Intratumoral IL-33⁺CD4⁺FoxP3⁺ regulatory T cell infiltration determines poor clinical outcomes and intensive immunoevasive contexture in patients with pancreatic cancer after surgical resection: a cohort study

Abstract

Background:

Interleukin-33 (IL-33), a member of the IL-1 cytokine family, is constitutively expressed in barrier cells such as endothelial cells and fibroblasts. While the expression of IL-33 in regulatory T cells (Tregs) has been previously reported, its clinical significance in pancreatic ductal adenocarcinoma (PDAC) remains unclear. This study aims to investigate the clinical relevance and biological role of IL-33⁺ Tregs in PDAC.

Methods:

Infiltration of IL-33⁺ Tregs was assessed by immunohistochemistry in 215 patients from our institute. The correlation between IL-33⁺ Tregs and clinical characteristics was analyzed. Additionally, the functional status of cytotoxic T cells in relation to IL-33⁺ Treg infiltration was examined. The impact of IL-33⁺ Tregs on the tumor microenvironment (TME) was further evaluated both in silico and in vitro.

Results:

IL-33⁺ Tregs infiltration was confirmed in PDAC tissues, and its abundance was positively associated with microvascular invasion, perineural invasion, and elevated serum CA19-9 levels. Patients with higher tumor-infiltrating IL-33⁺ Tregs demonstrated poorer overall survival (OS) and recurrence-free survival (RFS) compared to those with lower infiltration levels. Multivariate analysis confirmed IL-33⁺ Tregs as an independent prognostic factor for both OS and RFS, with improved survival prediction when combined with tumor differentiation. Subgroup analyses indicated that serum CA19-9 was not a useful risk stratification tool in patients with high IL-33⁺ Treg infiltration, and these patients showed limited survival benefit from adjuvant chemotherapy. Furthermore, increased IL-33⁺ Treg infiltration was associated with more pronounced immunosuppressive TME, marked by a reduction in cytotoxic phenotypes and an upregulation of exhausted markers on CD8⁺ T cells.

Conclusion:

Our findings identify IL-33⁺ Tregs as a novel subtype of Tregs, with strong prognostic value for survival risk stratification and therapeutic response prediction in PDAC. IL-33⁺ Tregs exhibit more pronounced immunosuppressive capabilities, impairing CD8⁺ T cell function. With further investigation, IL-33⁺ Tregs may represent a promising immunotherapeutic target for PDAC.

Introduction

Pancreatic ductal adenocarcinoma (PDAC) is among the most deadly cancers, with a 5-year survival rate below 13%, despite advances in surgery, chemotherapy, and radiotherapy^[1,2]. The aggressive nature of PDAC is largely driven by its distinctive tumor microenvironment (TME), which is marked by a dense fibrotic stroma and a strongly immunosuppressive setting that promotes tumor growth and metastasis^[3,4]. Immune cells within this TME, including regulatory T cells (Tregs), myeloid-derived suppressor cells, and tumor-associated macrophages (TAMs), are key players in dampening anti-tumor responses and enabling immune evasion^[5].

The transcription factor Foxp3, which is encoded by an X chromosome-linked gene and exhibits lineage specificity, is primarily expressed in CD4⁺CD25⁺ T lymphocytes. Foxp3 is both necessary and sufficient for the development and suppressive function of CD4⁺CD25⁺ Tregs^[6]. Consequently, CD4⁺CD25⁺ T cells exhibiting Foxp3 expression and immunosuppressive activity are now defined as Tregs. Tregs are essential for maintaining immune balance by controlling excessive immune reactions. In cancers, however, Tregs often shift this function towards supporting immune tolerance of tumors, thereby aiding disease progression^[7]. In PDAC, higher levels of Treg infiltration have been linked to worse outcomes and resistance to immunotherapy, as shown in our previous studies^[8,9]. Recent research has identified a subset of Tregs that express interleukin-33 (IL-33), which appears to create an even more suppressive immune environment^[10].

IL-33, a member of the interleukin-1 (IL-1) family due to structural similarities, is constitutively expressed in barrier cells, such as endothelial cells, epithelial cells, and stromal fibroblasts^[11,12]. However, without a secretory export sequence, nuclear IL-33 can be passively released by damaged membranes during non-lethal events like cellular stretch^[13], oxidative stress, or necrosis^[14]. This characteristic led to IL-33 being classified as an “alarmin” cytokine, similar to IL-1α. Additionally, IL-33 can be secreted into the extracellular environment in response to calcium influx triggered by Adenosine Triphosphate (ATP) or purinergic receptor signaling, confirming its role as a traditional cytokine, not merely an alarmin^[15]. Secreted IL-33 promotes tumor proliferation and invasion by directly targeting tumor cells and supporting angiogenesis through activation of endothelial cells via the suppression of tumorigenicity 2 (ST2) receptor. It also shapes the TME by activating various immune cells, including T cells, basophils, macrophages, eosinophils, and type 2 innate lymphoid cells through ST2 signaling. Both naïve and regulatory CD4⁺ T cells express ST2, making them targets of IL-33. Previous studies have shown that IL-33/ST2 signaling recruits GATA3⁺ Tregs, enhancing the immunosuppressive environment^[16]. More recent findings suggest that disrupting IL-33, rather than ST2, leads to tumor regression by inducing a “fragile” Treg phenotype through regulation of FoxP3 homeostasis^[10].

Given the pivotal role of Tregs and IL-33 in shaping an immunosuppressive TME in PDAC, it is essential to investigate their contributions to the poor clinical outcomes associated with this cancer. The specific impact of IL-33⁺ Tregs in PDAC remains inadequately studied. These cells might not only signify a more aggressive tumor phenotype but also reveal how the tumor manipulates its immune environment to support its growth and spread. This paper seeks to clarify the prognostic significance of tumor-infiltrating IL-33⁺ Tregs and their immunosuppressive characteristics. Additionally, it will explore potential therapeutic strategies aimed at targeting this pathway to improve anti-tumor immune responses.

