Introduction

Bladder cancer (BC) is the ninth most frequently diagnosed cancer in the world and the second most common urologic malignancy, with approximately 614,000 new cases and 220,000 deaths occurring in 20221. Non-muscle invasive bladder cancer (NMIBC) accounts for about 75% of patients with BC; for younger patients (< 40 years) this percentage is even higher2. The recurrence rate of NMIBC at 5 years is as high as 31% to 78%, and a progression rate of 1% to 45%3. The 5-year survival rate for NMIBC (stage 0 and 1) ranges from 88 to 98%, and once NMIBC progresses into muscle-invasive (stages 2, 3, and 4), the 5-year survival rate decreases to 15% to 63%4. Most patients die from metastatic bladder cancer (MBC), the main reason for which is the presence of micrometastases before surgery5.

Circulating tumor cells (CTCs) are cancer cells that detach from primary tumors and enter the bloodstream, serving as precursors to metastasis. Recent research has explored the prognostic value of CTCs in NMIBC, but results have been inconsistent due to small sample sizes, heterogeneous patient groups, and limited follow-up durations6. Meta-analyses, however, suggest that CTCs are a promising biomarker for predicting poor survival and aggressive tumor progression in bladder cancer patients7. Most of these studies have used the CellSearch system, which detects CTCs based on epithelial cell surface markers. Due to the heterogeneous nature of CTCs, including cells undergoing epithelial-mesenchymal transition (EMT), this method may miss certain CTC populations, leading to an underestimation of the actual count.8.

Alternative CTC detection methods include reverse transcriptase-polymerase chain reaction (RT-PCR), flow cytometry, and next-generation sequencing (NGS). However, these approaches have notable limitations: RT-PCR is prone to RNA degradation or contamination, flow cytometry lacks sensitivity, and NGS is costly9. A newer technique, subtraction enrichment and immunostaining-fluorescence in situ hybridization (SE-iFISH), developed in 2015, offers significant advantages. SE-iFISH combines nucleic acid, protein, and cell morphology analysis, enabling in situ phenotypic identification of tumor biomarkers, assessment of cell size, and chromosomal analysis of CTCs and circulating tumor endothelial cells (CTECs)9,10.

SE-iFISH technology can be used to detect the presence and dynamic changes of target proteins in CTCs as biomarkers related to treatment response and prognosis of patients. For example, vimentin expression in CTCs associated with liver metastases predicts poor progression‑free survival in patients with advanced lung cancer11. Continuous examining human epidermal growth factor receptor-2(HER2) phenotype on CTCs could help monitor resistance to therapy in advanced gastric cancer patients in real time12.

Despite the promising capabilities of SE-iFISH, its application for analyzing CTCs in NMIBC remains underexplored. In this study, we utilize the SE-iFISH method to detect postoperative CTC levels in high-risk NMIBC patients, aiming to determine the correlation between CTC count and tumor prognosis.

Materials and methods

A total of 59 fresh peripheral blood samples were collected from 18 healthy volunteers, 2 patients with benign urinary lesions, 20 patients with newly diagnosed BC, and 19 NMIBC patients receiving repeated-transurethral resection of bladder tumor(R-TURBT), from April 2021 to July 2022. A detailed flowchart was shown in Figure S1. The detailed information about the material and methods can be found in the Supporting Information.

Results

Patient characteristics

A total of 59 subjects were included in the current study, out of which 18 were healthy donors, 2 were cystitis patients, 20 were newly diagnosed BC patients, and 19 were NMIBC patients receiving R-TURBT. Patient characteristics are presented in Table 1. We included 11 cases of G3 high-grade non-muscle-invasive bladder cancer (NMIBC), 3 cases of T1NMIBC, and 2 cases of T1G3NMIBC. Cystitis patients and patients with newly diagnosed BC were detected before treatment. CTCs were counted in 16 patients with high-risk NMIBC and 3 patients with unknown invasion depth of bladder cancer R-TURBT. Theses 19 samples were collected within a window from the second to the fourth postoperative day. Specifically, 11 samples were collected on the third day after surgery, 4 on the second day, and 4 on the fourth day. The 19 post-surgery patients were followed up with a median follow-up time of 8 months (range 3–12), in which 3 high-risk NMIBC patients recurred.

Table 1 Clinical characteristics of 59 enrolled individuals.
Full size table

Identification of CTCs and CTECs

CTCs were identified as having an abnormal chromosome 8 karyotype. Figure 1 shows some representative images of different CTCs. With the general size of WBCs as the threshold, CTCs were identified as either large (> 5 µm) (Fig. 1A) or small (≤ 5 µm) (Fig. 1B). CTCs were further divided into triploid, tetraploid, and pentaploid karyotypes and beyond (≥ pentaploidy). We did not analyze the data of haploid CTCs, CTC cluster, and CTECs as they were hardly detected in our research.

