Risk of progression in chronic phase‐chronic myeloid leukemia patients eligible for tyrosine kinase inhibitor discontinuation: Final analysis of the TFR‐PRO study

Disease progression to accelerated/blast phase (AP/BP) in patients with chronic phase chronic myeloid leukemia (CP‐CML) after treatment discontinuation (TD) has never been systematically reported in clinical trials. However, recent reports of several such cases has raised concern. To estimate the risk of AP/BP among TD‐eligible patients, we conducted TFR‐PRO, a cohort retro‐prospective study: 870 CP‐CML patients eligible for TD formed a discontinuation cohort (505 patients) and a reference one (365 patients). The primary objective was the time adjusted rate (TAR) of progression in relation to TD. Secondary endpoints included the TAR of molecular relapse, that is, loss of major molecular response (MMR). With a median follow up of 5.5 years and 5188.2 person‐years available, no events occurred in the TD cohort. One event of progression was registered 55 months after the end of TD, when the patient was contributing to the reference cohort. The TAR of progression was 0.019/100 person‐years (95% CI [0.003–0.138]) in the overall group; 0.0 (95% CI [0–0.163]) in the discontinuation cohort; and 0.030 (95% CI [0.004–0.215]) in the reference cohort. These differences are not statistically significant. Molecular relapses occurred in 172/505 (34.1%) patients after TD, and in 64/365 (17.5%) patients in the reference cohort, p < .0001. Similar rates were observed in TD patients in first, second or third line of treatment. CML progression in patients eligible for TD is rare and not related to TD. Fears about the risk of disease progression among patients attempting TD should be dissipated.


| INTRODUCTION
Chronic myeloid leukemia (CML) patients in chronic phase (CP) now enjoy a normal life expectancy thanks to the routine use of tyrosine kinase inhibitors (TKIs). 1,2These drugs are able to induce not only a complete cytogenetic response (CCyR) in over 80% of subjects, 3 but also a deep molecular response (DMR or MR4), defined as BCR::ABL1 transcript levels ≤0.01%according to the International Scale (IS), in more than 50% of patients. 4,5e possibility of treatment discontinuation (TD) was pioneered by the French CML group 6 and subsequently confirmed by several studies, [7][8][9][10][11][12][13][14][15] which consistently indicate that approximately one half of patients who try TD can achieve a treatment-free remission (TFR) at 3 years (48%-61% in several studies), 16 defined as the maintenance of at least a major molecular response (MMR or MR3). 10 Adequate disease control off therapy affords several benefits for patients, such as a lower occurrence of drug-related AEs and a feeling of cure.A minimum of 5 years of TKI therapy and a stable MR4 for at least 2 years before attempting TD are recommended by the European Society for Medical Oncology (ESMO) 17 and European LeukemiaNet (ELN), 18 while even less stringent criteria are proposed by the US National Comprehensive Cancer Network (NCCN). 19der the recommended conditions and with optimal molecular monitoring, TD is generally considered safe.In fact, about 95% of patients who experience molecular recurrence regain their initial molecular level after restarting TKI therapy. 18In addition, the occurrence of disease progression (i.e., the transformation of CP-CML to the advanced state of accelerated phase [AP] or blast phase [BP]) was considered virtually impossible for a long time. 111][22][23] In all cases, the transformation to AP/BP dramatically changed the prognosis of the disease.These are case reports from which a precise assessment of the denominator, that is, the number of patients at risk, and the time of observation, is missing.A large-scale, precise quantification of the risk of CML progression after TD in the setting of TFR has not been performed.Sudden blastic transformation has also been reported as a rare event in patients in optimal molecular response who did not discontinue therapy, 24,25 suggesting that transformation may occur independently of TD. 26 The main goal of the Treatment-Free Remission-PROgression (TFR-PRO) study is to obtain a precise assessment of the risk of disease progression in a cohort of CP-CML patients eligible for TD, regardless of whether they underwent a TD.

