FRα expression in borderline ovarian tumors and low-grade serous ovarian carcinoma: prognostic and immune microenvironmental associations

Abstract

Background

Folate receptor alpha (FRα) is overexpressed in various epithelial ovarian carcinomas and has emerged as a potential therapeutic target. However, its expression pattern and clinical significance in borderline ovarian tumors (BOTs) and related low-grade serous tumors remain unclear. This study aimed to investigate the association between FRα expression and clinicopathologic characteristics of BOTs and low-grade serous ovarian carcinomas (LGSOC), and to explore the features of the tumor immune microenvironment in relation to FRα expression.

Results

A total of 100 surgically resected ovarian epithelial tumors (2014–2024) were retrospectively analyzed, including 91 BOTs and 9 LGSOCs. FRα expression was heterogeneous across histologic subtypes. High FRα expression (immunoreactive score ≥ 8) was detected in 13% (13/100) of cases and was predominantly observed in serous tumors (15.6%, 12/77). Within the serous subset, tumors with micropapillary features showed higher FRα expression, approaching levels observed in LGSOCs. FRα-high cases demonstrated a significantly higher recurrence rate compared with FRα-low tumors (42% vs. 14%, P = 0.003), particularly in serous subset. FRα expression was inversely correlated with immune cell infiltration and positively correlated with PD-L1 (r = 0.36, P < 0.001) and PD-1 expression. Moreover, a high proportion of FRα⁺PD-L1⁺ double-positive tumor epithelial cells in serous tumors was associated with shorter disease-free survival.

Conclusions

FRα expression was heterogeneous across BOT subtypes and tended to be higher in SBOTs with micropapillary features and LGSOCs. High FRα expression is associated with tumor recurrence and an immunosuppressive microenvironment, suggesting that FRα may serve as a prognostic biomarker and a potential therapeutic target in BOTs and LGSOCs.

Introduction

Borderline ovarian tumors (BOTs) represent a distinct category of epithelial ovarian neoplasms characterized by epithelial proliferation and nuclear atypia in the absence of destructive stromal invasion [1]. Although the prognosis is generally favorable, BOTs exhibit heterogeneous clinical behavior, with recurrence occurring in a subset of patients, particularly among serous tumors with micropapillary features, which share biological and clinical similarities with low-grade serous ovarian carcinomas (LGSOCs) [2, 3]. Reliable biomarkers that improve risk stratification and guide surveillance are lacking.

Folate receptor alpha (FRα), encoded by FOLR1, is a high-affinity folate-binding protein overexpressed in several epithelial malignancies, including high-grade serous ovarian carcinoma (HGSOC) [4]. In addition to serving as a prognostic factor, FRα has emerged as a clinically meaningful therapeutic target, with multiple FRα-directed antibody–drug conjugates (ADCs) demonstrating encouraging efficacy in recent clinical trials [5, 6]. Recent clinical results from FRα-targeted ADC programs—including AZD5335 in platinum-resistant recurrent ovarian cancer—have demonstrated antitumor activity across varying FRα expression levels [7]. These developments have encouraged strong translational interest in FRα and support further exploration of its biological and clinical significance across the spectrum of serous ovarian neoplasia.

Despite these developments, the pattern of FRα expression in BOTs and LGSOCs and its relationship with recurrence remain unclear, and data on expression differences across BOT histologic subtypes and in LGSOCs are limited. Furthermore, the relationship between FRα and the tumor immune microenvironment has not been investigated in BOTs and LGSOC within a unified analytical framework. Advances in multiplex immunofluorescence (mIF) and spatial analytic platforms now enable high-resolution characterization of immune cell subsets and their spatial interactions with tumor cells, providing an opportunity to contextualize FRα expression within the immune landscape [8]. However, a comprehensive characterization of FRα expression and its spatial and clinical relevance in BOTs and LGSOCs is currently lacking.

In this study, we examined FRα expression in a well-annotated study cohort comprising BOTs and LGSOCs using immunohistochemistry and multiplex immunofluorescence to further explore the clinical and immune microenvironmental associations of FRα. We evaluated how varying levels of FRα expression are related to clinicopathologic features and recurrence, and explored whether FRα expression or its spatial relationships with immune cell populations are related to patient outcomes.

