Transcriptomic remodeling of bone marrow mesenchymal stromal cells in pediatric B-cell acute lymphoblastic leukemia: a four-gene signature

Introduction

Acute lymphoblastic leukemia (ALL), particularly B-cell ALL, is the most common pediatric malignancy and contributes substantially to disease burden and late-effect morbidity worldwide (1,2). Recent gene expression- and genomics-informed frameworks have clarified molecular subtypes and risk stratification in pediatric B-ALL largely from leukemic blast-centric profiles (3-6), underscoring the need to integrate the stromal microenvironment as a complementary dimension of disease biology. Contemporary classifications emphasize the molecular heterogeneity of pediatric B-ALL, and risk stratification increasingly depends on genomic features (3-6). Despite treatment advances, relapse remains a major cause of mortality (7). High-throughput sequencing and multi-omics studies have refined subtype assignments and uncovered diagnostic markers (3-8), yet many baseline risk factors fail to anticipate therapy resistance or early relapse (9). Increasing evidence indicates that leukemic progression is shaped not only by blasts but also by the bone marrow microenvironment, including immune and stromal elements that support survival and drug resistance (10-12). Although contemporary protocols yield 5-year overall survival rates near 80–90%, 15–20% of children relapse or develop resistance, and treatment-related toxicities such as cardiotoxicity, neurocognitive impairment, and secondary malignancies remain substantial (13-16). Current diagnostic frameworks based on morphology, immunophenotyping, and cytogenetics are indispensable but do not fully capture biological heterogeneity or predict early relapse (7). Consequently, there is an unmet need for biomarkers that improve risk stratification and illuminate microenvironmental drivers of persistence.

Most omics-driven biomarker studies in pediatric B-ALL have focused on leukemic blasts and cell-intrinsic signaling, with limited attention to stromal components. Recent pediatric ALL studies have proposed transcriptome-informed molecular classifiers and epitranscriptomic biomarkers (17,18), while microRNA-based biomarker discovery remains active in childhood ALL (19-22). Immune-checkpoint pathways (e.g., CD300a) are increasingly recognized as contributors to immune evasion (23). However, these approaches still struggle to integrate microenvironmental biology into risk models and to explain therapy resistance not captured by blast-centric profiles. Large-scale transcriptomic profiling has begun to identify gene signatures and computational biomarkers in pediatric ALL (7-8,18-24). These studies demonstrate the promise of high-dimensional data for stratification, yet they often prioritize blast-derived signals and are sensitive to cohort size, platform effects, and overfitting. Integrating niche-derived signals may therefore improve model generalizability and clinical interpretability.

Emerging work on the bone marrow microenvironment indicates that stromal and immune compartments modulate leukemic survival, therapy response, and immune escape (10-12). Mesenchymal stromal cells (MSCs) are central organizers of the hematopoietic niche and provide cytokines, extracellular matrix cues, and metabolic support that shape leukemic cell fate. Yet systematic MSC-centered transcriptomic analyses in pediatric B-ALL are scarce, and it remains unclear whether MSCs exhibit reproducible disease-associated states across datasets or species. This gap limits translation of microenvironmental insights into biomarkers or therapeutic targets. From a methodological standpoint, identifying robust MSC signals is challenging because available cohorts are modest, profiling platforms vary, and in vitro expansion can alter MSC phenotypes. These constraints make it difficult to distinguish disease-associated remodeling from culture artifacts and to generalize findings across datasets. Public datasets often lack matched clinical metadata, limiting mechanistic inference.