HIGHLIGHTS

• Tumor-infiltrating IL-33⁺ Tregs emerged as a distinct subtype of Tregs with significant prognostic implications in PDAC.

• The presence of tumor-infiltrating IL-33⁺ Tregs provides a robust framework for survival risk stratification and predicting therapeutic responses, particularly to adjuvant chemotherapy, thereby aiding in personalized and precision medicine.

• IL-33⁺ Tregs exhibit a more pronounced immunosuppressive capabilities in PDAC, impairing CD8⁺ T cell function, which underscores the potential as an appealing target for future immunotherapy strategies.

Methods and materials

Study cohort and tissue microarray

This study utilized two previously mentioned tissue microarray (TMA) cohorts. The first cohort, collected from March 2015 to July 2018 at our institute, included 192 patients with pathologically confirmed PDAC who underwent radical resection. Six patients were excluded due to poor-quality tissue spots^[17]. The second cohort, from September 2018 to October 2019, consisted of 29 additional PDAC patients, which was further incorporated into analysis^[9]. The inclusion and exclusion criteria for both cohorts were as follows: (1) histologically confirmed PDAC with complete R0 resection; (2) complete clinical and follow-up data available; (3) no history of other primary malignancies; (4) no preoperative neoadjuvant therapies or involvement in clinical trials with immune or targeted therapies; (5) no evidence of distant metastasis. The protocol was registered in Research Registry (researchregistry11246).

Tumor tissues were fixed in formalin and embedded in paraffin. Representative tumor areas, selected by an experienced pathologist, were cut into 2.0 mm diameter cylindrical samples for TMA production. Clinicopathological data – including age, sex, surgical procedure, preoperative serum carbohydrate antigen 19-9 (CA19-9) levels, American Joint Committee on Cancer (AJCC) eighth edition of TNM stage, tumor differentiation, microvascular and perineural invasion, postoperative adjuvant radiotherapy (ART) and chemotherapy (ACT), and follow-up data – were collected by one surgeon and verified by another. This study has been reported in line with the STROCSS guidelines^[18].

Survival follow-up

The last follow-up date of this study was conducted in February 2024. All patients were regularly monitored following surgical resection, as previously described^[17]. Overall survival (OS) was defined as the period from the date of surgery to death from any cause, or the last follow-up, whichever came first. Recurrence-free survival (RFS) was measured from the date of surgery to either the first instance of relapse or death, or the last follow-up. These definitions align with those used in our previous reports^[17,19].

Immunofluorescence staining and assessment

Formalin-fixed, paraffin-embedded (FFPE) tissue specimens of the TMAs were subjected to immunofluorescence (IF) staining, following established protocols^[9]. Anti-CD4 (Maxim, RMA-0620), anti-FoxP3 (Cell Signaling Technology, 98377), and anti-IL-33 (Proteintech, 12372-1-AP) antibodies were used to identify IL-33⁺CD4⁺FoxP3⁺ Tregs. The antibodies were incubated with hydrated TMA sections overnight at 4°C, followed by labeling with Alex Fluor 647, Alex Fluor 555, and Alex Fluor 488. Afterward, the slides were counterstained with DAPI for 5 minutes. Following washing and mounting, the sections were scanned using a 3D HISTECH Pannoramic SCAN fluorescence scanner. For each spot, three independent and representative high-power fields (HPF) at 200× magnification (0.305 mm² per field) were captured. The results were assessed and quantified as previously described^[17].

Flow cytometry

Freshly resected PDAC tissues were dissociated, and the isolated cells were stained with surface markers (CD45, CD4, and CD25) at 4°C for 30 minutes. The cells were then fixed using IC Fixation Buffer (eBioscience, 00-8222-49). After three washes with PBS, the cells were permeabilized with Permeabilization Buffer (eBioscience, 00-8333-56) and stained for intracellular markers (FoxP3 and IL-33). Data acquisition was conducted using the BD FACS Arial Flow Cytometer (BD Biosciences, USA), and data analysis was performed with FlowJo V.10 software (BD Biosciences, USA). The antibodies used in this study included FITC anti-human CD4 (Biolegend, 317408), APC/Cyanine7 anti-human CD25 (Biolegend, 302613), Alexa Fluor 647 anti-human FOXP3 (Biolegend, 320213), PE anti-human IL-33 (Invitrogen, MA5-40993), and PerCP/Cyanine 5.5 anti-human CD45 (Biolegend, 304027).

In our previous study, we dissociated sixteen fresh PDAC tissues and analyzed tumor-infiltrating CD8⁺ T cells for cytotoxic markers (TNF-α, IFN-γ, Granzyme B, and Perforin) and exhaustion markers (PD-1, CTLA-4, TIM-3, and TIGIT)^[17]. For the current study, FFPE slides from these cases were further used for IF staining of IL-33⁺ Tregs. After quantifying the tumor-infiltrating IL-33⁺ Tregs, the cases were divided into two cohorts based on their levels. Correlation analysis was then conducted to evaluate the potential immunomodulatory effects of IL-33⁺ Tregs on CD8⁺ T cells.

In addition, we dissociated eight fresh PDAC tissues and sorted Tregs as CD4^highCD25^highCD127^low by MACS Treg magnetic separation (Miltenyi Biotec, 130-109-557) kit according to manufacturer instruction. The proportion of IL-33⁺ Tregs was assessed by flow cytometry, and categorized into IL-33⁺ Tregs high and low groups. These sorted Tregs were then co-cultured with CD8⁺ T cells, which were amplified from healthy donor peripheral blood mononuclear cells (PBMCs), as previously described^[20]. After co-culturing for 48 hours, the cytotoxic profiles of CD8⁺ T cells, including TNF-α, IFN-γ, Granzyme B, and Perforin expression, were analyzed by flow cytometry.

