Rationale and motivating factors for treatment-free remission in chronic myeloid leukemia

Abstract

With BCR-ABL1 tyrosine kinase inhibitors (TKIs), such as imatinib, nilotinib, dasatinib, bosutinib, and ponatinib, many patients with chronic myeloid leukemia in chronic phase (CML-CP) can expect to live near-normal life spans. Current treatment recommendations of the National Comprehensive Cancer Network and the European LeukemiaNet state that patients with CML-CP should remain on TKI therapy indefinitely. However, there is increasing evidence from clinical trials that some patients with sustained deep molecular responses may be able to achieve treatment-free remission (TFR), whereby they can suspend TKI therapy without losing previously achieved responses. With many patients achieving deep molecular responses to TKI therapy, there is growing interest in whether such patients can achieve TFR. In addition, adverse events (AEs) with long-term TKI therapy, including both the potential for later-emerging AEs and chronic, low-grade AEs, represent a major motivator for oncologists and their patients to investigate the feasibility of TFR. In this review, we provide an overview of data from TFR clinical trials, discuss the importance of achieving a deep molecular response to TKI treatment, and consider potential reasons for investigating TFR following TKI therapy.

Keywords:

• Chronic myeloid leukemia
• treatment-free remission
• deep molecular response
• tyrosine kinase inhibitors

Introduction

With tyrosine kinase inhibitor (TKI) therapies targeting BCR-ABL1 (imatinib, nilotinib, dasatinib, bosutinib, and ponatinib) for the treatment of chronic myeloid leukemia in chronic phase (CML-CP),[1] many patients can be expected to live near-normal life spans with continual TKI treatment.[2] However, there is increasing evidence demonstrating that, within the context of a clinical trial and with rigorous molecular monitoring, some patients may be able to suspend TKI therapy without losing their responses,[3–17] a state known as treatment-free remission (TFR).

Historically, a high percentage of patients with CML-CP who received allogeneic hematopoietic stem cell transplant (HSCT) were able to remain treatment free for many years following transplant.[18,19] However, the toxicity of the procedure limits patient eligibility and can lead to substantial posttransplant morbidity and mortality.[19,20] The possibility of TFR without allogeneic HSCT was previously demonstrated in patients with CML-CP treated with interferon alfa (IFN) therapy.[21,22] In a study of 21 IFN-treated patients who had achieved and sustained undetectable BCR-ABL1 transcripts for 2 years, 20 were able to remain in TFR (median follow-up of 8 years), with approximately half showing BCR-ABL1 positivity in ≥1 assessment, whereas 1 progressed to advanced CML after 5 years and underwent allogeneic HSCT.[21] Although this study demonstrated the possibility of TFR in some IFN-treated patients, very few patients are able to achieve undetectable BCR-ABL1 transcripts on IFN.[23] In an analysis of 512 patients with CML-CP treated with IFN between 1981 and 1995, only 27% achieved a complete cytogenetic response (CCyR; 0% Philadelphia chromosome–positive [Ph+] metaphases) at any time; among 78 patients with CCyR who were evaluated for BCR-ABL1 transcript levels, undetectable BCR-ABL1 transcripts were observed in ≥1 real-time quantitative polymerase chain reaction (RQ-PCR) assessments for 46 patients.[23] As such, only a small minority of IFN-treated patients would be able to attempt TFR.

More recently, there has been renewed interest in TFR as a therapeutic goal due to the higher rates of deep molecular responses observed with TKIs, particularly second-generation TKIs, relative to historical data from IFN-treated patients.[24–27] On TKI therapy, most patients rapidly achieve CCyR and major molecular response (MMR; BCR-ABL1 ≤ 0.1% on the International Scale [IS]),[28–30] and with long-term TKI therapy, many patients are able to achieve deep molecular responses, such as molecular response 4 (MR⁴; BCR-ABL1^IS ≤0.01%), molecular response 4.5 (MR^4.5; BCR-ABL1^IS ≤0.0032%), or undetectable minimal residual disease (UMRD; i.e. no detectable BCR-ABL1 transcripts; Figure 1). In addition, some TKIs are more potent than others; for example, nilotinib and dasatinib result in higher rates of deep molecular responses than imatinib in the frontline setting,[30,31] suggesting that patients on these agents may be more likely to be eligible to attempt TFR than patients on imatinib. Neither bosutinib nor ponatinib has been approved for the frontline treatment of CML-CP,[32,33] although some frontline clinical trial data are available for each.[34,35]

Figure 1. Molecular response depth and the number of CML cells over time. CML: chronic myeloid leukemia; IRIS: International randomized study of interferon versus STI571; IS: International Scale; MMR: major molecular response (BCR-ABL1^IS ≤0.1%); MR⁴: molecular response 4 (BCR-ABL1^IS ≤0.01%) MR^4.5: molecular response 4.5 (BCR-ABL1^IS ≤0.0032%); MR⁵: molecular response 5 (BCR-ABL1^IS ≤0.001%). BCR-ABL1^IS 100% and the corresponding number of leukemic cells refer to the median pretreatment disease level among 30 patients enrolled in the IRIS trial.[70,95,96]

