NOTCH1 mutations are associated with favourable long-term prognosis in paediatric T-cell acute lymphoblastic leukaemia: a retrospective study of patients treated on BCH-2003 and CCLG-2008 protocol in China

Authors

  • Chao Gao,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Shu-Guang Liu,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Rui-Dong Zhang,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Wei-Jing Li,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Xiao-Xi Zhao,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Lei Cui,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Min-Yuan Wu,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Hu-Yong Zheng,

    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    Search for more papers by this author
  • Zhi-Gang Li

    Corresponding author
    1. Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, Beijing, China
    • Correspondence: Dr Zhi-Gang Li, Beijing Key Laboratory of Paediatric Haematology Oncology, Key Laboratory of Major Diseases in Children, Ministry of Education, National Key Discipline of Paediatrics, Haematology and Oncology Centre, Beijing Children's Hospital, Capital Medical University, 56 Nanlishi Road, Beijing 100045, China.

      E-mail:ericlzg70@hotmail.com

    Search for more papers by this author

  • Zhi-Gang Li, Min-Yuan Wu and Hu-Yong Zheng share the senior authorship of this article.

Summary

Activating mutations of NOTCH1 are a common occurrence in T-cell acute lymphoblastic leukaemia (T-ALL), but its impact on T-ALL treatment is still controversial. In this study, the incidence, clinical features, and prognosis of 92 Chinese children with T-ALL treated using the Beijing Children's Hospital-2003 and Chinese Childhood Leukaemia Group-2008 protocols were analysed. NOTCH1 mutations were found in 42% of T-ALL patients and were not associated with clinical features, prednisone response, and minimal residual disease (MRD) at day 33 and 78. However, proline, glutamate, serine, threonine (PEST)/transactivation domain (TAD) mutations were associated with younger age (15/16 mutant vs. 48/76 wild-type, = 0·018) and more central nervous system involvement (4/16 mutant vs. 3/76 wild-type, = 0·016); while heterodimerization domain (HD) mutations were associated with KMT2A-MLLT1 (MLL-ENL; 4/30 mutant vs. 1/62 wild-type, = 0·037). Furthermore, prognosis was better in patients with NOTCH1 mutations than in those with wild-type NOTCH1 (5-year event-free survival [EFS] 92·0 ± 4·5% vs. 64·0 ± 7·1%; = 0·003). Long-term outcome was better in patients carrying HD mutations than in patients with wild-type HD (5-year EFS 89·7 ± 5·6% vs. 69·3 ± 6·2%; = 0·034). NOTCH1 mutations and MRD at day 78 were independent prognostic factors. These findings indicate that NOTCH1 mutation predicts a favourable outcome in Chinese paediatric patients with T-ALL on the BCH-2003 and CCLG-2008 protocols, and may be considered a prognostic stratification factor.

T cell acute lymphoblastic leukaemia (T-ALL) accounts for approximately 15% of newly diagnosed childhood ALL (Pui et al, 2011). The use of intensive chemotherapy has improved the long-term event-free survival (EFS) of nearly 70% of T-ALL patients, which is a little less than the percentage of B-cell precursor ALL (BCP-ALL) patients benefiting from intensive treatment (Gao et al, 2012). In BCP-ALL, prognostically significant molecular risk factors of therapy failure, such as BCR-ABL1 fusion and KMT2A (MLL) rearrangements, have been known for more than 10 years. However, few such factors are known in T-ALL. Biomolecular markers of outcome prediction are needed to improve T-ALL patient stratification.

NOTCH1 is a single transmembrane receptor and controls haematopoietic cell proliferation, maturation, adhesion and apoptosis (Demarest et al, 2008). Regulated NOTCH1 signalling is important for normal T cell development, and its activation is necessary for leukaemogenesis even in the presence of T-ALL-associated genetic abnormalities (Weerkamp et al, 2006; Demarest et al, 2011). Activating mutations of NOTCH1 are found in more than 50% of paediatric patients with T-ALL and cluster in the heterodimerization domain (HD), leading to ligand-independent cleavage and constant release of intracellular domain (ICN), as well as the C-terminal PEST [proline (P), glutamate (E), serine (S), threonine (T)] domain resulting in a prolonged half-life of ICN (Jenkinson et al, 2013).

