Systematic Review and Meta-Analysis of New-Generation Tyrosine Kinase Inhibitors versus Imatinib for Newly Diagnosed Chronic Myeloid Leukemia

Backgrounds: We performed this systematic review and me- ta-analysis to compare the efficacy of new-generation tyro- sine kinase inhibitors (NG-TKIs; including dasatinib, nilotinib, bosutinib, radotinib, and ponatinib) versus imatinib for pa- tients with newly diagnosed chronic myeloid leukemia (CML). Summary: We identified randomized controlled trials comparing the efficacy of NG-TKIs versus imatinib as the first-line treatment for CML patients by searching the PubMed, Cochrane library, and EMBASE databases. Two re- viewers independently extracted data and assessed study quality. A meta-analysis was performed to calculate risk ra- tios and 95% CIs using a fixed-effects model. Our study in- cluded 10 trials. Overall, treatment with NG-TKIs significant- ly improved the major molecular response and MR4.5 at all time points, and early molecular response at 3 months. Im- portantly, overall survival (OS) was significantly higher with the NG-TKIs at 12 months. Besides, NG-TKI-treated patients showed a significantly lower CML-related death and pro- gression to the accelerated phase/blast crisis. Key Messages: In first-line treatment, NG-TKIs are superior to imatinib re- garding OS at 12 months, and because molecular response rates were higher with the NG-TKIs at all time points, the NG- TKIs favor treatment-free remission. © 2019 S. Karger AG, Basel

Chronic myeloid leukemia (CML), a myeloprolifera- tive disorder [1], accounts for approximately 10% of all leukemias, with 8,430 new cases expected each year [2]. The clinical effect of several earlier treatments, such as hydroxyurea and interferon-α, were less than satisfactory. The development of tyrosine kinase inhibitors (TKIs) revolutionized the treatment of CML, and patients with CML who received TKI treatment now live almost as long as the general population [3].As the first TKI approved for first-line therapy of pa- tients with CML in the chronic phase (CP) [4, 5], imatinib proved an unprecedented efficacy on both hematological, Key Laboratory of Organ Regeneration and Transplantation of the Ministry of Education and Department of Epidemiology and Biostatistics, School of Public Health Jilin University, 1163 Xinmin Street, Changchun, Jilin 130021 (China)E-Mail jinln @; yaoyan @ cytogenetic, and molecular responses. A phase III trial (IRIS) showed that imatinib was associated with a higher level of progression-free survival (PFS) and long-term survival outcomes compared with interferon-α: 10-year overall survival (OS) was 83.3% [6–8]. However, 45% of patients in the IRIS trial randomized to imatinib discon- tinued therapy or crossed over due to disease progression, drug resistance, or adverse events [9], highlighting the need for new-generation (NG)-TKIs.Dasatinib and nilotinib were initially introduced as second-line therapy for patients with imatinib resistance or intolerance, and have recently been approved as first- line treatment for CML-CP by the US Food and Drug Administration (FDA) based on the results of two multi- center randomized trials [10, 11].

