Association of 12q21.32 cn-LOH Affecting the KITLG p53 Response Element with Therapy Resistance in Acute Leukemia

The tumor suppressor protein p53 is a central regulator of cellular defense against cancer, earning it the designation “the guardian of the genome.” Its primary function is to serve as a sequence-specific transcription factor [1]. In response to cellular stresses such as DNA damage, oncogene activation, or hypoxia, p53 becomes stabilized and activated. It subsequently binds to specific DNA sequences known as p53 response elements (p53 REs) [2]. These response elements have a consensus sequence of RRRCWWGYYY, followed by 0 to 13 random bases, and then another RRRCWWGYYY sequence (where R = A/G, W = A/T, Y = C/T) [3]. These response elements are typically located within the regulatory regions, such as promoters or enhancers, of target genes. Upon binding to its target response elements, p53 functions as a master transcriptional regulator, orchestrating a complex anti-cancer program by either activating or repressing specific genes.

Owing to its potent anti-proliferative effects, the p53 gene (TP53) is the most frequently mutated gene in human cancers. When p53 is inactivated, these critical protective pathways fail, allowing damaged cells to proliferate uncontrollably. One of the best-characterized p53 response elements in KITLG contains the single-nucleotide polymorphism rs4590952 (A>G), which has been shown to enhance p53 binding, increase KITLG expression, and potentially contribute to oncogenic effects. The KITLG gene (also known as Stem Cell Factor or Steel Factor), which encodes the ligand for the KIT receptor tyrosine kinase, is essential for critical physiological processes such as hematopoiesis, melanogenesis, gametogenesis, and mast cell function. However, this vital signaling pathway exhibits a dual nature, as its dysregulation through overexpression or constitutive activation can be co-opted in oncogenesis, thereby promoting tumor cell proliferation and survival, angiogenesis, and metastasis [4].

Consequently, the tight regulatory control of KITLG expression is paramount, as its loss can lead to significant oncogenic consequences. The contribution of the KITLG polymorphism rs4590952 to cancer risk is complex and appears to be context-dependent. One body of research has identified it as a key functional variant that enhances p53 binding, significantly upregulating KITLG expression and conferring one of the largest inherited risks for certain cancers, notably testicular cancer. This perspective is supported by evolutionary evidence of positive selection, indicating a past beneficial effect, and by the general rarity of such polymorphisms due to negative selection, highlighting their potential deleterious impact [5]. However, this conclusion is not universal. Contrary to the established mechanism, a focused study on breast cancer in a Chinese Han population found no significant association between rs4590952 and disease risk, even when stratified by hormone receptor status [6]. This discrepancy suggests that the oncogenic impact of this polymorphism may be highly cancer-type specific, indicating that heterogeneous mechanisms contribute to the etiology of different cancers and cautioning against broad generalizations of its risk profile.

The SNP rs4590952 is located at chr12:88,559,882 (GRCh38) within an intronic region of the KITLG gene. This polymorphism is characterized by a canonical allelic change of A > G, and its minor allele frequency demonstrates substantial variation across global populations. For instance, the frequency of the ‘A’ allele ranges from as low as 14.5% in a Northern Swedish cohort to over 43% in the PAGE_STUDY and nearly 48% in the TOMMO cohort, highlighting substantial inter-population diversity [7]. Despite its intronic location, this variant is of high functional significance, as it resides within a critical p53 transcription factor binding site, where the A > G change increases p53 binding affinity and modulates KITLG gene expression.

Building on prior data from pilot cohorts of patients with AML, B-ALL, and T-ALL, and leveraging chromosomal microarray analysis (CMA) based on single nucleotide polymorphism (SNP) arrays, this study aimed to investigate the allelic variants of rs4590952 (included in the CytoScan HD CMA panel as marker AX-17068144) within these patient groups. Special attention was paid to chromosomal events involving the KITLG locus. Key prognostic covariates (age, initial leukocyte count, cytogenetic/molecular risk, baseline MRD, and ELN risk in AML) were evaluated in our previous studies, with no association found with therapy response in the intermediate-prognosis groups of ALL and AML [8-11]. To establish the baseline frequency of KITLG RE allelic variants in the Russian population, a control cohort of individuals without a diagnosis of oncopathology who underwent genetic testing for family planning purposes was analyzed. The study aimed to assess the association between rs4590952 allelic variants and specific types of acute leukemia, as well as to investigate their potential relationship with therapy response.