Many recent clinical studies have investigated the prognostic importance of NOTCH1 mutations in paediatric T-ALL patients. Notably, long-term outcome in these patients depends largely on the therapeutic protocol applied. The prognosis of patients was favourable with the Berlin-Frankfürt-Münster ALL-BFM 2000 protocol (Kox et al, 2010), but unfavourable with the Dutch Childhood Oncology Group (DCOG) protocol (Zuurbier et al, 2010). As the prognostic significance of NOTCH1 mutations in paediatric T-ALL is still controversial and their significance in Chinese paediatric patients has not been reported, we investigated the incidence of NOTCH1 mutations in a large number of paediatric patients treated with the Beijing Children's Hospital (BCH)-2003 and Chinese Childhood Leukaemia Group (CCLG)-2008 protocols and analysed their correlation with clinical features and prognosis.

Materials and methods

Patients and treatment protocols

This study included 92 paediatric patients [aged 1–16 years (median, 7·5)] newly diagnosed with T-lineage ALL treated at the Haematology and Oncology Centre of Beijing Children's Hospital, Capital Medical University, between February 2003 and July 2012. The BCH-2003 protocol was used in 45 patients between January 2003 and March 2008 and the CCLG-2008 protocol was used in 47 patients after April 2008. Details of the stratification and treatment regimens of the two protocols have been published previously (Gao et al, 2012) and are outlined in Table 1. All patients in this study were included in the Intermediate-risk Group at initial diagnosis. Patients who had poor prednisone response (defined as the presence of >1·0 × 109/l peripheral blasts on day 8), incomplete morphological remission (defined as the presence of >5% bone marrow blasts at day 33), or high level of minimal residual disease (MRD, defined as >1% blasts at day 33 or >0·1% blasts at day 78) (Flohr et al, 2008; Cui et al, 2010; Schrappe et al, 2011) were then re-classified into the High-risk Group and received more intensive chemotherapy. The BCH-2003 and CCLG-2008 protocols were approved by the Beijing Children's Hospital Institutional Ethics Committee. Informed consents were signed by the parents or guardians of each patient.

Table 1. BCH-2003 and CCLG-2008 treatment protocol.
 Remission inductionEarly intensificationConsolidationDelay intensification IMaintenance IDelay intensification IIMaintenance II
  1. BCH, Beijing Children's Hospital; CCLG, Chinese Childhood Leukaemia Group; VDLP, vincristine + daunorubicin +L-asparaginase + prednisone; CAM, cyclophosphamide + cytarabine + 6-mercaptopurine (6-MP); HD-MTX, high-dose methotrexate; VDLD, vincristine + daunorubicin + L-asparaginase + dexamethasone; VD, vincristine + dexamethasone; IT, intrathecal injection with dexamethasone and methotrexate; VDLA, vincristine + cytarabine + L-asparaginase + dexamethasone; VM26, teniposide; HD-Ara-C, high dose cytarabine; CA, cyclophosphamide + cytarabine; TIT, intrathecal injection with dexamethasone, methotrexate and cytarabine; I′, Berlin-Frankfürt-Münster (BFM) High Risk block-1′; II′, BFM High Risk block-2′; III′, BFM High Risk block-3′.

BCH-2003
Standard-riskVDLP (DNR×2)CAMHD-MTX3 g/m2 × 4VDLD+CAM6-MP+MTX/VD+ITHD-MTX6-MP+MTX/VD+IT
Intermediate-riskVDLP (DNR×4)CAM×2HD-MTX5 g/m2 × 4VDLD+CAM6-MP+MTX/VD+ITVDLA+VM26+HD-Ara-C6-MP+MTX/VD+IT
High-riskVDLP (DNR×4)I′ II′ III′I′ II′ III′VDLD+CAM//6-MP+MTX/VD+IT
CCLG-2008
Standard-riskVDLD (DNR×2)CAMHD-MTX2 g/m2 × 4VDLD+CAM//6-MP+MTX/VD+IT
Intermediate-riskVDLD (DNR×4)CAM×2HD-MTX5 g/m2 × 4VDLD+CAM6-MP+MTXVDLD+CAM6-MP+MTX/VD+TIT
High-riskVDLD (DNR×4)CAM×2I′ II′ III′×2VDLD+CAM//6-MP+MTX/CA/VD+TIT

