Ronald Cheong Kin Chan

Introduction: The presence of non‐progressing ductal carcinoma in situ (DCIS) cases has raised concerns of overtreatment, necessitating comprehensive risk prediction. The tumour microenvironment, especially cancer‐associated fibroblasts (CAFs), plays a key role in DCIS progression. Fibroblast activation protein alpha (FAP), a CAF biomarker, has shown prognostic potential, but its clinical significance and spatial organization in DCIS remain rarely discussed. Methods: In this study, we employed a digital pathology pipeline to quantify stromal FAP expression and map its spatial organization. A total of 102 breast excision specimens were analysed, including pure DCIS and DCIS with co‐existing invasive carcinoma (either nearby or distant). FAP expression was quantified and correlated with clinicopathological features and survival outcomes. Results: Our results showed that FAP expression was significantly higher in DCIS with nearby invasion compared to DCIS with distant invasion ( P = 0.014) and pure DCIS ( P = 0.006), reaching levels comparable to the invasive component itself. Within these high‐FAP regions, no expression gradient from the invasive front was observed. In multivariable analysis, high FAP expression ( P = 0.053) and nuclear grade ( P = 0.043) were associated with invasion in older patients (≥50 years). Additionally, high FAP expression correlated with high nuclear grade ( P = 0.008), papillary DCIS ( P = 0.029) and HER2‐enriched subtype ( P = 0.0497). Although FAP was not an independent predictor of overall survival, it showed a trend towards poorer survival when combined with invasion. Conclusions: In conclusion, FAP is a promising biomarker for predicting nearby invasive carcinoma. Its integration into the existing risk stratification models could enhance clinically meaningful prediction.

Background: There were limited data on zinc alpha-2-glycoprotein (ZAG) levels and their association with peritoneal characteristics and clinical outcomes in patients undergoing peritoneal dialysis (PD). We aimed to quantify and explore the correlation between ZAG levels in PD effluent (PDE), adipose, and serum samples from PD patients, and to evaluate their association with peritoneal characteristics and outcome.Methods: Incident adult PD patients were enrolled in this prospective cohort study. We measured ZAG levels in PDE, serum, and pre-peritoneal fat tissue samples. Peritoneal characteristics were determined by the standard peritoneal equilibrium test. Primary and secondary outcomes were the 2-year peritonitis rate and peritonitis-related clinical outcomes, respectively.Results: We analyzed 126 patients. At baseline, PDE, adipose, and circulating ZAG levels were 1.0 ± 1.2 μg/mL, 15.3 ± 140.5-fold, and 74.4 ± 21.5 μg/mL, respectively. PDE ZAG correlated with adipose ZAG (r = 0.21, p = 0.02), but not with serum ZAG (p = 0.30). PDE ZAG independently predicted small solute peritoneal transport, i.e., dialysate-to-plasma ratio of creatinine at 4 hours (p = 0.01) and mass transfer area coefficient of creatinine (p = 0.04), but not peritoneal protein clearance (p = 0.90). After 222.7 patient-years of follow-up, 40 patients (31.7%) developed 66 peritonitis episodes, and nine patients died. PDE ZAG predicted peritonitis (unstandardized B = 0.446, p < 0.001) and peritonitis-free survival (hazard ratio, 1.36; p = 0.02) after adjustment for confounders.Conclusion: PDE ZAG level significantly correlated with peritoneal transport characteristics and was associated with peritonitis and peritonitis-free survival in PD patients.

Accurate histopathological diagnosis often requires multiple differently stained tissue sections, a process that is time-consuming, labor-intensive, and environmentally taxing due to the use of multiple chemical stains. Recently, virtual staining has emerged as a promising alternative that is faster, tissue-conserving, and environmentally friendly. However, existing virtual staining methods face significant challenges in clinical applications, primarily due to their reliance on well-aligned paired data. Obtaining such data is inherently difficult because chemical staining processes can distort tissue structures, and a single tissue section cannot undergo multiple staining procedures without damage or loss of information. As a result, most available virtual staining datasets are either unpaired or roughly paired, making it difficult for existing methods to achieve accurate pixel-level supervision. To address this challenge, we propose a robust virtual staining framework featuring cascaded registration mechanisms to resolve spatial mismatches between generated outputs and their corresponding ground truth. Experimental results demonstrate that our method significantly outperforms state-of-the-art models across five datasets, achieving an average improvement of 3.2% on internal datasets and 10.1% on external datasets. Moreover, in datasets with substantial misalignment, our approach achieves a remarkable 23.8% improvement in peak signal-to-noise ratio compared to baseline models. The exceptional robustness of the proposed method across diverse datasets simplifies the data acquisition process for virtual staining and offers new insights for advancing its development.

