This article is mentioned in:

Introduction

L-asparaginase (L-asp) is a fundamental drug in the systemic treatment of childhood B cell precursor acute lymphoblastic leukemia (BCP-ALL). The introduction of L-asp improved the 5-year event-free survival (EFS) by 20% and boosted treatment response rates by 60% (1,2). However, in genetic subtypes of ALL with a high risk of relapse, L-asp treatment did not meet expectations, with drug resistance being one of the major mechanisms contributing to treatment failure (2–6). Although L-asp resistance is multifaceted and still not fully explained, several potential mechanisms have been proposed, such as hydrolysis of asparaginyl bonds and enzymatic degradation of the drug particle itself, immunological response against the drug triggered by presentation of L-asp to dendritic cells through MHC class II complex and increased production of asparagine as a response to its depletion in the blast environment during L-asp treatment (7–9). Previous studies have suggested that the genes involved in the degradation of L-asp and subsequent minimization of its effectiveness, may be upregulated by primary genetic abnormalities and copy number variations (CNVs) identified in BCP-ALL, such as hypodiploidy, breakpoint cluster region and Abelson murine leukemia viral oncogene homolog 1 (BCR-ABL1) fusion, lack of ETS variant 6 and runt-related transcription factor 1 (ETV6-RUNX1) fusion, BCR-ABL1-like phenotype and ikaros family zinc finger protein 1 (IKZF1) deletion (1,10,11).

Of the genes identified to be involved in the mechanisms of resistance to L-asp treatment, three key examples are asparagine synthetase (ASNS), legumain (LGMN) and cathepsin B (CTSB). ASNS synthesizes asparagine to reverse the antineoplastic action of L-asp, which depletes the surrounding leukemic cells of asparagine (12,13). LGMN is a cysteine protease that inactivates drugs by specifically hydrolyzing asparaginyl bonds; it is also involved in the major histocompatibility complex (MHC) class II antigen presentation process, which may account for the allergic reactions to L-asp (14). CTSB encodes a peptidase that degrades L-asp (9). Although the aforementioned genes act in opposition to L-asp, no data on functional associations between them are currently available.

The role of the aforementioned genes in high-risk genetic subtypes of childhood BCP-ALL has not been fully elucidated. Studies of ASNS expression in groups with a number of genetic alterations are ambiguous (2,5,15). In addition, poor response to L-asp treatment is associated with ASNS overexpression at the protein, but not mRNA level (5,12,15,16). LGMN has been demonstrated to be upregulated in the intrachromosomal amplification of chromosome 21 (iAMP21) subtype of ALL, which is associated with high risk of relapse when treated with a low or standard risk protocol (14). No studies on the relevance of CTSB expression levels in BCP-ALL patients are currently available in the literature to the best of our knowledge, and a key aim of the present study was to determine the expression levels of the selected candidate genes underlying L-asp resistance in patients with different genetic types of BCP-ALL.

Materials and methods

Study group

A group of 52 patients ≤18 years were recruited retrospectively from three pediatric oncology centers in Poland: Lodz, Zabrze and Lublin. The patients were diagnosed with BCP-ALL between December 2005 and January 2016 and treated with the Berlin-Frankfurt-Münster backbone protocols with L-asp as a core regimen (ALL IC, 2002 and 2009) (17,18). Matched DNA and RNA samples from the bone marrow aspiration at the point of diagnosis, as well as clinical data, were acquired. BCR-ABL1 fusion, ETV6-RUNX1 fusion, hypodiploidy and hyperdiploidy were assessed routinely by fluorescence in situ hybridization and karyotyping performed in a certified laboratory (Genetic Laboratory at Collegium Medicum; Bydgoszcz, Poland) at the time of diagnosis. Each patient and/or their parents provided signed informed consent in writing prior to inclusion in the study.