The genotypic distribution within the reference cohort (n = 200) is summarized below. The homozygous A/A genotype was observed in 7 subjects (3.5%), one of whom exhibited a recorded cn-LOH. The heterozygous A/G genotype was found in 72 subjects (36%), and the homozygous G/G genotype was the most common, found in 121 subjects (60.5%). Notably, a cn-LOH spanning the KITLG gene was identified in 53 (43.8%) G/G homozygotes (Figure 2). The corresponding allele frequencies, derived from a total of 400 alleles, were 0.215 for allele A and 0.785 for allele G. The association between cn-LOH and demographic parameters such as gender and age was not revealed (Table 1).

In the T-ALL cohort (n = 38), the genotype distribution was: A/A in 3 patients (7.9%), A/G in 6 (15.7%), and G/G in 27 (71%). Сn-LOH covering the KITLG gene was detected in 17 (63%) of the G/G homozygotes. Additionally, two patients carried a duplication containing the KITLG gene, with allelic variants G/G/G and A/G/G. The overall allele frequency, including these duplications, was 0.166 for the A allele (13/78) and 0.833 for the G allele (65/78).

Genotype analysis of the B-ALL cohort (n = 37) revealed an A/A genotype in 2 patients (5.4%), A/G in 14 (37.8%), and G/G in 21 (56.8%). Furthermore, cn-LOH at the KITLG locus was detected in 10 (47.6%) of the G/G homozygotes. The resultant overall allele frequencies were 0.243 for the A allele and 0.757 for the G allele.

In the AML cohort (n = 35), the genotype distribution was as follows: A/A in 1 patient (2.9%), A/G in 16 (45.7%), and G/G in 18 (51.4%). Cn-LOH spanning the KITLG locus was identified in 9 (50.0%) of the G/G homozygous patients. The overall allele frequency was 0.257 for the A allele (18/70) and 0.743 for the G allele (52/70). Deviation from the Hardy-Weinberg equilibrium was tested in each population. The results, including observed allele frequencies, chi-square (χ²) statistics, and p-values, are presented in Table 2. Notably, significant deviation from HWE was observed only in the T-ALL cohort (p = 0.005), which also demonstrated the highest frequency of cn-LOH events in the 12q21.32 region. This concordance suggests tumor-specific selective pressure favoring homozygosity at this locus in T-cell leukemia.

The distribution of allelic variants is shown in Figure 3. Comparison of allele genotypes between the acute leukemia groups and the reference group using the chi-square test revealed no statistically significant differences.

However, evaluation of the cn-LOH 12q21.32 frequency showed a statistically significant increase of this event in the T-ALL group (OR = 0.4; 95% CI: 0.2–0.9; p = 0.02). Consequently, the identified factor, rather than the allelic variants, was used to assess its association with therapy response.

A significant association between MRD+ status and cn-LOH KITLG was found specifically in the T-ALL group (OR = 11; 95% CI: 2–62; p = 0.005), unlike in B-ALL (Figure 4) or AML groups (Figure 5).

While no significant impact on outcome was observed in ALL cohorts, likely due to subsequent targeted therapy or HSCT according to RALL-2016m protocol, cn-LOH KITLG was significantly associated with poor chemotherapy response in AML (p = 0.01) (Figure 5).

CMA analysis of matched-pair DNA samples from non-tumor tissue revealed cn-LOH at 12q21.32 in 8 out of 9 AML patients with KITLG cn-LOH.

Analysis of 52 healthy individuals (26 with and 26 without cn-LOH KITLG) demonstrated a statistically significant increase in allelic imbalance at heterozygous SNP markers within the 2 Mbp region flanking rs4590952 (±1 Mbp) in the cn-LOH group (Mann-Whitney U test, p = 0.003) (Figure 6A,B and Figure 7).

