Fibroblast Growth Factor 23 is a Potential Prognostic Biomarker in Uterine Sarcoma.

Uterine sarcoma (US) is one of the most aggressive and lethal gynecologic malignancies, comprising approximately 1% of all female reproductive tract malignancies and 3-7% of all uterine cancers. 1 Presently, the prognosis of US patients is extremely poor due to their delayed diagnosis attributed to non-specific early symptoms mimicking uterine myoma or adenomyosis. More than 50% of all US cases are diagnosed by chance. 2 Within US subtypes, the 5-year overall survival (OS) rates for low-grade endometrial stromal sarcoma (LG-ESS), high-grade endometrial stromal sarcoma (HG-ESS), and undifferentiated uterine sarcoma (UUS) were 94%, 53%, and 25%, respectively. 3 Residual tumors frequently occurred. Despite adjuvant chemotherapy (CT) or radiotherapy (RT) for most patients, the prognosis remains dismal even in stage I. 4 Therefore, novel potential cost-effective biomarkers and powerful treatments are needed to improve diagnosis and prognosis in cases of US. Nevertheless, the discovery of new biomarkers and specific targeted therapies is still challenging owing to the low incidence of rare uterine tumor. 5 Human fibroblast growth factor 23 (FGF23) is a member of the fibroblast growth factor (FGF) family. 6 It is secreted by bone cells and osteoblasts and is responsible for regulating phosphorus levels in a hormone-like manner. Research indicated a strong correlation between FGF23 and gynecological malignancies. Studies have revealed that FGF23 is closely related to gynecological tumors. Serum or plasma FGF23 levels were notably increased in patients with advanced ovarian cancer as compared to those with early-stage ovarian cancer, non-malignant disease, or healthy controls. Additionally, a positive correlation was observed between FGF23 and the stage of the disease. 7 Moreover, a distinct single-nucleotide polymorphism (SNP) of FGF23 exhibited a significant correlation with treatment response following platinum-based CT and surgery as well as the OS of patients with ovarian cancer. 8 Nonetheless, the expression of FGF23 in US remains unclear. In this study, to clarify the role of FGF23 in US, we explored the relationship between FGF23 expression and the prognosis of US.

According to the guidelines of the FIGO, the clinical data collected from the patients’ medical records were analyzed, including age at diagnosis, histologic types, clinical and pathologic staging, and adjuvant CT or radiation treatment (RT). These baseline characteristics related to two cohorts, which were from the TCGA database, including 57 UCS patients, and a Suining cohort including 44 US cases from the Department of Pathology of Suining Central Hospital. Supplementary Table S1 shows the baseline characteristics of 57 patients in the TCGA cohort. All patients involved in the study were female and over age 50. Most of them (n = 56, 98.25%) were postmenopausal. Regarding race, most of the patients (n = 44, 77.19%) were White. Peritoneal wash included positive (n = 12, 21.05%), negative (n = 25, 43.86%), and unknown (n = 20, 35.09%). The FIGO stage included 21 (36.84%) I, 6 (10.53%) II, 20 (35.09%) III, and 10 (17.54%) IV. Surgical approaches included minimally invasive (n = 30, 52.63%), open (n = 20, 35.09%), and unknown (n = 7, 12.28%). Person neoplasm cancer status included 9 (15.79%) with tumors, 11 (19.30%) tumor free, and 37 (64.91%) unknown. In terms of adjuvant treatment, 11 patients received RT (19.30%), 10 patients (17.54%) never used tamoxifen, and 47 patients unknown (82.46%). From an initial cohort of 56 patients, 44 cases were identified with a definite pathological subtype of US. The baseline characteristics of 44 US patients from the clinical (Suining) cohort are presented in Table 1 . Baseline Characteristics of 44 US Patients from the Suining Cohort. TH, total hysterectomy; TH + BSO, total hysterectomy + bilateral salpingo-oophorectomy; PLA, pelvic lymphadenectomy; US, uterine sarcoma; UCS, uterine carcinosarcoma; LMS, leiomyosarcoma; ESS, endometrial stromal sarcoma; AS, adenosarcoma. The mean age of all participants was 51.5 ± 10 years (range, 35 to 73 years). Regarding menopausal status, 20 (45.45%) patients were postmenopausal, and 24 (54.55%) patients were