Hyaluronan accumulation is associated with reduced hyaluronidase expression in renal cell carcinoma, with CD44, HAS1, and HYAL2 emerging as prognostic markers

Abstract

Hyaluronan (HA), a large extracellular matrix glycosaminoglycan, is associated with malignant features in several human cancers. The accumulation of HA in renal cell carcinomas (RCC) correlates with unfavorable outcomes, higher tumor grade, and more advanced disease stages. However, the mechanisms responsible for HA buildup in these neoplasms remain unclear, and studies on the expression of hyaluronan‐metabolizing and ‐degrading enzymes are either lacking or conflicting. This study aims to address this knowledge gap. Formalin‐fixed paraffin‐embedded (FFPE) RCC samples of various histological subtypes from 315 patients were immunohistochemically stained for CD44 (the main receptor of HA), hyaluronan‐synthesizing enzymes HAS1–3, and degrading enzymes HYAL1–2. Protein expression levels were correlated with clinicopathological variables and their prognostic significance was evaluated. Additionally, the mRNA expression levels of these proteins were examined using RNA extracted from the same samples and publicly available data from the cancer genome atlas (TCGA). CD44 protein expression was associated with increased tumoral HA content, poor prognosis, higher tumor grade, advanced stage, and sarcomatoid/rhabdoid changes. HYAL1 and HYAL2 protein levels were reduced in HA‐positive tumors, and low HYAL2 expression predicted worse prognosis. Elevated HAS2 protein expression was associated with poor differentiation, while low HAS1 protein levels were associated with reduced survival. mRNA levels of CD44 and HYAL2 correlated with their respective protein expression levels, and CD44 mRNA expression was also associated with HA content. In RCC, HA accumulation appears to be primarily driven by decreased degradation. HAS1 and HYAL2 were identified as novel prognostic biomarkers. These findings provide new insights into HA metabolism in RCC and open potential avenues for better understanding and management of these tumors.

Introduction

Renal cell carcinoma (RCC) is the most prevalent malignant neoplasm of the human kidney, accounting for approximately 2% of global cancer incidence and cancer‐related mortality [1]. Kidney cancer ranks as the seventh most common malignancy in males and the tenth most common malignancy in females [2]. The most frequent histological subtype is clear cell RCC (ccRCC, ~70%), followed by papillary RCC (pRCC, 10–15%), and chromophobe RCC (chRCC, 3–5%) [3]. In recent decades, novel molecularly defined RCC entities have been identified and incorporated into the latest (5th) edition of the World Health Organization (WHO) Classification of Tumors of Urinary and Male Genital Organs [4, 5].

Approximately 20% of patients with newly diagnosed RCC present with metastatic disease, and over one‐third of patients with initially localized disease ultimately develop metastases following surgical intervention [6]. The survival rate of RCC is estimated at over 90% in early‐stage disease; however, the 5‐year survival rate drops significantly to around 25% in advanced disease [7]. Surgery is the standard treatment for localized RCC, with a focus on nephron‐sparing procedures to preserve renal function. Active surveillance or ablative techniques are usually considered for small renal masses (<4 cm), especially in frail patients [8].

For metastatic RCC, the first targeted therapies included tyrosine kinase inhibitors targeting the vascular endothelial growth factor receptor (VEGFR) and the mammalian target of rapamycin (mTOR) inhibitors [8, 9, 10]. In recent years, immune checkpoint inhibitors have demonstrated significant survival benefits, and current guidelines recommend immune checkpoint inhibitor‐based combination therapies as the preferred first‐line treatment for patients with metastatic disease. Local therapy for metastases and cytoreductive nephrectomy may be considered in selected cases with oligometastatic disease [8, 11, 12]. Nonetheless, resistance to these therapies eventually develops, highlighting the need for novel prognostic and therapeutic biomarkers.

Hyaluronan (HA) is a large glycosaminoglycan naturally found in the extracellular matrix (ECM) of the skin, serous cavities, and interstitial spaces. It is synthesized by hyaluronan synthase enzymes (HAS1, HAS2, and HAS3) and degraded by a family of hyaluronidases, among which HYAL1 and HYAL2 are the most well characterized [13, 14]. HA is a major ECM component that contributes to cancer progression and is associated with a poor prognosis in multiple malignancies, including breast, colorectal, pancreatic, and prostate carcinomas [15, 16, 17, 18]. Conversely, reduced HA content has been reported in head and neck squamous cell carcinomas and melanomas [19, 20]. HA has been hypothesized to function as a barrier that protects tumor cells from the immune system and anti‐cancer drugs [21]. Accordingly, targeting HA and its primary receptor, CD44, has been proposed as a potential therapeutic strategy to enhance the efficacy of anti‐cancer treatments [22, 23]. Elevated CD44 levels have been demonstrated to correlate with poor prognosis in RCC; however, there is a lack of research exploring the potential co‐occurrence of HA and CD44 [24, 25].

In our previous study, we demonstrated that hyaluronan accumulation in RCCs is an independent factor for poor prognosis [26]. Increased cellular HA levels were associated with higher tumor grade and more advanced disease stage. Notably, HA accumulation predicted worse outcomes in low‐grade RCC [International Society of Urological Pathology (ISUP) grades 1 and 2]. Additionally, we found that HA accumulation in RCCs is associated with numerous biological pathways related to the ECM, epithelial‐mesenchymal transition (EMT), and tumor–stroma interactions [27]. We also identified a hyaluronan‐related gene signature that stratified patients into prognostically distinct cohorts. However, the relative contributions of HA‐synthesizing and HA‐degrading enzymes to HA accumulation, and their prognostic roles in RCC, remain unclear. This study aims to address that gap.

Materials and methods

Patient data and sample collection

Clinical and histopathological data from 315 patients with RCC who underwent surgery between 2000 and 2013 were collected from the Biobank of Eastern Finland. The study (Hyaluronan in Renal Cell Carcinoma, HARCC) was conducted in accordance with the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of the Northern Savo Hospital District (379/2016, November 1, 2016). The requirement for patient consent was waived because the data were obtained from the Biobank of Eastern Finland. Clinical data are presented in Table 1.

Table 1.

Clinicopathological variables

	n = 315	(%)	Mean	SD
Sex
Male	169	(53.7)
Female	146	(46.3)
Age (years)			65.3	10.57
Histology
Clear cell	284	(90.2)
Papillary	18	(5.7)
Chromophobe	8	(2.5)
Other	5	(1.6)
ISUP grade
Grade 1	23	(7.5)
Grade 2	165	(53.8)
Grade 3	61	(19.9)
Grade 4	58	(18.9)
Not graded (chromophobe RCCs)	8
Sarcomatoid change
Yes	34	(10.8)
No	281	(89.2)
Rhabdoid change
Yes	27	(8.6)
No	288	(91.4)
Clinical Stage (UICC)
I	160	(51.1)
II	39	(12.5)
III	62	(19.8)
IV	52	(16.6)
Metastasis at diagnosis (M1)
Yes	50	(15.9)
No	265	(84.1)
Follow‐up time (months)	315		98.8	70.3
Disease relapse
Yes	68	(25.3)
No	201	(74.7)
Disease related death
Yes	95	(30.2)
No	220	(69.8)

ISUP, International Society of Urological Pathology; UICC, Union for International Cancer Control.

Surgical specimens were processed according to the routine clinical diagnostic protocol of the Department of Clinical Pathology at Kuopio University Hospital, which has remained consistent throughout the sample collection period. Diagnoses and tumor grades were re‐evaluated according to the WHO Classification of Tumours of the Urinary System and Male Genital Organs (4th ed., 2016) [28]. Clinical follow‐up data were collected between the years 2000 and 2024.

