Children with chronic myeloid leukaemia treated with front-line imatinib have a slower molecular response and comparable survival compared with adults: a multicenter experience in Taiwan

Abstract

Background

The direct comparison of molecular responses of front-line imatinib (IM) monitored at the same laboratory between children and adults with chronic phase (CP) of chronic myeloid leukaemia (CML) had not been reported. In this multicenter study, we compared the landmark molecular responses and outcomes of paediatric and adult CML-CP cohorts treated with front-line IM in whom the BCR::ABL1 transcript levels were monitored at the same accredited laboratory in Taiwan.

Methods

Between June 2004 and July 2020, 55 newly diagnosed paediatric and 782 adult CML-CP patients, with molecular diagnosis and monitoring at the same reference laboratory in Taiwan, were enrolled. The criteria of 2020 European LeukemiaNet were applied to evaluate the molecular responses.

Results

By year 5, the cumulative incidences of IS <1%, MMR, MR^4.0 and MR^4.5 of paediatric patients were all significantly lower than those of adult patients (58 vs 75%, 48 vs 66%, 25 vs 44%, 16 vs 34%, respectively). The 10-year progression-free survival (PFS) (90%) and overall survival (OS) (94%) of paediatric patients did not differ from those (92%) of adult patients.

Conclusions

We demonstrated the paediatric cohort had slower molecular responses to front-line IM and similar outcomes in 10-year PFS and OS in real-world practice.

Introduction

Chronic myeloid leukemia (CML) is a clonal hematopoietic cell disorder driven by the constitutively activated BCR::ABL1 tyrosine kinase oncoprotein. Since the introduction of the tyrosine kinase inhibitor (TKI) in 2001, imatinib (IM) has been recommended as one of the first-line treatments for newly diagnosed CML patients in the chronic phase (CP) [1–3]. The molecular monitoring of BCR::ABL1 transcript level with real-time quantitative polymerase chain reaction (RT-qPCR) has become essential in patient care for assessing treatment response and guiding further therapy [4].

CML is a rare disease in children and constitutes 2–3% of paediatric leukemia in United States [5]. In Taiwan, 1.2–3.1% of CML occurred in patients younger than 15 years between 2014 and 2018 [6]. Most paediatric oncologists have followed the studies or practice guidelines that originally designed for adult patients [7–10]. Hijiya et al. found that children, adolescents and young adults tended to have more aggressive clinical presentations than older adults and required a different approach [8]. Limited multicenter studies with varying number of cases reported the molecular responses to front-line IM in children with CML [11–14]. The only international study CML-PAED-II [14], conducted in nine western countries, showed that front-line IM at the recommended dose resulted in both excellent response and tolerable side effects in 140 newly diagnosed paediatric CML-CP but with a short median follow-up time of 25 months. The study focusing on the comparison of molecular responses between paediatric and adult CML is extremely rare [15, 16]. The GIMEMA CML Working Party analysed 2784 patients ≥18 years of age and found that young adults (18–29 years) had a lower complete cytogenetic response (CCyR) and major molecular response (MMR) rates compared with older adults [17]. Millot et al. [18] reported 37% of children treated with IM failed to achieve BCR::ABL1 transcript levels ≤10% at 3 months, which was higher than that of a previous adult study (24%) [19]. In this long-term follow-up multicenter study, we aimed to retrospectively compare the molecular responses and treatment outcomes between paediatric and adult CML-CP patients treated with front-line IM in whom the BCR::ABL1 transcript levels were monitored at the same single reference laboratory in Taiwan.

