This Week in Pathology — Sep 19, 2026

Welcome to This Week in Pathology. This week we're covering 10 notable papers spanning three broad themes: the unglamorous but decisive questions of which stain and which block we choose, transcription-factor and copy-number subtyping of aggressive tumours, and a set of newly sharpened entities and risk-prediction findings. Let's dive in.

We start with the bench decisions that change a report. In Histopathology, Law and colleagues asked a question most of us have never had good data on: when a guideline tells us to order an elastin stain for large vessel invasion in colorectal cancer, which elastin stain? They took 78 colorectal resections of at least pT1 stage and generated 155 matched sets of orcein, elastic Verhoeff-Van Gieson, and recut haematoxylin and eosin, with vessel invasion adjudicated by at least two pathologists [1]. Using orcein plus haematoxylin and eosin as the reference, routine haematoxylin and eosin alone missed about three in ten positive cases and, notably, over-called invasion in a fifth of negatives. But the more interesting comparison is between the two elastin stains: the Verhoeff-Van Gieson stain missed roughly a fifth of the cases orcein picked up, and orcein gave circumferential staining of the vessel wall in just over three quarters of sets, compared with well under half for Verhoeff-Van Gieson. Verhoeff-Van Gieson remains highly specific, so a positive result is trustworthy, but if your laboratory has standardised on it, this paper is an argument for a conversation with your histotechnology team about orcein instead.

The companion question to which stain is which block, and two papers land squarely on that. Also in Histopathology, Vyas, Chen and Gonzalez looked at 75 jejunoileal well-differentiated neuroendocrine tumours with nodal disease or mesenteric tumour deposits, staining the largest primary, the largest node and the largest deposit for Ki67 and using digital image analysis validated against manual counts [2]. In about two thirds of cases the proliferation index of a node or deposit exceeded that of the primary, and on multivariable analysis only the Ki67 of the mesenteric tumour deposit predicted progression; the primary index, the nodal index, the highest index in the case, and the index of the largest focus all failed. Deposit size independently predicted overall survival. The practical instruction is simple: when a jejunoileal neuroendocrine tumour resection contains a mesenteric deposit, stain the largest deposit, not reflexively the primary. Meanwhile in The American Journal of Surgical Pathology, Martin and colleagues examined 32 treatment-naive mismatch repair-deficient prostate cancers and found the phenotype is broader than the textbook suggests — about a quarter were grade group 2, and intraductal or invasive cribriform carcinoma was present in roughly four fifths of cases overall and equally in the grade group 2 tumours [3]. Loss predominantly involved MSH2 and MSH6, a third of those who underwent germline testing had Lynch syndrome, and a third had concurrent homologous recombination repair alterations. Most striking for daily practice: where immunohistochemistry was run on multiple blocks, mismatch repair status was discordant between foci in well over half of cases, and deficiency was consistently confined to the highest-grade focus. In multifocal prostate cancer, test the highest-grade tumour, or you will call a Lynch syndrome patient proficient.

The second theme is subtyping aggressive tumours by transcription factor and by genome-wide copy number, and two papers on neuroendocrine carcinoma converge nicely from different organs. In Histopathology, Tenace and colleagues applied the small cell lung cancer transcriptional taxonomy to 88 small cell carcinomas of the bladder, characterising 79 by immunohistochemistry [4]. They resolved six subgroups — ASCL1-driven cases were the largest at just over 40 percent, followed by NEUROD1 and POU2F3 — which collapse usefully into a high neuroendocrine group and a low neuroendocrine group. Membranous Nectin-4 expression was low or absent overall and significantly higher in the low neuroendocrine tumours, which matters because it tempers enthusiasm for enfortumab vedotin in this setting even though NECTIN4 amplification occurred in about a fifth of cases, comparable to conventional urothelial carcinoma. DLL3 and SLFN11 were overexpressed in the high neuroendocrine group, pointing toward DLL3-directed agents and platinum or PARP-based strategies instead. In The American Journal of Surgical Pathology, Asai and colleagues did the analogous exercise in 42 biliary tract neuroendocrine carcinomas [5]. NEUROD1 type was commonest at 40 percent, POU2F3 type was enriched in the gallbladder, and the null type independently predicted shorter disease-specific survival, tripling the hazard. The diagnostic warning is important: several POU2F3-type tumours expressed none of chromogranin A, synaptophysin or INSM1, so a morphologically convincing neuroendocrine carcinoma with negative conventional markers is not automatically something else.

