Prognostic significance of therapy-induced senescence and SASP dynamics in acute myeloid leukemia: a retrospective cohort study Authors: Bing Ma, Lihong Zhang Published: 2026 ABSTRACT AbstractBackgroundChemotherapy constitutes the backbone of Acute Myeloid Leukemia (AML) treatment but often induces cellular senescence. While senescence initially halts tumor proliferation, the accumulation of senescent cells and the Senescence-Associated Secretory Phenotype (SASP) may promote leukemic survival and relapse. The clinical correlation between post-treatment senescence burden and AML prognosis remains poorly characterized.MethodsWe conducted a single-center retrospective study analyzing 128 newly diagnosed AML patients treated with standard induction chemotherapy between January 2019 and December 2023. Bone marrow samples were assessed at diagnosis and post-induction. Senescent cell burden was quantified via p16INK4a and p21Cip1 expression, alongside Senescence-Associated β-galactosidase (SA-β-gal) activity using flow cytometry. Immune sub-clustering was performed to evaluate senescence in progenitor versus mature populations. SASP factors (IL-6, IL-8, TNF-α) were measured in bone marrow plasma by ELISA. Patients were stratified into High-Senescence and Low-Senescence groups based on median cut-off values.ResultsInduction chemotherapy significantly increased the proportion of senescent cells and normalized SASP levels compared to baseline (P < 0.001), with LSC progenitors exhibiting the highest senescence burden. High post-induction senescence burden was positively correlated with adverse cytogenetics and minimal residual disease (MRD) positivity. Senescence markers strongly correlated with IL-6 and IL-8, but not TNF-α. Patients in the High-Senescence group exhibited significantly inferior Event-Free Survival (median EFS: 8.5 vs. 18.2 months, P = 0.002) and Overall Survival (median OS: 14.2 vs. 26.5 months, P = 0.004). Multivariate Cox regression identified high IL-6 levels (HR 2.14, 95% CI 1.32–3.48) and elevated p16INK4a (HR 1.98, 95% CI 1.15–3.40) as independent predictors of poor prognosis.ConclusionTherapy-induced senescence creates a pro-tumoral microenvironment via SASP in AML. High senescence burden post-induction is a robust biomarker for poor outcomes, suggesting that senolytic strategies targeting these cells could improve therapeutic efficacy.Keywords: Therapy-induced senescence, Senescence-associated secretory phenotype, Acute myeloid leukemia, p16INK4a / SA-β-gal, IL-6 / IL-8, p21 Introduction Acute Myeloid Leukemia (AML) remains a formidable challenge in hematologic oncology, characterized by the clonal expansion of undifferentiated myeloid precursors. Despite advancements in targeted therapies and hematopoietic stem cell transplantation (HSCT), the standard-of-care induction chemotherapy—typically the “7 + 3” regimen involving cytarabine and an anthracycline—remains the primary strategy for achieving remission [1]. While 60–80% of younger patients achieve complete remission (CR), relapse rates remain high, and long-term survival is dismal, particularly in elderly patients or those with adverse genetic risks [2]. The mechanisms underlying chemotherapy resistance and subsequent relapse are multifaceted, involving leukemic stem cell (LSC) quiescence, clonal evolution, and bone marrow microenvironment (BMM) alterations, as highlighted by recent ELN guidelines and emerging therapeutic reviews [2, 3]. Recent research has illuminated the “double-edged sword” nature of Cellular Senescence (CS) in cancer therapy. Classically defined as a state of stable cell cycle arrest, senescence was initially viewed as a desirable outcome of chemotherapy, termed Therapy-Induced Senescence (TIS), preventing the proliferation of malignant cells [4]. However, senescent cells are metabolically active and secrete a complex cocktail of pro-inflammatory cytokines, chemokines, growth factors, and matrix-remodeling proteases, collectively known as the Senescence-Associated Secretory Phenotype (SASP) [5, 6]. While the SASP can recruit immune cells to clear tumors, chronic accumulation of senescent cells and persistent SASP can remodel the BMM, induce epithelial-to-mesenchymal transition (EMT), suppress immune surveillance, and paradoxically stimulate the growth of residual resistant clones [7, 8]. In the context of AML, the bone marrow niche plays a pivotal role. Experimental models have suggested that senescent bone marrow stromal cells or senescent blast cells can protect LSCs from subsequent chemotherapy cycles via paracrine IL-6 and IL-8 signaling [9, 10]. Furthermore, the interaction between LSCs and the aged or senescent microenvironment has been identified as a key driver of therapy failure, with cytotoxic