Detection of NOTCH1 mutations

Total RNA was extracted from bone marrow mononuclear cells at diagnosis and reversed transcribed into cDNA, as previously described (Gao et al, 2012). Polymerase chain reaction (PCR) was performed in a 50-μl reaction mixture containing 1 μl of cDNA equivalent to 160 ng of RNA, 0·4 μl of Pyrococcus furiosus (Pfu) DNA polymerase (5 μ/μl, kindly provided by Prof. Shen-Tao Li, Capital Medical University, Beijing, China), 0·1 μmol of each dNTP, 0·02 μmol of forward and reverse primers, and 25 μl of 2 × GC buffer I (Takara Biotechnology, Dalian, China) on a GeneAmp PCR system 2700 (Applied Biosystems, Foster City, CA, USA). Exons 26 (N-terminal end of the HD domain), 27 (C-terminal end of the HD domain), and 34 [transactivation domain (TAD) and the PEST domain] of NOTCH1 were amplified. The primers and conditions used for amplification are presented in Table 2. The PCR products were sent to Beijing Genomics Institute-Beijing Technology & Service Co. Ltd. (Beijing, China) for sequencing in both directions using an ABI PRISM 3730 automated sequencer (Applied Biosystems). To further confirm the insertion or deletion mutations, we cloned the PCR product into the pEASY™-T5 Zero cloning vector (TransGen, Beijing, China), and used it for transformation into Trans5a chemically competent cells (TransGen) followed by culture at 37°C for 15 h on agar gel with ampicillin. Then individual colonies were selected and sequenced. The sequences were analysed using BLAST software (http://blast.ncbi.nlm.nih.gov/Blast.cgi) to identify mutations in the NOTCH1 gene (the single-nucleotide polymorphism, C5094T, was not included).

Table 2. Primers and conditions used for amplification of NOTCH1.Thumbnail image of

Statistical analysis

The reference date for the end of data collection for statistical analysis purposes was 1 February 2014. Comparisons between the patients with or without NOTCH1 mutations and associations of pre-treatment characteristics and response to treatment were evaluated by non-parametric tests. Relapse was defined as the reappearance of leukaemic cells in bone marrow (>5% blasts) and/or the reappearance of clinical evidence of the disease. EFS was defined from the date of diagnosis to the date of relapse, death, or induction failure, whichever came first, or the last contact with patients in continuous haematological complete remission. EFS distribution of the patients with or without NOTCH1 mutation was estimated with the Kaplan–Meier procedure; comparisons between groups were performed using the log-rank test. The Cox proportional hazards regression model was used to evaluate the significance of differences in survival among patients with different types of NOTCH1 mutations and other clinical indicators. Two-sided P values <0·05 were considered statistically significant. SPSS 16.0 software (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.

Results

Incidence and characteristics of NOTCH1 mutations

In general, 47 mutations in the NOTCH1 gene were detected in 39 (42·4%, 39/92) of the 92 patients analysed. There were 30 mutations (63·8%, 30/47) in the HD domain,including 23 (76·7%, 23/30) point mutations resulting in missense amino acid conversions and 7 (23·3%, 7/30) in-frame insertions or deletions. There were 15 mutations (31·9%, 15/47) in the PEST domain, including 7 (46·7%, 7/15) missense, 7 (46·7%, 7/15) frame-shift and one (6·7%, 1/15) nonsense mutation that caused truncation of the protein. We detected one missense and one nonsense mutation in the TAD and two synonymous mutations in the PEST domain. See Table SI for the details of NOTCH1 mutations.

