Long-term outcomes of third-line therapy with tyrosine kinase inhibitors in chronic phase chronic myeloid leukemia: A real-life experience

Introduction Tyrosine kinase inhibitor (TKI) therapy has greatly improved the prognosis of patients with chronic myeloid leukemia (CML), improving the survival expectancy of patients with chronic phase (CP) CML to that of the general population. However, despite these advances, nearly 50% of patients with CP CML experience failure to respond to frontline therapy, and most fail to respond to the subsequent second-line TKI. Treatment guidelines for patients failing second-line therapy are lacking. This study aimed to determine the efficacy of TKIs as third-line therapy in a “real-world” clinical practice setting and identify factors favorably influencing the long-term outcomes of therapy. Methods We have retrospectively analyzed the medical records of 100 patients with CP CML. Results The median age of the patients was 51 (range, 21–88) years, and 36% of the patients were men. The median duration of the third-line TKI therapy was 22 (range, 1– 147) months. Overall, the rate of achieving complete cytogenetic response (CCyR) was 35%. Among the four patient groups with different levels of responses at baseline, the best results were achieved in the groups with any CyR at the baseline of third-line therapy. Thus, СCyR was reached in all 15 and 8/ 16 (50%) patients with partial cytogenetic response (PCyR) or minimal or minor CyR (mmCyR), respectively, whereas CCyR was detected only in 12/69 (17%) patients without any CyR at baseline (p < 0.001). Univariate regression analysis revealed that the factors negatively associated with CCyR achievement in thirdline TKI therapy were the absence of any CyR on first- or second-line TKI therapy (p < 0.001), absence of CHR prior to third-line TKI (p = 0.003), and absence of any CyR prior to third-line TKI (p < 0.001). During the median observation time from treatment initiation to the last visit [56 (4–180) months], 27% of cases progressed into accelerated phase or blast phase CML, and 32% of patients died. Discussion Progression-free survival (PFS) and overall survival (OS) were significantly higher in patients with CCyR on third-line than in the group without CCyR on third-line therapy. At the last visit, third-line TKI therapy was ongoing in 18% of patients, with a median time of treatment exposure of 58 (range, 6–140) months; 83% of these patients had stable and durable CCyR, suggesting that patients without CHR at baseline and without CCyR at least by 12 months on third-line TKI should be candidates for allogeneic stem cell transplantation, third-generation TKIs, or experimental therapies.

Results: The median age of the patients was 51 (range, 21-88) years, and 36% of the patients were men. The median duration of the third-line TKI therapy was 22 (range, 1-147) months. Overall, the rate of achieving complete cytogenetic response (CCyR) was 35%. Among the four patient groups with different levels of responses at baseline, the best results were achieved in the groups with any CyR at the baseline of third-line therapy. Thus, СCyR was reached in all 15 and 8/ 16 (50%) patients with partial cytogenetic response (PCyR) or minimal or minor CyR (mmCyR), respectively, whereas CCyR was detected only in 12/69 (17%) patients without any CyR at baseline (p < 0.001). Univariate regression analysis revealed that the factors negatively associated with CCyR achievement in thirdline TKI therapy were the absence of any CyR on firstor second-line TKI therapy (p < 0.001), absence of CHR prior to third-line TKI (p = 0.003), and absence of any CyR prior to third-line TKI (p < 0.001). During the median observation time from treatment initiation to the last visit [56 (4-180) months], 27% of cases progressed into accelerated phase or blast phase CML, and 32% of patients died.
Discussion: Progression-free survival (PFS) and overall survival (OS) were significantly higher in patients with CCyR on third-line than in the group without CCyR on third-line therapy. At the last visit, third-line TKI therapy was ongoing in 18% of patients, with a median time of treatment exposure of 58 (range, 6-140) months; 83% of these patients had stable and durable CCyR, suggesting that patients without CHR at baseline and without CCyR at least by 12 months on third-line TKI should be candidates for allogeneic stem cell transplantation, third-generation TKIs, or experimental therapies. KEYWORDS chronic myeloid leukemia, chronic phase, efficacy of therapy, complete cytogenetic response, third-line therapy