Methods

Study design and patient cohort

This retrospective study included patients with ovarian epithelial tumors who underwent primary surgical resection at Jiangsu Province Hospital between October 2014 and November 2024. A total of 121 cases were initially reviewed. Twenty-one patients were excluded because of a diagnosis of invasive ovarian carcinoma, including high-grade serous ovarian carcinoma (HGSOC, n = 9), mucinous ovarian carcinoma (MOC, n = 8), and seromucinous carcinoma (SMOC, n = 4). The remaining 100 patients composed the overall study cohort (Fig. 1A), including 91 BOTs and 9 LGSOCs. LGSOC cases were intentionally retained in the overall study cohort for analyses of the serous spectrum. All histopathological diagnoses were independently confirmed by two experienced gynecologic pathologists according to the World Health Organization (WHO) classification criteria.

Fig. 1.

Study cohort selection and analytical workflow A Inclusion and exclusion criteria and cohort composition of borderline ovarian tumors (BOTs). A total of 121 ovarian epithelial tumors were initially reviewed. After exclusion of invasive ovarian carcinomas, including high-grade serous ovarian carcinoma (HGSOC), mucinous ovarian carcinoma (MOC), and seromucinous carcinoma (SMOC), 100 tumors were included in the final analytical cohort, comprising 91 BOTs and 9 LGSOCs. BOTs were classified as serous (SBOT), mucinous (MBOT), and seromucinous (SMBOT) subtypes. Serous tumors were further subclassified as typical SBOT and SBOT with micropapillary pattern (SBOT-MP) based on histopathologic review. The cohort included mucinous BOTs (MBOTs), seromucinous BOTs (SMBOTs), and tumors within the serous spectrum. The serous spectrum comprised typical SBOT, SBOT-MP, and LGSOC B Schematic overview of the study workflow. Tissue microarrays (TMAs) were constructed from formalin-fixed, paraffin-embedded specimens. FRα expression was assessed by immunohistochemistry on TMA sections and quantified using both the immunoreactive score (IRS) and H-score. IRS categories were used to describe FRα expression patterns across histologic groups, whereas the cohort median H-score was used to dichotomize tumors into FRα-high and FRα-low groups for subsequent clinicopathologic and survival analyses. Serial sections were subjected to multiplex immunofluorescence (mIF) staining for FRα, PD-1, PD-L1, PanCK, and immune markers. Whole-slide imaging and cell-level segmentation within PanCK-defined epithelial compartments were performed to enable quantitative phenotyping of tumor and immune cell populations. Abbreviations: FRα, folate receptor alpha; IRS, immunoreactive score; TMA, tissue microarray; mIF, multiplex immunofluorescence

This study was approved by the Institutional Review Board of Jiangsu Province Hospital (approval No. 2025-SR-452). The requirement for informed consent was waived because this was a retrospective study using archived formalin-fixed, paraffin-embedded specimens. All procedures were conducted in accordance with the Declaration of Helsinki.

Pathologic review and histologic classification

All cases were reviewed independently by two experienced gynecologic pathologists and classified according to the World Health Organization (WHO) criteria. The overall study cohort included 100 patients, comprising 68 serous borderline ovarian tumors (SBOTs), 17 mucinous borderline ovarian tumors (MBOTs), 6 seromucinous borderline ovarian tumors (SMBOTs), and 9 LGSOCs.

Within the SBOTs subset (n = 68), cases were further subclassified as typical SBOTs (n = 32), or serous borderline ovarian tumor with micropapillary pattern (SBOT-MP, n = 36), as shown in the study flow diagram (Fig. 1A). SBOTs with micropapillary pattern were diagnosed according to established and widely accepted criteria, as defined in the WHO classification, characterized by a hierarchical micropapillary architecture lacking fibrovascular cores and measuring at least 5 mm in greatest dimension or comprising ≥ 10% of the tumor.

Clinical data and follow-up

Clinicopathologic data were retrieved from electronic medical records and included age at diagnosis, serum tumor markers, International Federation of Gynecology and Obstetrics (FIGO) stage, surgical procedures, and recurrence status. Disease-free survival (DFS) was defined as the interval from the date of initial surgery to the date of documented recurrence or the last follow-up. The median follow-up duration for the overall study cohort was 41 months.