Methods

Patients and treatment

In this study, we retrospectively enrolled 55 newly diagnosed paediatric (<18 years) CML-CP patients with molecular diagnosis and monitoring at the Chang Gung Memorial Hospital (CGMH)-Linko laboratory between June 2004 and July 2020. And 782 adult CML-CP patients, newly diagnosed between June 2004 and November 2018 and monitored at the CGMH laboratory, were also included. All these patients took IM as the front-line TKI therapy for more than 3 months. Written informed consent was obtained from the patients, parents, or guardians. The presence of BCR::ABL1 transcripts and subtype of fusion transcripts were detected by RT-PCR from peripheral blood (PB). Patients with e14a2 and/or e13a2 transcripts were enrolled and those wth rare BCR::ABL1 fusion transcripts, including e14a3, e13a3, e19a2 and e1a2 were excluded.

Monitoring of molecular response

Following IM therapy, BCR::ABL1 levels of PB were measured by TaqMan RT-qPCR assay every 3 months according to treatment guideline at the GGMH laboratory. The RT-qPCR assays were developed in CGMH laboratory and used to monitoring molecular response in real-world clinical practice since 2004. The leftover mononuclear cells were all freshly frozen in RPMI with 10% dimethylsulfoxide and 10% foetal bovine serum at −80 °C or in liquid nitrogen in the cell bank. The CGMH laboratory obtained the lab-specific conversion factor for International Scale (IS) from the reference laboratory at Adelaide, Australia [20], then BCR::ABL1 levels were expressed as IS, and we also retrospectively converted all the data examined before 2006 to IS. The criteria of optimal response based on the 2020 European LeukemiaNet (ELN) were adopted: IS <10% at 3 months, IS <1% at 6 months and IS <0.1% (MMR) at 12 months [3]. The deep molecular response was also defined: MR^4.0, detectable disease with <0.01% and >0.0032% BCR::ABL1 IS or undetectable disease with >10,000 and <32,000 ABL1 transcripts; MR^4.5, undetectable or detectable disease with <0.0032% BCR::ABL1 IS and >32,000 ABL1 transcripts in the same volume of cDNA used to test for BCR::ABL1 [2]. MR^4.0 and MR^4.5 were validated through a proficiency test of College of American Pathologists and materials provided by the Secondary Reference Panel (Lyo panel) Study II [21]. No bone marrow samples were examined for monitoring of molecular response.

Prognostic scores

In adult patients, three prognostic scores: Sokal score [22], the European Treatment and Outcome Study for CML (EUTOS) score [23] and EUTOS Long-Term Survival (ELTS) score [24] were applied for risk classification at diagnosis. The Sokal score was generated from four prognostic parameters of age, spleen size, platelet count and circulating blasts at diagnosis. The EUTOS score was developed by two parameters of the percentage of basophils and spleen size. The ELTS score was calculated by the parameters including age, spleen size, blasts and platelets in PB.

Statistical analysis

Differences between categorical variables were examined with Fisher’s exact or χ² test. The duration of overall survival (OS) was defined as the time from diagnosis to the date of death. The duration of progression-free survival (PFS) was defined as the time from diagnosis to the date of accelerated phase (AP), blast crisis (BC) or death. The Kaplan–Meier method supplemented with 95% confidence intervals obtained from Greenwood’s estimate of standard error was used to estimate PFS and OS rates with differences compared by the log-rank test. The causes of death unrelated to CML were considered as competing event outcomes. The molecular response of a patient was censored if the patient received hematopoietic stem cell transplantation (HSCT), switched to 2nd-generation (2G) TKIs, or died. Patients still on IM treatment without achieving a response criterion were censored at the end of the follow-up period. Gray’s test was used for estimation and comparison of cumulative incidences of IS <1%, MMR, MR^4.0 and MR^4.5, under consideration of competing risk factors, defined by receiving HSCT, switching to 2G TKIs, or death. All reported P values were based on two-sided tests and considered statistically significant if they were less than 0.05. Data were analysed with IBM^®SPSS^® Statistics version 20 (IBM, Armonk, New York) and R4.1.2 software package.

