Prognostic impact of gained chromosomes in high-hyperdiploid childhood acute lymphoblastic leukaemia: a collaborative retrospective study of the Tokyo Children's Cancer Study Group and Japan Association of Childhood Leukaemia Study

Authors


Although the prognosis of high-hyperdiploid (HHD) acute lymphoblastic leukaemia (ALL) is excellent, relapse occurs in 10–15% of cases (Look et al, 1985). A gained chromosome is commonly found (Heerema et al, 2000; Kawamata et al, 2008), and previous studies reported a correlation between other chromosome combinations and outcomes. The Pediatric Oncology Group (POG) and Children's Oncology Group (COG) demonstrated that the combined gain of chromosomes 4, 10 and 17 (termed as triple trisomy) was associated with a better prognosis (Sutcliffe et al, 2005), and POG data suggested that +4 and +10 (termed as double trisomy) patients had a very low risk of relapse (Harris et al, 1992).

In this study, we performed a retrospective analysis with the Tokyo Children's Cancer Study Group (TCCSG) cohort as a test set to investigate the relationship between a combination of specific chromosome gains and disease outcomes, and used patients included in the Japan Association of Childhood Leukaemia Study (JACLS) as a validation set.

Paediatric ALL patients (aged 1–18 years old) enrolled in the TCCSG L95-14 (Igarashi et al, 2005) (n = 597, 1995–99), L99-15 (Manabe et al, 2008; Hasegawa et al, 2012; Kato et al, 2014) (n = 770, 1999–2003), and JACLS ALL97 (Suzuki et al, 2010) (n = 674, 1997–2002) trials were analysed. Based on cytogenetic data obtained with the G-banding test, HHD was defined as a modal chromosome number of 51 or more (Heerema et al, 2000), and 186 and 75 HHD patients were analysed from the TCCSG and JACLS cohorts, respectively.

Associations between the gained chromosome pattern and outcomes were investigated with all two combinations of each chromosome in the TCCSG cohort. Combinations with a small number of patients (<25% of all patients, i.e. 47 patients) were excluded. Fifteen combinations were found to have a significant impact on outcome; the association of these combinations and outcomes in the JACLS cohort was analysed for validation (Table SI and Figure S1).

Event-free survival (EFS) was calculated using Kaplan–Meier estimates, and the log-rank test was used to detect significant differences. Multivariate analysis was performed using the Cox proportional hazard regression model. All statistical analyses were performed with the statistical software r (version 2.13.0; The R Foundation for Statistical Computing, Vienna, Austria). A two-sided P-level of <0·05 was considered significant for all analyses.

The characteristics of the paediatric patients with HHD ALL are shown in Table 1. The median follow-up period was 2160 d in the TCCSG cohorts. The 6-year EFS and overall survival (OS) for HHD patients in the TCCSG cohort was 79·7 ± 3·2% and 91·2 ± 2·3% respectively, compared to 74·4 ± 1·3% and 83·4 ± 1·1%, respectively, for non-HDD patients. The distribution of gained chromosomes in the TCCSG cohort is shown in Fig 1A. In the JACLS cohort, the median follow-up period was 2791 d, and the 6-year EFS and OS was 86·6 ± 4·0% and 97·2 ± 2·0%, respectively.

Table 1. Characteristics and outcomes of patients with high-hyperdiploidy acute lymphoblastic leukaemia.
  n Median age at diagnosis, years (range) P Median WBC at diagnosis, ×109/l (range) P 6-year EFS P a
  1. TCCSG, Tokyo Children's Cancer Study Group; JACLS, Japan Association of Childhood Leukaemia study; WBC, white blood cell count.

  2. a

    Log-rank test.

TCCSG cohort
All1864 (1–14) 5·8 (0·3–16·5) 79·7 ± 3·2% 
+11 or +17
Yes1354 (1–14)0·856·2 (0·3–16·5)0·2983·2 ± 3·5%0·027
No513 (1–12)4·7 (0·8–46·4)70·8 ± 6·7%
JACLS cohort
All753 (1–15) 5·6 (1·6–970) 86·6 ± 4% 
+11 or +17
Yes563 (1–15)0·635·3 (1·6–970)0·5291·1 ± 3·8%0·045
No194 (1–9)6·3 (1·9–60·1)73·0 ± 10·4%
Figure 1.