Of the 39 patients with mutated NOTCH1, 31 had a single mutation (79·5%, 31/39), including 23 (74·2%, 23/31) and 8 (25·8%, 8/31) patients with mutations in the HD or TAD/PEST domain, respectively. The remaining eight patients (20·5%, 8/39) had two mutations, one in the HD and one in the PEST domain in five cases, one in the HD and one in the TAD in two cases, and both in the PEST domain in one case (Table SII).

Association of clinical characteristics, early treatment response and NOTCH1 mutations status

There was no difference in white blood cell (WBC) count, age, gender, central nervous system (CNS) involvement at diagnosis and risk factors used for stratification between the patients with or without NOTCH1 mutations. Neither STIL-TAL1 nor KMT2A-MLLT1 fusion was associated with NOTCH1 mutation status. There was no between-subgroup difference in early treatment responses, such as prednisone response, and MRD at the end of remission induction (day 33) and before consolidation (day 78). The clinical characteristics and early treatment response of the T-ALL patients subgrouped by NOTCH1 mutation status are presented in Table 3.

Table 3. Clinical characteristics and early treatment response of the T-ALL patients according to NOTCH1 mutation status.
SubgroupTotalNOTCH1 genotypeHD genotypePEST/TAD genotypeHD+PEST/TAD genotype
MUT (%)WT (%) P MUT (%)WT (%) P MUT (%)WT (%) P MUTBoth (%)MUTSingle or WT (%) P
  1. Bold values denote statistical significance.

  2. MUT, mutated; WT, wild type; EFS, event-free survival.

Patient Number92 (100)39 (42)53 (58)30 (33)62 (67)16 (17)76 (83)7 (8)85 (92)
White blood cell count
≥50 × 109/l57 (62)26 (67)31 (58)0·51621 (70)36 (58)0·36010 (63)47 (62)0·9995 (71)52 (61)0·705
<50 × 109/l35 (38)13 (33)22 (42)9 (30)26 (42)6 (37)29 (38) 2 (29)33 (39)
Age (years)
1–1063 (68)28 (72)35 (66)0·65219 (63)44 (71)0·48115 (94)48 (63) 0·018 6 (86)57 (67)0·426
≥1029 (32)11 (28)18 (34)11 (37)18 (29)1 (6)28 (37)1 (14)28 (33)
Gender
Male71 (77)32 (82)39 (74)0·45225 (83)46 (74)0·14712 (75)59 (78)0·7555 (71)66 (78)0·657
Female21 (23)7 (18)14 (26)5 (17)16 (26)4 (25)17 (22)2 (29)19 (22)
Central nervous system (CNS)
CNS1/285 (92)34 (87)51 (96)0·13027 (90)58 (94)0·67912 (75)73 (96) 0·016 5 (71)80 (94)0·087
CNS37 (8)5 (13)2 (4)3 (10)4 (6)4 (25)3 (4)2 (29)5 (6)
Risk Group
Intermediate59 (64)24 (62)35 (66)0·66718 (60)41 (66)0·6459 (56)50 (66)0·5693 (43)56 (66)0·245
High33 (36)15 (38)18 (34)12 (40)21 (34)7 (44)26 (34)4 (57)29 (34)
Genetic Subtype
STIL-TAL1 POS 21 (23)9 (23)12 (23)0·9995 (17)16 (26)0·4304 (25)17 (22)0·7550 (0)21 (25)0·345
STIL-TAL1 NEG 71 (77)30 (77)41 (77)25 (83)46 (74)12 (75)59 (78)7 (100)64 (75)
KMT2A-MLLT1POS5 (5)4 (10)1 (2)0·1594 (13)1 (2) 0·037 0 (0)5 (7)0·5830 (0)5 (6)0·999
KMT2A-MLLT1NEG87 (95)35 (90)52 (98)26 (87)61 (98)16 (100)71 (93)7 (100)80 (94)
Prednisone Response
Good59 (66)25 (66)34 (67)0·99919 (66)40 (67)0·99910 (67)49 (66)0·9994 (67)55 (66)0·999
Poor30 (34)13 (34)17 (33)10 (34)20 (33)5 (33)25 (34)2 (33)28 (34)
Unavailable312121212
Minimal residual disease at day 33
<10−413 (19)7 (24)6 (15)0·5566 (25)7 (15)0·3371 (11)12 (20)0·9990 (0)13 (20)0·466
10−4–10−242 (60)17 (59)25 (61)15 (63)27 (59)6 (67)36 (59)4 (100)38 (58)
≥10−215 (21)5 (17)10 (24)3 (12)12 (26)2 (22)13 (21)0 (0)15 (22)
Unavailable221012616715319
Minimal residual disease at day 78
<10−355 (89)23 (85)32 (91)0·68920 (87)35 (90)0·9997 (88)48 (89)0·9994 (100)51 (88)0·999
≥10−37 (11)4 (15)3 (9)3 (13)4 (10)1 (12)6 (11)0 (0)7 (12)
Unavailable301218723822327
Outcomes
5-Year EFS9292·0 ± 4·5%64·0 ± 7·1% 0·003 89·7 ± 5·6%69·3 ± 6·2% 0·034 86·7 ± 8·8%74·1 ± 5·3%0·32771·4 ± 17·1%76·8 ± 4·8%0·711