Results

Clinical features

The clinical features of paediatric and adult cohorts are summarised in Table 1. Of the 55 newly diagnosed paediatric CML-CP patients, there were 14 patients aged <10 years. The median WBC count was 303.7 × 10⁹/L (range, 24.5–806.5) and <250 × 10⁹/L in 20 patients; 250–500 ×  10⁹/L in 30 and ≥500 × 10⁹/L in 5.

Table 1.

Summary of clinical features, molecular responses and outcomes of CML-CP patients treated with front-line imatinib.

Features	Paediatric cohort	Adult cohort
No. of patients	55	782
Median age (range), years	13.5 (1.4–17.9)	47.5 (18–88)
Female/male	23 (42%)/32 (58%)	339 (43%)/443 (57%)
BCR::ABL1 transcript type
e14a2/e13a2/e14a2 + e13a2	34 (62%)/20 (36%)/1 (2%)	499 (64%)/253 (32%)/30 (4%)
Sokal score, low/Int/high	NA	227 (40%)/186 (33%)/152 (27%)*
Median follow-up time, months	70.2 (3.0–166.0)	59.2 (3.4–168.0)
Median time to MMR, months	16.5 (3.3–59.1)	12.0 (2.3–79.6)
Median time to MR4.0, months	43.6 (6.2–68.0)	22.0 (3.7–96.3)
Median time to MR4.5, months	54.2 (6.2–103.0)	34.2 (3.7–92.0)
AP/BC progression (%)	1 (2%)/2 (4%)	19 (2%)/29 (4%)
Death (%)	1 (2%)	35 (4%)

AP accelerated phase, BC blast crisis, CP chronic phase, Int intermediate, MR molecular response, MR^4.0 International Scale (IS) <0.01%, MR^4.5 IS <0.0032%, NA not available.

*Patient who did not have complete data were not included in the analysis.

In the paediatric cohort, the median dosage of IM was 340 mg/m²/day (range, 205–340). Six patients underwent HSCT with a median of IM treatment of 8.7 months (range, 5–24). Twenty-four (43.6%) patients switched to 2G TKIs (18 dasatinib and 6 nilotinib) with a median time of IM therapy of 28.9 months (range, 3–141), including 3 achieved optimal responses because of IM intolerance (renal dysfunction, severe headache and severe myalgia for each one patient) (Fig. 1). In the adult cohort, the dosage of IM ranged from 200 to 600 mg daily (median, 400) with a mean of 395 mg. Among 262 (33.5%) adult patients switching to 2G TKIs, 35 had intolerable adverse effects of IM though their optimal responses were achieved. The rates of switching to 2G TKI did not significantly differ between paediatric and adult cohort (43.6% vs 33.5%, P = 0.142).

Fig. 1. Flow diagram of paediatric CML-CP cohort.

2G 2nd-generation, AP accelerated phase, BC blast crisis, CP chronic phase, HSCT hematopoietic stem cell transplantation.

Molecular responses and outcomes

The median BCR::ABL1 levels of paediatric and adult cohorts at different time points after IM therapy are shown in Fig. 2. Children with CML exhibited a similar pattern of molecular responses but with a slower slope of decline. In paediatric CML, the cumulative incidences of IS <1% by 6 months and 12 months were 16.8% (95% confidence interval [CI] 10.7–22.9%) and 40.7% (95% CI, 31.5–49.9%), respectively; of MMR by 12 months and 18 months were 18.5% (95% CI, 10.2–26.8%) and 25.9% (95% CI, 16.3–35.5%). In the adult group, the cumulative incidences of IS <1% by 6 months and 12 months were 30.0% (95% CI, 26.7–33.3%) and 54.2% (95% CI, 50.6–57.8%), respectively; of MMR by 12 months and 18 months were 27.2% (95% CI, 24.1–30.3%) and 38.2% (95% CI, 34.7–41.7%), respectively. By year 5, the cumulative incidences of IS <1%, MMR, MR^4.0 and MR^4.5 of paediatric CML were all significantly lower than those of adult CML (Fig. 3a–d). The estimated PFS and OS at 10 years did not show statistical significance between paediatric and adult patients (Fig. 4a, b).