Staying with high-risk molecular subgroups, Modern Pathology brings two papers that add resolution where current classifiers are blunt. Htut and colleagues characterised 96 patients with acute myeloid leukaemia harbouring KMT2A amplification, median age 68 [6]. Every case had a highly complex karyotype, TP53 alteration was present in over nine in ten patients, and all cases assessed by optical genome mapping showed chromoanagenesis involving 11q23. Survival was dismal — a median of about five and a half months at diagnosis and just over two months in relapsed or refractory disease — and intensive chemotherapy conferred no survival advantage over lower-intensity therapy, while venetoclax-based regimens were associated with modestly better overall and event-free survival. If you report cytogenetics, KMT2A amplification should be flagged as a TP53-associated, chemoresistant entity rather than lumped with KMT2A rearrangement, which it does not resemble biologically or clinically. In the same journal, Bataillon and colleagues tackled TP53-mutated multiple-classifier endometrial carcinomas — the awkward tumours that current algorithms assign to the POLE or mismatch repair-deficient bins [7]. Using shallow whole-genome sequencing in 33 cases, they found about a third were copy-number high, and those tumours were more often non-endometrioid, high-grade and advanced stage, with higher TP53 variant allele fractions. Over a short median follow-up of about 13 months, all recurrences and all disease-related deaths occurred in the copy-number-high group, with none among copy-number-low patients. That is a small cohort with immature follow-up, and the authors rightly call for multicentre validation, but it suggests genome-wide copy-number profiling can separate a genuinely aggressive subset that the four-tier classifier currently absorbs.

The final theme is entity definition and risk prediction. In The American Journal of Surgical Pathology, Tsai and colleagues sequenced 60 inflammatory hepatocellular adenoma-like hepatocellular carcinomas and 16 inflammatory adenomas, finding JAK-STAT pathway alterations in well over half of the carcinomas, with JAK1 mutation in about 37 percent [8]. JAK1-mutated tumours had a recognisable cytology — abundant eosinophilic cytoplasm, vesicular chromatin, prominent central nucleoli — and recurrent hotspots at S703, S729 and L910, with the first two associated with chronic viral hepatitis and cirrhosis in Asian patients. That morphology is a prompt to sequence, because JAK-targeted therapy becomes a conversation. Modern Pathology reports six cases of ERBB2-altered, S100 and SOX10-positive uterine sarcoma, a recently proposed entity that Lyu and colleagues flesh out [9]: mostly cervical stromal tumours with infiltrative interdigitating borders, brisk mitotic activity, and multinucleated giant cells with a floret-like nuclear arrangement in every case, all carrying oncogenic ERBB2 missense mutations including the canonical V777L variant plus CDKN2A loss. Behaviour was aggressive in several patients, and the alterations are actionable — so an S100 and SOX10-positive uterine spindle cell tumour deserves sequencing rather than a descriptive diagnosis. Finally, in Human Pathology, Ijaz and colleagues reviewed 46 patients whose most significant biopsy lesion was pleomorphic or florid lobular carcinoma in situ [10]. Roughly a quarter of the pleomorphic cases upgraded to invasive carcinoma at excision, and upgrade was significantly more common in the small group carrying germline pathogenic variants; half of the patients who later developed ipsilateral recurrence carried a variant. The numbers are small, but it supports genetic testing as part of the risk conversation in these lesions.

If you only have time for one paper this week, make it the orcein versus Verhoeff-Van Gieson comparison in Histopathology [1]. It answers a question that affects staging and adjuvant therapy decisions in one of the commonest cancers we report, and acting on it costs nothing more than a change of standing order in your laboratory.

Here are the key takeaways from this week in Pathology. First, for large vessel invasion in colorectal cancer, haematoxylin and eosin alone is not enough, and among elastin stains orcein is the more sensitive choice. Second, tissue selection is a diagnostic act: stain the largest mesenteric deposit for Ki67 in jejunoileal neuroendocrine tumours, and test the highest-grade focus for mismatch repair in multifocal prostate cancer. Third, transcription-factor subtyping is moving from lung small cell carcinoma into the bladder and biliary tract, and it carries real therapeutic biomarker consequences — including the caution that POU2F3-driven tumours may be negative for all conventional neuroendocrine markers. Fourth, KMT2A-amplified acute myeloid leukaemia is a TP53-enriched, chemorefractory subgroup that should be reported as such. And fifth, distinctive morphology remains the trigger for molecular testing, whether that is eosinophilic hepatocellular carcinoma with JAK1 hotspots or a floret-cell, SOX10-positive uterine sarcoma with targetable ERBB2 mutations.

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.