Material and Methods: We retrospectively analyzed 129 meningioma cases resected between 2018 and 2024. Digital PR expression was quantified using the FDA-cleared Ventana uPath PR 1E2 algorithm, originally developed for breast carcinoma. Subsequently, tumors were stratified by age, sex, WHO grade, histological subtype, and localization, while Ki-67 index and mitotic counts were also recorded. Associations with digital PR H-scores were evaluated using non-parametric tests, correlation analysis, and ordinal logistic regression.
Results: Digital PR H-scores progressively decreased across WHO grades, with Grade 1 tumors showing the highest values, followed by Grade 2 and Grade 3 (p=0.002). PR expression was inversely correlated with proliferative markers, including Ki-67 index (ρ = –0.42, p<0.001) and mitotic count (ρ = –0.35, p<0.01). No significant differences were observed by age or sex. Convexity meningiomas tended to have higher scores than skull base and spinal tumors.
Conclusion: Digital PR H-score assessment confirmed the inverse association between PR expression and WHO grade, as well as its correlation with proliferative activity. Using an FDA-cleared algorithm originally developed for breast carcinoma, this method provides objective and reproducible evaluation in meningiomas.
In most published reports, PR expression has been evaluated by manual IHC scoring, which is affected by inter-observer variation and the typical limitations of semi-quantitative assessment. Indeed, reviews note that IHC scoring of PR expression is often subjective and inconsistent[6], and studies evaluating inter-observer agreement in PR scoring report only moderate correlation (e.g. r = 0.63) between pathologists [7]. Digital pathology provides an opportunity to assess PR expression more consistently using digital Hscore methods, although its application in meningiomas has been relatively limited to date. Integrating digital PR scoring with clinicopathological parameters—including age, sex, tumor localization, molecular subtypes, and proliferation indices—may provide refined insights into tumor behavior and prognostication. This study therefore aims to evaluate the digital H-score of PR expression in a large series of meningiomas, and to elucidate its associations with WHO grade, histological subtype, location, Ki-67 index, mitotic activity, and demographic factors.
Digital PR Evaluation and H-score Calculation
Immunohistochemical staining for progesterone receptor
(PR) was performed using the Ventana® PR (clone 1E2)
assay. Whole-slide images of PR-stained sections were
acquired with the Ventana DP200 digital slide scanner at
×20 magnification. Digital analysis was conducted with
the FDA-cleared Ventana uPath PR 1E2 algorithm, originally
developed for breast carcinoma. The algorithm has
been previously compared with manual scoring to demonstrate
its applicability to meningioma specimens[8]. For
each case, three representative regions of interest (ROIs)
measuring approximately 5 mm² were selected from tumor
areas by a pathologist, avoiding necrosis, hemorrhage, or
non-neoplastic tissue. The software automatically quantified
nuclear staining intensity as weak, moderate, or strong.
The H-score was then calculated for each ROI using the
formula (% weak × 1) + (% moderate × 2) + (% strong × 3),
resulting in a continuous score ranging from 0 to 300. For
each case, the mean of the three ROIs was used as the final
digital PR H-score.
Demographic Data
For statistical analysis, patients were stratified into three
age groups: <40 years (n=24), 40–59 years (n=50), and ≥60
years (n=55). The study cohort included 90 female and 39
male patients (female-to-male ratio: ~2.3:1).
Tumor Localization
Tumor localization was categorized into five groups: convexity
(n=66; including frontal, parietal, temporal, occipital
convexity and intraventricular meningiomas, as well as
those described as arachnoid in origin), parasagittal/falx
(n=9), skull base (n=36; including suprasellar, sphenoid
wing, clinoidal, petroclival, olfactory groove, tentorial,
foramen magnum, cerebellopontine angle, optic nerve–
adjacent, and other cranial base sites including medulla
oblongata), spinal (n=14; including cervical, thoracic, lumbar,
and sacral as well as vertebra-related and posterior cerebral/cerebellar artery–adjacent lesions), and unknown/
other (n=4).
