Prognostic impact of IKZF1 deletion in adults with common B-cell acute lymphoblastic leukemia

Interrogate the impact of IKZF1 deletion on therapy-outcomes of adults with common B-cell acute lymphoblastic leukemia. One hundred sixty-five consecutive adults with common B-cell ALL were tested for IKZF1 deletion and for BCR/ABL. Deletions in IKZF1 were detected using multiplex RQ-PCR, multiplex fluorescent PCR, sequence analysis and multiplex ligation-dependent probe amplification (MLPA). BCR/ABL was detected using RQ-PCR. All subjects received chemotherapy and some also received an allotransplant and tyrosine kinase-inhibitors. Multivariate analyses were done to identify associations between IKZF1 deletion and other variables on non-relapse mortality (NRM), cumulative incidence of relapse (CIR), leukemia-free survival (LFS) and survival. Amongst subjects achieving complete remission those with IKZF1 deletion had similar 5-year non-relapse mortality (NRM) (11 % [2–20 %] vs. 16 % [4–28 %]; P = 0.736), a higher 5-year cumulative incidence of relapse (CIR) (55 % [35–76 %] vs. 25 % [12–38 %]; P = 0.004), and worse 5-year leukemia-free survival (LFS) (33 % [16–52 %] vs. 59 % [42–73 %]; P = 0.012) and survival (48 % [33–62 %] vs. 75 % [57–86 %]; P = 0.002). In multivariate analyses IKZF1 deletion was associated with an increased relapse (relative risk [RR] =2.7, [1.4–5.2]; P = 0.002), a higher risk of treatment-failure (inverse of LFS; RR = 2.1, [1.2–3.6]; P = 0.007) and a higher risk of death (RR = 2.8, [1.5–5.5]; P = 0.002). The adverse impact of IKZF1 deletion on outcomes was stronger in subjects without vs. with BCR-ABL1 and in subjects receiving chemotherapy-only vs. an allotransplant. IKZF1 deletion was independently-associated with a higher relapse risk and worse LFS and survival in adults with common B-cell ALL after adjusting for other prognostic variables and differences in therapies. These data suggest IKZF1 deletion may be a useful prognostic variable in adults with common B-cell ALL, especially in persons without BCR-ABL1 and those receiving chemotherapy-only. Transplants appear to overcome the adverse impact of IKZF1 deletion on therapy-outcomes but confirmation in a randomized study is needed. The trial was registered in 2007 with the Beijing Municipal Government (Beijing Municipal Health Bureau Registration N: 2007–1007).


Background
About 85 % of cases of adult acute lymphoblastic leukemia (ALL) develop from B-cells. Identification of biological determinants of treatment-outcomes is incomplete. Consensus prognostic variables include age, gender, WBC level at diagnosis, cytogenetic abnormalities, BCR-ABL1, central nervous system (CNS) leukemia, time to initial response and detection of measureable residual disease (MRD) after induction and consolidation therapies [1][2][3][4][5]. However, analyses of receiver-operator characteristic (ROC) curves indicate only about one-half of the variability in subject-level outcomes is explained by known prognostic variables [6].
Genome-wide analyses indicate IKZF1 mutations are common in ALL and are associated with BCR-ABL1 [7,8]. IKZF1 encodes the Ikaros transcription factor, a member of a family of zinc-finger nuclear proteins required for normal lymphoid development [9]. Intra-genic deletions in IKZF1 generate aberrant isoforms [8,10]. The role of IKZF1 deletions is extensively-evaluated in children with ALL and is the most common genetic marker associated with a poor prognosis [11][12][13][14][15]. IKZF1 mutations are also common in adults with B-cell ALL and are also associated with poor prognosis in persons with concurrent BCR-ABL1 [16][17][18]. Few data are available on the role of IKZF1 deletions in adults with common B-cell ALL. In preliminary analyses we found frequent IKZF1 deletions in this population. Based on these data we interrogated associations between IKZF1 deletion and prognosis in a series of adults with common B-cell ALL.

Subjects
Between April, 2007 and December, 2012 165 newlydiagnosed subjects ≥14 years old with common B-cell ALL were seen at Peking University People's Hospital, Peking University Institute of Hematology (Beijing, China). 3 subjects with severe co-morbidities were ineligible for induction chemotherapy (2 had an IKZK1 deletion) and excluded from further analyses. The 162 eligible subjects received remission-induction chemotherapy. Those achieving complete remission received 2 cycles of consolidation chemotherapy after which they could continue receiving chemotherapy or receive an allotransplant based on donor availability, finances and physician and subject choice. Physicians and subjects were blinded to results of IKZF1testing. Subjects were followed until death, loss to follow-up or December, 2014. Written informed consent was obtained from adult subjects and from parents of minors in accordance with the Declaration of Helsinki.

