HLA Polymorphisms Are Associated with Treatment-Free Remission Following Discontinuation of Tyrosine Kinase Inhibitors in Chronic Myeloid Leukemia

This article is featured in Highlights of This Issue, p. 1

ABL tyrosine kinase inhibitors (TKI) markedly improve the prognosis of patients with chronic phase chronic myeloid leukemia (CML-CP; ref. 1). Several clinical trials investigating treatment-free remission (TFR) after TKI discontinuation revealed that approximately 40%–60% of the patients who achieved a durable deep molecular response (DMR) during TKI treatment show sustained molecular remission after TKI discontinuation (2–6). Male sex (2), long-term TKI treatment (2, 6, 7), deeper molecular remission status (undetectable BCR-ABL1 transcript levels) before TKI discontinuation (8, 9), and an increased number of natural killer (NK) cells (4, 10) and a reduced number of CD4-cells (11) were favorable prognostic factors for TFR in the patients with CML-CP. Recently, BCR-ABL1 transcriptional detection methods using the digital PCR (dPCR) have emerged (12) and may be a useful tool for prediction of TFR in patients with CML-CP (13, 14). Although TFR is now the therapeutic goal of CML-CP, the factors that predict TFR remain unclear because it has not been analyzed comprehensively.

Sensitive BCL-ABL1 transcriptional detection methods (15) can detect residual leukemic cells in patients who achieve TFR. Given that residual leukemic cells exist in these patients, the host immune system appears to prevent CML development or relapse mediated by residual leukemic cells (16); therefore, it is reasonable to assume that cancer immunosurveillance plays an important role in achievement of TFR. The human immune system comprises innate and adaptive arms, which provide functional diversity (16, 17). T cells are major components of the adaptive immune system; these antigen-specific lymphocytes undergo clonal expansion and induce target cell death. Allogenic stem cell transplantation (allo-SCT) has been used as a curative treatment for patients with CML-CP in the past. The beneficial effects of this treatment can be partly attributed to donor T-cell–mediated graft-versus-leukemia effects. Indeed, leukemia relapse is more common in CML cases undergoing T-cell–depleted allo-SCT (18). BCR-ABL1 chimeric proteins induce expansion of clonal human leucocyte antigen (HLA)-restricted cytotoxic T cells (CTL), which may control CML (19). Furthermore, the magnitude of HLA-restricted CTL depends on the HLA allele (20, 21). In particular, HLA-A*02:01, *24:02, and *11:01 can drive expansion of HLA-restricted cytotoxic T cells in patients with CML (22). Thus, T-cell immunity, possibly enhanced by particular polymorphisms, is indispensable for immune reaction in CML.

NK cells play a pivotal role in innate immunity against virally infected or malignant cells by discharging cytotoxic granules. Patients with newly diagnosed CML-CP show NK cell dysfunction; however, TKI treatment can restore NK cell function in patients with CML-CP (23), leading to molecular remission. Furthermore, CML recipients showing early recovery of NK cell populations after allo-SCT have a more favorable outcome (24), and the number of NK cells positively correlates with achievement of TFR (4, 25). Taken together, these findings suggest that NK cell–mediated immunity also plays an important role in CML.

NK cell activity is determined by the balance between activating and inhibitory signals induced via surface receptors (26). Among these, KIRs are the most polymorphic molecules that bind to MHC class I molecules [known as the human leukocyte antigen (HLA) system in human; ref. 27]. HLAs also harbor abundant polymorphisms; thus, the combination of KIRs and HLAs determines the magnitude of NK cell responses in patients with autoimmune diseases and cancers (28). Recently, next-generation sequencing (NGS) revealed that KIRs harbor abundant allelic polymorphisms that affect NK cell activity (29) as well as HLAs. We reported previously that allelic polymorphisms of KIRs and HLAs are associated with achievement of DMR in patients with CML-CP (30). We also found that high NK (11) and low CD4⁺ T-cell counts (4) were favorable prognostic factors for TFR in these patients.

However, it is unclear which cell-mediated immune response, T-cell or NK, is more important for achieving TFR in patients with CML, and no detailed analysis of polymorphisms in KIRs and HLAs in patients with CML after discontinuing TKIs [only KIR genotyping or haplotyping analysis has been reported; refs. 31, 32] has been undertaken. Therefore, we aimed to undertake subgroup analysis to examine the association between TFR and HLA and KIR polymorphisms in patients who discontinued TKIs, based on our previous analysis of DMR and KIRs/HLAs (30).