therapy-induced senescence in stromal components contributing to leukemogenesis and niche-mediated resistance mechanisms [11, 12]. Despite compelling preclinical evidence, clinical data correlating TIS markers and SASP levels with patient outcomes in AML are scarce. Most studies rely on in vitro lines or murine models, and the translational relevance of senescence biomarkers in actual clinical cohorts has not been systematically evaluated [13]. Therefore, this retrospective study aimed to quantify the changes in senescent cell burden (p16INK4a, p21Cip1, SA-β-gal) and key SASP factors (IL-6, IL-8, TNF-α) in the bone marrow of AML patients following induction chemotherapy. We hypothesized that a higher burden of residual senescent cells and an inflammatory SASP profile post-induction are associated with inferior survival outcomes, providing a rationale for future senolytic adjuvant therapies. Materials and methods Study design and population This single-center retrospective cohort study was conducted at the Department of Hematology of The First Hospital of Hebei Medical University. We reviewed the medical records of adult patients (≥ 18 years) newly diagnosed with de novo AML between January 2019 and December 2023. The study protocol was approved by the Institutional Ethics Committee and followed the Declaration of Helsinki. The reporting of this study adheres to the STROBE guidelines (Strengthening the Reporting of Observational Studies in Epidemiology) [14]. Inclusion and exclusion criteria Inclusion criteria (1) Morphologically confirmed diagnosis of AML according to WHO 2016 classification; (2) Received standard induction chemotherapy (Idarubicin 12 mg/m² days 1–3 and Cytarabine 100–200 mg/m² days 1–7); (3) Available paired bone marrow samples at diagnosis (Dx) and post-induction (PI, day 28 ± 7); (4) Complete clinical follow-up data. Exclusion criteria (1) Acute Promyelocytic Leukemia (APL); (2) Prior history of myelodysplastic syndrome (MDS) or other malignancies; (3) Active severe infection or uncontrolled autoimmune disease at diagnosis (to avoid confounding cytokine levels); (4) Early death (< 14 days) during induction. Data collection and follow-up Clinical characteristics, including age, gender, FAB classification, cytogenetic risk (ELN 2017 stratification), and initial white blood cell (WBC) count, were extracted from electronic medical records. The primary endpoints were Overall Survival (OS), defined as time from diagnosis to death from any cause, and Event-Free Survival (EFS), defined as time to induction failure, relapse, or death. Follow-up concluded on June 30, 2025. Biomarker analysis Cryopreserved bone marrow mononuclear cells (BMMCs) and bone marrow plasma from the biobank were utilized. Cellular senescence markers BMMCs were thawed and analyzed by multiparametric flow cytometry. Leukemic blasts were initially gated based on their characteristic CD45dim/SSClow properties. Senescent cells were identified as CD34+/CD38- (stem/progenitor enriched) or CD45 + blast populations expressing high levels of p16INK4a (abcam, clone 2D9A12), p21Cip1 (abcam, clone EPR362), and SA-β-gal activity (CellEvent™ Senescence Green Probe, Thermo Fisher) [15]. Additionally, mature immune cell subsets were classified into myeloid (CD33+/CD45high) and lymphoid (CD3+/CD19+) lineages to compare senescence burden across different cellular compartments. Results were expressed as Mean Fluorescence Intensity (MFI) relative to isotype controls. SASP cytokine quantification Bone marrow plasma levels of IL-6, IL-8, and TNF-α were quantified using a high-sensitivity Luminex xMAP multiplex assay (R&D Systems). To accurately reflect the secretory capacity of the residual microenvironment, raw cytokine concentrations (pg/mL) were normalized against the viable bone marrow mononuclear cell count obtained from concurrent flow cytometry and hemocytometer counts, expressed as pg/mL per 106 viable BMMCs. Statistical analysis Continuous variables were tested for normality using the Shapiro-Wilk test and compared using the Mann-Whitney U test or Wilcoxon signed-rank test (for paired samples). Categorical variables were analyzed using the Chi-square or Fisher’s exact test. Correlations were assessed using Spearman’s rank correlation coefficient. Survival curves were estimated using the Kaplan-Meier method and compared via the Log-rank test. Optimal cut-off values for senescence markers were determined using time-dependent ROC curve analysis. Univariate and multivariate Cox proportional hazards models were employed to identify independent prognostic factors. Data analysis was performed using SPSS 26.0 and GraphPad Prism 9.0. P < 0.05 was considered statistically significant. Results Patient characteristics A total of 128 AML