Integration Of Cytogenetic and Copy Number Alteration (CNA) Data Into a Single Risk Profile For Paediatric Acute Lymphoblastic Leukaemia (ALL)

Chromosomal abnormalities are established indicators of outcome in paediatric ALL. Recently numerous CNA have been reported which target genes involved in leukaemogenesis. The prognostic impact of these CNA has yet to be fully elucidated. We have used a novel approach to simultaneously assess the prognostic impact of the most prevalent CNA and integrate the results with cytogenetic risk groups. The CNA status of 8 genes/loci (CDKN2A/B, IKZF1, PAX5, ETV6, PAR1 (P2RY8-CRLF2), EBF1, BTG1 & RB1) was determined by multiplex ligation-dependent probe amplification (P335 kit, MRC Holland) in a representative cohort of 864 B-cell precursor (BCP) ALL patients treated on MRC ALL97/99. The presence or absence of each CNA was determined and each individual 8-gene combination was assessed and classified into risk categories on the basis of patient numbers, hazard ratio (HZR) for an event and p value. Using these risk categories patients were assigned to 3 groups: good (CNA-GR, 64% cases); intermediate (CNA-IR, 25%); poor (CNA-PR, 11%). The 5 year event free survival (EFS) rates were: CNA-GR 83%; IR 70%; PR 55%. The HZR comparing CNA-GR/IR and CNA-PR/IR were 0.54 (99% CI 0.40-0.73) p<0.001 & 1.76 (1.23-2.53) p=0.002. Next, we validated these groups using an independent and representative cohort of 781 BCP-ALL patients treated on UKALL2003. The distribution of patients across the three groups was similar: 64%/28%/9%. The EFS rates were: CNA-GR 93%; IR 82%; PR 90%. The HZR comparing CNA-GR/IR and CNA-PR/IR were 0.35 (0.22-0.56), p<0.001 and 0.60 (0.27-1.35), p=0.221. This validation confirmed the utility of CNA risk grouping to identify robust good and intermediate, but not poor, risk groups.

We previously defined 3 cytogenetic risk groups: CYTO-GR - ETV6-RUNX1, high hyperdiploidy; CYTO-PR - BCR-ABL1, MLLtranslocation, near haploidy, low hypodiploidy, intrachromosomal amplification of chromosome 21 (iAMP21), t(17;19)(q23;p13) and, in the absence of GR abnormalities, 13q deletions and 17p abnormalities; CYTO-IR - other cytogenetically normal or abnormal cases. There was a significant correlation between the CNA and cytogenetics risk groups (p<0.001). Using the ALL97/99 cohort, we cross-tabulated the CNA and cytogenetic risk groups, examined the outcome of all 9 combinations and then derived 3 genetic risk groups: GEN-GR (72%); IR (16%); and PR (13%) which had differential EFS rates: 85%, 65% and 47%. The GEN-GR/IR and GEN-IR/PR HZR were 0.34 (0.24-0.47) p<0.001, 1.69 (1.17-2.43) p=0.005. These 3 genetic risk groups were then validated on the UKALL2003 cohort where they identified patient groups with differential EFS: GEN-GR 93%, IR 82%; PR 71%. The HZR for GEN-GR/IR and GEN-PR/IR were 0.33 (0.20-0.56) p<0.001 and 1.89 (1.01-3.57), p=0.048.

The GEN-PR and CYTO-PR groups were equivalent indicating that the integration of CNA data did not identify any new high risk (HR) patient groups. In addition, all the CYTO-GR patients were classified in the GEN-GR group confirming their good prognosis. However, the integration of CNA data allowed the CYTO-IR group to be split; identifying a novel group of patients which accounted for 18% of BCP-ALL and was associated with inferior outcome. UKALL2003 patients in this group (n=141) had a mean age of 9.4 years, 60% were NCI-HR and 54% were MRD-HR. Compared with other UKALL2003 CYTO-IR patients those in this novel subgroup had a 3-fold increased risk of a relapse (14% v 5% at 5years, HZR 3.14 (1.25-7.93) p=0.015), an inferior EFS (82% v 94%, HZR 3.52 (1.52-8.17), p=0.003) and lower survival (86% v 98%, HZR 9.32 (2.17-40.03), p=0.003) (all at 5 years).

In conclusion, the genetic heterogeneity in paediatric ALL can be used to classify patients into clinically relevant groups. The integration of CNA data into our previously identified cytogenetic risk system enabled refinement of the CYTO-IR risk group to identify a subgroup of patients with inferior outcome.