The clinical features of IKZF1-mutated AML
In total, 530 newly diagnosed AML patients were involved in our study, and the mutational study was displayed for all of them (Figure S1). IKZF1 mutation was identified in 22 patients (4.15%, 22/530) (Table 1). Though no bias of distribution in gender was found in IKZF1-mutated (IKZF1MUT) and IKZF1-wild type (IKZF1WT) group, AML patients with IKZF1MUT showed one relatively young median age compared to those with IKZF1WT (42.5 years of IKZF1MUT vs. 50 years of IKZF1WT, P = 0.027). IKZF1 mutation preferred to affect de novo AML (19/22, 86.4%), while the frequency of MLL in IKZF1MUT group (2/22, 9.1%) was significantly higher than it in IKZF1WT group (6/508, 1.2%) (P = 0.012). Though AML with IKZF1 mutation mostly presented the morphologic subtype of AML-M2 (7/22, 31.8%), M4 (5/22, 22.7%), or M5 (5/22, 22.7%), no imbalanced distribution was found according to French-American-British (FAB) classification between two groups. In addition, there were no significant differences in the level of white blood cell (WBC) (P = 0.171), hemoglobin (HB) (P = 0.656), platelet (PLT) (P = 0.887) as well as the percentage of BM blast (P = 0.339) between IKZF1MUT and IKZF1WT groups (Table 2). Therefore, besides of younger morbid age and high frequency of MLL involvement, the basic clinical presentation of IKZF1MUT AML patients was similar to IKZF1WT patients.
Table 1
IKZF1-mutated AML patients in our cohort
UPN
|
Age/Sex
|
Diagnosis
|
Karyotype
|
IKZF1 mutation
|
Other genetic mutation
|
MutFreq
|
cHGVS
|
pHGVS
|
1
|
60/M
|
de novo AML
|
46,XY
|
0.01
|
c.1233delC
|
p.L411fs
|
MPL, CEBPA, CCDC168
|
0.20
|
c.1054dupA
|
p.H351fs
|
2
|
52/F
|
de novo AML
|
46,XX,t(8;21)
|
0.02
|
550C > T
|
p.R184W
|
NRAS, KRAS, ASXL2
|
3
|
45/M
|
de novo AML
|
46,XY
|
0.45
|
c.49_50insGC
|
p.S17fs
|
NOTCH1, DNMT3A, CSF3R
|
4
|
30/F
|
de novo AML
|
46,XX
|
0.03
|
c.1505_1511del
|
p.R502fs
|
CEBPA
|
0.02
|
637C > T
|
p.R213X
|
5
|
23/M
|
de novo AML
|
46,XY,9q-
|
0.44
|
c.663delG
|
p.E221fs
|
SRCAP, FLT3-ITD, CEBPA
|
6
|
52/F
|
de novo AML
|
45–47,XX,der(1),der(5),7p-,+8,r(12),-13,13q-,-15,18q-,-22,+20q-[cp18]/46,XX[2]
|
0.23
|
476A > G
|
p.N159S
|
TP53
|
7
|
55/F
|
de novo AML
|
46,XX
|
0.40
|
c.909_910del
|
p.N303fs
|
NRAS, KIT, CEBPA
|
8
|
34/F
|
de novo AML
|
NA
|
0.04
|
c.253_260del
|
p.L85fs
|
KRAS, KIT, FLT3-ITD, CSF3R, CEBPA
|
9
|
40/M
|
de novo AML
|
46,XY,t(11;19)
|
0.46
|
573C > A
|
p.H191Q
|
WT1, TET2, NRAS, KIT, FAT1, CEBPA
|
10
|
23/M
|
de novo AML
|
46,XY
|
0.06
|
815C > T
|
p.A272V
|
STAG2, NRAS, NF1
|
11
|
28/F
|
de novo AML
|
45,X,-X
|
0.11
|
484C > T
|
p.R162W
|
NRAS, KIT, EZH2, CSF3R, CEBPA
|
12
|
60/F
|
MLL
|
46,XX[20]
|
0.70
|
c.331C > T
|
p.R111X
|
NRAS, IDH2, DNMT3A
|
13
|
52/F
|
de novo AML
|
46,XX[20]
|
0.43
|
476A > G
|
p.N159S
|
SF3B1, PTPN11, ETNK1
|
14
|
61/M
|
de novo AML
|
47,XY,+3(q21)[20]
|
0.38
|
c.214G > T
|
p.E72X
|
SF3B1, PTPN11, FLT3, BCOR
|
0.42
|
c.1150delT
|
p.S384fs
|
15
|
45/M
|
