CREBBP is a Major Prognostic Biomarker for Relapse in Childhood B-cell Acute Lymphoblastic Leukemia: A National Study of Unselected Cohort

Acute lymphoblastic leukemia (ALL) is the most prevalent form of cancer among children, comprising 25% of all childhood malignancies, with a consistently increasing incidence rate over the years [1, 2]. It arises following the clonal proliferation of immature B and/or T lymphoid cells, with around 80% of the cases being of B lineage origin [3, 4]. The most common initial genetic lesions are chromosomal loss (hypodiploidy), gain (hyperdip-loidy), or fusion genes, leading to a pre-leukemic clone. A subsequent second hit, either a copy number alteration (CNA) or single nucleotide variant (SNV), is believed to be the cause of lymphoid arrest and the development of symptomatic disease [4, 5].

Conventional (karyotyping, fluorescence in situ hybridization - FISH) and molecular (reverse transcription quantitative polymerase chain reaction - qRT-PCR, multiplex ligation-dependent probe amplification - MLPA) techniques are routinely used for the identification of numerical and structural chromosomal abnormalities, allowing for the detection of several disease subtypes with different prognostic and therapeutic associations [3, 5, 6]. Among these, high-hyperdiploidy is the most common subtype in childhood B-ALL (25-30%), which is associated with favorable prognosis [6–8]. The structural chromosomal abnormalities involve genes that regulate hematopoiesis and lymphoid development (RUNX1, ETV6), activate oncogenes (MYC), or constitutively activate tyrosine kinases (ABL1). In particular, recurring translocations leading to different subtypes in B-ALL include t(12;21)(p13;q22) encoding ETV6::RUNX1 associated with favorable prognosis, t(1;19)(q23;p13) encoding TCF3::PBX1 with intermediate prognosis, t(9;22)(q34;q11) encoding BCR::ABL1, and rearrangements of MLL at 11q23 with different partner genes, both associated with poor prognosis [7, 8]. More recently, genomic profiling has led to the identification of new abnormalities that are not detectable by conventional methods, resulting in more than 20 disease subtypes [9, 10].

The risk classification at diagnosis of patients with ALL varies among different treatment protocols, but in general this includes the patient’s age, white blood cell (WBC) count, and presence of specific disease subtypes. In general, patients are classified as standard-risk if diagnosed at the age 1-5, presenting a WBC count of < 20 x 10^9/L and the absence of iAMP21, IKZF1 deletion or CRLF2 overexpression. The high-risk group includes patients with either poor prednisolone response, hypodiploidy, BCR::ABL1, TCF3::HLF or MLL::AFF1 subtypes, or any other MLL rearrangement in patients younger than 1 year, while all other patients are considered intermediate-risk. Although the identification of disease subtypes conveying prognostic significance along with minimal residual disease (MRD) assessment represent cornerstones for disease stratification, approximately half of the relapses occur in patients from standard-risk groups [11]. Additionally, it has been shown that more than half of the relapse samples have at least one genetic alteration originating either from the leukemic or the pre-leukemic clone, potentially affecting disease progression and therapy response [12,13]. Many of these alterations affect transcription factors (ETV6, PAX5, and IKZF1), epigenetic regulators (ATF7IP, SETD2, KM-T2D, and CREBBP), cell cycle regulators (CDKN2A/B, BTG1, and RB1), RAS pathway genes (KRAS, NRAS, and PTPN11), and the tyrosine kinase FLT3 [14, 15]. Identification of the drivers of treatment failure is crucial for detection of high-risk clones at the time of diagnosis, which can also contribute to uncovering new therapeutic targets, personalization of the treatment protocol and reduction of the short- and long-term adverse effects of intensified chemotherapy.

In this prospective observational national study, we present the clinical variables, identify the most common molecular biomarkers and the individual therapy response (MRD) data, as well as their relation to the clinical status in a cohort of 55 children with B-ALL. Additionally, we conduct a more comprehensive analysis of the patients who experienced disease relapse using whole exome sequencing to detect other alterations that may prove useful in risk stratification and to potentially discover new altered pathways that could be targeted therapeutically.

This study presents data from all pediatric patients diagnosed with B-ALL in our country over a period of six years. All patients were treated according to the ALL-IC-BFM 2002 protocol, which was escalated to high-risk protocol in eight patients. After a median follow-up of 46 months, five patients (9%) experienced disease relapse. In general, the patients with relapse were diagnosed before the age of 6; none presented with CNS infiltration at diagnosis, and the WBC count was slightly higher than 20 x 10^9/L in only one patient. Initial high-risk features (poor prednisolone response and BCR::ABL1 hybrid transcript) were detected in only two of the five patients with relapse. These findings support the need for inclusion of new molecular biomarkers to help identify the high-risk clones at diagnosis and redefine the stratification [6,21].

