HOXA-activated early T-cell progenitor acute lymphoblastic leukemia: predictor of poor outcome?

The outcome for T-cell acute lymphoblastic leukemia (T-ALL) has strongly improved over the last decades using high-intensity treatment protocols approaching cure rates of 80% for pediatric patients and 60% for adult patients. Fifteen percent of pediatric ALL patients present with T-ALL, and they represent nearly half of the ALL patients who require the most intensive treatment. Intensive chemotherapy increases the risk for treatment related morbidity and mortality. For relapsed patients, the outcome is poor, as T-ALL cells in those patients are highly resistant to further treatment. Therefore, patient-tailored treatment and the introduction of high precision medicines remain important. Molecular cytogenetic characterization of T-ALL has greatly increased our understanding of the pathogenic events that drive this disease. In contrast to precursor B-ALL, this improved insight into T-ALL has not yet yielded prognostic factors that allow for the identification of patients at high-risk of relapse and who may be eligible to receive alternative treatment, including allogeneic stem cell transplantation.

One cytogenetic entity in pediatric and adult T-ALL patients that has been suspected to cause poor outcome include patients bearing a CALM-AF10 (PICALM-MLLT10) fusion as a consequence of a t(10;11)(p13.14;q14-21) chromosomal translocation.1 A first systematic study comprising unselected pediatric and adult T-ALL patients treated on FRALLE-93, FRALLE 2000 or LALA-94 protocols identified the CALM-AF10 fusion in approximately 9% of patients. This fusion is associated with early and late T-cell developmental arrest in the γδ lineage. In this study, late CALM-AF10⁺ T-ALL patients responded well to therapy, but 2 out of 12 CALM-AF10 patients with an immature phenotype did not respond to therapy, and another 8 patients with an immature phenotype relapsed.2 Therefore, CALM-AF10 may be associated with poor outcome in T-ALL. The 3 CALM-AF10 patients treated in the Dutch Childhood Oncology Group (DCOG) ALL-9 protocol demonstrated early relapses during therapy.3 Another study comprising 187 children treated on the AIEOP-BFM ALL 2000 or AIEOP R-2006 protocols identified CALM-AF10 fusions in 14 children, 8 of whom presented with high-risk features, including high white blood cell counts and prednisone poor responses. However, event-free survival and cumulative incidence of relapse were comparable for CALM-AF10 and CALM-AF10-patients.4 In contrast to the earlier study in which immature CALM-AF10 patients were associated with poor outcome,2 only 1 out of the 14 CALM-AF10 patients in these protocol studies had an immature T-cell immunopheno-type.4

In 2005, two independent studies led to the discovery of a chromosomal inversion on chromosome 7 (inv(7)(p15;q34)) in T-ALL patients.65 This fusion leads to ectopic activation of HOXA genes (HOXA5-10 genes in particular) by a cis-acting mechanism due to the close proximity of the TCRB enhancer region.65 A high activation of HOXA genes was observed previously in MLL-rearranged T- and B-cell neoplasms,7 but gene expression profiling studies revealed that HOXA-deregulation was a more common feature among T-ALL patients bearing inv(7), MLL-rearrangements, CALM-AF10 or SET-NUP214 gene fusions.985 The activation of HOXA genes therefore seems to play an important role in the cellular transformation of thymocytes. Various other HOXA-activating events have nowadays been identified in T-ALL including a TCRD-HOXA translocation10 and novel MLLT10 gene fusions. MLLT10 was identified fused to XPO1/CRM1, which encodes for a nuclear export protein,11 to NAP1L1 that encodes for a nucleosome assembly protein,12 and to HNRNPH1 and DDX3X genes that are both involved in RNA processing.13 In the latter study by Brandimarte et al.,13 all MLLT10-translocated HOXA cases clustered separately from other HOXA T-ALL cases including MLL-rearranged, SET-NUP214 and inv(7) cases. These MLLT10-translocated cases highly expressed the hematopoietic stem cell homeobox HHEX and MEF2C genes, two genes that are commonly expressed in immature T-ALL denoted as early T-cell precursor ALL (ETP-ALL).1412

Regarding this issue, Bond and colleagues further delineate the ambiguous nature of HOXA-activated T-ALL patients with respect to T-cell developmental arrest and outcome.15 They extended their previous observations regarding CALM-AF10 cases in a cohort of 190 adult TALL patients that were treated in LALA-94/GRAALL03-05 protocols. In that study, 42 T-ALL patients expressed an ETP-ALL immunophenotypic profile (CD5, CD1a, CD8-and expression of CD34, CD13, CD33 and/or CD117) whereas 148 patients had arrested at later stages of maturation.16 ETP-ALL patients fared equally well compared to non-ETP-ALL patients. The ETP-ALL group comprised 9 CALM-AF10 patients who were TCR negative due to absent or incomplete TCRδ or TCRγ recombinations (i.e. IM, IMδ or IMγ). In the non-ETP-ALL group, 2 out of 5 CALM-AF10 patients had a similar TCR-negative genotype while the remaining 3 had the more mature sCD3/TCR or cortical/pre-αβ TCR-genotype. ETP-ALL patients who were positive for CALM-AF10 (and were TCR-) had a significantly higher risk for adverse events and a trend toward reduced overall survival compared to CALM-AF10 ETP-ALL patients. CALM-AF10 ETP-ALL patients fared equally well compared to non-ETP-ALL patients regardless of their CALM-AF10 status.16