Except for the superior response rates, these TKIs demonstrated increasing effi- cacy toward certain types of mutations developed during imatinib therapy, and have been widely used as the first and second line of treatment for CML patients. Subse- quently, other second- and third-generation TKIs, in- cluding bosutinib, radotinib, and ponatinib, were also de- veloped. Randomized controlled trials (RCTs) were initi- ated to evaluate the response rates and long-term outcomes of these TKIs compared with imatinib as the first-line treatment in patients with CML-CP [12–14]. These clinical trials demonstrated the superiority of NG- TKIs and their potential to replace imatinib.Many factors should be taken into consideration when choosing the best treatment scheme for CML patients, such as treatment goals and comorbidity and side effect profiles. The application of TKIs has greatly changed the treatment goal of CML – from single prolongation of sur- vival to the achievement of cytogenetic and molecular re- sponses, the prevention of disease progression and treat- ment-free remission (TFR) [15]. In an age of multiple generations of TKIs, it is paramount we comprehensively understand the efficacy of NG-TKIs versus imatinib for newly diagnosed CML-CP patients. The meta-analyses by Gurion et al. [16] and Yun et al. [17] , which each in- cluded 8 trials, showed an advantage of NG-TKI versus imatinib in terms of molecular response, progression to accelerated phase (AP)/blast crisis (BC), cytogenetic re- sponse, and CML-related death, respectively, but no dif- ference in OS. Due to the recent publication of data in- volving NG-TKIs versus imatinib and the data obtained from long-term results of trials included in a previous meta-analysis [18–20], we aim to conduct an updated sys- tematic review and meta-analysis to assess the cytogenet- ic and molecular responses, disease progression, and sur- vival outcomes. We searched the PubMed, Cochrane library, and EMBASE da- tabases to identify published systematic reviews or meta-analyses comparing the efficacy of NG-TKIs with imatinib as the first-line treatment for newly diagnosed, previously untreated CML-CP pa- tients from the inception dates to April 30, 2019. The keywords used for literature research included CML, TKI, imatinib, nilo- tinib, dasatinib, bosutinib, ponatinib, and radotinib. Afterwards, we identified relevant original RCTs included in the systematic reviews or meta-analyses. An additional search was performed to identify recently published RCTs (within the last 5 years) meeting the inclusion criteria, using the databases and keywords described above.

Only English publications were considered. Besides, we searched the following conference proceedings from the past 3 years to identify any relevant studies: Annual Meeting of the American Society of Hematology, American Society of Clinical Oncology, and the Annual Meeting of the European Hematology Association. In addition, we searched databases of ongoing and unpublished trials ( to identify all completed clinical trials.Inclusion CriteriaWe included RCTs comparing the efficacy of any NG-TKIs versus imatinib. Trials must have enrolled adults older than 18 years with newly diagnosed CML-CP who had never been treated with any TKI. Trials were excluded if complete cytogenetic re- sponse (CCyR) and/or major molecular response (MMR) out- comes data were not reported.Data Extraction and Risk of Bias AssessmentThe following information was extracted from each study by 2 researchers: lead author, publication year, clinical trial registra- tion number, inclusion and exclusion criteria, sample size, par- ticipant characteristics, and doses of TKIs. Besides, we extracted unpublished data from forest plots of the meta-analysis after de- termining that trials met our inclusion criteria. Any disagreement was resolved by consensus after group discussion and a full-text review.Two researchers independently assessed the quality of the RCTs included in our study based on the Cochrane handbook, ver- sion 5.1.0 [21]. The following domains were evaluated: random sequence generation, allocation concealment, blinding of partici- pants and personnel, blinding of outcome assessment, incomplete outcome data reporting, and selective outcome reporting. Each item was graded as low risk, high risk, or unclear risk. The includ- ed trials were graded as low quality, high quality, or moderate qual- ity based on the criteria of Zhao et al. [22].

