The historical arc traced by the guideline is striking. Clinical trials of anti-cancer drugs began in the 1940s and 1950s, when researchers such as Gilman and Philips used nitrogen mustard to treat lymphoma, an event regarded as the birth of modern chemotherapy. Over the following eight decades, the field evolved from nonrandomized, single-center, retrospective studies into randomized, international, multicenter, prospective trials. The regulatory scaffolding matured alongside: in 1991 the International Conference on Harmonization produced its E6 regulation on good clinical practice, and in 1993 the World Health Organization issued its own guidelines, both of which remain reference standards for multinational trials today. China’s role in this ecosystem has expanded dramatically. Between 2018 and 2022 the country recorded 5,773 investigational new drug applications and 266 new drug applications for innovative oncology therapies, while the number of lead clinical research institutions grew at an average annual rate of 34 percent. Since 2015, reform of the drug review and approval system by the National Medical Products Administration has accelerated approvals, and antineoplastic drugs have become the most heavily invested research area among all therapeutic fields.

Before any new compound can touch a patient, the guideline insists, preclinical evidence must establish biological plausibility of antineoplastic activity, reasonable expected safety, expected patient benefit and a defensible starting dose. Pharmacology studies must outline in vivo and ex vivo effects and mechanisms of action, using accepted test systems and, wherever possible, updated in vivo models. Toxicology programs must characterize the extent, severity and duration of toxic reactions, their dose correlation, reversibility and any species or sex differences, with particular attention to repeated-dose toxicity, animal mortality, pathological findings and local tolerance. Animal pharmacokinetic studies must describe absorption, tissue distribution, metabolism and excretion, and ideally correlate drug exposure with changes in target tissues through PK-PD analyses. Such translational work directly informs dose selection in humans. Biomarker-driven development has already proven its value: several approved anticancer drugs, including EGFR inhibitors developed for resistant non-small-cell lung cancer, were identified through biomarker screening of patient populations, improving trial success rates while sparing unlikely responders from unnecessary risk.

Once human testing begins, the guideline maps the familiar yet intricate staging system. Phase 0 trials, positioned between preclinical work and Phase I, administer subtherapeutic microdoses to first-in-human participants, extending beyond simple pharmacokinetic profiling to front-load information on mechanism of action and target engagement, and requiring ultra-sensitive tools such as positron emission tomography and accelerator mass spectrometry. Phase I trials are the first true human studies, focusing on single and multiple dose escalation, safety, tolerability and pharmacokinetics, with the principal goals of determining the maximum tolerated dose and the recommended Phase II dose. Participants are typically patients with advanced malignancies who have exhausted standard options, an ethical choice that prioritizes those with unmet needs. Escalation designs range from rule-based approaches such as the classic 3+3 scheme to model-based methods like the continuous reassessment method and model-assisted frameworks including mTPI and BOIN, with dose-limiting toxicity serving as the key stopping criterion. Extended cohorts then explore safety and antitumor activity further, supported by rigorous risk management.

Phase II trials divide into exploratory efficacy studies (IIa) and dose-finding studies (IIb), enrolling the target indication population and often employing single-arm designs such as the Simon two-stage method, which limits patient exposure to ineffective drugs, or randomized controlled designs when time-to-event endpoints such as progression-free survival are involved. Objective response rate frequently serves as the primary endpoint, ideally with independent imaging review. Phase III confirmatory trials then establish clinical benefit in large, randomized, often double-blind studies designed to support marketing approval. Overall survival remains the gold-standard primary endpoint, although well-validated surrogates such as progression-free survival, disease-free survival and objective response rate may be accepted by regulators, and multiple-endpoint designs combining survival and progression measures are increasingly common. Phase IV postmarketing studies round out the sequence, monitoring long-term safety, rare adverse reactions, special populations, drug interactions and pharmacoeconomic outcomes under conditions of widespread use.