The samples were previously stained with a biotinylated hyaluronan‐binding complex (bHABC, 1.25 μg/ml), as described by Tammi et al and reiterated by Siiskonen et al [26, 29, 30]. Tumor‐infiltrating lymphocytes (TILs) were previously assessed from whole‐slide images (WSIs), following the guidelines of the International Immuno‐Oncology Biomarkers Working Group [31].

Additionally, RNA sequencing data from 96 tumors were obtained, and differential expression analysis (DEA) was conducted on 48 HA‐positive and 48 HA‐negative samples, as described in our previous study [27]. The resulting list of differentially expressed genes was used to perform cluster analysis on the cancer genome atlas (TCGA) kidney renal clear cell carcinoma (KIRC) dataset, identifying two sample cohorts: one with a gene expression profile predictive of hyaluronan positivity and the other with predicted hyaluronan negativity [32].

Tissue microarray construction

The formalin‐fixed paraffin‐embedded (FFPE) material used in this study was obtained from the Biobank of Eastern Finland. The specimens included cores obtained from 315 patients. One pathologist (OJ) annotated the core locations used for the tissue extraction. The selection was conducted to ensure that a minimum of two 1‐mm diameter cores were obtained from each tumor, with emphasis on hyaluronan‐positive areas when available. Additional cores were extracted from the same tumor if higher‐grade regions were present, thereby accounting for intratumoral heterogeneity. Furthermore, 1–2 tissue punches were obtained from lymphovascular invasive foci and distant metastases when available.

Tissue microarrays (TMAs) were constructed using the Galileo CK4500 TMA platform at the Biobank of Eastern Finland. A total of 785 punches were compiled into eight TMA blocks, each arranged in a 13 × 8 grid. A total of 660 cores from primary tumors, 79 from venous invasion, and 45 from distant metastases were obtained from 315, 42, and 23 patients, respectively.

Immunohistochemistry and hyaluronan staining

The TMA material was immunohistochemically stained for CD44, HYAL1, HYAL2, HAS1, HAS2, and HAS3. The antibodies used are listed in Table 2. Immunohistochemical (IHC) staining was performed using an Autostainer Link 48 (Agilent, Santa Clara, CA, USA) and BenchMark Ultra (Roche Ventana, Tucson, AZ, USA). To confirm the HA status in the individual TMA cores, the samples were restained with bHABC (1.25 μg/ml) using a previously described protocol [26, 29, 30].

Table 2.

Primary antibody details

Antibody	Clonality	Host species	Company/Vendor	Concentration	Catalog number	LOT number
CD44	Monoclonal	Mouse	BioSB	Recommended dilution 1:250–1:1000	BSB‐6236	6238TOA05
HYAL1	Polyclonal	Rabbit	Sigma‐Aldrich	0.2 mg/ml	HPA002112‐100	C115919
HYAL2	Polyclonal	Rabbit	Abcam	1 mg/ml	Ab68608	GR3267063‐2
HAS1	Monoclonal	Mouse	GeneTex	NA	GTX82799‐100	822402988
HAS2	Monoclonal	Mouse	GeneTex	1 mg/ml	GTX60647‐100	822402988
HAS3	Monoclonal	Mouse	GeneTex	1 mg/ml	GTX60588‐100	822402988

TMA slides were pretreated with the following reagents: (1) anti‐HYAL1 (polyclonal, Sigma‐Aldrich, St. Louis, MO, USA), HYAL2 (polyclonal, Abcam, Cambridge, UK), anti‐HAS1 (clone 3E10, GeneTex), and anti‐HAS3 (clone 3C9, GeneTex, Irvine, CA, USA) in a pretreatment module PT Link (Agilent) with EnVision FLEX Target Retrieval Solution High pH for 20 min at 97 °C (Agilent); (2) anti‐HAS2 (clone 4E7, GeneTex) in a pretreatment module PT Link (Agilent) with EnVision FLEX Target Retrieval Solution Low pH for 20 min at 97 °C (Agilent); and (3) anti‐CD44 (clone BSB‐12, Bio SB) using the CC1 buffer (Roche Ventana) onboard the automated staining platform for 64 min at 98 °C.

Primary antibodies were stained using Autostainer Link 48 (Agilent); endogenous peroxidase activity was blocked with Dako REAL Peroxidase‐Blocking Solution (Agilent), and antigens were visualized using the EnVision FLEX+, Mouse, High pH (Link) Kit (K8002, Agilent). The incubation time for the HYAL1 (1:100), HYAL2 (1:400), HAS1 (1:100), HAS2 (1:150), and HAS3 (1:150) antibodies was 30 min at room temperature. Slides were counterstained with EnVision FLEX Hematoxylin (Link) (K8008, Agilent).

Sections were stained with primary antibody against CD44 (1:700) using a BenchMark Ultra (Roche Ventana). The OptiView DAB IHC Detection Kit (#760‐700, Roche Ventana) contained all reagents for endogenous peroxidase blocking and visualization. The primary antibody incubation time was 32 min without heating. Sections were counterstained with hematoxylin II (#790‐2208, Roche Ventana).

In all cases, EnVision FLEX Antibody Diluent (K8006, Agilent) was used as a diluent, and the slides were mounted using Tissue‐Tek Film (Sakura Finetek, Tokyo, Japan). IHC protocols were verified through a combination of literature review, empirical optimization, and the use of appropriate control tissues. The expected localization of each antigen in normal human tissues and subcellular compartments was confirmed based on published data and manufacturer guidelines. In the accepted IHC setting, antigens were detected at the expected histological and cellular locations [33]. Nonspecific background staining or staining of cell components assumed to be negative for the antibody used was not accepted. Control tissues (kidney, tonsil, skin, and breast) were included in every IHC staining replicate as patch control slides [34]. Staining protocols, including antigen retrieval and antibody dilution, were optimized for each antibody to achieve clear signal with minimal background. For bHABC staining, specificity was validated by performing negative controls in which the bHABC was omitted from the staining protocol.

TMA evaluation and scoring

Two independent observers (OJ and RS) evaluated the stained TMA slides. Discrepancies between evaluations were discussed and resolved. The cytoplasmic, membrane, and nuclear compartments were evaluated separately. The staining intensity was scored using a semi‐quantitative four‐tier system as 0 = absent, 1 = weak, 2 = moderate, and 3 = strong. In addition, the percentage of staining coverage was evaluated for each core and rounded to the nearest 10%.

To make the data comparable across cores, a histochemical score (H‐score) was calculated by multiplying the percentage of cells at each staining intensity (0–3) by the corresponding intensity score, yielding a range of 0–300. For statistical analyses, an H‐score of 0 was considered negative, an H‐score of 1–100 was considered weak, and a score ≥100 was considered strong. Owing to the very low number of negative samples for HYAL1, HYAL2, HAS1, and HAS3 staining, the negative and weak samples were combined for statistical analyses. If both membranous and cytoplasmic staining were observed, the highest compartmental H‐score was used for statistical analysis.

Immunohistochemical staining results from each individual core were used to assess correlations with core‐specific variables, including ISUP grade, sarcomatoid change, and rhabdoid change. For case‐level variables such as prognosis, stage, histological subtype, tumor size, metastasis status, and TIL score, the highest staining value across all cores from a given case was used when variation was present. This approach was chosen based on the rationale that focal areas of high marker expression are most likely to represent regions of greatest biological and clinical relevance.

The bHABC histochemical staining was assessed according to a previous publication, with any amount of positivity in tumor cells considered positive [26]. For statistical purposes, a sample was considered positive for hyaluronan if any core showed positive staining. TIL percentages from previously evaluated WSIs were aggregated into four‐tier categorical variables ranging from 0 (none) to 3 (high).