To further investigate the clinical characteristics of patients with specific mutant locations of NOTCH1, patients were divided into those with and without (i) HD involvement, (ii) PEST/TAD involvement and (iii) simultaneous HD and PEST/TAD involvement. The PEST/TAD mutations were associated with younger age (<10 years old, 15/16 mutant vs. 48/76 wild-type, = 0·018) and more CNS involvement (CNS3 disease, 4/16 mutant vs. 3/76 wild-type, = 0·016). In addition, KMT2A-MLLT1 fusion was associated with HD mutation. Four of 30 (13·3%) patients with HD mutations were KMT2A-MLLT1 positive, whereas only 1 of 62 (1·6%, 1/62) patients with wild-type HD domains carried this KMT2A rearrangement (= 0·037). Other clinical characteristics and early treatment response were irrelevant (see Table 3).

Impact of NOTCH1 mutation status on treatment outcome

The 5-year EFS rate of the 92 patients was 76·5 ± 4·6%, with a median follow-up of 70 months. Clinical outcome was more favourable in patients with NOTCH1 mutations than those without mutations (5-year EFS 92·0 ± 4·5% vs. 64·0 ± 7·1%; = 0·003, Table 3 and Fig 1A). As patient outcome is highly dependent on the treatment received, and two protocols, BCH-2003 and CCLG-2008, were used in our study, we also analysed the prognostic significance of NOTCH1 mutations in patients treated with each of these two protocols. T-ALL patients with NOTCH1 mutations had an excellent outcomes when treated with the BCH-2003 protocol (5-year EFS rate 94·7 ± 5·1% vs. 60·0 ± 9·8%; = 0·011, Fig 1B). We also found a trend of favourable outcome in patients treated with the CCLG-2008 protocol (5-year EFS rate 86·9 ± 8·7% vs. 64·8 ± 13·1%; = 0·095, Fig 1C). One explanation for this finding may be that follow-up time was shorter (i.e., <3 years) in more than one-third (36·2%, 17/47) of patients treated with the CCLG-2008 protocol. With respect to the relationship between prognosis and distinctive mutant domains of NOTCH1, prognosis was better in patients with HD domain mutations than in those with wild-type HD (5-year EFS 89·7 ± 5·6% vs. 69·3 ± 6·2%; = 0·034, Table 3, Fig 1D). In addition, no difference in prognosis was found between different PEST/TAD or HD+PEST/TAD genotype subgroups (Table 3).

Figure 1.

Event-free survival (EFS) of paediatric T-ALL patients according to NOTCH1 mutation status. (A) EFS of all patients stratified by NOTCH1 mutation status. (B) EFS of patients treated using the BCH-2003 protocol stratified by NOTCH1 mutation status. (C) EFS of patients treated using the CCLG-2008 protocol stratified by NOTCH1 mutation status. (D) EFS of patients stratified by NOTCH1 heterodimerization domain (HD) mutant status.