Here we tested the hypothesis that pediatric B-ALL is accompanied by a distinct MSC transcriptional state. We analyzed GSE101425 (GPL570) MSC microarrays from pediatric B-ALL patients and healthy donors, integrating differential expression, enrichment analysis, random forest feature ranking, consensus clustering, and gene set enrichment analysis (GSEA). We further contextualized key genes in pediatric single-cell atlases and performed an exploratory peripheral blood (PB) quantitative assay to examine systemic expression shifts. This integrative strategy aims to reposition MSCs as a measurable component of pediatric B-ALL biology and to define a compact four-gene signature that complements blast-centric diagnostics. By combining bulk MSC transcriptomics, single-cell context and exploratory PB validation, we provide a resource for future mechanistic and translational studies while explicitly acknowledging limitations related to cohort size and lack of functional perturbation. Because public MSC datasets often lack matched leukemic blast molecular subtype annotations, we did not perform subtype-stratified analyses and instead frame our results as MSC-centric state signatures that complement, rather than replace, established blast-derived subtype frameworks. We present this article in accordance with the TRIPOD reporting checklist (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0040/rc).

Methods

Data acquisition and differentially expressed genes (DEGs) analysis

Gene expression data were obtained from Gene Expression Omnibus (GEO) (GSE101425; GPL570), profiling in vitro-expanded bone marrow-derived MSCs from pediatric B-ALL at diagnosis (day 0, n=35), remission (n=29), relapse (n=6), and healthy donors (n=16). MSC identity was confirmed in the original study by absence of hematopoietic markers and positivity for mesenchymal markers. We used the GEO series matrix (log₂ processed values), mapped probes to gene symbols with GPL570 annotation, and collapsed multiple probes per gene by median expression (CCN4 is represented as WISP1 on GPL570). Differential expression comparing day 0 B-ALL-associated MSCs versus healthy MSCs was performed using limma. DEGs were defined as adjusted P<0.05 and |log_1.5(Fold Change)| >0.59.

Functional enrichment analysis

Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed separately using the clusterProfiler R package for significantly upregulated and downregulated genes. GO enrichment assessed over-representation in Biological Process, Molecular Function, and Cellular Component categories. KEGG analysis identified relevant signaling and metabolic pathways. Enriched terms were considered significant at adjusted P<0.05.

Machine learning for key gene identification

To prioritize DEGs associated with the MSC state, we used a Random Forest model (randomForest R package) for feature ranking with ntree =500 and mtry =3. Feature importance was evaluated by mean decrease in node purity (IncNodePurity), and candidate genes were selected using IncNodePurity >0.4 as a ranking threshold. The dataset was split into 70% training and 30% validation subsets for RF model fitting (25,26). We used Random Forest solely for feature ranking; discrimination performance was evaluated using cross-validated logistic regression rather than the Random Forest training set. RF-based feature importance offers a non-parametric ranking that can accommodate non-linear associations and feature interactions, which is advantageous for high-dimensional transcriptomic profiles; here it is used as a screening step prior to downstream cross-validated discrimination assessment.

Within-dataset discrimination assessment

Within the GSE101425 cohort, we evaluated each key gene's ability to distinguish B-ALL-associated MSCs from healthy donor MSCs using receiver operating characteristic (ROC) analysis (pROC). For the four-gene signature, we used 5-fold stratified cross-validation with logistic regression; areas under the ROC curve (AUCs) were computed from out-of-fold predictions with 95% bootstrap confidence intervals (CIs) (1,000 resamples). Active B-ALL MSCs (day 0 + relapse) were compared with healthy MSCs.

Consensus clustering and GSEA

Unsupervised consensus clustering was performed on expression of the key genes (ConsensusClusterPlus) to define MSC transcriptional states within B-ALL-associated MSC samples. The analysis included 1,000 repetitions and tested up to k=10. The optimal k was determined by consensus matrices and cumulative distribution function plots. UMAP (umap) was used for visualization. GSEA compared biological pathways between MSC states.

Single-cell RNA sequencing (scRNA-seq) data analysis

To provide cellular context for the key genes in the B-ALL microenvironment, publicly available pediatric B-ALL scRNA-seq datasets were queried from TISCH (http://tisch.comp-genomics.org/). Pre-processed and annotated data from GSE132509 and GSE154109 were examined. The TISCH interface was used to visualize gene expression across cell populations. These analyses were descriptive and intended to contextualize microenvironmental expression, not to validate MSC-specific expression.