Deep molecular responses to TKI therapy

The likelihood of obtaining deep molecular responses differs among the TKIs available to treat CML-CP. Many patients with CML-CP are able to achieve MR^4.5 with long-term imatinib therapy.[36,37] In an analysis of 423 patients treated with imatinib 400, 600, or 800 mg daily in clinical trials, 51.7% achieved MR^4.5 by 8 years, although only 36.5% of patients had stable MR^4.5 (no loss of response for ≥2 years).[36] Furthermore, through the first 3 years of treatment, approximately 20% of patients achieved confirmed MR^4.5 and <10% achieved stable MR^4.5.[36] In the German CML-Study IV, more patients achieved MR^4.5 by 6 years with high-dose imatinib (800 mg daily; 66%) than with standard-dose imatinib (400 mg daily; 56%), and patients achieved MR^4.5 more quickly with high-dose imatinib (median time to MR^4.5: imatinib 400 mg daily, 5.05 years; imatinib 800 mg daily, 3.47 years; p = 0.013);[37,38] however, imatinib 800 mg daily is not an approved treatment dose for patients with CML-CP.[39]

Although a similar length of follow-up has not been reported for second-generation TKIs, frontline trials of second-generation TKIs versus imatinib showed that patients achieved deeper molecular responses with second-generation TKIs, and they achieved such responses more rapidly [29–31,40–43] (Table 1). In the Evaluating Nilotinib Efficacy and Safety in Clinical Trials–Newly Diagnosed Patients (ENESTnd) trial, more than half of patients receiving nilotinib 300 mg twice daily (the approved frontline dose [44]) or nilotinib 400 mg twice daily achieved MR^4.5 by 6 years (55.7% and 54.8%, respectively), compared with 32.9% of patients receiving imatinib 400 mg once daily.[31] These results were confirmed by the single-arm Evaluating Nilotinib Efficacy and Safety in Clinical Trials as First-Line Treatment (ENEST1st) trial, which reported similar rates of MR^4.5 with frontline nilotinib through 2 years of follow-up.[45] Results from the final analysis of the Dasatinib Versus Imatinib Study in Treatment-Naive CML Patients (DASISION) trial showed that 42% of patients receiving dasatinib 100 mg once daily versus 33% of patients receiving imatinib 400 mg once daily achieved MR^4.5 by 5 years.[30]

Table 1. Cumulative rates of MR^4.5 according to frontline TKI^a.

Patients, %	1 Year	2 Years	3 Years	4 Years	5 Years	6 Years
ENESTnd [29,31,40–42]
Nilotinib 300 mg twice daily (n = 282)	11	25	32	40	53.5	55.7
p value versus imatinib	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001	<0.0001
Nilotinib 400 mg twice daily (n = 281)	7	19	28	37	52.3	54.8
p value versus imatinib	<0.0001	0.0006	0.0003	0.0002	<0.0001	<0.0001
Imatinib 400 mg once daily (n = 283)	1	9	15	23	31.4	32.9
ENEST1st^b [45]
Nilotinib 300 mg twice daily (n = 1052)	20.7	38.6	NR	NR	NR	NR
DASISION^c [30,43]
Dasatinib 100 mg once daily (n = 259)	5	19	24	34	42	NR
Imatinib 400 mg once daily (n = 260)	3	8	13	23	33	NR
p value	<0.2394	<0.0008	<0.0013	<0.0055	<0.0251	NR
CML-Study IV^d [38]
Imatinib 400 mg once daily (n = 337)	NR	20.9	NR	40.5	NR	55.6

AP: accelerated phase; BC: blast crisis; CML: chronic myeloid leukemia; DASISION: Dasatinib Versus Imatinib Study in Treatment-Naive CML Patients; ENEST1st: Evaluating Nilotinib Efficacy and Safety in Clinical Trials as First-Line Treatment; ENESTnd: Evaluating Nilotinib Efficacy and Safety in Clinical Trials–Newly Diagnosed Patients; MR^4.5: molecular response 4.5 (BCR-ABL1 ≤ 0.0032% on the International Scale); NR: not reported; TKI: tyrosine kinase inhibitor.

a Long-term rates of MR^4.5 have not been reported for bosutinib and ponatinib.

b In ENEST1st, patients were followed for only 2 years.

c In DASISION, patients were followed for only 5 years.

d In CML-Study IV, cumulative incidences were calculated by competing risk analysis, considering AP, BC, and death.

Bosutinib and ponatinib are not approved for frontline treatment of CML-CP; however, data from their respective frontline clinical trials, while limited, suggest that many patients may be able to achieve deep molecular responses on these agents as well.[34,35] In the Bosutinib Efficacy and Safety in Newly Diagnosed Chronic Myeloid Leukemia (BELA) trial, 16% of 250 bosutinib-treated patients and 12% of 252 imatinib-treated patients achieved MR⁴ at 24 months.[34] Rates of MR^4.5 in BELA have not been published to date. Further data on frontline bosutinib are anticipated from the ongoing Bosutinib Trial in First Line Chronic Myelogenous Leukemia Treatment (BFORE) trial of bosutinib versus imatinib in patients with newly diagnosed CML-CP.[46] The Evaluation of Ponatinib versus Imatinib in Chronic Myeloid Leukemia (EPIC) trial was terminated due to concerns about arterial thrombotic events observed in ponatinib-treated patients.[35] However, among evaluable patients who had been randomized to receive frontline ponatinib or imatinib prior to the study’s termination, overall rates of MR^4.5 were 15% (22/149) for ponatinib-treated patients and 0% (0/142) for imatinib-treated patients, with a median follow-up of 5.1 months (range, 0.03–17.6 months).[35]