The guideline also categorizes drugs by mechanism, reflecting the therapeutic revolution of recent decades. Cytotoxic chemotherapy remains a cornerstone but has been joined by novel structures including ruthenium-based agents and antibody-drug conjugates. Endocrine therapy for tumors of endocrine target organs such as breast, prostate and thyroid cancers has entered the targeted-therapy era. Targeted agents exploit molecules specifically and highly expressed on or within tumor cells, blocking growth and metastasis or inducing apoptosis while sparing normal tissue. Immunotherapy, which recruits the patient’s own immune system, demands a distinct evaluation framework because of pseudoprogression, where tumors appear to enlarge before responding; the immune-modified RECIST criteria, iRECIST, allow re-evaluation of apparent progression after at least four weeks, and immune-related adverse events must be systematically captured. Gene therapy, spanning oncolytic viruses and tumor vaccines, presents unique design challenges: maximum tolerated dose is often hard to define, pharmacokinetics are difficult to characterize, and because genetic modifications may persist indefinitely, long-term follow-up of participants is mandatory to detect delayed adverse events. Generic drugs and biosimilars, meanwhile, follow their own evidentiary pathways, with biosimilars requiring multi-level structural characterization, comparative clinical pharmacology and especially vigilant postmarketing surveillance for immunogenicity.

Ethics permeates the entire framework, anchored in good clinical practice and its thirteen ICH E6 principles. Every study must be reviewed and approved by an independent ethics committee before implementation and supervised throughout, guided by instruments such as the Declaration of Helsinki and China’s national quality and ethical review standards. The do-no-harm/benefit principle requires minimizing risk and maximizing benefit; the principle of respect safeguards informed consent, which must convey voluntariness, confidentiality, purpose, procedures, risks and the right to withdraw without prejudice; and the principle of justice demands fair participant selection, with special protection for vulnerable oncology patients against coercion or undue inducement. Placebo controls in cancer trials are confined to settings where no effective therapy exists, with standard treatment otherwise serving as the comparator. Recruitment must respect privacy and voluntariness, and informed consent documents must be written in language participants truly understand, with re-consent required whenever new safety information emerges.

Statistical rigor receives equally detailed treatment. Phase 0 studies typically involve a single cohort of 4 to 12 participants and can conclude within six to eight months; Phase I trials enroll roughly 20 to 40 patients using 3+3, CRM or newer Bayesian interval designs; Phase II studies range from 60 to 300 cases with endpoints analyzed using confidence intervals and Kaplan-Meier methods; and Phase III trials demand hundreds to thousands of participants, with sample sizes calculated from Type I error thresholds of at most 5 percent, Type II error caps of 20 percent, anticipated effect sizes, crossover and dropout rates. Analysis populations follow intention-to-treat, per-protocol and safety conventions, and innovative designs—basket trials testing one targeted therapy across multiple tumor types, umbrella trials testing multiple therapies within a single tumor type, and platform trials permitting treatments to enter or exit via adaptive algorithms—are reshaping the efficiency of development, though they demand sophisticated planning of interim analyses, error allocation and sample size re-estimation.

The guideline closes with warnings that resonate far beyond China. Older adults, who carry the highest cancer risk, appear in only about a quarter of oncology trials, hampered by protocol exclusions, comorbidities and financial barriers, prompting calls to remove age limits and adopt geriatric assessments. Pediatric oncology trials remain scarce despite cancer being a leading cause of death in children, constrained by limited commercial incentives and difficult guardian-consent processes. Efficacy assessment itself continues to evolve: RECIST 1.1 remains the standard for solid tumors, iRECIST addresses immunotherapy’s atypical patterns, the Lugano and LYRIC criteria govern lymphoma, and RANO and iRANO frameworks guide neuro-oncology, allowing continued treatment when imaging progression occurs alongside clinical stability. Adverse event causality, judged through expert, algorithmic or probabilistic methods, is complicated by polypharmacy and overlapping toxicities, and the authors urge standardized baseline collection and patient-reported outcome tools such as PRO-CTCAE to sharpen accuracy. In an era when China’s industry is shifting from generics to innovation and high-throughput sequencing makes human genetic resources central to drug development, the authors argue that rigorous, ethical and adaptive clinical research is not bureaucratic overhead but the very mechanism by which new anticancer medicines earn—and keep—their promise to patients.

