Tumor distribution affects bladder recurrence but not survival outcome of multifocal upper tract urothelial carcinoma treated with radical nephroureterectomy

Tumor multifocality and location are prognostic factors for upper tract urothelial carcinoma (UTUC). However, confounding effects can appear when these two factors are analyzed together. Therefore, we aimed to investigate the impact of tumor distribution on the outcomes of multifocal UTUC after radical nephroureterectomy. From the 2780 UTUC patients in the Taiwan UTUC Collaboration Group, 685 UTUC cases with multifocal tumors (defined as more than one tumor lesion in unilateral upper urinary tract) were retrospectively included and divided into three groups: multiple renal pelvic tumors, multiple ureteral tumors, and synchronous renal pelvic and ureteral tumors included 164, 152, and 369 patients, respectively. We found the prevalence of carcinoma in situ was the highest in the synchronous group. In multivariate survival analyses, tumor distribution showed no difference in cancer-specific and disease-free survival, but there was a significant difference in bladder recurrence-free survival. The synchronous group had the highest bladder recurrence rate. In summary, tumor distribution did not influence the cancer-specific outcomes of multifocal UTUC, but synchronous lesions led to a higher rate of bladder recurrence than multiple renal pelvic tumors. We believe that the distribution of tumors reflects the degree of malignant involvement within the urinary tract, but has little significance for survival or disease progression.

www.nature.com/scientificreports/ cancer-specific and disease-free survival, but there was a significant difference in bladder recurrencefree survival. The synchronous group had the highest bladder recurrence rate. In summary, tumor distribution did not influence the cancer-specific outcomes of multifocal UTUC, but synchronous lesions led to a higher rate of bladder recurrence than multiple renal pelvic tumors. We believe that the distribution of tumors reflects the degree of malignant involvement within the urinary tract, but has little significance for survival or disease progression.
Upper tract urothelial carcinoma (UTUC) comprises renal pelvic tumors (RPTs) and ureteral tumors (UTs). It is a relatively rare but aggressive urological cancer with a poor prognosis 1,2 . The current standard treatment for high-risk lesions is radical nephroureterectomy (RNU). However, disease recurrence or metastasis occurs in approximately 20-30% of patients [3][4][5] . Hence, to provide patients with better management options, a more precise and comprehensive UTUC assessment is needed. The fact that UTUC spans two neighboring localities has led to the comparison of RPTs and UTs. The incidence of RPTs seem to be higher 3,6,7 than that of UTs and RPTs often present a more advanced disease stage at the time of diagnosis 6,7 . Conversely, UTs are more likely to be organ-confined and have a greater association with bladder tumor history or recurrence 5,7 . Whether the prognosis of the two tumors is different has therefore been a matter of debate, but recent studies agree that RPTs are associated with better outcomes 3,5,6,8,9 . A possible explanation for this includes the difference in the anatomical nature of RPTs vs UTs. The renal pelvis has tougher physical surroundings that can facilitate a more complete dissection and prevent tumor extrusion, whereas the adventitial layer of the ureter is too thin to be a solid barrier 10 . The differences between RPTs and UTs have been validated, and tumor location is considered a profound factor for the evaluation of UTUC.
With this in mind, we aimed to develop a more explicit classification for multifocal tumors based on tumor distribution. We conducted a retrospective study using the nationwide multi-institution UTUC database to elucidate whether tumor location influences the disease outcomes of patients with multifocal lesions.

Methods
This nationwide study was conducted by the Taiwan UTUC Collaboration Group and was approved by the Kaohsiung Medical University Hospital Institutional Review Board [KMUHIRB-E(I)-20180214], which waived the need for formal informed consent due to the retrospective nature of the study. All personally identifiable information was removed. The study was performed according to the 1964 Helsinki declaration and strictly followed institutional guidelines and regulations.
A total of 16 participating Taiwanese hospitals contributed to the comprehensive database, which contained data on 2780 patients entered between 1988 and 2020. Exclusion criteria included those who did not receive RNU, those with only a single tumor lesion, and those without complete clinicopathological information. The remaining 685 patients all had multifocal tumors (defined as more than one tumor, regardless of location) and were divided into three groups according to tumor distribution as follows: multiple (defined as more than one tumor in a single anatomical site) RPTs, multiple UTs, and synchronous tumors (RPT + UT). Each group consisted of 164, 152 and 369 patients, respectively.
The database included the following parameters: age, gender, comorbidities [coronary artery disease, endstage renal disease, hypertension, diabetes mellitus, gout, and non-urothelial carcinoma (UC) malignancy], tumor distribution (multiple or synchronous), history of bladder tumor, pathological tumor grade, pathological tumor stage, lymph node status, histologic variant, carcinoma in situ (CIS), lymphovascular invasion (LVI), tumor necrosis, mortality from UTUC, disease recurrence, and bladder cancer recurrence. Pathological examinations were performed by genitourinary pathologists at each institution based on the same criteria. Tumor staging and grading strictly followed the 2010 American Joint Committee on Cancer TNM classification and the 2004 World Health Organization/International Society of Urologic Pathology consensus classification, respectively.
Regular follow-up appointments were arranged by urologists after RNU. Physical, laboratory and instrumental examinations were all performed according to standard guidelines. Patients with disease recurrence suffered from local relapse in the tumor bed or metastasis to the regional lymph nodes or distant organs; bladder recurrence was regarded as a separate clinical event. The cause of death was determined by attending physicians or death certificates.
For comparison of the demographic and clinicopathological factors of the three groups, Pearson's chi-square test and Student's t test were used for categorical and continuous variables, respectively. The Kaplan-Meier method was used to estimate the incidence of disease outcomes, including cancer-specific survival (CSS), diseasefree survival (DFS) and bladder-recurrence free survival (BRFS) rates, and the Cox proportional hazards model was selected to evaluate the adjusted survival and recurrence rates. The association of each clinicopathological characteristic with prognosis was examined in the univariate analysis, and those characteristics that showed statistical significance were included in the multivariate analysis. All statistical assessments were two-sided, and p < 0.05 was considered statistically significant. SPSS software, version 26 (IBM; Armonk, NY, USA), was used to perform all analyses.  Table 1 shows the demographic and clinicopathological features of the three groups based on tumor distribution. There were significant differences in history of bladder tumor (p < 0.001), tumor grade (p = 0.002), CIS (p = 0.028), LVI (p = 0.016), and tumor necrosis (p = 0.006) among the three groups.