Bosutinib and ponatinib are both indicated for second- or later-line treatment of CML: bosutinib for patients with prior treatment resistance or intolerance [33] and ponatinib for patients with the T315I mutation or for whom no other TKI is indicated.[32] Data from trials supporting these indications further demonstrated that patients may also be able to achieve deep molecular responses with bosutinib or ponatinib.[47,48] In a trial investigating second-line bosutinib in patients with CML-CP with resistance or intolerance to imatinib, 28% (55/200) of evaluable patients achieved undetectable BCR-ABL1 transcripts (with RQ-PCR sensitivity of ≥5 logs) by 2 years’ minimum follow-up.[47] In the Ponatinib Ph + Acute Lymphoblastic Leukemia and CML Evaluation (PACE) trial of ponatinib in heavily pretreated patients, 20% (53/267) of patients with CML-CP achieved MR^4.5 on ponatinib, with a minimum of 23.9 months of follow-up.[48] Achievement of MMR or deeper molecular responses has been reported in studies of second-line dasatinib [49] and nilotinib [50–52] as well. However, because the eligibility criteria for second-line TKI trials differ substantially, these results cannot be readily compared.

Molecular responses deeper than MMR are not currently described as treatment goals according to National Comprehensive Cancer Network (NCCN) or European LeukemiaNet (ELN) recommendations.[1,53] However, some studies have suggested that patients with deep molecular responses may have an especially good prognosis,[38,54,55] although additional studies are needed to further investigate the potential clinical importance of deep molecular responses. For example, an analysis of imatinib-treated patients in the German CML-Study IV (including patients treated with high-dose imatinib [800 mg once daily], standard-dose imatinib [400 mg once daily], or imatinib 400 mg once daily after IFN failure or in combination with IFN or cytarabine) found that achievement of confirmed MR^4.5 at 4 years was associated with a higher 8-year rate of overall survival (OS) than achievement of BCR-ABL1^IS 0.1% to 1% at 4 years (92% versus 83%, respectively; p = 0.047), and no patient with MR^4.5 experienced disease progression.[38] Similarly, a retrospective analysis by Etienne et al. found a higher 5-year rate of event-free survival (events for this study included death, disease progression, hematologic/cytogenetic relapse, detection of an imatinib-resistant mutation, and toxicity-related treatment cessation) among patients who achieved UMRD with RQ-PCR sensitivity to MR^4.5 at any time on imatinib (98.4%) versus those who achieved CCyR and MMR without UMRD (79.7%; p = 0.001) or versus those who achieved CCyR but not MMR or UMRD (27.7%; p < 0.0001).[54] In a separate retrospective analysis of patients with CML-CP who achieved CCyR, patients with a best molecular response of UMRD (in a standardized RQ-PCR test with sensitivity to detect 1 BCR-ABL1 transcript in a background of 100,000 ABL1 copies) had a better 6-year OS probability than patients with a best response of MMR (p < 0.0001 versus UMRD) or patients with no molecular response (p < 0.0001 versus UMRD); however, no association was detected between depth of molecular response (among patients with CCyR) at 18 or 24 months and rates of transformation-free survival or OS.[55]

Based on data demonstrating achievement of deep molecular responses with TKI treatment, studies evaluating the feasibility of TFR have been conducted with frontline imatinib and front- or second-line nilotinib and dasatinib. Published findings from these studies are discussed later. The possibility of TFR with bosutinib or ponatinib in the second- or later-line setting has not been investigated. Attempting TFR in the context of a clinical trial should only be considered in patients who have received treatment regimens for which achievement and maintenance of stable deep molecular response have been clearly demonstrated.

Who are the best candidates for treatment-free remission?

A sustained deep molecular response is considered necessary for successful TFR and most TFR trials have required this response prior to enrollment; however, the most appropriate criteria for study entry are still being determined (Table 2).[4–17] In reports of patients who stopped imatinib without achieving deep molecular responses, rapid relapses requiring treatment reinitiation were typical.[56,57] In addition to achievement of deep molecular responses, other potential predictors of TFR success have been assessed in a number of TFR studies (Table 2). In the Stop Imatinib 1 (STIM1) trial, patients with low Sokal risk had a higher probability of successful TFR than those with high Sokal risk.[17] In the Stop Second-Generation TKI (STOP 2G-TKI) and Dasatinib Discontinuation (DADI) trials, a history of imatinib resistance or suboptimal response was associated with a lower rate of TFR.[13,14] These results suggest that patients with a more aggressive CML biology may be less likely to maintain TFR. Results from the TWISTER trial, among patients with prior IFN treatment, showed an increased probability of maintaining TFR for patients with an IFN treatment duration of >12 months.[7] In addition, baseline immune markers, including high natural killer (NK) cell counts and low γδ + T-cell and CD4 + regulatory T-cell counts prior to stopping treatment, were associated with an increased likelihood of TFR in DADI.[14] Immunologic predictive factors of TFR maintenance have also been identified in substudies from STIM1 and the European Stop TKI (EURO-SKI) trial.[58–60] These findings suggest that immunogenic effects of treatment may affect the probability of TFR success. Other factors associated with successful TFR include achievement of UMRD ≤9 months after switching from IFN to imatinib,[7] duration of imatinib ≥62 months,[15] imatinib withdrawal syndrome (i.e. musculoskeletal pain and/or pruritus after stopping imatinib),[15] negative digital PCR at enrollment,[15,16] and age ≥45 years.[16]

Table 2. Data from key trials of treatment-free remission.