Fig. 1
Fig. 1
Full size image

Represent CTCs and CTEC detected by SE-iFISH in the peripheral blood of bladder cancer patients. (A) ≥ Pentaploid (CEP8 ≥ 5), large CTC (diameter > 5 µm). (B) Tetraploid (CEP8 = 4), small CTC (diameter ≤ 5 µm). (C) CTC cluster (≥ 2 CTCs). (D) CTEC.

Aneuploidy analysis of CTCs of different sizes

The content of CTCs in the 20 newly diagnosed bladder cancer patients (cancer group) ranged from 1 to 17, and the median number was 9. The CTCs contained in 20 non-cancer patients (control group) ranged from 0 to 10, with a median of 1.5, and the difference was statistically significant (P < 0.0001) (Fig. 2). In the cancer group, more than half of the total detected CTCs were large CTCs, with a proportion of 58.8%, while small CTCs comprised 41.2%. Karyotyping revealed that ≥ pentaploid CTCs accounted for the largest proportion (61.2%) of the entire CTCs in cancer group. Triploid and tetraploid accounted for 23.0% and 15.8%, respectively. However, the majority of the total CTCs in the control group were large CTCs, accounting for 83.0% and small CTCs accounted for only 17%. The largest proportion of aneuploid CTCs was ≥ pentaploid CTCs (72.3%), followed by triploid (12.8%) and tetraploid (14.9%).

Fig. 2
Fig. 2
Full size image

Distribution of CTCs and CTCs subtypes in different groups. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

To further understand the significance of various subtypes of CTCs in BC, cell size and chromosome 8 ploidy were analyzed, and the distribution of 11 subgroups of CTCs in control and BC groups is shown in Fig. 2. Both large and small CTCs had utility in distinguishing between the control group and BC (P = 0.0003 and 0.0002, respectively). And ≥ pentaploid and ≥ pentaploid large CTCs (but not ≥ pentaploid small CTCs) also were statistically significant in the BC group and control group (P < 0.0001 and 0.0034, respectively). However, other subtypes of CTCs did not have an advantage in identifying non-cancer and BC patients (P > 0.05) (Fig. 2).

CTC subtypes and PFS

Among 19 NMIBC patients treated with R-TURBT, 3 high-risk NMIBC patients (16%) had recurrence during the follow-up period, and the remaining 16 patients (84%) had no recurrence or metastasis. Figure 3A shows the proportion of aneuploid cells of different sizes in the non-recurrence and recurrence groups. The total CTC count of patients without recurrence accounted for 84% of large CTCs and 16% of small CTCs. While in the recurrence group, large CTCs accounted for 29% and small CTCs accounted for 71%.

Fig. 3
Fig. 3
Full size image

CTC subtypes correlated with prognosis. (A,B) Publication of CTC subtypes in recurrence and non-recurrence groups. (C) ROC curves of small CTCs count, the sum of tetraploid and ≥ pentaploid small CTCs, the sum of triploid and ≥ pentaploid small CTCs, and tetraploid small CTCs count. (D) Small CTCs number ≥ 5 showed poor prognosis with shorter PFS in NMIBC patients receiving R-TURBT.

Concerning the distribution of small CTC subtypes between non-recurrence and recurrence groups, in the non-recurrence group, triploid small CTCs had the highest proportion (68%) among small CTCs, tetraploid and ≥ pentaploid small CTCs accounts for 29% and 3% respectively; While in recurrence group, tetraploid small CTCs had the highest proportion (40%), triploid small CTCs followed (35%), and the proportion of ≥ pentaploid small CTCs was the least (25%) (Fig. 3B).

ROC analysis was also performed to evaluate the ability of small CTCs and subtypes to discriminate non-recurrence patients from recurrence patients (Fig. 3C). The best results showed that the AUCs were 0.9792 for small CTCs (95% CI = 0.9194–1.000, P = 0.01), cut-off: 5 cells/6 mL, sensitivity: 100.00%, specificity: 93.75%. We also found that combined tetraploid small CTCs and ≥ pentaploid small CTCs could show better clinical value as indicated by the AUC of 0.9583 (95%CI = 0.8611–1.000, P = 0.01), which was higher than that of triploid or ≥ pentaploid small CTCs alone. The AUC of triploid small CTCs combined with ≥ pentaploid small CTCs is 0.9375 (95%CI = 0.8169–1.000, P = 0.02). Only the tetraploid small CTC subtypes showed good clinical value with AUC of 0.8750 for small CTCs (95%CI = 0.6909–1.000, P = 0.04). There was no significant association between total or large CTC count and prognosis during postoperative follow-up (data is not fully shown).