Cox proportional hazards regression analysis identified the presence of NOTCH1 mutations and MRD at day 78 as independent prognostic factors [odds ratios of 0·113 (95% confidence interval (CI), 0·014–0·897), 4·214 (95% CI, 1·185–14·983); = 0·039 and 0·026, respectively]. Other clinical features at the time of initial diagnosis, such as WBC count, gender, age and CNS leukaemia, as well as other genetic markers, such as STIL-TAL1 and KMT2A-MLLT1 were not independent prognostic factors. The independent prognostic significance of responses to prednisone treatment and MRD at day 33 was not validated.

Discussion

T-ALL is a heterogeneous malignancy, and risk-adapted therapeutic strategies have been introduced into the treatment of paediatric T-ALL, which rely on measures of therapeutic response, such as MRD. However, no genetic markers of prognosis have been applied for stratification. Some studies have suggested that NOTCH1 mutations may be prognostic factors (Breit et al, 2006; Asnafi et al, 2009; Baldus et al, 2009; Park et al, 2009; Kox et al, 2010; Ben Abdelali et al, 2011; Jenkinson et al, 2013; Trinquand et al, 2013). In this study, NOTCH1 mutations were found in 42·4% of paediatric T-ALL patients, which is lower than the incidence reported for some Western countries (>50%) (Breit et al, 2006; Mansour et al, 2006, 2009; van Grotel et al, 2008; Asnafi et al, 2009; Baldus et al, 2009; Clappier et al, 2010; Kox et al, 2010; Zuurbier et al, 2010; Ben Abdelali et al, 2011; Jenkinson et al, 2013; Trinquand et al, 2013). However, our results are similar to the mutation incidences reported in Shanghai, South China (37·7%), and Japan (40·0%) (Zhu et al, 2006; Park et al, 2009). The reason may be partly the racial difference between Asians and Caucasians. Furthermore, most patients (79·5%, 31/39) carried a single mutation and only eight patients carried two mutations simultaneously. These mutations occurred mainly in the HD domain, followed by the PEST domain, which is in line with previous reports (Breit et al, 2006; Mansour et al, 2006, 2009; Zhu et al, 2006; Malyukova et al, 2007; Thompson et al, 2007; van Grotel et al, 2008; Asnafi et al, 2009; Baldus et al, 2009; Larson Gedman et al, 2009; Park et al, 2009; Clappier et al, 2010; Erbilgin et al, 2010; Kox et al, 2010; Zuurbier et al, 2010; Ben Abdelali et al, 2011; Mansur et al, 2012; Jenkinson et al, 2013; Trinquand et al, 2013). Mutations in these two domains of the NOTCH1 receptor were documented to associate with up-regulation of NOTCH1-dependent signal transduction. The mutations in the HD domain were of the missense point type or short in-frame indel type and possibly increased susceptibility to ligand-independent cleavage and subsequently increased release of ICN (Sanchez-Irizarry et al, 2004). Mutations in the PEST domain were missense or nonsense mutations, as well as frame-shift deletions and insertions, probably leading to the increased half-life of ICN (Weng et al, 2004).

Although we found no significant association between presence of NOTCH1 mutation and clinical presentation in our study, which is consistent with previous findings, we discovered that patients with PEST/TAD mutations were more likely to develop CNS involvement. Furthermore, 15 of our 16 patients with PEST/TAD mutation were younger than 10 years old. Similarly, about 70% of the patients with KMT2A-rearranged leukaemia were infants (Hilden et al, 2006), and ETV6-RUNX1 (TEL-AML1) fusion was more frequent in preschoolers with BCP-ALL (Gao et al, 2012). These phenomena might reflect the harbouring of PEST/TAD mutations by a distinct subgroup of T-ALL with a distinct mechanism of leukaemogenesis.