Figure 6A shows a significant decrease in the number of heterozygous SNP markers with a BAF close to 0.5 and a concurrent significant increase in the number of heterozygous SNP markers with a BAF below 0.3 and above 0.7 compared to the data presented in Figure 6B. This is particularly notable given that a precise, patient-specific analysis of the cn-LOH region was not performed; instead, a common interval of 2 million base pairs centered on rs4590952 was applied to the entire cohort.

This graph provides visual support for the findings. The pink distribution (cn-LOH group) is shifted sharply to the right compared to the blue distribution (normal group). This means individuals in the cn-LOH group have a significantly higher number of SNPs with allelic imbalance. The difference in the range (0–9 vs. 0–33 markers per person) provides concrete, quantitative evidence of allelic imbalance in the cn-LOH group. The Mann-Whitney U test indicated that the difference between the two groups was statistically significant, suggesting that it is unlikely to have occurred by chance. In summary, this plot provides both visual and statistical evidence that the 12q21.32 region in the “pink” group exhibits a genetic profile characterized by high allelic imbalance, consistent with somatic cn-LOH rather than normal inherited homozygosity.

The KITLG gene encoding stem cell factor (SCF) is an evolutionarily significant locus that has undergone a hard selective sweep in human populations over the past millennia, which is associated with adaptations in skin pigmentation and other features [15]. The KITLG locus has undergone strong positive selection throughout human history and also demonstrates remarkable stability, with its adaptive haplotype maintaining a high frequency and persistent selection signal over thousands of years, despite substantial demographic changes [15]. This evolutionary stability underscores the fundamental importance of the gene and explains why its aberrant regulation is frequently observed in pathologies, including malignant neoplasms. The extended homozygosity encompassing the KITLG gene, as identified in our analysis, aligns with established signatures of recent positive selection in Eurasian populations, a pattern previously linked to adaptations such as skin pigmentation and winter climate by Yang et al. This provides a crucial evolutionary backdrop, demonstrating that the KITLG locus is under strong germline selection [16].

The critical importance of the KITLG signaling pathway is normally emphasized by its key role in regeneration processes. As shown in the study of bone regeneration, KITLG, expressed by specialized endothelial cells, is the central organizer of the osteovascular niche, recruiting precursors and directing osteogenic differentiation [17]. This physiological function of KITLG (SCF) signaling as a niche organizer and survival factor for stem/progenitor cells helps explain why aberrant activation of this pathway is often observed in oncologic contexts, where the tumor “captures” these normal support mechanisms.

To investigate the potential clinical significance of a specific p53 response element in the KITLG gene, the study focused on its well-established role as a direct transcriptional target of p53. The genome-wide mapping study by Tebaldi et al. [18] aimed to rank p53 response elements (REs) based on their transactivation potential. Within this research, the gene KITLG was robustly validated as a direct p53 transcriptional target. This conclusion is supported by multiple lines of evidence: the p53 retriever algorithm predicted a high-ranking p53 binding site in the KITLG promoter, independent ChIP-seq data confirmed direct p53 binding, and microarray analysis demonstrated increased KITLG mRNA levels following p53 activation by doxorubicin. Furthermore, subsequent qPCR validation in isogenic cell lines with differing p53 status confirmed that KITLG induction was strictly p53-dependent. The study, therefore, definitively classifies KITLG as a direct p53 target gene, linking p53 pathway activation to the regulation of this key player in cell cycle control and tumor progression. Findings in leukemia are complemented by data from solid tumors, where high KITLG expression has been shown to be a hallmark of type A and AB thymomas, promoting oncogenesis through MAPK pathway activation [19].