Fig. 2. Median BCR::ABL1 levels (IS) in paediatric and adult CML-CP treated with front-line imatinib.

The ranges of IS in the paediatric cohort are also shown. *The median IS could not be reliable because the number of paediatric patients is too small. CP chronic phase, IS International Scale.

Fig. 3. Cumulative incidences of landmarks of molecular response to front-line imatinib in paediatric and adult CML-CP.

a IS <1%; b MMR; c MR^4.0; d MR^4.5. CP chronic phase, MMR major molecular response, MR^4.0 International Scale (IS) <0.01%, MR^4.5 IS <0.0032%.

Fig. 4. Outcomes of paediatric and adult CML-CP treated with front-line imatinib.

a Progression-free survival; b overall survival. CP chronic phase, OS overall survival, PFS progression-free survival.

For 21 children and 237 adults switched to 2G TKI due to failure or waring response of IM, the molecular response at the time of 2G TKI switch and the best molecular responses after 2G TKI therapy were shown in Supplemental Table 1. Compared to lower molecular responses to front-line IM in the paediatric cohort, there was no significant difference in the cumulative incidences of MMR, MR^4.0 or MR^4.5 between 21 paediatric and 237 adult CML patients after switching to 2G TKI (Supplemental Fig. 1).

Impacts of early molecular response and prognostic scoring systems

Adult patients with early molecular response, i.e., IS ≤10% at 3 months after starting IM correlated well with higher IS <1% (P < 0.0001) and MMR (P < 0.0001) by 12 months and better PFS (P < 0.0001) and OS (P = 0.0002) at 10 years (Supplemental Table 2 and Supplemental Fig. 2A, B). Paediatric patients with early molecular response also had higher rates of IS <0.1% (P = 0.002) and MMR (P = 0.008) by 12 months compared with those of patients with IS >10% (Supplemental Table 2). In contrast to the adult cohort, the PFS (P = 0.350) and OS (P = 0.176) of the paediatric cohort at 10 years did not have significant differences (Supplemental Fig. 2C, D).

Prognostic scoring systems could be applied in 565 (72%) adult patients who had complete diagnostic parameters for risk classification. The Sokal risk group could predict BCR::ABL1 IS level <1% at 3 months (P = 0.036), the rates of MMR (P < 0.001), MR^4.0 (P = 0.002) and MR^4.5 (P = 0.027) in the adult cohort (Supplemental Table 3). There was no significant difference in OS or PFS between Sokal risk groups in the adult cohort (Supplemental Fig. 3).

Discussion

The clinical features, molecular responses and survivals of our adult cohort were comparable to those of our previous report [25] and international adult studies [26–28]. The following discussion will mainly focus on the results of the paediatric cohort and the comparison of molecular response between the two cohorts.

The published paediatric CML studies addressing the molecular responses of front-line IM therapy in CML-CP have been increasing recently [11–14, 29–35]. However, so far only four studies reported on national or international multicenter cohorts comprising ≥40 paediatric patients [14] (Supplemental Table 4), of which all studies were from western countries. The median follow-up time ranged from 23 to 52 months (range, 1–146). Our paediatric cohort had the longest median follow-up time of 70 months (range, 3–166).

The reported rates of CCyR or IS <1% by 6 months and 12 months varied considerably from 26 to 93% and 61–96%, respectively; of MMR by 12 months and 18 months, 31–67% and 55–82%, respectively [11–14]. Our paediatric patients had a lower rate of IS <1% (41% by 12 months) or MMR (26% by 18 months), compared with the patients of western countries [11–14], whereas the molecular responses of our adult CML patients were comparable with those of published international studies [28]. Of note, the comparison between different trials might be hampered by differences in reporting results either as a cumulative response analysis (“by” a given timeframe) or as the efficacy of treatment (“at” specific time points) [14].