de novo AML
|
46,XY,der(3)(q27)[10]
|
0.45
|
c.184_185insAA
|
p.Q62fs
|
SF3B1, PTPN11
|
0.47
|
550C > T
|
p.R184W
|
16
|
24/M
|
de novo AML
|
46,XY,del(8)(q22){5}/46,XY{5}
|
0.02
|
427C > T
|
p.R143W
|
WT1, CEBPA
|
0.03
|
637C > T
|
p.R213X
|
17
|
34/F
|
de novo AML
|
46,XX[20]
|
0.20
|
c.336delinsGCCCG
|
p.L112fs
|
WT1, CTCF, CSF3R, CEBPA
|
18
|
50/M
|
s/t-AML
|
46,XY[20]
|
0.04
|
c.637C > T
|
p.R213X
|
WT1, TET2, SF3B1, GATA2, DNMT3A
|
19
|
21/M
|
MLL
|
46,XY[20]
|
0.23
|
c.472G > A
|
p.G158S
|
GATA2, CEBPA, CCND3
|
20
|
23/M
|
de novo AML
|
46,XY[20]
|
0.49
|
472G > A
|
p.G158S
|
RUNX1, GATA2, CEBPA
|
21
|
15/F
|
de novo AML
|
46,XX,-2,-6,-11,del(13)(q13q22), +3mar[6]/46,XX[4]
|
0.03
|
c.520A > C
|
p.K174Q
|
CCND3
|
22
|
66/F
|
de novo AML
|
46,XX[20]
|
0.01
|
c.472G > A
|
p.G158S
|
/
|
0.14
|
c.482T > G
|
p.L161R
|
M, male; F, female; AML, acute myeloid leukemia; MLL, mix lineage leukemia; s/t-AML, secondary/therapy-related AML; NA, not available. |
Table 2
The baseline characteristics of our AML cohort.
Characteristic
|
IKZF1WT group
|
IKZF1MUT group
|
P
|
N, % of total
|
508, 95.8%
|
22, 4.2%
|
|
Age (years)
|
50 (11–82)
|
42.5 (15–66)
|
0.027
|
Gender
|
|
|
|
Male (N)
|
284 (55.9%)
|
11 (50%)
|
0.585
|
Female (N)
|
224 (44.1%)
|
11 (50%)
|
Peripheral blood
|
|
|
|
White blood cells (109/L)
|
11.35 (0.4-484.8)
|
17.72 (1.67-428.01)
|
0.171
|
Hemoglobin (g/L)
|
83 (2-204)
|
87.5 (57–148)
|
0.656
|
Platelets (109/L)
|
51 (2-565)
|
59 (7-917)
|
0.887
|
Bone marrow blasts (%)
|
59 (11.5–98)
|
61.9( 20–96)
|
0.339
|
Diagnosis (N)
|
|
|
|
De novo AML
|
481 (94.7%)
|
19 (86.4%)
|
0.012
|
Mixed lineage leukemia
|
6 (1.2%)
|
2 (9.1%)
|
Secondary/therapy-related AML
|
21 (4.1%)
|
1 (4.5%)
|
French-American-British (N)
|
|
|
|
M0
|
21 (4.1%)
|
2 (9.1%)
|
0.728
|
M1
|
27 (5.3%)
|
1 (4.5%)
|
M2
|
170 (33.5%)
|
7 (22.7%)
|
M4
|
101 (19.9%)
|
5 (22.7%)
|
M5
|
161 (31.7%)
|
5 (22.7%)
|
M6
|
11 (2.2%)
|
0 (0%)
|
M7
|
1 (0.2%)
|
0 (0%)
|
Undefine
|
16 (3.1%)
|
2 (9.1%)
|
Cytogenetics (N)
|
|
|
|
Normal karyotype
|
250 (49.2%)
|
12 (54.5%)
|
0.886
|
Complex karyotype
|
44 (8.7%)
|
2 (9.1%)
|
0.736
|
Monosomal karyotype
|
16 (3.1%)
|
0 (0%)
|
0.793
|
-5/5q-/monosomy 5
|
21 (4.1%)
|
0 (0%)
|
0.637
|
-7/monosomy 7
|
17 (3.3%)
|
0 (0%)
|
0.758
|
-17/17p abnormalities
|
11 (2.2%)
|
0 (0%)
|
0.989
|
Chromosome 3 abnormalities
|
16 (3.1%)
|
2 (9.1%)
|
0.417
|
Gene fusions (N)
|
|
|
|
RUNX1-RUNX1T1
|
65 (12.8%)
|
1 (4.5%)
|
0.376
|
CBFB-MYH11
|
36 (7.1%)
|
0 (0%)
|
0.365
|
BCR-ABL1
|
10 (2.0%)
|
0 (0%)
|
0.489
|
KMT2A rearrangements
|
19 (3.7%)
|
0 (0%)
|
0.712
|
European Leukemia Net 2017 (N)
|
|
|
|
Low
|
151 (29.7%)
|
3 (13.6%)
|
0.005
|
Intermediate
|
214 (42.1%)
|
17 (77.3%)
|
High
|
143 (28.2%)
|
2 (9.1%)
|
Complete remission (N)
|
400 (84.6%)
|
15 (68.2%)
|
0.041
|
No complete remission (N)
|
73 (15.4%)
|
7 (31.8%)
|
The genetic features of IKZF1-mutated AML