Multiplex ligation-dependent probe amplification (MLPA)

For all collected bone marrow samples, MLPA analysis with P327-B1 iAMP21-ERG probe mix was performed according to the manufacturer's protocol (MRC-Holland BV). Additionally, MLPA P329 CRLF2-CSF2RA-IL3RA and P335 ALL-IKZF1 (MRC-Holland BV) probe mixes for cytokine receptor-like factor 1, IKZF1, IKZF2, IKZF3, cyclin-dependent kinase inhibitor (CDKN) 2A/2B, retinoblastoma 1 (RB1), PAX5 and ETV6 amplification or deletion status detection were used. Absolute fluorescence was normalized by comparing peak patterns of DNA in the sample of interest with a DNA isolated from blood samples from age- and gender-matched healthy individuals who provided written consent to participate in the study. The relative probe ratio of the tested samples was compared with the average relative probe ratio in the reference samples to calculate the dosage quotient. Data analysis and interpretation were conducted using GeneMarker v2.7.4 software (Softgenetics, LLC) according to the manufacturer's protocol.

Gene expression measurement

To obtain cDNA for quantitative PCR, all available RNA samples that were extracted using the TRIzol reagent (Thermo Fisher Scientific, Inc.) were reverse transcribed using a High-Capacity cDNA Reverse Transcription kit (Thermo Fisher Scientific, Inc.). The thermal conditions were 30 min at 50°C, 15 min at 95°C, followed by storage at 4°C. TaqMan Gene Expression assays (Applied Biosystems; Thermo Fisher Scientific, Inc.) for ASNS (cat. no. Hs04186194_m1), LGMN (cat. no. Hs00271599_m1) and CTSB (cat. no. Hs00947433_m1) were used; GAPDH (cat. no. Hs02786624_g1) was selected as a control. The thermocycling conditions were as follows: 50°C for 2 min and 95°C for 10 min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. All samples were tested in duplicate and the 2−ΔΔCq value was calculated (19) with GAPDH used for normalization. The relative mRNA expression was used for further comparisons.

Transcription factor binding sites search

The search for PAX5 binding sites throughout the human and mouse genome matrix was conducted using the HOCOMOCO v11, TRANSFAC database version 2018.3 (GeneXplain, GmbH). This is an online engine for searches through multiple module databases and was used as previously described (20).

Statistical analysis

Statistica 12.0 software (TIBCO Software, Inc.) was used for all analyses. Nominal variables are presented as percentages; differences between the groups were evaluated using the χ2 test. Continuous variables are presented as medians with interquartile range (IQR); differences between groups were evaluated using the Kruskal-Wallis test and post-hoc Tukey's HSD test or the Mann-Whitney U test for paired groups were used to establish groups with significant differences. Correlations between continuous variables were calculated with the Spearman's rank-correlation coefficient. Kaplan-Meier curves were used to present relapse-free survival (RFS) time, and hazard ratios (HRs) were calculated using Cox proportional hazards. P<0.05 was considered to indicate a statistically significant difference. For the graphical presentation of result, GraphPad Prism 7.05 software (GraphPad Software, Inc.) was used.

Results

Clinical and genetic characteristics

A total of 52 children with BCP-ALL were enrolled in the study with a median age at diagnosis of 6.54 years (IQR, 2.97–12.38 years). The group was predominantly male (60.40%). The median white blood cell (WBC) count at diagnosis was 1.40×104 cells/µl (IQR, 0.41–9.68×104 cells/µl) and the median leukemic cell percentage in bone marrow collected at diagnosis assessed by flow cytometry was 93.10% (IQR, 83.50–96.70%). The mean follow-up time was 4.43 years (range, 0.24–9.6 years).

The patient cohort was divided into groups based on the major genetic abnormalities detected in leukemic cells that have prognostic value within the used treatment protocols: The BCR-ABL1 fusion-positive group (n=5; 9.62% of all patients), hyperdiploid group (n=7; 13.46%), hypodiploid (n=2; 3.85%), ETV6-RUNX1 fusion (n=3; 5.77%) and other BCP-ALL group with no clear primary genetic aberrations (n=35; 67.31%) (Table I). The subgroups differed significantly in age at diagnosis, with the BCR-ABL1 fusion group being the oldest at diagnosis (median, 13.76 years; IQR, 6.76–13.90 years; P=0.04).

Table I. Clinical characteristics of the analyzed cohort of 52 patients with BCP-ALL.

Table I.

Clinical characteristics of the analyzed cohort of 52 patients with BCP-ALL.

