a). According to gene function categories, the order of incidence is as follows: signaling pathways (72.7%), epigenetic regulators (71.4%), tumor suppressors (31.9%) and myeloid transcription factors (8.8%, Fig. 2b). DNMT3A (104, 43.7%), FLT3-ITD (95, 39.9%) and FAT1 (57, 23.9%) represented the top three most frequently mutated genes (more details on relatively common genes with a mutation frequency > 5% are listed in Fig. 3). Analysis of gene-gene relationship across NPM1mut coexisting mutations showed that FLT3-ITD and DNMT3A had significant accompaniment (P = 0.005), while FLT3-ITD was mutually exclusive of FLT3-nonITD (P < 0.001), NRAS (P < 0.001), PTPN11 (P = 0.017) and IDH1 (P = 0.005, Fig. 4).
Association between NPM1mut coexisting mutations and immunophenotypic markers
Our results have shown that the expression of CD34 and CD7 is significantly associated with FLT3-ITD. Because NPM1mut AML mostly occurs in the context of NK, we hypothesize that the diversity of antigen expression in leukemia cells to a certain extent is determined by heterogeneity of coexisting mutations. In order to rule out the influence of abnormal karyotype on immunophenotype, as well in view of the deductively insufficient pathogenicity of NPM1mut missense mutations, only patients with NK and NPM1mut indels type were included for subsequent analysis. A total of 205 NPM1mut patients meeting the above conditions were submitted to the distributional crosstabulation between immunophenotypic markers and coexisting mutations. The significant results from χ2 test and multivariate analysis are shown in Table 2. Logistic analysis showed that, in entire NPM1mut cohort, FLT3-ITD was positively correlated with CD34 and CD7 (HR = 5.29 [95% CI 2.64–10.60], P < 0.001; HR = 3.47 [95% CI 1.79–6.73], P < 0.001). Ras-pathway mutation was positively correlated with HLA-DR expression (HR = 4.05 [95% CI 1.70–9.63], P = 0.002) and negatively with MPO expression (HR = 0.18 [95% CI 0.05–0.62], P = 0.007) in entire NPM1mut cohort. Stratified analysis indicated that this effect was only seen in NPM1mut/FLT3-ITD(−) group (HR and P values are detailed in Table 2), but not in NPM1mut/FLT3-ITD(−) group.
Table 2
Correlation between immunophenotypic markers with mutations of coexisting genes in NPM1mut AML
Association | NPM1mut/FLT3-ITD(−) | | NPM1mut/FLT3-ITD(+) | | Entire cohort |
χ2 P | HR (95%CI) | P | χ2 P | HR (95%CI) | P | χ2 P | HR (95%CI) | P |
CD34 (N = 202) | ×FLT3-ITD | NA | NA | NA | NA | NA | NA | < 0.001 | 5.29 (2.64–10.60) | < 0.001 |
×DNMT3A | NS | NA | NA | 0.028 | 2.60 (1.00-6.79) | 0.051 | 0.026 | NA | NA |
×TET2/IDH1 | NS | NA | NA | 0.005 | 0.21 (0.06–0.71) | 0.012 | 0.001 | 0.26 (0.11–0.62) | 0.002 |
CD7 (N = 186) | ×FLT3-ITD | NA | NA | NA | NA | NA | NA | < 0.001 | 3.47 (1.79–6.73) | < 0.001 |
×DNMT3A | 0.008 | NA | NA | 0.007 | 3.30 (1.15–9.46) | 0.026 | < 0.001 | NA | NA |
×DNMT3A-R882 | 0.002 | 3.93 (1.61–9.59) | 0.003 | 0.009 | NA | NA | < 0.001 | 3.59 (1.80–7.16) | < 0.001 |
