Data was collected from patients diagnosed with urothelial carcinoma at Shandong Provincial Hospital who received treatments from January 2022 to December 2023. Based on the patients’ clinical conditions, physical status, and individual needs, our center provided a range of treatment options, including transurethral resection of bladder tumor (TURBT), partial cystectomy, radical cystectomy, nephroureterectomy, and chemotherapy. The inclusion criteria encompassed patients meeting the following conditions: (1) a pathological diagnosis of urothelial carcinoma, and (2) positive FISH test results. A total of 149 patients meeting these dual criteria were included in the study. Among them, data from 119 patients treated during the first 20 months were allocated to the training set, while the remaining 30 patients were assigned to the test set.

200 ml of morning urine was collected with stringent measures to prevent contamination during sample collection. Samples were promptly sent for testing to ensure their integrity. Urine collection was conducted prior to any procedures that could potentially impact test outcomes, including enhanced CT or MRI scans, cystoscopy, ureteroscopy, prostate rectal examination, or catheterization.

The urine samples were processed into cytospin slides for analysis as exfoliated cell specimens. Following enzymatic digestion and ethanol dehydration, the cytospin slides underwent CSP3/CSP7 and CSP17/GLPp16 fluorescent probe hybridizations, washings, restainings, and observation of FISH staining results.

Urine samples were collected from 20 healthy volunteers. The FISH experiments were performed as previously described, with 100 clearly defined signals observed for each probe combination. The percentages of cells exhibiting various types of abnormalities were calculated, and a threshold was established, defined as the mean plus three times the standard deviation (M + 3SD). The interpretation of the FISH results required analyzing 100 non-overlapping urothelial cells for each probe in each sample. This analysis involved calculating the percentage of cells with different abnormal conditions. Results exceeding the threshold were designated as positive, while those below the threshold were classified as negative.

All pathologies were surgically removed or extracted during cystoscopy by the chief urology physician at Shandong Provincial Hospital. Subsequently, they were reviewed and diagnosed by the senior doctor of the pathology department at the same hospital.

For the patients in the training set, a univariate analysis was conducted to examine the relationship between chromosome mutations and pathological grading. Relevant chromosomal variations, age, and gender were jointly analyzed through a multifactor analysis, gradually eliminating non-significant factors and retaining influential factors with p < 0.05 to establish a logistic regression model. Utilizing the decision curve analysis (DCA) to calculate the net benefit within the threshold probability range. Establishment of a calibration curve and performing the Hosmer-Lemeshow test to assess the model’s fit. Performance evaluation of the model in the training and test sets using the receiver operating characteristic (ROC) curve.

Freshly collected urothelial carcinoma tissues were initially fixed in 4% paraformaldehyde for 3 hours. Subsequently, the tissues underwent a series of dehydration processes: immersion in 75% ethanol for 1.5 hours, 95% ethanol for 1.5 hours, 95% ethanol for 1 hour, anhydrous ethanol for 1.5 hours, anhydrous ethanol for 1 hour, and xylene I and xylene II for 0.5 hours each. After dehydration, the tissues were embedded in paraffin and sectioned into 4-micrometer slices. The slices were subjected to high-temperature antigen retrieval at 65°C for 20 minutes, followed by deparaffinization in xylene for 1 to 3 minutes and rehydration through an ethanol gradient. Microwave antigen retrieval was then performed, and endogenous peroxidase and avidin activities were blocked using an endogenous peroxidase blocking buffer (purchased from Dowobio, China, DW 2176). Afterward, the cells were blocked with normal goat serum (purchased from ZSGB-BIO, China, ZLI-9022) for 30 minutes at room temperature. Following the blocking step, the cells were incubated with the primary antibody against p19, (dilution multiple: 1:200) (purchased from CST, USA, #18769) at 4°C for 16 hours. Post incubation, the cells were exposed to mouse horseradish peroxidase solution at room temperature for 120 minutes and stained with 3,3’-diaminobenzidine (purchased from Beyotime Biotechnology, China, P0202) for 5 minutes. After counterstaining with hematoxylin for 2 minutes, the excess dye was removed with PBS washing, and differentiation was performed in 1% hydrochloric acid alcohol for 3 seconds. Finally, the cells underwent dehydration in 70% ethanol for 1 minute, 80% ethanol for 1 minute, 95% ethanol for 2 minutes, anhydrous ethanol for 4 minutes, xylene I for 3 minutes, and xylene II for 3 minutes before the sections were sealed. The results were observed with the optical microscope. Three high-power fields were assessed for each section, and each measurement was conducted in triplicate.

The collected data were statistically analyzed using SPSS 20.0 software. Chi-square tests were utilized to assess the relationship between FISH detection results and the TNM staging and pathological grading of urothelial carcinoma. Multivariate analysis using R packages was employed to examine the correlation between multiple factors and the pathological grading of urothelial carcinoma, followed by binary Logistic regression analysis. A significance level of p < 0.05 was used to determine statistical significance.

