Frontiers in Oncology最新文献

Objective: This study aims to systematically evaluate the efficacy of different treatments after non-curative endoscopic resection for superficial esophageal carcinoma.

Methods: Databases including PubMed, the Cochrane Library, Embase, China National Knowledge Infrastructure (CNKI), and Wanfang Data were searched from inception to November 25, 2025. The meta-analysis was performed using Review Manager version 5.3, with Stata version 15 employed for supplementary statistical assessments.

Results: A total of 13 studies involving 1,304 patients were included. The results of the meta-analysis showed that the recurrence rate after non-curative endoscopic resection of superficial esophageal carcinoma was significantly higher in the observation group than in the adjuvant treatment group (OR = 3.25, 95% CI: 1.88-5.62, P < 0.0001). However, there was no significant difference in recurrence rate between the chemoradiotherapy group and the surgery group (OR = 0.65, 95% CI: 0.23-1.84, P = 0.42). Patients with lymphovascular invasion had a higher recurrence rate, which was statistically significant (OR = 4.01, 95% CI: 1.48-10.84, P = 0.006).

Conclusion: Lymphovascular invasion is a risk factor for recurrence in patients. The adjuvant treatment group shows significantly reduced recurrence rates compared to the observation group, with no significant difference in recurrence rates between surgical and chemoradiotherapeutic approaches.

Systematic review registration: https://www.crd.york.ac.uk/prospero/, identifier CRD420251108299.

Background: Colorectal cancer (CRC) is a common and highly lethal malignancy worldwide. Even after curative resection, patients with stage I-III disease remain at substantial risk of recurrence and mortality. The Prognostic Immune and Nutritional Index (PINI) and lymphocyte-to-monocyte ratio (LMR) have been validated as prognostic markers in cancer, yet their individual predictive performance remains limited. We developed a novel Immune-Nutritional Prognostic Ratio (INPR) integrating PINI and LMR to provide a more comprehensive assessment of immune, nutritional, and inflammatory status. This study further evaluated its value in predicting 1-, 3-, and 5-year survival in stage I-III CRC.

Methods: We retrospectively analyzed data from 556 colorectal cancer patients at two hospitals, with one serving as the validation cohort. Receiver operating characteristic (ROC) curves were used to determine optimal cutoff values for PINI and LMR, and the area under the curve (AUC) was applied to assess predictive performance. KKaplan-Meier analysis showed that lower PINI and LMR were associated with shorter overall survival (OS). The INPR, integrating both markers, demonstrated superior accuracy. Variables linked to OS were selected using the Boruta algorithm and multivariable Cox regression, and a nomogram model was developed and validated internally and externally.

Results: The Youden index identified optimal cutoff values of 3.50 for PINI and 2.65 for LMR, with low levels independently predicting shorter OS. The INPR, integrating both, stratified patients into low-, intermediate-, and high-risk groups, with 5-year OS rates of 93.30%, 59.35%, and 28.57% in the training cohort (p<0.001). INPR outperformed either marker alone, showing higher AUC. A nomogram incorporating variables selected by the Boruta algorithm and multivariable Cox regression demonstrated stable and superior prognostic performance in both internal and external validation.

Conclusion: Our findings demonstrate that INPR is a simple, accessible, and effective prognostic tool for postoperative risk stratification in stage I-III CRC patients, providing valuable guidance for optimizing individualized treatment strategies.

[This corrects the article DOI: 10.3389/fonc.2025.1625797.].

Background/objectives: Pancreatic cancer (PC) remains among the most lethal malignancies worldwide, with a persistently low 5-year survival rate despite advances in systemic therapies and surgical innovation. Machine learning (ML) has emerged as a transformative tool for early detection, prognostic modelling, and treatment planning in PC, yet widespread clinical use is constrained by the "black box" nature of many models. Explainable artificial intelligence (XAI) offers a pathway to reconcile model accuracy with clinical trust, enabling transparent, reproducible, and clinically meaningful predictions.

Methods: We reviewed literature from 2020-2025, focusing on ML-based studies in PC that incorporated or discussed XAI techniques. Methods were grouped by model architecture, data modality, and interpretability framework. We synthesized findings to evaluate the technical underpinnings, interpretability outcomes, and clinical relevance of XAI applications.

Results: Across 21 studies on ML in PC, only three studies explicitly integrated XAI, primarily using SHAP and SurvSHAP. These methods helped identify key biomarkers, comorbidities, and survival predictors, while enhancing clinician trust. XAI approaches were categorized by staging (ante-hoc vs. post-hoc), compatibility (model-agnostic vs. model-specific), and scope (local vs. global explanations). Barriers to adoption included methodological instability, limited external validation, weak workflow integration, and lack of standardized evaluation.

