- Design
- retrospective single-institution cohort with integrated immunohistochemistry, targeted sequencing and shallow whole-genome sequencing
- Population
- 33 TP53-mutated multiple-classifier endometrial carcinomas profiled between 2022 and 2025
- Primary outcome
- copy-number status against clinicopathological features, recurrence and disease-related death
- Effect
- 12/33 (36.4%) copy-number-high; all 6 recurrences and all 3 deaths in that group at median 12.8 months follow-up
TP53-mutated multiple-classifier endometrial carcinomas are tumours carrying a TP53 alteration alongside a POLE mutation or mismatch repair deficiency; current ESGO and FIGO algorithms assign them to the POLE-mutated or mismatch repair-deficient category, which carries a favourable prognosis. Thirty-three such cases identified through routine molecular profiling between 2022 and 2025 were reanalysed with histopathology, immunohistochemistry, targeted sequencing and shallow whole-genome sequencing.
Copy-number-alteration-high status was prespecified as five or more large-scale genomic alterations, defined as gains or losses of 3 Mb or more within a chromosomal arm, excluding whole-arm changes. Twelve tumours (36.4%) were copy-number-high and 21 (63.6%) were copy-number-low. The copy-number-high tumours were more often non-endometrioid, high-grade and advanced-stage on FIGO 2023, and carried higher TP53 variant allele frequencies and higher ratios of that frequency to tumour cellularity.
Outcomes separated completely, and that is both the striking finding and its main limitation. All six recurrences and all three disease-related deaths occurred in the copy-number-high group; none occurred among the copy-number-low patients. Median follow-up, however, was 12.8 months — too short for endometrial carcinoma, where recurrence in a low-risk group would not yet be expected either way.
The practical implication today is interpretive rather than procedural. Shallow whole-genome sequencing is not a routine test and the authors call for validation in larger multicentre cohorts. But it argues for reporting TP53 variant allele frequency relative to tumour cellularity where that is available, and for treating the favourable classification of a multiple-classifier tumour as an algorithmic output rather than a biological conclusion — particularly in a tumour that is high-grade, non-endometrioid and advanced.
- Flag a multiple-classifier tumour that is non-endometrioid, high-grade or advanced-stage in the report rather than letting the algorithm's favourable category stand alone
- Report TP53 variant allele frequency alongside tumour cellularity where your panel provides both
- Do not order shallow whole-genome sequencing on this basis — it is not validated for this use
- Raise these cases at the multidisciplinary meeting, where the discordance between morphology and molecular class can be discussed
- Expect the definition of copy-number-high used here (five or more alterations of 3 Mb or more) to differ between studies
Why it matters
The molecular classification exists to avoid overtreating good-prognosis tumours, and multiple-classifier cases are where it may be doing the opposite.
Don't overread it
With 33 cases, nine events and under 13 months of follow-up, this identifies a hypothesis rather than a prognostic group.
The statistics, in plain English
Thirty-three cases with nine events total is too small to compare groups formally, and no hazard ratio is reported. That all events fell in one group looks decisive but is compatible with chance at these numbers — six recurrences distributed by luck would land this way reasonably often. A median follow-up of 12.8 months is the more serious limit: the copy-number-low group has simply not been observed long enough for its recurrences to appear. The copy-number threshold was prespecified, which is in the study's favour, but it was applied to the same cohort used to describe the outcome difference.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for pathology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free