Various imaging techniques combined with machine learning (ML) models have been used to build computer-aided diagnosis (CAD) systems for breast cancer (BC) detection and classification. The rise of deep learning models in recent years, represented by convolutional neural network (CNN) models, has pushed the accuracy of ML-based CAD systems to a new level that is comparable to human experts. Existing studies have explored the usage of a wide spectrum of CNN models for BC detection, and supervised learning has been the mainstream. In this study, we propose a semi-supervised learning framework based on the Vision Transformer (ViT). The ViT is a model that has been validated to outperform CNN models on numerous classification benchmarks but its application in BC detection has been rare. The proposed method offers a custom semi-supervised learning procedure that unifies both supervised and consistency training to enhance the robustness of the model. In addition, the method uses an adaptive token sampling technique that can strategically sample the most significant tokens from the input image, leading to an effective performance gain. We validate our method on two datasets with ultrasound and histopathology images. Results demonstrate that our method can consistently outperform the CNN baselines for both learning tasks. The code repository of the project is available at https://github.com/FeiYee/Breast-area-TWO.

Increasing evidence indicates that Epithelial–mesenchymal transition (EMT) can be regulated by microRNAs (miRNAs). MiR-449a is a liver abundant miRNA. However, the role of miR-449a in the metastasis of hepatocellular carcinoma (HCC) remains largely unknown. The expression levels of miR-449a were first examined in HCC cell lines and tumour tissues by real-time PCR. The in vitro and in vivo functional effect and underlying molecular mechanisms of miR-449a were examined further. In the present study, we found that miR-449a was significantly decreased in HCC cells and tissues, especially in those with the portal vein tumor thrombus. In HCC cell lines, stable overexpression of miR-449a was sufficient to inhibit cell motility in vitro, and pulmonary metastasis in vivo. In addition, ectopic overexpression of miR-449a in HCC cells promoted the expression of epithelial markers and reduced the levels of mesenchymal markers. Further studies revealed that the reintroduction of miR-449a attenuated the downstream signaling of Met, and consequently reduced the accumulation of Snail in cell nucleus by targeting the 3’-untranslated regions (3’-UTR) of FOS and Met. Our data highlight an important role of miR-449a in the molecular etiology of HCC, and implicate the potential application of miR-449a in cancer therapy.

Immune checkpoint inhibitors (ICIs) have recently been approved as part of a neoadjuvant therapy plan for triple‑negative breast cancer (TNBC). Consequently, the concurrent use of radiotherapy (RT) and ICIs is becoming increasingly common in clinical practice. Combining RT with ICIs may increase the risk of radiation pneumonitis. However, evidence remains limited regarding whether concurrent ICIs administered during postoperative RT increase the risk of radiation pneumonitis in patients with TNBC. The aim of the present preliminary, single‑centre, retrospective, hypothesis‑generating study was to evaluate the incidence of radiation pneumonitis in patients with TNBC treated with concurrent ICIs and RT. Patients with TNBC, who were treated at Hubei Cancer Hospital (Wuhan, China) between 2022 and 2024, were retrospectively enrolled. The inclusion criteria were receipt of neoadjuvant chemotherapy with or without ICIs, followed by postoperative adjuvant RT. Patients were divided into two groups: Postoperative RT with concurrent ICIs (ICIs‑RT) and postoperative RT alone (RT‑alone group). Baseline clinical characteristics, treatment responses, RT parameters and follow‑up CT scans for the assessment of radiation pneumonitis were documented. The χ² test and Student's t‑test were used for the statistical analyses. A total of 132 patients with TNBC were enrolled. The pathological complete response rate following neoadjuvant therapy was significantly higher in patients who subsequently received postoperative ICIs‑RT than in those who received RT alone (46.7 vs. 22.5%; P=0.01). Of the 132 patients, 30 received ICIs‑RT and 102 received RT alone. In the ICIs‑RT group, 5 patients received pembrolizumab, 22 received tislelizumab, 1 received camrelizumab and 2 received toripalimab. Follow‑up assessments, including contrast‑enhanced CT scans, were performed every 3 months for 1 year. The incidence of grade 1 radiation pneumonitis was 63.3% (19/30) in the ICIs‑RT group and 46.1% (47/102) in the RT‑alone group, with no statistically significant difference between the groups. No grade 2 or higher radiation pneumonitis occurred in either group. Hypothyroidism was observed in 33.3% (10/30) of patients in the ICIs‑RT group, which was significantly higher than the 11.8% (12/102) observed in the RT‑alone group (P=0.01). The incidence of grade 1 radiation dermatitis was 10.0% (3/30) in the ICIs‑RT group and 9.8% (10/102) in the RT‑alone group, whereas the incidence of grade 2 radiation dermatitis was 90.0% (27/30) and 90.2% (92/102), respectively, with no significant difference between the groups. In patients with TNBC previously treated with neoadjuvant ICIs, the addition of concurrent ICIs to postoperative RT did not significantly increase the incidence of radiation dermatitis. Regarding radiation pneumonitis, no statistically significant increase was observed; however, a clinically meaningful trend toward a higher incidence of grade 1 radiation pneumonitis cannot be excluded. An increased risk of hypothyroidism was observed, likely attributable to immune‑related adverse effects of ICIs. These preliminary findings require validation in larger prospective trials.

The NSABP B-51 trial left unresolved whether postmastectomy radiotherapy benefits ypN0 patients after neoadjuvant chemotherapy. Meanwhile, the PREDICT-DNA study showed that ctDNA clearance after neoadjuvant therapy identifies patients with favorable outcomes irrespective of pathologic response. We propose integrating these datasets: ctDNA status may serve as a tiebreaker for radiotherapy decisions, particularly in patients with residual disease, offering a molecularly guided framework for de-escalation.