Histopathological Classification
Histopathological classification was performed according
to the 2021 WHO Classification of Tumours of the Central
Nervous System (CNS5)[1]. Accordingly, meningiomas
were assigned as Grade 1 (benign), Grade 2 (atypical/intermediate),
or Grade 3 (anaplastic/malignant).
• Grade 1 tumors (n=97): meningothelial (n=27), transitional (n=50), fibrous (n=6), psammomatous (n=7), angiomatous (n=4), and microcystic (n=3).
• Grade 2 tumors (n=26): atypical meningiomas (n=23), chordoid meningiomas (n=2), and clear cell meningioma (n=1).
• Grade 3 tumors (n=6): anaplastic (n=5), one papillary meningioma (n=1). No rhabdoid subtype was identified.
Proliferation Markers
Ki-67 proliferation index was recorded in all 129 cases (median
2%, range 1–40). Mitotic count was available for all
tumors (median 1 per 10 HPF, range 0–47). Distribution
according to WHO-relevant thresholds was as follows: 116
tumors with 0–3 mitoses, 10 tumors with 4–19 mitoses,
and 3 tumors with ≥20 mitoses per 10 HPF.
The detailed distribution of demographic, clinical, and pathological characteristics of the study cohort is summarized in Table I.
Statistical Analysis
All clinicopathological variables, including age, sex, tumor
localization, WHO grade, histological subtype, Ki-67 index,
and mitotic count, were included in the statistical analyses.
Continuous variables were summarized as median [IQR]
and categorical variables as counts (%). Between-group
comparisons of digital PR H-score across WHO grade, histological
subtype, and location categories were performed
using the Kruskal–Wallis test with Dunn`s post-hoc tests;
p-values were adjusted for multiple comparisons using the
Benjamini–Hochberg procedure. Correlations between Hscore
and proliferative markers (Ki-67, mitotic count) were
assessed with Spearman`s rank correlation. Ordinal logistic
regression models were fitted to examine the independent
association between H-score and WHO grade (1<2<3)
while adjusting for prespecified covariates. All tests were
two-sided with α=0.05. Analyses were conducted in Python
(pandas, SciPy, statsmodels).
Association with Histological Subtypes
When analyzed by histological subtypes according to the
2021 WHO classification, Grade 1 variants (meningothelial,
transitional, fibrous, psammomatous, angiomatous,
and microcystic) showed the highest digital PR H-scores.
In contrast, Grade 2 tumors (atypical, chordoid, and clear
cell) showed significantly lower scores, while Grade 3 tumors
(anaplastic and papillary) had the lowest values.
There was a statistically significant difference among the
histological subtypes (Kruskal–Wallis p<0.01). Post-hoc
analysis revealed that Grade 1 subtypes differed significantly
from both Grade 2 and Grade 3 tumors (p<0.05 for each
comparison), whereas the difference between Grade 2 and
Grade 3 subtypes was not statistically significant. Within
Grade I variants, fibrous and microcystic meningiomas exhibited
comparatively lower digital PR H-scores (fibrous:
median 114, IQR 105–148; microcystic: median 134, IQR
118–165) relative to other Grade I subtypes, although these differences did not reach statistical significance after multiple-
testing correction. Representative examples are shown
in Figure 2A-D.
Association with Tumor Localization
When stratified by anatomical site, digital PR H-scores
varied across localization groups. The highest scores were
observed in convexity meningiomas (n=66), followed by
parasagittal/falx (n=9) and skull base tumors (n=36). Spinal
meningiomas (n=14) and the miscellaneous/unknown
group (n=4) had the lowest H-scores. The overall difference
across groups reached statistical significance (Kruskal–Wallis
p<0.05). However, post-hoc pairwise comparisons did
not demonstrate statistically significant differences between
individual localization groups after correction for multiple
testing. These findings suggest that digital PR expression
may be relatively higher in convexity meningiomas, whereas
spinal and other rare sites show lower levels, although the
small sample sizes limit definitive conclusions.