Subject of Research: Clinical research methodology and guideline development for antineoplastic drugs

Article Title: Anti-cancer drug clinical research

Article References: Ma, F., Pan, H., Li, J., Zhang, Y., Zhao, H., Xiong, J., Liu, T., Chen, J., Ba, Y., Su, C., Deng, Y., Li, W., Gu, K., Chang, J., Hu, X., Liu, Y., Wang, J., Wang, Z., Wu, J., … Zhou, C. (2026). Anti-cancer drug clinical research. Holistic Integrative Oncology, 5(1), Article 72. https://doi.org/10.1007/s44178-026-00289-2

Image Credits: AI Generated

DOI: 10.1007/s44178-026-00289-2

Keywords: antineoplastic drugs, clinical trials, oncology drug development, phase I trials, dose escalation, GCP, informed consent, RECIST, immunotherapy, biomarkers, basket trials, pharmacovigilance

Multiple myeloma, a malignancy of plasma cells, has long challenged clinicians due to its heterogenous clinical course and unpredictable outcomes. POD24, defined as disease progression within two years post-diagnosis, has been widely recognized as a harbinger of poor prognosis. However, prior investigations have largely relied on traditional survival analyses without delving into the nuanced layers of patient data that machine learning can unravel. This study’s dual approach offers a robust framework to decode complex prognostic patterns that classical analyses might overlook.

The investigative team retrospectively assembled a dataset encompassing clinical information from 155 patients diagnosed with multiple myeloma and stratified them into POD24 and non-POD24 cohorts. Employing Kaplan-Meier survival curves and Cox proportional hazards regression models, they demonstrated a statistically significant reduction in overall survival for patients experiencing POD24, echoing earlier reports but with enhanced confidence due to a rigorous data curation and analysis pipeline.

Pushing beyond conventional statistics, the researchers implemented ten different machine-learning algorithms to gauge their efficacy in predicting overall survival outcomes based on the clinical variables. Among these, the Artificial Neural Network (ANN) emerged as the superior model, showcasing its ability to capture complex nonlinear relationships within the multivariate data. This finding underscores the growing utility of machine learning in oncology prognostication, where intricate biological interplay often defies linear modeling.

Furthering interpretability, the study harnessed Principal Component Analysis (PCA) for dimensionality reduction and visualization. PCA plots clearly delineated class separation between POD24 and non-POD24 groups, affirming that the selected features and model predictions preserved the intrinsic structure of the clinical data. This visual confirmation bolsters confidence in the machine learning model’s discriminative power and highlights the latent patterns distinguishing early progressors from their counterparts.

A hallmark of this research is the application of SHapley Additive exPlanations (SHAP), a game-theory-based method to demystify complex model outputs. SHAP values unequivocally identified POD24 status as the most influential predictive feature driving mortality risk in this patient cohort. This interpretable layer allows clinicians and researchers to understand the weight of POD24 relative to other clinical variables, enhancing trust in model recommendations and facilitating translational adoption.

The study also deployed force plots to visually encapsulate individual patient-level predictions, revealing how non-POD24 status significantly lowers predicted mortality risk. These intuitive visualizations serve as practical tools for personalized risk assessment, potentially guiding more tailored therapeutic strategies and monitoring intensities.

By integrating ANN-based mortality prediction with SHAP-driven interpretability, this work sets a precedent for transparent yet sophisticated prognostic modeling in hematological malignancies. It bridges the gap between black-box AI models and actionable clinical insights, a crucial step for precision medicine advancement.

Moreover, the evidence presented invigorates the notion that POD24 is not merely a temporal milestone but a pivotal biomarker intrinsically linked to disease aggressiveness and patient survival. Recognizing its prognostic strength through dual analytic lenses could inform future clinical trial designs, therapeutic decision-making, and patient counseling.

The implications extend to risk stratification, whereby patients identified as POD24 positive might benefit from intensified treatment regimens, closer surveillance, or novel therapies aimed at mitigating early relapse. As machine learning models mature and integrate larger datasets, personalized medicine in multiple myeloma could reach unprecedented accuracy levels.

The study’s robust methodology—combining retrospective clinical data with advanced algorithmic validation—establishes a paradigm for future research endeavors seeking to meld traditional epidemiological approaches with artificial intelligence frameworks. Such synergy promises enhanced predictive analytics capable of capturing intricacies in disease behavior.

Importantly, the authors emphasize the importance of model transparency, highlighting how explainable AI techniques like SHAP can unravel the decision-making process of complex neural networks. This transparency fosters clinician acceptance and sparks interdisciplinary collaboration between data scientists and healthcare providers.

While the cohort size of 155 patients offers valuable insights, the authors acknowledge the need for validation in larger, multicenter populations to reinforce generalizability. Additionally, integrating molecular and genomic data could further elucidate the biological underpinnings of POD24 and refine predictive accuracy.

