The UKALL-HeH model could potentially redefine low-risk high hyperdiploidy in pediatric acute lymphoblastic leukemia

Approximately 30% of pediatric patients with acute lymphoblastic leukemia (ALL) have a high hyperdiploid (HeH) karyotype, defined as the presence of 51 to 67 chromosomes. Although HeH is widely associated with a favorable outcome in both pediatric and adult ALL, its prevalence means that HeH accounts for ~25% of all relapses. Numerous studies have assessed specific trisomies, model chromosome number, and pattern of gain as potential risk factors with the aim of identifying robust groups with low risk of relapse. Despite all these studies, there is no universally accepted definition of low-risk HeH, other than the traditional modal chromosome number.

Anthony Moorman presented the results of a study, UKALL-HeH, at the 26th Congress of the European Hematology Association (EHA2021), which evaluated all relevant risk factors to define a very low-risk subgroup within HeH with low risk of relapse, with the aim to use it prospectively to identify patients eligible for treatment de-escalation.¹

Methods

To evaluate all possible significant independent risk factors, an analysis was performed using data from the UKALL97/99 (n = 456, discovery) and UKALL2003 (n = 725, validation) studies, with a median follow-up of 10.6 and 9.4 years, respectively.

Event-free survival (EFS), relapse rate (RR), and overall survival (OS) rates were calculated at 10 years, and data were compared using the univariate and multivariate analysis Cox regression models.

Area under ROC curve, a general linear model, and targeted projection pursuit were used to establish the optimal number of chromosomes required to accurately predict outcomes.

Results

Using the discovery cohort, analysis of all possible combinations of one to six trisomies revealed that four chromosomal trisomies produced the maximum prediction power.

The multivariate Cox modelling identified that chromosomes +5, +17, +18, and +20 trisomies were independent factors that could significantly predict outcome; adding further chromosomes did not improve the prognostic model.

Cluster analysis of the four identified chromosomes revealed that two groups were associated with different risks of relapse. A good risk (GR) and a poor risk (PR) HeH profiles, comprising the following, were developed:

• GR profile: Patients had (i) +17 and +18 trisomies together, or (ii) either +17 or +18 but not +5 or +20
• PR profile: All other HeH trisomies

To validate the profiles, the defined parameters were applied to 725 patients with HEH from the validation cohort (UKALL2003) that had similar risk group sizes relative to the discovery cohort (Table 1).

Table 1. Validation of UKALL-HeH classifier to UKALL2003 (validation cohort)*

Adapted from Moorman et al.1
Parameter	Good risk	Poor risk
Total, n (%)	579 (80)	146 (20)
Relapse rate
Rate at 10 years, %	5	16
Hazard ratio (95% CI)	3.80 (2.14−6.75)
p value	< 0.001
Event-free survival
Rate at 10 years, %	92	81
Hazard ratio (95% CI)	2.68 (1.67−4.3)
p value	< 0.001
Overall survival
Rate at 10 years, %	96	86
Hazard ratio (95% CI)	3.42 (1.86−6.27)
p value	< 0.001

• Patients categorized to the HeH-PR profile had higher RR in both the discovery and validation cohorts: hazard ratio (HR) 2.50 (95% CI, 1.51−4.14) vs 3.80 (2.14−6.75), p < 0.001, independent of measurable residual disease (MRD).
• Threshold analysis revealed that 0.03% was the optimal MRD threshold. The RR in MRD-negative patients was 4% for HeH-GR, and 11% for HeH-PR, p = 0.001. For MRD-positive patients, the RR was 8% for HeH-GR, and 27% for HeH-PR, p = 0.003.
• The UKALL-HeH model showed that there was no correlation between HeH risk groups regarding age, sex, or white blood cell count.
• Patients classified as HeH-GR who received the least intensive therapy had favorable outcomes, compared with the outcome of HeH-PR patients.
• The area under the curve ROC analysis revealed that the UKALL-HeH model had higher specificity, with a better prediction accuracy, compared with previously published risk profiles, including triple trisomies (Table 2).

Table 2. Comparison of HeH in pediatric ALL*

Conclusion

The authors developed and validated a model that was able to define two distinct clinically relevant risk groups. The UKALL-HeH profile demonstrated that patients with a GR profile had more favorable outcomes, with a lower risk of relapse, and could be considered for treatment de-escalation. By contrast, patients with the UKALL-HeH PR profile had outcomes that were similar to patients with intermediate-risk cytogenetics. The UKALL-HeH could potentially represent a new model for defining risk groups and help guide treatment.