RNA sequencing and expression analysis

Ninety‐six samples with available RNA data were divided into two groups based on staining results. For CD44, samples were categorized as positive or negative, whereas for other proteins, they were classified as low or high. DEA was then performed on the stratified cohorts. The materials and sequencing pipeline, including quality control and count normalization, used to obtain RNA data were described in detail in our previous article [27]. Additionally, the Wilcoxon rank‐sum test was used to compare normalized transcripts per million values between groups stratified by staining results.

For in silico analyses, TCGA KIRC samples were used to compare the RNA expression levels of the genes encoding the proteins of interest between two previously identified cohorts: one with a gene expression profile predicting hyaluronan positivity and the other with predicted hyaluronan negativity [32].

Statistical analysis

All statistical analyses were conducted using RStudio (2024.09.1 + 394) and R (version 4.4.1) for Windows [35, 36]. Associations between clinicopathological variables and staining results were evaluated using Kruskal–Wallis test, Mann–Whitney U test, chi‐square test, and Fisher's exact test when appropriate. Statistical significance was set at p < 0.05, and the Benjamini‐Hochberg procedure was applied to correct for multiple tests.

Survival rates were evaluated using Kaplan–Meier plots, and the groups were compared using the log‐rank test. Survival analysis was performed using the Cox proportional hazards regression model. The endpoint of disease‐specific survival (DSS) analysis was death from RCC. Metastasis‐free survival (MFS) events were defined as the date of discovery of metastases or local recurrence. Censored events included loss to follow‐up, death from other causes, or survival at the time of the last follow‐up.

DEA was conducted using edgeR (v4.4.1), with p value adjustment performed using the Benjamini‐Hochberg method to control the false discovery rate [37]. Only genes with mean counts >1 in all the samples were included. In addition, the Wilcoxon rank‐sum test was used to compare the normalized RNA and protein expression levels.

Results

Protein expression patterns

CD44 protein expression in RCC cells was observed primarily on the plasma membrane, with 190 (60.3%) cases being CD44‐negative, 79 (25.1%) having weak positivity, and 46 (14.6%) having strong CD44 expression. A minority of cases [29/315 (9.2%)] showed cytoplasmic staining. No nuclear staining was observed. The mean membranous H‐score was 29.1 ± 69.7.

HYAL1 staining was predominantly positive [304/315 (96.5%)], with 242/315 (76.8%) cases showing cytoplasmic staining and 176/315 (55.9%) displaying membranous staining. No nuclear staining was observed. The intensity of staining was generally weak, with mean cytoplasmic and membranous H‐scores of 81.4 ± 88.0 and 59.5 ± 75.2, respectively. Tumors with eosinophilic cytoplasm exhibited significantly higher cytoplasmic HYAL1 expression (mean H‐score 119.8 ± 100.9) compared to non‐eosinophilic tumors (69.2 ± 80.1; Wilcoxon rank‐sum test, p = 2.97 × 10⁻⁵) irrespective of histological subtype (Figure 1).

Figure 1.

Collage of staining results in chromophobe renal cell carcinoma (RCC), low‐grade (ISUP grade 2) papillary RCC, and high‐grade (ISUP grade 3) papillary RCC with eosinophilic features.

For HYAL2, the vast majority (306/315, 97.1%) of cases were positive, and staining was observed both on the membrane [258/315 (81.9%)] and in the cytoplasm [238/315 (75.6%)]. Nuclear staining was observed in most cases [221/315 (70.2%)]. Nuclear staining intensity was generally weak, with a mean H‐score of 62.9 ± 74.0. The mean membranous H‐score was higher at 123.6 ± 97.5, and the mean cytoplasmic score was 72.4 ± 78.1. Spearman's rank correlation showed a positive correlation between nuclear and cytoplasmic/membranous expression (ρ = 0.60, p < 2.2 × 10⁻¹⁶, n = 313).

Due to tissue loss in deeper TMA sections, the number of evaluable samples was slightly lower for HAS1–3 than for CD44 and HYAL1–2.

HAS1 protein was ubiquitously expressed in RCCs, with almost all cases [311/314 (99.0%)] showing cytoplasmic staining and most [280/314 (89.2%)] exhibiting membranous staining. The staining intensity was high [280/314 (89.2%)]. The mean H‐scores for the primary tumors were 163.8 ± 85.7 and 187.3 ± 90.9 for cytoplasmic and membranous staining, respectively. Predominantly weak nuclear staining was observed in 73/301 (24.3%) samples.

HAS2 protein expression was heterogeneous, with 19 (6.1%) out of 313 cases being negative, 139 (44.4%) weakly positive, and 155 (49.5%) strongly positive. The mean H‐scores for primary tumors were 62.3 ± 74.9 and 106.9 ± 79.1 for cytoplasmic and membranous staining, respectively. Strong nuclear staining was observed in eight cases (2.6%).

HAS3 was expressed in most of the samples. Of the 314 cases, 3 (1.0%) were negative, 212 (67.5%) had low expression, and 99 (31.5%) had high expression. The mean H‐score for cytoplasmic staining was 50.5 ± 58.6 and for membranous staining was 86.5 ± 53.0. A total of 276 (87.9%) cases were weakly positive for nuclei, with a mean H‐score of 61.3 ± 48.5.

Association with hyaluronan content

The bHABC staining results were re‐evaluated from the TMA cores, and the core staining results were consistent with those of the WSIs. A summary of the protein staining results with respect to hyaluronan status is shown in Figure 2. Higher CD44 protein expression was observed in hyaluronan‐positive RCCs. Among the CD44‐high cases, 26 out of 31 (83.8%) were hyaluronan positive. In comparison, 41 out of 66 (61.2%) CD44‐low cases and 89 out of 216 (41.2%) CD44‐negative cases were hyaluronan positive (Table 3).

Figure 2.

Association of protein expression levels with hyaluronan status and clinicopathological variables in RCCs. The y‐axis denotes proportional ratios. The total number of samples in each category is indicated within the respective bar. Statistical significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Table 3.

Associations between CD44, HYAL1, and HYAL2 protein expression and clinicopathological variables

