Summary
A HER2-low–focused IHC scoring system was validated by nine breast pathologists using digitized images of HER2 IHC slides. The system demonstrated high performance metrics, including accuracy, sensitivity, and specificity, across two data sets, validating its effectiveness. The validated approach will be used for peer training and updating the national HER2 IHC external quality assurance program.
Editors: Rouzan Karabakhtsian, MD, PhD, professor of pathology and director of the Women’s Health Pathology Fellowship, Albert Einstein College of Medicine, Montefiore Medical Center, Bronx, NY; S. Emily Bachert, MD, associate pathologist, Brigham and Women’s Hospital, Boston; Amarpreet Bhalla, MD, assistant professor of pathology, Albert Einstein College of Medicine, Montefiore Medical Center; Divya Sharma, MD, associate professor, Department of Pathology and Laboratory Medicine, University of Cincinnati Medical Center; and Paula Toro, MD, gastrointestinal and hepatobiliary fellow, Cleveland Clinic.
Independent validation of a HER2-low–focused IHC scoring system
November 2025—For two decades, the American Society of Clinical Oncology–College of American Pathologists human epidermal growth factor receptor 2 testing criteria have included 0 and 1+ scores, but this distinction was inconsequential. Outcome data from the Destiny-Breast04 (DB-04) trial demonstrated improved survival for patients with metastatic breast cancer who had low levels of HER2 protein expression and were treated with trastuzumab-deruxtecan. Discerning 0 from 1+ IHC staining is challenging because human epidermal growth factor receptor 2 (HER2) low is not a biologically distinct cancer subset; no reference standards or controls exist; and second-tier tests, such as in situ hybridization, do not apply. Prior reports cast doubt on the reliability of pathologists’ IHC scoring, with resulting treatment misalignments. With institutional review board approval, the authors (nine breast pathologists from eight Australian laboratories) had previously established HER2-low–focused scoring conventions, based on the American Society of Clinical Oncology-College of American Pathologists 2018 HER2 guidelines, that included common staining pitfalls. They reported the results of the first set of 60 breast cancers evaluated with these methods. After a five-month washout period, the authors conducted another validation study in which they compiled a second set of 64 HER2-negative invasive breast cancer core biopsies, all assessed using the Ventana 4B5 HER2 assay. Each pathologist scored digitized images of HER2 IHC slides of the cases. Using majority opinion as the target score, they calculated the performance metrics. The authors compared the results of their performance in sets one and two to assess the effectiveness and retention of their approach. The cases in this validation set included 40 (62.5 percent) HER2 low, 10 (17.2 percent) ultralow, and 13 (18.8 percent) null cancers. Concordance was not achieved in one case. For distinguishing HER2 low from other cancers (ultralow and null combined), the mean values of the authors’ performance metrics were accuracy of 89.58 percent, sensitivity of 90.83 percent, specificity of 87.50 percent, positive predictive value of 95.63 percent, negative predictive value of 83.59 percent, and Cohen kappa score of 0.81. Comparing these results with those of their initial study, the authors determined that they had maintained their high level of performance across these parameters. The mean kappa score was in the excellent range for concordance. The authors concluded that maintaining high performance across a range of measures in two data sets validates the effectiveness of their HER2-low–focused scoring conventions. Having validated their approach, the authors will use these reference case sets with expert-level consensus scores for peer training and updating their national HER2 IHC external quality assurance program. In ongoing studies, they are also assessing the suitability of software algorithms for prescreening HER2 IHC slides.