patients meeting the criteria were included in the final analysis. The median age was 56 years (range 22–74). The patient selection process is illustrated in Fig. 1. Baseline demographic and clinical characteristics are summarized in Table 1. The cohort was representative of a standard AML population, with 23.4% falling into the Adverse Risk category according to ELN 2017 criteria. Study flow diagram illustrating patient selection and analysis. A total of 165 newly diagnosed AML patients were screened between January 2019 and December 2023. After applying exclusion criteria (APL, prior malignancies, missing paired samples, or early death), 128 patients were included in the final analysis. Patients were subsequently stratified into High-Senescence and Low-Senescence groups based on post-induction biomarker levels Table 1Baseline characteristics of the study cohort (N = 128)CharacteristicValue (n, %) or Median (IQR)Age, years56 (42–65)Gender (Male)72 (56.3%)WBC count at Dx (x109/L)18.5 (4.2–45.6)Bone Marrow Blasts (%)64% (45–82%)ELN 2017 Risk StratificationFavorable38 (29.7%)Intermediate60 (46.9%)Adverse30 (23.4%)Treatment ResponseCR/CRi after Cycle 189 (69.5%)Refractory Disease39 (30.5%) Baseline characteristics of the study cohort (N = 128) Induction chemotherapy increases senescence markers and SASP levels We first analyzed the dynamic changes of senescence markers before and after chemotherapy. Representative flow cytometry plots (Fig. 2A) demonstrate a significant increase in the proportion of p16INK4a/SA-β-gal double-positive cells from diagnosis (median 5.2%) to post-induction (median 48.6%, P < 0.001). In paired samples (N = 128), the expression of p16INK4a in bone marrow cells significantly increased post-induction compared to diagnosis (Median MFI: 1250 vs. 480, P < 0.001). Simultaneously, the expression of p21Cip1, another classical DNA-damage senescence mediator induced by anthracyclines, was prominently upregulated post-treatment (Median MFI: 980 vs. 310, P < 0.001). Similarly, SA-β-gal activity was markedly elevated in the post-treatment samples (Fig. 2B). Dynamic changes of senescence markers and SASP factors post-induction chemotherapy. (A) Representative bivariate flow cytometry dot plots detailing the gating strategy for pre- and post-treatment samples, highlighting the shift of cellular populations into the p16INK4a and SA-β-gal double-positive quadrant (Q2). (B) Paired dot and violin plots showing significant increases in intracellular senescence markers (p16INK4a, p21Cip1, and SA-β-gal MFI) from diagnosis (Pre) to post-induction (Post). (C) Quantification of normalized SASP cytokines (IL-6, IL-8, TNF-α) in bone marrow plasma (expressed as pg/mL per 106 viable cells), demonstrating significant therapy-induced elevation per cell. (D) Bar chart illustrating senescence burden across different cellular sub-populations (CD34+ LSCs/progenitors vs. mature Myeloid and Lymphoid cells), indicating that primitive blasts exhibit the highest propensity for senescence. ***P < 0.001 As expected, the median proportion of leukemic blasts in the bone marrow drastically decreased from 64% at diagnosis to 4.5% post-induction (P < 0.001). While raw plasma concentrations of cytokines showed variable dynamics due to the drastic reduction in total bone marrow cellularity post-treatment, normalization of these cytokines to the residual viable mononuclear cell count revealed a robust and significant upregulation of the SASP secretory profile per cell post-induction: normalized IL-6 (median 45.2 vs. 12.4 pg/mL per 106 cells, P < 0.001) and normalized IL-8 (median 112.5 vs. 34.6 pg/mL per 106 cells, P < 0.001). Normalized TNF-α levels also showed a modest but statistically significant increase (Fig. 2C). Furthermore, we evaluated the senescence burden across different cellular compartments. Post-induction, the progenitor-enriched fraction (CD34+/CD38-) exhibited a significantly higher p16INK4a MFI compared to mature CD45high myeloid and lymphoid cells (Median MFI: 1450 vs. 620 and 410, respectively; P < 0.001) (Fig. 2D), suggesting that primitive leukemic stem/progenitor cells are more prone to entering a senescence-like state following therapy. Notably, correlation analyses revealed that a strong positive correlation was observed between p16INK4a levels and normalized IL-6 concentration (r = 0.68, P < 0.001) as well as IL-8 (r = 0.52, P < 0.001). However, TNF-α showed only a weak, non-significant correlation with p16INK4a expression (r = 0.15, P = 0.12) (Fig. 3), suggesting that the residual senescent cells are a major source of these selected inflammatory cytokines. Correlation between post-induction cellular senescence burden and SASP cytokine levels. Scatter plot analysis demonstrating the correlation