	Total, N (%)	CD44 H‐score, N (%)			p	HYAL1 H‐score, N (%)		p	HYAL2 H‐score, N (%)		p
	315 (100.0)	0 (negative)	1–100 (low)	≥101 (high)		0–100 (low)	≥101 (high)		0–100 (low)	≥101 (high)
ISUP grade
I	23 (7.5)	20 (9.6)	2 (3.6)	1 (2.4)	<0.001 *	17 (10.2)	6 (4.0)	0.005 **	7 (5.3)	16 (8.7)	0.022 **
II	165 (53.7)	139 (66.4)	29 (51.8)	2 (4.8)		93 (56.0)	77 (51.7)		62 (47.3)	108 (58.7)
III	60 (19.5)	35 (16.8)	17 (30.4)	10 (23.8)		18 (10.8)	43 (28.9)		27 (20.6)	35 (19.0)
IV	59 (19.2)	23 (10.3)	18 (32.1)	18 (42.9)		38 (22.9)	23 (15.4)		35 (26.7)	25 (13.6)
Stage
I	160 (51.1)	134 (61.8)	21 (32.8)	5 (16.1)	<0.001 **	79 (47.6)	81 (55.1)	0.624**	51 (38.9)	109 (59.9)	0.00267 **
II	39 (12.5)	23 (10.6)	11 (17.2)	5 (16.1)		25 (15.1)	14 (9.5)		15 (11.5)	24 (13.2)
III	62 (19.8)	32 (14.8)	18 (28.1)	11 (35.5)		33 (19.9)	29 (19.7)		33 (25.2)	29 (15.9)
IV	52 (16.6)	28 (12.9)	14 (21.9)	10 (32.3)		29 (17.5)	23 (15.6)		32 (24.4)	20 (11.0)
Histology
ccRCC	284 (91.6)	197 (92.5)	58 (89.2)	28 (90.3)	0.593*	155 (95.7)	127 (87.0)	0.041 *	111 (86.0)	172 (95.6)	0.0121 *
Papillary	18 (5.8)	10 (4.7)	5 (7.7)	3 (9.7)		6 (3.7)	12 (8.2)		12 (9.3)	6 (3.3)
Chromophobe	8 (2.6)	6 (2.8)	2 (3.1)	0 (0.0)		1 (0.6)	7 (4.8)		6 (4.7)	2 (1.1)
Sarcomatoid
No	281 (89.2)	208 (95.9)	56 (84.9)	18 (58.1)	<0.001 *	146 (88.0)	136 (91.3)	0.624**	107 (81.7)	174 (94.6)	0.00267 **
Yes	34 (10.8)	9 (4.2)	10 (15.2)	13 (41.9)		20 (12.0)	13 (8.7)		24 (18.3)	10 (5.4)
Rhabdoid
No	288 (91.4)	206 (94.9)	61 (92.4)	20 (64.5)	<0.001 *	150 (90.9)	137 (91.9)	0.963**	117 (89.3)	171 (92.9)	0.354**
Yes	27 (8.6)	11 (5.1)	5 (7.6)	11 (35.5)		15 (9.1)	12 (8.1)		14 (10.7)	13 (7.1)
Metastasis at diagnosis
M0	265 (84.1)	191 (88.0)	52 (78.8)	21 (67.7)	0.0109 *	139 (83.7)	126 (84.6)	0.963**	101 (77.1)	164 (89.1)	<0.0118 **
M1	50 (15.9)	26 (12.0)	14 (21.2)	10 (32.3)		27 (16.3)	23 (15.4)		30 (22.9)	20 (10.9)
Tumor size
T1	170 (54.5)	142 (65.4)	22 (34.4)	6 (20.0)	<0.001 *	82 (49.4)	88 (60.3)	0.261**	56 (43.1)	114 (62.6)	0.00375 **
T2	44 (14.1)	29 (13.4)	10 (15.6)	5 (16.7)		30 (18.1)	14 (9.6)		18 (13.8)	26 (14.3)
T3	87 (27.9)	41 (18.9)	29 (45.3)	16 (53.3)		47 (28.3)	40 (27.4)		49 (37.7)	38 (20.9)
T4	11 (3.5)	5 (2.3)	3 (4.7)	3 (10.0)		7 (4.2)	4 (2.7)		7 (5.4)	4 (2.2)
Hyaluronan
Negative	158 (50.3)	127 (58.8)	25 (37.9)	5 (16.1)	<0.001 **	72 (43.4)	88 (59.1)	0.00382 **	41 (31.3)	117 (63.6)	<0.001 **
Positive	156 (49.7)	89 (41.2)	41 (62.1)	26 (83.9)		94 (56.6)	61 (40.9)		90 (68.7)	67 (36.4)
TILs
None	191 (61.0)	141 (65.0)	37 (56.9)	12 (40.0)	0.127*	97 (58.4)	94 (63.9)	0.579**	65 (50.0)	126 (68.9)	0.0118 **
Low	54 (17.3)	36 (17.0)	12 (18.5)	6 (20.0)		26 (15.7)	28 (19.0)		31 (23.8)	23 (12.6)
Moderate	35 (11.2)	21 (9.7)	7 (10.8)	7 (23.3)		22 (13.3)	13 (8.8)		17 (13.1)	18 (9.8)
High	33 (10.5)	19 (8.8)	9 (13.9)	5 (16.7)		21 (12.7)	12 (8.2)		17 (13.1)	16 (8.7)
Type
Primary	315 (83.1)	190 (86.8)	79 (79.8)	46 (76.7)	0.176*	166 (85.1)	149 (81.4)	0.782**	131 (82.4)	184 (84.0)	0.132**
LVI	42 (11.1)	20 (9.1)	11 (11.1)	10 (16.7)		19 (9.7)	22 (12.0)		22 (13.8)	19 (8.7)
Metastasis	22 (5.8)	9 (4.1)	9 (9.1)	4 (6.7)		10 (5.1)	12 (6.6)		6 (3.8)	16 (7.3)

ccRCC, clear cell renal cell carcinoma; ISUP, International Society of Urological Pathology; LVI, lymphovascular invasion; TIL, tumor‐infiltrating lymphocytes.

p‐values with statistical significance (p < 0.05) are shown in bold.

^*Fisher's exact test.

^**Pearson's χ ² test.

Lower HYAL1 and HYAL2 protein expression levels were observed in the hyaluronan‐positive samples (Table 3, Figure 2). The Wilcoxon rank‐sum test was used to compare the mean H‐scores. The mean HYAL1 cytoplasmic scores were 81.9 ± 78 for HA‐positive and 82.1 ± 95.5 for HA‐negative samples (p = 0.070). However, the corresponding membranous H‐scores were 44.1 ± 67.5 for HA‐positive and 72.7 ± 78.8 for HA‐negative samples (p = 4.46 × 10⁻⁷).

The mean HYAL2 cytoplasmic H‐scores were 65.1 ± 66.1 for HA‐positive and 79.3 ± 86.3 for HA‐negative samples (p = 0.23). For membranous staining, the corresponding scores were 102.8 ± 80.4 and 158.6 ± 102.5, respectively (p = 6.82 × 10⁻¹³). Nuclear staining mean H‐scores were 40.7 ± 63.4 for HA‐positive samples and 81.7 ± 77.0 for HA‐negative samples (p = 1.14 × 10⁻¹⁴).

A significant reduction in HAS1 expression was observed in HA‐positive cases when the individual cores were analyzed. However, this difference was not observed when comparing the highest HAS1 H‐score per patient. The mean cytoplasmic H‐score for HA‐positive samples was 142.4 ± 77.6 compared to 147.5 ± 81.1 in HA‐negative samples (p = 0.51). For membranous staining, the scores were 156.1 ± 81.2 and 178.1 ± 88.2, respectively (p = 0.00053).

On average, HA‐positive cases had lower HAS3 membranous H‐scores (76.3 ± 47.3 versus 94.9 ± 55.8) but higher cytoplasmic scores (53.3 ± 55.8 versus 48.3 ± 60.7) than HA‐negative cases (p = 3.29 × 10⁻⁵ for membranous and p = 0.048 for cytoplasmic staining) (Table 4). However, the median value and interquartile range (IQR) were the same in both groups (median = 100, IQR 30–100).

Table 4.

