This week in Pathology

This AudioScholar briefing is generated by artificial intelligence for healthcare professionals and trainees. It is not medical advice.

Welcome to This Week in Pathology. This week we're covering 10 notable papers spanning molecular and morphologic risk stratification in gynaecologic and colorectal cancers, the push to quantify what we see on the slide, and new immunohistochemical tools for some of our most stubborn differential diagnoses. Let's dive in.

We start with gynaecologic and colorectal pathology, where molecular classification is reshaping how precursors and carcinomas are sorted. In Virchows Archiv, Barbe and colleagues tested the fully automated Idylla real-time PCR assay for POLE hotspot mutations against targeted next-generation sequencing in 40 colorectal and endometrial cancers [1]. The cartridge picked up POLE hotspot variants with 96 percent sensitivity and perfect specificity, but limit-of-detection work showed it was reliable only when DNA concentration exceeded 5 nanograms per microlitre and tumour content exceeded 20 percent. The authors frame it as a possible front-line screen to speed molecular classification, with sequencing still needed for low-input samples; this is a small retrospective validation, and it detects hotspots only, so it supplements rather than replaces sequencing. Also in Virchows Archiv, Travaglino and colleagues examined 33 complex endometrial non-morular metaplasias, the tubal, eosinophilic, mucinous and related lesions that often sit awkwardly between benign and precancerous [2]. Roughly four in five of these lesions carried KRAS mutations, compared with about one in five conventional atypical hyperplasias, and most of the KRAS-wild-type lesions had other MAPK pathway mutations. PTEN and PIK3CA alterations were common, none had CTNNB1 mutations, and on subsequent hysterectomy close to 40 percent of cases showed endometrioid carcinoma, with a further 44 percent showing atypical hyperplasia. The authors propose calling these lesions metaplastic atypical hyperplasia or endometrioid intraepithelial neoplasia; the follow-up cohort was only 18 hysterectomies, so the proposal rests on a modest series, but it gives a molecular rationale for treating these metaplasias as genuine precancers. Moving to the vulva, Lengyel and colleagues in The American Journal of Surgical Pathology assessed tumour budding and tumour-stroma ratio in 172 vulvar squamous cell carcinomas [3]. High budding activity, at 20 or more buds per ten high-power fields, was associated with shorter recurrence-free survival, especially in the HPV-independent, p53-abnormal group, and that association held after adjusting for lymphovascular invasion and stage. Budding also tracked with worse disease-specific and overall survival. This is a single retrospective cohort spanning three decades, and the authors explicitly call for validation before budding enters routine vulvar reporting.

Our second theme is quantification, the effort to turn morphologic impressions into reproducible, prognostically meaningful numbers. In Histopathology, Sur and colleagues report on behalf of the International Immuno-Oncology Biomarker Working Group with recommendations for assessing tumour-infiltrating lymphocytes in colorectal carcinoma [4]. The report acknowledges that despite strong prognostic data, there is still no widely accepted definition of colorectal TILs and they are not used systematically in practice; it proposes a standardised assessment method, reviews relationships with other biomarkers, and highlights artificial intelligence-based quantification and spatial technologies. It is a consensus document rather than new outcome data, but it is the kind of groundwork that precedes adoption into reporting standards. Also in Histopathology, Buzoianu and colleagues used a commercial machine learning algorithm to measure Gleason pattern 4 on biopsies from 726 men with Grade Group 2 to 4 prostate cancer who went on to prostatectomy [5]. They compared 15 ways of measuring pattern 4, differing mainly in how gaps between cancer foci were handled, and found that a pixel-counting approach discriminated best, though the gains were modest: pattern 4 quantification edged out Grade Group alone for predicting adverse prostatectomy pathology, and the amount of pattern 3 added nothing once pattern 4 was known. The authors present this as proof of principle for machine learning as a research tool to settle the measurement question, not yet as a reporting method. Haematopathology contributes a related idea from Virchows Archiv, where Hsu and colleagues digitally quantified eleven BCL-2 family proteins on diagnostic marrow biopsies from 52 elderly patients with acute myeloid leukaemia treated with venetoclax-based regimens [6]. Higher expression of the pro-apoptotic proteins BAD, HRK and NOXA was associated with longer survival, while higher BCL-XL and MCL-1 were linked to poorer outcomes; for context, adverse cytogenetic risk was associated with roughly six times the risk of death. With 52 patients from a single centre and data-derived cutoffs, this is hypothesis-generating, but it suggests routine marrow immunohistochemistry could carry prognostic information for venetoclax therapy.

Our final theme gathers diagnostic adjuncts for differentials that morphology alone struggles to resolve. In The American Journal of Surgical Pathology, Huang and colleagues applied TRBC1 immunohistochemistry to early mycosis fungoides [7]. They describe what they call spatial clonal discordance: monotypic TRBC1 expression confined to intraepidermal T cells while the dermal infiltrate remains polytypic, a pattern that could explain why bulk PCR clonality assays so often come back equivocal in paucicellular early lesions. Tumour-stage disease and Sézary syndrome showed restriction across all compartments, and digital analysis supported thresholds of 25 percent or less, or 75 percent or more, TRBC1-positive cells to define clonality. The authors acknowledge a small, single-centre retrospective cohort, so the thresholds need external testing, but the approach addresses a long-standing diagnostic gap. In the kidney, Nguyen and colleagues in Virchows Archiv assembled 51 eosinophilic vacuolated tumours from 21 institutions alongside 149 differential diagnostic tumours [8]. CD56 was expressed in nearly all of the eosinophilic vacuolated tumours, mostly in a strong diffuse pattern, but that same strong pattern appeared in about a third of oncocytomas and a quarter of FLCN-mutated tumours, while low-grade oncocytic tumours were negative. Every sequenced tumour carried TSC1, TSC2 or MTOR variants. The authors position CD56 as a useful adjunct within a panel, not a standalone discriminator, particularly given the oncocytoma overlap. Rakha, also in Virchows Archiv, offers a narrative review of papillary breast neoplasms reflecting the sixth edition WHO classification [9], arguing for viewing papillary neoplasia as a biological continuum and cautioning against applying conventional invasion criteria outside the context in which they were developed, with practical guidance on core biopsies, mimics and staging. And in Cancer Cytopathology, Wojcik's commentary on urinary cytology [10] draws on two recent studies, one showing keratin 17 immunocytochemistry improved performance alongside cytology in subtle cases, the other showing the methylation-based Bladder EpiCheck assay improved risk stratification in indeterminate cytology with negative cystoscopy. The common thread with the renal and lymphoma papers is that biomarkers add most when integrated with morphology rather than used in isolation; as an editorial, it synthesises rather than adds new data.

If you only have time for one paper this week, make it the TRBC1 study in early mycosis fungoides from Huang and colleagues [7]. It offers a spatially resolved explanation for one of dermatopathology's most frustrating problems, discordant clonality in early lesions, and reopens the question of how much weight bulk molecular assays should carry in that setting.

That's your roundup for This Week in Pathology. The full transcript and references are available on the episode page. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if you have a paper you have been meaning to read, upload the PDF, or paste any link, at audioscholar dot C C. We will turn it into audio like this one, in any of thirty-one languages.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.