Cervical cancer is a leading malignancy in female reproductive system. While AI-assisted cytology offers a cost-effective and non-invasive screening solution, current systems struggle with generalizability in complex clinical scenarios. To address this issue, we introduced Smart-CCS, a generalizable Cervical Cancer Screening paradigm based on pretraining and adaptation to create robust and generalizable screening systems. To develop and validate Smart-CCS, we first curated a large-scale, multi-center dataset named CCS-127K, which comprises a total of 127,471 cervical cytology whole-slide images collected from 48 medical centers. By leveraging large-scale self-supervised pretraining, our CCS models are equipped with strong generalization capability, potentially generalizing across diverse scenarios. Then, we incorporated test-time adaptation to specifically optimize the trained CCS model for complex clinical settings, which adapts and refines predictions, improving real-world applicability. We conducted large-scale system evaluation among various cohorts. In retrospective cohorts, Smart-CCS achieved an overall area under the curve (AUC) value of 0.965 and sensitivity of 0.913 for cancer screening on 11 internal test datasets. In external testing, system performance maintained high at 0.950 AUC across 6 independent test datasets. In prospective cohorts, our Smart-CCS achieved AUCs of 0.947, 0.924, and 0.986 in three prospective centers, respectively. Moreover, the system demonstrated superior sensitivity in diagnosing cervical cancer, confirming the accuracy of our cancer screening results by using histology findings for validation. Interpretability analysis with cell and slide predictions further indicated that the system's decision-making aligns with clinical practice. Smart-CCS represents a significant advancement in cancer screening across diverse clinical contexts.

Aims: This study evaluates the utility of breast specific markers in identifying breast cancer subtypes within metastatic settings. The subtype alteration in metastatic disease and its consequent impact on breast‐specific marker expression is also examined.Materials and methods: GATA‐binding protein 3 (GATA3), mammaglobin (MMG), transcriptional repressor GATA binding 1 (TRSP1) and SRY‐box transcription factor 10 (SOX10) expression were assessed in a large cohort of metastatic breast cancer (MBC) cases and correlated with the characteristics of both MBC and primary breast cancer (PBC).Results: GATA3 was the most sensitive in MBC (83.1%), followed by TRPS1 (77.0%), MMG (58.5%) and SOX10 (7.1%). This trend was consistent in hormonal receptor (HR)+ and HR− MBC. Combining GATA3/TRPS1 yielded the highest detection rates in the overall cohort (90.1%) and HR+ MBC (97.1%), while TRSP1/MMG was most effective in HR− (76.2%) and TN (71.1%) MBC. Marker expression did not correlate with metastatic site, except SOX10 in lung metastases (P = 0.031). Subtype discordance between MBC and PBC occurred in 43 cases (24.4%), with GATA3 expression in HR− MBC significantly linked to subtype discordance (P = 0.005). Conversely, SOX10 expression was significantly associated with subtype concordance in HR− and TNBC (P ≤ 0.003). Despite a higher expression of GATA3 in all HR− cases, TRSP1 outperformed GATA3 in detecting concordant HR− cases (64.0% versus 38.5%). TRPS1 and SOX10 were expressed in more than 50% of concordant TNBC cases.Conclusions: The expression of breast‐specific markers is mainly determined by the PBC subtype. GATA3 retains high sensitivity in HR+ cancers, even after HR loss during metastasis. TRPS1 and SOX10 are identified as valuable markers in TNBC metastasis.