Associations between HAS1‐3 protein expression and clinicopathological variables

	Total, N (%)	HAS1 H‐score, N (%)		p	HAS2 H‐score, N (%)			p	HAS3 H‐score, N (%)		p
	315 (100.0)	0–100 (low)	≥101 (high)		0 (negative)	1–100 (low)	≥101 (high)		0–100 (low)	≥101 (high)
ISUP grade
I	23 (7.5)	3 (8.8)	20 (7.1)	0.0422 *	5 (26.3)	15 (10.9)	3 (1.9)	<0.001 *	18 (8.4)	4 (4.0)	0.690**
II	165 (53.7)	16 (47.1)	153 (54.6)		11 (57.9)	87 (63.0)	70 (45.2)		116 (54.0)	54 (54.5)
III	60 (19.5)	2 (5.9)	59 (21.1)		2 (10.5)	18 (13.0)	41 (26.5)		40 (18.6)	21 (21.2)
IV	59 (19.2)	13 (38.2)	48 (17.1)		1 (5.3)	18 (13.0)	41 (26.5)		41 (19.1)	20 (20.2)
Stage
I	160 (51.1)	12 (35.3)	147 (52.9)	0.1593*	10 (52.6)	78 (56.5)	71 (46.1)	0.293*	108 (50.5)	51 (52.0)	0.497**
II	39 (12.5)	3 (8.8)	36 (12.9)		2 (10.5)	21 (15.2)	16 (10.4)		29 (13.6)	10 (10.2)
III	62 (19.8)	10 (29.4)	52 (18.7)		4 (21.1)	24 (17.4)	33 (21.4)		47 (22.0)	15 (15.3)
IV	52 (16.6)	9 (26.5)	43 (15.5)		3 (15.8)	15 (10.9)	34 (22.1)		30 (14.0)	22 (22.4)
Histology
ccRCC	284 (91.6)	29 (87.9)	253 (92.0)	0.347*	17 (94.4)	128 (94.8)	136 (91.3)	0.413*	196 (92.9)	86 (88.7)	0.497*
Papillary	18 (5.8)	2 (6.1)	16 (5.8)		0 (0.0)	6 (4.4)	7 (4.7)		11 (5.2)	7 (7.2)
Chromophobe	8 (2.6)	2 (6.1)	6 (2.2)		1 (5.6)	1 (0.7)	6 (4.0)		4 (1.9)	4 (4.1)
Sarcomatoid
No	281 (89.2)	26 (76.5)	256 (91.4)	0.0422 *	19 (100.0)	129 (92.8)	133 (85.8)	0.120*	192 (89.3)	89 (89.9)	1.000**
Yes	34 (10.8)	8 (23.5)	24 (8.6)		0 (0.0)	10 (7.2)	22 (14.2)		23 (10.7)	10 (10.1)
Rhabdoid
No	288 (91.4)	27 (79.4)	260 (92.9)	0.0422 *	19 (100.0)	130 (93.5)	138 (89.0)	0.305*	197 (91.6)	91 (91.9)	1.000**
Yes	27 (8.6)	7 (20.6)	20 (7.1)		0 (0.0)	9 (6.5)	17 (11.0)		18 (8.4)	8 (8.1)
Metastasis at diagnosis
M0	265 (84.1)	26 (76.5)	238 (85.0)	0.334**	16 (84.2)	124 (89.2)	122 (78.7)	0.120*	185 (86.0)	79 (79.8)	0.497**
M1	50 (15.9)	8 (23.5)	42 (15.0)		3 (15.8)	15 (10.8)	33 (21.3)		30 (14.0)	20 (20.2)
Tumor size
T1	170 (54.5)	14 (41.2)	155 (56.0)	0.0494 *	11 (57.9)	81 (58.7)	77 (50.3)	0.430*	114 (53.3)	55 (56.7)	0.456*
T2	44 (14.1)	3 (8.8)	41 (14.8)		2 (10.5)	23 (16.7)	19 (12.4)		32 (15.0)	12 (12.4)
T3	87 (27.9)	13 (38.2)	74 (26.7)		5 (26.3)	31 (22.5)	50 (32.7)		64 (29.9)	23 (23.7)
T4	11 (3.5)	4 (11.8)	7 (2.5)		1 (5.3)	3 (2.2)	7 (4.6)		4 (1.9)	7 (7.2)
Hyaluronan
Negative	158 (50.3)	13 (38.2)	146 (52.1)	0.253**	11 (57.9)	74 (53.2)	75 (48.4)	0.588**	95 (44.2)	65 (65.7)	0.00639 **
Positive	156 (49.7)	21 (61.8)	134 (47.9)		8 (42.1)	65 (46.8)	80 (51.6)		120 (55.8)	34 (34.3)
TILs
None	191 (61.0)	17 (50.0)	173 (62.2)	0.332**	13 (68.4)	99 (71.7)	77 (50.5)	0.024 *	130 (60.7)	60 (61.2)	0.515**
Low	54 (17.3)	10 (29.4)	44 (15.8)		4 (21.1)	21 (15.2)	29 (18.8)		41 (19.2)	13 (13.3)
Moderate	35 (11.2)	4 (11.8)	31 (11.2)		1 (5.3)	10 (7.2)	24 (15.6)		24 (11.2)	11 (11.2)
High	33 (10.5)	3 (8.8)	30 (10.8)		1 (5.3)	8 (5.8)	24 (15.6)		19 (8.9)	14 (14.3)
Type
Primary	315 (83.1)	34 (63.0)	280 (87.5)	<0.001 *	19 (95.0)	139 (85.8)	155 (81.2)	0.453*	215 (85.7)	99 (80.5)	0.497**
LVI	42 (11.1)	14 (25.9)	24 (7.5)		0 (0.0)	14 (8.6)	24 (12.6)		21 (8.4)	17 (13.8)
Metastasis	22 (5.8)	6 (11.1)	16 (5.0)		1 (5.0)	9 (5.6)	12 (6.3)		15 (6.0)	7 (5.7)

ccRCC, clear cell renal cell carcinoma; ISUP, International Society of Urological Pathology; LVI, lymphovascular invasion; TIL, tumor‐infiltrating lymphocytes.

p‐values with statistical significance (p < 0.05) are shown in bold.

^*Fisher's exact test.

^**Pearson's χ ² test.

No significant differences were found in the HAS2 staining patterns between HA‐positive and HA‐negative samples.

Association with clinicopathological variables

Higher CD44 protein expression was associated with higher tumor grade, advanced stage, increased tumor size, sarcomatoid and rhabdoid changes, and metastasis at diagnosis (Table 3, Figures 2 and 3). No correlation was observed between CD44 protein expression and histological subtype or TIL count. Furthermore, no statistically significant differences were found between the primary tumor and lymphovascular invasion or between the primary tumor and distant metastases.

Figure 3.

Collage of staining results in low‐grade clear cell renal cell carcinoma (RCC), RCC with sarcomatoid features, and RCC with rhabdoid features.

Chromophobe carcinomas exhibited higher HYAL1 expression levels than clear cell carcinomas (p = 0.044) (Table 3, Figure 1). An increase in HYAL1 expression was observed between ISUP grade 1 and 3 tumors (p = 0.0020); however, this trend was not observed between ISUP grades 1 and 2, or grades 1 and 4 (Figure 2). No statistically significant differences were observed between HYAL1 expression and stage, tumor size, TIL count, metastasis status at diagnosis, or between primary tumors and their lymphovascular invasion or distant metastasis.

Lower HYAL2 accumulation was associated with higher tumor grade, sarcomatoid change, higher stage, metastases at diagnosis, and a higher TIL count (Table 3). No association was observed between HYAL2 expression and rhabdoid change. Additionally, no significant differences were found between the histological subtypes, primary tumors, and their lymphovascular invasive foci or distant metastases.

A significant reduction in HAS1 staining was observed in sarcomatoid, rhabdoid, and ISUP grade 4 carcinomas (Table 4, Figure 2). Additionally, larger tumors tended to exhibit lower HAS1 levels. No correlation was found between HAS1 expression and stage, metastatic status, TIL count, or histological subtypes. The median cytoplasmic H‐score decreased significantly from 200 (IQR: 140–200) in primary tumors to 110 (IQR: 100–180) in lymphovascular invasive foci (p = 0.0029, paired Wilcoxon signed‐rank test). In contrast, the median cytoplasmic H‐score increased significantly from 140 (IQR: 70–200) in primary tumors to 180 (IQR: 102.5–232.5) in metastatic lesions (p = 0.022, paired Wilcoxon signed‐rank test).

HAS2 expression was positively correlated with increased tumor ISUP grade, sarcomatoid change, and TIL count. However, statistical significance for sarcomatoid change was lost when cores were aggregated. No significant correlations were found with stage, metastatic status, rhabdoid change, or histological subtype. No significant differences in HAS2 expression were observed when comparing primary tumors with lymphovascular invasive foci or distant metastases when analyzed separately (Table 4).

