IKZF1 belongs to the IKAROS family of transcription factors. It contains four zinc fingers at the N-terminal that directly bind to DNA at the core motif A/GGAAA and additional two zinc fingers at the C-terminal required for forming homo- and hetero-dimerization between different IKZF proteins [1]. DNA binding activity of IKZF1 can be enhanced by its dimerization, so both DNA-binding and dimer-forming defects alter IKZF1 function. IKZF1 deletions and mutations have been reported to affect B-cell precursor ALL and contribute to its poor prognosis [2]. IKZF1 alterations are less studied in AML. Recurrent IKZF1 deletions have been identified in AML [3, 4], but whether IKZF1 mutations affect AML in general remains unknown. Herein, we analyzed IKZF1 mutation in AML.

A total of 193 adult AML patients, who subjected to TES, were retrospectively analyzed in our center (01/05/2018–29/02/2020), while APL was excluded. Among these patients, 100 were male and 93 were female, and the median age was 56 (range 18–82). A total of 169 patients were diagnosed with de novo AML, 10 with refractory/relapsed AML, 6 with MDS/AML, 5 with MLL (5 de novo cases), and 3 with MS/AML (1 de novo case, 2 refractory/relapsed cases). The panel of TES included 236 genes recurrently mutated in hematological malignancies, and TES was displayed by NovaSeq platform (Illumina). The average raw sequencing depth on target per sample was ≥ 1000, and VAF ≥ 1% was considered significant. For TES, 184 samples were collected from BM and 9 samples from PB. In addition, fusion gene screening for common rearrangements in AML was employed.

IKZF1 mutation affected 2.6% (5/193) or 1.8% (3/169) of all AML patients or de novo AML patients from our cohort, respectively (Fig. 1a and Table 1). Totally, 7 different types of IKZF1 mutations were found, and 5 were frameshift or nonsense mutations, while 2 were missense mutations (Fig. 1b). Interestingly, IKZF1 mutation was absent in fusion gene-positive AML, while IKZF1 mutation co-occurred with PTPN11, SF3B1, bi-allelic CEBPA or WT1 mutation in our study (Fig. 1c). Their association was further determined by Chi-square test with continuity correction, and OR was calculated. In 154 fusion gene-negative patients, we found that SF3B1 and bi-allelic CEBPA but not PTPN11 or WT1 mutations were the significant concomitant genetic alteration with IKZF1 mutation (P < 0.05; OR > 1) (Fig. 1d–f). In clinic, treatment response was evaluated in 4/5 patients with IKZF1-mutated AML, and CR was achieved in 2 patients. Notably, 2 primary chemotherapy-resistant patients had IKZF1, PTPN11 and SF3B1-mutated AML, so this subtype of AML seemingly exhibited an aggressive clinical course. However, the impact of IKZF1 mutation in AML could not be determined in our study due to limited positive cases and short follow-up duration.

Fig. 1
figure 1

IKZF1 mutation in AML. a The frequency of IKZF1 mutation in our AML cohort and literature reports. b The type of IKZF1 mutation identified in our cohort and literature reports. c Genetic lesion in AML with IKZF1 mutation. d The frequency of PTPN11, SF3B1, bi-allele CEBPA and WT1 mutations in IKZF1mut-AML and IKZF1wt-AML from our cohort, respectively. e The frequency of IKZF1 mutation in PTPN11, SF3B1, bi-allele CEBPA, WT1-wild type/-mutated AML from our cohort, respectively. f The statistical significance of associations between IKZF1 and other gene mutations in our study was assessed by Chi-square test with continuity correction. Odds ratio was also calculated to define whether the correlation was positive or negative. g The frequency of SF3B1, FLT3, EZH2, KRAS, BCOR and DNMT3A mutations in IKZF1mut-AML and IKZF1wt-AML from OHSU study, respectively. h The frequency of IKZF1 mutation in SF3B1, FLT3, EZH2, KRAS, BCOR and DNMT3A-wild type/-mutated AML from OHSU study, respectively. i Associations between IKZF1 and other gene mutations in OHSU study were also analyzed as f indicated. NS, P ≥ 0.05; *, P < 0.05; **, P < 0.01; ***, P < 0.001

Full size image
Table 1 Acute myeloid leukemia with IKZF1 mutation in our cohort
Full size table

In addition to our study, we also used the cBioPortal tool to analyze the frequency of IKZF1 mutation in other three independent studies (OHSU [5], TCGA [6] and TARGET [7]). The frequency was 1.35% (8/593), 0.5% (1/200) and 4.21% (4/95), respectively, while the relatively high frequency in our study was possibly attributed to the criterion of enrollment and the limited cases (Fig. 1a). In total, 13 mutations were found in these studies, but there were no patients with 2 different mutations simultaneously (Additional file 1: Table S1). Of these 13 patients, 6 had frameshift or nonsense mutations and the rest 7 had missense mutations. IKZF1N195S was a hotspot mutation with the frequency of 38.5% (5/13), but it was absent in COSMIC and our study (Fig. 1b). Due to limited positive cases in TCGA and solely pediatric cases in TARGET, we further analyzed the related genetic events of IKZF1 mutations in OHSU and found that SF3B1 and EZH2, but not KRAS, BCOR, FLT3 or DNMT3A mutations were the significant concomitant alteration with IKZF1 mutation (Fig. 1g–i). Remarkably, SF3B1 mutation appeared in both concomitant alteration lists of our study and OHSU, suggesting their strong synergy in leukemogenesis.

Compared to AML, IKZF1 alteration is well studied in ALL. Churchman et al. reported that IKZF1 alteration affected 25% of childhood and 44% of young adult pre-B-cell ALL, especially BCR-ABL1-positive ALL with frequency of over 80%. In ALL, the most common type of alterations in IKZF1 is deletions, whereas IKZF1 mutations accounted only 2.6% of childhood and 3.4% of young adult ALL. The latter were observed in 11.9% of BCR-ABL1-negative and 2.2% of BCR-ABL1-positive ALL cases [8]. The frameshift or nonsense mutations of IKZF1 often occurred at the N-terminal or the region between DNA binding and dimerization domains, while missense mutations affected both domains. Consistently, IKZF1 mutations followed the same pattern in AML. Similarly to IKZF1N159S in AML, IKZF1N159Y is a hotspot mutation in ALL that affects its DNA binding domain. IKZF1N159Y-ALL exhibited one unique transcriptional profile characterized by downregulation of B-cell receptor and JAK-STAT signaling and upregulation of SALL1 [9]. Nevertheless, whether IKZF1N159S-AML could be defined as one independent subtype remains to be investigated.

In conclusion, besides of IKZF1 deletion, IKZF1 mutation is also recurrent in AML.