This parallel underscore the broader role of KITLG as a potential diagnostic biomarker and therapeutic target across multiple malignancies. The controversial role of KITLG as an oncogene or suppressor is further illustrated by its expression in breast cancer: it is highly expressed and plays an unfavorable role in the HR+ subtype [20], but is expressed at low levels and associated with poor outcomes in TNBC [21]. In colorectal cancer, for example, increased KITLG levels expressed by fibroblasts and tumor endothelium correlate with the activation of infiltrating mast cells and, paradoxically, with improved patient survival, emphasizing its potential protective immunomodulatory role in this context [22]. This contrast illustrates the critical importance of the cellular source of KITLG and the tumor type in determining the final effect of its signal. In addition to its immunomodulatory functions, KITLG can also act as a classical oncogenic driver. In nasopharyngeal carcinoma (NPC), its overexpression in tumor cells directly correlates with lymph node metastasis and poor prognosis, and KITLG suppression inhibits invasion and metastasis, probably through activation of the JAK/STAT pathway [23]. Independent confirmation of the central role of KITLG in NPC pathogenesis was obtained through co-expression network analysis (WGCNA), which identified this gene as a key hub associated with histological grade and tumor stage. Interestingly, this analysis also linked the KITLG-containing module to the p53 pathway and cell cycle regulation [24], consistent with evidence that its expression is regulated via the p53 response element. These findings indicate that tumor cells can autonomously exploit the KITLG-cKIT pathway to enhance their aggressiveness.

Stimulation of tumor cell proliferation through KITLG is often mediated by activation of the MAPK/ERK pathway. In multiple myeloma (MM), this activation occurs through a unique mechanism: the oncogenic splicing factor DAZAP1 directly regulates alternative splicing of KITLG mRNA, ultimately leading to ERK phosphorylation and accelerated tumor growth [25]. The role of alternative splicing as a critical regulator of KITLG’s oncogenic function has been convincingly demonstrated in lung adenocarcinoma. The tumor suppressor, lncRNA SPAT, by inhibiting the SF1 splicing factor, shifts KITLG splicing towards the formation of the less active isoform KITLG-201, while suppressing the production of the oncogenic isoform KITLG-205. Increased levels of KITLG-205 directly correlate with activation of the ERK/MAPK pathway, increased cell migration, and poor patient prognosis [26]. This mechanism highlights that therapeutic strategies targeting the KITLG pathway may include modulating its splicing to suppress specific, most aggressive isoforms. Thus, the KITLG-ERK pathway represents a common endpoint for diverse regulatory aberrations.

The study of Zhang et al. [27] demonstrates a critical interaction between germline genetic variation in the p53 pathway and somatic TP53 mutations, collectively influencing cancer risk, progression, and therapy response.

The oncogenic role of KITLG (SCF) is convincingly confirmed not only in acute leukemias, but also in chronic lymphoproliferative diseases. In chronic lymphocytic leukemia (CLL), leukemic cells predominantly overexpress the SCF membrane isoform. This overexpression is an independent unfavorable prognostic factor correlating with a short time to start therapy and overall survival, and also serves as a key regulator of interactions with the microenvironment. Notably, SCF levels decrease in response to the BTK inhibitor ibrutinib, a standard therapy for CLL, indicating the involvement of this pathway in the drug’s mechanism of action and highlighting KITLG as a potential target for therapeutic intervention in this disease [28]. In AML, single-cell RNA sequencing (scRNA-seq) revealed that low expression of the key stromal factor KITLG in bone marrow aspirates is an independent prognostic marker associated with significantly improved overall survival [29,30]. The role of cn-LOH 12q in the progression of leukemia is confirmed in the work of Sinclair et al. In B-ALL with iAMP21, the cn-LOH 12q event, which results in homozygous mutations in SH2B3, constitutes an unfavorable prognostic factor [31], consistent with findings on the significance of cn-LOH 12q21.32 in AML and T-ALL.

Therefore, strategies aimed at suppressing this pathway, whether by direct inhibition or indirect regulation of its expression, are of therapeutic interest. For example, in glioblastoma, one mechanism underlying the antitumor effect of MSC therapy is the downregulation of the KITLG gene [32]. Precision medicine algorithms are starting to consider KITLG as a predictor biomarker. For example, the DDPP algorithm for predicting the outcome of therapy identified the KIT-KITLG pair as the optimal two-gene signature correlating with the duration of progressive survival (PFS) in patients receiving the antiangiogenic drug axitinib [33]. These findings indicate that the transcriptomic status of the KITLG pathway may serve as a tool for patient stratification and the selection of optimal targeted therapies. Importantly, chemotherapy itself can enhance the pathogenic effects of KITLG. Sublethal doses of doxorubicin induce the release of extracellular vesicles enriched in KITLG from melanoma cells, which directly impair cardiomyocyte mitochondrial function, revealing a novel mechanism of tumor-mediated cardiotoxicity [34].