This study retrospectively recruited the patients who had serial MRD monitoring at our laboratory. The clinical information was provided and treatment discretion were made by the treating physicians. We did not have data of drug adherence in the retrospective study. Our paediatric patients received a higher median IM dosage of 340 mg/m² (range, 205–340), which had been reported tolerable in children and adolescents [12, 36]. However, it was noted that the dosage of paediatric patients was capped to 400 mg daily which resulted in the actual dosage <260 mg/m² in six adolescents. Optimal response was achieved in none of these six patients in comparison with 25 of 49 patients with dosages ≥260 mg/m² (P = 0.027). Three of the six patients switched to dasatinib and the other three underwent HSCT. Both PFS and OS did not differ significantly in these two groups. In real-world practice, drug incompliance and lower dosage could partly explain the lower molecular response rate and higher rate (43.6%) of switching to 2G TKIs in our paediatric cohort compared with the international Phase III trial of CML-PAED-II [14] (27.1%) (P = 0.026). The current recommendation of Children’s Oncology Group (COG) CML Working Group was 340 mg/m² of IM with a maximum of 600 mg daily [9], which could minimise the number of patients treated with lower dosages.

ASXL1 is among the most frequently mutated genes in CML, however, the clinical relevance of mutated ASXL1 at the time of CML diagnosis is unclear [37]. We performed ASXL1 mutational analysis on 51 paediatric CML patients by next-generation sequencing (data not shown). We found that four (7.8%) had ASXL1 mutations at diagnosis which was lower than 6 of 21 (29%) patients of the paediatric study by Ernst et al. [38], but comparable to the adult patients in the study of Roche–Lestienne et al. (8.8%) [39] and the very recent report of Schonfeld et al. (9.0%) [40]. Of note, three of the four ASXL1-mutated patients in our study failed to achieve IS <1% at 12 months but this proportion was not different to the paediatric patients without ASXL1 mutations (21/47; P = 0.331). Similarly, there was no difference in the achievement of MMR at 18 months (P = 0.625). Furthermore, PFS and OS of these four patients did not differ significantly with those of non-mutated patients, in line with the paediatric study of Ernst [38]. Thus ASXL1 mutations do not explain the lower molecular response to front-line IM in our paediatric cohort.

Very few studies specifically investigated the difference in molecular responses between adult and paediatric/adolescent CML patients treated with first-line IM [15, 16]. In a single-institution study, Cortes et al. reported an unfavourable trend in outcome for adolescent and young adult (AYA, 15–29 years) CML treated with first-line TKI [15]. The rates of CCyR, MMR and MR^4.5 were significantly lower in AYA (N = 61) compared to older patients (N = 407), but PFS and OS were similar for the two groups. However, the detailed outcomes of 35 AYA patients treated with front-line IM in that study were not available. To the best of our knowledge, only one German study quantitatively compared the median response in 25 paediatric (age 1–18 years) and 55 adult CML patients on IM therapy [16]. Proschmann et al. found no difference in the median response kinetics between paediatric and adult cohorts during the time interval of 36 months [16]. In the current Taiwanese multicenter study, all molecular response landmarks of paediatric and adult patients were measured at the same accredited laboratory using IS, and the optimal molecular responses at different landmarks were defined following ELN 2020 criteria [3]. The cumulative incidences of IS <1% by 6 months and 12 months; of MMR by 12 months and 18 months of the paediatric cohort were lower than those of the adult cohort. Our paediatric cohort achieved MR^4.0 and MR^4.5 with longer median times of 44 and 54 months, respectively, as compared with the adult cohort. In contrast to the German study [16], the ratio of median BCR::ABL1 IS levels of our paediatric cohort to adult cohort was mostly between 2 and 4 at different time points after 6 months of IM therapy (Fig. 2). Our data indicated that Taiwanese children with CML did have slower molecular responses to front-line IM compared to adult patients. But, the PFS and OS of our paediatric cohort did not differ from those of the adult cohort and the published paediatric multicenter studies [10–14, 29–35].