Bioinformatic analysis

Transcriptomic and survival data for patients with PDAC were obtained from public databases, including the Cancer Genome Atlas (TCGA, https://cancergenome.nih.gov/), the International Cancer Genome Consortium (ICGC, https://dcc.icgc.org/), and the NIH GEO database (Accession #GSE71729). These datasets were analyzed to assess the prognostic significance and immunoregulatory function of IL-33⁺ Tregs. Using a Treg-specific gene set from MSigDB (GSE22045_Treg_VS_TCONV_UP) combined with the IL-33 gene, we developed an IL-33⁺ Treg signature. The GSVA algorithm was employed to calculate an IL-33⁺ Treg score for each patient, with individuals stratified into high- and low-infiltration groups based on the optimal threshold, determined using the “surv_cutpoint” function from the “survminer” R package. Differentially expressed genes were identified within the ICGC dataset using the “limma” package, and ranked by log₂ fold change (log₂FC). To explore the biological processes associated with different levels of IL-33⁺ Treg infiltration, Gene Set Enrichment Analysis (GSEA) was performed based on Gene Ontology biological processes.

Statistical analysis

All statistical analyses were conducted using SPSS 21.0 and R version 4.3.0. Receiver operating characteristic (ROC) curve analysis was used to determine optimal cut-off values for continuous variables, including CA19-9 levels, total tumor-infiltrating Tregs, and IL-33⁺ Tregs. Associations between tumor-infiltrating Treg cells (total and IL-33⁺) and clinicopathological characteristics were assessed using Pearson’s chi-square or Fisher’s exact tests, depending on the data distribution. For continuous variables, comparisons between groups were made using either Student’s t-test or the Mann–Whitney U test, as appropriate. Kaplan–Meier survival curves were generated and analyzed using log-rank tests. Independent prognostic factors for OS and RFS were identified through univariate and multivariate Cox proportional hazards analyses. Interaction analyses were performed using Cox models adjusted for potential confounding factors. A P value of less than 0.05 was considered statistically significant.

Results

Demographical and clinicopathological characteristics

A total of 215 PDAC patients with comprehensive follow-up data were included in this study. The cohort was 63.7% male, with a median age of 64 years (interquartile range [IQR]: 58–70). Pathological grading revealed 4 patients with well-differentiated tumors, 91 with moderately differentiated tumors, and 120 with poorly differentiated tumors. According to the AJCC TNM staging system (eighth edition), 104 patients (48.4%) were classified as stage I, 92 (42.8%) as stage II, and 19 (8.8%) as stage III. Most patients (75.8%) received ACT, while 17.2% received ART. The median pre-operative serum CA19-9 level was 135.7 U/mL (IQR: 35.0–374.7). Additional clinicopathological details are provided in Supplemental Digital Content Table 1, available at: http://links.lww.com/JS9/E310.

Identification and correlations of tumor-infiltrating IL-33⁺ Tregs with clinicopathological characteristics

The presence of IL-33⁺CD4⁺FoxP3⁺ Tregs in the PDAC microenvironment was identified through triple-color IF staining (Fig. 1A) and flow cytometry (Supplemental Digital Content Figure 1, available at: http://links.lww.com/JS9/E309). In the TMA slices, the median count of tumor-infiltrating IL-33⁺ Tregs was 5.7 cells per HPF (IQR: 3.0–9.0), while the median count of total conventional Tregs was 13.7 cells/HPF (IQR: 9.0–19.0) (Supplemental Digital Content Table 1, available at: http://links.lww.com/JS9/E310). Flow cytometry of fresh PDAC tissues revealed that made up a median proportion of 46.6% (IQR: 37.6%–52.6%) of total conventional Tregs (Fig. 1B).

Figure 1.

The presence, clinicopathological correlation, and prognostic significance of tumor-infiltrating IL-33⁺ Tregs in the Zhongshan cohort. (A) Representative immunofluorescence staining images showing IL-33⁺ and IL-33⁻ Tregs in pancreatic cancer tissue. (B) Proportion of IL-33⁺ Tregs in the total Treg population. (C–E) Tumor-infiltrating IL-33+ Treg density comparisons in relation to different clinical parameters: microvascular invasion status, perineural invasion status, and serum CA19-9 levels. Statistical significance is indicated as ***P < 0.001, ** P < 0.01. (F, G) Kaplan–Meier plots depicting OS and RFS based on varying densities of tumor-infiltrating IL-33⁺ Tregs via the log-rank test. OS refers to overall survival, RFS refers to recurrence-free survival.

Based on ROC curve analysis, the optimal cut-off values were determined for serum CA19-9 (231.4 U/mL), tumor-infiltrating conventional Tregs (13.2 cells/HPF), and tumor-infiltrating IL-33⁺ Tregs (5.5 cells/HPF) as reported in our previous studies^[9,17]. Using these thresholds, patients were divided into low (104 patients) and high (111 patients) IL-33⁺ Treg infiltration groups. Correlation analysis indicated that IL-33⁺ Treg infiltration was significantly associated with serum CA19-9 levels (P = 0.007), microvascular invasion (P = 0.038), and perineural invasion (P < 0.001), while other clinicopathological factors showed no significant associations (Table 1). Tumors with microvascular invasion (P = 0.006, Fig. 1C) or perineural invasion (P < 0.001, Fig. 1D) exhibited notably higher IL-33⁺ Tregs infiltration. Additionally, higher serum CA19-9 levels were predictive of increased IL-33⁺ Tregs infiltration (P = 0.002, Fig. 1E).

Table 1.

