Progesterone Receptor Isoforms as Predictors of Endometrial Hyperplasia Relapse

Study Overview

• Type: Retrospective cohort analysis nested within a prior Norwegian multicentre randomised trial (2005–2012).

• Focus: Determine whether pre-treatment immunohistochemical (IHC) expression of progesterone receptor isoforms PR-A and PR-B predicts relapse after successful 6-month progestin therapy for endometrial hyperplasia (EH).

• Setting: University Hospital of North Norway and collaborating centres; ethics approval $ext{REK NORD 25/2004}$; trial registry $ext{NCT01074892}$.

Background & Rationale

• EH = recognised precursor of endometrial carcinoma (EC); untreated progression risk ≈ $20\%$ (Kurman 1985).

• Pathogenesis: continuous estrogen exposure unopposed by progesterone.

• Progestin therapy (especially LNG-IUS) shows dose-dependent regression but relapse occurs once therapy stops; reported relapse rates $13.7\text{–}41\%$.

• Nuclear progesterone receptors (PRs) mediate anti-proliferative actions. Two main isoforms:

  • PR-A (truncated)

  • PR-B (full-length, transcriptionally stronger)

• Balanced PR-A & PR-B expression needed for normal endometrial differentiation; imbalance observed in early carcinogenesis.

• Few human data on PR isoforms as predictors of relapse. Present study addresses this gap.

Trial Design & Population

• Original RCT enrolled women aged $30\text{–}70$ years with WHO-94 low- or medium-risk EH.

• Randomisation ➔ three 6-month progestin regimens:

  1. LNG-IUS $52\,\text{mg}$ (Mirena®)

  2. Oral medroxyprogesterone acetate (MPA) $10\,\text{mg}$ daily (continuous)

  3. Oral MPA $10\,\text{mg}$ for 10 days/cycle (cyclic)

• Numbers:

  • $n=153$ treated

  • $n=135$ ( $88\%$ ) achieved complete histological response after 6 months.

  • $n=135$ entered 24-month post-withdrawal surveillance with Pipelle® resampling every 6 months.

  • Relapse observed in $n=55$.

• IHC material available for $n=94$ responders (insufficient tissue in $n=41$); final analytic set:

  • Relapse $n=40$

  • No-relapse $n=54$

Baseline Characteristics (pre-treatment)

• Age: Relapse group younger (mean $45.4\,\text{y}$) vs no-relapse (mean $48.4\,\text{y}$).

• Menopausal status: $82.5\%$ pre-menopausal in relapse group vs $48.1\%$ in no-relapse.

• Estradiol: median $0.34\,\text{nmol·L}^{-1}$ vs $0.16\,\text{nmol·L}^{-1}$.

• WHO-94 histology: Atypical hyperplasia (AH) more common in relapse ( $15\%$ ) than no-relapse ( $6\%$ ).

• BMI: similar means ( $27.5$ vs $26.7\,\text{kg·m}^{-2}$ ) but higher proportion $\ge 26\,\text{kg·m}^{-2}$ in relapse group.

Immunohistochemistry Protocol

• Tissue: $4\text{–}5\,\mu\text{m}$ paraffin sections on Superfrost + slides.

• De-paraffinisation & antigen retrieval in Ventana Benchmark Ultra (CC1 buffer, $95^{\circ}\text{C}$, $48\,\text{min}$).

• Primary antibodies (dilution $1{:}150$):

  • PR-A: Clone 16 (Novocastra)

  • PR-B: Clone hPRa 2 (ThermoFisher)

• Incubation $60\,\text{min}$ at $37^{\circ}\text{C}$; DAB detection; haematoxylin counter-stain.

Scoring System

• Histological score (H-score): $HS = \sum(P_i \times i)/100$

  • $P_i$ = percentage of positive nuclei

  • $i$ = intensity ( $1=\text{weak}, 2=\text{moderate}, 3=\text{strong}$ )

  • Range $0\text{–}3$

• Separate assessment in endometrial glands & stroma.

• Two blinded assessors (gynaecologic pathologist & chief engineer).

Statistical Analysis

• Outcome: Binary relapse (Yes/No) within 24 months.

• Tests:

  • Univariable & multivariable logistic regression.

  • Covariates: age, WHO-94 class, BMI, menopausal status (estradiol excluded—collinear).

  • Receiver Operating Characteristic (ROC) analysis ➔ Area Under Curve (AUC).

• Interpretation guide: \text{AUC} <0.5 not useful; $0.8\text{–}0.9$ very good; etc.

• Software: IBM SPSS v24; significance $\alpha =0.05$.