premenopausal. The symptoms included colporrhagia (n = 20, 45.45%), stomachache/bloating (n = 7, 15.91%), abdominal mass (n = 11, 25.00%), and asymptomatic (n = 6, 13.64%). In terms of CA125 levels, 14 (31.82%) patients had CA125 levels >35 U/ml, 18 (40.91%) patients had CA125 levels <35 U/ml, and 12 (27.27%) patients were unknown. The FIGO stage included 27 (61.36%) I, 6 (13.64%) II, 5 (11.36%) III, and 6 (13.64%) IV. Surgical approaches included total hysterectomy (TH, n = 1, 2.27%), TH and bilateral salpingo-oophorectomy (TH + BSO, n = 24, 54.55%), TH and BSO and pelvic lymphadenectomy (TH + BSO + PLA, n = 11, 25.00%), TH and BSO and PLA and omentectomy (n = 5, 11.36%), and TH and BSO and PLA and omentectomy and appendectomy (n = 3, 6.82%). A total of 43.18% (n = 19) of all patients underwent lymphadenectomy, and 6.82% (n = 3) experienced lymphatic metastasis. In addition, for adjuvant treatment, half of the patients (n = 22, 50.0%) received CT with doxorubicin and ifosfamide as the primary protocol, while the rest did not. Eleven patients (25.0%) underwent RT, while 33 patients (75.0%) did not. Forty-four patients were categorized according to pathological type into groups of AS (n = 1, 2.27%), UCS (n = 13, 29.55%), ESS (n = 14, 31.82%) (including 12 LG-ESS and 2 HG-ESS), and LMS (n = 16, 36.36%). A cohort of 57 UCS patients and corresponding necessary clinical data were downloaded from the TCGA database. We analyzed FGF23 expression levels between tumor tissues and normal tissues in 57 UCS samples and 78 normal tissues from US patients which download from TCGA and GTEx database, respectively, were analyzed and found that there was no significant difference in FGF23 gene expression between tumor and normal tissues ( P  > 0.05; Supplementary Figure S1 ). Immunohistochemistry was conducted to examine the expression of FGF23 in 44 full-face tissue sections obtained from US patients, including 32 paired samples of uterine normal smooth muscle (UNSM) and six paired samples of uterine leiomyoma (ULM). Positive expression of FGF23 protein was detected in the cell membrane and cytoplasm, appearing as light yellow, brownish yellow, or dark brown particles. Notably, only one ESS sample displayed sporadic and subdued heterogeneous staining in 26-50% of cells, which was classified as weakly positive. Additionally, all specimens from US patients showed varying levels of positivity, ranging from moderate to strong. But we found that there was no significant difference in FGF23 gene expression between tumor and normal tissues based on IHC or RNAseq from TCGA cohort ( P  > 0.05; Supplementary Figure S1 ). As shown in Figure 1 , among six paired samples, the expression level of FGF23 exhibited a significant increase in the US group compared to both the UNSM and ULM groups ( P  = 0.005 and P  = 0.006, respectively) and in the ULM group compared to the UNSM group ( P  = 0.02). Furthermore, among 32 paired samples, similar results were obtained in the US group compared to the UNSM group ( P  < 0.001). (A) IHC indicated the expression levels of FGF23 in the UNSM, ULM, and US groups. (B) The box plot shows FGF23 expression in US, ULM, and UNSM among six paired samples. (C) The box plot shows FGF23 expression in US and UNSM among 32 paired samples. US, uterine sarcoma; IHC, immunohistochemistry; UNSM, uterine normal smooth muscle; ULM, uterine leiomyoma. As shown in Figure 2 , immunohistochemistry and statistical analysis revealed a significant difference ( P  = 0.01) in FGF23 expression levels among AS, ESS, LMS, and UCS. The expression of FGF23 in AS showed moderate positivity and was significantly lower than that in ESS, LMS, and UCS, although there was only a single patient. Furthermore, the expression of FGF23 in ESS was substantially lower than that observed in LMS and UCS. Obviously, the findings indicated that UCS patients exhibited the highest levels of FGF23 expression when compared with other types ( Figure 2b ). (A) IHC assay indicated FGF23 expression in different uterine sarcoma subtypes. (B) The box plot and scatter plot shows the differential expression of FGF23 in AS, ESS, LMS, and UCS. FGF23 intensity: 1+ = weak, 2+ = moderate, 3+ = strong, FGF23 expression: 1+ = 1-25%, 2+ = 25-50%, 3+ = 50-75%, 4+ = ＞75%. IHC, immunohistochemistry; UCS, uterine