The relationships between tumor-infiltrating Tregs and IL-33⁺ Tregs and clinicopathological characteristics

Variables	Low Tregs (n = 101)	High Tregs (n = 114)	P value	Low IL-33+ Tregs (n = 104)	High IL-33+ Tregs (n = 111)	P value
Sex			0.919			0.836
Male/female	64/37	73/41		67/37	70/41
Age (years)			0.481			0.423
<65/ ≥ 65	58/43	60/54		60/44	58/53
Surgery			0.776			0.718
Whipple/DP/TP	57/43/1	63/48/3		58/45/1	62/46/3
Differentiation			0.101			0.136
I/ II/III	4/43/54	0/48/66		4/44/56	0/47/64
T stage			0.478			0.828
T1/T2/T3	31/57/13	36/57/21		32/57/15	35/57/19
N stage			0.908			0.168
N0/N1/N2	62/30/9	67/37/10		68/26/10	61/41/9
TNM stage			0.998			0.316
I/II/III	49/43/9	55/49/10		55/39/10	49/53/9
CA19-9			0.070			0.007
<231.4/ ≥ 231.4 U/L	68/33	63/51		73/31	58/53
ART			0.389			0.746
No/yes	86/15	92/22		87/17	91/20
ACT			0.534			0.065
No/yes/unknown	10/77/14	16/86/12		7/84/13	19/79/13
Microvascular invasion			0.239			0.038
No/yes	86/15	90/24		91/13	85/26
Perineural invasion			0.001			<0.001
No/yes	33/68	15/99		36/68	12/99
Tumor IL-33+ Tregs			<0.001			/
Low/high	88/13	16/98		/	/
Tumor Tregs			/			<0.001
Low/high	/	/		88/16	13/98

Bold values represent that the factors have significant difference (p < 0.05).

Prognostic value of tumor-infiltrating IL-33⁺ Treg in PDAC patients

The median follow-up duration for this cohort was 72.0 months (95% CI: 68.8–75.2). Median OS was 28.5 months, while median RFS was 16.8 months. The 1-, 3-, and 5-year OS rates were 81.2%, 43.1%, and 27.9%, respectively. Similarly, the 1, 3, and 5-year RFS rates were 59.7%, 26.2%, and 17.7%, respectively. Patients in the high tumor-infiltrating conventional Treg group had significantly shorter median OS (21.1 vs. 41.0 months, P < 0.001, hazard ratio [HR] = 2.103, 95% CI: 1.534–2.883, Supplemental Digital Content Figure 2A, available at: http://links.lww.com/JS9/E309) and RFS (11.9 vs. 29.0 months, P < 0.001, HR = 2.141, 95% CI: 1.537–2.894, Supplemental Digital Content Figure 2B, available at: http://links.lww.com/JS9/E309) compared to the low-infiltration group, consistent with our previous findings^[8,9]. More pronounced differences were observed between the high and low tumor-infiltrating IL-33⁺ Treg groups, with a median OS of 17.9 months versus 56.4 months (P < 0.001, HR = 3.436, 95% CI: 2.475–4.769, Fig. 1F) and a median RFS of 10.7 months versus 32.9 months (P < 0.001, HR = 3.280, 95% CI: 2.393–4.495, Fig. 1G).

Univariate Cox analysis identified several factors significantly associated with both OS and RFS, including tumor differentiation (both P = 0.002), N stage (both P < 0.001), TNM stage (both P < 0.001), serum CA19-9 levels (both P < 0.001), ACT (P = 0.019 and P = 0.012, respectively), perineural invasion (P = 0.001 and P = 0.005, respectively), tumor-infiltrating IL-33⁺ Tregs (both P < 0.001), and tumor-infiltrating conventional Tregs (both P < 0.001) (Table 2). In multivariate Cox analysis, independent prognostic factors for both OS and RFS were identified as N stage (P = 0.022, HR = 1.903, 95% CI: 1.096–3.303 and P = 0.045, HR = 1.690, 95% CI: 1.012–2.823, respectively), ACT (P < 0.001, HR = 0.309, 95% CI: 0.194–0.493 and P = 0.018, HR = 0.577, 95% CI: 0.366–0.908, respectively), and tumor-infiltrating IL-33⁺ Tregs (P < 0.001, HR = 4.323, 95% CI: 2.514–7.436 and P < 0.001, HR = 4.170, 95% CI: 2.523–6.894, respectively). Additionally, tumor differentiation (P = 0.031, HR = 1.422, 95% CI: 1.033–1.956) and serum CA19-9 levels (P = 0.009, HR = 1.536, 95% CI: 1.111–2.124) were identified as independent risk indicators for RFS alone (Table 2).

Table 2.

Univariate and multivariate analysis of prognostic indicators associated with overall survival and recurrence-free survival

Variables	Overall survival			Recurrence-free survival
	Univariate P value	Multivariate P value	Multivariate HR (95% CI)	Univariate P value	Multivariate P value	Multivariate HR (95% CI)
Sex
Male/female	0.329	NA		0.535
Age (years)
<65/ ≥ 65	0.438	NA		0.293
Surgery
Whipple/DP/TP	0.610	NA		0.555
Differentiation
I + II/III	0.002	0.078	1.364 (0.966–1.926)	0.002	0.031	1.422 (1.033–1.956)
T stage
T1/T2/T3	0.192	NA		0.245
N stage
N0/N1/N2	<0.001	0.022	1.903 (1.096–3.303)	<0.001	0.045	1.690 (1.012–2.823)
TNM stage
I/II/III	<0.001	0.537	0.838 (0.479–1.467)	<0.001	0.645	0.884 (0.523–1.495)
CA19-9
<231.4/ ≥ 231.4 U/L	<0.001	0.059	1.387 (0.988–1.947)	<0.001	0.009	1.536 (1.111–2.124)
ART
No/yes	0.604	NA		0.725
ACT
No/yes/unknown	0.019	<0.001a	0.309 (0.194–0.493)	0.012	0.018a	0.577 (0.366–0.908)
Microvascular invasion
No/yes	0.775	NA		0.475
Perineural invasion
No/yes	0.001	0.204	1.349 (0.850–2.140)	0.005	0.386	1.201 (0.794–1.817)
Tumor IL-33+ Tregs
Low/high	<0.001	<0.001	4.323 (2.514–7.436)	<0.001	<0.001	4.170 (2.523–6.894)
Tumor Tregs
Low/high	<0.001	0.056	0.615 (0.373–1.013)	<0.001	0.118	0.693 (0.437–1.098)

^aBold values represent that the factors have significant difference (p < 0.05).

