This Week in Pathology — Oct 3, 2026

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 predictive biomarker scoring in gastroesophageal cancer, prognostic features that sit outside conventional staging, and the shifting boundaries of tumour classification and its mimics. Let's dive in.

We start with predictive biomarkers in gastric and gastroesophageal junction adenocarcinoma, where pathologists' immunohistochemistry reads increasingly decide who gets targeted therapy. In Human Pathology, Sy and colleagues asked six gastrointestinal pathologists to score claudin 18.2 staining on digitally scanned slides from two cohorts totalling 95 adenocarcinomas, recording both an H-score and whether each case met the zolbetuximab eligibility cutoff of more than 75 percent of tumour cells with moderate or strong membranous staining [1]. Agreement on the overall H-score was good, and agreement on the eligibility call itself was substantial, with all six raters fully concordant in about two thirds of cases in the nonconsecutive cohort. The weak point was the middle of the intensity scale: agreement on the proportion of cells staining 1+ and 2+ was poor to fair, and tumours with heterogeneous staining showed noticeably lower agreement on eligibility. The authors argue for standardized scoring and a consensus approach in difficult cases; this is a reader study on scanned slides rather than an outcomes study, but it does locate where the misclassification risk concentrates. A second Human Pathology paper, from Zhang and colleagues, looked at FGFR2b, an emerging target, in 169 patients with paired primary and metastatic specimens [2]. Positivity was uncommon, around 14 percent of primaries at the any 2+ or 3+ cutoff, and expression was patchy in space, with fewer positive cells in superficial than in deeper tumour regions. Discordance between primary and metastasis was marked and ran mostly in one direction: more than half of the primaries positive at the any-staining cutoff became negative in their metastases. FGFR2b status was independent of HER2, claudin 18.2, mismatch repair and PD-L1, and a third of FGFR2b-positive tumours carried no other currently targetable biomarker. All tumours with homogeneous FGFR2b staining harboured FGFR2 amplification. Read together, these two papers point the same way: superficial biopsies and heterogeneous staining are where treatment-eligibility calls are least secure, although neither study links scoring variation to patient outcomes.

Our second theme is prognostic information that lives outside the dominant lesion or the conventional staging variable. In Histopathology, Shu and colleagues used the Netherlands Cancer Registry to follow 3,505 colorectal cancer patients without distant metastases at diagnosis [3]. Tumour deposits were independently associated with about a third higher risk of developing distant metastases, and among patients who did metastasize, those with both nodal disease and tumour deposits had the largest share with four or more involved sites, about 18 percent, against roughly 11 percent with nodal disease alone. This is retrospective registry data from a single half-year of diagnoses, so it supports the argument that tumour deposits may mark a distinct metastatic phenotype rather than proving it, and the authors frame it as a case for future risk stratification rather than a staging change. In The American Journal of Surgical Pathology, Shah and colleagues addressed a question many genitourinary pathologists face at prostatectomy sign-out: what does a positive margin mean when it involves only a secondary tumour while the dominant tumour is cleanly excised [4]? Among 1,449 multifocal prostatectomies, about one in twenty had a positive margin in a secondary tumour, usually grade group 1 or 2. After adjustment for grade, volume, extraprostatic extension and seminal vesicle invasion, that secondary-tumour margin roughly tripled the risk of biochemical recurrence. Yet the dominant-tumour features carried far heavier weight, with grade groups 3 through 5 associated with roughly ten- to sixteen-fold higher risk, and the authors conclude that secondary-tumour margins have a lesser effect than dominant tumour characteristics. Follow-up was short, around two years, so the finding supports reporting these margins but leaves their long-term weight unsettled. The week's highest-profile paper, in JAMA, from de Avila and colleagues, makes a related point in the liver: the histologic variable, not the body habitus, carries the prognosis [5]. In more than 18,000 patients with biopsy-confirmed metabolic dysfunction-associated steatotic liver disease from 41 countries, just under seven percent were lean by body mass index, most often in Asia. Lean patients had milder histology, with less diabetes, less advanced fibrosis and lower activity scores, but lean status was not independently associated with death or liver-related events. Advanced fibrosis, by contrast, roughly doubled the risk of death and more than tripled the risk of clinical events. The fibrosis-4 score was somewhat less accurate in lean patients, while transient elastography performed at least as well, if not better. It is observational and retrospective, but its size gives weight to the conclusion that fibrosis stage remains the anchor of the biopsy report regardless of weight.