Exploratory PB real-time quantitative PCR (RT-qPCR) analysis

For exploratory assessment of systemic expression differences, PB samples were collected from 8 pediatric B-ALL patients and 8 age- and sex-matched healthy controls. This PB analysis probed compartment-dependent expression patterns and does not validate MSC-specific expression. PB mononuclear cells were isolated by Percoll density gradient centrifugation. After centrifugation at 1,500 rpm for 5 min, the supernatant was discarded, the pellet was treated with 2 mL red blood cell lysis buffer, and centrifuged again under identical conditions. Total RNA was extracted using the NcmSpin Rapid RNA Extraction Kit and reverse-transcribed into cDNA (New Cell & Molecular Biotech Co., Ltd., Suzhou, China). RT-qPCR was performed using 2x Q3 SYBR qPCR Master Mix (Vazyme, Nanjing, China). Data were analyzed by the 2^−ΔΔCt method. Primers (Table 1) were synthesized by Tsingke Biotechnology (Beijing, China).

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Wuhan Union Hospital, Tongji Medical College, Huazhong University of Science and Technology [No. 2025(0566)]. Written informed consent was obtained from the parents or legal guardians of all participants.

Statistical analysis

All statistical analyses were conducted using R and Python (scikit-learn). RT-qPCR results were analyzed by Student’s t-test or Wilcoxon rank-sum test, as appropriate. Gene-level comparisons in the GEO cohort used the Mann-Whitney U test, and AUC CIs were estimated by bootstrap resampling. Data visualization was performed with ggplot2. Two-tailed P<0.05 was considered significant.

Results

DEGs identified in B-ALL-associated MSCs

The research flowchart is shown in Figure S1. We analyzed GSE101425 bone marrow MSCs from day 0 B-ALL (n=35), remission (n=29), relapse (n=6), and healthy donors (n=16). Expression of canonical MSC markers (CXCL12, NT5E, THY1, ENG, LEPR) were higher than hematopoietic markers (PTPRC, CD14), supporting MSC identity. This supports that the analyzed expression profiles represent stromal MSCs rather than leukemic blasts. After batch correction (Figure S2), differential expression analysis identified 78 DEGs among 17,270 genes using adjusted P<0.05 and |log_1.5(Fold Change)| >0.59, including 32 upregulated and 46 downregulated genes in B-ALL-associated MSCs versus healthy MSCs (Figure 1).

Figure 1 Volcano plot of DEGs. DEGs, differentially expressed genes; NS, not significant.

Functional enrichment analysis of MSC DEGs

GO and KEGG enrichment of the 78 MSC DEGs revealed significant alterations in biological processes, molecular functions, cellular components, and pathways (Figure 2). Among the 32 upregulated DEGs in B-ALL-associated MSCs, enriched GO terms included regulation of cardiac muscle tissue (biological process), G-protein alpha-subunit binding (molecular function), and collagen-containing extracellular matrix (cellular component), with PPAR signaling as the top KEGG pathway. The 46 downregulated DEGs were enriched for antigen processing and presentation (biological process), peptide binding (molecular function), and Golgi apparatus subcompartment (cellular component), with rheumatoid arthritis as the top KEGG pathway.

Figure 2 GO and KEGG enrichment of DEGs. Upregulated DEGs: GO (A) and KEGG (B). Downregulated DEGs: GO (C) and KEGG (D). DEGs, differentially expressed genes; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

Machine learning prioritizes four MSC-associated genes

Random Forest feature ranking prioritized 13 genes (Figure 3). Using IncNodePurity >0.4, we identified four genes: DKK1 and RGS2 (upregulated in B-ALL-associated MSCs) and CCN4/WISP1 and LYZ (downregulated) (Figure S3). In active B-ALL-associated MSCs versus healthy MSCs, log₂FCs were 1.07 (DKK1), 1.04 (RGS2), −1.12 (CCN4/WISP1), and −0.78 (LYZ) (all P<5×10⁻⁴). A 5-fold cross-validated logistic regression model using these four genes achieved an AUC of 0.883 (95% CI: 0.783–0.966), indicating strong within-cohort separation rather than clinical diagnostic performance (Figure 4). Effect sizes and cross-validated ROC are summarized in Figure S4.