Introduction
Tyrosine kinase inhibitors (TKIs) have significantly improved the outcomes of patients with chronic myeloid leukemia (CML), seemingly increasing their life expectancy to that of the general population (1). However, 35%-45% of patients fail to respond to first-line TKI within 10 years of therapy (2,3). Most patients after first-line treatment discontinuation are switched to second-generation TKI. Overall, nilotinib, dasatinib, and bosutinib have demonstrated similar response rates, with probabilities of obtaining a complete cytogenetic response (CCyR) of approximately 50%. After 5 years of observation, up to 60% of patients discontinue second-line therapy mainly due to primary or secondary resistance (4)(5)(6). Treatment and response recommendations focus on firstand second-line therapy with firstgeneration (imatinib) or second-generation (nilotinib, dasatinib, or bosutinib) TKIs (7,8). Currently, the ELN 2020 guidelines suggest the use of the third-generation TKI ponatinib as third-line therapy in patients without specific mutations rather than alternative secondgeneration TKIs. Patients with suboptimal response to two or more previous TKIs should be transferred to allogeneic stem cell transplantation (allo-SCT) (7). Allo-SCT may be considered for patients with de-novo blast phase (BP) CML, preferably after achieving some response to a TKI-based therapy, those with accelerated phase (AP) CML who are not responding well to current therapy, those who progressed to AP/BP CML while receiving TKI therapy, and those with resistance or intolerance to TKIs. It may also be used in patients with T315I mutation after an inadequate response to attempted ponatinib therapy (9). However, the possibilities for its implementation are limited due to the risks associated with age, comorbidities, and/or lack of matched donors. Globally, only a small proportion (≤1%) of patients with chronic phase (CP) CML undergo allo-SCT after failure of second-line therapy, while most patients continue to receive other TKIs (10).
It is very important to identify a group of patients in whom third-line therapy with TKIs is associated with equal or even better long-term outcomes than using allo-SCT. As a reference point for determining the effectiveness of third-line TKI, an assessment of the achievement of CCyR has been proposed. CyR, especially CCyR, has historically been and is currently associated with a significant survival advantage in patients with CP CML (11,12).
This study aimed to evaluate the long-term efficacy of third-line therapy with available TKIs and identify the factors that could favorably influence CCyR achievement and survival prognosis.

Patients
This multicenter retrospective observational study that aimed to evaluate the long-term efficacy of third-line therapy with TKI (third TKI) in patients with CP CML in real-life practice was conducted between 2019 and 2022. The inclusion criteria were as follows: 1) age ≥18 years, 2) CP CML (ELN criteria 2013), 3) treatment with available third-line TKI outside clinical trials with any reason of previous discontinuation, and 4) absence of CCyR at baseline. The exclusion criteria were 1) history of AP/BP and 2) prior allo-HSCT. All patients gave informed consent prior to participation. All study procedures were performed in accordance with the institutional and national ethical standards on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Overall, 100 [men, n = 36 (36%)] patients from six centers of the Saint Petersburg and Leningrad region were included. The data of 73 patients were collected primarily in 2019 and updated in 2022. The data of the other 27 patients were collected in 2022. In all cases, the diagnosis of CML was confirmed using cytogenetic and molecular analysis (13). At the time of diagnosis and time of initiation of third-line TKI, the median age of the patients was 45 (range, 12-82) and 51 (range, 21-88) years, respectively. First-line TKI treatment for most patients was imatinib (n = 97, 97%). Nilotinib, dasatinib, imatinib, and bosutinib were used as second-line therapy in 69 (69%), 25 (25%), 1 (1%), and 5 (5%) patients, respectively. The median duration of CML from the time of diagnosis to first-line TKI and third-line TKI and from the initiation of first-line TKI and second-line TKI therapy to the initiation of third-line therapy was 2 (range, 1-245), 64 (range, 7-316), 48 (range, 7-156), and 17 (3-96) months, respectively. The main characteristics of the patients prior to thirdline TKI initiation are shown in Table 1.
Overall, any time before the initiation of third-line TKI therapy, BCR::ABL mutational analysis was performed in 91/100 (91%) patients. BCR::ABL mutations were evaluated on both firstand second-line  At the baseline of third-line TKI therapy, 35 (35%), 34 (34%), 16 (16%), and 15 (15%) patients did not have a complete hematologic response (CHR) and had CHR without cytogenetic response (CyR), minor or minimal CyR (mmCyR), and partial CyR (PCyR), respectively. No patients with CCyR or deeper responses at baseline were included in the study according to the inclusion criteria.