The universal role of KITLG in stromal cell pathology is also confirmed in a non-oncological context. In age-related osteoporosis, it has been identified as one of the key genes whose expression changes in aging BMSCs within the miRNA-mRNA regulatory network affecting the PI3K-Akt pathway [35]. This indicates that abnormalities in the regulation of KITLG in the stromal niche can lead not only to malignant transformation (leukemia) but also to degenerative processes associated with aging. Similarly, in renal fibrosis associated with diabetic nephropathy, KITLG functions as a key mediator of tissue damage [36]. The diagnostic potential of alterations at the KITLG locus is not limited to genomic variations. Thus, in ovarian seminoma, analysis of KITLG promoter methylation in cfDNA from liquid biopsies (blood plasma) is being explored as a non-invasive biomarker approach [37], illustrating the progression from studying gene function to developing practical diagnostic applications.

The significance of KITLG as a cancer-associated gene is evident even in veterinary medicine: germline copy number variations (CNVs) of this gene are an established predisposition factor for digital squamous cell carcinoma in Schnauzer dogs [38].

Our study, focused on assessing copy number variations (CNV) and cn-LOH at the KITLG locus, employed CMA based on high-resolution SNP microarrays This technology is fundamentally based on the hybridization of fragmented, fluorescently labeled sample DNA to specific oligonucleotide probes immobilized on a chip. Subsequent analysis of signal intensities and genotype calls (B-allele frequency) enables the simultaneous detection of a broad spectrum of genomic alterations, ranging from large copy number variations (CNVs) to subtle copy-neutral events [39-41]. This comprehensive genomic profiling capability was pivotal for our investigation. The SNP array platform offers dual analytical capabilities: first, it enables high-throughput genotyping of the rs4590952 SNP across the entire patient and control cohort. Secondly, and crucially, it enabled a detailed analysis of the molecular genotype across the 12q21.32 locus. This specific feature, the ability to detect regions of homozygosity through shifts in B-allele frequency without corresponding changes in copy number, enabled the unambiguous identification of acquired cn-LOH, a key finding of the study.

A combined analytical approach is essential to distinguish between evolutionarily selected germline ROH (regions of homozygosity) and acquired somatic cn-LOH in the study. Population genetics provides the initial filter: the frequency and genomic architecture of the homozygous segment can be indicative. Long, frequent ROH tracts are often signatures of identity-by-descent due to shared ancestry [42]. However, conclusive evidence for a postzygotic, somatic event comes from the detection of allelic imbalance at heterozygous sites within the region. The presence of heterozygous SNPs with allelic ratios systematically deviating from the expected 0.5, as observed in data from healthy individuals with cn-LOH, is a hallmark of somatic genetic alteration [43,44]. This mosaicism, detectable even in non-malignant cells, confirms the event as acquired rather than inherited. Therefore, integrating population-based allele frequencies with high-resolution analysis of allelic ratios using techniques such as microarrays or next-generation sequencing enables robust discrimination of cn-LOH from ROH [45]. The study of Kim & Suyama [46] provides critical population-level context, revealing that cn-LOH is a frequent and intrinsic genetic phenomenon in healthy individuals, with an average of 40.7 events per genome. A majority (65%) of these events were classified as gonosomal mosaicism, present in both germline and somatic cells. Importantly, their work suggests that the occurrence of cn-LOH is influenced by genomic architecture, with a tendency to increase in GC-rich regions and on chromosomes where homologs are in closer spatial proximity.

This study investigated the nature of the 12q21.32 homozygosity region. Specifically, is it a result of evolutionary selection for a single allele in the population, a classic homozygote inheriting identical alleles from both parents, or is it a somatic event that occurred during early post-zygotic stages? A direct answer to this question would have been provided by genotyping the parents of the individuals included in the study. However, we had neither the opportunity nor the objective to do so.