Figure 3 Identification of key MSC-associated DEGs using machine learning. DEGs, differentially expressed genes; MSC, mesenchymal stromal cell.

Figure 4 Within-dataset discrimination of the four-gene MSC signature. (A) DKK1, (B) RGS2, (C) CCN4, (D) LYZ. AUC, area under the curve; MSC, mesenchymal stromal cell.

Four-gene signature defines two MSC transcriptional states

We next examined whether the four-gene signature could stratify B-ALL-associated MSC samples into distinct transcriptional states. Consensus clustering identified k=2 as optimal (Figures 5A-5D). UMAP visualization separated MSC samples into two states based on the four-gene signature (Figures 5E-5H).

Figure 5 Determining the optimal number of MSC transcriptional states by consensus clustering. (A) k=2, (B) k=3, (C) k=4, (D) k=5. (E-H) UMAP visualization of clustering across k values. MSC, mesenchymal stromal cell; UMAP, Uniform Manifold Approximation and Projection.

MSC transcriptional states show divergent pathway enrichment

To explore functional differences between the two MSC transcriptional states, we performed GSEA. The analysis showed distinct enrichment patterns between states (Figure 6), indicating divergent biological programs. Notably, enrichment analyses highlighted differences in antigen presentation and major histocompatibility complex (MHC)-related gene programs and immune-associated disease pathways. These results suggest distinct immunoregulatory programs within MSCs, but functional validation is required to define their biological consequences.

Figure 6 GSEA of GO and KEGG gene sets between the two MSC transcriptional states. (A) GO, (B) KEGG. GO, Gene Ontology; GSEA, Gene Set Enrichment Analysis; KEGG, Kyoto Encyclopedia of Genes and Genomes; MSC, mesenchymal stromal cell.

Single-cell data provide cellular context for key genes

To provide cellular context, we examined pediatric B-ALL scRNA-seq data from TISCH. RGS2 expression was primarily detected in B cells and monocyte/macrophage populations, and LYZ was predominantly observed in monocyte/macrophage populations (Figure 7). CCN4 expression was minimal in the immune cell populations analyzed, and DKK1 transcripts were low, suggesting enrichment in stromal compartments rather than immune cells. These observations contextualize the microenvironmental distribution of the signature genes but do not validate MSC-specific expression. Because MSCs are typically rare in scRNA-seq datasets and subject to dropout, low detection of DKK1/CCN4/WISP1 transcripts should be interpreted cautiously and may reflect technical sensitivity rather than true absence; we therefore treat the single-cell results as contextual support rather than direct validation.

Figure 7 Single-cell microenvironment analysis. (A) DKK1, (B) RGS2, (C) LYZ. TPM, transcripts per million.

Exploratory PB RT-qPCR shows compartment-dependent expression

To explore systemic expression patterns, we performed RT-qPCR in PB mononuclear cells from 8 pediatric B-ALL patients and 8 healthy controls. RGS2 and LYZ were significantly downregulated, whereas CCN4 showed a nonsignificant upward trend in PB from B-ALL patients (P=0.053; Figure 8). DKK1 was not reliably detected (high Ct). These PB findings are exploratory and indicate compartment-dependent expression rather than validation of MSC-specific changes.

Figure 8 RT-qPCR analysis in peripheral blood samples. (A) RGS2, (B) LYZ, (C) CCN4. *, P<0.05; ****, P<0.0001; ns, not significant. ALL, acute lymphoblastic leukemia; HC, healthy control; RT-qPCR, real-time quantitative polymerase chain reaction.