TFR trial	Patients	Response required to attempt TFR	Trigger to reinitiate therapy	Longest duration of follow-up reported, median (range), months	Probability of maintaining TFR, % (95% CI)	Factors associated with TFR success
STIM1 [4,5,10,17,58]	Adults with CML-CP or -AP (N = 100) with ≥3 years of imatinib treatment	UMRD with ≥50,000 ABL1 copies for ≥2 years	Loss of UMRD (2 consecutive assessments)^a or Loss of MMR (1 assessment)	65 (9–84)	6 months: 59 24 months: 41^b	Low Sokal risk, high NK cell counts^c
TWISTER [7]	Adults with CML-CP (N = 40) with ≥3 years of imatinib treatment	UMRD with sensitivity to BCR-ABL1^IS ≤ 0.0032% maintained for ≥2 years (minimum of 2 RQ-PCR tests per year)	Loss of UMRD (2 consecutive assessments) or Loss of MMR (1 assessment)	42 (15–72)	24 months: 47.1 (31.5–62.7)^d	Duration of IFN treatment >12 months and achievement of UMRD ≤9 months after switching from IFN to imatinib treatment^e
A-STIM [9]	Adults with CML-CP or -AP (N = 80) with ≥3 years of imatinib treatment	Confirmed (2 consecutive RQ-PCR tests) UMRD with ≥40,000 ABL1 control gene copies^f	Loss of MMR (1 assessment)	31 (8–92)	12 months: 64 (54–75)^d 24 months: 64 (54–75)^d 36 months: 61 (51–73)^d
KIDS [15]	Adults with CML-CP (N = 156) with ≥3 years of imatinib treatment	UMRD (≥ MR^5 sensitivity) for≥2 years	Loss of MMR (2 consecutive assessments)	26.6 (12.6–58.1)	12 months: 62.2^g 24 months: 58.5^g	Negativity of digital PCR at screening, duration of imatinib therapy ≥62 months, and imatinib withdrawal syndrome (i.e. aggravation or development of musculoskeletal pain and/or pruritus)^h
ISAV [16]	Patients with CML-CP or -AP (N = 112) with >2 years of imatinib therapy	UMRD (confirmed with ≥3 RQ-PCR tests) for ≥18 months	2 consecutive BCR-ABL1+ samples (≥ 1 with BCR-ABL1 >0.1%)	28.0 (95% CI, 25.5–30.1)	24 months: 50.9	Negativity of digital PCR prior to imatinib discontinuation and patient age (≥ 45 years)
EURO-SKI [12,59,60]	Adults with CML-CP (n = 200) with ≥36 months of TKI treatment and with no prior TKI failure^i	Confirmed (3 consecutive RQ-PCR tests) MR^4 maintained for≥1 year	Loss of MMR (1 assessment)	NR; 6–month minimum follow–up	6 months: 61.5 (54.4–68.3)	High NK cell counts, low frequency of CD86-positive plasmacytoid dendritic cells^j
STOP 2G-TKI [8,13]	Adults with CML-CP (n = 52^k) on nilotinib or dasatinib and with ≥36 months of treatment (prior imatinib allowed)	UMRD (≥ MR^4.5 sensitivity) maintained for ≥24 months	Loss of MMR (1 assessment)	32 (12–56)	12 months: 61.4 (48.1–74.6)^l 24 months: 57.0 (43.3–70.6)^l	History of suboptimal response or resistance to imatinib^m
DADI [14]	Patients with CML-CP (age ≥15 years; N = 63) on dasatinib following imatinib failure/intolerance	BCR-ABL1^IS < 0.0069% sustained for ≥1 year, as determined by a central laboratory	BCR-ABL1^IS ≥ 0.0069%(1 assessment)	20 (IQR, 16.5–24.0)	12 months: 48 (35–59)	History of imatinib resistance,^m high NK cell counts, low γδ+ T-cell count, low CD4+ regulatory T-cell count

AP: accelerated phase; A-STIM: According to STIM; CI: confidence interval; CML: chronic myeloid leukemia; CP: chronic phase; DADI: Dasatinib Discontinuation; EURO-SKI: European Stop Tyrosine Kinase Inhibitor; IFN: interferon alfa; IQR: interquartile range; ISAV: Imatinib Suspension and Validation; KIDS: Korean Imatinib Discontinuation Study; MMR: major molecular response (BCR-ABL1 ≤ 0.1% on the International Scale [IS]); MR⁴: molecular response 4 (BCR-ABL1^IS ≤ 0.01%); MR^4.5: molecular response 4.5 (BCR-ABL1^IS ≤ 0.0032%); NK: natural killer; NR: not reported; RQ-PCR: real-time quantitative polymerase chain reaction; STIM: Stop Imatinib; STOP 2G-TKI: Stop Second-Generation Tyrosine Kinase Inhibitor; TFR: treatment-free remission; TKI: tyrosine kinase inhibitor; UMRD: undetectable minimal residual disease.

a A 1-log BCR-ABL1 transcript increase between assessments was also required.

b With 65 months’ median follow-up, no molecular relapse has occurred since month 22.