It was hypothesized that in cases of post-zygotic cn-LOH, which implies mosaicism and the concurrent existence of a cell population retaining the allele, there would be an increase in the number of heterozygous markers exhibiting allelic imbalance. In other words, an increased number of AG markers with B-allele frequencies (BAF) below 0.3 and above 0.7 would be expected. Conversely, in the case of an inherited homozygosity region, the number of heterozygous markers with allelic imbalance would not exceed this same parameter in the group where no loss of heterozygosity was observed. To minimize technical batch effects when comparing the two groups, the B-allele frequency (BAF) of heterozygous SNP markers from the CytoScan™ HT-CMA panel was analyzed exclusively for samples processed within a single CMA batch (a run of 96 samples). Consequently, the cohort was structured to include 26 samples with 12q21.32 loss of heterozygosity (27% of 96) and 26 samples without it. Our study, which identified a high frequency of 12q21.32 cn-LOH, aligns with this ACMG technical standard. Our conclusion that this homozygosity is not due to consanguinity but is a recurrent somatic event (postzygotic cn-LOH) is fully consistent with the interpretation principles outlined by the ACMG. The significant allelic imbalance observed at heterozygous SNP markers within the locus provides direct evidence against inherited ROH and confirms the somatic origin of cn-LOH, in accordance with ACMG guidelines for accurate diagnosis [47].

Mentions of 12q21.32 cn-LOH are scarce in the literature because, in CMA diagnostics, loss-of-heterozygosity events smaller than 3 million base pairs are typically disregarded due to their frequent occurrence in the healthy genome. However, Wen et al. identified a recurrent ROH region at 12q21.3 with a frequency greater than 1% in a cohort of 958 cases with normal karyotype and aCGH results, included in the study over the four years from 2014 to 2017; this cohort consisted of 142 parental, 500 postnatal/pediatric, 195 prenatal CVS, and 121 prenatal AF cases [48]. It is hypothesized that this genomic feature may not have phenotypic consequences in healthy individuals, but in patients with acute leukemias, it could be indirectly associated with poorer responses to chemotherapy.

In conclusion, the data suggest that acquired 12q21.32 cn-LOH is likely a more important determinant of chemotherapy response than the germline allelic variants of the p53 response element in KITLG. The population frequency of the G allele falls within the range established for other populations and shows no association with leukemia. It can be inferred that the rs4590952 G allele, particularly the GG genotype, is not causative but rather serves as a marker for another factor or event within the 12q21.32 cn-LOH region, with the emergence of the homozygous p53 response element genotype occurring in close proximity. However, we are aware of the limitations of our work. Firstly, the small sample size of patients made it difficult to perform statistical analysis and prevented us from conducting a reliable multivariate analysis that included standard prognostic covariates. Secondly, the retrospective nature of the study prevented us from analyzing functional data, such as KITLG expression, KIT pathway activation, and downstream signaling pathways, after identifying aberrations affecting KITLG.

This study is a pilot exploratory investigation aimed at identifying novel potential risk factors. Future validation of 12q21.32 homozygosity/cn-LOH phenomenon in larger, prospective acute leukemia cohorts, with mandatory profiling of matched tumor-free samples, could establish it as a novel biomarker to guide the inclusion of targeted agents into first-line therapy for patients carrying this specific genetic marker.

The population frequency of the rs4590952 G allele (controls 0.785; T-ALL 0.833; B-ALL 0.757; AML 0.743) falls within the range reported for other populations and, in the patient, cohorts studied, shows no significant association with acute leukemia (p > 0.05). The GG genotype frequently arises from acquired cn-LOH at 12q21.32, observed in 45% of T-ALL and 26–27% of other cases (p = 0.02). This pattern is consistent with a recurrent postzygotic event rather than evolutionarily selected germline homozygosity, as indicated by significant allelic imbalance at heterozygous SNPs within the region. Critically, 12q21.32 cn-LOH is associated with MRD positivity in T-ALL and poor chemotherapy response in AML. This treatment-response marker, present irrespective of leukemic status, warrants further validation in expanded cohorts.