Gained chromosomes in HHD-ALL and their prognostic impact. (A) Frequency of gained chromosomes in 186 HHD-ALL patients in the TCCSG cohort. Event-free survival in the TCCSG and JACLS cohorts is shown. Concurrent absence of a gain of both chromosomes 11 and 17 was associated with poor outcome in the TCCSG cohort (B) and JACLS cohort (C). (D) Event-free survival of ALL patients with or without double trisomy (+4 and +10) in the TCCSG cohort. HHD, high-hyperdiploidy; ALL, acute lymphoblastic leukaemia; TCCSG, Tokyo Children's Cancer Study Group; JACLS, Japan Association of Childhood Leukaemia study.

The absence of +11 and +17 was associated with a poorer outcome in the TCCSG cohort (Table 1 and Fig 1B). Fifty-one of 186 (27·4%) patients with HHD-ALL and no extra copies of these two chromosomes had a significantly poorer prognosis, with a 6-year EFS of 70·8 ± 6·7% compared to HHD patients with either +11 or +17 (83·2 ± 3·5%, P = 0·027). However, no significant difference was observed in OS between the two groups (83·5 ± 5·9% with no +11 and no +17, and 94·3 ± 2·1% with +11 or +17, P = 0·09). Mulitvariate analysis failed to identify these chromosome gains as statistically significant due to small sample size (Table SII).

The correlation was concordant with that found in the JACLS cohort. Nineteen of 75 (25·3%) HHD patients had neither +11 nor +17, and EFS was inferior to that of patients with +11 or +17 (73·0 ± 10·4% and 91·1 ± 3·8%, P = 0·045) (Fig 1C). No significant difference was observed in age, leucocyte count at diagnosis, or gender between the two groups (Table 1).

In contrast to the findings of previous studies by the POG and COG (Harris et al, 1992; Sutcliffe et al, 2005), double/triple trisomy was not correlated with outcome in our cohort. In the TCCSG cohort, 105 of 186 (56·5%) patients had both +4 and +10, and EFS at 6 years was 81·2 ± 4·2%, whereas EFS of 81 patients without +4 or +10 was 77·8 ± 4·8%, which was not significantly different (P = 0·64) (Fig 1D).

Based on the finding that the gain of chromosomes is non-random, it is assumed that the gain of specific chromosomes contributes to leukaemogenesis, while the gain of other chromosomes is a “passenger” event, which is a by-product of leukaemic cell development.

This study showed an association between +11 and +17 and EFS probability in two independent patient cohorts, each of which received different treatments. Data from the POG and COG showed that +4, +10, and +17 was associated with outcome (Heerema et al, 2000; Sutcliffe et al, 2005), and another report on the Berlin-Frankfürt-Münster study cohort demonstrated that patients with neither +17 nor +18 had poor outcomes (Kawamata et al, 2008). Thus, the gain of chromosome 17 may be the most important ‘driver’ abnormality in HHD pathogenesis and provide a favourable phenotype. However, it should be noted that no relapse was observed in 21 patients with +11 and +17 of the TCCSG cohort, which suggests that +11 is still associated with better prognosis even in patients with +17.

Some of our findings were inconsistent with those of previous studies. This suggests that the association between the gained chromosome combination and outcome may be influenced by the treatment regimen or ethnicity because most of the enrolled patients in the present study were Asian. The significance of the gain of specific chromosomes in HHD ALL on prognosis should be considered carefully.

Acknowledgement

All authors have no conflict of interest to disclose. We thank Kaori Itagaki for preparing and refining the protocol data for ALL in the TCCSG. We also thank Ms. Noriko Sato for the data management of JACLS ALL97. We thank Dr. Ryo Inuzuka for technical assistance. The TCCSG study was supported in part by the Children's Cancer Association of Japan.

Author contributions

A.M., K.K., H.T., T.T., Y.H., M.T. and A.O. designed the TCCSG study and collected the data. T.I., Y.H., A.S., H.H., M.I., K.H., S.K. and M.O. designed the JACLS study and collected the data. M.K. and T.I. analysed the data. M.K., T.I. and A.M wrote the paper. All authors discussed the results and commented on the manuscript.

Ancillary