We identified the presence of alterations in the CREBBP gene in 80% (4 out of 5) of the patients with relapse in our cohort, all of which occurred in the HAT domain of the gene. These alterations were found at the time of initial diagnosis in 2 and at relapse in another 2 patients. Notably, none of the specimens from patients in remission featured alterations in this gene. The CREBBP gene has been recognized as one of the most common relapse-enriched genes in ALL [22–24], and its association with relapse is particularly evident in the high-hyperdiploid subtype [10, 25]. Alterations in this gene affect the response to one of the key components of the treatment protocol, glucocorticoids, leading to a treatment failure [22]. The mechanisms through which CREBBP contribute to glucocorticoid resistance are considered to be associated with its activity as a transcriptional co-activator, which interacts with the glucocorticoid receptor (GR) and potentially modulates its transcriptional activity. When glucocorticoids bind to the GR, the receptor undergoes a conformational change allowing it to interact with co-activators like CREBBP, which in turn can acetylate histones and other chromatin regulators, promoting a more accessible chromatin structure and facilitating gene transcription of target genes involved in processes like apoptosis. Therefore, alterations in this gene can lead to disruption of normal cellular processes including transcriptional regulation, chromatin remodeling and apoptosis. Importantly, mutations in the CREBBP gene also have a therapeutic significance, as it has been found that different CREBBP inhibitors and histone deacetylase inhibitors can alleviate chemotherapy resistance and may become a successful approach for the treatment of relapsed ALL [26].

We also detected alterations in several other genes with potential prognostic value. These include deletions in TP53 in two patients with relapse (in one of them relapsespecific) which were infrequent in the rest of the patients, and deletion in EBF1 gene in one patient with relapse which was absent in the rest of the patients in our cohort. Previous studies have also associated these alterations with disease progression and reduced overall survival rates [3, 5, 22, 27, 28]. Furthermore, NRAS gene mutation was found at relapse only, in one of the patients. Mutations in this gene have been observed in high-risk ALL by others and have been reported as important biomarkers for poor relapse-free survival [23, 29, 30]. However, NRAS mutation was also lost in the relapse clone in another patient from our cohort, indicating its sub-clonal nature and uncertain role in clonal chemoresistance. By contrast, deletions of IKZF1 and CDKN2A/2B genes, individually reported as high-risk markers for disease progression and correlated with poor outcome in several studies [31–34], were not found to independently influence prognosis in our study. Their prognostic significance has been further refined with the detection of the IKZF1^plus profile [20] which, however, was not present in our cohort. Additionally, the occurrence of the IKZF1 deletion in the high-hyperdiploid subtype, which was also associated with an increased relapse risk in a recent large prospective study [10], was detected in one patient in our cohort, who is still in remission after a follow-up of 46 months.

Concerning the evolutionary mechanisms of the clones from diagnosis to relapse, we found that none of the patients with relapse in our cohort experienced expansion of a novel clonal population completely distinct from the population present at diagnosis. Rather, in most of them (4 out of 5), in addition to the same clonal rearrangement and initiating genetic event (CNA, hybrid transcript), we observed novel alterations at relapse that were not detected in the primary clone/s. This either indicates that they were present in minor sub-clones, not detectable with the applied method, survived chemotherapy, and arose as dominant clone/s due to the presence of chemotherapy resistance mutations, or that they were acquired during chemotherapy (treatment-induced) [13, 22]. However, the absence of other alterations in three of these patients suggests that the clonal evolution from an ancestral sub-clone was probably the mechanism for relapse, which has also been described as the most frequent event by others [13, 24, 35]. Only one patient retained all the alterations within the diagnostic and relapse clones, suggesting a linear evolution [35].

The strengths of this study are that it includes an unselected cohort of pediatric patients with B-ALL who were uniformly treated, and that it provides comprehensive data for all patients along with detailed molecular characterization for those with relapse. The limitations of the study include the lack of complete knowledge of the frequency of SNVs at diagnosis in patients without relapse, the insufficient depth of the WES analysis to detect mutations present in minor sub-clones (<10%), and the relatively small patient cohort.

In conclusion, we identified that alterations in the epigenetic regulator CREBBP were the most frequent event in the patients with relapse, either appearing at diagnosis or being acquired at relapse. Screening for alterations in this gene at the beginning, and/or at multiple time-points during chemotherapy, could be incorporated into treatment protocols, as they may contribute to the identification of significant number of patients with predefined or acquired chemoresistant clones. In addition, detection of deletions in the TP53 and EBF1 genes in the CREBBP-negative patients could further help identify patients at increased risk for relapse. Finally, screening for clinically actionable alterations in these and other pathways and genes (RAS, MMR genes), could be of substantial significance for patients with relapse in the coming years by offering a more individualized targeted therapy or immunotherapy therapeutic approach.

Table S1: Type of the single nucleotide variants detected in the patients with relapse.

Table S2: The fragment lengths and Sanger sequencing details (V-D-J genes and CDR3 sequences) of the clonal immunoglobulin heavy chain gene rearrangements present in the matched diagnosis-relapse samples.