Early molecular response defined by a BCR::ABL1 transcript level ≤10% at 3 months after commencing IM therapy had been reported to predict better overall survival in adults treated with front-line IM [19]. Consistent with this finding in adult CML, Millot et al. showed that paediatric CML with early molecular response correlated well with better CCyR and MMR at 12 months and better PFS both at 36 and 48 months [18]. The results of our adult cohort were in line with those of the published adult study [19], whereas in our paediatric cohort, early molecular response significantly predicted higher rate of IS <1% and MMR by 12 months, but not of PFS and OS.

In the pre-TKI era, a meta-analysis of large-scale cohorts of patients with CML was conducted to develop reliable prognostic scoring systems for predicting survival. The Sokal score could predict molecular response, risk of progression to AP or BC, and OS in adult patients treated with IM, and thus remained a useful prognostic marker in the TKI era in some studies [41–44]. The EUTOS score was initially developed to identify high-risk and low-risk groups with significantly different probabilities of CCyR after 18 months IM therapy and 5-year PFS in adult patients [23]. The ELTS score was introduced in 2016 to distinguish three risk groups with significantly different probabilities of disease-specific death of adults with CML [24]. The ELTS score was later proved to have more predictive value than Sokal score and the EUTOS score for predicting survival in CML [45]. However, the results of correlation of prognostic score systems with outcomes in the paediatric cohort were conflicting. Gurrea Salas et al. showed a high discordance when Sokal, Hasford [46] and EUTOS scores were applied on paediatric patients at diagnosis [47]. The International Registry for Chronic Myeloid Leukemia (I-CML-Peds study) showed the ELTS score could have better discrimination of PFS than previous scoring systems in 309 children with CML [48]. But it was not supported by a recent Hong Kong study of 36 patients [49]. The current recommendation of COG did not use the Sokal, Hasford, and EUTOS scores for risk assessment or taking treatment decisions for children with CML. And more data are required to confirm the application of ELTS for children and adolescents with CML [9].

In conclusion, multicenter studies reporting molecular responses to front-line IM in paediatric CML-CP were limited, and all were conducted in western countries. Further, a comparison of serial molecular response landmarks between paediatric and adult CML-CP patients monitored at the same laboratory is not yet reported. In this Taiwanese multicenter study of CML-CP patients, the molecular responses were presented according to the recent ELN recommendation and expressed as IS. Our paediatric patients were followed up for the longest time among the published multicenter studies. Compared with the adult cohort monitored at the same accredited laboratory, we demonstrated that the paediatric cohort had slower molecular responses to front-line IM and similar outcomes in 10-year PFS and OS in real-world practice, incomplete IM compliance and lower dosage could have the impact in the teenagers.

Author contributions

H-CL and M-CK: provision of materials, data curation and writing of the manuscript. K-HW, T-YC, J-SC, M-CW, T-LL, YSY, M-CM, P-NW, J-MS, S-CW, S-HC, T-HJ, C-NC and T-CY: provision of materials and approval of the manuscript. T-HL: data analysis and approval of the manuscript. L-YS: concept design, methodology, funding acquisition, supervision, data curation and writing of the manuscript.

Funding

This work was supported by grants from Chang Gung Memorial Hospital, Taiwan (CMRPG350071 and XMRPG1A0081).

Data availability

The datasets generated and/or analysed during this study are available from the corresponding author on reasonable request.

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

Written informed consent was obtained from the patients, parents, or guardians. Chang Gung Medical Foundation Institutional Review Board approved the study and the committee’s reference numbers are 96-0358B and 100-0927B. This study was performed in accordance with the Declaration of Helsinki.

Consent to publish

Not applicable.

Supplementary information

The online version contains supplementary material available at 10.1038/s41416-023-02162-9.