Patients with Philadelphia chromosome-positive CML-CP (n = 76) who were treated with TKIs (imatinib, dasatinib, or nilotinib) between April 2002 and August 2017 at Saga University Hospital were enrolled. NGS of DNA from the patients (n = 76) was performed to evaluate the association between KIR/HLA polymorphisms and TFR. CML was diagnosed according to the World Health Organization classification of myeloid neoplasms and acute leukemia (33). Baseline patient characteristics were obtained from hospital records; these included general characteristics (age and sex), laboratory data (complete cell counts, blast counts, and percentage of eosinophils and basophils), molecular diagnosis (BCR-ABL1 transcript levels), and spleen size. Written informed consent was obtained from all patients before registration, as required by the Declaration of Helsinki. This study was approved by the institutional review board of Saga University (UMIN-CTR, ID: R000020356).

Response criteria were defined according to BCR–ABL1 mRNA transcript levels [as measured by real time quantitative-PCR (RQ-PCR)] adjusted to the international reporting scale (IS), and/or using the transcription-mediated amplification (TMA) method. A major molecular response (MMR) was defined as ≤0.1% (on the IS) or a BCR-ABL1 transcript level of <50 copies/0.5 μg RNA (TMA). MR^4.0 was defined as ≤0.01% (on the IS) or undetectable BCR-ABL1 (TMA). MR^4.5 was defined as ≤0.0032% (on the IS). DMR was defined as MR^4.0 or a deeper response, and molecular relapse was defined as loss of DMR at two consecutive time points or loss of MMR at a single time point. The TMA method for quantification of BCR-ABL1 transcripts was the only method approved by the national insurance body of Japan until June 2015 (34), and before the implementation of international standards (11, 35) in our institute. This method was used only to confirm DMR before TKI discontinuation. Monitoring after TKI discontinuation was assessed by RQ-PCR, as stipulated by international standards. RQ-PCR analyses were done every 3 months during the consolidative TKI treatment (patients with durable DMR before TKI discontinuation), every month for the first year after TKI discontinuation, and every 3 months after the second year.

TKI was discontinued in patients who enrolled in TKI stop clinical trials (n = 12) or in those not in clinical trials (n = 21) who attended Saga University Hospital. Patients in clinical trials discontinued TKIs according to individual study protocols (4, 7, 11). Patients not in clinical trials discontinued TKIs if they so wished 20 patients met the criteria for TFR as a treatment option according to the European Society for Medical Oncology guidelines (36); one patient experienced adverse events after durable DMR after at least 3 years of TKI treatment.

Allelic genotyping of HLA and KIR genes was performed by NGS of DNA samples using Illumina MiSeq technology, as previously described (30).

The following monoclonal antibodies were used to identify their respective antigens: FITC–anti-CD3 (OKT3, TONBO Biosciences); PE-Cy7–anti-CD56 (HCD56) and Pacific Blue–anti-CD107a (H4A3; both from BioLegend); and APC-Cy7–anti-CD16 (3G8) and V450–IFNγ (B27; both from BD Biosciences). Fixation and permeabilization buffer sets (Thermo Fisher Scientific) were used to stain IFNγ. Stained samples were assessed using a FACSVerse cytometer (BD Biosciences) and data were analyzed using FlowJo software (Tree Star).

The activation status of NK cells was evaluated by flow cytometry analysis of CD107a degranulation and intracellular IFNγ secretion by CD3⁻CD16⁺CD56⁺ NK cells, as previously described (30). The experiments were performed at the next visit after registration in the study. The correlation between NK cell activation and achievement of DMR and/or TFR was examined.

Cumulative incidence probabilities were calculated using the Kaplan–Meier method and differences were analyzed using the log-rank test. Cox's proportional hazard model was used to evaluate TFR, and two-sided P values <0.05 were considered statistically significant. The Mann–Whitney U test was used to determine statistically significant differences between two groups or variables. All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University), a graphical user interface for R.

A total of 33 of 76 CML-CP patients discontinued TKIs. The median follow-up time was 7.61 years (range, 4.37–16.30 years); the median age was 65 years (range, 39–86 years); 17 patients were male and 16 were female; and 14, 12, and seven patients had low, intermediate, and high Sokal scores, respectively. The first-line TKI was imatinib in 17 cases, dasatinib in 11, and nilotinib in five. The TKI used before discontinuation was imatinib in nine cases, dasatinib in 17, and nilotinib in seven; eight patients who began treatment with imatinib were switched to second-generation TKIs due to suboptimal responses or adverse events. Overall, 21 of 33 patients achieved TFR [63.6%; 95% confidence interval (CI), 44.9%–77.5%] at 1 year (Fig. 1). The detailed clinical characteristics are summarized in Supplementary Table S1.

