Should I rock the boat? When to stop TKIs in CML

In this issue of Blood, Shanmuganathan et al used early BCR-ABL1 kinetics to predict the likelihood of treatment-free remission (TFR) in patients with chronic myeloid leukemia receiving tyrosine kinase inhibitors (TKIs).¹ 

The overall survival of chronic myeloid leukemia (CML) patients was remarkably improved after the introduction of TKI therapy. TKI therapy was initially thought to mean that patients must continue on treatment indefinitely. Stopping treatment safely would reduce cost, improve quality of life, and allow younger patients to conceive safely. Thus, sustained TFR, defined as maintaining a major molecular response (MMR) with BCR-ABL1 ≤0.1% for at least 12 months off TKI therapy, became an important goal.²  Several studies have shown that TKIs can be discontinued safely in ∼40% to 65% of patients who achieve a deep molecular response (DMR, defined as either MR4.0, BCR-ABL1 ≤0.01% on the international scale or MR4.5, level ≤0.0032%) during their therapy.^3-6  Although some clinical variables had a strong correlation with sustained TFR such as longer duration of TKI therapy and a longer duration of maintaining DMR, other variables such as age, withdrawal syndrome secondary to TKI cessation, and Sokal risk score had a weaker correlation.^3-6  More importantly, today, there is no clinical decision tool that can accurately predict which patient can sustain MMR after TKI cessation, and most guidelines recommend monthly BCR-ABL1 transcript level monitoring for the first 12 months after stopping TKI to detect early loss of MMR.

In this study, the authors performed a retrospective analysis of data from 386 chronic-phase CML patients who received TKI therapy at 2 hospitals in South Australia between January 2008 and October 2019. The TFR eligibility included a minimum of 3 years of TKI therapy and at least 2 years of sustained MR4.5. All patients attempting TFR had monthly BCR-ABL1 testing for a minimum of 12 months. Sustained TFR was defined as maintaining MMR at 12 months after stopping the therapy. The primary end point of the study sustained TFR at 12 months. The secondary objective was molecular recurrence-free survival, defined as the time between cessation of TKI therapy to the first detection of loss of MMR.

The authors calculated the rate of decline in the BCR-ABL1 transcript level in the first 3 months using the halving time formula: Halving time = ln(2)*x/ln(y/z) in which x = the number of days between the baseline and 3-month assessment, y = baseline BCR-ABL1 value, and z = 3-month BCR-ABL1 value. They compared it with the fold-reduction in BCR-ABL1 transcript derived from dividing the baseline value over 3-month value (which does not incorporate the time differences between the 2 measurements). A total of 130 patients met the TFR eligibility criteria and 115 had ≥12 months of follow-up after discontinuation. The median duration of follow up after TKI discontinuation was 51.1 months (range, 12.0-153.2), the median duration of TKI therapy before TFR was 7 years (range, 3.1-17.6), and the median duration of sustained MR4.5 was 3.3 years (range, 2.2-12.5). Of 130 patients who stopped their TKIs, 55% remained off therapy at 12 months. Loss of MMR resulted in restarting the TKI in 87% (n = 45) of molecular relapses during the first 12 months.

In a multivariate analysis, the halving time and transcript type were the only significant predictors of sustained TFR. Although the fold-reduction value from baseline to 3 months also was significant in multivariate analysis (after exclusion of halving time), the c-index for the final model was higher using the halving time, suggesting that it is a better predictor of sustained TFR. More important, sustained TFR at 12 months was lower in patients with longer halving times. The authors then divided the patients into quartiles based on halving time with 80% probability of sustained TFR the first quartile (halving time <9.35 days) compared with 4% in patients in the fourth quartile (halving time >21.85 days), P < .001. The authors also validated their results in 26 patients who attempted TFR in Korea. Finally, the authors built a model that uses the halving time, transcript type, and the duration of TKI exposure, which provides the probability of sustained TFR. The authors gave an example of a patient with an e14a2 transcript and an initial BCR-ABL1 halving time of 5 days. This patient has a 30% probability of sustained TFR, if attempted after 3 years of TKI exposure, whereas this probability increases to ∼65% if the TFR attempt is deferred to 8 years of treatment.

Developing a clinical decision tool that can aid patients and their physicians of predicting the ability to successfully attempt and sustain TFR is clinically important. Such a model can help patients understand their disease trajectory and establish realistic expectations regarding their ability to stop their treatment in the future (a critical conversation for younger patients who are trying to conceive or plan their future). Further, stopping TKIs can be a difficult decision especially for patients who have been tolerating the therapy without any side effects (“Why rock the boat?”). Several questions remained. First, what if the model predicted a longer time for the probability of achieving a TFR? Would switching to another TKI expedite the process? Second, should starting a second-generation TKI be the standard for patients who want to be eligible for TFR in a short time frame given that these patients may achieve lower levels of BCR-ABL1 transcript at 3 months and could achieve a DMR faster compared with imatinib? And, finally, what about the patients who have multiple discontinuations of their therapy because of toxicity during their first 12 months? Should a switching strategy be attempted early to accelerate their chance of achieving a DMR that qualify them for a TFR?