Discussion

In this study, we analyzed bone marrow-derived MSC transcriptomes in pediatric B-ALL to delineate microenvironmental remodeling and identify MSC-associated genes. We identified 78 DEGs in B-ALL-associated MSCs compared with healthy donor MSCs. Enrichment analysis suggested alterations in extracellular matrix organization and immune-related gene programs, consistent with microenvironmental remodeling (10-12). These pathway shifts are consistent with MSC remodeling toward an extracellular matrix and niche-supportive state. Upregulated PPAR-related programs may reflect altered stromal metabolism (27-30), while downregulated antigen presentation suggests reduced immune signaling in MSCs. Such changes could reshape niche stiffness, cytokine gradients, and MSC-immune cross-talk, thereby influencing leukemia persistence and treatment response.

Using Random Forest feature ranking, we prioritized four MSC-associated genes (DKK1, RGS2, CCN4/WISP1, LYZ) that separated B-ALL-associated MSCs from healthy MSCs within the discovery cohort. Cross-validated discrimination for active B-ALL-associated MSCs versus healthy MSCs reached AUC 0.883 (95% CI: 0.783–0.966), but this reflects within-cohort separation and should not be interpreted as clinical diagnostic performance without external validation. At the gene level, DKK1 is a Wnt pathway antagonist implicated in tumor progression and niche remodeling; increased DKK1 in leukemia-associated MSCs may favor a leukemia-permissive niche (31-34). RGS2 modulates GPCR and cAMP signaling and can tune stromal responses to inflammatory cues (35,36). CCN4/WISP1 is a matricellular protein that integrates Wnt and ECM signaling; its downregulation in MSCs may indicate altered matrix deposition or adhesion dynamics (37-39). LYZ is low in MSCs but enriched in immune compartments; its reduction in MSCs alongside immune expression in scRNA-seq supports compartment-specific regulation rather than MSC identity. We note that this compact signature may partially reflect compositional shifts, culture-associated stromal activation, or microenvironmental crosstalk rather than a single causal MSC-intrinsic mechanism; accordingly, we temper mechanistic claims and present the signature primarily as a candidate MSC-associated state readout.

The four-gene signature also stratified B-ALL-associated MSCs into two transcriptional states with distinct pathway enrichment. Importantly, this MSC-centric stratification does not replace established B-ALL molecular subtypes that are defined from leukemic cells and used in clinical risk stratification (3-6). Future studies integrating leukemic cell subtypes with microenvironmental MSC states may clarify how stromal programs influence disease behavior. Single-cell analyses provided microenvironmental context, showing RGS2 and LYZ expression in immune compartments and low expression of DKK1 and CCN4 in those cells. The PB RT-qPCR results showed compartment-dependent expression patterns (RGS2 and LYZ downregulated, CCN4 showing a nonsignificant upward trend, P=0.053), underscoring that PB measurements do not validate MSC-specific expression.

Mechanistically, we interpret the four-gene set as a compact readout of coordinated stromal remodeling rather than a definitive causal module. DKK1 (a Wnt antagonist) and CCN4/WISP1 (a matricellular ECM-associated CCN family member) point to altered Wnt-ECM signaling and niche architecture, whereas RGS2 suggests rewiring of GPCR/cAMP-responsive stress programs in MSCs. In contrast, LYZ is enriched in immune compartments in single-cell data, which cautions against treating it as an MSC-intrinsic marker and instead supports a compartment-dependent signature component. Accordingly, we propose testable hypotheses (e.g., a DKK1-driven suppression of osteogenic/stromal-support programs coupled with CCN4/WISP1-linked ECM remodeling) that may shape leukemic support and immune crosstalk.