Statistical analysis
The cumulative rate of CCyR achievement on third-line TKI was assessed as the primary efficacy endpoint of third-line TKI. All decimal numbers were converted to the nearest whole number. CCyR was defined as 0% of Ph+ metaphases, with at least 20 metaphases required, according to bone marrow cytogenetics or 2log reduction of molecular response (MR2), with a BCR::ABL/ABL ratio of ≤1% in peripheral blood by reverse transcription quantitative real-time polymerase chain reaction test (14).
To test the type of distribution of quantitative variables, the Shapiro-Wilk W-test was used. Variables did not have a normal distribution in any of the cases. Continuous variables were reported as medians and ranges. Categorical variables were reported as absolute values and percentages. Differences between categorical variables were calculated using the c 2 test, and differences between continuous variables were calculated using the Mann-Whitney U test.
Overall survival (OS) was defined as the time from the initiation of third-line TKI therapy to death. Progression-free survival (PFS) was defined as the time from the initiation of third-line TKI therapy to disease progression or death, whichever occurred first. Event-free survival (EFS) was defined as the time from the initiation of thirdline therapy to therapy withdrawal, disease progression, or death, whichever occured first (15). OS and PFS were calculated using the Kaplan-Meier method. The log-rank test was used to assess the significance of differences in survival between the patient groups.
The influence of possible risk factors on CCyR achievement was assessed by univariate analysis.

Results
At the time of the last data collection, the median duration of third-line TKI therapy was 22 (range, 1-147) months, and the median time of follow-up from the initiation of third-line TKI and from diagnosis to the last visit was 56 (range, 4-180) months and 125 (range, 27-434) months, respectively. Only four patients were lost to follow-up after a median observation time of 29 (range, 5-79) months.

CHR
At baseline, 65 patients had CHR, and 26/35 (74%) patients who did not have CHR at baseline achieved it by 3 months of therapy. CHR was subsequently lost in 9/26 (35%) and 5/65 (8%) patients without and with CHR at baseline, respectively (p = 0.002). The influence of the baseline level of response to CCyR in patients undergoing third-line TKI therapy was assessed. The median time on third-line TKI therapy for patients without CHR, with CHR but no CyR, and with any CyR (mmCyR or PCyR) was 20 (range, 1-95), 18 (range, 1-147), 34 (range, 5-140), and 32.5 (range, 3.5-121) months, respectively. CCyR was reached in 30/65 (46%) and 5/35 (14%) patients with and without CHR at baseline (p = 0.02). No differences in CCyR achievement were observed in patients without CHR and with CHR but no CyR (p > 0.05). Meanwhile, CCyR was significantly higher in patients with any baseline CyR than in patients with no CyR with or without CHR (p < 0.05). Moreover, all patients with any CyR at baseline reached CCyR during therapy (p = 0.013). The rate of CCyR in the patient groups is presented in Table 2.

Progression to advanced phases and PFS
The expected PFS was similar and was very high among patients with any CyR at baseline. It was lower in patients with CHR but no CyR than in other patients. There were no significant differences between patients with different levels of cytogenetic resistance at baseline. The worst PFS was observed in patients with hematologic resistance (Figure 4).
on treatment and 88%, 72%, 63%, and 52% in patients with no CCyR on treatment, respectively, with significant differences at any time point ( Figure 5).

OS
With the median observation time from the beginning of thirdline therapy to the last visit of 56 (range, 3.5-179.5) months, 32/100 (32%) deaths occurred, 21 of which were due to CML progression, 6 occurred after complications of allo-SCT (n = 3, allo-SCT was performed due to BP), and 5 occurred for other reasons [cardiovascular disease, n = 3; other secondary neoplastic processes, n = 1; and SARS-CoV-2 infection, n = 1 (1)]. Drugrelated toxicity did not cause any deaths ( Table 4). The causes of death in different groups are shown in Table 4.
After a nearly equal observation time from third-line treatment, it was found that the rate of death was lower in the group with any depth of CyR than in the group with no CyR but with CHR and the group without CHR at baseline [2/31 (6.5%) vs. 10/34 (29%) vs. 20/ 35 (57%), p < 0.01]. Simultaneously, patients with mmCyR and PCyR have a similar mortality rate of 1/16 (6%) vs. 1/15 (7%) (p = 0.9), respectively.
The expected OS was the lowest in patients with no initial CHR. It was higher in patients with any CyR than in patients with CHR but no CyR (Figure 7).