Characteristic	BCR-ABL1, n=5	ETV6-RUNX1, n=3	Hyperdiploid, n=7	Hypodiploid, n=2	Other BCP-ALL, n=35	P-value
Male, %	60.00	66.67	71.43	50.00	57.14	0.89
Mean age at diagnosis, years	13.76	8.54	2.97	1.62	7.03	0.04
WBC, ×104 cells/µl	5.27	0.81	0.62	5.70	2.35	0.33
Blast count, %	88.00	83.25	96.80	94.00	94.50	0.64
Poor steroid response, %	40.00	33.33	0	50.00	10.71	0.23
MRD at day 15, %	0.40	11.34	0.71	29.45	2.23	0.51
Relapses, %	0	0	0	0	22.86	0.04
Deaths, %	0	0	0	0	20.59	0.07

[i] BCR-ABL1, breakpoint cluster region and Abelson murine leukemia viral oncogene homolog 1 fusion; ETV6-RUNX1, ETS variant 6 and runt-related transcription factor 1 fusion; BCP-ALL, B cell precursor acute lymphoblastic leukemia; WBC, white blood cell count at diagnosis; MRD, minimal residual disease.

The incidence of CNVs frequently observed in childhood ALL varied between the groups. All patients in the ETV6-RUNX1 fusion group carried IKZF1 and ETV6 deletions in the leukemic clone, and 2 out of the 3 patients displayed concomitant CDKN2A/2B deletion. The majority of patients with BCR-ABL1 fusion exhibited IKZF1 and PAX5 deletions (both were present in 3 out of 5 patients) and CDKN2A/2B, CRLF2, ETV6, RB1 and transcriptional regulator ERG (ERG) deletions were detected in 1 out of 7 patients each. In the patients from the hyperdiploid group, IKZF1 (3 out of 7), CRLF2 (2 out of 7), CDKN2A/2B (1 out of 7), PAX5 (1 out of 7) and Janus kinase 2 (JAK2; 1 out of 7) deletions were identified. The other BCP-ALL group was the most heterogenic among all groups; all patients exhibited CNVs in at least one of the tested genes with an incidence of 29 for IKZF1, 2 for IKZF2, 2 for IKZF3, 17 for CDKN2A/2B, 8 for PAX5, 4 for CRLF2, 4 for JAK2, 9 for ETV6, 4 for ERG and 6 for RB1 out of 35 patients in this group (Table SI).

Expression analysis

Expression levels of ASNS, LGMN and CTSB were measured in diagnostic bone marrow samples from the 52 patients. The median relative expression value for ASNS was 2.57 (IQR, 2.26–3.26), for LGMN was 1.48 (IQR, 1.21–2.14) and for CTSB was 1.14 (IQR, 0.79–1.77). The median relative mRNA expression of ASNS correlated significantly with LGMN and CTSB (r=0.59 and r=0.62, respectively; P<0.05, data not shown. No correlations were identified between ASNS, LGMN and CTSB expression levels and minimal residual disease (MRD) on day 15 of the treatment protocol or RFS time (data not shown). In addition, no relevant differences in the median relative mRNA expression levels of the evaluated genes between the distinct molecular groups of BCP-ALL were identified (Table SII; Fig. 1A-C).

Figure 1. ASNS, LGMN and CTSB expression levels in patients with BCP-ALL with various genetic aberrations. Expression of (A) ASNS, (B) LGMN and (C) CTSB in every group according to the primary genetic abnormality identified in the leukemic clone of BCP-ALL. Sample sizes: BCR-ABL1 fusion, N=5; hyperdiploidy, N=7; hypodiploidy, N=2; ETV6-RUNX1 fusion, N=3; other, N=35. Data were obtained by reverse transcription-quantitative PCR and are presented as relative mRNA expression. Horizontal line, median; whiskers, minimum to maximum; boxes, interquartile range. ASNS, asparagine synthetase; LGMN, legumain; CTSB, cathepsin B; BCP-ALL, B cell precursor acute lymphoblastic leukemia; BCR-ABL1, breakpoint cluster region and Abelson murine leukemia viral oncogene homolog 1 fusion; ETV6-RUNX1, ETS variant 6 and runt-related transcription factor 1 fusion.

