BCR-ABL1 Transcript Kinetics and Treatment-Free Remission in Patients with Chronic Myeloid Leukemia

Lotte Peeters; Nina Kovacic; Elodie Chevalier

doi:10.5281/zenodo.22227740

Abstract

Early BCR-ABL1 kinetics predict which patients will reach a deep molecular response, and reaching one is the entry requirement for attempting to stop treatment. It does not follow that kinetics predict who will stay in remission after stopping, and the largest comparison found that they do not. Objectives: To review the published evidence on BCR-ABL1 transcript kinetics and treatment-free remission in chronic myeloid leukaemia: what proportion of patients maintain remission after stopping, when relapse occurs, which variables predict it, and whether early kinetic measures add anything to the prediction. Material and Methods: Narrative review of discontinuation trials, real-world cohorts, modelling studies and consensus recommendations indexed in PubMed and the major haematology journals. Studies were eligible if they reported treatment-free remission outcomes after tyrosine kinase inhibitor discontinuation in a defined chronic myeloid leukaemia population, or the predictive performance of molecular response measures. No new patient data were generated and no patients were recruited; every estimate quoted is one published by the original investigators.

Results: In the largest discontinuation trial, 434 of 728 patients (61 per cent, 95% CI 57-64) remained in major molecular response at 6 months and 309 of 678 (46 per cent, 42-49) at 36 months. Duration of tyrosine kinase inhibitor treatment and of deep molecular response before stopping were confirmed as predictors of loss at 6 months, with transcript type added as a third. In the earliest trial, molecular relapse occurred in 61 of 100 patients and 80 per cent of those relapses fell within the first 6 months. Where deep response had been sustained for only one year and total treatment was under three years, relapse-free survival was 16.7 per cent at 12 months and the trial was terminated early. Comparing early measures in 408 real-world patients, halving time and residual disease at month 3

performed similarly and residual disease at month 6 performed best for predicting deep response, with areas under the curve of 0.81 to 0.92, but the initial molecular response did not predict treatment-free remission maintenance. Conclusion: Early kinetics identify who will achieve a deep response and are useful for that purpose. What predicts staying in remission after stopping is duration, of treatment and of deep response, rather than how quickly the transcript fell in the first months. Abbreviations: CML – Chronic Myeloid Leukaemia, CML-CP – Chronic Phase CML, TKI – Tyrosine Kinase Inhibitor, TFR – Treatment-Free Remission, MMR – Major Molecular Response, DMR – Deep Molecular Response, MR4 – BCR-ABL1 IS ≤0.01 Per Cent, MR4.5 – BCR-ABL1 IS ≤0.0032 Per Cent, IS – International Scale, HT – Halving Time, ELTS – EUTOS Long-Term Survival Score, KDm – Kinase Domain Mutation.

Keywords: Chronic Myeloid Leukaemia; BCR-ABL1; Treatment-Free Remission; Halving Time; Deep Molecular Response

Introduction

stopped, in people who may otherwise take a daily drug for forty years [3]. baseline [4,5]. Deep molecular response at MR4 or MR4.5 became the entry criterion for stopping attempts [6,7].

Alongside this, early transcript kinetics were shown to predict which patients would reach those depths. The transcript level at three or six months, and more recently the time taken for the transcript to halve, both carry prognostic weight for subsequent response [8,9,10]. It has been natural to assume the same measures would predict who could stop successfully.

This review sets out what discontinuation actually achieves, when relapse occurs, which variables predict it, and whether the early kinetic measures transfer from predicting depth to predicting durability.

Material and Methods

Search strategy: PubMed and the archives of the major haematology journals were searched for studies of treatment discontinuation in chronic myeloid leukaemia. Search terms combined chronic myeloid leukaemia with treatment-free remission, tyrosine kinase inhibitor discontinuation, deep molecular response, BCR-ABL1 halving time, transcript kinetics and molecular relapse. Reference lists of retrieved trials and recommendations were screened by hand.