FRα immunohistochemistry

FRα immunohistochemistry was performed on formalin-fixed, paraffin-embedded (FFPE) tissue microarray (TMA) Sect.  (4 μm thick). After deparaffinization in xylene and rehydration through graded ethanol, antigen retrieval was performed by heating the sections in citrate buffer (pH 6.0) at 95–100 °C for 20 min, followed by cooling at room temperature. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 10 min. The sections were then incubated overnight at 4 °C in a humidified chamber with a rabbit recombinant monoclonal anti-FOLR1 antibody (catalog no. A22419, Abclonal, Wuhan, China; clone ARC50863) at a dilution of 1:100. After washing with phosphate-buffered saline (PBS), the sections were incubated with an HRP-conjugated secondary antibody for 30 min at room temperature. Immunoreactivity was visualized using 3,3′-diaminobenzidine (DAB), followed by hematoxylin counterstaining, dehydration, and mounting. Sections of high-grade serous ovarian carcinoma (HGSC) with known high FOLR1 expression were included in each staining run as positive controls and processed under identical conditions. For negative controls, the primary antibody was replaced with PBS, while all other steps were kept identical. All control and study sections were stained simultaneously to ensure comparability. FRα expression was evaluated in tumor cells on the basis of membranous staining only; cytoplasmic staining was not considered positive. FRα expression was evaluated in tumor epithelial cells on FRα-immunostained TMA cores using both an immunoreactive score (IRS) and an H-score. The IRS was determined by multiplying the staining intensity score (0 = no staining, 1 = weak, 2 = moderate, 3 = strong) by the percentage score of positive tumor cells (0 = 0–5%, 1 = 6–25%, 2 = 26–50%, 3 = 51–75%, 4 = > 75%), yielding a total score ranging from 0 to 12. For descriptive analyses of expression patterns across histologic groups, IRS values were further categorized as low (0–3), intermediate (4–7), and high (8–12). The H-score was calculated as 1 × (% of weakly stained tumor cells) + 2 × (% of moderately stained tumor cells) + 3 × (% of strongly stained tumor cells), yielding a score ranging from 0 to 300. Because relatively few tumors met the IRS-high category, subsequent clinicopathologic comparisons and survival analyses were performed using the cohort median FRα H-score as the dichotomizing cutoff for FRα-high and FRα-low groups.

Multiplex immunofluorescence staining and imaging

Multiplex immunofluorescence (mIF) was chosen for immune profiling because it enabled simultaneous visualization of multiple immune and checkpoint markers on the same tissue section, cell-level phenotyping, and colocalization analysis within PanCK-defined tumor epithelial and non-epithelial compartments. The mIF staining was performed using a tyramide signal amplification (TSA)–based cyclic staining protocol (Immunoway Biotechnology, Plano, TX, USA) with internally validated antibodies. Sequential rounds of antigen retrieval, primary antibody incubation, TSA signal amplification, and antibody stripping were carried out according to optimized and standardized procedures. The mIF panel included FRα, PD-1, PD-L1, CD3, CD4, CD8, FOXP3, CD20, CD56, CD68, and pancytokeratin (PanCK). DAPI was used as a nuclear counterstain. After completion of all the staining cycles, the slides were counterstained with DAPI and mounted using antifade mounting medium.

The TMA slides were scanned using a fluorescence whole-slide scanner under standardized acquisition settings, generating digital whole-slide images from which individual TMA cores were subsequently analyzed. The predominant cellular localization and biological relevance of each marker are summarized in Table 1.

Table 1.

Markers included in the multiplex immunofluorescence panel and their biological relevance

Marker	Predominant cellular localization	Biological relevance
FRα	Cell membrane	Tumor epithelial cells
PD-L1	Cell membrane	Tumor cells, antigen-presenting cells
PD-1	Cell membrane	T cells
PanCK	Cytoplasmic	Epithelial tumor cells
CD3	Cell membrane	Total T cells
CD4	Cell membrane	Helper T cells
CD8	Cell membrane	Cytotoxic T cells
FOXP3	Nucleus	Regulatory T cells
CD20	Cell membrane	B cells
CD56	Cell membrane	NK cells/NK-like cells
CD68	Cytoplasmic	Macrophages
DAPI	Nucleus	Nuclear counterstain

Abbreviations: FRα Folate receptor alpha, PD-1 Programmed cell death protein 1, PD-L1 Programmed cell death ligand 1, PanCK Pan-cytokeratin, FOXP3 Forkhead box P3, NK Natural killer, DAPI 4′,6-diamidino-2-phenylindole

Image analysis and cell phenotyping

All the stained images were analyzed using QuPath software [9]. For tumor-cell analyses, tumor epithelial regions were defined by PanCK positivity and used as masks. FRα⁺PD-L1⁺ tumor cells were defined as PanCK⁺ epithelial cells coexpressing FRα and PD-L1. For immune-cell analyses, nucleated cells within each evaluable TMA core were phenotyped on the basis of marker expression. Specifically, CD3⁺ cells were classified as total T cells, CD4⁺ cells as helper T cells, CD8⁺ cells as cytotoxic T cells, FOXP3⁺ cells as regulatory T cells, CD20⁺ cells as B cells, CD56⁺ cells as NK/NK-like cells, and CD68⁺ cells as macrophages. Marker positivity thresholds were interactively optimized within QuPath by an experienced pathologist on the basis of visual inspection and negative controls. For each TMA core, the number of positive cells for each phenotype was quantified and expressed as the proportion of total nucleated cells within the analyzed core; for tumor-cell analyses, proportions were calculated within PanCK-defined epithelial regions.