Analysis of the gene expression profiles based on common CNVs revealed a significantly higher median LGMN relative expression level in patients with PAX5 deletion compared with patients with wild-type PAX5 in all BCP-ALL subgroups (median, 1.93 and 1.34, respectively; P=0.0282; Table SIII). Patients with PAX5 deletion also exhibited a significant correlation of CTSB median expression level with ASNS and LGMN (r=0.73 and r=0.64, respectively; P<0.05, data not shown). By contrast, deletion of IKZF1, IKZF2, IKZF3, CDKN2A/2B, CRLF2, JAK2, ETV6, RB1 or ERG did not exhibit predictive significance for ASNS, LGMN and CTSB median expression levels as a single factor (Table SIII).

Differences in ASNS, LGMN and CTSB expression levels in the other BCP-ALL group were identified when the patients were divided into subgroups based on PAX5 deletion status; 8/35 (24%) patients carried PAX5 deletions, whereas 27/35 (76%) did not. Patients with PAX5 deletions exhibited a higher relative expression level of ASNS compared with the wild-type PAX5 group (3.14 vs. 2.47; P=0.0117; Fig. 2A). Similar results were observed for LGMN expression levels, which were higher in patients with PAX5 deletion compared with patients with wild-type PAX5 (2.24 vs. 1.45; P=0.0029; Fig. 2A). In addition, median CTSB expression levels were higher in patients with PAX5-deletions compared with patients with wild-type PAX5 (2.08 vs. 1.04; P<0.0001; Fig. 2A). No significant associations between the median expression levels of the assessed genes were identified when the other BCP-ALL patient group was divided based on IKZF1 deletion status (82.86% of the group carried deletions; Fig. 2B). However, in cases with concomitant deletions of IKZF1 and PAX5 (14.29% of the group-5 patients carried both deletions), significant differences were observed in ASNS and CTSB expression levels (P=0.0334 and P=0.0358, respectively) but not in LGMN expression levels (P=0.0735) (Table SIV; Fig. 2C).

Figure 2. Expression of ASNS, LGMN and CTSB in patients carrying PAX5 and IKZF1 deletions. (A) Patients from the other BCP-ALL group with no identified primary genetic abnormality were separated into subgroups based on PAX5 deletion (N=8) or wild-type (N=27). Relative expression of ASNS (3.21 vs. 2.50), LGMN (2.32 vs. 1.43) and CTSB (1.91 vs. 1.03) in patients with PAX5 deletions was significantly higher compared with that in patients with wild-type PAX5. (B) Patients from the other BCP-ALL group were divided according to the IKZF1 deletion status (IKZF1 deletions, N=29; IKZF1 wild-type, N=6). No significant differences were observed in the relative expression levels of ASNS, LGMN and CTSB. (C) Relative expression levels of ASNS, LGMN and CTSB in patients from the other BCP-ALL group with concomitant IKZF1 and PAX5 deletions (N=5). Middle horizontal line, median value; whiskers, interquartile range. ASNS, asparagine synthetase; LGMN, legumain; CTSB, cathepsin B; PAX5, paired box 5; IKZF1, ikaros family zinc finger protein 1; wt, wild-type; Δ, deletion; BCP-ALL, B cell precursor acute lymphoblastic leukemia.

Of the 12 BCP-ALL patients with PAX5 deletion, four deletions were partial and heterozygous and eight affected the whole gene: Seven were heterozygous and one was homozygous. Additionally, TRANSFAC software (GeneXplain, GmbH) was used to test the PAX5 transcription factor binding sites (TFBS) for a possible association with ASNS, LGMN or CTSB. A search through human and mouse databases revealed that the three genes of interest were not directly regulated by PAX5.

Outcome description

In the analyzed cohort of 52 patients, eight relapsed. Median time to relapse was 2.27 years (IQR, 1.37–3.04 years). All relapses occurred in patients in the other BCP-ALL group. RFS did not differ significantly when patients in the other BCP-ALL group were divided according to PAX5 deletion status; median RFS was 5.14 in cases with PAX5 deletion and 5.10 in PAX5 wild-type cases (P=0.4540; HR, 1.56; 95% CI, 0.24–10.20). Out of the eight relapses, six (75%) occurred in patients with wild-type PAX5, whereas two (25%) occurred in patients with PAX5 deletions (P=0.0002). When the impact of PAX5 deletion on RFS was analyzed in the entire cohort, no significant differences were identified (median RFS for wild-type PAX5, 4.69 vs. 4.26 years in cases with PAX5 deletion; P=0.7084; HR, 1.50; 95% CI, 0.22–10.03).