Association with Age and Sex
When stratified by age categories, median H-scores were
198 in patients <40 years (n=24), 184 in those aged 40–59 years (n=50), and 176 in patients ≥60 years (n=55). Although
there was a trend toward lower H-scores in older
patients, the differences across age groups did not reach
statistical significance (Kruskal–Wallis p>0.05).
Digital PR H-scores were similar between sexes, with median values of 192 in female patients (n=90) and 188 in male patients (n=39). No significant sex-related differences in H-scores were detected (Mann–Whitney p>0.05).
Association with Proliferative Markers
Digital PR H-scores demonstrated a significant negative
correlation with proliferative activity. H-score correlated
inversely with Ki-67 index (Spearman`s ρ = –0.42, p<0.001),
indicating that tumors with higher proliferative labeling
tended to show lower PR expression. Similarly, an inverse
correlation was observed with mitotic count (Spearman`s
ρ = –0.35, p<0.01). When categorized, tumors with Ki-67
≤3% exhibited higher H-scores compared to those with
4–10% or >10%, and a comparable trend was observed
across increasing mitotic count strata (0–3, 4–19, ≥20 per
10 HPF), although groupwise differences were attenuated
by the small number of high-mitotic tumors.
Multivariable Analysis
In ordinal logistic regression including age, sex, localization,
histological subtype, Ki-67 index, and mitotic count
as covariates, digital PR H-score remained an independent
predictor of WHO grade. Each 10-point increase in
H-score was associated with a 15% reduction in the odds
of being classified as a higher WHO grade (OR 0.85, 95%
CI 0.76–0.95, p=0.004). Neither age nor sex showed significant
associations with grade, whereas both Ki-67 index
(p=0.01) and mitotic count (p=0.02) were independently
linked to higher WHO grade. These findings indicate that
lower digital PR expression is an independent marker of
tumor aggressiveness.
All statistical comparisons of digital PR H-scores with clinicopathological variables, including WHO grade, histological subtype, tumor localization, age, sex, Ki-67 index, mitotic count, and multivariable regression results, are summarized in Table II.
Table II: Statistical Associations of Digital PR H-scores with Clinicopathological Variables
A combined overview of demographic, pathological, and statistical findings is presented in Table III, providing an integrated summary of the study results.
Table III: Combined Overview of Demographics, Pathological Variables, and Key Statistical Findings
The inverse correlation between PR expression and proliferative indices such as Ki-67 and mitotic count has been well documented. Our results corroborate these associations: tumors with higher Ki-67 or mitotic activity had significantly lower digital PR H-scores. This is in line with reports that PR-negative meningiomas are more likely to be atypical or anaplastic and to recur earlier[4,9].
Localization and histological subtype have been variably linked to PR expression in prior reports[2,3]. In our series, convexity meningiomas tended to display higher PR scores compared with skull base and spinal tumors, although these differences did not remain significant in post-hoc tests, likely reflecting limited sample sizes. Among Grade 1 subtypes, fibrous meningiomas showed a trend toward lower PR expression, similar to earlier observations that fibrous morphology may be associated with lower PR levels [4]. The microcystic Grade I variant likewise exhibited comparatively lower digital PR expression (median 134, IQR 118–165; vs overall Grade I median 206); however, these subtype differences did not persist after multipletesting adjustment, likely owing to the rarity of the microcystic variant, and its small representation in our cohort (n=3). Collectively, these patterns point to subtype-specific heterogeneity in PR signaling.
A major strength of this study is the use of an FDA-cleared digital algorithm (Ventana uPath PR 1E2), which was originally validated for breast carcinoma. Applying this standardized platform to meningiomas reduces observer bias and increases reproducibility, providing a methodological advance over manual, semi-quantitative scoring. In addition, the adoption of an intensity and percentage-weighted H-score approach ensured further standardization and objectivity in quantifying PR expression. Furthermore, the relatively large sample size (n = 129) and the integration of multiple clinicopathological variables strengthen the robustness of our findings.