This study exemplifies the transformative potential of combining statistical rigor with machine learning ingenuity in oncology research. It charts a promising path toward harnessing big data analytics for practical clinical prognostication, ultimately striving to improve outcomes in patients battling multiple myeloma.

As the field advances, integrating such AI-driven prognostic models into electronic health records and clinical workflows might enable real-time risk assessment, empowering clinicians to enact timely, evidence-based interventions personalized to individual patient risk profiles.

In conclusion, Zhang et al.’s investigation into the prognostic value of POD24 encapsulates a significant leap forward in multiple myeloma research, merging comprehensive statistical analyses with machine learning sophistication. Their findings underscore the vital role of early progression as a mortality predictor and illuminate the path for AI-enhanced oncology precision diagnostics.

Subject of Research: Evaluation of the prognostic significance of progression within 24 months (POD24) for overall survival in multiple myeloma, integrating traditional statistical analyses with machine learning approaches.

Article Title: The prognostic value of POD24 for multiple myeloma: a comprehensive analysis based on traditional statistics and machine learning.

Article References:
Zhang, Q., Wang, Y., Chen, Q. et al. The prognostic value of POD24 for multiple myeloma: a comprehensive analysis based on traditional statistics and machine learning. BMC Cancer 25, 1652 (2025). https://doi.org/10.1186/s12885-025-15089-6

Image Credits: Scienmag.com

DOI: https://doi.org/10.1186/s12885-025-15089-6

The Golgi apparatus, traditionally recognized as the cellular logistics hub responsible for modifying, sorting, and packaging proteins and lipids, has steadily gained attention for its multifaceted role in oncogenesis. While previous investigations hinted at its involvement in tumor progression, the extent to which Golgi apparatus-associated molecular signatures impact gastric cancer development remained elusive—until now.

Employing robust statistical methodologies, including LASSO (least absolute shrinkage and selection operator) and multivariate Cox regression analyses, the research team meticulously curated a gene signature reflective of Golgi function. This seven-gene panel serves as the cornerstone for the GARS. Intriguingly, all these genes are significantly overexpressed in tumor tissues, underscoring their potential roles in driving malignancy or shaping the disease microenvironment.

Validation of GARS across patient cohorts illuminated its remarkable prognostic utility. Patients categorized into the low-risk group by GARS exhibited markedly better overall survival rates compared to their high-risk counterparts. This stratification invites a paradigm shift, enabling clinicians to tailor risk assessment strategies more precisely based on cellular organelle-linked molecular profiles rather than conventional clinicopathological features alone.

Beyond prognosis, GARS demonstrated predictive value in therapeutic contexts. The low-risk cohort showcased enhanced sensitivity not only to frontline chemotherapeutic agents such as 5-fluorouracil and paclitaxel but also to cutting-edge immune checkpoint inhibitors. This dual predictive capability signifies a leap forward toward personalized oncology, where treatment regimens could be optimized by harnessing the molecular characteristics of the Golgi apparatus in tumor cells.

Central to this gene signature is F2R (coagulation factor II receptor), a gene known for its roles in signaling pathways that govern proliferation and migration. Employing targeted gene silencing, the researchers experimentally validated that diminishing F2R expression in gastric cancer cell lines significantly curbed both cellular proliferation and migratory potential. This not only reinforces the biological relevance of F2R within the GARS framework but also highlights it as a promising therapeutic target.

The implications of linking Golgi apparatus features to gastric cancer extend beyond biomarker discovery. The organelle’s involvement in intracellular trafficking and post-translational modifications may influence the tumor’s immune microenvironment, affecting antigen presentation, immune evasion, and response to immunotherapies. By integrating such cellular nuances into prognostic models, this study paves the way for a more nuanced comprehension of tumor-immune interactions.

Moreover, the study’s findings highlight how alterations in Golgi apparatus dynamics might contribute to chemoresistance mechanisms. By correlating GARS scores with chemotherapy sensitivity, the authors suggest that aberrations in protein processing and secretion could modulate drug efficacy, offering a mechanistic foothold to develop novel sensitizing agents or combinatorial therapies.

This research leverages high-throughput genomic data and rigorous bioinformatics pipelines, epitomizing the fusion of computational and experimental cancer biology. The methodological approach underscores the trend of extracting organelle-centric molecular information from bulk tumor analyses, which may revolutionize biomarker development in oncology.

