Genetic markers in the prognosis of patients with ALL

Despite much research, the mutational landscape of patients with acute lymphoblastic leukemia (ALL) remains relatively poorly characterized, especially for rarer subtypes. Identifying prognostic markers to aid risk stratification and predict response to treatment is invaluable to further enable the development of targeted therapies.

During the 62nd American Society of Hematology (ASH) Annual Meeting and Exposition, three abstracts were presented on novel genetic markers found to be associated with clinical outcomes in patients with ALL. Martin Neumann discussed how mutations in the TLX1 and NKX2-1 subgroups may confer favorable outcomes in patients with T-cell ALL, whilst Yuya Sasaki presented data from patients with Philadelphia chromosome-positive (Ph+) B-cell ALL treated with tyrosine kinase inhibitors, demonstrating a link between deletions in IKZF1 and poor survival. Lastly, Judith Boer shared work on NUTM1 fusion genes associated with survival benefit in pediatric patients with B-cell precursor (BCP) ALL. Key findings from these presentations are shown below.

Molecular subgroups of T-cell ALL¹

Targeted therapeutic options for patients with T-cell ALL are limited, with risk stratification usually based on immunophenotype. Molecular subgroups, according to aberrant oncogene expression, are not well defined, and to date few studies have been performed in adults.

Key findings

In this study, whole RNA transcriptome sequencing was performed on bone marrow samples from 163 patients with T‑cell ALL. Median age was 30 years (range, 17–83 years) and 121 patients were male. Molecular subgroups were assigned to 162 of these patients based on oncogene overexpression, as summarized in Table 1.

Table 1. Molecular subgroups assigned by RNA sequencing to patients with T-cell acute lymphoblastic leukemia¹

Molecular subgroup	Patients, n (N = 162)	Frequency of gene fusion events, %
TLX1	37	51
HOXA	37	70
TAL1/LMO1	34	44
LYL1/LMO2	32	9
TLX3	17	24
NKX2-1	4	50
TAL2	1	—

There was an age-dependent component to the molecular subgroups identified, with more patients aged 16–25 years in the TAL1/LMO1 subgroup than those aged >35 years (35% vs 3%; p = 0.001). There was also a trend for more older patients to be allocated to the LYL1/LMO2 and HOXA subgroups (40%) than younger patients (23%).

The mutational landscape was analyzed using deep targeted DNA sequencing in 84 patients, revealing a high frequency of NOTCH1 (71%) and PHF6 (67%) mutations in the TXL1 subgroup. Within the TLX3 subgroup, JAK/STAT (30%) and SUZ12 (40%) mutations were common, while mutations in the epigenetic regulators KMT2D (20%) and DNMT3A (20%) were associated with the more immature LYL1/LMO2 subtype. Interestingly, the TAL1/LMO1 subgroup showed only a few mutations, with the exception of alterations in the epigenetic regulator PTEN gene (29%).

Profiling by DNA methylation confirmed the mutational analysis and showed a hypomethylated signature in the LYL1/LMO2 and TAL1/LMO1 subgroups, while all other subgroups had a more homogenous methylation profile.

Clinical outcomes are shown in Table 2 for patients with data available.

• Overall survival analysis differentiates the favorable TLX1 and NKX2 subgroups, with a 5-year overall survival (OS) of 94%, which correlated with complete molecular response rates of 93% in the TLX1 group.
• The 5-year OS for the combined subgroups TAL/LMO, HOXA, TLX3, TAL2, and LYL1/LMO2 was only 62% (p = 0.007), correlating with lower complete molecular response rates.

Table 2. Correlation of molecular subgroups and clinical responses (adapted from Neumann et al.¹)

CR, complete remission; OS, overall survival. *Defined as minimal residual disease negative after consolidation one.
Response	n = 129
Cytologic response, n (%)
CR	126 (98)
Failure	2 (2)
Death	1 (1)
Molecular CR* by subgroup, n (%)
TLX1 (n = 30)	28 (93)
HOXA (n = 18)	11 (58)
LYL1/LMO2 (n = 6)	2 (33)
5-year OS by subgroup, %
TLX1	93
NKX2-1	100
TLX1/NKX2-1	94
TAL/LMO	76
All other subgroups	62

Genetic alterations in Ph+ B-cell ALL²

Genomic studies have detected deletions in IKZF1, CDKN2A/2B, PAX5, BTG1, and EBF1 in patients with Ph+ B‑cell ALL. In particular, patients with IKZFI deletions show poor prognosis when treated with regimens that include chemotherapy and the tyrosine kinase inhibitors (TKIs) imatinib or dasatinib. However, little is known about genetic prognostic markers in patients on ponatinib-based treatment regimens. Sasaki and colleagues used targeted capture DNA sequencing and copy number alteration analysis with single nucleotide polymorphism microarray or whole exome sequencing to investigate molecular determinants of outcomes in adults with Ph+ B‑cell ALL who were treated with hyper-CVAD plus dasatinib in NCT00390793 or hyper-CVAD plus ponatinib in NCT01424982.

