A reciprocal translocation between chromosomes 9 and 22 results in the fusion gene BCR-ABL1, which is the genetic hallmark of chronic myeloid leukemia (CML) (1, 2). The breakpoints of BCR gene cluster occur primarily within a 5.8-kb region known as the major breakpoint cluster region (M-BCR) that spans exons e12-16 (historically named b1-5); the breakpoints in the ABL1 gene are similarly variable (3). Of note, e13 or e14 is more prone to fuse with ABL exon 2 (a2), giving rise to the e13a2 or the e14a2 transcripts. According to statistics, more than 90% of CML patients carry either the e14a2 or the e13a2 transcript alone. The co-expression of both transcripts (e14a2 and e13a2) can also be found in approximately 5-10% of patients. Both transcripts are translated into constitutively active proteins of 210 kDa which serve as targets for tyrosine kinase inhibitors (TKIs). The life expectancy of CML patients who may once have died within 7 years of diagnosis in the pre-TKI era is now more likely close to that of general population (4, 5). However, nearly 40% of CML patients treated with TKIs fail to achieve an optimal response throughout 5-year treatment period, or later relapse (6, 7). One possible hypothesis for the causes of resistance to TKIs could be due to the different protein tyrosine kinases (i.e., e13a2 and e14a2) that differ from one another by 75 base pairs. This structural difference may be related to the rates of transcription and translation, and the affinity of protein tyrosine kinases to TKIs, which may therefore affect the response to TKI treatment (8). If confirmed, transcript type could be used to guide clinical decisions in treating CML, especially at a time when treatment-free remission (TFR) is becoming the ultimate goal of therapy.

So far, the impact of BCR-ABL transcript type on outcome in CML patients has been investigated in few studies but was inconclusive in the TKI era. In three studies, no significant difference in major molecular response (MMR) was found between different transcripts (9–11), whereas seven studies found that superior MMR was observed in patients with e14a2 transcripts (12–18). Six studies reported that e14a2 was a better predictor of deep molecular response (DMR) (10, 12, 13, 15, 16, 18). However, Mulas et al. (19) and Marce et al. (11) showed that the transcript types did not affect DMR. In addition to molecular response, survival was also a major subject of debate. Event-free survival (EFS) was demonstrated to be the same in two studies (16, 18), but to be significantly better in e14a2 patients in four studies (13, 15, 20, 21). Progression-free survival (PFS) was demonstrated to be the same in three studies (11, 12, 19), but to be significantly better in e14a2 patients in two studies (13, 15). Overall survival (OS) was demonstrated to be the same in seven studies (12, 13, 16, 19–22), but to be significantly different in two (11, 15). To our knowledge, no meta-analysis has been conducted to summarize the conflicting evidence.

Given the inconsistency of the above findings, the aim of this meta-analysis is to evaluate whether the impact on response and survival in TKI-treated patients with CML varies by different transcript types (e13a2 vs e14a2 vs both).

This meta-analysis was performed according to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) (23). The research protocol was registered and approved in PROSPERO (CRD42021288440).

To our knowledge, this is the first meta-analysis aiming to evaluate the impact of BCR-ABL transcript type on outcome in TKI-treated patients with CML. To fully leverage available data and increase confidence in our results, several analytical approaches were used including traditional meta-analysis and network meta-analysis. Compared with e13a2 transcripts, there was a statistically significant advantage for patients with e14a2 (alone or with co-expressed e13a2) in terms of MMR and DMR at 6, 12 and 18 months. This benefit was sustained up to 5 years for patients with e14a2 transcripts, but not for patients with both transcripts. The expression of e14a2 also improved probability of EFS and OS throughout treatment period. Importantly, having e14a2 transcripts were associated with a higher rate of TFR in CML patients attempting TKI discontinuation. By network meta-analysis, results were similar, confirming the robustness of the results. Rank analyses showed that the e14a2 transcript had the highest probability to be the most favorable transcript type for the majority of outcomes, followed by both and e13a2.

As evidence continues to emerge from TFR studies, discontinuation of TKI therapy is feasible in the most of patients with sufficient TKI response. However, the eligibility criteria for patients maintaining optimal TKI discontinuation is rather strict (i.e., a minimum of 5 years of TKI therapy and (> 3 years of sustained MR⁴ or > 2 years of sustained MR^4.5)) (25). Our meta-analysis indicated that the DMR at 60 months was still higher in the e14a2 arm (OR 2.21, 95% CI: 1.71-2.87), suggesting 121% more patients expressing e14a2 transcripts qualified for entering TFR phase compared to those expressing e13a2 transcripts. So far, well-designed studies to directly investigate the correlation between different transcripts and the maintenance of TFR have less frequently been performed. Claudiani et al. (32) and Shanmuganathan et al. (31) found that probability of TFR was higher for patients expressing e14a2 transcripts, whereas D’Adda et al. (10) and Marce et al. (11) demonstrated that transcript types did not affect TFR rates. Our exploratory meta-analysis showed a 217% increase in the rate of TFR maintenance in e14a2 patients attempting TKI discontinuation. In addition, previous studies have demonstrated several factors contributing to successful TFR with duration of DMR appearing to be the most important predictor (33–36). Thus, taken together, it is suggested that e14a2 transcripts could serve as a strong parameter allowing for successful TFR. Of note, due to the lack of power stemming from the limited number of studies included and the small sample size, the result concerning correlation between the transcript type and TFR maintenance rate may not be robust (only 156 and 97 patients in the e14a2 and the e13a2 arms, respectively, were pooled). Taking important values of TFR in treatment of CML into consideration, further studies in larger patient cohorts are required to demonstrate this correlation.