Importantly, the study transcends mere prognostic correlations by anchoring its conclusions in functional experiments. The knockdown of F2R and consequent diminished tumor cell aggression cement the causal link between Golgi apparatus-associated genes and cancer progression. This adds a compelling dimension of translational relevance, as targeting such genes could translate into tangible clinical interventions.

Another striking facet is the potential of GARS to serve as an indicator for immune therapy responsiveness. Given the revolutionizing impact of immunotherapies in cancer treatment, the ability to predict which patients are more likely to benefit is of immense clinical value. The Golgi apparatus’s influence on antigen processing may underlie this predictive relationship, a hypothesis that merits further investigation.

The study also provides an avenue for rethinking gastric cancer heterogeneity. Dissecting tumors through the prism of organelle-specific signatures offers a more granular understanding of tumor biology, which is critical given the notoriously diverse nature of gastric cancer. Stratifying patients based on GARS could refine clinical trial designs and inform personalized medicine strategies.

Critically, the integration of chemotherapy sensitivity data within the GARS framework serves to bridge molecular profiling and real-world therapeutic outcomes. This nexus is essential for transitioning from bench to bedside, as it allows for data-driven clinical decision-making that improves patient survival and quality of life.

While the research opens exciting horizons, it naturally raises questions about the mechanisms through which Golgi apparatus perturbations orchestrate tumor behavior. Future studies may delve into how these seven signature genes influence intracellular pathways, interact with other oncogenic networks, and modulate the tumor milieu, including stromal and immune cell components.

In summary, this study marks a pivotal milestone by positioning the Golgi apparatus—not merely as a cellular organelle—but as a critical determinant of gastric cancer fate. Through the development of GARS and experimental validation of key genes like F2R, the authors provide a compelling framework that merges cellular biology, genomics, and clinical oncology. The translational impact of these findings proposes a future wherein treatment strategies in gastric cancer are finely tuned by the intricacies of subcellular organelle biology, ultimately improving patient prognosis and therapeutic outcomes.

Subject of Research:
The study investigates the role of Golgi apparatus-related gene signatures in predicting the prognosis, chemotherapy sensitivity, and immunotherapy response in gastric cancer.

Article Title:
A Golgi apparatus-related signature predicts the immune microenvironment and prognosis of gastric cancer.

Article References:
Wu, C., Sun, L., Zhu, W. et al. A Golgi apparatus-related signature predicts the immune microenvironment and prognosis of gastric cancer. Genes Immun (2025). https://doi.org/10.1038/s41435-025-00332-8

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41435-025-00332-8

Insulin resistance has long been implicated in various metabolic disorders, but its precise relationship with prostate cancer has remained elusive. This new research, led by Wang, An, and Tao, distinguishes itself by employing innovative non-insulin-based markers to gauge insulin resistance, avoiding confounding variables inherent in direct insulin measurement. The study examines four such indices—ZJU, TyG, TG/HDL-c, and METS-IR—to assess their associations with prostate cancer risk across a substantial cohort.

The investigative team recruited 354 men diagnosed with prostate cancer alongside 1,498 control participants devoid of the disease. Through meticulous inverse probability weighting, they managed to balance baseline discrepancies between groups, enhancing the reliability of their findings. This statistical rigor was complemented by advanced logistic regression models, which methodically tracked the impact of increasing insulin resistance measures on prostate cancer risk.

Results were striking. All four non-insulin-based insulin resistance indices correlated significantly with prostate cancer presence. The TyG index, in particular, demonstrated an adjusted odds ratio exceeding five, suggesting a fivefold increase in prostate cancer risk per unit increase in this marker when other variables were controlled. These elevated odds ratios underscore a robust and possibly causal link—a remarkable advance over previous studies that reported inconclusive or conflicting data.

Importantly, when these indices were partitioned into quintiles, those within the highest quintile displayed risk magnitudes that were sometimes over tenfold greater than the lowest group. For instance, individuals in the highest quintile of the ZJU index had an adjusted odds ratio surpassing 15, illustrating a potent graded relationship between insulin resistance severity and prostate cancer susceptibility.