Eligibility: Studies were included if they reported outcomes after tyrosine kinase inhibitor discontinuation in a defined chronic myeloid leukaemia population, or evaluated the predictive performance of molecular response measures for deep response or for treatment-free remission. Modelling studies are cited where they address prediction directly and are identified as such.

Data extraction and handling: Design, cohort size, entry criteria, definition of molecular relapse, follow-up duration and reported outcome estimates were extracted as published. Entry criteria are stated alongside every remission figure, because trials differ in required treatment duration, required depth and duration of deep response, and those differences drive most of the variation between reported success rates. No pooling or re-analysis was performed and no patient-level data were obtained. Figures 1 to 4 and Tables 1 to 3 reproduce published values only. This is a narrative synthesis and was not registered as a systematic review.

Results

What stopping achieves: The largest discontinuation trial enrolled patients treated with any tyrosine kinase inhibitor for at least three years who had sustained MR4 for at least one year. Among 728 evaluable patients, 434 (61 per cent, 95% CI 57 to 64) remained in major molecular response at 6 months, and 309 of 678 (46 per cent, 95% CI 42 to 49) at 36 months (Figure 1) [11,12].

The earlier trial that established feasibility enrolled 100 chronic-phase patients with undetectable transcripts sustained for at least two years on imatinib. Molecular relapse occurred in 61 patients, and treatment-free survival was 39 per cent at five years with median follow-up of 77 months [13,14]. A parallel study using similar criteria reported treatment-free remission of 47.1 per cent at 24 months among 40 patients [15].

Roughly half of carefully selected patients therefore stay off treatment long term, and the figure has been consistent across trials and drugs [11,13,16]. Restarting after molecular relapse restores response in almost all cases, which is what makes the attempt

Relapse is an early event: The timing is consistent and clinically important. In the earliest trial, 80 per cent of the 61 molecular relapses occurred within the first six months after stopping (Figure 2) [14], and in the largest trial loss of major molecular response occurred primarily within the same window [11,12]. Longer follow-up of the first cohort found no additional relapses beyond a median of 65 months [13].

That distribution dictates monitoring intensity. Transcript measurement is required monthly during the first six months, and the risk after two years of sustained remission is low enough that intervals can be extended [6,7]. Late losses do occur, and the factors predicting them differ from those predicting early loss, which is

Principal reported

Study n Entry criteria finding

Undetectable ≥2 39% treatment-free at STIM1 [13,14] 100 years five years Undetectable ≥2 TWISTER [15] 40 47.1% at 24 months years EURO-SKI MR4 ≥1 year, TKI 61% at 6 months, 46% at [11,12] ≥3 years 36 months Kinetics of early and late STIM2 [17] — First-line imatinib recurrence described MR4.5 ≥1 year, 16.7% at 12 months; D-FREE [18] 21 TKI <3 years terminated early Real-world cohort Early response predicts 408 Observational [19] depth, not maintenance Table 1: Principal discontinuation studies, with entry criteria and the finding as reported by the original investigators Duration is what predicts success: The final analysis of the largest trial confirmed duration of tyrosine kinase inhibitor treatment and duration of deep molecular response before stopping as significant predictors of losing major molecular response at six months, and added transcript type as a third factor, with e13a2 carrying worse prospects than e14a2 [12,20,21]. For losses occurring after six months, the significant factors were treatment duration, percentage of blasts in peripheral blood and platelet count at diagnosis [12].

The clearest demonstration is a trial that violated the duration requirement deliberately. Twenty-one patients who had maintained MR4.5 for one year discontinued dasatinib after less than three years of total treatment. Seventeen relapsed, giving estimated relapse-free survival of 16.7 per cent at 12 months, and the study was terminated early under its own safety monitoring rules (Figure 3) [18].

Setting that against 61 per cent at six months in the larger trial, where treatment had lasted at least three years, isolates duration as the operative variable [11,18]. Depth of response was if anything greater in the terminated study, since MR4.5 is deeper than the MR4 What early kinetics do and do not predict: Transcript halving time has been proposed as an early predictor, and it performs well for the endpoint it was developed against. Shorter halving time predicts achievement of molecular response in newly diagnosed patients [9], and early kinetics predict subsequent achievement of the deep response that makes a stopping attempt possible [10].