Third, kidney and adrenal tumours, where morphology, immunohistochemistry and genomics are being cross-referenced. In Human Pathology, Wen and colleagues pooled 84 BAP1-mutated clear cell renal cell carcinomas from a Chinese cancer centre and two public genomic datasets [6]. They describe six morphologic patterns, with macroacinar and mixed patterns the most common, and half of the mixed cases combining papillary and macroacinar architecture. In the institutional cases, every tumour showed complete nuclear BAP1 loss with predominantly diffuse strong AMACR staining, and well over half of the tumours in each cohort were high nuclear grade. Complementing this, Tanaka and colleagues in Histopathology examined 108 clear cell carcinomas verified molecularly by VHL alteration or 3p loss [7]. Keratin 7 was positive in about 30 percent overall, more often in primaries than metastases, in lower-grade architectural patterns, and in PBRM1 wild-type tumours, and CA9 negativity was seen only in metastases, about 15 percent of them. BAP1-mutated tumours in this series tended toward higher-grade architecture and loss of CD10, consistent with Wen's description. The shared message is that aberrant staining in true clear cell carcinoma is common and patterned, particularly in metastatic and molecularly defined subsets, though both studies are retrospective and modest in size. For adrenocortical carcinoma, Du and colleagues in Virchows Archiv reviewed 65 resected tumours from a single institution [8]. After adjusting for pT category, more than five mitoses per 50 high-power fields, nuclear beta-catenin expression, and a pathogenic or likely pathogenic alteration on clinical molecular testing were each associated with worse overall survival, the molecular finding with close to a fivefold higher risk. Ki-67 of ten percent or more, aberrant p53, necrosis and angioinvasion were not significantly associated with survival after adjustment, and mitotic activity's link with disease-specific survival narrowly missed significance. With small numbers, wide uncertainty and molecular testing in only half the cohort, the authors themselves call for larger standardized studies.

Finally, two papers on diagnostic boundaries. Chetty and Gill, writing in Histopathology, review how the 2026 WHO classification handles gastrointestinal tumours with mixed differentiation [9]. Mixed neuroendocrine-non-neuroendocrine neoplasm still requires two morphologically recognizable components, each making up at least 30 percent of the tumour, with clear neuroendocrine marker expression. A non-neuroendocrine component that is only a precursor or intramucosal lesion does not qualify, so the preferred wording is a neuroendocrine carcinoma arising from an adenoma. The new category of amphicrine-like carcinoma captures tumours where both lines of differentiation are intimately intermixed rather than zonal, and molecular data favour a common clonal origin over collision. This is a narrative review, but it maps directly onto reporting language. And in Human Pathology, Xiao and colleagues described 19 women with late breast hematomas showing papillary endothelial hyperplasia, presenting on average 18 years after surgery and 13 years after radiation, often on anticoagulation and imaged as suspicious masses [10]. The helpful features were reactive endothelium dissecting organizing fibrin rather than collagen, highlighted by Martius-Scarlet-Blue or trichrome stains, along with negative PRAME and wild-type MYC and p53 staining in all tested lesions. It is a small single-institution series, but it describes a mimic of radiation-associated angiosarcoma that an ageing survivor population may make more common.

If you only have time for one paper this week, make it the claudin 18.2 interobserver study from Sy and colleagues in Human Pathology [1]. It shows that pathologists largely agree on zolbetuximab eligibility, while pinpointing weak intensities and heterogeneous staining as the zone where a treatment decision can hinge on who reads the slide.

Here is what this week's evidence adds up to in Pathology. First, biomarker-driven gastroesophageal therapy rests on reasonably reproducible claudin 18.2 scoring, but FGFR2b expression is heterogeneous and often lost in metastases, and neither study tests whether these discrepancies change outcomes. Second, features outside conventional staging, such as tumour deposits in colorectal cancer and positive margins in secondary prostate tumours, carry independent prognostic signal in retrospective data, though smaller than dominant-tumour features in the prostate. Third, a large multinational JAMA cohort indicates that fibrosis stage, not leanness, tracks with outcomes in steatotic liver disease, and that elastography holds up in lean patients. Fourth, molecularly anchored studies show that aberrant immunoprofiles in clear cell renal carcinoma are common and patterned, and that mitoses, nuclear beta-catenin and molecular alterations may add prognostic information in adrenocortical carcinoma, all from modest retrospective series. And finally, the 2026 WHO framework tightens what counts as a mixed neuroendocrine neoplasm, while a small breast series describes reproducible clues separating late hematomas from angiosarcoma.

That's your roundup for This Week in Pathology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And if it saves you time, a quick rating in your podcast app helps other physicians find the show.

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