Diagnosis and response criteria
Diagnosis of ALL was based on ≥20 % bone marrow lymphoblasts. Common-B-cell type was defined as cases with CD10, CD19, CD22 and CD79a expression and no surface immunoglobulin by flow cytometry [19]. Complete remission was defined bone marrow lymphoblasts < 5 %, neutrophils >1.0 × 10E + 9/L, platelets >100 × 10E + 9/L, no extra-medullary disease and no loss of these features for >4 weeks. Relapse was defined as leukemia recurrence at any site in persons in complete remission.

Chemotherapy
One hundred six subjects without BCR/ABL received induction chemotherapy with 1-2 cycles of CODPL (cyclophosphamide, vincristine, daunorubicin, prednisone and L-asparaginase). 56 subjects with BCR-ABL1 received induction chemotherapy with the same regimen without L-asparaginase and 48 also received imatinib, 400 mg/d, beginning at diagnosis. 5 were later switched to dasatinib, 1 to nilotinib and 1 to ponatinib because of resistant mutations. Data regarding IKZF1 deletion were not used to determine induction chemotherapy. Subjects achieving complete remission (CR) received 2 cycles consolidation chemotherapy with hyper-CVAD (cyclophosphamide, vincristine, doxorubincin, dexamethasone, cytarabine and methotrexate). All subjects received CNS prophylaxis with intrathecal methotrexate, cytarabine and dexamethasone for ≥8 doses during induction and consolidation therapy. Details of these regimens are reported [24]. Subjects remaining in CR continued receiving chemotherapy until the completed 6 more cycles, received an allotransplant or relapsed. Post-consolidation therapy was with 6-mercaptopurine, 50 mg/mE2/d PO and methotrexate, 20 mg/mE2/d PO, each once weekly. 28 subjects with BCR-ABL1 receiving an allotransplant also received posttransplant tyrosine kinase-inhibitor (TKI) therapy. 17 subjects with BCR-ABL1 receiving post-consolidation chemotherapy but no allotransplant also received postconsolidation TKI therapy. Decisions to give or not to give TKI therapy post-consolidation or posttransplant were based on finances and subject and/or physician preference and without knowing results of IKZF1 deletion-testing.

Allotransplants
Ninety-two subjects (64 %) received an allotransplant, 24 (26 %) from an HLA-identical sibling and 68 (74 %) from a HLA-haplotype-mismatched related donor (1 HLA-antigen mismatch [N = 4], 2 HLA-antigen mismatches [N = 20] and 3 HLA-antigen mismatches [N = 44]). Conditioning was with busulfan and cyclophosphamide. Anti-thymocyte globulin was given pretransplant when the donor was a HLA-haplotype-mismatched relative. Donors received recombinant human G-CSF after which bone marrow and blood cells were collected and infused into the recipient. Graft-vs.-host disease prophylaxis was with cyclosporine, mycophenolatemofetil and short-term methotrexate. Haematological relapses were treated with donor mononuclear cells infusions. Some subjects received a 2 nd transplant from the same or a different donor. Details are reported [25,26].

Endpoints and statistical analyses
Haematologic response was analyzed weekly for the first 3 months and monthly thereafter. Cytogenetic and molecular responses were analyzed every 3 months for the first 6 months and every 6-12 months thereafter. Leukemia-free survival (LFS) was calculated from the date of 1 st complete remission to the date of first relapse or death in complete remission. Survival was calculated from 1 st complete remission to the date of death from any cause. Observations were censored at the date of last contact or December, 2014 when no events were observed. In survival analyses subjects receiving an allotransplant who relapsed were censored at the time they received a donor blood cell infusion or a second allotransplant. Last follow-up was December, 2014. Relapse was defined as 1 st hematologic relapse regardless of site. Independence of categorized parameters was calculated using Chi-square test (or Fisher Exact test). Distribution of continuous variables was calculated using Wilcoxon two sample tests. Survival functions were estimated by the Kaplan-Meier method and compared by the log-rank test. Cumulative incidences were estimated for NRM and relapse to accommodate competing risks. NRM was a competing risk for relapse and death from any cause was a competing risk for relapse. A Cox proportional hazard regression model was used to determine associations between IKZF1 deletion and other variables with relapse, NRM, treatment-failure and death. The variables to be considered in the multivariate models were: IKZF1 deletion (Y/N), BCR-ABL1 (Y/N), gender, age (≥ vs. < 35 years), WBC (≥ vs. < 30 × 10E + 9/L), hypo-diploid karyotype (Y/N), complex karyotype and treatment (chemotherapy vs. allotransplant). CNS, MLL rearrangement or E2A-PBX1 transcripts and TKI were not considered in the multivariate analysis because of either too few subjects in a category or missing data. Assumption of proportional hazards for each factor in the Cox model was tested using time-dependent covariates. A stepwise model selection approach was used to identify all significant risk factors. Each step of model building contained the main effect for IKZF1 deletion. Factors which are significant at a P = 0.05 level were kept in the final model. Potential interactions between main effect and all significant risk factors were tested. Analyses were performed by SPSS software version 19.0 (Chicago, IL, USA), Graphpad Prism 5.01 (San Diego, California, USA) and R version 3.1.2. Difference with P < 0.05 was considered significant.