Discussion
Most patients were treated with dasatinib and nilotinib as thirdline TKI. Bosutinib was the treatment of choice only in 17% of patients. The drug was registered in our country much later than other TKIs and was available for fewer patients. Ponatinib was not registered in the country at the time of data collection. Only three patients were provided with ponatinib via different programs outside of clinical trials. Owing to recent advances in TKI therapy, patients with CP CML have a low incidence of progression to AP/BP and CML-related death. Moreover, up to 30% of patients with a stable deep molecular response have a chance of treatment-free remission. Meanwhile, up to 45% and 60% of patients, respectively, discontinue firstand second-line TKI therapy due to resistance or intolerance. Current guidelines consider ponatinib and allo-SCT as a third-line strategy, especially for patients with TKI resistance, but in real-life practice, nearly all patients are switched to third-line TKI therapy with second-generation TKI because of different socioeconomic status and inaccessibility of third-generation TKI. In our previous study (16), only 21% of new cases of CCyR among patients with CP CML undergoing second-generation TKI therapy continued to third-line therapy, and 21/53 (40%) patients experienced drug withdrawal in less than 2 years of observation. The 2-year OS was 67%. All patients with major CyR on third-line TKI were in stable CP and alive during observation (16). In our current retrospective study, CCyR was a significant marker for both PFR and OS. Among the 35 patients with CCyR, only one patient died due to progression to AP/BP and another one died for a reason not related to CML. Both PFS and OS were significantly higher at any time point in patients with CCyR than in patients with no CCyR during third-line treatment. In these groups of patients, the 5-year PFS and OS were 93% vs. 52% and 93% vs. 55%, respectively (p < 0.05). According to the results of our study, 27/100 (27%) patients progressed to advanced phases. Progression to BP occurred mostly among patients without CHR at baseline [n = 17/35 (49%)]. There were only two progressions in the mmCyR group [2/16 (12.5%)]. Disease progression did not occur among patients with PCyR at baseline. Similar results were shown by Garg et al. (17), where transformation to AP/BP was reported in 12/48 (25%) patients. Among patients with AP/BP progression, most [8/12 (67%)] did not achieve any CyR prior to third-line TKI therapy (17). In an article by Ribeiro et al. (18), among 18 patients with CP CML, only 3/18 (16%) progressed to BP on third-line therapy. Khan et al. (19) reported that the depth of response to therapy (both CCyR and MMR) was associated with improved OS in the univariate analysis. In a Progression-free survival (PFS) on third-line TKI therapy. PFS on third-line TKI therapy according to the depth of response at baseline. There were only 1/35 (3%) and 26/65 (40%) cases of AP/BP among patients who achieved or did not achieve CCyR on third-line TKI therapy (p = 0.0001). PFS on third-line therapy according to CCyR achievement on thirdline TKI. multivariate analysis, MMR was a favorable prognostic marker for OS (19). In our study group, MMR or better molecular responses were reached in 17/35 (49%) patients with CCyR. All patients were alive at the last visit, and 14/17 (82%) patients had stable MMR or better molecular response. CCyR loss was observed only in one out of 17 (6%) cases among patients with MMR achievement, as compared with 12/18 (67%) cases of CCyR without MMR achievement (p = 0.001). As only two patients died in the CCyR group, no difference was considered present between patients with and without MMR achievement during observation. In our observation, 35/100 (35%) patients reached CCyR. In our study, loss of CCyR on third-line therapy occurred in 13/35 (37%) patients. Ongoren et al. (20) reported that most patients on thirdline TKI therapy achieved CCyR [11/21 (52%)], and only one patient lost CCyR on third-line therapy [1/11 (9%)] (20).
While CCyR was a strong surrogate marker of survival on thirdline TKI, we have attempted to search for factors that influenced CCyR achievement. The baseline level of response and also prior responses and mutations were studied as prognostic factors for CCyR achievement. Among the four patient groups with different responses at baseline, the best results were achieved in the groups with any CyR at baseline. Thus, CCyR was reached in 23/31 (74%) patients with any level of CyR compared with 12/69 (17%) patients without any CyR at baseline (p = 0.0001). These results were compatible with the outcomes of the study of Ibrahim et al. (21), which showed that the rate of the 30-month cumulative incidence of CCyR in subgroups with and without any CyR was 71% and 0%, respectively (p = 0.0005). The significance of the depth of cytogenetic response before third-line therapy was also assessed in the study of Bosi et al. (22), where among the 13 patients who failed to respond to firstand second-line therapy, six with cytogenetic response before third-line therapy achieved a deeper cytogenetic response and higher overall survival on third-line TKI therapy.
We performed univariate analyses to identify pretreatment factors that predicted complete cytogenetic response. In the univariate analysis, the factors found to have a significant influence on the probability of achieving a complete cytogenetic response were age at third-line therapy, CyR on firstor second-line therapy, CyR before third-line therapy, and CHR before third-line therapy. The presence of kinase domain mutations prior to thirdline therapy did not affect the probability of achieving a complete cytogenetic response, as discussed previously. For the multiple regression analysis, we included also the time from the diagnosis to third-line therapy as a factor that reliably affects the achievement of a cytogenetic response according to the data of global clinical practice (23).