Additionally, funnel plots were applied to assess publication bias when the number of trials reporting the outcomes was 10 or more.The primary outcomes measured in the present study were MMR and CCyR rates at 12 months of TKI treatment (MMR, BCR-ABLIS ≤0.1%; CCyR, lack of Ph chromosome in ≥20 bone marrows). The secondary outcomes were as follows: MMR at 24 months and 3–5 years; CCyR at 24 months; MR4.5 (BCR-ABLIS≤0.0032%) at 12 and 24 months and 3–5 years; early molecular response (EMR; defined as BCR-ABLIS ≤10%) at 3 months (the definition was based on the European Leukemia Network; ELN [5]); the rate of patients progressing to AP/BC at 12, 24, and 36 months and 5 years; OS (death due to any cause while on treatment or during follow-up after the discontinuation of treatment) at 12, 24, and 36 months and 5 years; PFS (progression to AP/BC or death due to any cause while on treatment), and CML-related death at 12 and 24 months and 5 years.The efficacy of NG-TKI versus imatinib was assessed by calcu- lating risk ratios (RR) and 95% CIs using the Mantel-Haenszel sta- tistical method. The I2 statistic was applied to evaluate the hetero- geneity between summary data. A fixed-effects model was used to pool the data when heterogeneity was not present (I2 <50%), oth- erwise a random-effects model (REM) was used. We performed a sensitivity analysis to determine how robust the findings were by excluding low-quality studies or trials with characteristics differ- ent from the others. All meta-analyses were performed using Rev- man version 5.3 (Cochrane Collaboration). All tests were 2-tailed, and p < 0.05 was considered statistically significant. Results In total, 274 potentially eligible systematic reviews or meta-analyses were identified. The full texts of 25 records were read after screening the titles and abstracts of these records, and 11 met the inclusion criteria [16, 17, 23–31]. The 11 systematic reviews or meta-analyses included 9 RCTs that met the inclusion criteria. The searches for re- cently published RCTs yielded 310 records, and 34 full texts were reviewed. Finally, 8 trials met the inclusion cri- teria. After excluding duplicate trials, 10 RCTs were ulti- mately included in this meta-analysis (Fig. 1), 4 of dasat- inib [10, 32–38], 2 of bosutinib [12, 18, 39, 40], 2 of nilo-tinib [11, 19, 41–45], 1 of ponatinib [13], and 1 of radotinib[14, 20].A total of 4,272 patients were included in the meta- analysis. Among them, 2,317 had been treated with NG- TKIs (dasatinib, n = 810; bosutinib, n = 496; nilotinib, n = 697; ponatinib, n = 154; radotinib, n = 160) and 1,955 had been treated with imatinib. The median age of pa- tients in the included trials ranged from 41 to 58 years, and the percentage of male patients ranged from 32 to 68%. The baseline demographic characteristics in each study were well balanced between the experimental and control arms. A total of 3,360 patients had undergone risk assessment using the Sokal or Hasford risk score. The risk group stratification and the Eastern Coopera- tive Oncology Group performance status score were also well balanced between the treatment arms in each study. The characteristics of each trial are summarized in Table 1.All studies were randomized trials; 6 trials did not re- port methods of sequence generation and allocation con- cealment, therefore we judged them as unclear risk of se- lection bias [12, 18, 35, 37, 38, 45]. One trial described the methods used for allocation concealment without report- ing the sequence generation process [10]. Three trials re- ported methods both of sequence generation and alloca- tion concealment [11, 13, 14]. Thus, 3 trials were assessed as high quality, and the others were moderate quality. There was no evidence of publication bias (online suppl. Fig. 1; see for all online suppl. material).Major Molecular ResponseFig. 3. Forest plots of CCyR in patients with CML treated with NG-TKIs versus imatinib. The squares represent point estimates, their sizes reflect the weight in the pooled analysis, and the horizontal bars represent the 95% CI. Diamonds represent the pooled point estimate. Discussion A systematic review and meta-analysis was conducted using all available evidence from randomized trials. Our analysis comparing the efficacy of NG-TKIs versus ima- tinib for newly diagnosed CML-CP demonstrated that NG-TKIs were associated with higher rates of MMR and CMR at all time points, EMR at 3 months, and CCyR and OS at 12 months. Meanwhile, NG-TKIs showed lower progression to AP/BC and CML-related mortality. How- ever, adverse events leading to treatment discontinuation occurred more frequently with NG-TKIs when compared with imatinib.Response during therapy, the indicator for continu- ing or changing the treatment [46, 47], is also one of the most important predictors of long-term outcome [47– 49]. The CML Study IV found that patients who achieved MMR by 12 months had better PFS and OS at 3 years compared with those who did not [50]. The DA- SISION and ENESTnd clinical trials [34, 44] showed that patients with EMR were more likely to reach CCyR, MMR, and CMR by 5 years, and had significantly high- er PFS, OS, and lower rates of progression. The findings supported the notion of Marin et al. [51]. In our excite- ment, we found that NG-TKIs were associated with higher rates of molecular response, including EMR, MMR, and MR4.5, at all time points. These advantages give NG-TKIs a favorable position in clinical applica- tion.The OS of patients with CML treated with NG-TKI versus imatinib has been investigated in 2 other meta- analyses [16, 17]. The studies included 8 trials, respec- tively. Both of these studies, however, did not find dif- ferences in OS between the two groups at any time points, although patients treated with NG-TKI consis- tently had higher molecular and cytogenetic response rates and lower CML-related death. In contrast to the above studies, in addition to the advantages in EMR, MMR, MR4.5, and CML-related death, our meta-analy- sis found that patients who were treated with NG-TKIs had significantly higher OS than those treated with ima- tinib at 12 months after including two more studies [16, 18]. The survival advantage of NG-TKI makes it more possible to choose them as the first-line treatment for CML-CP patients. Unfortunately, the advantage in fa- vor of NG-TKI disappeared after 2 years. Since the max- imum follow-up period of clinical trials included in our study was 5 years, a longer period of follow-up or larger sample size may demonstrate a statistically significant difference.Although there were no differences in long-term OS, CML-related death was consistently lower in patients treated with NG-TKIs. The finding was consistent with Gurion et al. [16], who suggested that the mortality risks of CML-CP patients mainly come from general health conditions or coexisting conditions rather than CML. This notion has been confirmed by Saussele et al. [52], who found that comorbidities seem to have a greater im- pact on survival than CML itself. This also suggested that any measures conducive to the management of comor- bidities should be implemented without delay. As for PFS, NG-TKIs were superior to imatinib at 24 months, but there were no differences at any other time. However, considering the limited trials included in our study (only two trials included in the analysis of 24 months and 5 years), the results might be biased.Another important finding lies in CCyR. The achieve- ment of CCyR at 12 months confers better OS and PFS [33, 50]. We found that although CCyR was higher with NG-TKIs at 12 months, the statistically significant differ- ence was not observed at 24 months. However, cytoge- netic data at 24 months were available from four studies only, thus limiting the use of long-term analysis, which could be biased.An optimal TKI should have both effectiveness and safety. However, the greater efficacy of NG-TKIs was at the cost of more adverse events requiring treatment discontinuation. Other less threatening adverse events, such as the grade 3–4 gastrointestinal events, including diarrhea and nausea or vomiting, occurred more frequently in patients treated with NG-TKIs. This would bring more pain to patients. Besides, the rates of grade 3–4 aminotransferase elevation was higher in NG-TKI arms. This suggested that patients with a history of hepatic dysfunction should be cau- tious about NG-TKIs. Although patients with CML can live almost as long as the general population with lifelong TKI therapy, many patients consider discontinuing treatment due to adverse events or economic reasons [53]. TFR has be- come a treatment goal for many patients with CML, and clinical trials have confirmed the feasibility of TFR [54]. Analysis of these trials indicate that the longer durations of DMR and TKI therapy were associated with a higher probability of successful TFR [55], which means that the faster and deeper molecular response of NG-TKIs may make patients more likely to be eligible for TFR. In addi- tion, the ENESTfreedom observed that patients previ- ously treated with nilotinib showed a high rate of TFR, despite the duration of nilotinib therapy and MR4.5 be- ing shorter [56]. This suggests that, while NG-TKIs did not appear to significantly improve the rate of TFR, it shortened the therapy duration. This is probably the most important reason why NG-TKIs are considered as first-line treatment for patients with CML despite their higher toxicity and cost.Some limitations should be noted in the present study. First, long-term follow-up data available in our study were limited, such as OS, with only two trials pro- viding data for 5 years [34, 44]. Therefore, our results might be biased. Second, Ota et al. [57] reported that the response rates and OS were lower in patients with co- morbidities and older age. According to usual practice, the proportion of older people in the trials included in our study was low. Thus, our conclusions are difficult to apply to elderly patients. Lastly, some heterogeneity could not be avoided since NG-TKIs were considered as a category. Conclusion Our meta-analysis of RCTs of patients with CML-CP found that NG-TKIs were associated with superior OS at 12 months. Although there was no long-term survival ad- vantage, patients treated with NG-TKIs had a greater mo- lecular response at all time points. Furthermore, NG- TKIs were better at “softer” clinical outcomes, such as progression to AP/BC and Bosutinib CML-related mortality. Fur- ther large-scale clinical trials with longer follow-up peri- ods are required to assess long-term survival outcomes.