| Study design and participants
TFR-PRO is a multicenter, international, retro-prospective, observational study.
Eligible patients were aged 18 years or older and had a diagnosis of BCR::ABL1-positive CP-CML.The target population included all CML patients eligible for TD regardless of whether they had discontinued.To be considered eligible for TD, patients needed at least 4 years of TKI treatment (regardless of the number of TKIs used) and at least 18 months of DMR, determined through at least three consecutive molecular analyses by RT-qPCR performed at their own centers.
DMR was defined as MR4 (either detectable disease with BCR::ABL1 ratio ≤0.01% on the IS or undetectable disease in cDNA with at least 10 000 ABL1 copies).The date of the first DMR and the dates and level of molecular response by RT-qPCR during the 18 months of stable DMR were registered.Patients with previous allogeneic hematopoietic stem-cell transplantation (HSCT) and with a diagnosis of CML in advanced phase (AP or BP) were excluded.
Two cohorts were considered: patients that met the eligibility criteria started contributing to the non-discontinuation (reference) cohort; when patients started TD, they were moved to the discontinuation cohort.TD was terminated when the patients lost MR3 and resumed treatment.Events developing in patients after the end of TD were considered associated to discontinuation if they developed within 36 months from the end of discontinuation.Thereafter, they were no longer considered to be associated to discontinuation and the patient returned to the reference cohort for the remaining time.
This rule applied also to subsequent TD attempts.
Collection of data was both retrospective and prospective.
Patients registered before the opening of this study (July 24, 2020) contributed to the retrospective cohort and, after the opening of the study they also contributed to the prospective cohort.Those who were registered after the opening of the study contributed only to the prospective cohort.Monitoring of disease status was performed to assess the maintenance of the molecular remission during the study period, according to the policies of each center.Data from patients belonging to the retrospective cohort were collected starting from 2011.The cut-off date for this analysis was January 1, 2023.
A flexible statistical approach allowed attributing patients to the two groups in different periods according to their condition concerning TD (yes or no).
The ethics committee at each center approved the protocol in accordance with local rules and regulations.The informed consent for the prospective cohort had to be signed and dated before starting the collection of data.

| Outcomes
The primary endpoint of TFR-PRO was the time-adjusted rate (TAR) of progression to AP or BP.Definitions of AP and BP were made according the 2006 ELN criteria. 27Secondary endpoints included: progression-free survival (PFS), defined as time from eligibility to progression to AP or BP; the rate of molecular relapse, defined as the loss of MMR after TD (in the discontinuation cohort) or after eligibility (in the reference cohort); the percentage of patients with molecular relapse who achieved a new MMR after losing it.

| Statistical analysis
Time to molecular relapse was measured from the date on which the patient satisfied the eligibility criteria to the date of the RT-qPCR assessment which defined molecular relapse.Time to disease progression was measured from the date the patient satisfied the eligibility criteria to the minimum between the date of AP or BP diagnosis and the end of the follow-up.Study group classification in the analysis of time to molecular relapse was defined as a binary, non-reversible, time-varying factor, where eligible patients were divided into two groups at the beginning of the follow-up depending on the presence or absence of TD, and in the latter case could start discontinuation during the follow-up and thus dynamically changed study group classification.Study group classification in the analysis of time to progression was defined as in the case of the analysis of time to molecular relapse, but it was a reversible binary variable where patients who developed molecular relapse under discontinuation and restarted treatment were kept under discontinuation for a further time lag of 36 months and then reclassified dynamically as absence of TD.
Descriptive statistics on the main endpoints were obtained by the "Simon-Makuch" version of the Kaplan-Meier curves and log-rank test and time adjusted rates accounting for the time varying nature of the classification according to presence or absence of TD. 28 Kaplan-Meier curves and log-rank test were used to describe the cumulative incidence of achievement of novel MMR since molecular relapse in patients who did/did not have a prior discontinuation.
The follow-up length was described using the "reverse Kaplan-Meier" method.The exponential model and Clopper Pearson exact method were used for significance and 95% confidence interval calculation of rates and proportions, respectively.All the analyses and graphics were performed by using Stata software version 18.

| Patients and follow-up
Between July 2020 and November 2022, we registered 907 patients with CP-CML from 20 centres in five countries (Italy, France, Germany, Spain, and Canada).Thirty-seven patients were excluded after screening as not eligible or without follow up information.Thus, data were available for 870 patients at the time of the analysis.A total of 505 patients (58.0%) attempted TD during the observation period (43 patients had two TD attempts) and 365 (42.0%) never attempted TD (Figure S1).Median number of RT-qPCR analyses performed for monitoring of BCR::ABL1 transcript levels was three per year (IQR 2-4) during the entire study.However during the period of TKI discontinuation (that is from discontinuation onwards for patients who never developed molecular relapse, or from discontinuation to the date of molecular relapse for patients who developed it), almost all patients had 3 RT-qPCR assessment per year (median 3, IQR 3-3).