Main Results

• Mean H-scores (relapse vs no-relapse):

  • PR-A glands: lower in relapse ($P=0.013$)

  • PR-A stroma: markedly lower (P<0.001)

  • PR-B glands: higher in relapse ($P=0.001$)

  • PR-B stroma: no difference ($P=0.720$)

Logistic Regression (key adjusted odds ratios)

• PR-A stroma: $\text{OR}=0.15$ per 1-unit H-score increase ( 95\%\,\text{CI }0.05\text{–}0.39; P<0.001 ).

• PR-B glands: $\text{OR}=2.91$ per 1-unit increase ( $95\%\,\text{CI }1.11\text{–}7.62; P=0.030$ ).

• Age & menopausal status lost significance after multivariable adjustment.

Subgroup Findings (by Treatment Arm)

• Stromal PR-A predictive across all regimens:

  • LNG-IUS: $\text{OR}=0.21$

  • MPA continuous: $\text{OR}=0.10$

  • MPA cyclic: $\text{OR}=0.14$

• Glandular PR-B significant with LNG-IUS ( $\text{OR}=4.38$ ) & cyclic MPA ( $\text{OR}=8.30$ ).

PR-A : PR-B Ratio Analysis

• Each $0.1$ increase in $\dfrac{\text{PR-A}<em>{\text{glands}}}{\text{PR-B}</em>{\text{glands}}}$ ➔ $19\%$ lower odds of relapse (adjusted $P=0.002$).

• Combined glands + stroma ratio:

  • ROC AUC $=0.771$ (moderate diagnostic value).

  • Optimal cut-off $\le 1$ ➔ Sensitivity $75\%$, Specificity $78\%$.

  • $\text{OR}=11.05$ for relapse when $\text{PR-A:PR-B}\le1$ vs >1.

  • Cumulative relapse rates: $71\%$ (ratio $\le1$) vs $19\%$ (ratio >1).

Mechanistic & Translational Insights

• Animal data: PR-A essential for anti-proliferative regulation; PR-B can drive EH under estrogenic milieu (Mulac-Jericevic 2000).

• PR-A may directly inhibit estrogen receptor signalling; loss favours continued proliferation.

• Stromal PR-A appears critical; stromal microenvironment modulates epithelial progestin response.

• Overall, PR-A deficit ± PR-B excess create a pro-growth, pro-relapse milieu.

Strengths

• Nested within a rigorously conducted RCT; uniform 24-month biopsy follow-up.

• Separate quantification of isoforms in glands vs stroma.

• Blinded dual pathology review; standardised H-score.

• Practical, low-cost biomarker easily added to routine diagnostic workflow.

Limitations

• Therapy heterogeneity (three regimens); however, relapse previously shown regimen-independent and subgroup trends consistent.

• $41$ women excluded for insufficient tissue (may introduce selection bias).

• Lack of core outcome set in EH literature hampers direct comparison with other studies.

• Observational nature ⇒ cannot prove causality; validation in larger prospective cohorts needed.

Clinical Implications

• Pre-treatment IHC for PR-A & PR-B can stratify relapse risk.

• Women with $\text{PR-A:PR-B}\le1$ may benefit from:

  • Extended progestin therapy beyond 6 months.

  • More frequent surveillance post-withdrawal.

• Could reduce overtreatment in low-risk (ratio >1) patients.

• Supports move toward personalised EH management.

Research Directions

• Prospective validation of PR-A/PR-B ratio as a decision tool.

• Explore modulators that up-regulate PR-A or down-regulate PR-B.

• Functional co-culture studies to dissect stromal–glandular cross-talk.

• Integration with other biomarkers (ER, COX-2, mismatch repair) for multiparametric models.

Quick Reference: Key Numbers & Equations

• H-Score formula: $HS = \dfrac{\sum(P_i \times i)}{100}$

• Relapse odds (adjusted):

  • $\text{OR}_{\text{PR-A stroma}} = 0.15$ per 1-unit increase

  • $\text{OR}_{\text{PR-B glands}} = 2.91$ per 1-unit increase

• Ratio cut-off: $\text{PR-A:PR-B} \le 1$ ➔ $\text{LR}^+ = 3.4$, $\text{LR}^- = 0.32$

• ROC interpretation scale: Not useful <0.5, Bad $0.5\text{–}0.6$, Sufficient $0.6\text{–}0.7$, Good $0.7\text{–}0.8$, Very good $0.8\text{–}0.9$, Excellent >0.9.

These bullet-point notes consolidate all major & minor findings, methodological details, statistical outcomes, mechanistic explanations, and clinical implications from the transcript so they can serve as a standalone study guide.