Of note, most patients with CCyR on third-line TKI reached it within the first 12 months of therapy. Thus, the cumulative incidence of CCyR was 66% and 86% by 6 and 12 months of treatment, respectively.
Prior cytogenetic response is the most robust positive prognostic factor identified in patients with CML receiving TKIs in third-line therapy. This fact has been confirmed by our research, in addition to other studies. Ibrahim et al. (21) reported that CCyR on firstor second-line therapy was the only factor that affected the outcomes of third-line therapy (21). Third-and fourth-line therapy with bosutinib was also successful in the subgroup with any cytogenetic response at baseline. CCyR was achieved in 94% (31/ 33) of patients with any CyR at baseline in contrast, and the probability to achieve CCyR in the group without any CyR at baseline was 25% (7/28) (24). Again, in the study of Russo Rossi et al. (25), patients who did achieve a CyR on imatinib or had low and intermediate Sokal risk had a higher probability of achieving CyR with third-line TKI therapy (p < 0.001).
Baseline factors also influenced PFS and OS. The 5-year OS on third-line therapy in our study was 68%. Ribeiro et al. (18) reported higher 5-year OS rates among patients with CP CML (86%) than among patients with AP/BP CML. In the study of Cortes et al. (26), the 4-year OS among patients treated with third-line bosutinib was 78%. We could speculate that some patients took a longer time to reach third-line TKI, had more clinically significant BCR::ABL gene mutations, especially with T315I, or had other prior unfavorable  Overall survival (OS) on third-line therapy.
factors compared with the patients in the aforementioned study, although there were not enough data for comparison. When comparing baseline factors, hematologic resistance was associated with the worst outcomes pertaining to PFS and OS, although we did not find any statistical differences between groups with any level of cytogenetic responses at third-line TKI initiation. Most of our patients who failed to respond to therapy were switched to thirdgeneration TKIs as their fourth or even fifth form of therapy within clinical trials; accordingly, we suspect that the high survival rate may be associated with subsequent therapies. Although we did not include these data in this analysis, many patients continued conservative TKI therapy without any CyR but were still in CP. We can speculate that patients with unfavorable biological changes, including genetic aberrations outside of BCR::ABL, progressed earlier while on firstor second-line TKIs and did not reach third-line treatment. Accordingly, a population-based study to evaluate the real unmet needs of patients on third-line therapy is warranted. Another obvious limitation of our study is its retrospective nature; accordingly, there were some gaps in data for some indicators and analysis. Moreover, the sample size was small; therefore, some of the analyses lacked statistical power to draw a clear conclusion. However, in our opinion, we have obtained interesting findings through the current retrospective research, which may allow us, along with the data of other authors, to define a therapeutic strategy in patients who responded to the first two generations of TKI therapy. In our observation, long-term treatment outcomes are not favorable as more than 80% of the patients discontinued third-line TKIs mainly due to resistance. These results can be improved by using ponatinib or asciminib. OS on third-line therapy according to the depth of response at baseline. OS on third-line therapy according to CCyR achievement on thirdline TKI. EFS on third-line therapy in the whole group. Long-term outcomes in the whole group on and after third-line TKI.

Conclusion
TKI has revolutionized CML management, and the availability of different TKIs provides patients with some alternatives after failure to respond to firstand second-line therapy (27). The data reviewed here suggest that independent prognostic factors may be helpful for the identification of patients who are unlikely to achieve CCyR response on TKIs after failure to respond to imatinib or a prior second-generation TKI. The cytogenetic response prior to third-line therapy, as well as some degree of cytogenetic response during firstor second-line therapy, was proven to be highly informative. Patients with these two criteria had a much higher probability of achieving CCyR on third-line therapy. Patients with CCyR and those without baseline hematologic resistance had better PFS and OS than the remaining patients. Patients with any level of CyR response on previous TKIs, especially with PCyR, are seemingly good candidates for continued conservative therapy, even with second-generation TKIs. These patients should be closely monitored for CyR and molecular responses. Patients without CCyR by 6 months or especially 12 months on third-line TKI, as well as patients with unfavorable baseline prognostic factors, especially those without CHR, should be switched to other available therapies, including allo-SCT, third-generation TKIs, or experimental drugs.

Data availability statement
The datasets presented in this article are not readily available because all patient data is anonymized. Requests to access the datasets should be directed to chitanava_tv@almazovcentre.ru.