Univariate analysis of the clinical characteristics for TFR (including sex, age, Sokal-risk score, use of first-line TKIs, TKI used before discontinuation, DMR time before TKI discontinuation and duration of TKI treatment) identified male sex as a significant prognostic variable for a lower likelihood of molecular relapse [hazard ratio (HR), 0.141; 95% CI, 0.031–0.645; P = 0.012; Table 1; Fig. 2A], whereas age, Sokal-risk score, type of TKI, and TKI treatment duration had no effect.

Univariate analysis of KIR and HLA profiles revealed that KIR genotype, KIR allele type, KIR and HLA licensing status, KIR3DL1 expression levels, KIR3DL1 and HLA-Bw interaction avidity [determined by measuring the strength of interactions (strong, weak or no interaction) between KIR3DL1 (high, low or null) and HLA-B (HLA-Bw4-80T, 80I or Bw6); e.g., KIR3DL1high binds HLA-Bw4-80I more strongly than HLA-Bw4-80T as previously described; refs. 30, 37], and HLA-C did not affect the incidence of molecular relapse (Table 2). TFR in patients with HLA-Bw6 was lower than in patients with HLA-Bw4-80I (45.5%; 95% CI, 16.7–70.7; vs. 78.9%; 95% CI, 53.2–91.4; P = 0.043; HR, 3.496; 95% CI, 0.980–12.48; P = 0.054) and patients lacking HLA-A*02:01, *24:02, or *11:01 were more likely to experience molecular relapse (HR, 5.546; 95% CI, 1.885–20.092; P < 0.001; Table 2; Fig. 2B; Supplementary Table S2). Furthermore, homozygous (HLA-A*02:01, *24:02, or *11:01) individuals tended to experience less molecular relapse than heterozygous individuals [TFR; homozygous (n = 12), 91.7% (95% CI, 53.9–98.8); heterozygous (n = 14), 64.3% (95% CI 34.3–83.3); P = 0.12; Table 2; Fig. 2C]. These results indicate that HLA-A profile plays an important role in TFR, and the degree of heterozygosity may be correlated with TFR in patients with CML-CP.

Multivariate analysis identified male sex (HR, 0.157; 95% CI, 0.031–0.804; P = 0.003) and HLA-A*02:01, *11:01, or *24:02 (HR, 6.386; 95% CI, 1.701–23.980; P = 0.006) as independent favorable prognostic factors for TFR in patients with CML (Table 3), indicating that sex-specific or HLA-restricted T-cell–mediated immune responses may be involved in TFR.

Previously, we reported that the degree of NK cell activation contributes to achievement of MR^4.0. Here, we found that patients who discontinued TKIs (achievement of durable deep molecular remission; n = 20) exhibited higher levels of CD107a degranulation (Fig. 3A) and secretion of IFN-γ (Fig. 3B) than the patients who did not (n = 22). By contrast, there were no significant differences in CD107a degranulation and IFN-γ secretion between patients who achieved TFR (n = 13) and those who experienced molecular relapse (n = 8, Fig. 3C and D). These results suggest that NK cell activation contributes to achievement of DMR but not TFR.

Here, we identified that sex (male) and HLA polymorphisms (HLA-A*02:01, *24:02, and *11:01) were associated with TFR in the patients with CML-CP. The finding that male sex is a favorable predictive factor for TFR is consistent with the results of the STIM trial (2); however, the majority of TKI discontinuation studies suggest that a patient's sex does not affect TFR (2, 4–6), even in clinical trials using a sensitive dPCR method to detect BCR-ABL1 (13, 14). Immune cell populations (38) and immune responses (39) differ in males and females; indeed, female sex is a favorable prognostic factor for DMR in the patients who are treated with TKIs (30, 40, 41). Anyway, further investigation is still needed to conclude whether patient sex is involved in achievement of TFR.

Immune cell profiles, especially the magnitude of NK cell responses modulated by TKIs (42) in patients with CML-CP may determine the responses of TKIs treatment (23, 43). Previously, we reported that NK cell immune responses contribute to achievement of DMR (MR^4.0) in patients with CML-CP; this finding was associated with several favorable KIR allelic haplotypes and with NK cell activation status (30). Here, the patients with durable DMR exhibited higher NK cell activation (CD107a degranulation and secretion of IFNγ) than those with less molecular responses, indicating that NK cell activation with favorable KIR alleles (30) is necessary for achievement of durable DMR in the patients with CML-CP. By contrast, we found no differences in NK cell activation status and KIR polymorphisms between patients with TFR and those with molecular relapse, which indicates that different immune responses contribute to achievement of DMR and maintenance of TFR.

