Early Disease Relapse after Tyrosine Kinase Inhibitor Treatment Discontinuation in CML Is Related Both to Low Number and Impaired Function of NK-Cells

Background: Recent reports suggest that approximately 40% of CML patients who have achieved sustained complete molecular remission are able to stop TKI treatment without disease relapse. However, there are no predictive markers for successful therapy discontinuation. Therefore, we set up an immunological sub-study in the ongoing pan-European EURO-SKI stopping study. Our aim was to identify predictive biomarkers for relapse/non-relapse and to understand more on the mechanisms of immune surveillance in CML.

Methods: The EURO-SKI study started in 2012, and patients included were at least three years on TKI and at least one year in MR4 or deeper before the study entry. Basic lymphocyte immunophenotyping (the number of NK-, T- and B-cells) was performed at the time of therapy discontinuation and 1, 6, and 12 months after the TKI stop and in case of relapse (defined as loss of MMR, BCR-ABL1>0.1% IS). In addition, from a proportion of patients more detailed immunophenotypic and functional analyses (cytotoxicity of NK-cells and secretion of Th1 type of cytokines IFN-γ/TNF-α) were done at the same times.

Results: Thus far 119 Nordic patients (imatinib n=105, dasatinib n=12, nilotinib n=2) who have discontinued TKI treatment within the EURO-SKI study have been included in the lymphocyte subclass analysis (results are presented from patients who have reached 6 months follow-up). Immunophenotyping analysis demonstrates that imatinib treated patients who were able to maintain remission for 6 months (n=36) had increased NK-cell counts (0.26 vs. 0.15x10⁹cells/L, p=0.01, NK-cell proportion 18.9% vs. 11%, p=0.005) at the time of drug discontinuation compared to patients who relapsed early (before 5 months n=22).

Furthermore, the phenotype of NK-cells was more cytotoxic (more CD57+ and CD16+cells and less CD62L+cells), and also their IFN-γ/TNF-α secretion was enhanced (19.2% vs. 13%, p=0.02). Surprisingly, patients who relapsed more slowly (after 5 months, n=16) had similar baseline NK-cell counts (0.37x10⁹cells/L), NK-cell proportion (21.2%), and phenotype and function as patients, who were able to stay in remission. No differences in the NK-cell counts were observed between patients who had detectable or undetectable BCR-ABL1 transcripts at the baseline (0.22 x10⁹cells/L vs. 0.31 x10⁹cells/L, p=0.61). Interestingly, NK-cell count was higher in patients with low Sokal risk score than in patients with intermediate risk (0.33 x10⁹cells/L vs. 0.20 x10⁹cells/L, p=0.04). Furthermore, there was a trend that male patients had a higher proportion of NK-cells than females (21.6% vs. 15.7%, p=0.06). Pretreatment with IFN-α or the duration of imatinib treatment did not have an effect on NK-cell count or proportion. In comparison to the imatinib group, dasatinib treated patients had higher NK-cell counts at the baseline (median 0.52x10⁹cells/L vs. 0.26x10⁹cells/L, p=0.02), and also the proportion of CD27 (median 50% vs. 16%, p=0.01) and CD57 expressing (median 79% vs. 74%, p=0.05) NK-cells was higher. The follow-up time of dasatinib treated patients is not yet long enough to correlate the NK-cell counts with the success of the treatment discontinuation.

The absolute number of T-cells or their function did not differ significantly between relapsing and non-relapsing patients at the time of treatment discontinuation. However, both CD4+ and CD8+ T-cells tended to be more mature in patients who stayed in remission compared to patients who relapsed early (CD4+CD57+CD62L- median 5.7% vs. 2.4%, p=0.06, CD8+CD62L+CD45RA+ 13% vs. 26.7%, p=0.05). The analysis of follow-up samples showed that in patients who stayed in remission the Th1 type cytokine (IFN-γ/TNF-α) secretion of CD8+T-cells increased at 6 months compared to baseline (23.6 vs. 18.5%, p=0.07). Same phenomenon was observed in the late relapsing group at relapse compared to baseline (37.9 vs. 13.5%, p=0.03). No similar increase was observed in the early relapsing group.

Conclusions: Low NK-cell numbers and poor cytokine secretion may predict early disease relapse after TKI discontinuation. However, patients who relapse later have high numbers of normally functioning NK-cells. Further research (detailed phenotypic analysis of NK- and T-cells including activating and inhibitory receptors and immune checkpoint molecules) and correlation of biomarker data with clinical parameters are ongoing to understand the ultimate determining factors of relapse.