Article Text

Download PDFPDF

2022-RA-1463-ESGO An algorithm based on CD8 and CD68/PD-1 expression within the tumour immune microenvironment detected by multiplexed immunofluorescence can discriminate FIGO IA and IB endometrioid endometrial carcinomas
  1. Julian Sanz1,
  2. Inigo Espinosa2,
  3. Isabel Guerenu1,
  4. Teresa Iscar1,
  5. Enrique Chacon1,
  6. Francisco Guillen-Grima1,
  7. Antonio Gonzalez-Martin1,
  8. Luis Chiva1 and
  9. Carlos DeAndrea1
  1. 1Anatomia Patologica, Clinica Universidad Navarra, Madrid, Spain
  2. 2Anatomia Patologica, Hospital sant Pau, barcelona, Spain


Introduction/Background Most low-risk, early-stage disease endometrioid endometrial carcinomas (FIGO stage 1 EEC) have a good clinical outcome. 5–10% of these patients will suffer recurrence. The tumour immune microenvironment is closely associated with the tumour biology and has been shown to be a powerful prognostic tool in several tumour types.

Methodology A retrospective cohort of hysterectomy specimens (n=75) with FIGO stage I EEC from our institution were identified. 77,8% were stage IA and 22,2% stage IB. MELF (Microcystic, elongated, and fragmented) pattern of myoinvasion was detected in 21%. A TMA with 3 regions of interests (ROI) was constructed. Multiplexed immunofluorescence was used to assess staining for CD3, CD8, CD20, CD68, PD-1, PD-L1, FOXP3 and pan-cytokeratin on the Vectra Polaris™ platform and analyzed using QuPath software. Discriminant function analysis was performed to classify patients in ‘Stage IA’ vs. ‘Stage IB’, according to invasion. All the analyses were performed with IBM SPSS version 20 and Openepi. We fit a logarithmic curve to predict invasion in mm based on CD8.

Results Statistical analyses (chi-square test) showed that CD20, CD68, Foxp3, Foxp3/PD1, CD68/PD1 and CK/PD1 were associated (p<0.05) with MELF but not with depth of invasion (mm). The discriminant model (Stage IA vs IB) obtained the following equation: Classification score =1.081 + 0.002 (mean_CD8) – (0.005* mean_CD68_PDL1). Sensitivity and specificity for the score were: 68.42% (95% CI 52.54, 80.92) and 57.14% (95% CI 32.59–78.62) . Positive Predictive Value was 81.25%. We fitted a logarithmic curve for the prediction of invasion the model was the following mm invasion = 13.725 – 1.446 (ln(mean_CD8)) (figure).

Conclusion An algorithm based on combined CD8 and CD68/PD1 expression can discriminate the likelihood of Stage IA or stage IB tumours from a small tissue sample. In addition, mean of CD8 expression was associated with mm of invasion [AG1] (curve, figure).

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.