Patient demographics, clinical characteristics, and follow-up information are reported in
At eligibility, 496 patients (57.0%) were on treatment with imatinib, 182 (20.9%) with nilotinib, 140 (16.1%) with dasatinib, 39 (4.5%) with bosutinib, and six (0.7%) with ponatinib.A total of 245 patients (28.2%) were receiving their TKI as a second or subsequent line of therapy.The two cohorts were homogeneous for the main demographic and clinical characteristics, except for the type of TKI at eligibility.Compared with the reference cohort, in the discontinuation cohort significant differences existed regarding the TKI being used: there was a tendency towards higher percentages of patients in the T A B L E 1 Patients' characteristics and follow-up information.Furthermore, patients who were on treatment with a TKI as a second or subsequent line of therapy at the time of eligibility (for the reference cohort) or at the time of TD (for the discontinuation cohort) were less represented in the discontinuation cohort (24.2%) than in the reference cohort (33.7%) (Table 1).
Median duration of follow-up since eligibility was 5.5 years (IQR 2.6-8.6) in the overall cohort, with 5188.2 person-years available for the analysis of the primary endpoint, of which approximately 75% were collected retrospectively and 25% prospectively.

| Survival
A total of 24 deaths were reported, for a TAR of 0.

| Progression
One patient progressed to lymphoid BP 55 months after the end of TFR.Because this event developed more than 36 months after the end of TFR, as per study protocol, the patient was contributing to the reference cohort at the time of progression.Therefore, the risk of progression was 1/870 or 0.1% (95% CI [0.0-0.6]) in the overall cohort.This value ranges from 0/505 or 0%, (95% CI [0.0-0.6]) for  (A) The current management of CML includes the possible discontinuation of TKI treatment in a sizeable proportion of patients.This practice carries clear advantages, including sparing the patient potential longterm adverse events and significantly reducing the cost of treatment.About half of patients attempting TD, regardless of the TKI being used, must resume therapy due to loss of MMR, 16 but most achieve a new MMR once back on treatment. 18wever, anecdotal reports of CML patients experiencing a transformation to acute leukemia during or shortly after TD were alarming because they suggested the possibility of a progression of CML from a safe condition such as DMR to acute leukemia, which dramatically worsens patient prognosis and quality of life.At the time of presenting these results, described cases of progressions grew to 12. 10,[20][21][22][23][29][30][31][32] The TFR-PRO results presented here indicate that there is no cause for alarm. TFR-PR represents the first study in which consecutive patients eligible for TD were enrolled and followed.More than 5000 patient-years were available for analysis.
In this study we report that disease progression among patients in TD or among patients eligible for TD represents an exceedingly rare event, but a possible one.The TAR data suggest, for example, that in a center managing 100 patients eligible for TD, at most one event of progression could happen once every 8 years.
In the TFR-PRO cohort, the relationship between TD and progression is doubtful.The only patient who experienced progression attempted TD but his progression occurred almost 5 years after MMR was lost, therefore outside of the 36-month limit the protocol set for considering an event linked to TD.The analysis of this patient's clinical course clearly points to the lack of compliance to treatment after the end of TD as the main cause for the subsequent progression.This emphasizes a key point regarding selecting patients for TD.While multiple variables have been included in guidelines for TD, 18,19 the willingness of the patient to resume treatment in the case of TD failure must also be evaluated.It is assumed that patients who achieve a DMR are more likely to be compliant but, as shown from the case presented, patient compliance can change over time.
The population studied within TFR-PRO showed an excellent health status, with a 95% survival rate at 10 years and no CML-related death.
TFR-PRO followed patients treated as per standard of care for their CML, in whom monitoring does not always match recommended guidelines, as previously reported. 18,19For instance, almost 50% of patients did not have a Sokal score evaluated at diagnosis to assess risk of treatment failure.In addition, 25% of patients in the TFR-PRO study had less than two RT-qPCR tests per year in the whole study period, although during the discontinuation period this number raised to three.Although these numbers are lower than what recommended in many TFR guidelines, 18,19 it is noteworthy that the prognosis for patients remained very good even in this out-of-trial cohort and that no progression linked to TD was observed.
