To the editor:
Recently, Gale and Chen published their perspective on “How should we interpret conclusions of TKI-stopping studies” [1]. The authors focused on the EURO-SKI trial as the largest study on treatment-free remission (TFR) after stopping a tyrosine kinase inhibitor (TKI), claimed the study to be biased, and criticised outcome interpretation. We explain why their allegation of bias is unjustified and clarify the interpretation of the trial.
Study population
For the EURO-SKI trial starting May 2012, physicians from 61 centres in 11 European countries screened 868 patients with chronic myeloid leukaemia and included those who were eligible and consenting in the trial [2]. No existing “observational database” was used.
Gale and Chen entertained the suspicions that not all consecutively potentially eligible patients were screened and that subjects with the longest TKI-therapy or deep molecular response (DMR, BCR::ABL1 transcripts ≤ 0.01% on the International Scale, IS [3]) intervals were not included, knowing the trial involved stopping therapy [1]. These allegations lack any merit. Why should potentially eligible patients not have been screened by an investigator, who has agreed to the protocol of a trial on stopping therapy? And why precisely would inclusion depend on the length of treatment or DMR intervals? It was planned to register 500 patients within two years. In fact, 821 patients were registered within 2.5 years. Patients and physicians were more enthusiastic about the trial than hesitant for obscure reasons.
As a crucial measure against bias, the EURO-SKI trial was designed as a prospective trial. All values of baseline variables, including the candidate prognostic factors, were known at registration. In contrast, TFR outcome was observed post-registration. “Unavoidable limitations and biases” of observational studies do not apply to the EURO-SKI trial. At the time of inclusion, no subjective decisions regarding patient treatment were made. All patients uniformly discontinued TKI-treatment, irrespective of any (prognostic) patient profile. Physicians had no influence on the prospective outcome. The restart of TKI-therapy in 13 cases while still in major molecular response (MMR, BCR::ABL1 transcripts ≤ 0.1% on the IS) constituted the only deviations. Their influence on outcome was and is negligible (Fig. 3 in [2]).
Interpretation of the results of the EURO-SKI trial
Gale and Chen questioned the representativeness of the EURO-SKI data, claiming that therapy outcomes in North America would fall behind those in Germany, referencing Radivoyevitch et al. [4]. However, these authors neither specifically addressed German patients nor conducted an adjusted comparison of molecular recurrence-free survival (MRecFS, recurrence: loss of MMR) between patient samples from different countries. Nevertheless, what really matters for future patients considering TKI-discontinuation is not the repeatability of MRecFS probabilities itself but rather the consistency of prognostic factors associated with MRecFS.
Indeed, the EURO-SKI patient sample is quite unique considering its size and its heterogeneity, with a median duration of TKI-therapy of 7.5 years (interquartile range [IQR]: 5.0–9.9 years) and a median duration of DMR before TKI-discontinuation of 4.7 years (IQR: 2.9–6.9 years). Sample size and heterogeneity made the EURO-SKI trial particularly suitable for studying prognostic factors. With 180 events (losses of MMR) in 405 patients, the sample size was clearly sufficient to examine 10 prognostic factors [5]. Sample size and the heterogeneity in the durations of TKI-therapy and DMR, enabled logistic regression based on a broad range of values, thus providing sufficient protection against bias in modelling. Even occasional deliberate protocol deviations in patient selection would not have introduced a substantial bias.
Considering the appropriateness of patients and temporal perspective for the EURO-SKI trial, survivorship bias is not relevant. The trial only addresses patients who were in DMR. The appropriate patients were basically defined by the inclusion and exclusion criteria (Table 1). Modelling MRecFS probabilities for all patients from the time of diagnosis is not reasonable since it included patients for whom discontinuation of therapy will never be relevant. It would also unnecessarily boost model uncertainty. The selection of prognostic factors for MRecFS probabilities should only depend on the patients the model is intended for.