carcinosarcoma; LMS, leiomyosarcoma; ESS, endometrial stromal sarcoma; AS, adenosarcoma. Based on the mean value of FGF23 expression, the patients were categorized into high-FGF23 and low-FGF23 expression groups, and we performed Kaplan-Meier analysis in three cohorts from TCGA, GEO, and clinical cohorts. Kaplan-Meier analysis indicated that 41 UCS patients from the TCGA database with high-FGF23 expression levels were unrelated to poor OS ( P  = 0.22, HR: 2.01, 95% CI: 0.66-6.14; Figure 3a ), and the same results were obtained in 50 UUS patients from the GSE119043 microarray ( P  = 0.065, HR: 1.87, 95% CI: 0.96-3.64; Figure 3b ). Then, the values of OS and PFS in Suining cohort including 44 patients suggested similar results as shown in Figure 3c ( P  = 0.38, HR: 1.53, 95% CI: 0.59-3.99 and P  = 0.66, HR: 1.26, 95% CI: 0.45-3.51, respectively). The median OS time of patients with low-FGF23 expression was not reached (NR), indicating a longer but not statistically significant survival compared to the median OS (48.7 months) in the high expression group. The median PFS times of patients with high- or low-FGF23 expression were NR. (A) K-M analysis suggested overall survival of FGF23 high or low expression in 41 UCS patients from the TCGA cohort. (B) K-M analysis revealed the overall survival by FGF23 high or low expression in 50 UUS patients from the GEO database. (C) K-M analysis indicated no significant correlation between FGF23 expression levels and OS or PFS of US patients from the Suining cohort. (D) K-M analysis showed the differential OS and PFS between LG-ESS and other types of US patients from the Suining cohort. US, uterine sarcoma; TCGA, The Cancer Genome Atlas; FGF23, fibroblast growth factor 23; GEO, Gene Expression Omnibus; UCS, uterine carcinosarcoma; OS, overall survival; PFS, progression-free survival; LG-ESS, low-grade endometrial stromal sarcoma; UUS, undifferentiated uterine sarcoma. However, we analyzed the various pathological subtypes of 44 US patients from Suining cohort and found that LG-ESS showed significantly better survival than other subtypes in terms of OS and PFS ( P  = 0.006, HR: 0.26, 95% CI: 0.10-0.68 and P  = 0.003, HR: 0.21, 95% CI: 0.07-0.59, respectively; Figure 3d ). Univariate and multivariate Cox proportional hazard regression analyses were carried out with UCS patients from the TCGA database. The univariate analysis indicated that tumor stage (HR = 1.39; P  = 0.029; 95% CI: 1.03-1.87), neoplasm cancer status (HR = 0.16; P  = 0.031; 95% CI: 0.03-0.85), peritoneal wash (HR = 0.19; P  = 0.003; 95% CI: 0.07-0.57), and adjuvant radiotherapy (HR = 0.46; P  = 0.006; 95% CI: 0.26-0.79) were significantly associated with OS in UCS patients, with no significant differences in other clinical indicators such as age, menopausal status, race, BMI, and surgical approaches in relation to OS ( P  > 0.05). The findings of the multivariate analysis were similar with the univariate analysis. Nevertheless, no substantial correlation between FGF23 expression and OS were observed (HR = 1.06; P  = 0.532; 95% CI: 0.47-2.41). The findings are presented in Supplementary Table S2 . Furthermore, we conducted univariate and multivariate Cox proportional hazard regression analyses of OS and PFS in US patients from the clinical cohort. As shown in Supplementary Tables S3 and S4 demonstrated that tumor stage (HR = 3.62; P  = 0.007; 95% CI: 1.42-9.22), CA125 (HR = 3.23; P  = 0.001; 95% CI: 1.58-6.62), tumor type (HR = 5.71; P  = 0.021; 95% CI: 1.30-25.13), and tumor recurrence (HR = 8.47; P  < 0.001; 95% CI: 3.08-23.27) were significantly associated with OS and PFS in US patients. Multivariate analysis showed that only CA125 and tumor recurrence remained significant predictors of OS, while only tumor relapse had prognostic significance for PFS ( P  < 0.05). Nevertheless, the results indicated an absent relationship between FGF23 expression and OS (HR = 1.58, P  = 0.383, 95% CI: 0.59-3.99) or PFS (HR = 1.07, P  = 0.59, 95% CI: 0.88-1.29) in US patients consistent with prior study on the TCGA cohort. Moreover, additional clinical variables, such as age, menopausal status, tumor size, lymphadenectomy, lymphatic metastasis, adjuvant CT, and adjuvant radiotherapy, did not influence OS and PFS ( P  > 0.05).