The loss of MMR was clearly associated with TD.However, loss of MMR was also observed among patients who did not try TD.The reasons for this are heterogeneous and may range from a lack of compliance with treatment, to the need to interrupt treatment because of comorbidities or surgical procedures, or to the possible induction of increased metabolism of TKI over time.
Imatinib represented the most common TKI used by patients eligible for TFR-PRO, in whom it was given as a first line therapy, as expected.It is interesting to note that patients in second or subsequent lines of therapy had similar RFS values than patients in first line treatment once they decided to start TD.This result indicates that once patients became eligible for TD the line of treatment had a limited impact on outcome, although most patients enrolled in the TFR-PRO study were on first line treatment.
Within this group, the RFS probability resulted significantly lower in patients with a previous history of TKI resistance, compared to those without such a history.This could reflect a possible higher biologic instability of the leukemic stem cells in the first group of patients, which could require a closer monitoring during TD.
Imatinib and nilotinib tended to be overrepresented in patients who attempted TD compared to patients who did not.
The TFR-PRO study represents the only study which enrolled consecutive CML patients eligible for TD independently of their decision to stop therapy; the results emanating from it will therefore contribute to our knowledge and will help physicians treating CML patients.
The low likelihood of disease progression and the lack of an evident relationship between disease progression and TD represent useful and encouraging information that should allay fears about a disease progression among patients who attempt TD.
Among 172 patients who experienced molecular relapse after TD, 170 (98.8%) resumed TKI treatment after a median time of 0.7 months after molecular relapse (IQR 0.3-1.4).The remaining two eligibility (years)F I G U R E 1Overall survival in the total cohort.The results of the Kaplan-Meier estimates for overall survival among patients of the whole cohort.[Color figure can be viewed at wileyonlinelibrary.com] patients had fluctuating RT-qPCR values around the MMR threshold, they never confirmed MMR loss on two consecutive assessments and were still in MMR at the last follow up.Therefore, they never resumed treatment.The proportions of patients in the discontinuation cohort who achieved MMR once again after molecular relapse were 78.2% (95% CI [71.3-84.4])at 6 months, 92.1% (95% CI [87.1-95.7])at 12 months, and 96.0%(95% CI [92.1-98.4])at 24 months.In comparison, the proportions of patients in the reference cohort who achieved a new MMR after MMR loss were 63% (95% CI [49.4-76.7])at 6 months, 84.8% (95% CI [73.0-93.3])at 12 months, and 89.6% (95% CI [78.6-96.4])at 24 months.Figure 2B presents the Kaplan-Meier survival curves relative to the proportion of patients who achieved a new MMR after MMR loss in the two cohorts.

the discontinuation cohort, to 1 /
365 or 0.3% (95% CI [0.0-1.5]) for the reference cohort.The TAR of progression was 0.019/100 personyears (95% CI [0.003-0.138])for the overall cohort, ranging from 0.0 (95% CI [0-0.163]) for the discontinuation cohort to 0.030 (95% CI [0.004-0.215])for the reference cohort.No statistically significant differences exist between the two cohorts.The subject who developed disease progression was a male patient, diagnosed with CP-CML in January 2009, when he was 45 years old.He had started TKI therapy with imatinib, achieving an MMR 3 months later.In February 2013 TD was initiated.After 20 months, in October 2014, he experienced a molecular relapse.The patient was recommended to resume imatinib but he demonstrated a very poor compliance to retreatment, which was reflected by the subsequent RT-qPCR assessments, revealing fluctuating BCR::ABL1 transcript levels, ranging from MR1 to MR4 (Figure S2 in the Supplementary Appendix).In March 2019 he lost MMR for the fifth time and in May (55 months after the first molecular relapse), he developed an abrupt lymphoid BP.He received chemotherapy combined with ponatinib and in October 2019 he underwent allogeneic HSCT.At the last follow-up in September 2022, the patient was alive in good general condition, with a complete hematological response, undetectable levels of BCR::ABL1 by RT-qPCR and no signs of GVHD.
a Data were missing for 393 patients.b