Complete cytogenetic response (CCyR) results were not calculated, since they are insufficiently sensitive to monitor response. On the other hand, CCyR is also not optimal in our study comparing differences between typical BCR-ABL1 transcripts. The up-dated 2020 ELN guidelines recommend the BCR-ABL1 transcript levels at specific time points as the monitoring milestones for treating CML and achieving MMR from 12 months onward is regarded as the optimal response (25). As far as MMR rates are concerned, our results showed superiority which is sustained over 60 months in the e14a2 arm compared to the e13a2 arm. Indeed, patients who maintain MMR throughout long-term follow-up are not likely to progress, but rather show good clinical results. So far, e14a2 has not yet been evaluated and included in prognostic systems, i.e., Sokal, Euro, EUTOS and ELTS (37–40). For high-and intermediate-risk patients stratified by current risk scores, the physicians are more likely to use new-generation TKIs for treating them in case of treatment failure or in order to achieve TFR. However, since the increased number of CML-unrelated deaths occurred in CML patients who are still in remission, the physicians making clinical decisions also consider the patient’s characteristics, comorbidity and the distinct toxic effect profile of the different TKIs. For example, pleuro-pulmonary disease and arteriovascular disease are strong contraindications to dasatinib and nilotinib, respectively (41–43). In many situations, the benefit versus risk is difficult to balance. Our results provided a possibility that patients with e14a2 transcripts could reduce risk stratification to a certain extent, thereby receiving effective imatinib therapy with relatively less toxicity. In addition, Jain et al. (13) even found that the rates of MMR for patients with e14a2 transcripts treated with imatinib were similar to that of patients treated with second-generation TKIs. Thus, to obtain more accurate benefit-risk profile, we strongly suggest incorporating the e14a2 transcript into the prognostic system and giving it appropriate weight.

Although EFS and OS in each study were no statistical difference, the pooled results resulted significant (Figure 6). There may be several reasons for this contrary observation. First, statistical heterogeneity was mild (I² = 0.00%) and the point estimates (rectangles) from almost all studies were located on the left of the vertical line. Second, life expectancy of patients with typical BCR-ABL1 transcripts treated with all TKIs is close to that of the general population. Also, progression to AP/BC and CML-related mortality are rarely encountered after 12 months of TKI therapy (44). If differences in long-term survival outcomes exist, less statistical power owing to fewer events would need a large patient sample to discover such differences. Our meta-analysis increased the sample size and reduced the widths of CI, therefore providing statistically significant results.

In our meta-analysis, the advantage of the e13a2 and e14a2 co-expression was demonstrated within 18 months of treatment in terms of both MMR and DMR. Even so, no significant difference was found in evaluated outcomes at 60 months as compared to e13a2, neither for molecular response nor for survival. This phenomenon is reasonable because e13a2 cells are more persistent and the prognosis of patients with e13a2 is worse (45). Thus, we suppose that patients with co-expression of both transcripts after experiencing long-term TKI treatment would probably require more careful molecular monitoring or derive more benefit from switching to new-generation TKIs.

Our meta-analysis has several limitations: 1) The definition of DMR were not uniform such that DMR was defined as MR⁴ in some studies and as MR^4.5 in others, potentially reducing precision. 2) As with any meta-analysis, our study lacked individual data that might have provided additional details such as sustained DMR which was more critical for patients attempting TKI discontinuation. 3) Studies were pooled with different characteristics of the patients and designs such as TKI type, median follow-up and amplification efficiency. Despite this limitation, heterogeneity was mild for the primary outcomes across these studies; we also minimized the influence of heterogeneity by using a random-effects model, especially for the secondary outcomes with severe heterogeneity. Additionally, subgroup analyses were performed according to characteristics of the patients and designs, and these provided concordant results. 4) For some outcomes, the number of studies included was limited, which could increase uncertainty of the results. 5) Our meta-analysis might have been limited by the retrospective nature of the included studies.

In conclusion, the expression of e14a2 is related to a faster, deeper and more sustained molecular response compared to e13a2. This superiority in response translates in improved long-term EFS and OS. From 12 months onward, having the e14a2 transcript ranked first to achieve all outcomes. Importantly, despite relatively small numbers, the expression of e14a2 may have a positive impact on TFR. Our meta-analysis shows that the e14a2 transcript can be added to the list of prognostic factors to guide clinical decisions in treating CML, especially at a time when TFR is becoming the ultimate goal of therapy.

The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding authors.

KC, FC, and QC contributed to the conception and design this study. WH and JL carried out the development of the methodology. NX, KT, and PX analyzed and interpreted the data. KC and YR wrote the manuscript and approved the final submission of the study. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.