Adjusted P value for receiving ACT compared to those not received.

Survival predictive value in combination of tumor-infiltrating IL-33⁺ Tregs and tumor differentiation

Tumor differentiation, a key indicator of the tumor’s inherent aggressiveness, has long been recognized as a prognostic factor in PDAC^[21]. Our interaction analysis demonstrated a significant association between tumor-infiltrating IL-33⁺ Tregs and tumor differentiation, with both showing a strong correlation with OS and RFS (both p < 0.001), after adjusting for confounding variables. These findings confirmed that both tumor-infiltrating IL-33⁺ Tregs and tumor differentiation were independent prognostic indicators of RFS, which indicated that tumor-infiltrating IL-33⁺ Tregs could serve as a crucial biomarker for tumor malignancy. By combining these two biological characteristics – tumor-infiltrating IL-33⁺ Tregs and tumor differentiation – we achieved a more precise stratification of patient prognosis. The median OS for the subgroups defined by IL-33⁺ Treg levels and tumor grades were as follows: 15.5 months for IL-33⁺ Treg^high and G3, 26.3 months for IL-33⁺ Treg^high and G1-2, 46.9 months for IL-33⁺ Treg^low and G3, and not yet reached for IL-33⁺ Treg^low and G1-2 (Fig. 2A). The HRs for OS in the IL-33⁺ Treg^high and G3, IL-33⁺ Treg^high and G1-2 and IL-33⁺ Treg^low and G3 subgroups, using IL-33⁺ Treg^low and G1-2 as a reference, were 6.610, 3.912, and 1.727, respectively. Similarly, the median RFS for these subgroups was 9.0, 13.0, 28.3, and 41.4 months, respectively (Fig. 2B), with corresponding HRs of 6.395, 4.088, and 1.762. These results demonstrate a stronger and more refined risk stratification when IL-33⁺ Tregs are assessed alongside tumor differentiation, highlighting the interplay between malignant tumor biology and immunosuppressive TME.

Figure 2.

Survival predictive value of combining intratumoral IL-33⁺ Tregs with tumor differentiation. (A, B) Kaplan–Meier plots depicting OS and RFS based on the density of tumor-infiltrating IL-33⁺ Tregs and tumor differentiation via the log-rank test. (C, D) Receiver operating characteristic curves for 1-, 3-, and 5-year OS and RFS, showing the predictive performance of IL-33⁺ Tregs, tumor differentiation, and their combination. OS refers to overall survival, RFS refers to recurrence-free survival.

Furthermore, the combined predictive model for 1-, 3-, and 5-year OS achieved area under the curve values of 0.745, 0.788, and 0.817, respectively, outperforming either factor alone (Fig. 2C). Decision curve analysis further underscored the greater net benefit of the combined model across a wider range of threshold probabilities when compared to tumor-infiltration IL-33⁺ Tregs or tumor differentiation in isolation (Supplemental Digital Content Figure 3A, available at: http://links.lww.com/JS9/E309). Consistent results were observed for RFS (Fig. 2D and Supplemental Digital Content Figure 3B, available at: http://links.lww.com/JS9/E309). Therefore, these findings highlight the worst survival outcomes in patients with high IL-33⁺ Treg levels and poorly differentiated tumors, emphasizing the predictive value of combining immunosuppressive and tumor biological features.

Prognostic variations of serum biological CA19-9 level intercepted by tumor-infiltrating IL-33⁺ Tregs

CA19-9 is the most widely used tumor biomarker for PDAC and the only FDA-approved marker for diagnosis, assessing tumor burden, and monitoring therapeutic response. Numerous studies have validated its pre- and peri-operative prognostic value for patient outcomes^[22,23]. In our subgroup analysis, adjusted for all confounding variables, interaction analysis revealed that the prognostic utility of serum CA19-9 may be influenced by the level of tumor-infiltrating IL-33⁺ Tregs (both P < 0.001 for OS and RFS). Specifically, in the low IL-33⁺ Treg group, CA19-9 demonstrated strong predictive power. Patients with high CA19-9 levels had a significantly worse median OS compared to those with low levels (34.8 vs. NA. months, P < 0.001, Fig. 3A) and a similarly worse RFS (20.6 vs. 48.6 months, P < 0.001, Fig. 3B). However, in the high IL-33⁺ Treg group, CA19-9 levels did not significantly correlate with prognosis (OS: 16.7 vs. 20.8 months, P = 0.423, Fig. 3C; RFS: 10.0 vs. 11.5 months, P = 0.222, Fig. 3D). These results suggest that tumor-infiltrating IL-33⁺ Treg play a critical role in modulating the predictive accuracy of CA19-9. These observations emphasize the dynamic crosstalk between the TME and intrinsic tumor biological behavior, implying that TME-mediated immunosuppression may exert dominant role in malignant progression.

Figure 3.

Survival analysis based on intratumoral IL-33⁺ Tregs infiltration level and serum CA19-9 level stratification. (A, B) Kaplan–Meier plots depicting OS and RFS in patients with low tumor-infiltrating IL-33⁺ Tregs density according to serum CA19-9 levels via the log-rank test. (C, D) Kaplan–Meier plots depicting OS and RFS in patients with high tumor-infiltrating IL-33⁺ Tregs density according to serum CA19-9 levels via the log-rank test. OS refers to overall survival, RFS refers to recurrence-free survival.