For HAS3, no statistically significant differences were observed with respect to ISUP grade, sarcomatoid change, rhabdoid change, tumor size, stage, histological subtype, lymphovascular invasion, or distant metastases (Table 4).

Prognostic significance

For Kaplan–Meier plotting, the samples were divided into three categories based on CD44 H‐scores: CD44‐negative, CD44‐low, and CD44‐high. Higher CD44 levels were associated with worse DSS and MFS (Figure 4). Multivariate analysis confirmed these results, revealing that CD44 positivity, along with higher ISUP grade and tumor stage, was a statistically significant independent predictor of poor survival (Table 5).

Figure 4.

Impact of CD44, HYAL2, and HAS1 protein levels on disease‐specific and metastasis‐free survival in RCCs.

Table 5.

Univariate and multivariate analysis of disease‐specific survival (DSS) in RCC patients

Variables	Categories	HR (95% CI)	p
Univariate analysis
Age	≥65/<65	0.94 (0.63–1.41)	0.76
Sex	Male/Female	1.68 (1.11–2.54)	0.015
WHO/ISUP grade	3–4/1–2	3.86 (2.55–5.85)	1.94 × 10 −10
Stage	3–4/1–2	5.7884 (3.75–8.94)	2.25 × 10 −15
HYAL2 protein expression	High/Low	0.45 (0.30–0.67)	0.00011
CD44 protein expression	Positive/Negative	2.79 (1.85–4.21)	9.51 × 10 −7
HAS1 protein expression	High/Low	0.54 (0.32–0.93)	0.025
Hyaluronan staining	Positive/Negative	3.89 (2.45–6.19)	9.18 × 10 −9
Multivariate analysis
Age	≥65/<65	1.01 (0.85–2.03)	0.49
Sex	Male/Female	1.31 (1.21–2.03)	0.22
WHO/ISUP grade	3–4/1–2	2.51 (1.62–3.88)	3.48 × 10 −5
Stage	3–4/1–2	4.06 (2.57–6.41)	1.75 × 10 −9
HYAL2 protein expression	High/Low	0.64 0.42–0.97)	0.037
Multivariate analysis
Age	≥65/<65	1.08 (0.72–1.64)	0.70
Sex	Male/Female	1.08 (0.69–1.69)	0.73
WHO/ISUP grade	3–4/1–2	2.35 (1.51–3.64)	0.00013
Stage	3–4/1–2	4.12 (2.639–6.46)	7.13 × 10 −10
CD44 protein expression	Positive/Negative	1.73 (1.12–2.66)	0.012

CI, confidence interval; HR, hazard ratio; ISUP, International Society of Urological Pathology; WHO, World Health Organization.

p‐values with statistical significance (p < 0.05) are shown in bold.

The prognostic value of HA was assessed using both the log‐rank test (p < 0.0001) and the Cox regression model. Positive hyaluronan content was associated with significantly worse prognosis in both univariate (Table 5) and multivariate models, after adjustment for grade, stage, age, and sex [HR 1.86 (95% CI 1.23–3.06), p = 0.016].

For survival analysis of HYAL2, samples were stratified into low and high protein expression categories. Low HYAL2 expression was associated with DSS and MFS (Figure 4). Among low‐grade (ISUP1–2) carcinomas, low HYAL2 expression also predicted poorer survival (log‐rank test, p = 0.016). Moreover, low HYAL2 expression was identified as an independent predictor of poor survival in multivariate analysis (Table 5).

Lower HAS1 protein expression was also correlated with worse DSS and MFS (p = 0.023 and p = 0.028) (Figure 2). This association remained significant in univariate analysis; however, statistical significance was lost in the multivariate model (Table 5).

HYAL1, HAS2, and HAS3 protein levels did not predict DSS or MFS.

Transcriptomic analysis and correlation with protein expression

To investigate transcriptomic correlations with protein expression, we performed DEA on RNA sequencing data. The analysis showed a log₂ fold change (log₂FC) of 0.76 in CD44 mRNA expression between the CD44‐positive and CD44‐negative groups (p = 3.58 × 10⁻⁸), indicating higher CD44 expression in the positive group. CD44 expression was also compared between hyaluronan‐positive and hyaluronan‐negative samples, showing a log₂FC of 0.59 (p = 9.64 × 10⁻⁵), again indicating higher expression in HA‐positive samples.

The DEA between HYAL2‐positive and HYAL2‐negative samples revealed a log₂FC of 0.45 (p = 0.0031), reflecting increased HYAL2 mRNA expression in the HYAL2‐positive group. No statistically significant differences in HYAL2 mRNA levels were observed with respect to hyaluronan status. Additionally, no significant differences in mRNA expression of HYAL1 or HAS1–3 were found in relation to either protein staining or hyaluronan content.

Figure 5 shows boxplots of normalized gene expression values of CD44, HYAL1‐2, and HAS1‐3 grouped by protein expression level in our dataset (Figure 5A–C) and by predicted HA status in the TCGA KIRC cohort (Figure 5D–I). The plots also display the significance of differences between groups, as determined by the Wilcoxon rank‐sum test. CD44 mRNA expression was significantly higher in the CD44‐positive cohort compared with the CD44‐negative cohort (p < 5.13 × 10⁻¹⁶) (Figure 5B). Higher CD44 expression was also observed in HA‐positive samples. The HYAL2‐high protein expression group exhibited elevated HYAL2 mRNA expression compared to the HYAL2‐low group (Figure 5C). No statistically significant differences were observed between the mRNA and protein expression levels of HYAL1, HAS1, HAS2, and HAS3.

Figure 5.

Comparison of RNA and protein expression levels of CD44 and hyaluronan pathway genes using our data from 96 ccRCC samples (A–C) and TCGA data (D–I). Box plots depict RNA expression levels in relation to hyaluronan (HA) status and protein expression. (A) CD44 RNA expression in HA‐negative and HA‐positive ccRCC samples. (B) CD44 RNA expression stratified according to protein expression levels. (C) HYAL2 RNA expression in relation to protein expression levels. (D–I) TCGA data showing CD44 (D), HYAL2 (E), HAS2 (F), HYAL1 (G), HAS1 (H), and HAS3 (I) RNA expression in relation to predicted hyaluronan (HA) status. For TCGA data, the HA‐negative group includes 221 samples, and the HA‐positive group includes 151 samples. Statistical significance is indicated as follows: *p < 0.05, **p < 0.01, ***p < 0.001.

Discussion

In our previous study, we showed that a subset of RCCs accumulated cellular hyaluronan; however, prior studies have not clarified the mechanisms or the relative contributions of hyaluronan‐synthesizing and ‐degrading enzymes underlying intratumoral hyaluronan accumulation [26]. To the best of our knowledge, this is the first study to demonstrate that hyaluronan accumulation in RCCs correlates with increased expression of its receptor CD44 and reduced protein expression of its degrading enzymes, HYAL1 and HYAL2.

Furthermore, high CD44 protein expression, along with low HYAL2 and HAS1 expression levels, predicted poor outcomes. CD44 overexpression was associated with adverse clinicopathological variables, including higher tumor grade and sarcomatoid change, while similar trends were observed for higher HAS2 protein expression and lower HYAL2 protein expression. These findings suggest that HA metabolism plays an important role in RCC progression and prognosis.

CD44–HA interaction in RCC

This study supports the negative prognostic impact of CD44 in RCC and its association with higher tumor grade, stage, metastasis, and poor prognosis [24, 25, 38]. We also demonstrate a simultaneous increase in RNA and protein levels, suggesting transcriptional upregulation of CD44. Although previous studies have shown that CD44 binds to HA in non‐neoplastic tubular epithelial cells, and that CD44 and HA are co‐expressed in human malignancies such as non‐small cell lung carcinoma, colorectal carcinoma, and endometrial carcinoma, this phenomenon has not been previously observed in RCCs [16, 39, 40, 41].