Conclusions: XAI has the potential to serve as a cornerstone for advancing transparent, trustworthy ML in PC prediction. By clarifying model reasoning, XAI enhances clinical interpretability and regulatory readiness. This review provides a technical and clinical synthesis of current XAI practices, positioning explainability as essential for translating ML innovations into actionable oncology tools.

RCC represents the predominant form of kidney cancer, with rising global incidence and notable mortality despite advancements in diagnosis and treatment. Traditional imaging and histopathological techniques, while foundational, face limitations in early detection, subtype differentiation, and treatment personalization. This review comprehensively explores RCC's clinical and pathological landscape, then transitions to focus on recent innovations in microfluidics and molecular diagnostics that are reshaping kidney cancer management. Microfluidic platforms facilitate efficient, minimally invasive analysis of biomarkers such as CTCs, ctDNA, and exosomes, enabling real-time disease monitoring and drug response assessment. Biomolecular technologies-including single-cell sequencing, spatial transcriptomics, and next-generation sequencing-offer deeper insights into tumor heterogeneity and therapeutic resistance. The integration of multi-omics data and emerging platforms like kidney cancer-on-a-chip highlight the promise of precision medicine. Challenges in clinical translation, including assay standardization and regulatory hurdles, are also addressed. Together, these developments underscore a paradigm shift toward individualized, biomarker-driven care in RCC.

Endometriomas, commonly known as ovarian chocolate cysts, are a prevalent condition in women of reproductive age. They are cysts formed by the ectopic growth of endometrial tissue within the ovary, often leading to dysmenorrhea and infertility. Dienogest is a first-line medication for the long-term management of endometriomas, control of associated pain, and prevention of postoperative recurrence. A 34-year-old woman presented to our hospital with dysmenorrhea and was diagnosed with an endometrioma. She subsequently commenced dienogest treatment for a total duration of thirty-two months. During this period, the cyst gradually decreased in size and eventually became undetectable. However, a follow-up ultrasound indicated the recurrence of the endometrioma after twenty-seven months. Five months later, a subsequent ultrasound revealed papillary growth with internal blood flow within the ovarian endometrioma. The patient underwent immediate surgical intervention. The postoperative pathology indicated a borderline clear cell tumor. Consequently, the patient promptly underwent comprehensive staging surgery, with the final pathology confirming no residual tumor. This case demonstrates that despite long-term and effective dienogest treatment, endometriomas retain the potential for malignant transformation. Therefore, regular monitoring during treatment and prompt intervention upon suspicion of malignancy are indispensable.

Objective: Ectopic parathyroid gland-induced hypercalcemia is unusual, whereas hypercalcemia from a gastrointestinal stromal tumor (GIST) is extremely rare. This study aims to present a rare case of simultaneous ectopic parathyroid adenoma and GIST, associated with two episodes of hypercalcemia, and to review imaging techniques for ectopic parathyroid localization and the mechanism of hypercalcemia linked to GISTs.

Methods: The clinical manifestations, diagnostic workup, therapeutic interventions, and outcomes of the present case were analyzed. To evaluate advanced imaging modalities, particularly four-dimensional computed tomography (4D-CT) and ¹⁸F-fluorocholine (FCH) PET/CT, for ectopic parathyroid localization, a PubMed search for literature in English from inception to July 2025 was conducted using the terms ("4D-CT" AND "ectopic parathyroid") or ("¹⁸F-fluorocholine PET/CT" AND "ectopic parathyroid"). Additional keywords related to parathyroid imaging, including "FCH-PET/CT", "¹⁸F-fluorocholine PET/CT", and "4D-CT", were incorporated to broaden the search. Reports of GIST-related hypercalcemia were also identified to summarize underlying mechanisms and management approaches.

Results: An 87-year-old man presented with progressive renal dysfunction and hypercalcemic hyperparathyroidism. ^99mTc-sestamibi single photon emission computed tomography/computed tomography (SPECT/CT) identified an ectopic parathyroid lesion in the anterior mediastinum, which was successfully treated with video-assisted thoracoscopic surgery, resolving hypercalcemia. Two years later, recurrent hypercalcemia occurred with reduced parathyroid hormone levels. A CT scan and biopsy revealed a GIST in the pelvis, an extremely rare cause of hypercalcemia. Imatinib normalized calcium and parathyroid hormone levels and induced tumor regression. Nineteen reports showed that 4D-CT or FCH-PET/CT successfully localized ectopic parathyroid lesions after conventional imaging modalities were inconclusive. In 9 cases of GIST-associated hypercalcemia, pathophysiology may involve parathyroid hormone-related protein (PTHrP) or 1-alpha-hydroxylase, with glucocorticoids having a potential role in treatment.