The cohort was divided based on the expression levels of ASNS, LGMN and CTSB, with the cut-off value set as the median expression for each gene (ASNS, 2.57; LGMN, 1.48; CTSB, 1.14). When median RFS was analyzed in patients with high and low ASNS expression, high ASNS expression was associated with a longer time to relapse (P=0.0315; HR, 0.19; 95% CI, 0.04–0.86; Fig. 3A). The 5-year RFS rate of patients with high ASNS expression was 90.15% (95% CI, 87.90–92.40%), and the 5-year RFS rate of patients with low ASNS expression was 73.34% (95% CI, 69.40–77.30%) (P=0.0164). No significant differences in RFS between patients with high and low LGMN or CTSB expression were observed (P=0.2878; HR, 2.36; 95% CI, 0.54–10.37; and P=0.7714; HR, 1.25; 95% CI, 0.28–5.49, respectively; Fig. 3B and C). Patients with high LGMN expression exhibited a 5-year RFS rate of 88.46% (95% CI, 85.70–91.20%), and those with low LGMN expression exhibited a 5-year RFS rate of 77.54% (95% CI, 74.30–80.80%). Similarly, the 5-year RFS rate was 84.71% (95% CI, 81.70–87.70%) for patients with high expression of CTSB and 80.54% (95% CI, 77.40–83.70%) for those with low CTSB expression.

Figure 3. Kaplan-Meier curves for RFS based on ASNS, LGMN and CTSB expression levels. (A) For ASNS, the 5-year RFS rate was 90.15% (95% CI, 87.90–92.40%; N=24) in the high expression group and 73.34% (95% CI, 69.40–77.30%; N=25) in the low expression group. (B) For LGMN, the 5-year RFS rate in the high and low LGMN expression groups was 88.46% (95% CI, 85.70–91.20%; N=23) and 77.54% (95% CI, 74.30–80.80%; N=25), respectively. (C) In the high and low CTSB expression groups, 5-year RFS rate was 84.71% (95% CI, 81.70–87.70%; N=23) and 80.54% (95% CI, 77.40–83.70%; N=26), respectively. ASNS, asparagine synthetase; LGMN, legumain; CTSB, cathepsin B; RFS, relapse-free survival; HR, hazard ratio; CI, confidence interval.

Discussion

To evaluate the differences in ASNS, LGMN and CTSB gene expression in different types of BCP-ALL with possible additional effects of PAX5 and/or IKZF1 gene deletions, the patient cohort was grouped according to the major genetic abnormalities identified in leukemic cells such as BCR-ABL1 fusion, ETV6-RUNX1 fusion, hypodiploidy and hyperdiploidy (21). A large proportion of patients in the current study were classified into the other BCP-ALL group, in which none of the aforementioned aberrations were identified. However, considering the high incidence of IKZF1 and PAX5 deletions in these patients, it is likely that the group mostly comprised the BCR-ABL1-like type BCP-ALL (3). However, due to a lack of data on the gene expression profile, it is impossible to conclusively confirm this hypothesis, which is one of the limitations of the present study. Additionally, the small number of patients in the ETV6-RUNX1 fusion group substantially limited the comparisons between the remaining groups; therefore, the clinical characteristics of the group may not fully reflect the favorable profile described in previous reports (21,22).

The median expression of ASNS did not significantly differ between the groups; however, relative mRNA expression of ASNS was the highest in the ETV6-RUNX1 fusion group, which was consistent with previous observation (5). Previous studies have suggested the existence of an association between L-asp resistance and the upregulation of ASNS protein rather than mRNA (5,12,15,16); however, the data obtained in the present study were insufficient to verify this hypothesis, which was an additional limitation of this study. Despite this, longer RFS was observed in patients with high ASNS gene expression compared with those with low ASNS expression, which corresponded with the aforementioned findings (5).