The overall study workflow is summarized in Fig. 1B.

Cutoff determination and survival analysis

The proportion of FRα⁺PD-L1⁺ tumor cells was calculated for each patient. The optimal cutoff value for stratifying patients into FRα⁺PD-L1⁺ high and low groups was determined using X-tile software, with disease-free survival as the endpoint. The resulting cutoff value (5.6%) was applied consistently across the overall study cohort and histological subgroups. Survival curves were generated using the Kaplan–Meier method and compared using the log-rank test.

Statistical analysis

Univariate Cox proportional hazards regression was performed to evaluate factors associated with disease-free survival (DFS). Variables with P < 0.10 in univariate analysis were subsequently included in a multivariate Cox regression model to identify independent prognostic factors. Hazard ratios (HRs) and 95% confidence intervals (CIs) were reported. Statistical analyses were performed using SPSS version 26.0. Continuous variables were compared using Student’s t-test or the Mann–Whitney U test, as appropriate. Categorical variables were analyzed using the chi-square test or Fisher’s exact test. Correlations between FRα and PD-L1 positive tumor cell proportions and immune parameters were evaluated using Spearman’s rank correlation analysis, and correlation coefficients (r) and corresponding P values were reported. Continuous variables were presented as mean ± standard deviation (SD) or median (interquartile range, IQR), as appropriate. All tests were two-sided, and P < 0.05 was considered statistically significant.

Results

1. FRα expression status and clinicopathologic characteristics in the overall study cohort

FRα immunoreactivity was heterogeneous across histologic subtypes in the overall study cohort. FRα expression was initially assessed using an immunoreactive score (IRS; range 0–12), a semiquantitative system in which an IRS ≥ 8 is commonly regarded as indicating strong positivity, and categorized as low (0–3), intermediate (4–7), or high (8–12) (Fig. 2A). Using an IRS ≥ 8 threshold, high FRα expression was observed in 12/77 (15.6%) serous tumors, 1/17 (5.9%) MBOTs, and 0/6 (0%) SMBOTs. Within the serous spectrum, FRα-high cases were identified in 1/32 typical SBOTs (3.1%), 6/36 SBOT-MPs (16.7%), and 5/9 LGSOCs (55.6%) (Fig. 2A).

Fig. 2.

FRα expression patterns and clinical relevance across BOT histotypes and the serous spectrum A Distribution of FRα expression assessed by immunoreactive score (IRS) across histologic subtypes of borderline ovarian tumors, including serous (SBOT), mucinous (MBOT), and seromucinous (SMBOT) tumors, as well as across serous tumor subgroups, including typical SBOT, SBOT with micropapillary pattern (SBOT-MP), and low-grade serous ovarian carcinoma (LGSOC). FRα IRS was categorized as low (0–3), intermediate (4–7), or high (8–12) B Representative hematoxylin and eosin (H&E) staining and FRα immunohistochemical staining of borderline ovarian tumor tissue microarray cores demonstrating low and high FRα expression. Scale bar, 250 μm C Kaplan–Meier analyses of disease-free survival (DFS) stratified by FRα expression (high vs. low, dichotomized by the cohort median H-score) in the overall study cohort and selected histologic subgroups. DFS was defined as the time from primary surgery to recurrence or last follow-up. Statistical significance was assessed using the log-rank test

Given the relatively small number of tumors meeting the IRS-high definition and the resulting imbalance between groups, subsequent clinicopathologic and survival analyses were performed using the median FRα H-score as the cutoff, and the cohort was classified into FRα-high (n = 50) and FRα-low (n = 50) groups. Baseline clinicopathologic characteristics were comparable between the FRα-high and FRα-low groups, with no significant differences detected in age, serum CA125, CA199, or HE4 levels, FIGO stage, bilaterality, or fertility-sparing surgery rate (all P > 0.05; Table 2). In contrast, recurrence occurred more frequently in the FRα-high group than in the FRα-low group (42% vs. 14%, P = 0.003), indicating that only the clinicopathologic variable was significantly associated with the FRα status (Table 2).

Table 2.