c High NK cell counts were associated with a higher rate of TFR in a subset of 51 patients enrolled in STIM1.[58]

d Kaplan–Meier estimates of treatment-free remission.

e Among the 21 patients enrolled in TWISTER with IFN treatment prior to imatinib.

f Patients with occasional unconfirmed BCR-ABL1+ samples (BCR-ABL1^IS < 0.1%) were eligible.

g Probability of sustained MMR among 90 patients without hematopoietic cell transplantation with ≥12 months’ follow-up.

h In a previous analysis from KIDS, longer durations of imatinib therapy and UMRD prior to suspension of imatinib were associated with increased probability of sustained MMR.[94]

i Data from EURO-SKI are from a preplanned interim analysis of the first 200 patients with a 6-month assessment. Of these patients, 97% received frontline imatinib, 1.5% received frontline dasatinib, and 1.5% received frontline nilotinib; 24 patients had switched to second-line therapy due to intolerance (16 to dasatinib, 2 to imatinib, and 6 to nilotinib).

j High NK cell counts and low frequency of CD86-positive plasmacytoid dendritic cells were associated with increased rate of successful TFR in 2 substudies of 105 and 123 patients, respectively, enrolled in EURO-SKI [59,60].

k Data from STOP 2G-TKI were reported for 52 patients with ≥12 months of follow-up.

l Treatment-free survival without MMR loss.

m Associated with a significantly lower chance of successful TFR.

TFR remains investigational,[1,53] and as such, the most appropriate triggers for reinitiating treatment are unclear. Although there is no established definition, the term molecular relapse usually refers to the study protocol-defined BCR-ABL1 transcript level at which a patient should restart TKI treatment. Molecular relapse is generally not associated with cytogenetic or hematologic relapse [3,9] and does not indicate treatment failure or disease progression according to NCCN or ELN guidelines. In early imatinib TFR studies (STIM1 and TWISTER trials), treatment was restarted following confirmed BCR-ABL1 positivity by RQ-PCR;[4,5,7] however, loss of MMR was shown to be a safe trigger for treatment reinitiation in subsequent studies, including the According to STIM (A-STIM), Korean Imatinib Discontinuation Study (KIDS), EURO-SKI, and STOP 2G-TKI trials.[9,12,13,15] These studies demonstrated that a proportion of patients may have detectable BCR-ABL1 transcripts following TKI suspension without further loss of response.[8,9,12,13,15]

Long-term TKI therapy can eliminate the majority of CML cells but may leave behind a small population of TKI-resistant, quiescent leukemic stem cells.[61] As such, in comparison to patients with newly diagnosed CML-CP, patients with persistent minimal residual disease (MRD) after receiving TKI therapy may have a higher proportion of TKI-resistant leukemic stem cells.[61,62] However, although a small number of CML cells appear to persist in some patients with deep molecular responses on TKI therapy, a growing body of data suggests that some patients may be able to remain in TFR despite low levels of detectable BCR-ABL1 transcripts.[8,9,12,13,15] The mechanism by which patients maintain TFR despite low levels of residual disease is unknown but may involve immunologic control.[63] IFN has been investigated as a potential maintenance therapy for patients who stop TKI treatment, with 13 of 20 patients remaining relapse free through 7.9 years’ median follow-up after imatinib discontinuation in 1 study.[64] Another ongoing study is evaluating IFN as a maintenance therapy following nilotinib discontinuation.[65]

Results from published trials of treatment-free remission

Although the eligibility criteria and treatment-reinitiation triggers differ across studies, approximately 40–60% of patients with CML-CP enrolled in TFR studies have been able to successfully maintain TFR after stopping TKI therapy.[3–16] Longer follow-up is needed to determine how long patients will ultimately be able to maintain TFR. At present, available evidence shows that most molecular relapses occurred within 6 months of TKI discontinuation.[7,9,13,14,17] With 65 months’ median follow-up in STIM1, 61 of 100 patients had experienced molecular relapse (defined as BCR-ABL1 positivity in 2 consecutive assessments with a 1-log increase in BCR-ABL1 transcripts between assessments, or loss of MMR in a single assessment) and 39 remained free from molecular relapse (median duration of follow-up among patients remaining in TFR, 67 months).[17] Among the 61 patients with molecular relapse in STIM1, 95% (58 of 61) experienced relapse within 7 months of stopping imatinib, whereas 3 relapses occurred between 18 and 22 months.[17] In TWISTER, 22 of 40 enrolled patients experienced relapse (i.e. loss of UMRD in 2 consecutive assessments or loss of MMR in a single assessment) with a median follow-up of 42 months; 68% of relapses (15 of 22) occurred during the first 6 months off treatment, and the latest relapse was observed at 27 months after stopping imatinib.[7] In A-STIM, with 31 months’ median follow-up, molecular relapse (defined as loss of MMR) was observed in 29 of 80 patients, with a median time to relapse of 4 months (range, 2–17 months).[9] In the Imatinib Suspension and Validation (ISAV) study, 53 of 108 patients (49.1%) experienced a molecular relapse (2 consecutive BCR-ABL1–positive samples, with 1 sample >0.1%) by 2 years; 73.6% of molecular relapses occurred within 9 months of imatinib discontinuation.[16]