Farshid G, Armes J, Dessauvagie B, et al. Independent validation of a HER2-low focused immunohistochemistry scoring system for enhanced pathologist precision and consistency. Mod Pathol. 2025;38. doi.org/10.1016/j.modpat.2024.100693
Correspondence: Dr. Gelareh Farshid at [email protected]
Relevance of complex pathological findings in alcoholic foamy degeneration of liver
It is important to differentiate alcoholic foamy degeneration from severe alcoholic hepatitis. This is partly because corticosteroid treatment is recommended for some patients with severe alcoholic hepatitis, while symptoms and biochemical abnormalities of alcoholic foamy degeneration (AFD) rapidly improve after abstinence from alcohol, with almost no risk of mortality. Diagnosis of AFD requires histological confirmation of characteristic foamy hepatocytes with numerous minute fat droplets in the cytoplasm (microvesicular steatosis). The authors performed a retrospective study to better characterize clinicopathological features of AFD. They found that patients with the condition presented with biochemical liver dysfunction (n = 1) or clinical jaundice (n = 8). The patient without jaundice had mixed macrovesicular and microvesicular bland steatosis. Seven patients with jaundice had lobular inflammation, acidophilic bodies, cholestasis, and lobular distortion. The affected hepatocytes were extensively enlarged and had clear cytoplasmic change, and they somewhat resembled ballooning degeneration. This was primarily due to accumulated lipid droplets, or pseudoballooning. The one remaining patient had predominant changes of AFD, but a few of that patient’s foci showed classical ballooning hepatocytes and Mallory–Denk bodies, in keeping with mixed AFD and steatohepatitis. When compared with patients who had severe alcoholic hepatitis, those with AFD had lower white blood cell and neutrophil counts and higher cholesterol levels (all p < .001). On imaging, ascites and varices were less common with AFD than with severe alcoholic hepatitis (11 versus 75 percent, p = .014; zero versus 67 percent, p = .008, respectively). All seven patients with AFD who abstained from alcohol experienced rapidly improved liver function. The authors proposed that it is important to recognize the AFD pattern for patients with clinical jaundice and to be aware of the following diagnostic criteria for AFD: acute or acute-on-chronic presentation with jaundice; a history of excessive alcohol consumption; microvesicular steatosis in 20 percent or more of hepatocytes; nonballooning, foamy hepatocyte enlargement that is at least two times greater than that of nonsteatotic hepatocytes; and absence or only focal presence of steatohepatitis. The criterion of microvesicular steatosis in 20 percent or more of hepatocytes is provisional. Additional studies are needed to determine whether some patients with AFD have less than 20 percent microvesicular steatosis.
Jeyanesan D, Antonello A, Cannon M, et al. Rethinking alcoholic foamy degeneration of the liver: a study of nine cases highlighting complex pathological findings. J Clin Pathol. 2025. doi:10.1136/jcp-2024-209939
Correspondence: Dr. Yoh Zen at [email protected]
Use of deep-learning–informed multimodal fusion of radiology and pathology in HPV-associated OPSCC
The authors conducted a study to predict outcomes of human papillomavirus-associated oropharyngeal squamous cell carcinoma (OPSCC), a subtype of head and neck cancer typically characterized by favorable clinical outcome and treatment response. Pathology- and radiology-focused artificial intelligence-based prognostic models have been independently developed for OPSCC, but the combined prognostic implications of the primary tumor and lymph node involvement across multiscale imaging has received less attention. For the study, the authors investigated the prognostic value of the Swin Transformer-based multimodal and multiregion data fusion framework (SMuRF). SMuRF integrates features from computed tomography corresponding to the primary tumor and lymph node, as well as whole slide pathology images from the primary tumor as a predictor of survival and tumor grade in human papillomavirus-associated OPSCC. SMuRF employs cross-modality and cross-region window-based, multi-head, self-attention mechanisms to capture interactions between features across tumor habitats and image scales. Developed and tested on a cohort of 277 patients with OPSCC with matched radiology and pathology images, SMuRF demonstrated strong performance (C-index = 0.81 for disease-free survival prediction; area under the curve [AUC] = 0.75 for tumor grade classification) and emerged as an independent prognostic biomarker for disease-free survival (hazard ratio = 17; 95 percent confidence interval [CI], 4.9–58; p < .0001) and tumor grade (odds ratio = 3.7; 95 percent CI, 1.4–10.5; p = .01) after controlling for other clinical variables—that is, T- and N-stage, age, smoking, sex, and treatment modalities. Importantly, SMuRF outperformed unimodal models derived from radiology or pathology alone. The authors’ findings underscore the potential of multimodal deep learning for stratifying OPSCC risk, informing tailored treatment strategies, and refining treatment algorithms.
Song B, Leroy A, Yang K, et al. Deep learning informed multimodal fusion of radiology and pathology to predict outcomes in HPV-associated oropharyngeal squamous cell carcinoma. eBioMedicine. 2025;114. doi.org/10.1016/j.ebiom.2025.105663
Correspondence: Dr. Anat Madabhushi at [email protected]