FGF23 is significantly upregulated in US and serves as a potentially novel diagnostic and prognostic biomarker.

US is one of the most aggressive soft tissue sarcomas in adult women, with an increasing incidence and high mortality rate of up to 30% due to its highly aggressive nature. 13 Typically, this tumor is diagnosed in older women among whom the median age ranges from 62 to 67 years. Nonetheless, our study revealed a younger patient population, with a median age of 51.5 years and with a higher proportion of early-stage disease, which was observed in 75% (n = 33) of patients versus 58-66% in the published literature. 14 Our results revealed that tumor stage, CA125, tumor type, tumor recurrence, neoplasm cancer status, and peritoneal wash were significantly associated with OS in US patients. There were no significantly relations among OS and other clinical parameters, such as age, menopausal status, race, BMI, surgical approaches, tumor size, lymphadenectomy, lymphatic metastasis, and adjuvant radiotherapy. Currently, the cornerstone of US treatment is hysterectomy and BSO, bilateral salpingo-oophorectomy. However, no standardized procedure for U.S. standard therapy is currently established, with varying perspectives on ovarian removal and pelvic lymphadenectomy. Machida's study revealed that lymph node positivity was a significant independent risk factor. 15 Nonetheless, Nasioudis et al separately analyzed patient data per subtype. In 3517 LMS patients, survival rate was superior in the group without lymphadenectomy, as compared to another group. In 46 HG-ESS patients, the lymphadenectomy group showed better survival. In 1168 LG-ESS and 1321 UAS patients, there were no significant differences in survival rates between two groups. 16 We found that surgical approaches and lymphadenectomy were not associated with OS, which was distinct from previous studies. Several prior studies justified that radiation therapy in patients with US can improve survival, 17 – 19 similar results were found in TCGA cohort. However, the findings indicated that adjuvant radiotherapy did not improve the survival of US patients in Suining cohort compared with previous studies. 20 , 21 The role of radiation therapy in patients with US still needs further investigation. 22 In addition, our research showed the opposite outcome compared with a prior study, which found that poor survival was associated with old age and Black race. 17 Previous research confirmed that histological type and tumor size were independent prognostic factors for both OS and PFS, and tumor stage was significantly associated with PFS, 23 our study exhibited similar results except for tumor size. Currently, there is a lack of research on the expression of FGF23 in various subtypes of US, and the relationship between FGF23 expression and the prognosis of US remains unknown. In this study, we investigated the relationship between FGF23 expression and the clinical characteristics and survival of US patients. Consistently, tumor stage, neoplasm cancer status, peritoneal wash, CA125, and tumor recurrence were independent prognostic factors in US survival. LMS was the most prevalent histology in previous reports and was also observed in our investigation. At the time of analysis, 18 patients had died, and the 3-year OS rate was 45.45% (n = 20). Furthermore, our findings demonstrated that LG-ESS had longer PFS and OS than other pathological subtypes in the US, which is consistent with other published literature. 17 , 24 The principal surgical intervention, performed in 97.73% (n = 43) of patients, was hysterectomy with bilateral salpingo-oophorectomy. The rarity and heterogeneity of US presents tremendous challenges in identifying appropriate biomarkers and treatment modalities, resulting in limited progress in survival rates in recent decades. 5 Carcinogenesis is a multifaceted process involving numerous genes and factors. Among these, FGF23 and phosphate have been confirmed as cancer-related factors, with FGF23 being particularly relevant in bone-related cancers such as multiple myeloma or those with bone metastasis. Certain indications showed that oncogenic osteomalacia and elevated FGF23 levels may be relevant to breast cancer. 25 Breast cancer cells exhibit elevated FGF23 mRNA expression, and FGF secreted by malignant cells contributes to the formation of metastatic lesions. 