Although the majority of TFR data reported to date have been from imatinib-treated patients, reports of TFR in nilotinib- or dasatinib-treated patients demonstrate the feasibility of TFR in these patients as well. With 32 months’ median follow-up in STOP 2G-TKI, 24 of 52 patients experienced molecular relapse (i.e. loss of MMR) after stopping frontline or second-line nilotinib or dasatinib; the median time to loss of MMR was 4 months (range, 1–38 months).[13] With 12 months’ follow-up in DADI, 33 of the 63 patients who discontinued second-line dasatinib experienced molecular relapse (defined as BCR-ABL1^IS ≥ 0.0069% at any analysis point), all occurring within 7 months of stopping dasatinib.[14] Results from a number of ongoing studies (e.g. ENESTfreedom, ENESTop, ENESTgoal, ENESTpath, DASFREE) may provide further evidence of the feasibility of TFR following treatment with nilotinib or dasatinib in the front- or second-line setting.[66–69]

Of the patients who experienced molecular relapse in clinical studies, the majority were able to regain deep molecular responses after restarting therapy. Only 1 patient has been reported to experience disease progression after stopping TKI treatment in a TFR study; this patient restarted imatinib and reachieved MMR but later progressed to blast crisis and received an allogeneic HSCT.[9] In STIM1, 57 of the 61 patients with molecular relapse reinitiated TKI therapy, and 55 regained UMRD.[17] All patients in TWISTER who experienced molecular relapse (n = 22) had ≥1 UMRD sample with imatinib retreatment, with a median time of 3 months to achieve UMRD.[7] In A-STIM, all 31 patients who reinitiated treatment following loss of MMR regained the response, and 23 regained UMRD.[9] With 26.6 months’ median follow-up in KIDS, all 37 patients who were retreated with imatinib following loss of MMR regained MMR (median time to regain MMR, 3.9 months) and 32 regained UMRD (median time to regain UMRD, 7.2 months).[15] Of the 24 patients who experienced molecular relapse in STOP 2G-TKI, all but 1 regained MMR after a median of 3 months of retreatment.[13] In the DADI study, 29 of 33 patients who reinitiated TKI therapy (dasatinib, n = 32; nilotinib, n = 1) after molecular relapse regained BCR-ABL1^IS < 0.0069% within 3 months of retreatment.[14] Overall, these results demonstrate that many patients with sustained deep molecular responses can maintain TFR for at least several years, although how long patients will ultimately be able to maintain TFR remains unknown.

Molecular monitoring of patients with CML-CP

Reliable methods for measuring molecular response are needed, both for identifying patients who obtain deep responses and also for monitoring disease state after suspension of therapy in patients attempting TFR. Molecular responses are measured using RQ-PCR and are reported based on the ratio of BCR-ABL1 transcripts to control gene transcripts standardized to the IS.[70,71] Regular molecular monitoring of BCR-ABL1 transcript levels is considered standard care for patients with CML-CP according to the NCCN and the ELN.[1,53] The NCCN recommends that patients receive molecular monitoring via RQ-PCR testing of BCR-ABL1 in an IS-standardized laboratory every 3 months upon initiation of TKI treatment; after patients achieve CCyR, BCR-ABL1 testing is recommended every 3 months for 2 years and every 3–6 months thereafter while on TKI treatment.[1] NCCN guidelines do not consider the absence of MMR a sign of treatment failure in patients with CCyR.[1] Per ELN recommendations, patients on TKI therapy should receive molecular monitoring every 3 months until MMR is achieved and then every 3 to 6 months after achieving MMR.[53] According to the ELN, MMR at 12 months and thereafter is considered an optimal response to TKI therapy.[53] However, for patients who attempt TFR following achievement of stable deep molecular remission, molecular monitoring should be performed monthly after stopping TKI treatment for the first 6 months to identify those at risk of early relapse. Afterward, a decrease in the frequency of monitoring may be possible because the majority of observed molecular relapses in TFR trials have occurred by 6 months;[7,9,13,14,17] however, most TFR trials have continued monthly monitoring beyond this time. For example, in the STIM1, A-STIM, and TWISTER TFR trials, patients who stopped imatinib were monitored monthly for the first year, every 2 months for the second year, and every 3 months thereafter.[4,7,9]