Therapeutic response to ACT in PDAC patients based on tumor-infiltrating IL-33⁺ Tregs

Chemotherapy has been shown to influence the TME in PDAC^[24], while the TME, in turn, can affect chemotherapy efficacy^[25,26]. Immune cells within the TME can either hinder or enhance therapeutic response by altering their patterns and functions^[27]. Our interaction analysis, adjusted for confounding factors, revealed a significant relationship between tumor-infiltrating IL-33⁺ Tregs and the impact of ACT on patient outcomes (both P < 0.001 for OS and RFS). In patients with low IL-33⁺ Treg infiltration, those receiving ACT had significantly longer OS (59.2 vs. 16.9 months, P = 0.002, Fig. 4A) and RFS (32.4 vs. 9.8 months, P = 0.012, Fig. 4B) compared to those who did not receive ACT. In the high IL-33⁺ Treg subgroup, patients receiving ACT also had a longer OS compared to those without ACT (22.5 vs. 10.0 months, P < 0.001, Fig. 4C). However, while there was a significant difference in OS within the high IL-33⁺ Treg subgroup, the OS benefit remained relatively limited. Meanwhile, no significant difference in RFS was observed between ACT and non-ACT groups in this subgroup (10.7 vs. 6.9 months, P = 0.123, Fig. 4D). These findings suggest that PDAC patients with high IL-33⁺ Tregs infiltration may derive limited benefit from ACT, whereas those with lower IL-33⁺ Treg levels are more likely to experience a favorable response to ACT.

Figure 4.

Survival analysis based on intratumoral IL-33⁺ Tregs infiltration and chemotherapy. (A, B) Kaplan–Meier plots depicting OS and RFS in patients with low tumor-infiltrating IL-33⁺ Tregs according to adjuvant chemotherapy or not. (C, D) Kaplan–Meier plots depicting OS and RFS in patients with high tumor-infiltrating IL-33⁺ Tregs according to adjuvant chemotherapy or not. Log-rank test was applied to Kaplan–Meier curves. OS refers to overall survival, RFS refers to recurrence-free survival.

Tumor-infiltrating IL-33⁺ Treg as a robust immunosuppressive phenotype in PDAC

To clarify the specific role of IL-33⁺ Tregs in the PDAC microenvironment, we retrospectively assessed the level of tumor-infiltrating IL-33⁺ Tregs in sixteen PDAC patients using their FFPE slides. In these patients, the activation and exhaustion profiles of tumor-infiltrating CD8⁺ T cells had been previously analyzed through flow cytometry, as reported in our earlier study^[17]. Representative images of high and low tumor-infiltrating IL-33⁺ Tregs were displayed in Figure 5A. Intriguingly, patients with high IL-33⁺ Treg infiltration exhibited significantly higher levels of CD8⁺ T cell exhaustion markers, including TIM3 and PD-1, compared to those with low IL-33⁺ Treg infiltration (Fig. 5B). Furthermore, the cytotoxic phenotype, indicated by GzmB expression in CD8⁺ T cells, was markedly reduced in the high IL-33⁺ Treg group (Fig. 5C).

Figure 5.

Tumor-infiltrating IL-33⁺ Tregs as a pronounced immunosuppressive phenotype in PDAC. (A) Representative images of high and low tumor-infiltration IL-33⁺ Tregs in PDAC tissues. Flow cytometry to compare the levels of exhausted markers (B) and cytotoxic cytokines (C) in tumor-infiltrating CD8⁺ T cells between IL-33⁺ Tregs high and low groups. *P < 0.05, **P < 0.01, ***P < 0.001, ns refers to not significant.

To investigate the clinical impact of IL-33⁺ Tregs, we applied an IL-33⁺ Treg gene signature to stratify PDAC patients into high and low-infiltration groups using three independent transcriptomic datasets. In all datasets, patients with high IL-33⁺ Treg infiltration had significantly worse OS, as demonstrated in the TCGA cohort (P < 0.001, Fig. 6A), GSE71729 cohort (P = 0.035, Fig. 6B), and ICGC cohort (P = 0.002, Fig. 6C). To further explore biological function of IL-33⁺ Tregs in PDAC, we then conducted differential gene expression analysis between the high and low IL-33⁺ Treg groups in these datasets. Genes related to extracellular matrix (ECM) remodeling, including collagen (FAP, COL11A1, COL17A1) and integrins (ITGB6, ITGA2), were significantly enriched in the high IL-33⁺ Treg group. TGFB1, a critical cytokine for Treg differentiation, immunosuppression, and ECM remodeling, was also up-regulated in this group (Fig. 6D). GSEA revealed that pathways associated with cell cycle, DNA replication, Th17/IL-17 signaling, and TGF-β production were significantly enriched in the high IL-33⁺ Treg group. In contrast, pathways related to pancreatic secretion, digestion, and absorption were enriched in the low IL-33⁺ Treg group (Fig. 6E). To validate the immunosuppressive properties of IL-33⁺ Tregs, we isolated tumor-infiltrating Tregs from fresh PDAC tissues, quantified their proportions via flow cytometry, and co-cultured them with human PBMC-derived CD8⁺ T cells, as previously described^[20]. As anticipated, a high proportion of IL-33⁺ Tregs significantly suppressed the cytotoxic phenotypes of CD8⁺ T cells (Fig. 6F). In summary, IL-33⁺ Tregs have a more pronounced capacity for immunosuppression, and appear to correlate with ECM remodeling, and tumor proliferation process in PDAC.

Figure 6.

Validation of the prognostic value and biological function of IL-33⁺ Tregs. (A–C) Kaplan–Meier plots depicting OS in patients with different IL-33⁺ Treg infiltration level within the TCGA, GSE71729, and ICGC cohort. (D) Volcano plot depicting the differentially expressed genes in IL-33⁺ Tregs high groups compared to IL-33⁺ Tregs low groups within the TCGA, GSE71729, and ICGC cohorts. (E) Gene Set Enrichment Analysis (Go biological process) on genes ranked by log2 fold change between IL-33⁺ Treg high and low groups within the TCGA, GSE71729, and ICGC cohorts. (F) Flow cytometry analysis for amplified CD8⁺ T cells cytotoxic phenotype after co-culturing with Tregs sorting from PDAC tissues. The proportion of IL-33⁺ Tregs in each sample was evaluated immediately after sorting, and divided into IL-33⁺ Tregs^high (>50%) and IL-33⁺ Tregs^low (<50%) groups. OS refers to overall survival, TCGA refers to The Cancer Genome Atlas; ICGC refers to International Cancer Genome Consortium.

