This Week in Pathology — Aug 23, 2026

Welcome to This Week in Pathology. This week we're covering ten notable papers spanning predictive biomarker testing in gastrointestinal and gynaecological cancers, the diagnostic pitfalls of immunohistochemistry when a single stain is asked to do too much, and molecular characterisation of rare and borderline tumours. Let's dive in.

We'll start with therapeutic biomarkers, where the pathologist's report increasingly determines drug access. In Modern Pathology, Ma and colleagues profiled Claudin 18.2 immunohistochemistry in 564 surgically resected gastric and gastro-oesophageal junction adenocarcinomas using a clinical-trial-validated monoclonal antibody [2]. Crisp membranous staining of any degree was present in about six in ten tumours, but when they applied the stringent approved threshold — at least three quarters of tumour cells with moderate-to-strong membranous staining — only about a fifth of gastric and just over a quarter of junctional cancers qualified. That gap between any staining and threshold-positive staining is the practical message: as trials of moderate- and low-expressing tumours read out, the denominator of eligible patients could roughly triple, and laboratories should be recording the proportion and intensity of staining rather than a binary result. High-threshold positivity tracked with Epstein-Barr virus status and, interestingly, with stage one disease in gastric primaries, while transcriptomic analysis confirmed good concordance between the stain and Claudin 18 messenger RNA. Staying with treatment-relevant molecular pathology, The Journal of Pathology reports work from Guo and colleagues on SMARCA4-deficient undifferentiated malignancies of the digestive tract [7]. In a retrospective cohort of forty-three undifferentiated digestive tumours, loss of SMARCA4 by immunohistochemistry was found in about thirty percent, and those patients had roughly three times the risk of death and around two and a half times the risk of recurrence, independent of lineage markers or microsatellite status. Linked analysis of The Cancer Genome Atlas showed SMARCA4-mutated tumours carried a higher tumour mutational burden and co-occurring alterations in ERBB2, MET, RET, BRCA2 and the mismatch repair genes — so a SMARCA4-deficient undifferentiated gut tumour is not merely a bad-prognosis label but an argument for broad sequencing.

The second theme is molecular testing in gynaecological pathology, and specifically what happens when the pathologist, rather than the oncologist, drives it. In Human Pathology, Mills and colleagues audited every endometrial carcinoma over roughly three years in which the reporting pathologist recommended sequencing in their comment [3]. Sequencing was suggested in thirty-eight cases but actually pursued in only sixteen — fewer than half of the recommendations were acted on. Of those tested, testing changed the classification in half: four tumours proved POLE-mutated and four had molecular findings that contradicted the immunostains. Every POLE-mutated case was a multiple classifier in a woman over fifty, half were stage three, and none had an adverse outcome, while two thirds of the no-specific-molecular-profile tumours did. There were incidental wins too — a BRCA2 mutation that unmasked a germline syndrome, and a tumour reclassified from high-grade carcinoma to malignant PEComa on the basis of TSC2 alterations. The practice point is that POLE testing should not be reserved for young women with low-grade tumours, and that a pathologist's recommendation only helps if the clinical team acts on it.

That leads naturally into a run of papers about the limits of single immunostains. In the American Journal of Surgical Pathology, Shakiba and colleagues examined cervical squamous carcinomas with negative or merely focal p16 staining, which accounted for about five percent of two hundred and one cases [5]. Crucially, this group was mixed: four were high-risk HPV-associated despite losing p16, six were HPV-independent, and one was linked to low-risk HPV. Three of the four HPV-associated p16-negative cancers also showed loss of methylthioadenosine phosphorylase, pointing to co-deletion of the locus, and that same co-deletion appeared in precursor high-grade lesions — whereas all six HPV-independent carcinomas retained that marker. TERT promoter mutations were common in both groups, and PIK3CA hotspot mutations appeared only in the HPV-independent tumours. The actionable message is that a negative p16 stain does not exclude HPV-associated carcinoma, and correlation with direct HPV testing or a prior liquid-based cytology result is warranted before reclassifying a tumour. A parallel cautionary tale comes from Histopathology, where Orsatti and colleagues assessed SOX2 in sixty-one ovarian teratomas [10]. SOX2 was diffusely positive not only in immature neuroepithelium but also in mature glia, ependymal structures and some non-neural epithelium, with a distinctive layer-restricted pattern in retina and consistent negativity in cerebellum. When heterogeneous SOX2 staining in rosette-like structures triggered systematic morphological review, more than half of the cases originally called immature teratoma — five of nine — were reclassified as mature. SALL4, by contrast, was absent from all mature components. Given that immature teratoma grading drives chemotherapy decisions in young women, treating SOX2 as a lineage marker rather than a maturity marker, and reading it alongside morphology and SALL4, is the safeguard against overdiagnosis.