Side effects of TKI therapy may be motivators for attempting TFR

Evidence suggests that long-term TKI use may be associated with unique long-term adverse events (AEs), particularly cardiopulmonary or atherosclerotic events, which may contribute to both morbidity and even mortality with long-term use.[30,31,35,48,72] The most serious of these have included myocardial infarctions, cerebrovascular events, peripheral vascular events, pulmonary hypertension, and significant pleural or pericardial effusions occasionally requiring invasive interventions.[30,31,35,48,72,73] Cardiovascular events (CVEs) have been more frequently reported with second-generation TKIs than with imatinib.[30,31,35,72] Through 6 years of follow-up in ENESTnd, CVEs have been reported in 28 (10.0%) and 44 (15.9%) patients in the nilotinib 300 mg twice-daily and nilotinib 400 mg twice-daily arms, respectively, compared with 7 (2.5%) patients in the imatinib 400 mg once-daily arm.[31] The majority of these CVEs were ischemic heart disease (nilotinib 300 mg, n = 14; nilotinib 400 mg, n = 28; imatinib 400 mg, n = 6), although peripheral artery disease (PAD) was observed in 12, 9, and 0 patients, respectively, and ischemic cerebrovascular events were observed in 4, 9, and 1 patients, respectively, in the nilotinib 300-mg twice-daily, nilotinib 400-mg twice-daily, and imatinib 400-mg once-daily arms.[31] In the final, 5-year analysis from DASISION, CVEs were also more commonly observed with dasatinib than with imatinib: 12 (5%) patients on dasatinib and 6 (2%) patients on imatinib had arterial ischemic events.[30] These events included CV ischemic events (dasatinib, n = 10; imatinib, n = 4), transient ischemic attack (dasatinib, n = 2; imatinib, n = 0), and PAD (dasatinib, n = 0; imatinib, n = 2).[30] Long-term safety data on frontline bosutinib and ponatinib are not yet available, but CVEs have been reported in early follow-up analyses. With 2 years’ follow-up in BELA, CVEs (most commonly hypertension and palpitations) were reported in 25 (10%) and 19 (8%) patients in the bosutinib and imatinib arms, respectively; no patients had myocardial infarction or PAD.[34] The EPIC trial was terminated early due to the high rate of CVEs observed with ponatinib: with 5.1 months’ median follow-up, 11 (7%) patients in the ponatinib arm and 3 (2%) patients in the imatinib arm experienced arterial thrombotic events.[35]

Other long-term safety concerns have been linked primarily to dasatinib treatment, including pulmonary hypertension and pleural effusions.[30,74,75] Through 5 years of follow-up in DASISION, 14 (5%) patients in the dasatinib arm experienced pulmonary hypertension,[30] an increase from the 8 cases of pulmonary hypertension reported after 3 years of follow-up.[75] No patients in the imatinib arm of DASISION had pulmonary hypertension through 3 years of follow-up,[75] and the rate in the imatinib arm was not provided in the 5-year report.[30] The rate of pleural effusions in the dasatinib arm increased with each year of follow-up in DASISION, with 74 (29%) patients experiencing pleural effusion by 5 years.[30] The rate of pleural effusions on imatinib was not reported by 5 years,[30] although <1% of patients on imatinib experienced pleural effusions through 3 years.[75] Pleural effusions with dasatinib may be dose dependent,[76] and data have indicated that the incidence of pleural effusions could be reduced through dose-optimization strategies that use plasma drug-level monitoring,[77] although drug-level monitoring is not widely available in the United States.

With long-term therapy, even low-grade AEs can have significant impacts on patients’ quality of life (QoL).[78–80] In an analysis of 448 patients with CML on long-term imatinib therapy (median duration of 5 years) using the Medical Outcomes Study 36-Item Short Form (SF-36) Health Survey to assess health-related QoL, severe AEs reported in >25% of patients included muscle cramps, fatigue, edema, and musculoskeletal pain.[80] Fatigue of any grade was reported by 82% of patients.[80] Compared with control populations, female patients with CML reported greater reductions in health-related QoL than male patients with CML.[80] In addition, patients with CML between the ages of 18 and 39 years reported more QoL impairments than any other age group relative to age-matched population controls.[80] In an analysis of 152 patients with CML (average time from diagnosis to study entry, 72.5 months) monitored every 2 weeks for 1 year using a CML-specific module for the MD Anderson Symptom Inventory (MDASI), 30% of patients reported persistent moderate-to-severe symptoms that interfered with daily functioning.[79] Overall, patients receiving imatinib reported higher CML-specific symptom severity (1.75) versus patients receiving dasatinib or nilotinib (1.04; p = 0.002) as well as more severe diarrhea (2.06 versus 0.45; p < 0.001), muscle soreness and cramping (2.44 versus 1.42; p = 0.017), and swelling (2.08 versus 1.27; p = 0.044). Guérin et al. used the SF-36 and Functional Assessment of Cancer Therapy–Leukemia (FACT-Leu) surveys to assess health-related QoL in 593 patients enrolled in ENESTnd and evaluated the impact of low-grade AEs on health-related QoL in these patients (with 48 months of follow-up in ENESTnd).[78] Impairments in health-related QoL due to low-grade AEs in the following categories were detected: gastrointestinal disorders (e.g. diarrhea, nausea), blood/lymphatic disorders (e.g. thrombocytopenia, neutropenia), musculoskeletal disorders (e.g. muscle spasms, arthralgia), psychiatric disorders (e.g. anxiety, insomnia), and general disorders/administration-site conditions (e.g. fatigue, peripheral edema).[78] Incidence rates of AEs in each category that were found to negatively impact health-related QoL were similar or lower in patients receiving nilotinib 300 or 400 mg twice daily versus patients receiving imatinib.[78]

For patients experiencing chronic AEs with deep molecular responses, the potential to safely attempt TFR may be very attractive. In a survey of 40 patients with CML-CP conducted by Boquimpani et al., relief of TKI side effects was the most frequently cited (75%) positive impact anticipated from successful TFR.[81] Although it may be anticipated that AEs will resolve in patients in TFR,[81] stopping TKI therapy may also lead to the development of AEs in some patients.[12,15,82–84] The development of a potential imatinib withdrawal syndrome was first noted in the Swedish cohort of EURO-SKI, in which 30% of patients (15/50) developed musculoskeletal pain within 6 weeks of stopping TKI therapy.[12,83] Since this report, evidence of musculoskeletal pain following imatinib discontinuation has been reported elsewhere.[15,82,84] The potential for a TKI withdrawal syndrome requires further investigation in ongoing TFR trials. The alleviation or avoidance of chronic or long-term AEs provides a compelling reason for attempting TFR; however, TFR should be attempted only in a clinical trial where patients receive frequent molecular monitoring and can reinitiate TKI therapy promptly following molecular relapse.