Remarkable strides in computational pathology have been made in the task-agnostic foundation model that advances the performance of a wide array of downstream clinical tasks. Despite the promising performance, there are still several challenges. First, prior works have resorted to either vision-only or vision-captions data, disregarding invaluable pathology reports and gene expression profiles which respectively offer distinct knowledge for versatile clinical applications. Second, the current progress in pathology FMs predominantly concentrates on the patch level, where the restricted context of patch-level pretraining fails to capture whole-slide patterns. Here we curated the largest multimodal dataset consisting of H\&E diagnostic whole slide images and their associated pathology reports and RNA-Seq data, resulting in 26,169 slide-level modality pairs from 10,275 patients across 32 cancer types. To leverage these data for CPath, we propose a novel whole-slide pretraining paradigm which injects multimodal knowledge at the whole-slide context into the pathology FM, called Multimodal Self-TAught PRetraining (mSTAR). The proposed paradigm revolutionizes the workflow of pretraining for CPath, which enables the pathology FM to acquire the whole-slide context. To our knowledge, this is the first attempt to incorporate multimodal knowledge at the slide level for enhancing pathology FMs, expanding the modelling context from unimodal to multimodal knowledge and from patch-level to slide-level. To systematically evaluate the capabilities of mSTAR, extensive experiments including slide-level unimodal and multimodal applications, are conducted across 7 diverse types of tasks on 43 subtasks, resulting in the largest spectrum of downstream tasks. The average performance in various slide-level applications consistently demonstrates significant performance enhancements for mSTAR compared to SOTA FMs.

Breast cancer has reached the highest incidence rate worldwide among all malignancies since 2020. Breast imaging plays a significant role in early diagnosis and intervention to improve the outcome of breast cancer patients. In the past decade, deep learning has shown remarkable progress in breast cancer imaging analysis, holding great promise in interpreting the rich information and complex context of breast imaging modalities. Considering the rapid improvement in deep learning technology and the increasing severity of breast cancer, it is critical to summarize past progress and identify future challenges to be addressed. This paper provides an extensive review of deep learning-based breast cancer imaging research, covering studies on mammograms, ultrasound, magnetic resonance imaging, and digital pathology images over the past decade. The major deep learning methods and applications on imaging-based screening, diagnosis, treatment response prediction, and prognosis are elaborated and discussed. Drawn from the findings of this survey, we present a comprehensive discussion of the challenges and potential avenues for future research in deep learning-based breast cancer imaging.

Background: Vaspin is an adipokine that regulates glucose and lipid metabolism. Plasma vaspin level is increased in chronic kidney disease but decreased in hemodialysis patients. However, plasma vaspin level in peritoneal dialysis (PD) patients, as well as its prognostic role, has not been studied. Methods: We recruited 146 incident PD patients. Their baseline plasma vaspin levels, body anthropometry, the profile of insulin resistance, bioimpedance spectroscopy parameters, dialysis adequacy, and nutritional indices were measured. They were followed for up to 5 years for survival analysis. Results: The average age was 58.4 ± 11.8 years; 96 patients (65.8%) were men, and 90 (61.6%) had diabetes. The median vaspin level was 0.18 ng/dL (interquartile range [IQR] 0.11 to 0.30 ng/dL). Plasma vaspin level did not have a significant correlation with adipose tissue mass or baseline insulin level. However, plasma vaspin level had a modest correlation with the change in insulin resistance, as represented by the HOMA-IR index, in non-diabetic patients (r = -0.358, p = 0.048). Although the plasma vaspin level quartile did not have a significant association with patient survival in the entire cohort, it had a significant interaction with diabetic status (p < 0.001). In nondiabetic patients, plasma vaspin level quartile was an independent predictor of patient survival after adjusting for confounding clinical factors (adjusted hazard ratio 2.038, 95% confidence interval 1.191–3.487, p = 0.009), while the result for diabetic patients was not significant. Conclusions: Plasma vaspin level quartile had a significant association with patient survival in non-diabetic PD patients. Baseline plasma vaspin level also had a modest inverse correlation with the subsequent change in the severity of insulin resistance, but the exact biological role of vaspin deserves further studies.