To directly evaluate functional relevance, future work should move beyond state discrimination and perform perturbation experiments in independent MSC cohorts. A practical next step is knockdown/overexpression of DKK1, RGS2, and CCN4/WISP1 in primary MSCs, followed by MSC-leukemia co-culture assays to quantify leukemic cell adhesion, proliferation, apoptosis, and drug sensitivity. These experiments would provide causal evidence for whether the signature genes modulate MSC support capacity. The PB qPCR results further underscore compartment-dependent expression and argue against equating PB signals with MSC expression.

There are several limitations in this study. The discovery cohort is modest and limited to in vitro-expanded MSCs, which may not fully capture in vivo stromal states. Relapse samples are few (n=6), limiting inference about relapse-specific programs. No external human MSC validation cohort or clinical outcome data were available, and the ROC analyses are restricted to within-cohort separation. CCN4 is represented as WISP1 on GPL570, underscoring the need for careful cross-platform gene mapping. The PB RT-qPCR results are exploratory and should not be interpreted as validation of MSC-specific changes. Functional assays will be required to define the mechanistic roles of these genes in the B-ALL microenvironment. In addition, the MSC dataset lacks matched patient-level molecular subtype metadata, precluding subtype-aware stratification in the present analysis.

Conclusions

Collectively, this study delineates transcriptional remodeling of bone marrow MSCs in pediatric B-ALL and nominates a four-gene MSC-associated signature. The work is hypothesis-generating and provides a foundation for future studies integrating independent MSC cohorts, leukemic cell subtypes, and functional validation.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0040/rc

Data Sharing Statement: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0040/dss

Peer Review File: Available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0040/prf

