IL-27 Gene Therapy Ameliorates IPEX Syndrome Caused by Germline Mutation of Foxp3 Gene: A Major Role for Induction of IL-10.

Inactivating mutations of Foxp3, the master regulator of regulatory T cell development and function, lead to immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX) syndrome in mice and humans. IPEX is a fatal autoimmune disease, with allogeneic stem cell transplant being the only available therapy. In this study, we report that a single dose of adeno-associated virus (AAV)-IL-27 to young mice with naturally occurring Foxp3 mutation (Scurfy mice) substantially ameliorates clinical symptoms, including growth retardation and early fatality. Correspondingly, AAV-IL-27 gene therapy significantly prevented naive T cell activation, as manifested by downregulation of CD62L and upregulation of CD44, and immunopathology typical of IPEX. Because IL-27 is known to induce IL-10, a key effector molecule of regulatory T cells, we evaluated the contribution of IL-10 induction by crossing IL-10-null allele to Scurfy mice. Although IL-10 deficiency does not affect the survival of Scurfy mice, it largely abrogated the therapeutic effect of AAV-IL-27. Our study revealed a major role for IL-10 in AAV-IL-27 gene therapy and demonstrated that IPEX is amenable to gene therapy.

Intraoperative Frozen Section Evaluation of Pancreatic Specimens and Related Liver Lesions.

CONTEXT.­: Frozen sections are essential in the surgical management of patients, especially those with pancreatic masses, because frozen sections can provide answers intraoperatively and aid in treatment decisions. Pancreas frozen sections are challenging because of the small tissue size, processing artifacts, neoadjuvant treatment effects, and concurrent pancreatitis-like obstructive changes. The authors present a review of intraoperative evaluation of pancreatic specimens. OBJECTIVES.­: To provide an approach to the diagnosis of pancreatic adenocarcinoma on frozen sections and to discuss commonly encountered pitfalls. Indications for pancreas frozen sections and specific margin evaluation will be discussed. We will also review frozen section diagnosis of subcapsular liver lesions and tumors other than metastases of pancreatic ductal adenocarcinoma. DATA SOURCES.­: Data sources included a literature review and the personal experiences of the authors. CONCLUSIONS.­: The features for diagnosis of pancreatic adenocarcinoma include disordered architecture, glands at abnormal locations, and atypical cytology. It is important to be aware of the pitfalls and clues on frozen section. The evaluation of resection margins can be challenging, and in the setting of the resection of cystic tumors, the key is the diagnosis of high-grade dysplasia or cancer. Finally, it is vital to remember the differential diagnosis for subcapsular liver lesions because not all lesions will be metastases of adenocarcinomas or bile duct adenomas. Frozen sections remain a useful tool for the intraoperative management of patients with pancreatic tumors.

EPM2AIP1 immunohistochemistry is inadequate as a surrogate marker for MLH1 promoter hypermethylation testing in colorectal cancer.

MLH1 promoter hypermethylation (MPH) analysis is an essential step in the universal tumor testing algorithm for Lynch syndrome, the most common inherited predisposition to colorectal cancer (CRC). MPH usually indicates sporadic CRC. EPM2AIP1 gene shares the same promoter as MLH1, therefore MPH should also silence EPM2AIP1 transcription leading to loss of protein expression on immunohistochemistry (IHC). It has been previously reported that EPM2AIP1 IHC can be used as a surrogate for MPH in endometrial cancer. Our goal was to evaluate the feasibility of EPM2AIP1 IHC as a surrogate for MPH in CRC. 101 microsatellite instable CRC cases were selected, including 19 cases from whole tumor sections and 82 cases from tissue microarrays. 74 cases were with MPH and 27 without MPH. All 74 cases with MPH showed absent MLH1 by IHC, but only 47 (64%) exhibited loss of expression of EPM2AIP1. Of the 27 cases without MPH, 9 (33%) cases had unexpected loss of EPM2AIP1 expression. Of note, 10 cases were MLH1-mutated Lynch syndrome without MPH, and 2 of these cases showed unexpected loss of EPM2AIP1 staining. Of the 6 cases with double somatic mutations of MLH1 gene (without MPH), only 4 cases demonstrated intact expression of EPM2AIP1 as expected. Taken together, EPM2AIP1 loss was 64% sensitive and 67% specific for MPH, with an accuracy of 64%. We conclude that, unless stain quality improves with different clones or platforms, EPM2AIP1 IHC will likely not be useful as a surrogate test for MPH in CRC.

Adapting SAM to Histopathology Images for Tumor Bud Segmentation in Colorectal Cancer.

Colorectal cancer (CRC) is the third most common cancer in the United States. Tumor Budding (TB) detection and quantification are crucial yet labor-intensive steps in determining the CRC stage through the analysis of histopathology images. To help with this process, we adapt the Segment Anything Model (SAM) on the CRC histopathology images to segment TBs using SAM-Adapter. In this approach, we automatically take task-specific prompts from CRC images and train the SAM model in a parameter-efficient way. We compare the predictions of our model with the predictions from a trained-from-scratch model using the annotations from a pathologist. As a result, our model achieves an intersection over union (IoU) of 0.65 and an instance-level Dice score of 0.75, which are promising in matching the pathologist's TB annotation. We believe our study offers a novel solution to identify TBs on H&E-stained histopathology images. Our study also demonstrates the value of adapting the foundation model for pathology image segmentation tasks.

Enhancing Colorectal Cancer Tumor Bud Detection Using Deep Learning from Routine H&E-Stained Slides.

Tumor budding refers to a cluster of one to four tumor cells located at the tumor-invasive front. While tumor budding is a prognostic factor for colorectal cancer, counting and grading tumor budding are time consuming and not highly reproducible. There could be high inter- and intra-reader disagreement on H&E evaluation. This leads to the noisy training (imperfect ground truth) of deep learning algorithms, resulting in high variability and losing their ability to generalize on unseen datasets. Pan-cytokeratin staining is one of the potential solutions to enhance the agreement, but it is not routinely used to identify tumor buds and can lead to false positives. Therefore, we aim to develop a weakly-supervised deep learning method for tumor bud detection from routine H&E-stained images that does not require strict tissue-level annotations. We also propose Bayesian Multiple Instance Learning (BMIL) that combines multiple annotated regions during the training process to further enhance the generalizability and stability in tumor bud detection. Our dataset consists of 29 colorectal cancer H&E-stained images that contain 115 tumor buds per slide on average. In six-fold cross-validation, our method demonstrated an average precision and recall of 0.94, and 0.86 respectively. These results provide preliminary evidence of the feasibility of our approach in improving the generalizability in tumor budding detection using H&E images while avoiding the need for non-routine immunohistochemical staining methods.

Few-shot Tumor Bud Segmentation Using Generative Model in Colorectal Carcinoma.

Current deep learning methods in histopathology are limited by the small amount of available data and time consumption in labeling the data. Colorectal cancer (CRC) tumor budding quantification performed using H&E-stained slides is crucial for cancer staging and prognosis but is subject to labor-intensive annotation and human bias. Thus, acquiring a large-scale, fully annotated dataset for training a tumor budding (TB) segmentation/detection system is difficult. Here, we present a DatasetGAN-based approach that can generate essentially an unlimited number of images with TB masks from a moderate number of unlabeled images and a few annotated images. The images generated by our model closely resemble the real colon tissue on H&E-stained slides. We test the performance of this model by training a downstream segmentation model, UNet++, on the generated images and masks. Our results show that the trained UNet++ model can achieve reasonable TB segmentation performance, especially at the instance level. This study demonstrates the potential of developing an annotation-efficient segmentation model for automatic TB detection and quantification.