Key findings

Pretreatment bone marrow or peripheral blood specimens from 105 patients were analyzed, 55 of whom were treated with dasatinib and 50 with ponatinib. Median age was 51.9 years (range, 22–80 years), 39% were female, and cohorts were comparable with regards to blood markers and cytogenetic abnormalities.

Deletions in IKZF1 were found to be the most common alteration (n = 63; 60%). Within these patients, the Ik6 isotype (deletion of exon 4–7; 44%) and Ik2 (deletion of exon 2; 32%) were the most frequent. Interestingly, 75% of patients with an IKZF1 deletion also had a deletion in VPREB1, CDKN2A/2B, EBF1, or PAX5 (defined as IKZF1^plus). There was no significant difference in genetic alternations between patients treated with dasatinib and ponatinib.

Overall, patients with IKZF1^plus had inferior outcomes compared with those with IKZF1 deletions only or without IKZF1 deletions.

• IKZF1^plus was strongly associated with poorer OS (p = 0.0116).
• Stratification by TKI type revealed that, overall, patients treated with ponatinib had improved survival compared with those treated with dasatinib (p = 0.00046).
• With dasatinib treatment, there was no difference in OS between patients with IKZF1^plus and those with IKZF1 deletions only or without IKZF1 deletions (p = 0.581).
• However, in ponatinib-treated patients, IKZF1^plus was significantly associated with poorer survival (p = 0.00637).
• Multivariate analysis showed TKI type, MRD-positivity at 3 months, and IKZF1^plus status were independent factors associated with poor prognosis.

The authors therefore proposed that IKZF1^plus status could be used for risk factor stratification when using ponatinib‑based therapy in patients with Ph+ ALL.

NUTM1 rearrangements in BCP ALL

NUTM1 fusions were recently discovered as a novel genetic subtype of BCP ALL. Rearrangement of NUTM1 (NUTM1r) on 15q14 and subsequent fusion with a partner gene can cause upregulation of the proto-oncogene BMI1 and transcription of the HOXA gene cluster. Although this subtype is rare in pediatric patients, it is thought to be more common in infants negative for KMT2A rearrangement (KMT2Ar) than children.

Key findings

Boer and colleagues screened 161 patients with BCP ALL who were KMT2Ar-negative in Interfant-99/06 study groups and identified 35 as NUTM1r-positive and 126 as NUTM1r-negative using break-apart fluorescent in situ hybridization, RNA sequencing, and reverse transcription polymerase chain reaction. A further 11 infants and 39 children with BCP ALL within study groups in the Ponte di Legno consortium were also found to be NUTM1r-positive. This translates to a rate of ~5% of NUTM1 rearrangement in infant ALL, while in pediatric ALL, the rate was < 1%. No cases of NUTM1 rearrangement have been reported in adult ALL.

Ten partner genes of NUTM1 were identified in infants and children within the study cohort. ACIN1 (44%), BRD9 (26%), and CUX1 (15%) were the most common fusion genes in infants, whereas in children a more diverse range of fusion genes was found, including CUX1 (28%), ACIN1 (22%), and ZNF618 (22%). HOXA9 upregulation was detected in all infant cases of BCP ALL with NUTM1 fusions, and in ~50% of children.

Clinical characteristics are shown in Table 3.

• Within the Interfant cohort, NUTM1-positive cases were younger than the NUTM1-negative cases, and more patients were diagnosed at <6 months of age.
• NUTM1 rearrangements were associated with improved 4-year event-free survival and OS compared with patients without these fusions, suggesting that they have potential as prognostic markers.

Table 3. Clinical characteristics and responses of infants and children with BCP ALL (adapted from Boer et al.³)

Clinical characteristic	Interfant cohort				Consortium cohort
KMT2Ar-negative/ NUTM1r-postive (n = 35)	KMT2Ar-negative/ NUTM1r-negative (n = 126)	95% CI	p value	NUTM1r-positive (n = 43)
Median age, years (range)	5.6 (0.4–11.0)	9.3 (0.1–11.9)	—	<0.00001	—
Diagnosed <6 months of age, %	54	16	—	<0.0001	—
Male, %	51	57	—	0.57	—
4-year EFS, %	100	74	65–81	0.001	90.3
4-year OS, %	100	88	81–93	0.04	100
CCR, n	35	93	—	—	42
ALL, acute lymphoblastic leukemia; BCP, B-cell precursor; CCR, continuous complete remission; EFS, event-free survival; OS, overall survival.

Conclusion

These three abstracts have uncovered novel genetic markers that correlate with clinical outcomes in patients with subtypes of ALL.

• In T-cell ALL, patients with TLX1 and/or NKX2-1 overexpression represent a subgroup with superior survival. Identification of patients with specific mutations in these genetic subgroups may promote more accurate risk stratification.
• In Ph+ ALL, patients with IKZF1^plus mutations show inferior outcomes with ponatinib treatment, suggesting this abnormality could help to risk stratify patients and target treatment accordingly.
• In infant BCP ALL, 5% of patients harbor a NUTM1 rearrangement, which is associated with excellent outcomes, indicating a possible use as a prognostic marker.

These markers have the potential to aid in the selection of targeted therapy options and predict prognosis, and also highlight the need for further research into the relationship between mutational landscape and treatment response in such patients.