Baseline clinicopathologic characteristics of patients stratified by FRα expression

	FRα High (n = 50)	FRα Low (n = 50)	p-value
Age [M (P25, P75), years]	33 (29, 48)	32 (30, 46)	0.939
CA125 [M (P25, P75), U/mL]	235.3 (44.5, 673.9)	66.8 (38.7, 463.9)	0.196
CA199 [M (P25, P75), U/mL]	2.3 (1.6, 5.4)	2.5 (1.6, 10.4)	0.671
HE4 [M (P25, P75), pmol/L]	109.2 (54.1, 236.7)	89.2 (51.0, 162.9)	0.359
Pathological type, n (%)			0.173
SBOT
LGSOC	7 (14)	2 (4)
SBOT-MP	21 (42)	15 (30)
Typical SBOT	13 (26)	19 (38)
MBOT	6 (12)	11 (22)
SMBOT	3 (6)	3 (6)
FIGO stage, n (%)			0.123
I	19 (38)	26 (52)
II	6 (12)	9 (18)
III	25 (50)	15 (30)
Bilaterality, n (%)			0.139
Unilateral	13 (26)	21 (42)
Bilateral	37 (74)	29 (58)
FSS, n (%)			0.225
Yes	25 (50)	32 (64)
No	25 (50)	18 (36)
Recurrence, n (%)			0.003
Yes	21 (42)	7 (14)
No	29 (58)	43 (86)

Continuous variables are presented as medians (P25, P75), and categorical variables as number are presented as numbers (percentages). FRα-high and FRα-low groups were defined according to the median FRα H-score of the cohort (median = 40). Comparisons between the FRα-high and FRα-low groups were performed using the Mann–Whitney U test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables, as appropriate. The FRα-high and FRα-low groups correspond to those shown in the representative immunohistochemistry images in Fig. 2B

Abbreviations: FRα Folate receptor alpha; M (P25, P75), median (25th–75th percentile), SBOT Serous borderline ovarian tumor, SBOT-MP Serous borderline ovarian tumor with micropapillary features, MBOT Mucinous borderline ovarian tumor, SMBOT Seromucinous borderline ovarian tumor, LGSOC Low-grade serous ovarian carcinoma, HGSOC High-grade serous ovarian carcinoma, FSS Fertility-sparing surgery (defined as preservation of the uterus and at least one ovary), FIGO International Federation of Gynecology and Obstetrics, HE4 Human epididymis protein 4

According to the results of the Kaplan–Meier analyses, FRα-high tumors were associated with significantly shorter DFS in the overall study cohort (log-rank P = 0.0035) (Fig. 2C). Similar associations were observed for serous cohort (SBOT and LGSOC; P = 0.0211) and within SBOT-MP (P = 0.0408), whereas no significant relationship was detected for MBOT (P = 0.3044) (Fig. 2C). Consistent with these findings, Cox proportional hazards regression analyses revealed that a high FRα status remained independently associated with recurrence after adjustment for relevant clinicopathologic factors (multivariable HR 2.604, 95% CI 1.053–6.441; P = 0.038), along with FIGO stage III and fertility-sparing surgery (Table 3). Collectively, these data suggest that elevated FRα expression is associated with recurrence in the overall study cohort, particularly within the serous spectrum.

Table 3.

Univariate and multivariate Cox regression analyses of factors associated with recurrence

Variable	Recurrence, n	No Recurrence, n	Univariate Cox analysis			Multivariate Cox analysis
			HR	95% CI	p value	HR	95% CI	p value
Age, years
> 30	21	50	1.652	0.694–3.933	0.257	-	-	-
≤ 30	7	22	1.000	-	-	-	-	-
BMI, kg/m2
> 24	8	24	0.570	0.240–1.357	0.204	-	-	-
≤ 24	20	48	1.000	-	-	-	-	-
CA125, U/mL
> 35	28	63	25.031	0.125–5008.562	0.234	-	-	-
≤ 35	0	9	1.000	-	-	-	-	-
Tumor size, cm
> 15	19	26	2.391	1.077–5.311	0.032	1.250	0.465–3.361	0.658
≤ 15	9	46	1.000	-	-	1.000	-	-
Pathological type
Serous tumors	25	52	3.536	1.062–11.769	0.040	1.089	0.243–4.88	0.911
non-serous tumors	3	20	1.000	-	-	1.000	-	-
FIGO stage
I - II	15	45	0.404	0.184–0.887	0.024	0.281	0.109–0.727	0.009
III	13	27	1.000	-	-	1.000	-	-
Bilaterality
Unilateral	5	29	0.256	0.096–0.679	0.006	0.428	0.105–1.756	0.239
Bilateral	23	43	1.000	-	-	1.000	-	-
FSS
Yes	24	33	4.870	1.687–14.055	0.003	9.348	2.762–31.641	< 0.001
No	4	39	1.000	-	-	1.000	-	-
Chemotherapy
Yes	16	21	1.576	0.727–3.415	0.249	-	-	-
No	12	51	1.000	-	-	-	-	-
FRα H-score
High	21	29	2.988	1.247–7.155	0.014	2.604	1.053–6.441	0.038
Low	7	43	1.000	-	-	1.000	-	-