This study has some limitations. Molecular markers such as TERT promoter mutations or CDKN2A/B deletions, which are now included in the 2021 WHO classification, were not available for analysis[1]. In addition, Ki-67 evaluation was performed manually rather than digitally, which may introduce variability. Finally, the small number of Grade 3 cases (n=6) limits the statistical power to detect subtle differences between atypical and anaplastic tumors.
Despite these limitations, our results confirm that declining PR expression is closely associated with higher WHO grade and tumor progression in meningiomas. Beyond methodological advantages, digital PR scoring thus holds promise as a clinically relevant biomarker that could aid in prognostication and, with further validation, inform therapeutic strategies targeting hormone-related pathways.
Declaration of Generative AI Use
The authors declare that generative artificial intelligence (ChatGPT,
OpenAI, San Francisco, CA, USA) was used to assist in language
editing and improving the clarity of the manuscript. The authors
reviewed, verified, and approved all content generated, and take full
responsibility for the integrity and accuracy of the work.
Funding
This research did not receive any specific grant from funding agencies
in the public, commercial, or not-for-profit sectors.
Conflict of Interest
The authors declare no conflicts of interest.
Authorship Contributions
Concept: YA, Design: YA, Supervision: YA, Data collection and/
or processing: YA, EBG, Analysis and/or interpretation: YA, EBG,
Literature search: YA, Writing: YA, Approval: YA, EBG.
1) WHO Classification of Tumours Editorial Board: Central Nervous
System Tumours. 5th ed., Lyon, International Agency for
Research on Cancer, 2021.
2) Agopiantz M, Carnot M, Denis C, Martin E, Gauchotte G.
Hormone Receptor Expression in Meningiomas: A Systematic
Review. Cancers (Basel). 2023;15(3):980. doi: 10.3390/cancers15030980.
3) Maiuri F, Mariniello G, de Divitiis O, Esposito F, Guadagno E,
Teodonno G, Barbato M, Del Basso De Caro M. Progesterone
Receptor Expression in Meningiomas: Pathological and Prognostic
Implications. Front Oncol. 2021;11:611218. doi: 10.3389/
fonc.2021.611218.
4) Mukherjee S, Ghosh SN, Chatterjee U, Chatterjee S. Detection
of progesterone receptor and the correlation with Ki-67 labeling
index in meningiomas. Neurol India. 2011;59(6):817-22. doi:10.4103/0028-3886.91357. PMID: 22234191.
5) Mukhopadhyay M, Das C, Kumari M, Sen A, Mukhopadhyay B,
Mukhopadhyay B. Spectrum of meningioma with special reference
to prognostic utility of ER, PR and Ki67 expression. J Lab
Physicians. 2017;9(4):308-13. doi: 10.4103/JLP.JLP_158_16.
6) Li Z, Bui MM, Pantanowitz L. Clinical tissue biomarker digital
image analysis: A review of current applications. Hum Pathol
Rep. 2022;28:300633. doi:10.1016/j.hpr.2022.300633
7) Prasad K, Tiwari A, Ilanthodi S, Prabhu G, Pai M. Automation
of immunohistochemical evaluation in breast cancer using image
analysis. World J Clin Oncol. 2011;2(4):187-94. doi: 10.5306/
wjco.v2.i4.187. PMID: 21611095; PMCID: PMC3100486.
8) Akca Y, Gokce EB. Evaluation of a breast cancer–trained digital
progesterone receptor scoring algorithm in meningiomas: A
comparative study with manual assessment. Ann Diagn Pathol.
2025;152577. doi:10.1016/j.anndiagpath.2025.152577
){kind=link}
){kind=link}
){kind=link}
){kind=link}
){kind=link}