As for high-risk NMIBC patients undergoing R-TURBT, the peak risk of postoperative recurrence occurred at 3 months postoperatively13. PFS was significantly reduced in the Small CTCs ≥ 5 group (Fig. 3D). The hazard ratio (Mantel–Haenszel) was 362.1(95%CI:17.45–7512). Small CTC count (≥ 5CTCs/6 ml) predicted poor prognosis in high-risk NMIBC patients receiving R-TURBT. Moreover, we have conducted an ROC curve analysis comparing BC patients and healthy individuals for each CTC subtype. The result was shown in Fig. 4. We found that small cell CTCs still yielded the highest AUC value.

Fig. 4
Fig. 4
Full size image

ROC curves of different CTC subtypes.

The prediction of prognosis based on CTC subtypes

In order to extend the clinical applicability, the numbers of CTC subtypes cell count, cancer embolus and CTC were all included in a logistic ridge regression formula to predict the recurrence after TURBT. Due to the small clinical sample and the potential multicollinearity, the logistic ridge regression is suitable for the study, reducing the error of prediction. The ridge parameter λ was determined by cross-validation. By dividing the dataset into multiple subsets and testing the validity of different models with different λ, the model with the best performance is selected with its corresponding λ value. The partial regression coefficients the standard regression coefficients and are listed in Table 2. The results show that small CTC (β = 0.168), cancer embolus (β = 0.025), triploid CTC (β = 0.057), tetraploid CTC (β = 0.113), triploid small CTC (β = 0.103), tetraploid small CTC (β = 0.231), ≥ pentaploid small CTC (β = 0.626) is positively correlate with the recurrence, among which the ≥ pentaploid small CTC has the strongest effect (B = 0.514). The results in this part are basically correspondent to the ones mentioned above.

Table 2 Ridge regression model on the relationship between CTC subtypes and prognosis.
Full size table

The reliability of the prediction model was inspected by estimating accuracy. Leave one out cross validation (LOOCV) was utilized to assess the accuracy of the model. By setting a training set of certain proportion in the dataset randomly and calculating the accuracy with the testing set, the accuracy of different training–testing sets can be estimated (Table 2).

Discussion

Bladder cancer is a highly heterogeneous tumor with multiple treatment options. Despite advances in medicine, survival rates for bladder cancer have remained stagnant over the past three decades4. The main challenge of NMIBC management is its high recurrence rate. Currently, the determination of risk groups, treatment options, clinical trial eligibility, and follow-up protocols for NMIBC depends on data obtained from tumor tissue following transurethral resection of bladder tumor (TURBT). However, depending on these clinicopathological features alone, it is often difficult to choose the best treatment for the individual patient. Our understanding of bladder cancer biology is still emerging, and the search for biomarkers is in its early stages. Blood-based biomarkers for bladder cancer have shown a wide range of sensitivity and specificity (2.4–97.6% and 43.3–100%, respectively), with few achieving both high accuracy and independent validation14. Recent gene sequencing and expression studies have revealed several DNA, RNA, and protein markers, highlighting distinct molecular subtypes that can better predict disease progression and treatment response15. However, the European Association of Urology still regards these markers as promising but not yet suitable for routine use2,16.

Our study identified circulating tumor cells (CTCs) as predominant in patients with bladder cancer, particularly revealing significant differences in the prevalence of ≥ pentaploid CTCs between benign and cancerous cases. Aneuploid CTC subtypes may prove valuable in distinguishing malignant tumors. These cells likely have a lower malignancy potential than “true” CTCs and may be eliminated by the immune system17. Not all tumor cells that enter the circulation have metastatic capacity; the majority of cells released during surgical manipulation or spontaneously are either apoptotic, lack stem-like or EMT phenotypes, or are rapidly eliminated by immune surveillance mechanisms18,19,20. Only a small subset of ‘true’ CTCs is capable of escaping immune destruction and establishing metastases. However, the presence of small numbers of CTCs in healthy individuals and those with benign conditions requires further investigation. We believe that the preliminary findings from this study still offer valuable insights, and the expansion of the sample size in future work will help to validate and strengthen these findings.