In our cohort, 28 IKZF1 mutations were identified in 22 patients. In detail, 25 mutations were identified in 19 de novo AML patients (19/22, 86.4%), 2 mutations were in 2 MLL patients (2/22, 9.1%), and 1 mutation was in 1 secondary/therapy-related AML patient (1/22, 4.5%). Among those mutations, 13 were missense mutation, 5 were nonsense mutation, and 10 were frame-shift mutation. Missense mutation preferred to localize at the exon 5 (92.31%, 12/13), which influenced the DNA-binding of IKZF1, while only one affected the exon 7 (7.69%, 1/13). 40% (6/15) of nonsense mutation and frame-shift mutation disrupted the DNA-binding domain and lost the dimerization domain, while 60% (9/15) of them only disrupted the dimerization domain. Both of the DNA binding domain and the dimerization domain were required for the functional integrity of IKZF1, so all of above mutations possibly impaired its transcription activity (Fig. 1A and Table 1).
Cytogenetic aberrations were common in AML, and then we analyzed the relationship between IKZF1 mutation and them. Our results showed that normal karyotype (P = 0.886), complex karyotype (P = 0.736), monosomal karyotype (P = 0.793), -5/5q-/monosomy 5 (P = 0.637), and − 17/17p abnormalities (P = 0.989) did not preferred to co-occur with IKZF1 mutation significantly. -7/monosomy 7 led to IKZF1 haploid-insufficiency, and no IKZF1 mutation affected those patients, indicating that IKZF1 mutation possibly impaired the function of IKZF1WT, and did not need to disrupt the WT allele by loss of chromosome 7. IKZF1 mutation preferred to locate in AML with chromosome 3 abnormalities, but only two IKZF1MUT patients were found in this sub-group and no significant distribution preference of IKZF1 mutation was exhibited (P = 0.417). At the same time, IKZF1 mutation also showed rare co-occurrence with common gene fusion in AML, such as RUNX1-RUNX1T1, CBFβ-MYH11, BCR-ABL1, or KMT2A rearrangement, while only one IKZF1MUT and RUNX1-RUNX1T1-positive patient was identified in our cohort (Table 2). Therefore, the partner of IKZF1 mutation, which contributed to leukemogenesis, was not such cytogenetic aberrations.
Then, we analyzed the relationship of IKZF1 mutation with other gene mutations. IKZF1MUT significantly exhibited more preferable to co-occur with CEBPA (P < 0.000), SF3B1 (P = 0.001), and CSF3R (P = 0.008) mutation than IKZF1WT. Besides, NRAS (P = 0.645), WT1 (P = 0.260), GATA2 (P = 0.093), KIT (P = 0.638), EZH2 (P = 0.333), PTPN11 (P = 0.648), DNMT3A (P = 0.670), and TET2 (P = 0.987) mutations were exhibited to co-occur with IKZF1 mutation, but no preferable bias was found between IKZF1MUT and IKZF1WT groups. In contrast, IKZF1 mutation seemingly exhibited mutually exclusion with NPM1 (P = 0.063), ASXL1 (P = 0.171), IDH1 (P = 0.232), IDH2 (P = 0.576), TP53 (P = 0.730) SRSF2 (P = 1) or U2AF1 (P = 0.616) mutation though without significance (Fig. 1B and 1C). So, our results strongly indicated that the possible cooperation of IKZF1 mutation with CEBPA, SF3B1 or CSF3R mutation existed in AML pathogenesis.