Visualization of these effects through restricted cubic spline analysis further confirmed the trend—prostate cancer risk rose consistently as the insulin resistance indices increased, painting a clear dose-response pattern that bolsters the argument for a mechanistic link. This pattern was observed uniformly across all four markers, reinforcing the reliability of the observations.

Adding another dimension to their inquiry, the researchers explored interactions between insulin resistance and mean platelet volume, a hematologic measure known to reflect platelet size and activation status. Utilizing generalized additive models, they revealed a statistically significant interaction effect: the combination of higher TG/HDL-c ratios (a surrogate of insulin resistance) and lower MPV levels amplified prostate cancer risk.

This finding suggests that not only do metabolic dysfunction and platelet characteristics independently influence cancer risk, but their interplay may synergistically exacerbate oncogenic processes. Considering platelets’ role in inflammation, vascular function, and tumor microenvironment modulation, these insights open new investigative avenues into how systemic metabolic conditions contribute to tumor development.

The robustness of the findings was further validated through three separate sensitivity analyses, which all confirmed the stability and generalizability of the associations across various model configurations. This comprehensive analytical approach dispels lingering doubts about confounding variables or statistical artifacts driving the observed relationships.

Beyond establishing associations, this study underscores the potential for clinical applications. Non-insulin-based insulin resistance indices are simple to calculate from routine laboratory parameters, offering a cost-effective and accessible tool for identifying men at heightened risk for prostate cancer. Combined with MPV assessments, clinicians might in the future stratify patients with metabolic syndrome components more accurately for prostate cancer screening and preventative strategies.

The authors caution that, as a cross-sectional investigation, causality cannot be definitively inferred. However, the compelling data invite longitudinal studies and mechanistic explorations to confirm whether insulin resistance and platelet-volume interactions actively drive prostate tumorigenesis or reflect broader systemic changes linked to cancer development.

This research also challenges previous paradigms that discounted the metabolic system’s influence on specific cancers, urging oncologists and endocrinologists alike to consider shared pathways. Insulin resistance affects myriad biological processes—glucose metabolism, lipid homeostasis, inflammation—that are increasingly recognized as integral to the cancer ecosystem.

Future research inspired by these findings could explore therapeutic modulation of insulin resistance or platelet activation as adjunct strategies in prostate cancer management. Targeting these factors might reduce not only metabolic morbidity but also oncologic risk, aligning preventative medicine with cancer control.

Moreover, this study adds to the growing narrative that cancers should be contextualized within systemic physiological states rather than viewed in isolation. It challenges researchers to integrate metabolic health parameters into cancer risk assessments and to reconsider the multifactorial etiology of urologic malignancies.

In summary, the work by Wang and colleagues marks a significant leap forward in unraveling the metabolic underpinnings of prostate cancer. By harnessing non-insulin-based indices and linking them innovatively with platelet metrics, it presents a nuanced portrait of cancer risk shaped by interconnected biological domains.

As the medical community digests these revelations, the hope is that improved risk stratification and earlier detection strategies will emerge, ultimately reducing prostate cancer incidence and mortality worldwide. Such progress underscores the power of interdisciplinary research bridging endocrinology, hematology, and oncology.

This study invests in the promise of personalized medicine, where metabolic profiles could guide screening intensity and preventive tactics in men predisposed to prostate cancer. It also sets the stage for exploring novel biomarkers that reflect complex physiological interactions rather than single isolated parameters.

In a time where cancer burden continues to rise globally, unraveling subtle systemic contributors offers a beacon of hope. If insulin resistance and platelet characteristics define new frontiers in cancer risk, their modulation may unlock transformative advances in public health.

Ultimately, this comprehensive examination offers more than data—it provides a conceptual framework inviting continued exploration of metabolism-cancer interrelations, fueling innovative clinical approaches that extend beyond traditional boundaries.

Subject of Research:
Association between non-insulin-based insulin resistance indices, mean platelet volume, and prostate cancer risk

Article Title:
Association of non-insulin-based insulin resistance indices, mean platelet volume and prostate cancer: a cross-sectional study

Article References:
Wang, J., An, H. & Tao, N. Association of non-insulin-based insulin resistance indices, mean platelet volume and prostate cancer: a cross-sectional study. BMC Cancer 25, 795 (2025). https://doi.org/10.1186/s12885-025-13839-0

Image Credits:
Scienmag.com

DOI:
https://doi.org/10.1186/s12885-025-13839-0