The basic information of all patients is shown in Table 1 . In the training set, there were 119 patients with urothelial carcinoma, consisting of 101 males (84.87%) and 18 females (15.13%), with a mean age of 66.52 ± 12.62 years. According to the TNM staging classification, there were 4 patients with Ta stage (3.36%), 65 with T1 stage (54.62%), 22 with T2 stage (18.49%), 18 with T3 stage (15.13%), and 10 with T4 stage (8.40%). There was a total of 75 patients with high-grade urothelial carcinoma (63.03%) and 44 patients with low-grade urothelial carcinoma (36.97%). In the test set, there were 30 patients, including 23 males (76.67%) and 7 females (23.33%), with a mean age of 65.16 ± 9.43 years. The distribution of Ta, T1, T2, T3, T4 stages was 1 (3.33%), 5 (16.67%), 8 (26.67%), 11 (36.66%), 5 (16.67%) respectively, with 13 (43.33%) cases of low-grade urothelial carcinoma and 17 (56.67%) cases of high-grade urothelial carcinoma.

The results of FISH depicted in Figure 1 revealed diploid normal cells displaying 2 red/2 green fluorescent signals under the microscope when labeled with the probe. The CAS3 and CSP7 probes, respectively targeting chromosome 3 and chromosome 7, exhibited green and red fluorescence. Similarly, the CSP17 and GLPp16 probes, labeling the centromere of chromosome 17 and the p21 locus of chromosome 9, displayed green and red fluorescence, respectively. Anomalies in the deletion or amplification of red or green fluorescent signals detected by fluorescence signal implied abnormalities, signifying chromosomal deletions or amplifications. Notably, mutations were identified in chromosomes 3, 7, and 17, indicating amplifications, while chromosome 9 p21 site mutations involved both amplifications and deletions. Research data indicated that 89.92%, 93.28%, and 86.55% of patients respectively exhibited amplification of chromosomes 3, 7, and 17, while 85.71% of patients displayed a mutation in chromosome 9p21. Specifically, among all patients with mutations in GLPp16, 51.96% showed amplification while 48.04% exhibited deletion. ( Figure 2A ). The study further delineated four chromosomal mutation combinations observed in the 119 patients, as represented in Figure 2B . Among these, CSP3, CSP7, CSP17, and GLPp16 mutations were amplified in 51 patients. Additionally, 33 patients demonstrated amplifications in CSP3, CSP7, and CSP17 along with GLPp16 deletions, while 13 patients displayed amplifications in CSP3, CSP7, and CSP17 alongside GLPp16 unmutations. Furthermore, five patients exclusively exhibited GLPp16 deletions, without mutations in CSP3, CSP7, and CSP17.

Our study findings, illustrated in Figure 3A , indicated a significant correlation between urothelial carcinoma pathological grade and the amplification of CSP7 as well as aneuploidy at GLPp16 (χ2 (CSP7) = 9.39, χ2 (GLPp16) = 25.34, p < 0.05) within the entire training cohort. Conversely, the amplification of CSP3 and CSP17 did not exhibit statistical variance between the study groups (p > 0.05). Furthermore, comparisons between aneuploidy alterations at CSP3, CSP7, CSP17, and GLPp16 did not reveal significant discrepancies with respect to TNM staging of urothelial carcinoma ( Figure 3B , p > 0.05).

In our investigation utilizing logistic regression analysis, we assessed the influence of gender, age, and aneuploidy changes in CSP7 and GLPp16 on the occurrence of low-grade urothelial carcinoma. The analysis revealed that gender and amplification of GLPp16 were inversely associated with the development of low-grade urothelial carcinoma ( Figure 4A , p < 0.05), whereas age and amplification of CSP7 did not show a significant association with urothelial carcinoma stage (p > 0.05). Based on these findings, we constructed a nomogram model incorporating gender and GLPp16 ( Figure 4B ). The DCA results ( Figure 4C ) indicated that utilizing the nomogram model to make treatment decisions conferred greater net benefit across various threshold probabilities ranging from 36% to 95% compared to treating all patients or none. The calibration curve for the nomogram training set ( Figure 4D ) exhibited a non-significant Hosmer-Lemeshow test result (p = 0.436), signifying strong calibration ability and alignment between model predictions and actual observations. Furthermore, the ROC curve of the nomogram displayed an area under the curve (AUC) of 0.785 (95% CI: 0.707 - 0.863), demonstrating excellent discrimination ( Figure 4E ), while the ROC curve of the test set revealed an AUC of 0.740 (95% CI: 0.559 - 0.920), indicating robust discrimination capability in practical application ( Figure 4F ).

The P16 locus on chromosome 9 tended to be deleted in low-grade urothelial carcinoma ( Figure 5A ) and amplified in high-grade urothelial carcinoma ( Figure 5B ). Figure 5C shows that GLPp16 was unmutated in high-grade urothelial carcinoma.