Documentation
Methods and interpretation
WHO 2020 framework, Swedish national guidelines v3.2.
Classification algorithm
The WHO 2020 hierarchical sequential classifier assigns one primary subtype in strict priority order. If a sample meets criteria for multiple subtypes, the highest-priority class is assigned and the others are reported as secondary features.
| Priority | Subtype | Prognosis | Clinical implication |
|---|---|---|---|
| 1 | POLEmut | Best | No adjuvant treatment for stage I-II (Swedish guidelines) |
| 2 | MMRd | Intermediate | Checkpoint inhibitor benefit; Lynch syndrome screening |
| 3 | p53abn | Worst | Postoperative chemotherapy for stage I-II |
| 4 | NSMP | Intermediate | Diagnosis of exclusion; ER status sub-stratifies |
POLE assessment
Pathogenic mutations in the POLE exonuclease domain (exons 9-10-11-12-13-14, codons 268-471) define the POLEmut subtype.
Tier 1, established hotspots
Auto-classified as POLEmut regardless of other features.
Tier 2, confirmed pathogenic
Auto-classified as POLEmut. Based on Leon-Castillo et al. 2020.
VUS in exonuclease domain
Missense variants within codons 268-471 not in the curated list are flagged as VUS. They are not auto-classified. Secondary evidence required: TMB >100 mut/Mb, SBS10a/SBS10b, C>A >20%.
MMR / MSI assessment
MMR deficiency is assessed from three independent sources, in priority order:
1. MMR immunohistochemistry (gold standard)
When IHC results are provided, loss of any MMR protein (MLH1, MSH2, MSH6, PMS2) directly classifies as MMRd with high confidence. Lynch screening guidance is tailored to which protein is lost.
2. MSI marker percentage
- MSI-H: ≥20% unstable markers (configurable) → MMRd
- MSI-L: >0% but <20% → not classified as MMRd
- MSS: 0% → not classified as MMRd
3. MMR gene mutations + cMS proxy
When MSI markers are unavailable, frameshift indels at 11 high-specificity coding microsatellite indicator genes are counted. Biallelic MMR gene inactivation (2+ pathogenic hits) also supports MMRd.
TP53 assessment
DNA-binding domain: codons 102-292.
Pathogenic variants
Curated hotspots: R175H G245S R248W R248Q R249S R273H R273C R282W
All missense in the DNA-binding domain and all truncating variants (nonsense, frameshift, splice-site) are classified as pathogenic.
p53 IHC
When available, aberrant p53 IHC (overexpression or null pattern) directly classifies as p53abn, even without a detectable point mutation. This captures TP53 deletions and LOH that sequencing misses (~6% of p53abn cases).
Exclusions
- P72R/R72P: benign polymorphism, excluded.
- Low-VAF variants: below 5% VAF, flagged but not auto-classified.
Secondary evidence
Each classification is scored against expected genomic features. Concordance strengthens confidence; discordance may indicate misclassification or unusual biology.
| Subtype | TMB | CNA | Spectrum |
|---|---|---|---|
| POLEmut | >100 mut/Mb | Low | C>A >20%, indels <5% |
| MMRd | 10-100 | Low-int | Indel fraction >15% |
| p53abn | <10 | High | Clock-like (SBS1, SBS5) |
| NSMP | <10 | Low | Clock-like (SBS1, SBS5) |
Confidence levels
| Level | Meaning |
|---|---|
| High | Primary classifier + all secondary evidence concordant |
| Moderate | Primary classifier present, some evidence discordant or unavailable |
| Low | Borderline: POLE VUS, single heterozygous MMR hit, TP53 VUS |
| Discordant | Primary classifier contradicted by secondary evidence |
Limitations
- Research and clinical decision support only. Final classification must be confirmed by a molecular pathologist with histopathology and clinical context.
- MMR from sequencing is less reliable than IHC. Panel sequencing may miss MLH1 promoter methylation, large deletions, and other mechanisms of MMR loss.
- POLE VUS require secondary evidence (TMB >100, SBS10a/b) before classification.
- TP53 sequencing does not replace p53 IHC. ~92% concordance. TP53 deletions/LOH appear wild-type by sequencing but aberrant by IHC, particularly in serous carcinomas.
- Signature decomposition from small panels may be unreliable. The tool uses 6-class spectrum analysis as a more robust alternative.
- Multiple classifiers (3-6% of cases) are reported with all features. Prognosis follows the highest-priority class.
References
- Swedish national guidelines: Nationellt vardprogram livmoderkroppscancer v3.2 (2026-03-10)
- ESGO-ESTRO-ESP guidelines update 2025 (Lancet Oncology)
- FIGO 2023 staging system for endometrial cancer
- Leon-Castillo A, et al. J Pathol. 2020;250(3):323-335.
- Leon-Castillo A, et al. J Pathol. 2020;250(4):413-422.
- Cancer Genome Atlas Research Network. Nature. 2013;497(7447):67-73.
- WHO Classification of Female Genital Tumours, 5th ed. (2020)