The lower-risk category was used as the reference group (HR = 1.00). Age was dichotomized at 30 years, reflecting the predominantly young patient population and commonly used clinical thresholds in studies of borderline ovarian tumors. BMI was categorized using the upper limit of the normal range (24 kg/m²). Serum CA125 levels were dichotomized according to the clinical upper limit of normal (35 U/mL). Tumor size was dichotomized at the cohort median value (15 cm). FRα expression status was defined on the basis of the median H-score of the cohort. Variables with statistical significance in univariate Cox regression analyses were entered into the multivariate model

Abbreviations: HR Hazard ratio, CI Confidence interval, SBOT Serous borderline ovarian tumor, FSS Fertility-sparing surgery (defined as preservation of the uterus and at least one ovary), FIGO International Federation of Gynecology and Obstetrics

• 2.Correlation between FRα expression and immune cell subsets and PD-1/PD-L1 expression

To investigate the relationship between FRα expression and the immune landscape of the overall study cohort, we first evaluated the correlations between FRα levels and immune cell subsets (Fig. 3A). FRα expression was significantly and inversely correlated with the expression of multiple immune cell populations, including CD20⁺ B cells, CD68⁺ macrophages, and CD3⁺/CD4⁺ T cells. Notably, FRα expression was moderately negatively correlated with CD68⁺ macrophage infiltration (r = − 0.37, P < 0.001) (Fig. 3B).

Fig. 3.

Associations between FRα expression, immune cell infiltration, and immune checkpoint markers A Heatmap showing Spearman correlation coefficients between FRα, PD-1, PD-L1, and immune cell subsets, including CD3⁺, CD4⁺, CD8⁺ T cells, CD20⁺ B cells, CD56⁺ NK cells, CD68⁺ macrophages, and FOXP3⁺ cells. Red indicates positive correlations, whereas blue indicates negative correlations, with color intensity corresponding to the strength of the correlation B Selected scatter plots showing statistically significant inverse correlations between FRα expression and immune cell subsets, including CD20⁺ B cells, CD68⁺ macrophages, CD3⁺ T cells, and CD4⁺ T cells. (C) Selected scatter plots showing statistically significant positive correlations between FRα expression and immune checkpoint markers, including PD-1 and PD-L1. Correlations were evaluated using Spearman’s rank correlation analysis. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant

In parallel, we further assessed the association between FRα expression and immune checkpoint molecules. FRα expression was significantly and positively correlated with both PD-1 and PD-L1 levels (Fig. 3A). A moderate positive correlation was observed between FRα expression and PD-L1 expression (r = 0.36, P < 0.001), with a similar trend for PD-1 expression (Fig. 3C), indicating higher immune checkpoint marker expression in tumors with elevated FRα levels.

To examine internal consistency across markers, we evaluated correlations between PD-1/PD-L1 expression and immune cell infiltration. As expected, PD-1 expression and PD-L1 expression were inversely associated with CD3⁺ and CD4⁺ T-cell infiltration (Fig. 3A), a result that is in agreement with reported immune features in borderline tumors. Together, these findings indicate that FRα expression in the overall study cohort is linked to both reduced immune cell presence and increased immune checkpoint expression, suggesting distinct immune states across FRα-defined tumor subsets.

• 3.Prognostic significance of combined FRα and PD-L1 expression in the overall study cohort

To further explore the clinical relevance of FRα⁺PD-L1⁺ tumor cells in the overall study cohort, we assessed the prognostic impact of their combined expression. Representative multiplex immunofluorescence images revealed FRα and PD-L1 expression within PanCK⁺ tumor epithelial regions (Fig. 4A). Cell-level classification further confirmed the presence of FRα⁺PD-L1⁺ double-positive tumor cells, supporting subsequent survival stratification on the basis of the FRα⁺PD-L1⁺ status (Fig. 4B).

Fig. 4.

