Page 130 - Read Online

Santiago et al. J Transl Genet Genom 2021;5:380-95 https://dx.doi.org/10.20517/jtgg.2021.16 Page 382

MOLECULAR CLASSIFICATION OF ALL
B-ALL
[12]
In the early 2000s, Yeoh et al. confirmed that GEP is able to accurately classify known cytogenetic
subgroups and sometimes, rectify karyotyping misclassification for the ETV6-RUNX1 subgroup. They also
highlighted unique gene expression phenotypes identifying novel subgroups for previously unclassified B-
ALL . Furthermore, gene expression clustering correlates with outcome and can be used for prognostic
[12]
and risk-group classification [13,14] . Herein, we will only emphasize on the recently discovered molecular
subgroups, their potential prognostic association and their predilection according to age group. Figure 1
summarizes the B-ALL subgroups with their distribution by age group and a preliminary proposition of
risk-group classification.

Ph-like ALL
The discovery of Ph-like or BCR-ABL1-like ALL hails from genomic exploration of HR B-ALL [13-17] . This
subgroup is defined by an activated kinase gene expression profile, similar to that of Ph ALL but missing
+
the canonical BCR-ABL1 fusion. This subset was identified in 2009 by two independent ALL cohorts
defined by two distinct gene classifiers that shared only 7 genes in common [15,16] . Interestingly, Ph-like ALL
+
is associated with a worse prognosis than other HR B-ALL and comparable to that of Ph ALL . In
[15]
multivariable analysis, the prognostic significance of Ph-like ALL was retained as an independent adverse
[18]
outcome biomarker for relapse . This subgroup accounts for ~15% of pediatric B-ALL and increases with
age and risk group. In younger children, it represents 10% of standard-risk (SR) ALL and 13% among HR-
ALL. The prevalence increases to 21% among adolescents and 27% in young adults and then stabilizes
around 20% in older adults after the age of 40 [19,20] . Alterations (deletions and inactivating mutations) of
IKZF1 or other lymphoid transcription factors (CDKN2A/B, PAX5, ERG, ETV6) and kinase-activating
alterations constitute the molecular hallmark of Ph-like ALL [13,14,16,19] . CRLF2 rearrangements (e.g., IGH-
CRLF2 and P2RY8-CRLF2), conferring CRLF2 overexpression, comprise half of Ph-like ALL, often harbor
concomitant JAK mutations or other JAK-STAT pathway alterations (SH2B3, IL7R) in about 50% of the
CRLF2-rearranged cases . CRLF2 rearrangements are associated with a worse prognosis and are more
[19]
frequent in older children and in people with Hispanic or Native American origin . Overall, more than
[17]
90% of Ph-like ALL harbor a myriad of kinase-activating alterations that can be further divided in 2 major
categories: (1) alterations activating JAK-STAT signaling pathways, predominantly rearrangements of
CRLF2, JAK2 and EPOR; and (2) translocations involving ABL-class genes (ABL1, ABL2, CSF1R, PDGFRA,
PDGFRB, LYN). A small number of Ras pathway (KRAS, NRAS, PTPN11, NF1) mutations have been
identified; nevertheless, it remains unclear whether these mutations represent pathogenic drivers in Ph-like
ALL or only contribute to a similar kinase-activated signature. In addition, some rare fusions involving
other tyrosine kinases such as NTRK3, DKGH or FLT3 have been reported in Ph-like ALL [10,19,20] . IKZF1
(gene encoding for the transcription factor IKAROS) intragenic deletions and inactivating mutations are
preponderant in kinase-activating leukemia such as Ph or Ph-like ALL [14,16,21,22] . IKZF1 alterations are
+
+
present in up to 70% of Ph ALL and Ph-like ALL and confer a worse prognosis compared to their
respective counterparts with wild-type IKZF1 [19,23] . Nevertheless, the independent negative prognostic
impact of IKZF1 deletion in B-ALL is still debatable. IKZF1 was first described as a pejorative marker for the
occurrence of relapse , but multivariable analyses across different consortia had come to conflicting
[16]
conclusions. In the Dana Farber Cancer Institute (DFCI) ALL 05-001 study, IKZF1 deletion was associated
with poor survival irrespective of the presence of kinase fusion and MRD among 105 NCI HR B-ALL
cases [7,23] . However, the results from Children’s Oncology Group (COG) P9905/P9906 trials failed to confirm
the prognostic impact of IKZF1 alterations in multivariable analysis when analyzed by risk group [22,24] . The
enrichment of IKZF1 alterations in the already known unfavorable Ph-like ALL represent major
confounders and contribute to the uncertain prognostic impact of IKZF1. A new category, called IKZF1 ,
plus
regrouping IKZF1 deletions that co-occur with CDKN2A, CDKN2B, PAX5 or PAR1 deletion in the absence