The prognostic role of IKZF1 mutation in AML
ELN 2017 prognostic stratification well predicted the clinical outcome of AML patients17. Though compared to IKZF1WT group, IKZF1MUT group showed one high frequency of patients belonging to ELN-intermediate-risk group (77.3% of IKZF1MUT vs. 42.1% of IKZF1WT), and low frequency of ELN-low-risk (13.6% of IKZF1MUT vs. 29.7% of IKZF1WT) and ELN-high-risk group (9.09% of IKZF1MUT vs. 28.1% of IKZF1WT) (P = 0.005), the rate of CR in IKZF1MUT group was significantly more inferior to it in IKZF1WT group (68.2% of IKZF1MUT vs. 84.6% of IKZF1WT, P = 0.041) under our treatment strategy (Table 2). To further determine the prognostic role of IKZF1 mutation in AML, we compared the OS as well as RFS for IKZF1MUT and IKZF1WT group. Our results indicated that IKZF1MUT patients showed similar OS (P = 0.76) and RFS (P = 0.64) with IKZF1WT patients (Fig. 2A and 2B). Though IKZF1 mutation conferred one disadvantaged therapeutic response for AML patients in our cohort, overall, it finally did not influence their survival duration.
To interpret the contrast phenomena and define the prognostic role of IKZF1 mutation further, we displayed the subgroup analysis for IKZF1MUT patients according to the VAF, the type of mutation, or the mutational count in single patient. In VAF-based stratification, IKZF1MUT patients with high VAF (> 0.230) exhibited one lower CR rate than IKZF1MUT patients with low VAF (≤ 0.230), though without significance (Table S1). Notably, IKZF1MUT patients with high VAF showed one significantly inferior OS compared to those with low VAF or IKZF1WT (P = 0.047), but their RFS did not exhibit the statistic difference (P = 0.4) (Fig. 2C and 2D). However, the type of mutations, missense mutation or frame-shift mutation as well as nonsense mutation, did not influence the OS (P = 0.95) and RFS (P = 0.9) of IKZF1MUT patients (Figure S1C and S1D), while the mutational count in single patient, 1 or 2, also exhibited no significance in OS (P = 0.93) and RFS (P = 0.28) (Figure S1E and S1F). Therefore, high burden of IKZF1 mutation predicated one poor prognostic role in AML.
Univariate And Multivariate Analysis For Overall Survival Duration
To assess the true contribution of high burden of IKZF1 mutation to the poor prognosis of AML, we displayed univariate and multivariate analysis for OS, in which the baseline characteristics and genetic alterations were included. In univariate analysis, we identified 17 factors influencing OS of our AML cohort significantly, including IKZF1 mutation with high VAF. When they were submitted to multivariate analysis, our results strongly indicated that IKZF1 mutation with high VAF was one independent risky factor for the death of AML (hazard ratio [HR], 3.14; 95% CI, 1.27–7.75; P = 0.013). In the baseline characteristics, advanced age and high WBC count predicted increased risk of death, while bone marrow transplantation predicted the decreased risk. In the genetic alterations, core-binding factor rearrangement, IDH2 mutation, and bi-allele CEBPA mutation predicated one relatively good prognosis, while DNMT3A mutation, TP53 mutation, and aberrant karyotype exhibited the inverse effect on the OS of AML (Table 3). Thus, our results supported that the independent role of high burden of IKZF1 mutation on the poor OS of AML.