The appropriate temporal perspective is the day when a patient in DMR discontinues TKI. On that very same day, say, patient A and patient B were in continuous DMR for 5 years and 8 years, respectively. In accordance with the model obtained in the EURO-SKI trial, their probabilities of staying in MMR 6 months after TKI-discontinuation are estimated to be 56 and 65% (Table 2). We cannot tell patient A her chances of MMR maintenance improve by 9% if she stays on TKI for 3 more years as she might lose DMR in between. Patients, who have maintained DMR for 8 years, are likely to constitute a prognostically favourable subgroup of all patients who had been in DMR for 5 years, 3 years ago. Still, since we do not know the risk of losing DMR within these 3 years, the chances of maintaining MMR might improve by postponing treatment discontinuation. Furthermore, keeping DMR while under TKI-treatment and keeping MMR without treatment are two different conditions. A direct transfer of the factors influencing response maintenance between the two different conditions is not possible.
The EURO-SKI trial models the influence of different DMR durations in patients starting the second condition all on the same day. Like age, TKI and DMR duration have to be understood as fixed conditions when prognosticating the probability of MMR maintenance. Besides a somewhat elusive “prognostic profile”, there could be other unknown factors, not yet measured but potentially measurable in the future, hiding behind the duration of DMR.
The bottom line
There is no indication of relevant bias in the EURO-SKI patient sample. With 62% MRecFS (95% confidence interval: 54–68%), the trial confirmed the feasibility of controlled TKI-discontinuation [2]. The significant association of DMR duration before TKI-discontinuation with MRecFS at 6 months was validated within the trial and, recently, in an independent Polish patient sample [6]. At least for patients meeting the criteria of Table 1, the results presented in Table 2 support the decision on treatment discontinuation based on DMR duration – whatever variables hide behind it.
References
Gale RP, Chen J. How should we interpret conclusions of TKI-stopping studies. Leukemia. 2023;37:2343–5.
Saussele S, Richter J, Guilhot J, Gruber FX, Hjorth-Hansen H, Almeida A, et al. Discontinuation of tyrosine kinase inhibitor therapy in chronic myeloid leukaemia (EURO-SKI): a prespecified interim analysis of a prospective, multicentre, non-randomised, trial. Lancet Oncol. 2018;19:747–57.
Cross NC, White HE, Colomer D, Ehrencrona H, Foroni L, Gottardi E, et al. Laboratory recommendations for scoring deep molecular responses following treatment for chronic myeloid leukemia. Leukemia. 2015;29:999–1003.
Radivoyevitch T, Weaver D, Hobbs B, Maciejewski JP, Hehlmann R, Jiang Q, et al. Do persons with chronic myeloid leukaemia have normal or near normal survival? Leukemia. 2020;34:333–5.
Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165:710–8.
Machova Polakova K, Albeer A, Polivkova V, Krutska M, Vlcanova K, Curik N, et al. The SNP rs460089 in the gene promoter of the drug transporter OCTN1 has prognostic value for treatment-free remission in chronic myeloid leukemia patients treated with imatinib. Leukemia 2023; https://doi.org/10.1038/s41375-023-02109-2.
Baccarani M, Deininger MW, Rosti G, Hochhaus A, Soverini S, Apperley JF, et al. European LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013. Blood. 2013;122:872–84.
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
MP designed the work and interpreted the results, draughted the manuscript, approved the final version, agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work were appropriately investigated and resolved, and confirms that he has had full access to the data in the study and final responsibility for the decision to submit for publication. SS played an important role in interpreting the results, revised the manuscript, approved the final version, and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work were appropriately investigated and resolved. FXM played an important role in interpreting the results, revised the manuscript, approved the final version, and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work were appropriately investigated and resolved. JR played an important role in interpreting the results, revised the manuscript, approved the final version, and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work were appropriately investigated and resolved.
Corresponding author
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Pfirrmann, M., Saussele, S., Mahon, FX. et al. How to interpret the EURO-SKI study and its treatment-free remission outcome. Reply to R.P. Gale and J. Chen. Leukemia 38, 460–462 (2024). https://doi.org/10.1038/s41375-023-02124-3
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1038/s41375-023-02124-3