Prognostic relevance of FRα⁺PD-L1⁺ tumor epithelial cells in the overall study cohort A Representative multiplex immunofluorescence images showing FRα, PD-L1, PanCK, and DAPI expression within tumor epithelial regions, with magnified views of selected areas B Image-based cell-level segmentation and phenotyping within PanCK-defined compartments, illustrating the classification of tumor epithelial cells according to FRα and PD-L1 expression status, including FRα⁺PD-L1⁺ double-positive cells C Kaplan–Meier analyses of disease-free survival stratified by the proportion of FRα⁺PD-L1⁺ tumor epithelial cells (high vs. low) in the overall study cohort and selected histologic subgroups, including the SBOT and low-grade serous ovarian carcinoma (LGSOC), SBOT-MP, and MBOT cohorts. The cutoff value was determined using X-tile software, and survival differences were assessed using the log-rank test

Using an optimized cutoff value (5.6%) determined in the overall study cohort, patients were classified into FRα⁺PD-L1⁺ high and low groups. Kaplan–Meier analyses revealed that a high proportion of FRα⁺PD-L1⁺ tumor cells was significantly associated with poor survival in the overall study cohort (P = 0.0004) and in serous tumors. Similar survival differences were also observed for tumors with micropapillary features. In contrast, no significant difference in survival was detected between the FRα⁺PD-L1⁺ high and low groups of patients with mucinous tumors (Fig. 4C).

Collectively, these findings indicate that FRα⁺PD-L1⁺ tumor cell status stratifies clinical outcomes in the overall study cohort in a histology-dependent manner, particularly within clinically relevant subsets of the serous spectrum.

Discussion

In this study, we comprehensively evaluated FRα expression and its immune context in a well-characterized overall study cohort comprising BOTs and LGSOCs. Our findings demonstrate that elevated FRα expression is associated with an increased risk of recurrence in patients with BOTs, particularly those with serous tumors and those with micropapillary features. Importantly, we further show that the proportion of FRα⁺PD-L1⁺ tumor cells, quantified at single-cell resolution using QuPath-based cell segmentation within tumor epithelial compartments, provides additional prognostic stratification beyond FRα expression alone. These results position FRα, especially in combination with PD-L1, as a clinically relevant biomarker for identifying higher-risk subsets of BOTs.

Although BOTs are generally associated with a favorable prognosis, they frequently affect younger patients, making accurate risk stratification for recurrence clinically important, particularly for BOTs with specific histopathological features such as SBOT-MP [10]. While most BOTs follow an indolent course, a subset of BOTs exhibits recurrent disease or progression toward low-grade serous ovarian carcinoma [11]. Our observation that FRα expression is higher in SBOT-MP than in typical SBOT, and comparable to that in LGSOC, supports the concept that FRα expression increases along the serous tumor continuum. These findings align with prior clinicopathologic observations that SBOT-MP behaves more aggressively and shares biological features with LGSOC [12], reinforcing the relevance of FRα as a marker of adverse tumor biology within this spectrum. Notably, the association between FRα expression and recurrence was confined to serous BOTs and was not observed in mucinous tumors, indicating that FRα may be of particular relevance in the serous subtype.

In the phase III MIRASOL trial (NCT04209855) [6], platinum-resistant ovarian cancer patients were selected on the basis of high FRα expression (≥ 75% of tumor cells with staining intensity of ≥ 2+). Although only a minority of BOTs and LGSOCs in our cohort met the IRS-high definition (IRS ≥ 8), FRα-high cases were predominantly found within the serous spectrum, particularly in SBOT-MP and LGSOCs. These findings suggest that some serous BOTs exhibit FRα expression comparable to that of malignancies targeted by FRα-directed therapies. In addition, preliminary results from the FONTANA study (NCT05797168), presented at the ESMO annual meeting, indicated antitumor activity of the FRα-targeted antibody–drug conjugate AZD5335 in patients with platinum-resistant ovarian cancer. Notably, the drug showed efficacy across different FRα expression levels, even in patients with low FRα expression. Our findings also revealed that high FRα expression was associated with recurrence risk, supporting the potential value of FRα as a biomarker of aggressive tumor biology. However, because this study did not include treatment-response data, any therapeutic implications should be regarded as exploratory and hypothesis-generating only.