Discussion

IL-33, a cytokine from the IL-1 family, serves as a pivotal immunomodulatory factor that influences immune cells function within the TME. This study highlights the regulatory impact of IL-33⁺ Tregs and their crucial role in survival stratification, promoting tumor progression, underscoring their contribution to immune evasion. Recent research has shown that IL-33 can originate from multiple sources, including tumor cells, stromal cells, and other immune cells within the TME, resulting in varied effects across different cancer types^[14,28,29]. In line with these observations, our study demonstrates that IL-33⁺ Tregs are not only strong predictors of prognosis and treatment outcomes in PDAC but also modulate the pronounced immunosuppression, further driving tumor growth and progression.

One key characteristic of IL-33 is its diverse origins. Alam et al^[30]. observed that IL-33 produced by tumor cells can recruit and activate TH2 cells and group 2 innate lymphoid cells, which then release tumor-promoting cytokines like IL-4, IL-5, and IL-13, driving tumor growth. In contrast, Alonso-Curbelo et al^[31]. linked IL-33 activation in Kras-mutant pancreatic epithelial cells to the combined effects of tissue damage and Kras mutations, which together trigger early epigenetic reprogramming and tumorigenesis. Hatzioannou et al^[10]. found that IL-33 is produced directly by Tregs within tumors, where it plays a crucial role in maintaining their stability and suppressive function. This intrinsic production of IL-33 by Tregs supports their immunosuppressive activity within the TME, aiding immune evasion. Additionally, IL-33 can be secreted by other cells such as fibroblasts, dendritic cells, and macrophages, contributing to its varied effects depending on the source^[32–34]. In our study, we identified Tregs as one of the primary sources of IL-33, emphasizing their pivotal role in tumor immunosuppression. While CA19-9 maintains predictive value in PDAC patients with low IL-33⁺ Treg infiltration, this correlation is abolished in those with high IL-33⁺ Treg infiltration. This suggests that the immunosuppressive function of Tregs, rather than the tumor burden and malignant behavior, plays a dominant role in shaping the tumor’s progression and clinical outcomes.

IL-33 has been found to play a dual role in both promoting inflammation and driving immunosuppression within tumors. On the one hand, IL-33 enhances Treg proliferation and function through its receptor ST2, facilitating immune escape by tumors. Our study revealed that high infiltration of IL-33⁺ Tregs is often accompanied by reduced cytotoxicity and enhanced exhaustion of CD8⁺ T cells. These findings are consistent with previous research. Studies by Alam et al and Akimoto et al demonstrated that IL-33, through the ST2 receptor, fosters an immunosuppressive TME that supports tumor growth^[30,35,36]. Additionally, macrophages have been shown to secrete TGF-β, much like tumor-initiating cells, which induces invasive and drug-resistant properties while further increasing IL-33 expression^[29,37]. This elevated IL-33 level promotes the differentiation of macrophages into an immunosuppressive phenotype^[35,38]. On the other hand, IL-33’s pro-inflammatory role is evident in its ability to activate effector immune cells. For example, Song et al^[39] found that IL-33 enhanced CD8⁺ T cell activity, contributing to anti-tumor immune responses. Dixit et al^[40] observed in a CCR2 gene knockout mouse model that reduced monocyte recruitment led to lower TNF-α levels and increased IL-33 expression, which decreased metastasis and improved survival. In addition, Song et al^[39] showed that in colorectal cancer, IL-33 increased tumor sensitivity to 5-FU in the presence of T cells. However, most studies indicate that IL-33 predominantly exerts an immunosuppressive effect in tumors.

Treg cells are well known for their ability to suppress major inflammatory responses by regulating the activity of various cells in both the innate and adaptive immune systems^[41]. Previous research has established that IL-33 activates several downstream pathways and engages in crosstalk with other cells, contributing to tumor progression. A consensus nuclear localization sequence in the N-terminus of IL-33 directs it to the cell nucleus (nIL-33), where it functions as a repressive transcriptional regulator by binding to heterochromatin in endothelial cells^[42]. Intrinsic IL-33 has also been identified in Treg cells, and we observed that IL-33 significantly located in the nuclei of tumor-infiltrating Tregs. NF-κB activity is of great importance in Tregs generation and maintenance^[43,44]. In previous studies, intrinsic nIL-33 could interfere with the DNA-binding activity of NF-κB and regulated its downstream genes transcription^[45], this may be the potential mechanism that IL-33⁺ Tregs had stronger immunosuppressive capacity. However, the detailed mechanism of nIL-33 of Tregs and its molecular characteristics in PDAC needs to be further explored. Meanwhile, Xie et al^[46] discovered that in a tobacco smoke-induced liver cancer model, IL-33 activated the p38 MAPK pathway, which promoted both cancer stem cell expansion and Treg suppression, accelerating tumor progression. Additionally, the IL-33/SMAD signaling pathway, the HIF/IL-33/GATA3 axis, and the IL-33/COX2/PGE2 axis have all been validated as key mechanisms by which IL-33 drives tumor progression and suppresses immune responses against tumors^[47–49]. Given these findings, we propose that IL-33⁺ Tregs, a distinct Treg subtype, exert a pronounced immunosuppressive effect within tumors. This subtype likely plays a pivotal role in the immunosuppressive environment of PDAC. The consistent immunosuppressive activity observed may be due either to IL-33 enhancing the suppressive function of Tregs within the TME or to IL-33⁺ Tregs themselves exerting a particularly strong tumor-suppressive effect.