Other potential motivations for attempting TFR

There are additional treatment considerations that may motivate some patients to attempt TFR, as indicated by patient surveys.[81,85,86] In the survey by Boquimpani et al., positive expectations associated with TFR included relief of medication side effects (75%), reduced financial burden (58%), convenience (43%), positive emotional impact (43%), and increased activity level (30%).[81] Overall 58% of patients reported being cautiously positive about attempting TFR, and with a recommendation from their physician, 77% of patients aged ≥60 years and 52% of those aged <60 years reported that they would consider attempting TFR.[81] In addition to their physician’s recommendation, patients’ perception of whether TFR is safe may also influence their willingness to attempt TFR.[85,87] In a survey of 56 patients with CML, Sanford et al. found that 71.4% of patients would be willing to attempt TFR if they were assured that they would receive adequate monitoring for relapse.[85] Patient-reported willingness to attempt TFR in Sanford et al. was inversely associated with the likelihood of experiencing relapse.[85] In a single-institution survey of 84 US patients (of whom 82% were receiving TKI therapy) by Goldberg et al., if offered the opportunity to stop TKI therapy, 42% of patients would, 34% would remain on TKI therapy, and 25% were not sure.[86] The main patient-reported motivators for attempting TFR were fewer side effects (40%), lower costs (30%), and convenience (26%).[86] Similar to the results of other patient surveys,[85,87] safety of TFR was a major concern, with worry about the potential for disease recurrence reported as the most common reason (58%) to not attempt TFR.[86]

Due to the risks of TKI exposure to a fetus,[88,89] there may be interest in the possibility of TFR to accommodate a pregnancy. In women exposed to imatinib during pregnancy, high rates of fetal abnormalities and spontaneous abortion were observed.[88] Similar cases have been observed in women receiving dasatinib during pregnancy.[90] To minimize such cases, patients on TKI therapy should be educated regarding the risks of pregnancy and encouraged to use highly effective means of contraception.[32,33,39,44,91] The NCCN and ELN recommend that, on an individual basis, discontinuation of TKI therapy may be considered in patients responding to TKI therapy who wish to become pregnant.[1,53] A report by Kuwabara et al. followed 7 patients who discontinued imatinib due to pregnancy or plans to conceive.[56] Treatment responses prior to imatinib discontinuation were MMR (n = 3), CCyR (n = 3), or major cytogenetic response (MCyR; 0–35% Ph + metaphases; n = 1).[56] With a median time off imatinib of 9 months (range, 6–23 months), all patients experienced loss of CCyR prior to reinitiation of imatinib. All 3 patients with MMR prior to imatinib suspension achieved MMR following treatment reinitiation.[56] Among the 4 patients without MMR prior to treatment discontinuation, 2 failed to achieve MCyR with imatinib retreatment, including 1 who experienced a hematologic relapse.[56] This report highlights the importance of sustained deep molecular responses prior to TFR and also shows the need for established guidelines for safely attempting TFR.

An additional motivation for attempting TFR may be the alleviation of the financial burden of TKI therapy.[81,86,92] In one analysis of health insurance claims from 1541 patients with CML on imatinib in the United States, higher out-of-pocket costs were significantly associated with lower adherence to imatinib and a greater likelihood of stopping imatinib treatment.[93] In the survey by Boquimpani et al., 53% of patients noted the financial burden of CML treatment and 58% anticipated reduced financial burden as a benefit of TFR.[81] Sanford et al. reported that 44.6% of patients surveyed were concerned about their ability to afford CML treatment if they attempted TFR and experienced molecular relapse.[85] With a median follow-up of 50 months (range, 9–72 months) in the STIM1 trial, suspension of imatinib therapy for the 100 enrolled patients was estimated to have saved a total of 5.5 million Euros,[10] demonstrating the potential economic impact of TFR. However, the actual economic impact for patients may depend on their specific situation because potential savings from being treatment free may be counterbalanced by the cost of more frequent molecular monitoring required after stopping TKI therapy.

Conclusions

TFR may be an attractive treatment goal for oncologists and their patients for numerous reasons, including achievement of deep molecular responses, alleviation of long-term AEs, pregnancy, or economic considerations. For patients interested in attempting TFR within the context of a clinical trial, TKI choice may impact their likelihood of achieving the necessary level of deep molecular response, and the establishment of clear guidelines for attempting TFR could help guide treatment decisions. The rates of deep molecular response vary between different frontline TKIs, with a higher rate of MR^4.5 observed with the second-generation TKIs than with imatinib,[29–31,40–43] which may motivate patients and oncologists with TFR as a treatment goal to choose a second-generation TKI.

Acknowledgments

The authors thank Karen Kaluza, PhD, and Jonathan Morgan, PhD (Articulate Science), for medical editorial assistance with this manuscript. Financial support for medical editorial assistance was provided by Novartis Pharmaceuticals.

Potential conflict of interest

Disclosure forms provided by the authors are available with the full text of this article at http://dx.doi.org/10.1080/10428194.2016.1198959.