Our study further delineated the immune landscape associated with FRα expression in the overall study cohort. We observed that higher FRα expression was associated with reduced infiltration of multiple immune cell populations, including T cells, B cells, and macrophages, indicating that the tumor microenvironment is immunosuppressive in FRα-high tumors in our cohort. However, the immunologic significance of FRα expression remains incompletely understood. Prior studies have demonstrated marked heterogeneity and context-dependent associations between FRα expression and immune infiltration across different tumor types [13, 14]. How FRα overexpression contributes to the establishment of a suppressive immune microenvironment in BOTs and LGSOCs, and whether feedback regulatory loops exist between FRα signaling and immune cell exclusion, are important mechanistic questions that remain to be clarified. In addition, we found that FRα expression was positively correlated with PD-1 and PD-L1 expression in our cohort. Nevertheless, existing evidence regarding the relationship between FRα and PD-1/PD-L1 is limited and inconsistent across tumor types [14–16], suggesting that FRα-mediated modulation of the immune microenvironment may be highly tissue-specific. Thus, further mechanistic studies in BOTs and LGSOCs are warranted to elucidate how FRα expression is linked to the development of an immunosuppressive microenvironment. Moreover, by integrating FRα and PD-L1 expression at the tumor cell level, we observed that patients harboring a high proportion of FRα⁺PD-L1⁺ tumor cells had significantly worse outcomes in the overall study cohort as well as in the serous subtype, whereas this association was not seen in mucinous tumors. These findings suggest fundamentally different biological and immunologic responses between serous and mucinous BOTs. FRα⁺PD-L1⁺ tumor cells may constitute a biologically aggressive subset with enhanced immune-evasive potential and could serve as a clinically meaningful indicator for risk stratification. In the future, quantitative assessment of FRα⁺PD-L1⁺ tumor cells may help identify patients at increased risk of recurrence who could benefit from more intensive surveillance strategies.

Notably, FRα and the PD-1/PD-L1 axis are meaningful molecular targets in ovarian cancer, with FRα-targeted ADCs [5, 6]and PD-1/PD-L1 immune checkpoint inhibitors [17, 18] already undergoing active clinical investigation. In this context, our findings provide a rationale for future translational studies exploring whether FRα- and PD-L1-related therapeutic strategies may be relevant in selected recurrent tumors within the serous spectrum, including LGSOC; however, such implications remain exploratory.

Several limitations should be considered. This was a retrospective, single-institution study with a relatively limited sample size, particularly in nonserous subgroups. External validation in independent cohorts will be necessary to confirm the prognostic utility of FRα⁺PD-L1⁺ tumor cells. Moreover, the median follow-up duration of 41 months is relatively limited for borderline ovarian tumors, particularly for serous borderline tumors, in which late recurrences may occur. Therefore, the present survival findings should be interpreted cautiously and viewed as reflecting recurrence risk within the available follow-up period rather than definitive long-term prognostic estimates. Further, because this was an observational study without treatment-response data and the cohort consisted predominantly of borderline tumors, any therapeutic implications of FRα expression or FRα/PD-L1 co-expression should be considered exploratory and hypothesis-generating only. In addition, although our study revealed associations between FRα expression, immune features, and clinical outcomes, the mechanisms underlying the formation of the FRα-associated immune microenvironment remain unclear. Future studies integrating single-cell sequencing and functional analyses may further elucidate the role of FRα in mediating tumor–immune interactions.

In summary, our study demonstrated that high FRα expression in the serous spectrum is associated with an increased risk of recurrence and an immunosuppressive tumor microenvironment. Importantly, a high proportion of FRα⁺PD-L1⁺ tumor cells were identified as a marker of poor prognosis. These findings suggest that the FRα/PD-L1 axis may warrant further investigation in future translational studies of high-risk tumors within the serous spectrum.

Authors’ contributions

XX conceptualized and designed the study, performed data curation, conducted formal analysis, and drafted the manuscript. LY contributed to conceptualization, provided resources, and assisted with manuscript drafting. HYM contributed to conceptualization, methodology, and critically revised the manuscript. LZ contributed to conceptualization, methodology, and provided resources, and critically revised the manuscript. WJC contributed to conceptualization, methodology, supervised the project, and critically revised the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 82373387), China Postdoctoral Science Foundation (2022M711413), Natural Science Fund of Jiangsu Province (BK20231265, BE2023776), Jiangsu Provincial Key Medical Discipline Fund (ZDXK202210), Jiangsu Province Traditional Chinese Medicine Science and Technology Development Plan Project-Key Project (No. ZD202014), General Project of Yili Clinical Medical Research Institute (2022), TCM-Western Collaborative Clinical Project (ovarian cancer), the National Natural Science Foundation of China (No. 82404004), China Postdoctoral Science Foundation (No. 2024M751216).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the First Affiliated Hospital of Nanjing Medical University (Approval No. 2025-SR-452). The study was conducted in accordance with the Declaration of Helsinki. Given the retrospective nature of the study, the requirement for written informed consent was waived by the ethics committee. No animal experiments were involved in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.