The presence of NOTCH1 mutations was correlated with a favourable MRD status in the ALL-BFM 2000 and Medical Research Council (MRC) UKALL 2003 trials (Kox et al, 2010; Jenkinson et al, 2013), but NOTCH1 mutations were not correlated with good response to early treatment in our study. This finding may be partly explained by the placement of all T-ALL patients on initial diagnosis in an Intermediate-risk group to receive more intensive treatment, thereby concealing the good impact of NOTCH1 mutations on early treatment responses.

The impact of NOTCH1 mutations on long-term survival in paediatric T-ALL is still controversial among different treatment groups. Trials of the BFM-2000, MRC UKALL 2003 and Japan Association of Childhood Leukemia Study ALL-97 protocols, but not others, found that NOTCH1 mutations predicted good prognosis (Park et al, 2009; Kox et al, 2010; Jenkinson et al, 2013). Moreover, the European Organization for Research and Treatment of Cancer study reported more CNS relapses and a poor prognosis in high-risk patients with NOTCH1 mutations (Clappier et al, 2010). In our study, the presence of NOTCH1 mutations was significantly correlated with favourable long-term outcome, perhaps because two BFM-like protocols were used. Furthermore, the EFS of patients with mutations involving the HD domain was excellent. These results suggest the possibility that NOTCH1 is a useful genetic marker for risk stratification in T-ALL. The patients with NOTCH1 mutations might be treated with a less intense, lower toxicity regimen, such as the Standard-risk protocol (Gao et al, 2012). In addition, MRD at day 78 was also validated as an independent predictor of outcome, which is in line with the results in a large cohort of patients included in the BFM 2000 study (Schrappe et al, 2011).

Inactivating mutations of FBXW7, reported in about 20% of T-ALL patients (Mansour et al, 2009; Kox et al, 2010; Zuurbier et al, 2010; Jenkinson et al, 2013), result in the inability of FBXW7 to bind the NOTCH1 PEST domain to perform proteolytic cleavage function, thereby increasing the half-life of the ICN. Furthermore, reduction of FBXW7 activity could also play a role in the development of the disease (Jen et al, 2012). Therefore, many studies have investigated the impact of mutations of both FBXW7 and NOTCH1 on treatment outcome. In this study, because of the limited amounts of samples and low incidence of FBXW7 mutations, we did not investigate this gene abnormality. Further studies on FBXW7 mutations and their prognostic importance in Chinese patients are needed.

To date, this is the largest study of Chinese paediatric T-ALL patients treated in a single institution. Our focus was on long-term outcome in patients with or without NOTCH1 mutations. Our study demonstrated an association of activating NOTCH1 mutations with favourable long-term prognosis in Chinese paediatric T-ALL patients treated using BFM-like protocols. Our results suggest that NOTCH1 mutations can be regarded as not only a risk factor for stratification purposes, but also an indicator for reducing chemotherapy intensity. Thus, a prospective study using NOTCH1 mutations as a criterion for classification is needed to fully elucidate the prognostic importance of this gene abnormality. Furthermore, therapeutic intervention targeting NOTCH1 signalling may lead to marked improvement in the long-term outcome of paediatric T-ALL patients.

Acknowledgements

This work was supported in part by National Natural Science Foundation Project (No. 81300432), National Key Technologies Research & Development Programme of the 11th 5-Year Plan (No. 2007BAI04B03), National Science & Technology Major Project of the 12th 5-Year Plan (No. 2011ZX09302-007-01), Beijing Health Qualified Personnel Programme (No. 2011-3-049), Beijing Municipal Education Commission Key Scientific and Technological Project (No. KZ201210025031). We gratefully thank all staff of the Haematology Oncology Centre of Beijing Children's Hospital, Capital Medical University, for their help in sampling.

Authorship and disclosures

GC designed the study, performed the experiments, collected the data and wrote the paper. LSG, LWJ, ZXX and CL helped to collect the samples. WMY, ZHY and ZRD recruited the patients and drafted the BCH-2003 and CCLG-2008 protocols. GC and LSG participated in the statistical analysis. LZG reviewed the final manuscript and takes primary responsibility for the article. The authors reported no potential conflicts of interest.

Ancillary