The results of prior clinical observational studies of associations between KIRs and hematological malignancies have often been inconsistent, due to patient heterogeneity (44, 45) or HLA-associated factors such as the association between downregulation of HLAs and immune escape (46). The small cohort of patients in our study could have resulted in several factors canceling the clinical impact of KIR polymorphisms in patients with CML-CP and preventing them from achieving TFR. We have begun an analysis examining the impact of KIR and HLA polymorphisms in a larger cohort.

HLA-restricted antigen presentation is assumed to be important for the induction of T-cell immune responses in patients with CML (22). Antigen-specific cytotoxic T cells are more strongly induced in patients with HLA-A*02:01, *24:02 or *A11:01, whereas they are less induced in patients with other HLA alleles (20, 21); we therefore focused on HLA-A*02:01, *24:02, and *11:01. Our results indicate the HLA profile, that is, A*02:01, *24:02, or *11:01 in particular, may be an independent prognostic factor for achieving TFR. However, these three HLA alleles did not affect achievement of DMR in patients with CML-CP (30). T cells are major components of adaptive immunity (16) and HLA-restricted cytotoxic T cells (CTLs) that recognize BCR-ABL1 are detectable in patients with CML-CP (19). The magnitude of HLA-restricted CTL depends on the HLA allele (20, 21). In particular, HLA-A*02:01, *24:02, and *11:01 can drive expansion of HLA-restricted cytotoxic T cells (CTLs), which may be involved in long-lasting adaptive immune responses (47), and successful achievement of TFR. We previously reported that T-cell profiles were associated with successful achievement of TFR (11). These results suggest HLA-restricted T-cell–immune responses may also be important for achieving TFR (unlike at induction therapy) in patients with CML-CP. Peptide vaccination to induce T-cell immune response against CML may have the potential to increase achievement of TFR (48).

Previous reports indicated that an increase in the number NK cells is associated with achievement of DMR (4, 10), suggesting that NK cells might also have a role in achieving TFR. On the basis of the results of this article, we would like to propose that the strong molecular responses of undetectable BCR-ABL1 transcript levels (8, 9) are associated with TFR. Therefore, a higher number of NK cells immediately before TKI discontinuation may be associated with TFR (4, 10), through higher probability of DMR. Currently, we assume that T cells, not NK cells, contribute to TFR, but this will require further verification.

Several surrogate markers for successful achievement of TFR have been investigated (49). Lower BCR-ABL1 transcript levels detected by digital PCR (13, 14), e14a2 BCR-ABL1 transcript type (50), longer TKI treatment duration, time of DMR, presence of withdrawal syndrome, a deeper molecular response, a lower Sokal score, and interferon α treatment before TKI administration may be associated with a favorable TFR (49). However, different study designs have generated inconsistent data; therefore, further investigations are needed to identify factors that consistently favor achievement of TFR.

Our results indicate that NK cell–mediated innate immunity plays a pivotal role in the achievement of DMR in patients with CML-CP, whereas T-cell–mediated adaptive immunity contributes to TFR, but this will require further confirmation in a future study. Indeed, we are now examining KIR and HLA polymorphisms in a larger cohort patients enrolled in the DADI (4), DOMEST (7), and first-line DADI (11) trials. Moreover, we are also undertaking a further prospective TFR trial to investigate the gene expression profiles and the NK and T-cell subpopulations responsible for successful achievement of TFR. Further understanding of the role of KIR and HLA polymorphisms in patients with CML-CP could help optimize treatment strategies (e.g., duration of consolidation phase) for the achievement of successful TFR.

S. Kimura reports grants and personal fees from Bristol-Myers Squibb, Pfizer, and Novartis Pharmaceuticals outside the submitted work. No disclosures were reported by the other authors.

H. Ureshino: Data curation, formal analysis, funding acquisition, investigation, writing-original draft, writing-review and editing. T. Shindo: Conceptualization, writing-original draft, writing-review and editing. H. Tanaka: Writing-review and editing. H. Saji: Supervision. S. Kimura: Conceptualization, supervision, funding acquisition, writing-review and editing.

This work was supported by research grants from JSPS KAKENHI (19K17860; to H. Ureshino) and JSPS KAKENHI (17K09908; to S. Kimura).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.