between the mean fluorescence intensity (MFI) of p16INK4a and normalized bone marrow plasma SASP factors following induction chemotherapy. (A) A strong positive correlation with IL-6 (r = 0.68, P < 0.001). (B) A moderate positive correlation with IL-8 (r = 0.52, P < 0.001). (C) A weak, non-significant correlation with TNF-α (r = 0.15, P = 0.12). Red solid lines indicate the linear trend High senescence burden is associated with inferior survival outcomes Based on the ROC curve analysis, patients were stratified into “High-Senescence” (n = 64) and “Low-Senescence” (n = 64) groups using the median post-induction p16INK4a MFI value as the cut-off. We then evaluated the clinical impact of this stratification. Kaplan-Meier survival analysis revealed striking differences between the two groups. Patients in the High-Senescence group had a significantly shorter Overall Survival (OS) compared to the Low-Senescence group (Median OS: 14.2 months vs. 26.5 months; Log-rank P = 0.004) (Fig. 4A). Similarly, Event-Free Survival (EFS) was significantly compromised in the High-Senescence group (Median EFS: 8.5 months vs. 18.2 months; Log-rank P = 0.002) (Fig. 4B). Kaplan-Meier survival analysis stratified by senescence burden. (A) Overall Survival (OS) probability for patients in the High-Senescence group (red line) versus the Low-Senescence group (blue line). The High-Senescence group showed significantly inferior OS (Median: 14.2 vs. 26.5 months, Log-rank P = 0.004). (B) Event-Free Survival (EFS) probability showing a similar trend, with significantly shorter EFS in the High-Senescence group (Median: 8.5 vs. 18.2 months, Log-rank P = 0.002) Furthermore, we analyzed the relationship between senescence burden and clinical response. The rate of relapse within 12 months was significantly higher in the High-Senescence group (54.7% vs. 23.4%, P < 0.01). Interestingly, among patients classified as ELN Adverse Risk, those with high SASP levels had an exceedingly poor prognosis, with a 1-year survival rate of only 25%. Multivariate analysis of prognostic factors To determine if senescence burden is an independent prognostic factor, we performed a multivariate Cox proportional hazards regression analysis including age, WBC count, ELN risk, and senescence markers. As shown in Table 2, High post-induction IL-6 (HR 2.14, 95% CI 1.32–3.48, P = 0.002) and High p16INK4a expression (HR 1.98, 95% CI 1.15–3.40, P = 0.014) remained statistically significant predictors of mortality, independent of ELN risk stratification. Table 2Multivariate Cox regression analysis for overall survivalVariableHazard Ratio (HR)95% CIP-valueAge (> 60 vs. ≤60)1.651.10–2.450.015ELN Risk (Adverse vs. Fav/Int)2.851.80–4.52< 0.001Post-Induction p16INK4a (High vs. Low)1.981.15–3.400.014Post-Induction IL-6 (High vs. Low)2.141.32–3.480.002 Multivariate Cox regression analysis for overall survival Discussion In this retrospective study of 128 AML patients, we provide clinical evidence supporting the hypothesis that chemotherapy-induced senescence and the resulting SASP are associated with adverse outcomes. While induction chemotherapy effectively reduces tumor burden, our data suggests it simultaneously elevates p16INK4a and p21Cip1 expression and pro-inflammatory cytokines (IL-6, IL-8) in the bone marrow microenvironment. Crucially, high levels of these markers post-treatment were independent predictors of shorter OS and EFS, challenging the traditional view of senescence as a purely beneficial tumor-suppressive mechanism. This perspective aligns with recent reviews highlighting the complexity of senescence in cancer therapy [16, 17]. Our findings are consistent with emerging preclinical data describing the deleterious effects of the SASP. Duy et al. (2021) demonstrated that chemotherapy induces a “senescence-like” state in AML cells, allowing them to persist and eventually relapse with higher stemness potential [18]. Our novel flow cytometry sub-clustering observation—that CD34+/CD38- progenitor cells retain the highest senescence burden—directly supports this concept in a clinical cohort, indicating that these cells are more prone to entering a senescence-like state following therapy. Similarly, studies in solid tumors and aging models have shown that senescent stromal cells can shield residual cancer cells from immune clearance via PD-L1 upregulation, myeloid-derived suppressor cell (MDSC) recruitment, and general immune exhaustion [19, 20]. Furthermore, Liu et al. recently emphasized the link between autophagy and senescence in maintaining this resistant state [21]. Our observation of a strong correlation between p16INK4a and IL-6 reinforces the concept that the senescent cells may contribute to this inflammatory niche, consistent with the “inflamed” microenvironment theory proposed by Yousefzadeh et al. [20]. The role of SASP factors, particularly IL-6 and IL-8, in AML prognosis has been debated. Korbecki et al. (2023) indicated that IL-8 promotes cell migration and invasion in AML via CXCR2-mediated pathways in the tumor microenvironment [22], while recent studies have highlighted the IL-6/STAT3 signaling axis as a critical promoter of leukemic stem cell survival and chemoresistance [23]. Our study contextualizes these findings within the framework of therapy-induced senescence. By demonstrating that normalized levels of IL-6 and IL-8 are significantly elevated post-induction and correlate with senescence burden, we propose that TIS is a major source of these cytokines. Interestingly, TNF-α did not significantly correlate with p16INK4a. While this could suggest a specific SASP profile, it may also reflect differences in cytokine regulation, kinetics, or cellular sources, representing a limitation in defining a universal SASP signature in this dataset [15]. The clinical implication of our study is significant. If senescent cells act as a reservoir for relapse or a shield for LSCs, then eliminating them could improve outcomes. These findings raise the possibility that senolytic strategies could be explored as a complementary therapeutic approach [17]. Agents like Navitoclax (Bcl-2/Bcl-xL inhibitor) or Dasatinib have shown promise in preclinical models, particularly in targeting senescent stromal cells within the bone marrow microenvironment to disrupt AML blast survival and proliferation [24, 25]. Interestingly, Venetoclax, a Bcl-2 inhibitor widely used in AML [26], may already be exerting partial senolytic effects, as BCL-2 family inhibitors like Navitoclax have demonstrated efficacy in eliminating senescent cells across various cancer models, including hematologic malignancies [25, 27]. However, resistance remains an issue, possibly linked to alternative pathways like those targeted by Lenalidomide. Furthermore, innovative approaches such as Senolytic CAR-T cells are showing potential in reversing senescence-associated pathologies [28]. Our study has limitations. First, as a single-center retrospective analysis, selection bias is unavoidable. Second, we used p16INK4a, p21Cip1, and SA-β-gal as surrogate markers; however, no single marker is perfectly specific for senescence [29]. Future studies should employ multi-marker algorithms or single-cell RNA sequencing for more precise identification. Third, we did not evaluate the specific non-hematopoietic cell types (blasts vs. stromal cells) contributing to the SASP, a distinction that is crucial for targeted therapy, as the SASP from stromal cells (e.g., cancer-associated fibroblasts) often promotes tumor progression, whereas SASP from precancerous epithelial cells may initially suppress tumorigenesis [30]. Despite these limitations, our data provides a compelling rationale for integrating senescence biomarkers into clinical risk stratification. Conclusion In conclusion, this study demonstrates that high levels of senescence markers and SASP factors following induction chemotherapy are strongly associated with poor survival and increased relapse risk in AML. These findings highlight the need to monitor senescence dynamics during treatment and suggest that targeting therapy-induced senescence could be a novel therapeutic avenue to overcome chemoresistance. Acknowledgements Not applicable. Abbreviations Acute Myeloid Leukemia Bone Marrow Microenvironment Bone Marrow Mononuclear Cells Complete Remission Cellular Senescence Event-Free Survival Enzyme-Linked Immunosorbent Assay Epithelial-to-Mesenchymal Transition Hematopoietic Stem Cell Transplantation Leukemic Stem Cell Mean Fluorescence Intensity Overall Survival Senescence-Associated β-galactosidase Senescence-Associated Secretory Phenotype Therapy-Induced Senescence Authors’ contributions BM and LZ designed the study, performed the hematological analysis, interpreted the patient data, and were major contributors in writing the manuscript. BM supervised the project and provided clinical insights. Both authors read and approved the final manuscript. Funding Not applicable. Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study protocol was approved by the Institutional Ethics Committee of The First Hospital of Hebei Medical University. Informed consent was obtained from all individual participants included in the study (or their legal guardians) prior to data and sample collection. The study was conducted in accordance with the Declaration of Helsinki. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Bing Ma and Lihong Zhang contributed equally to this work. References Associated Data Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.