Funding: This work was supported by National Natural Science Foundation of China (Nos. 82374255, 82000175) and Hubei Pediatric Medical Alliance (No. HPAMRP202101).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tp.amegroups.com/article/view/10.21037/tp-2026-1-0040/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of Wuhan Union Hospital, Tongji Medical College, Huazhong University of Science and Technology [No. 2025(0566)]. Written informed consent was obtained from the parents or legal guardians of all participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Liu W, Fang J, Zhu M, et al. Global, regional, and National burden of childhood leukemia from 1990 to 2021. BMC Pediatr 2025;25:571. [Crossref] [PubMed]
2. Inaba H, Mullighan CG. Pediatric acute lymphoblastic leukemia. Haematologica 2020;105:2524-39. [Crossref] [PubMed]
3. Arber DA, Orazi A, Hasserjian RP, et al. International Consensus Classification of Myeloid Neoplasms and Acute Leukemias: integrating morphologic, clinical, and genomic data. Blood 2022;140:1200-28. [Crossref] [PubMed]
4. Alaggio R, Amador C, Anagnostopoulos I, et al. The 5th edition of the World Health Organization Classification of Haematolymphoid Tumours: Lymphoid Neoplasms. Leukemia. 2022;36:1720-48.
5. Gu Z, Churchman ML, Roberts KG, et al. Genomic analyses define distinct molecular subtypes of pediatric B-ALL. Nature genetics 2019;
6. Beder T, Abadi U, Lo Nigro L, et al. Molecular subtypes of B-ALL and clinical risk stratification. HemaSphere 2023;
7. Vllahu M, Savarese M, Cantiello I, et al. Application of Omics Analyses in Pediatric B-Cell Acute Lymphoblastic Leukemia. Biomedicines 2025;13:424. [Crossref] [PubMed]
8. Hu Z, Kovach AE, Yellapantula V, et al. Transcriptome Sequencing Allows Comprehensive Genomic Characterization of Pediatric B-Acute Lymphoblastic Leukemia in an Academic Clinical Laboratory. J Mol Diagn 2024;26:49-60. [Crossref] [PubMed]
9. Schwab C, Cranston RE, Ryan SL, et al. Integrative genomic analysis of childhood acute lymphoblastic leukaemia lacking a genetic biomarker in the UKALL2003 clinical trial. Leukemia 2023;37:529-38. [Crossref] [PubMed]
10. Ferrao Blanco MN, Kazybay B, Belderbos M, et al. Distinct stromal cell populations define the B-cell acute lymphoblastic leukemia microenvironment. Leukemia 2025;39:2622-39. [Crossref] [PubMed]
11. Song L, Zhang R, Pan L, et al. GREM1 deficiency induced bone marrow adipose niche promotes B-cell acute lymphoblastic leukemia disease progression. Int J Cancer 2025;157:559-72. [Crossref] [PubMed]
12. Ángel Cortés S, Rojas Zambrano PM, Vernot JP. Favorable and poor prognosis B-cell precursor acute lymphoblastic leukemia subtypes reveal distinct leukemic cell properties when interacting with mesenchymal stem cells, differentially modifying their cell stemness and leukemia chemoresistance. J Cell Commun Signal 2025;19:e70009. [Crossref] [PubMed]
13. Jelić M, Pavlović M, Mucavac L, et al. Causes of Death in Childhood Acute Lymphoblastic Leukemia: A Single-Center Experience. Medicina (Kaunas) 2025;61:1193. [Crossref] [PubMed]
14. Zhou O, Harila A, Hovén E, et al. Educational outcomes school year nine in children treated for acute lymphoblastic leukemia: A nationwide registry-based study from Sweden. Int J Cancer 2025;156:1347-57. [Crossref] [PubMed]
15. Roganovic J, Haupt R, Bárdi E, et al. Late Adverse Effects after Treatment for Childhood Acute Leukemia. Acta Med Acad 2024;53:59-80. [Crossref] [PubMed]
16. Al-Mahayri ZN, AlAhmad MM, Ali BR. Long-Term Effects of Pediatric Acute Lymphoblastic Leukemia Chemotherapy: Can Recent Findings Inform Old Strategies? Front Oncol 2021;11:710163. [Crossref] [PubMed]
17. Saluja S, Ganguly S, Singh J, et al. Aberrant overexpression of m6A writer and reader genes in pediatric B-Cell Acute Lymphoblastic Leukemia (B-ALL). Transl Oncol 2025;56:102403. [Crossref] [PubMed]
18. Hu Z, Jia Z, Liu J, et al. MD-ALL: an integrative platform for molecular diagnosis of B-acute lymphoblastic leukemia. Haematologica 2024;109:1741-54. [Crossref] [PubMed]
19. Rashed WM, Hammad AM, Saad AM, et al. MicroRNA as a diagnostic biomarker in childhood acute lymphoblastic leukemia; systematic review, meta-analysis and recommendations. Crit Rev Oncol Hematol 2019;136:70-8. [Crossref] [PubMed]