×TET2/IDH1 | 0.048 | NA | NA | 0.001 | 0.18 (0.05–0.60) | 0.005 | NS | 0.30 (0.14–0.62) | 0.001 |
HLA-DR (N = 200) | ×Ras pathways | 0.002 | 3.83 (1.40-10.46) | 0.009 | 0.055 | NA | NA | < 0.001 | 4.05 (1.70–9.63) | 0.002 |
×DNMT3A-R882 | < 0.001 | 26.77 (3.44-208.46) | 0.002 | < 0.001 | 8.65 (2.28–32.89) | 0.002 | < 0.001 | 13.41 (4.56–39.45) | < 0.001 |
×TET2/IDH1 | NS | NA | NA | 0.002 | 0.26 (0.09–0.78) | 0.016 | 0.046 | NA | NA |
MPO (N = 196) | ×KRAS | 0.002 | 0.13 (0.03–0.56) | 0.006 | NS | NA | NA | 0.003 | 0.18 (0.05–0.62) | 0.007 |
×DNMT3A | 0.003 | NA | NA | 0.071 | NA | NA | < 0.001 | 0.35 (0.17–0.70) | 0.003 |
×DNMT3A-R882 | 0.002 | 0.27 (0.10–0.74) | 0.011 | NS | NA | NA | 0.001 | NA | NA |
×TET2/IDH1 | 0.040 | NA | NA | 0.021 | 4.32 (1.16–16.15) | 0.029 | 0.001 | 3.52 (1.48–8.38) | 0.004 |
APL-like (N = 198) | ×Ras pathways | 0.008 | 0.32 (0.11–0.96) | 0.041 | 0.025 | NA | NA | < 0.001 | 0.22 (0.08–0.57) | 0.002 |
×DNMT3A-R882 | < 0.001 | NA | NA | < 0.001 | 0.04 (0.01–0.36) | 0.004 | < 0.001 | 0.02 (0.00-0.18) | < 0.001 |
×TET2/IDH1 | NS | NA | NA | < 0.001 | 6.73 (1.83–24.78) | 0.004 | 0.008 | 2.26 (1.07–4.78) | 0.033 |
Abbreviations: NS, no significance; NA, not applicable |
DNMT3A-R882 was positively correlated with CD7 and HLA-DR expression (HR = 3.59 [95% CI 1.80–7.16], P < 0.001; HR = 13.41 [95% CI 4.56–39.45], P < 0.001), and DNMT3A mutation negatively with MPO expression (HR = 0.35 [95% CI 1.48–8.38], P = 0.004). Stratified analysis according to FLT3-ITD status indicated that the independent correlation of DNMT3A mutation (especially DNMT3A-R882) with CD7 and with HLA-DR expression was significant both in NPM1mut/FLT3-ITD(+) and in NPM1mut/FLT3-ITD(−) groups (HR and P values are detailed in Table 2). TET2/IDH1 mutations were negatively correlated with CD34 and CD7 expression (HR = 0.26 [95% CI 0.11–0.62], P = 0.002; HR = 0.30 [95% CI 0.14–0.62], P = 0.001), and positively with MPO expression (HR = 3.52 [95% CI 1.48–8.38], P = 0.004). Stratified analysis indicated the above effects to be prominent only in the NPM1mut/FLT3-ITD(+) group (HR and P values are detailed in Table 2), but not in the NPM1mut/FLT3-ITD(−) group. There were no significant correlations between NPM1mut coexisting mutations with other antigen expressions.
We finally analyzed association of NPM1mut coexisting mutations with APL-like phenotype CD34(−)/HLA-DR(−)/MPO(str+), which has been reportedly to predict a presence of TET2/IDH1 mutation.(12) In entire cohort, mutations of Ras-pathway, DNMT3A-R882 and TET2/IDH1 were each significantly linked with APL-like phenotype. When layered by FLT3-ITD, in NPM1mut/FLT3-ITD(−) group, only Ras-pathway mutation presented an association with APL-like phenotype (HR = 0.32 [95% CI 0.11–0.96], P = 0.041). Comparatively, negative correlation of DNMT3A-R882 (HR = 0.04 [95% CI 0.01–0.36], P = 0.004) and positive correlation of TET2/IDH1 mutation (HR = 6.73 [95% CI 1.83–24.78], P = 0.004) with this phenotype were seen only in NPM1mut/FLT3-ITD(+) group, but not in NPM1mut/FLT3-ITD(−) group (Table 2).