In patients with recurrent NMIBC after surgery, we observed that the count of small CTCs detected using SE-iFISH was significantly higher than in non-recurrent patients. Small CTCs were correlated with poorer outcomes among high-risk NMIBC patients undergoing repeat TURBT. Aneuploid analysis revealed that small CTCs primarily exhibited chromosome 8 triploidy or tetraploidy, and tetraploid small CTCs were particularly associated with a poor prognosis21. Our prediction model introduced several other indicators, including cancer embolus and aneuploid CTCs, which enhance the possibility of earlier diagnosis and prediction of prognosis. These findings are consistent with research in other cancers, such as non-small cell lung cancer and hepatocellular carcinoma, where smaller circulating aneuploid cells, especially triploid ones, have been linked to shorter disease-free survival and increased likelihood of relapse22. Moreover, this study found the presence of CTCs in patients with benign urinary system diseases. Therefore, expanding the inclusion of a healthy control group to explore the quantity and subtype distribution of CTCs in non-cancerous populations will provide valuable insights and an experimental foundation for subsequent CTC screening in both healthy individuals and cancer patients.

EpCAM and vimentin have been identified as biomarkers for epithelial and mesenchymal cells, respectively. Studies suggest that tumor cells undergoing epithelial-mesenchymal transition (EMT) are smaller than those without EMT characteristics23. In our study, vimentin was predominantly found in small CTCs and CTC clusters, suggesting dynamic changes in both cell size and surface markers during EMT11,24. The presence of mesenchymal tumor-initiating cells (TICs) is known to enhance the metastatic capability of epithelial TICs, accelerating disease progression25.

The diagnostic utility of small CTCs is increasingly recognized, as advanced detection technologies have revealed that small CTCs may be present in a significant proportion of cancer patients and may even indicate aggressive disease phenotypes26. CTC size heterogeneity is influenced by biological processes such as epithelial-mesenchymal transition, apoptosis, and tumor cell plasticity, resulting in the coexistence of both small and large CTCs27. These findings underscore the need for sensitive, marker-independent detection approaches to fully capture the clinical significance of all CTC subpopulations.

Aneuploidy is a key characteristic of malignant cells, with approximately 90% of solid tumors and 75% of hematological cancers showing evidence of aneuploidy28. Chromosome instability—the driver of aneuploidy—promotes tumorigenesis by increasing genetic diversity and driving tumor evolution29. This makes aneuploidy a potential therapeutic target and a prognostic marker, as well as a possible source of drug resistance30. Our study found that only tetraploid karyotype small CTCs were linked to poorer outcomes in high-risk NMIBC patients receiving R-TURBT. In other studies, the ploidy type of chromosome 8 in CTCs among 19 different carcinoma patients were detected by SE-iFISH. Triploid and polyploid CTCs are the most common phenotypes, while tetraploid CTCs are more common in bladder cancer24. The clinical significance and role of CTC subtypes with different chromosome 8 aneuploidy in cancer development and progression should be elucidated.

SE-iFISH utilizes a negative enrichment strategy to identify and visualize diverse types of CTCs. This technique enables characterization of highly heterogeneous CTCs based on aneuploidy and tumor biomarkers10. Previous studies integrating SE-iFISH with high-throughput image scanning systems identified 71 distinct CTC and circulating tumor endothelial cell (CTEC) subtypes, each potentially linked to specific clinical outcomes, such as metastasis, relapse, drug sensitivity, or resistance31,32. Despite its utility, SE-iFISH has limitations, including a risk of false negatives31,33.

Our findings suggest that subclass analysis of CTCs based on chromosome 8 aneuploidy and cell size could offer new insights into the clinical management of bladder cancer. However, this study has certain limitations. First, the sample size was small, warranting further validation with larger cohorts. Second, we did not examine preoperative CTCs in high-risk NMIBC patients, which may limit the completeness of our conclusions. Lastly, the follow-up period was relatively short. A longitudinal study with an extended follow-up period is crucial to determine the true value of this test in ongoing patient surveillance, as well as its potential to predict recurrences and progression over time. Future studies should involve a larger number of patients with dynamic monitoring of CTCs to assess long-term outcomes more comprehensively.

Conclusions

In this study, we employed the SE-iFISH technique to detect circulating tumor cells (CTCs), circulating tumor endothelial cells (CTECs), and their subtypes in the peripheral blood of NMIBC patients. Our findings indicate that the count of CTCs and specific subtypes can effectively differentiate patients with bladder cancer from those without the disease. Additionally, the presence of small aneuploid CTCs, especially ≥ pentaploid small CTC, serves as a significant predictor for short-term recurrence in high-risk NMIBC patients undergoing repeat TURBT.