Conclusions: To our knowledge, this case represents the first reported coexistence of an ectopic parathyroid adenoma and a GIST. 4D-CT and FCH-PET/CT can be used as alternative imaging modalities following ^99mTc-sestamibi SPECT/CT to locate ectopic parathyroid lesions. The mechanism behind GIST-related hypercalcemia may involve the expression of PTHrP or 1-alpha-hydroxylase in tumor tissues.

Background: Extra-skeletal osteosarcoma (ESOS) is an uncommon malignant soft tissue tumor, primarily seen in the soft tissues of the extremities or the retroperitoneal region. Primary hepatic osteosarcoma (PHO), a unique subtype, is clinically exceedingly rare. Thus far, only isolated instances have been documented in the literature, with limited high-quality study data accessible. Furthermore, there is no definitive clinical consensus regarding the ideal characterization and management of PHO.

Case presentation: This case report details a 67-year-old male patient hospitalized for one month due to abdominal pain. Upon admission, the patient displayed an increased serum alkaline phosphatase level. Computed tomography (CT) and magnetic resonance imaging (MRI) identified a well-defined lesion in the left hepatic lobe. The patient underwent an open left hemihepatectomy to obtain a conclusive diagnosis. Postoperative histological and molecular pathology assessments verified the tumor as a PHO. The patient underwent transhepatic arterial chemotherapy with epirubicin 30mg/m² and cisplatin 40mg/m², succeeded by 3 cycles of MAP regimen chemotherapy (methotrexate, doxorubicin, cisplatin) in conjunction with sorafenib-targeted therapy, achieving a recurrence-free survival (RFS) of 21 months.

Conclusion: With a median age upon presentation of 61 years, PHO primarily affects men. It is identified radiologically by cystic-solid tumors, sometimes accompanied by calcifications, which facilitates early radiological detection. The literature analysis and our case report point to TP53 mutations and aberrant SATB2 expression as possible genetic markers that could close the diagnostic gap for this uncommon and frequently misdiagnosed illness. Preliminary findings indicate that multimodal therapies-surgery, chemotherapy, and targeted therapy-hold promise for improving patient survival despite PHO's high malignancy and poor prognosis.

Background: Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal malignancy, and traditional prognostic methods, such as TNM staging, often fail to accurately predict outcomes. This review evaluates the use of machine learning (ML) to improve PDAC prognosis.

Methods: A systematic literature search of PubMed and Google Scholar was conducted, identifying 12 studies that applied ML algorithms to predict survival, recurrence, and metastasis in patients with PDAC.

Results: Various algorithms, including Random Forests, XGBoost, and Deep Learning, demonstrated superior predictive performance compared to the TNM staging. Models using multimodal data-combining clinical, radiomic, and genomic features-yielded the highest accuracy for predicting overall survival and early liver metastasis.

Conclusion: ML offers a significant advantage in analyzing complex medical data to refine risk stratification and support personalized PDAC treatment. However, current models are limited by their small datasets and retrospective designs. Future research requires prospective validation to translate these ML tools into clinical practice.

Proton therapy (PT) has the potential to deliver conformal doses to the tumor while sparing normal tissue, but is highly susceptible to treatment uncertainties. The occurrence of anatomical changes during PT treatments has a major impact on the delivered dose, often necessitating plan adaptations that are typically performed offline and require a few days before the adapted plan is ready. In order to react promptly to detected anatomical changes, online adaptive proton therapy (OAPT) has been proposed with the goal of adapting the plan while the patient is on the treatment couch. First OAPT workflows for daily plan adaptation that are effective against interfractional anatomical variations have reached clinical application. However, even faster OAPT workflows are needed to cope with faster anatomical changes. Near real-time adaptive PT (NAPT) relying on online in vivo treatment verification can be a potential solution for many tumor entities (e.g., thoraco-abdominal tumors), which would greatly benefit from the conformality of PT, but are presently challenging to treat with proton beams due to the influence of intrafractional variations. In addition, NAPT offers the opportunity to achieve the long-awaited closed PT feedback loop. In this paper we review the required tasks and necessary components in an OAPT workflow for the application of near real-time adaptation, proceeding sequentially from volumetric imaging for online plan adaptation up to online verification during delivery. Available technology and upcoming developments are discussed. Several aspects regarding regulatory approval, cost-benefit related issues and additional beyond-the-loop tasks are also addressed.