HA–CD44 interaction has been linked to multiple intracellular pathways, including the anti‐apoptotic PI3K/AKT and EMT‐promoting Wnt/β‐catenin pathways [42, 43]. In our previous study, we found that the expression of EMT‐associated genes is enhanced in HA‐positive tumors [27]. Since CD44 is known to be upregulated in mesenchymal‐like cells, and sarcomatoid/rhabdoid RCCs have been shown to significantly upregulate EMT‐related gene sets, it is not surprising that CD44, HA, and unfavorable histological features co‐occur [44].

HAS1–3 expression in RCC

The precise role of each HAS isoform in cancer and other physiological processes remains largely unknown, as there are no isoform‐specific inhibitors of HAS activity [45]. Increased HAS1 and HAS2 protein expression has been associated with HA content in breast carcinoma cells, and the expression of all HAS isoforms in stromal cells correlates with stromal HA content [46]. HAS1 RNA overexpression has been reported in colon, ccRCC, and bladder cancer [38, 47, 48]. Nykopp et al investigated HAS1–3 mRNA levels in ovarian and endometrial cancers but found no correlation with corresponding protein levels [49, 50]. Although immunoreactivity for all HAS isoforms was stronger in endometrial cancer than in normal endometrium, only HAS2 protein levels correlated with HA content. Overall, the relationship between HAS1–3 protein and HA levels appears to vary across malignancies [51].

Compared to normal renal tissue, RCC reportedly shows elevated RNA levels of all HAS isoforms, as well as increased HAS3 protein expression [52]. In ccRCC and its associated venous thrombi, HAS1 RNA upregulation was observed alongside CD44 and has been suggested to play a role in tumor invasion [24]. In contrast to previous studies, we found that HAS1 protein is downregulated in high‐grade tumors and intravenous foci, and that reduced HAS1 expression is a marker of poor prognosis in RCC. Despite this, regardless of its expression level, HAS1 is unlikely to directly contribute to HA accumulation in RCC, as it has the lowest enzymatic activity among the HAS isoforms and shows no positive correlation with tumor HA content [13, 53]. Similarly, HAS3 levels did not correlate with tumor HA content, were consistently low throughout the tumors, and appeared to have no role in tumor progression.

HAS2 is considered the most abundant and functionally significant hyaluronan synthase in cancer [53]. In our study, most RCCs, particularly those of lower grades, exhibited low HAS2 protein levels, which likely contributes to the relative scarcity of HA in these tumors. In contrast, higher HAS2 protein expression in more aggressive tumors was consistent with the observed pattern of HA accumulation. Accordingly, HAS2 was the only isoform that correlated with higher tumor grade. It is possible that HAS2 dominance over the other isoforms, especially HAS1, enhances overall enzymatic activity. Mechanistically, this could be explained by the observation that HAS enzymes can form functional complexes with one another [54].

While our study provides insight into HAS1–3 expression, it may have limited capacity to accurately assess transcript–protein correlations due to the use of FFPE‐derived RNA, which is prone to degradation and reduced sensitivity, especially for low‐abundance transcripts such as those encoding HAS enzymes.

Biologically, the lack of correlation between HAS protein levels and hyaluronan content is not unexpected, and several factors may contribute to this discrepancy. HAS enzymes are subject to complex post‐transcriptional and post‐translational regulation. Various growth factors and cytokines influence transcription, while post‐translational modifications, such as O‐GlcNAcylation, among others, modulate enzymatic activity and protein stability [55]. In addition, the enzymatic activity of HAS proteins depends on their recycling dynamics between cytoplasmic compartments and the plasma membrane, the latter being the only site where the enzymes are active [56]. Moreover, HAS activity and hyaluronan synthesis are strongly influenced by the cellular availability of UDP‐sugars [57]. This is exemplified by human breast cancer biopsies, which show a strong correlation between HA content and UDP‐sugar levels, but not with HAS family RNA expression [58].

HYAL1 and HYAL2 in RCC

In other cancers, such as endometrial carcinomas, reduced HYAL1 protein expression is a poor prognostic factor, and decreased HYAL1 RNA expression correlates with elevated HA content [49, 59]. Similarly, low HYAL1 RNA expression correlates with increased HA content in serous ovarian tumors, and weak HYAL1 protein expression is associated with poor survival in ductal pancreatic carcinoma [18, 49]. However, contradictory findings have been reported. For example, Kohi et al showed that HYAL1 mRNA and protein levels are higher in the generally HA‐rich ductal pancreatic cancer than in normal pancreatic tissue [60]. There are only a few studies on HYAL2 expression; however, in triple‐negative breast cancer, increased HYAL2 protein levels have been associated with more advanced stages and poorer survival [61]. Overall, studies examining the correlation between HYAL2 mRNA/protein levels and HA accumulation remain insufficiently characterized.

In ccRCC, one study reported reduced HYAL1 RNA expression compared to normal renal tissue, with no significant change in HYAL2 levels [38]. In contrast, another study demonstrated lower expression levels of both HYAL1 and HYAL2 in ccRCC cell cultures compared to normal renal cells [62]. In the present RCC cohort, both HYAL1 and HYAL2 expression levels were lower in HA‐positive tumors. A decrease in HYAL2 was observed at both the RNA and protein levels, consistent with transcriptional regulation. Our results suggest that HYAL2 may play a more significant role than HYAL1 in RCC, as indicated by its stronger association with tumor size, stage, metastasis at diagnosis, sarcomatoid change, HA content, TIL score, and survival.

Intriguingly, HYAL1 and HYAL2 are located at chromosomal region 3p21.31, which also harbors the VHL tumor‐suppressor gene and is frequently lost in ccRCC and sarcomatoid subtypes [3]. However, our findings indicate that most RCCs express relatively high levels of HYAL2, and a substantial proportion of tumors also express HYAL1. Notably, loss of these proteins is more common in higher‐grade tumors and sarcomatoid/rhabdoid subtypes, suggesting the involvement of additional genetic alterations affecting HYAL1 and HYAL2. The presence of HYAL1 and HYAL2 in lower‐grade tumors implies that not all genes in the 3p region are necessarily lost.

HYAL2 likely contributes to the initial stage of HA degradation by facilitating endocytosis, which is subsequently followed by HYAL1, whose activity is optimal at lysosomal pH [63]. Alterations in their expression levels could lead to the production of HA fragments of varying sizes. These fragments, particularly if released into the extracellular space, may act as signaling molecules that induce cancer‐promoting inflammatory responses [51].

In addition to generating these signals, HA removal may facilitate invasive migration by exposing integrins and their binding sites within the ECM. However, HYAL1 and HYAL2 protein expression levels did not show a consistent directional relationship across all samples, nor did their expression consistently correlate with HA content. This suggests that mechanisms beyond the enzymatic activity of HYAL1 and HYAL2 also regulate HA accumulation. Indeed, HYAL2 signaling has been observed in the nucleus, where it influences the alternative splicing of CD44 pre‐mRNA and thereby regulates the expression of CD44 isoforms [64].

Taken together, HA‐binding (CD44) and HA‐degrading (HYAL) proteins are closely associated with inflammation‐like processes that enable cancer growth, dissemination, and immune evasion. There is a fine balance between these factors, which may either promote or suppress malignant progression depending on the cancer type and its stage of development. Nonetheless, it is evident that both HA synthesis and degradation are actively occurring in most carcinomas, including RCC.

Transcriptomic analysis and protein correlation

Based on sequencing studies, CD44 and HYAL2 mRNA levels correlated with their respective protein levels. Only CD44, HYAL1, and HYAL2 showed read counts sufficient for statistical analysis. HAS1–3 expression levels were generally low in our dataset, resulting in the exclusion of multiple samples during quality control.