Our final theme is the characterisation of rare and borderline entities, where classification systems are still moving. Virchows Archiv carries a review from Luchini synthesising the 2026 World Health Organization updates on pancreatic ductal adenocarcinoma and its subtypes, covering adenosquamous, colloid, medullary, hepatoid, poorly cohesive and undifferentiated variants, the reclassification of micropapillary morphology, and the persistently unresolved problem of scoring treatment response after neoadjuvant therapy [1]. In Modern Pathology, Acosta and colleagues report thirteen new cases of inflammatory and nested testicular sex cord tumour [6]. Of eight patients with adequate follow-up, seven developed retroperitoneal nodal metastases and two had lung disease — this is an aggressive tumour. Nested architecture with inflammation and thick collagenous stroma, positivity for inhibin-alpha, CD30, keratin and steroidogenic factor 1, and an EWSR1::ATF1 fusion or EWSR1 rearrangement in twelve of thirteen tumours define it; the Leydig-cell-like case in the series is the trap to remember. The Journal of Pathology also reviews SRF fusion genes in myoid soft tissue tumours, which behave very differently — predominantly in children, usually low-grade, and often cured by complete resection [8]. Two further papers deal with lesions that sit on the borderline. Hamada and colleagues in Human Pathology described thirty-nine thyroid nodules that had been labelled uncertain malignant potential but lacked the questionable capsular or angioinvasion the World Health Organization criteria require [4]; with a median follow-up of two and a half years there were no recurrences or metastases despite high-risk TERT promoter or PIK3CA alterations in about a fifth, supporting hemithyroidectomy and conservative management. And Ravisankar and colleagues used fluorescence in situ hybridisation for isochromosome 12p to test whether pagetoid involvement of the rete testis in seminoma is invasion or in-situ spread [9]. Isochromosome 12p was present in only one of fourteen evaluable pagetoid foci, against about a third of the paired invasive seminomas, supporting the long-held morphological assumption that pagetoid rete involvement represents spread of germ cell neoplasia in situ and should not be staged as invasion.

If you only have time for one paper this week, make it the Claudin 18.2 survey in Modern Pathology [2]. It tells you what proportion of your gastric and junctional adenocarcinomas will clear the current threshold, and how much that eligible population expands if lower cut-offs are approved — information you need before the requests arrive.

Here are the key takeaways from this week in Pathology. First, report Claudin 18.2 with intensity and percentage rather than as a binary call, because the threshold that defines eligibility is likely to shift. Second, a negative p16 does not rule out HPV-associated cervical carcinoma, and SOX2 positivity does not mean immature neuroepithelium — both stains need morphological and ancillary correlation. Third, POLE testing in endometrial cancer belongs in post-menopausal women with abnormal p53 and high-grade histology, and in selected stage three disease, but the recommendation must be followed through to be useful. Fourth, SMARCA4 loss in an undifferentiated gastrointestinal tumour predicts poor survival and should prompt broad sequencing for co-occurring actionable targets. And fifth, on the reassuring side, pagetoid rete testis involvement behaves as in-situ disease, and borderline thyroid nodules outside formal criteria appear indolent and can be managed conservatively.

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 a quick rating on Apple Podcasts or Spotify helps other physicians discover the show.

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