26 In endometrial cancer, the median FGF23 concentrations in serum did not exhibit any notable distinctions when compared to benign endometrial changes. 27 Conversely, advanced-stage epithelial ovarian cancer is characterized by an increase in FGF23 plasma concentration, 7 and a specific FGF23 SNP has been related to a more favorable prognosis in this particular tumor type. 8 In our study, we analyzed the expression of FGF23 in US and its correlation with the prognosis of patients for the first time. We evaluated the expression of FGF23 in specimens of US patients using immunohistochemical assays and found that FGF23 was expressed at low levels in UNSM and ULM but was highly expressed in US, while survival analysis in the TCGA, GEO, and clinical cohorts showed that there was no significant correlation between the expression level of FGF23 and the prognosis of US patients. Moreover, by univariate and multivariate Cox regression analyses, we could not determine the statistical significance of FGF23 expression in OS. Considering that there were only 57 patients in the TCGA cohort, 50 patients in the GSE119043 microarray and 44 patients in the Suining cohort, the sample size was small and may not have afforded sufficient statistical efficacy; the median PFS and OS were NR in the low-FGF23 expression group at the end of the study, despite a median follow-up of 62.2 (range 1-105.9) months; therefore, the conclusions may be biased. Significantly, protein expression of FGF23 exhibits notable disparities in normal tissue, ULM, and US tissue. Meanwhile, shorter OS and PFS were observed in the patients with high-FGF23 expression compare with low-FGF23 expression. Although Histopathology remains the gold standard for diagnosis, FGF23 may serve as a potential biomarker for auxiliary diagnosis and predicting prognosis in US. However, further studies should be performed to validate the role of FGF23 in US. Furthermore, in the US subgroups of Suining cohort, we found that LG-ESS showed significantly better OS and PFS than the other subtypes. This result was consistent with previous studies. 28 However, we deemed it inappropriate to further dissect the disparities between HG-ESS and LG-ESS due to the of two HG-ESS patients, potentially leading to sampling bias. Survival analysis with large sample sizes can be challenging due to the low incidence of US. Sample constraints of Suining cohort prohibited us from ascertaining FGF23 mRNA expression levels, presenting a further limitation in this study. It is possible that newly diagnosed clinical samples will be employed for sequencing to investigate the expression of FGF23 mRNA and its underlying molecular mechanism in the future.

This study included three cohorts in total. The gene expression profiles and corresponding clinicopathological information were downloaded from two open databases, including a cohort from The Cancer Genome Atlas (TCGA) data portal ( http://cancergenome.nih.gov/ ) and a microarray dataset ( GSE119043 ) from the Gene Expression Omnibus (GEO) ( https://www.ncbi.nlm.nih.gov/geo/ ), which contained data from 57 uterine carcinosarcoma (UCS) patients and 50 UUS patients, respectively. The above data are publicly available. A total of 44 diagnosed patients and clinicopathological information were collected from the Department of Pathology of Suining Central Hospital from May 2014 to January 2022. This study received approval from the Review Board of Suining Central Hospital and the Ethics Committee of Zunyi Medical University. The primary endpoint was OS, defined as days from initial diagnosis to death or last follow-up. Progression-free survival (PFS) was defined as days from initial diagnosis to disease progression or death, whichever occurred first (or the date of last follow-up if progression or death had not yet occurred). Staging was assessed according to the International Federation of Gynecology and Obstetrics (FIGO) stage system. 9 According to the National Comprehensive Cancer Network (NCCN) guidelines, 10 US pathological types are most commonly include leiomyosarcoma (LMS), endometrial stromal sarcoma (ESS), UUS and adenosarcoma (AS). Although UCS is classified as a dedifferentiated or metaplastic form of endometrial carcinoma, due to its more aggressive behavior compared with the usual type of endometrial carcinoma, carcinosarcoma is still included in most retrospective studies of US as well as in the separate section of “mixed epithelial and mesenchymal tumors” of the 2014 WHO classification. 