The only previous study of LGMN gene expression in patients with BCP-ALL reported significantly higher expression in patients with iAMP21-positive BCP-ALL across all major genetic variants (14), which was associated with poor prognosis if not treated aggressively. Due to the lack of iAMP21-positive cases in the cohort analyzed in the present study, it was not possible to support those conclusions. No significant differences in LGMN expression were observed between the BCP-ALL subgroups; however, the lowest level of amplification was identified in patients with ETV6-RUNX1 fusion.

To the best of our knowledge, the results of the present study are the first to describe the CTSB expression profile in patients with childhood ALL. Despite the lack of statistical significance, the findings suggest that high CTSB expression may be associated with positive prognostic factors such as ETV6-RUNX1 fusion and hyperdiploidy. These findings contradict our initial hypothesis of the association between high ASNS, CTSB and LGMN expression levels and an inferior outcome, although it is in line with the high ASNS expression observed in cases with ETV6-RUNX1 fusion, as well as the longer time to relapse noted in a previous study (5).

The results of the present study also suggested an association between PAX5 deletions and increased ASNS, LGMN and CTSB expression in the other BCP-ALL patient group, which has not previously been reported. ASNS, CTSB and LGMN expression was significantly higher in patients with PAX5 deletion compared with those with concomitant IKZF1 deletion, which suggested that PAX5 may influence the expression of L-asp resistance genes as an independent factor. Although these comparisons were statistically significant, further experiments including in vitro PAX5 silencing, as well as larger study cohorts are needed to support these results.

Although the outcome of PAX5 deletion varies depending on its location in the front, middle or end of the gene (23), any PAX5 deletion, regardless of its location and extent of deleted region, results in the dysfunction of the PAX5 protein. A small number of this subgroup enables to statistically connect the region of PAX5 deletion with gene expression levels or outcome. Thus, patients with PAX5 CNVs were analyzed as a single cohort, irrespective of the type of PAX5 deletion. This was a limitation of the present study, and in vitro tests (such as gene silencing) are required for further confirmation.

TRANSFAC analysis did not provide any information on the direct influence of PAX5 TFBS on ASNS, LGMN or CTSB and could not explain the results of the present study. This suggested that PAX5 may have an indirect effect on ASNS, LGMN or CTSB expression, and additional functional studies are required for clarification.

RFS was used for outcome assessment, as it is superior to overall survival when investigating the influence of gene expression on resistance to treatment; hence, relapse was the most accurate end point. High expression of the genes considered to be responsible for L-asp resistance did not correspond with shorter RFS or decreased 5-year RFS in this cohort, which rejected the initial hypothesis of the present study. In addition, higher 5-year RFS was observed in patients with high ASNS expression, although the results for CTSB and LGMN were not significant. Despite the differences in the median RFS in the subgroup analysis not being significant, patients in the other BCP-ALL group carrying PAX5 deletions exhibited a longer time to relapse compared with those with wild-type PAX5.

The main limitation of the present study was a small number of patients and the lack of BCR-ABL1-like subtype confirmation, which resulted in challenges in achieving significant differences between variables. The present study is the first to report CTSB expression among different genetic subtypes of BCP-ALL, and the findings encourage further research. The results of the present study confirmed the association between PAX5 deletions and the gene expression levels of ASNS, LGMN and CTSB, as well as between ASNS expression and 5-year RFS. The latter result, however, requires further study in a larger cohort for a more thorough explanation.

Supplementary Material

Supporting Data

Acknowledgements

Not applicable.

Funding

This study was supported by the PRELUDIUM grant from the National Science Center (grant. no. 2014/13/N/NZ5/03660) and the National Center of Research and Development PersonALL project (grant. no. STRATEGMED3/304586/5/2017).

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Authors' contributions

EW, WM and AP designed the study. ML, JK, LS and TS obtained clinical data and bone marrow samples, and reviewed the manuscript. EW, JM and AP conducted the experiments. EW, JM, JJ and BP collected the data and interpreted the results. EW performed statistical analyses. EW and WM prepared a final manuscript for publication.

Ethics approval and consent to participate

All procedures performed in this retrospective study involving human participants were in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments, and were approved by the Bioethical Committee of the Medical University of Lodz (approval no. RNN/155/13/KE). Informed consent was obtained from all individual participants and/or their parents included in the study.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References