Through comprehensive multidimensional analysis, our study established significant associations between IL-33⁺ Treg and clinicopathological characteristics including elevated serum CA19-9 levels, microvascular invasion, and perineural invasion in PDAC. The higher CA19-9 levels, microvascular and perineural invasion reflect the aggressive malignant behavior of PDAC, which hints the potential role of IL-33⁺ Tregs in PDAC progression. One recent study has described CA19-9 production at single-cell resolution and the dynamics of the immune atlas in terms of CA19-9, and found two high macrophage subtypes were associated with an unfavorable clinical prognosis in PDAC^[50]. Recent studies have also shown that perineural invasion is associated with impaired immune responses^[51,52]. For instance, Yang et al^[51] found that the perineural invasion is characterized by a reduction in CD8⁺ T cells and a lower Th1/Th2 ratio. Additionally, we observed that a high infiltration of IL-33⁺ Tregs is linked to poorer prognosis in PDAC patients. Historically, the academic community has long debated the role of IL-33 in PDAC, with differing views on its origins and whether it primarily promotes or suppresses tumor growth^{[30,32,33,40]}. Our research provides new evidence that IL-33⁺ Tregs act as adverse prognostic factors, indicating that this Treg subset, more than other IL-33 sources, plays a central role in tumor immune suppression in PDAC. In this study, we were the first to identify the distinctive prognostic role of IL-33⁺ Tregs, which proved to be more significant than that of conventional Tregs. By stratifying patients based on the level of tumor-infiltrating IL-33⁺ Tregs, we developed a more precise predictive model for clinical outcomes, identified a cohort with no survival benefit, irrespective of CA19-9 levels, and a specific group that would benefit from ACT. As we know, tumor differentiation and serum CA19-9 level both reflect the aggressive malignant behavior in PDAC^[21,53]. Our model, which incorporates tumor differentiation and IL-33⁺ Tregs, captures both the intrinsic tumor behavior and the immunosuppressive microenvironment, potentially offering greater accuracy than existing models that just reflect tumor malignancy. Importantly, tumor-infiltrating IL-33⁺ Tregs emerged as a critical factor influencing the predictive accuracy of CA19-9, highlighting the interaction between the TME and tumor self-characteristics. This suggests that the immunosuppressive components may play a more dominant role than the tumor’s biological properties in driving tumor progression. By leveraging these insights, clinicians may be better equipped to identify a specific cohort whose survival is unaffected by CA19-9 levels, allowing for more personalized treatment approaches. Additionally, increasing evidence suggests that the extent of immune cell infiltration and their functional status significantly impact the effectiveness of ACT. Di Caro et al^[54]. found that TAMs play a pivotal role in determining the response to ACT. In our previous study, we also demonstrated that lower levels of tumor-infiltrating CD161⁺CD8⁺ T cells were linked to a better response to ACT^[17]. Building on these findings, our current study proposes IL-33⁺ Tregs as a novel predictive marker for responsiveness to ACT, further refining the ability to tailor treatment strategies in PDAC.

However, several limitations of this study should be acknowledged. First, the retrospective, single-center design may restrict the generalizability of our findings. A larger, prospective, multicenter cohort is required to validate the robustness of these results. Additionally, the study focused on patients eligible for resection, predominantly those with stages I and II disease. To avoid potential effects on the TME, some stage III patients who had received neoadjuvant therapy for PDAC were excluded. Stage IV patients, being ineligible for surgical intervention, lacked available tumor specimens. Nevertheless, the distribution of patients across stages in our cohort is consistent with other studies^[25,55,56]. Moreover, our research focused on the translational relevance of IL-33⁺ Tregs, highlighting their potential clinical significance. Further studies are necessary to comprehensively understand the role of IL-33 within Tregs. Additionally, the potential involvement of IL-33⁺ Tregs in peripheral circulation and their relationship with tumor-infiltrating IL-33⁺ Tregs require further exploration. Investigating the potential of targeting these cells as a novel immunotherapy strategy presents a promising avenue for future research.

Conclusion

Our study highlights the significance of IL-33⁺ Tregs as a promising biomarker for predicting both survival and therapeutic response to ACT in PDAC. Additionally, their pronounced immunosuppressive role underscores their potential as an appealing target for future immunotherapy strategies.

Ethical approval

The protocol of this study has been approved by the institutional review board and ethics committee of Zhongshan Hospital, Fudan University (B2023-129). Written informed consent was obtained from individual or guardian participants. The protocol was registered in Research Registry (researchregistry11246).

Consent

Not applicable.

Sources of funding

This work was supported by grants from Shanghai Municipal Health Commission (20244Y0023, 201940019), National Natural Science Foundation of China (82103409, 82273382, 82473459), Beijing Xisike Clinical Oncology Research Foundation (Y-2022METAZQN-0003, Y-Gilead2024-PT-0002, Y-HR2022MS-0251), Shanghai “Rising Stars of Medical Talents” Youth Development Program, Open Research Fund of Key Laboratory of Gastrointestinal Cancer (Fujian Medical University), Ministry of Education (FMUGIC-202301), Fund of Fujian Provincial Key Laboratory of Translational Cancer Medicine (TCM2024-2), Program of Shanghai Academic/Technology Research Leader (23XD1400600), Shanghai Anticancer Association EYAS PROJECT (SACA-CY24B09), Youth Fund Project of Zhongshan hospital (2024ZSQN33, ZSZP202401) and Liu Liang Expert Workstation of Yunnan Province (202305AF150148).

Author contributions

N.P., H.Y., W.L., and L.L. contributed to funding acquisition, data acquisition, data analysis, manuscript writing, manuscript review. Q.C., J.H., and Z.X. were involved in data acquisition, data analysis, statistical analysis, manuscript writing. Z.J., T.H., Y.A., Y.X., W.G., H.W., W.W., W.W., Y.J., and J.Y. contributed to data acquisition, manuscript review. All authors read and approved the final version of the manuscript. N.P. supervised the study. L.L. is responsible for the overall content as guarantor.

Conflicts of interest disclosure

None declared.

Research registration unique identifying number (UIN)

Research Registry; Intratumoral IL-33⁺CD4⁺FoxP3⁺ regulatory T cell infiltration determines poor clinical outcomes and intensive immunoevasive contexture in patients with pancreatic cancer after surgical resection; researchregistry7046; https://www.researchregistry.com/browse-the-registry.

Guarantor

Liang Liu.

Provenance and peer review

Not commissioned, externally peer-reviewed.

Data availability statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy and ethical restrictions.