Discussion
IKZF1 deletion was independently-associated with higher risks of relapse, treatment-failure and death in adults with common B-cell ALL compared with similar subjects without IKZF1 deletion. Although IKZF1 deletion was confounded with BCR-ABL1, the adverse impact persisted after adjusting for BCR-ABL1 and parallels similar findings in children with B-cell ALL [12,15]. The adverse impact of IKZF1 deletion operated mainly in subjects without BCR-ABL1. In subjects receiving an allotransplant the adverse impact of IKZF1 deletion was less than in those receiving chemotherapy-only suggesting a benefit for using transplants in subjects with IKZF1 deletion.
Our analysis was complex because of confounding between IKZF1 deletion, BCR-ABL1 and post-remission therapy (chemotherapy-only vs. an allotransplant). We used multivariate analyses to resolve this confounding. We found IKZF1deletion, chemotherapy-only, gender and age were significantly associated with outcomes whereas other variables including WBC, cytogenetics and BCR-ABL1 were not. This is surprising, especially the lack of a significant association between BCR-ABL1 with outcomes. There are several possible explanations. One is insufficient statistical power; there were relatively few subjects with adverse cytogenetics, hypo-diploidly or a complex karyotype. However, low power cannot explain the lack of a significant association between BCR-ABL1 and outcomes. As indicated, IKZF1 deletion and BCR-ABL1 were confounded. Subjects with BCR-ABL1 received different initial therapy than subjects without BCR-ABL1 including TKIs in 90 % and an allotransplant in 58 %. If these interventions were highly-effective they could mitigate the adverse impact of BCR-ABL1. This notion is supported by the observation the adverse predictive impact of IKZF1 deletion was detected in subjects without but not in those with BCR-ABL1 and in subjects receiving chemotherapy-only but not in those receiving an allotransplant.
Our conclusions contrast with data from a study of 83 subjects with BCR-ABL1 which reported IKZF1 deletions were associated with more relapses and worse LFS [16]. However, there are several important differences in subjects (we studied only subjects with common B-cell ALL) and post-remission therapy (few of their subjects received an allotransplant). Our conclusion is similar to reports in children with BCR-ABL1 in whom IKZF1 deletion is associated with worse outcomes and with results of a recent meta-analysis indicating IKZF1 deletion is independentlyassociated with worse outcomes in children and adults with ALL [27,28].
There are several limitations to our study. One, as discussed, is confounding between IKZF1 deletion, BCR-ABL1 and post-remission therapy. Confounding is difficult to satisfactorily sort out in multivariate analyses but our data suggest that IKZF1 deletion was independentlyassociated with higher risks of relapse, treatment-failure and death in subjects without BCR-ABL1. Second, subjects with BCR-ABL1 received different induction, consolidation and post-remission therapies than those without BCR-ABL1. However, therapy-assignment was made without knowing IKZF1 deletion data and we tried to account for this complexity in multivariate analyses. Another limitation is we did not include several potentially important variables (CNS leukemia, MLL rearrangement, E2A-PBX1 transcripts and TKI therapy) in the multivariate analyses because of too few subjects or missing data. The sum of these potential limitations means our conclusions require confirmation in prospective, randomized studies.

Conclusion
In adults with common B-cell ALL achieving complete remission IKZF1 mutation was independently-associated with a higher CIR and worse survival than subjects without IKZF1 mutation. These differences were greater in subjects receiving post-remission chemotherapy than in subjects receiving an allotransplant. We suggest a randomized trial to confirm our observation.