11 Therefore, inclusion criteria were pathological confirmation of US per the 2020 WHO classification of US , 12 , including LMS, ESS, AS, UUS, and UCS in this study. Patients with metastatic sarcoma from other gynecological sides or without complete information (a definitive pathological diagnosis, clinical findings, and loss of follow-up data, etc.) were excluded from our research. We downloaded the RNA sequencing (RNA-seq) expression profiles and corresponding necessary clinical information from TCGA database and the Genotype-Tissue Expression project (GTEx) ( http://commonfund.nih.gov/GTEx/ ), which contained data about 57 US tissues (2 without follow-up data) and 78 normal tissues from US patients. The differential expression of FGF23 genes between tumor tissues and normal tissues in US patients was analyzed by using the R software edgeR package ( http://bioconductor.org/packages/edgeR/ ) according to the parameters |log 2 -fold change| > 1 and P -value <0.05. In this study, we thoroughly reviewed the medical records of 44 patients with pathologically confirmed US, including 16 samples of LMS, 14 of ESS, 13 of UCS, and 1 of AS, in the Department of Pathology of Suining Central Hospital. Smooth muscle tissue of the uterine wall and leiomyoma tissue were also obtained if present. Immunohistochemistry (IHC) was performed using a mouse anti­human FGF23 monoclonal antibody (NBP3-07378; Novus Biologicals, Colorado, USA). Tissue specimens were fixed with 10% formalin and preserved in paraffin. The paraffin-embedded blocks were sectioned continuously to a thickness of 4 µm and fixed on slides for immunohistochemical staining. Paraffin-embedded sections were dewaxed and then repaired using ethylenediaminetetraacetic acid antigen retrieval buffers. The primary antibody was used at a 1 : 100 dilution with an incubation of 2 h at room temperature. Secondary antibody and detection were performed using an automated immunostainer (VENTAN Bench Mark XT, Roche, Swiss), followed by counterstaining with hematoxylin. Normal human kidney tissue was used as a positive control. The higher the expression content of FGF23 was, the greater was the distribution density and the stronger the positive result would be. FGF23 expression was evaluated using the H­score, which is defined as the product of the staining intensity and the percentage of positively stained cells. The scoring system for staining intensity was as follows: 0 points for no staining, 1 point for light yellow, 2 points for brown‒yellow, and 3 points for brown. The percentage of positive cells was also considered, and the scores were as follows: 0 points, positive cells 5% or less; 1 point, positive cells between 6% and 25%; 2 points, positive cells between 26% and 50%; 3 points, positive cells between 51%-75%; and 4 points, positive cells greater than 75%. The resulting total scores were categorized as negative (0 points), weakly positive (1-4 points), moderately positive (5-8 points), and strongly positive (9-12 points). The scoring process was performed independently by two gynecologic pathologists using optical microscopy. The decision was made after discussion within the group if inconsistent. The statistical analysis was conducted using IBM® SPSS® Statistics software version 24 (IBM, Armonk, NY), R (version 4.2.2) and GraphPad Prism 9.5.1. The distributions of categorical variables and continuous variables between cases and controls were performed using Pearson χ 2 tests and Student's t tests, respectively. Categorical data are presented as counts and percentages and were analyzed using the Chi-square test and Fisher's exact test. The skewed distributions were described with medians and interquartile ranges. The expression levels of FGF23 in patients were determined using the ggplot2 package. Kaplan-Meier analysis was used to evaluate the relationship between OS and FGF23 expression level and pathological subtypes in the US. Univariate and multivariate Cox proportional hazard regression analyses were used to investigate FGF23 expression as an independent prognostic factor of US. P  < 0.05 was considered statistically significant.

Supplemental material, sj-docx-1-tct-10.1177_15330338241245924 for Fibroblast Growth Factor 23 is a Potential Prognostic Biomarker in Uterine Sarcoma by Ling Yang, Ying Cai, Yunjia Wang, Yue Huang, Chi Zhang, Hu Ma and Jian-Guo Zhou in Technology in Cancer Research & Treatment