Table 3
Univariate and multivariate analysis for overall survival duration
Variable
|
Univariate
|
Multivariate
|
HR
|
95% CI
|
P
|
HR
|
95% CI
|
P
|
ELN adverse risk
|
1.72
|
1.19–2.47
|
0.00000001
|
-
|
-
|
-
|
TP53 mutation
|
2.70
|
1.39–5.21
|
0.000004
|
1.98
|
1.20–3.28
|
0.008
|
CBF-AML
|
0.37
|
0.26–0.52
|
0.00002
|
0.40
|
0.23–0.68
|
0.00065
|
Advanced age
|
2.01
|
1.32–3.04
|
0.000046
|
1.64
|
1.13–2.39
|
0.010
|
Aberrant karyotype
|
1.94
|
1.18–3.18
|
0.0007
|
2.11
|
1.34–3.32
|
0.0013
|
High WBC count
|
1.65
|
1.20–2.28
|
0.0019
|
1.97
|
1.38–2.80
|
0.00017
|
DNMT3A mutation
|
1.73
|
1.11–2.70
|
0.0032
|
1.59
|
1.05–2.41
|
0.030
|
BMT
|
0.60
|
0.42–0.84
|
0.0078
|
0.56
|
0.36–0.85
|
0.0068
|
MLL
|
2.20
|
0.88–5.50
|
0.013
|
1.85
|
0.94–3.65
|
0.078
|
KRAS mutation
|
1.75
|
0.99–3.10
|
0.016
|
1.60
|
0.95–2.68
|
0.076
|
ASXL2 mutation
|
0.49
|
0.31–0.78
|
0.020
|
0.55
|
0.27–1.13
|
0.103
|
BCOR mutation
|
1.73
|
0.92–3.26
|
0.030
|
1.19
|
0.69–2.04
|
0.539
|
IKZF1 mutation with high VAF
|
2.38
|
0.70–8.17
|
0.031
|
3.14
|
1.27–7.75
|
0.013
|
Bi-allele CEBPA mutation
|
0.53
|
0.34–0.85
|
0.034
|
0.46
|
0.24–0.87
|
0.017
|
SF3B1 mutation
|
2.03
|
0.65–6.35
|
0.082
|
1.73
|
0.68–4.41
|
0.248
|
SRCAP mutation
|
0.55
|
0.32–0.95
|
0.095
|
0.56
|
0.27–1.17
|
0.124
|
IDH2 mutation
|
0.65
|
0.38–1.12
|
0.191
|
0.39
|
0.20–0.79
|
0.009
|
ELN, European Leukemia Net; CBF, core-binding factor; WBC, white blood cells; BMT, bone marrow transplantation; MLL, mixed lineage leukemia. |
The prognostic role of IKZF1 mutation in specific genetic AML subtypes
To analyze the prognostic role of IKZF1 mutation in some genetic subtype of AML, we further compared the clinical outcome between CSF3R, CEBPA or SF3B1-mutated patients with or without IKZF1 mutation due to their recurrent co-existences. In our cohort, CSF3R mutation did not confer one significant difference in the OS (P = 0.95) or RFS (P = 0.21) of AML (Figure S2A and S2B), and IKZF1 mutation also did not influence the OS (P = 0.92) or RFS (P = 0.35) of CSF3RWT or CSF3RMUT patients (Fig. 3A and 3B, Table S2). Bi-allele CEBPA mutation conferred one good prognostic role for AML patients18–21. Consistent with it, AML patients in our cohort with bi-allele CEBPA mutation showed one extremely good OS compared to those without it (P = 0.034), though RFS was similar between two subgroups (P = 0.23) (Figure S2C and S2D). In IKZF1MUT patients, bi-allele CEBPA mutation (11/13, 84.6%) was far more than single-allele CEBPA mutation (2/13, 15.4%). Interestingly, IKZF1 mutation conferred one relatively lower CR rate in CEBPAWT/single-allele CEBPAMUT group rather than bi-allele CEBPAMUT group (Table S3). However, both of CEBPAWT/single-allele CEBPAMUT and bi-allele CEBPAMUT AML showed similar OS (P = 0.052) and RFS (P = 0.65) between with or without IKZF1 mutation in survival analysis (Fig. 3C and 3D). Besides, SF3B1 mutation also showed no influence on the OS (P = 0.082) or RFS (P = 0.65) of our patients (Figure S2E and S2F). Strikingly, IKZF1WT/SF3B1MUT AML patients exhibited one low CR (50%, P = 0.33), while the therapeutic response was even worse in IKZF1MUT/SF3B1MUT AML and none of them achieved CR all the course (0%, P < 0.001) (Table S4). Consistently, IKZF1 mutation combined with SF3B1 mutation conferred on extremely poor OS on AML (P = 0.002), but the RFS of IKZF1MUT/SF3B1MUT AML patients was unavailable due to no CR achievement (P = 0.846) (Fig. 3E and 3F). Thus, IKZF1 mutation conferred one poor prognosis when combined with SF3B1 mutation.