In the study, Bond and colleagues extended their observations by analyzing the prognostic impact of all HOXA-positive cases in relation to ETP-ALL and outcome in a cohort of 209 adult T-ALL treated in the GRAALL-2003/2005 protocol.15 Fifty-five T-ALL cases (26%) expressed HOXA9 at levels similar to those in CALM-AF10 patients as defined by RT-QPCR. Apart from 8 patients with CALM-AF10 translocations (1 patient also had an inv(7)), 10 patients had inv(7), 9 were positive for SET-NUP214, and 6 patients had MLL-rearrangements. Further screening for alternative HOXA-activating events revealed XPO1-MLLT10, DDX3X-MLLT10 or NAP1L1-MLLT10 fusions in 1 patient each, whereas 1 additional patient had an unresolved MLLT10 rearrangement. Two patients were identified with NUP98-RAP1GDS1 fusions, whereas no HOXA-activating events were found in the remaining 16 cases. HOXA cluster activation in T-ALL patients by these oncogenic fusion products (denoted as trans-acting mechanism) is mostly linked with the presence of an immature TCR-genotype (IM0, IMδ or IMγ) and an ETP-ALL immunophenotype. In contrast, inv(7)-positive (cis-HOXA-acting) patients almost exclusively present with a sCD3/TCR or cortical/pre-αβ TCR-genotype. In line with previous observations,4 HOXA patients demonstrated increased resistance to corticosteroid treatment and chemotherapy and frequently remained MRD positive (>10) after induction therapy.15 Surprisingly, overall outcome (OS, EFS and DFS) for HOXA patients was identical to HOXA T-ALL patients. Further discriminating patients based on ETP-ALL immunophenotype revealed that HOXA/ETP-ALL patients had a significantly poor outcome as compared to HOXA/ETP-ALL patients (OS: 31.2% vs. 74.2%; EFS: 25% vs. 60.8%; DFS: 28.6% vs. 64.7% and CIR: 53.7% vs. 29.2%, respectively). The outcome for HOXA/ETP-ALL patients was as favorable as for non-ETP-ALL patients regardless of the presence of HOXA-activating rearrangements.15

The Children’s Oncology Group (COG) has now reported similar findings for pediatric T-ALL patients.17 This study investigated 100 children with T-ALL who were treated in the COG AALL0434 protocol, including 17 patients for whom initial treatment failed. Evidence for MLL-rearrangements were found in 12 patients in addition to 6 CALM-AF10 patients, 3 DDX3X-AF10 patients (1 case had a complex CASK-DDX3X-AF10 translocation), 2 patients with NUP98-rearrangements and 3 inv(7)+ patients. MLL-but not AF10-rearrangements were strongly associated with induction failure and inferior EFS in uni- and multivariate analyses. Expression of an ETP-ALL expression signature also predicted for inferior EFS, and trended towards enrichment of MLL-rearranged cases. MLL-rearrangements combined with an ETP-ALL expression profile most strongly associated with induction failure, refractory disease and relapse.17 Both studies therefore point to HOXA-activated ETP-ALL cases that are at a higher risk to fail on induction therapy or have inferior survival rates. Further studies are needed to investigate whether this can be attributed to specific HOXA-activating events: 13 HOXA-activated adult T-ALL patients that relapsed included 3 patients with SET-NUP214 fusions, 2 patients with MLLT10-rearrangements, 1 MLL-rearranged case and 1 patient with an inv(7),15 while in the pediatric study the MLL-AF6 or Del3’MLL rearranged patients were at the highest risk to fail on therapy.17

Based on these important findings in both studies, routine screening for HOXA-activation events and ETP-ALL profiles in future T-ALL patients may help to identify patients at risk for induction failure or relapse. For this to happen, several issues need to be resolved: What would be the best detection method to identify HOXA-activated ETP-ALL patients? Gene expression profiling to identify HOXA-activated T-ALL patients failed to classify 3 patients carrying HOXA-activating events in the COG series.17 On the other hand, not all HOXA-activating genetic events have been resolved.15 Also, what is the best method to identify ETP-ALL patients? Will this rely on ETP-ALL immunophenotypic markers18 or on the expression of ETP-ALL signature genes? Several retrospective studies on historical T-ALL samples did not consistently identify an ETP-ALL-specific immunophenotype for cases that expressed an ETP-ALL gene signature.2019 Finally, what alternative treatment should be given? HOXA-activated ETP-ALL patients may receive allogeneic stem cell transplantations if suitable donors are available, or receive precision medicine like the DOT1L inhibitor EPZ-5676 compound that is currently being tested in clinical trials.21 An additional important question that needs to be resolved in the future is which other genetic factors may cause HOXA-activated cases to arrest at the ETP-stage and define poor outcome, while other HOXA-activated T-ALL cases with seemingly identical chromosomal rearrangements arrest at late stages and have a better prognosis? The answer may reveal the true determinant that defines ETP-arrest and the high-risk of treatment failure for HOXA-activated ETP-ALLs.

Acknowledgements

This study was supported by the Children Cancer Free Foundation (Stichting Kinderen Kankervrij, KiKa) grants KiKa2008-29 and KiKa2013-116 (KC-B) and the Dutch Cancer Society KWF2010-4691 (EV).