20. Rzepiel A, Kutszegi N, Gézsi A, et al. Circulating microRNAs as minimal residual disease biomarkers in childhood acute lymphoblastic leukemia. J Transl Med 2019;17:372. [Crossref] [PubMed]
21. El-Khazragy N, Noshi MA, Abdel-Malak C, et al. miRNA-155 and miRNA-181a as prognostic biomarkers for pediatric acute lymphoblastic leukemia. J Cell Biochem 2019;120:6315-21. [Crossref] [PubMed]
22. Kyriakidis I, Papadovasilakis Z, Papoutsoglou G, et al. AutoML identification of microRNA biomarkers in high-risk pediatric acute lymphoblastic leukemia. Noncoding RNA Res 2025;15:120-31. [Crossref] [PubMed]
23. Peng JM, Liu HY. CD300a: An Innate Immune Checkpoint Shaping Tumor Immunity and Therapeutic Opportunity. Cancers (Basel) 2025;17:1786. [Crossref] [PubMed]
24. Iacobucci I, Zeng AGX, Gao Q, et al. Multipotent lineage potential in B cell acute lymphoblastic leukemia is associated with distinct cellular origins and clinical features. Nat Cancer 2025;6:1242-62. [Crossref] [PubMed]
25. Chen C, Kang D, Chen Z, et al. DLD is a potential therapeutic target for COVID-19 infection in diffuse large B-cell lymphoma patients. Apoptosis 2024;29:1696-708. [Crossref] [PubMed]
26. Ruan S, Wang S, Jiang Y. Machine learning and validation reveal that immune-related genes in systemic lupus erythematosus regulate apoptosis and cycle progression in diffuse large B-cell lymphoma. Front Med (Lausanne) 2025;12:1707913. [Crossref] [PubMed]
27. Zhang S, Tu Y, Lai H, et al. PPARG, GNG12, and CD19 are potential independent predictors of central nerve recurrence in childhood acute lymphoblastic leukemia. Hematology 2023;28:2182169. [Crossref] [PubMed]
28. Esmaeili S, Safaroghli-Azar A, Pourbagheri-Sigaroodi A, et al. Activation of PPARγ intensified the effects of arsenic trioxide in acute promyelocytic leukemia through the suppression of PI3K/Akt pathway: Proposing a novel anticancer effect for pioglitazone. Int J Biochem Cell Biol 2020;122:105739. [Crossref] [PubMed]
29. Mokhtari Y, Yousefi AM, Bashash D. Inhibition of PI3K Signaling Intensified the Antileukemic Effects of Pioglitazone: New Insight into the Application of PPARγ Stimulators in Acute Lymphoblastic Leukemia. Indian J Hematol Blood Transfus 2023;39:546-56. [Crossref] [PubMed]
30. Esmaeili S, Safaroghli-Azar A, Pourbagheri-Sigaroodi A, et al. Stimulation of Peroxisome Proliferator-Activated Receptor-Gamma (PPARγ) using Pioglitazone Decreases the Survival of Acute Promyelocytic Leukemia Cells through Up-Regulation of PTEN Expression. Anticancer Agents Med Chem 2021;21:108-19. [Crossref] [PubMed]
31. Janovská P, Bryja V. Wnt signalling pathways in chronic lymphocytic leukaemia and B-cell lymphomas. Br J Pharmacol 2017;174:4701-15. [Crossref] [PubMed]
32. Jiang H, Zhang Z, Yu Y, et al. Drug Discovery of DKK1 Inhibitors. Front Pharmacol 2022;13:847387. [Crossref] [PubMed]
33. Niu J, Li XM, Wang X, et al. DKK1 inhibits breast cancer cell migration and invasion through suppression of β-catenin/MMP7 signaling pathway. Cancer Cell Int 2019;19:168. [Crossref] [PubMed]
34. Fezza M, Moussa M, Aoun R, et al. DKK1 promotes hepatocellular carcinoma inflammation, migration and invasion: Implication of TGF-β1. PLoS One 2019;14:e0223252. [Crossref] [PubMed]
35. Cho J, Min HY, Lee HJ, et al. RGS2-mediated translational control mediates cancer cell dormancy and tumor relapse. J Clin Invest 2021;131:e136779. [Crossref] [PubMed]
36. Joshi JC, Joshi B, Zhang C, et al. RGS2 is an innate immune checkpoint for suppressing Gαq-mediated IFNγ generation and lung injury. iScience 2025;28:111878. [Crossref] [PubMed]
37. Yeger H. CCN proteins: opportunities for clinical studies-a personal perspective. J Cell Commun Signal 2023;17:333-52. [Crossref] [PubMed]
38. Maiese K. Wnt Signaling and WISP1 (CCN4): Critical Components in Neurovascular Disease, Blood Brain Barrier Regulation, and Cerebral Hemorrhage. Curr Neurovasc Res 2022;19:379-82. [Crossref] [PubMed]
39. Singh K, Oladipupo SS. An overview of CCN4 (WISP1) role in human diseases. J Transl Med 2024;22:601. [Crossref] [PubMed]