Using data from both our study and TCGA sequencing datasets, we found a consistent correlation between mRNA and protein expression levels in HA‐positive versus HA‐negative groups for all proteins analyzed, except for HAS1, which showed an inverse relationship between mRNA and protein expression. HAS1–3 mRNA levels did not consistently correlate with protein levels, suggesting the involvement of post‐transcriptional, post‐translational, and possibly epigenetic regulatory mechanisms [65, 66]. Furthermore, results from the TCGA in silico analysis should be interpreted with caution, as the predicted HA‐positive and HA‐negative groups have not yet been experimentally validated.

TILs and HA metabolism in RCC

Reduced HYAL2 and increased HAS2 expression were associated with a higher number of TILs. High‐molecular‐weight hyaluronan (HMW‐HA) has been shown to exert anti‐inflammatory effects, primarily by forming a barrier‐like, viscoelastic scaffold [67, 68, 69]. HAS2 synthesizes HMW‐HA, while HYAL2 degrades HA into low‐molecular‐weight fragments (LMW‐HA). Thus, a shift from LMW‐HA to HMW‐HA may occur, contributing to the effects discussed above.

However, while the pericellular HMW‐HA coat may shield tumor cells from immune attack, it can also facilitate immune cell adherence, allowing immune cells to persist within the tumor microenvironment. Moreover, the anti‐inflammatory effects of HMW‐HA may be less pronounced in RCC, which lacks the desmoplastic stromal reaction characteristic of other carcinomas. Additionally, the increased TIL presence may more accurately reflect the poor differentiation and neoantigen load of HYAL2‐low/HAS2‐high tumors, potentially overriding the immunosuppressive effects of HMW‐HA and rendering these tumors more responsive to immunomodulatory therapy.

HA metabolism in RCC subtypes

In this study, all histological subtypes were included in the statistical analyses to address the limited existing data on protein expression across RCC variants and to build on our prior work, which also encompassed all histological subtypes [26]. However, when the analysis was restricted to ccRCCs alone, the associations of CD44, HYAL2, and HAS1 with poor prognosis remained, supporting the robustness of these findings.

Curiously, relatively higher levels of HYAL1 protein expression were detected in tumors with eosinophilic cytoplasm compared to non‐eosinophilic tumors, regardless of histological subtype. Among the histological subtypes, only chromophobe RCCs demonstrated statistically significant higher HYAL1 protein levels compared to clear cell carcinomas. While this difference was not statistically significant in papillary RCCs overall, high‐grade papillary RCCs with eosinophilic features (previously classified as type 2 according to the 2016 WHO Classification of Tumours, 4th edition) showed a similar trend toward elevated HYAL1 expression (Figure 1).

These findings may be explained by the distinct molecular background of eosinophilic tumors, which often exhibit widespread gene and chromosome loss rather than 3p deletion [70]. Eosinophilic changes are attributable to a high mitochondrial content in tumor cells, which frequently harbor mutations affecting the mTOR complex 1 pathway [71]. However, the number of cases representing variant histological subtypes was relatively limited, and the resulting statistical power may have been insufficient to detect meaningful differences in staining profiles, although some non‐significant trends were observed.

Clinical significance

Given HA's role in RCC progression and immune modulation, targeting this pathway may have therapeutic potential. CD44 expression has been linked to epitope masking in breast and gastric cancers; thus, removing this barrier can enhance cancer cell sensitivity to anti‐cancer treatments [72, 73, 74]. Cancers expressing CD44 could also benefit from the use of drug‐carrying HA‐coated nanoparticles, with CD44 offering a specific intracellular entry point [75].

Targeting enzymes involved in HA metabolism, such as the HAS family, also holds promise for inhibiting cancer development and spread [45]. Knocking down HAS2 has been shown to increase tumor cell line sensitivity to radiotherapy. Additionally, the combination of 4‐methylumbelliferone, a competitive inhibitor of HA synthesis, and sorafenib inhibited HAS3 and HA synthesis, resulting in reduced tumor growth [76, 77]. Although there is some preliminary evidence that inhibition of hyaluronidase activity can suppress cancer progression, further studies are needed to clarify the potential of hyaluronidases as anti‐cancer targets [78].

In contrast to inhibiting endogenous HA metabolism, another promising approach involves enhancing HA degradation using exogenous hyaluronidases. This strategy may reduce immune suppression and remove physical barriers to drug delivery. The most extensively studied exogenous hyaluronidase, PEGylated recombinant human hyaluronidase (PEGPH20), has been shown to enhance tumor regression when combined with anti‐PD‐1 antibodies and to increase progression‐free survival in patients with pancreatic cancer [22, 79].

Strengths and limitations of the study

The strengths of this study include the large sample size and the availability of comprehensive clinical and follow‐up information. Additionally, the use of TMA material enabled spatial evaluation of HA accumulation and protein expression in sequential sections.

However, this study also has limitations. Although TMA allows spatial evaluation of protein expression relative to HA, consistent with the study's aims, it may overlook regional heterogeneity and introduce sampling errors. The representation of variant histological subtypes was relatively limited, and the statistical power was insufficient to detect significant differences in staining profiles across these subtypes, despite the observation of potential trends. Additionally, RNA analyses using FFPE material are often constrained by RNA degradation and crosslinking, which can reduce sensitivity, particularly for low‐expression genes.

Conclusions

This is the first study to simultaneously examine HA‐degrading and HA‐synthesizing enzymes in RCCs and to correlate their expression with HA content within the same tumors. Our findings suggest that HA accumulation in RCC is driven by reduced degradation via HYAL1 and HYAL2, increased synthesis by HAS2, or a combination of both. This study also identifies HYAL2 and HAS1 as novel biomarkers for RCC prognostication.

While CD44 and HYAL2 mRNA levels correlate with their respective protein levels, further research is needed to determine the extent to which mRNA levels influence the production of HYAL1–2 and HAS1–3 proteins. Studying the expression levels of HAS1–3 and HYAL1–2 in normal renal tissue may also provide valuable insights into baseline HA metabolism and help distinguish tumor‐specific alterations. Additionally, investigating potential differences among CD44 isoforms in relation to HA accumulation and HA‐mediated cancer progression in RCC would be valuable.

Future studies should also consider the functional consequences of HA fragment size and the subcellular localization of HAS and HYAL enzymes, both of which may critically affect enzymatic activity, signaling, and interactions within the tumor microenvironment. A more detailed characterization of immune cell populations and their correlation with HA could further clarify the immunological landscape of RCC. These insights offer a foundation for future work to unravel the therapeutic and diagnostic potential of HA metabolism in RCC.

Author contributions statement

Conceptualization: OJ, SP‐S, TKN and RS. Data curation: OJ, SR, SP‐S and RS. Formal analysis: OJ. Funding acquisition: OJ and TKN. Investigation: OJ, SR, TKN and RS. Methodology: OJ, TR, SR, SP‐S and RS. Project administration: OJ, TKN and RS. Resources: OJ, SR, SP‐S and TKN. Software: OJ and TR. Supervision: SP‐S, TKN and RS. Validation: OJ, TR, SR, SP‐S, TKN and RS. Visualization: OJ. Writing – original draft: OJ and SR. Writing – review and editing: OJ, TR, SR, SP‐S, TKN and RS. TKN and RS contributed equally to this study. All the authors have read and agreed to the published version of the manuscript.

Data availability statement

Restrictions apply to the availability of these data. Data were obtained from the Biobank of Eastern Finland and are available from the authors upon reasonable request with the permission of the Biobank of Eastern Finland. Please contact info@ita-suomenbiopankki.fi for access inquiries.