A real-world cohort of 408 patients compared the available early measures directly, testing a prognostic score, halving time, and residual disease at months 3 and 6 against three endpoints. Halving time and residual disease at month 3 performed similarly. Residual disease at month 6 performed best for predicting optimal response, with areas under the curve between 0.81 and 0.92 and cut-offs of 0.11 per cent for MR4 at month 24 and 0.12 per cent for MR4.5 at month 48 (Figure 4) [19].

The negative finding is the one that matters here. The same analysis found that initial molecular response predicted deep molecular response but not treatment-free remission maintenance [19]. Early kinetics identify who will reach the door; they do not identify who will stay outside it once through.

Figure 4: Discrimination of early measures for deep molecular response in 408 patients [19] Predicts deep Predicts treatment-free

Variable

The finding at the centre of this review is a negative one, and negative findings about biomarkers are easy to overlook. Halving time and early residual disease are good predictors of deep molecular response [8,9,10], deep molecular response is the entry criterion for stopping [6,7], and it is a short step to assuming that early kinetics therefore predict successful stopping. The direct comparison shows they do not [19].

What does predict success is duration: of treatment before stopping, and of deep response before stopping [11,12]. The trial that stopped dasatinib after less than three years despite MR4.5 sustained for a year makes the point unusually cleanly, with relapse-free survival of 16.7 per cent against 61 per cent where the duration requirement was met [11,18].

Why depth and durability come apart

A plausible account is that transcript level measures the differentiated leukaemic compartment while relapse originates in a quiescent stem cell population that the assay does not see [22,23]. Kinetics of transcript decline describe how efficiently the drug clears cells that are dividing; they say nothing about how many quiescent cells remain.

Duration of exposure, by contrast, plausibly does affect that reservoir, through attrition over years rather than through rapid killing. If so, duration and kinetics are measuring different compartments, and the observation that only duration predicts durability is what the biology would lead one to expect rather than an anomaly [22,23].

Study limitations

The limitations belong to the underlying literature. Discontinuation trials differ in required treatment duration, required depth, required duration of deep response and definition of molecular relapse, so their success rates are not directly comparable [11,13,15,18]. Most enrolled patients treated predominantly with imatinib, and whether findings transfer to second-generation inhibitors used first line is incompletely established [11,16]. Assay standardisation on the international scale is good but not perfect, and the deepest response categories approach the limit of quantification, where reproducibility falls [4,5]. The comparison of early predictive measures is from a single real-world cohort [19]. And patients with prior resistance or kinase domain mutations are generally excluded from stopping attempts, so outcomes in that group rest on small series [24].

Two gaps deserve naming. No trial has randomised the duration of deep molecular response required before stopping, so the current thresholds rest on observational association rather than comparison [11,12]. And no study reviewed here identifies a measure that predicts treatment-free remission independently of duration, which is what would allow patients to stop earlier than current criteria permit [19,25].

Conclusion

Roughly half of carefully selected patients with chronic myeloid leukaemia remain in remission after stopping tyrosine kinase inhibitor therapy, with 61 per cent still in major molecular response at six months and 46 per cent at three years in the largest trial. Relapse is an early event, with 80 per cent of losses occurring within six months, which is why monthly monitoring is required over that period. The variables that predict success are duration of treatment and duration of deep molecular response before stopping, together with transcript type; stopping after less than three years of treatment gave relapse-free survival of 16.7 per cent even where MR4.5 had been sustained. Early transcript kinetics, including halving time, predict achievement of deep molecular response well and do not predict maintenance of treatment-free remission, a distinction that the shared vocabulary of molecular response tends to obscure.

Acknowledgements

The authors thank the discontinuation trial groups and the laboratories maintaining international scale standardisation, whose published data form the substance of this article.

Funding

No dedicated funding from public, commercial or charitable sources supported this review.

Conflict of Interest

Ethics Statement

This article reviews previously published studies only. No new patient data were collected and no participants were recruited, so ethical approval was not required.