Results
Cancer-specific outcomes. Cancer-specific outcomes include CSS and DFS. Focusing on patients with multifocal tumors, Table 2 summarizes the univariate and multivariate analyses of the CSS and DFS rates. Regarding the CSS rate, tumor grade (p = 0.017), histologic variant (p < 0.001), and tumor necrosis (p < 0.001) were significant in the univariate analysis but not in the multivariate analysis. Non-UC malignancy (p = 0.002; p = 0.003), pathological T stage (both p < 0.001) and LVI (p < 0.001; p = 0.002) were significantly associated with CSS in both analyses. The overall 5-year CSS rate in this population was 75%. Figure 1 show the unadjusted and adjusted survival curves of the three groups. For the DFS rate, tumor grade (p < 0.001), pathological N stage (p = 0.001) and histologic variant (p < 0.001) were only significant in the univariate analysis, while end-stage renal disease on dialysis (p = 0.003; p = 0.046), pathological T stage (both p < 0.001), LVI (p < 0.001; p = 0.001), and tumor necrosis (p < 0.001; p < 0.04) were demonstrated to be independent prognostic factors for the DFS rate in both analyses (Table 2). Overall, the 5-year DFS rate was 60%, and the corresponding DFS rate curve is shown in Fig. 2.
Bladder recurrence. The overall 5-year BRFS rate in this cohort was 57%. In the univariate analysis, age (p = 0.014), gender (p = 0.001), gout (p < 0.001), history of bladder tumor (p = 0.002), non-UC malignancy (p = 0.022), and tumor distribution (p = 0.046) were significantly correlated with the BRFS rate (Table 3). Additionally, age (p = 0.014), gender (p = 0.020), gout (p < 0.001), history of bladder tumor (p = 0.002), and tumor distribution (p = 0.020) showed significant associations with the BRFS rate in the multivariate analysis (Table 3). Figure 3 demonstrates the cumulative BRFS of the three groups, and the adjusted curve shows that the recurrence rate of bladder cancer in the synchronous group is higher than that in the multiple RPT group (Fig. 3b, p = 0.018).

Discussion
Previous studies showed that multifocal tumors had a worse prognosis and higher progression rate than single lesions [12][13][14]19 . A possible explanation included the correlation of multifocal tumors with a larger volume of disease 20 . Moreover, 61.8% of the patients with multifocal tumors in this population presented with pT2 stage disease or higher at the time of diagnosis. This result was similar to that of the study by Chromecki et al. 12 , in which multifocal tumors were found to be more advanced than single lesions (65.1% vs 50.6% of tumors at stage pT2 or higher). Such extensive involvement and aggressive tumor behavior are likely to worsen patient prognosis. In addition, a correlation between the total tumor load and tendency of metastasis has been observed in other cancers 21 . We hypothesize that a similar trend might exist in UTUC, and this requires further research.
However, UTUC arises from two anatomical sites that present different biological contextures, and tumor location has been shown to have prognostic significance in recent studies 3,6,8,9 . It is unknown whether the distribution of lesions has an impact on the disease outcomes of multifocal UTUC. In previous analyses, the definitions of tumor multifocality were dissimilar. We reviewed the literature and found that tumor multifocality, tumor multiplicity, and synchronous RPT and UT were individually reported to be correlated with poor prognosis [3][4][5][6][11][12][13][14][15][16][17][18][19][20] . However, comparisons between multiple and synchronous tumors are lacking. Therefore, we divided patients with multifocal tumors into three groups according to tumor distribution for outcome analysis.
Under strict patient selection, we first sought to clarify whether the cancer-specific outcomes of multifocal tumors were affected by tumor distribution. We compared multiple RPTs, multiple UTs and synchronous tumors and found no difference in the CSS and DFS rates between the groups. Thus, we considered that tumor multifocality as a whole may exert a dominant impact on cancer-specific outcomes regardless of how the tumors are distributed. Moreover, the development of multiple lesions might also blur the distinction of localities, since multiple lesions in one location may encroach on the other location. Therefore, differences in the CSS and DFS rates among the three groups may have been obscured.
Although tumor distribution seemed to have little impact on cancer progression and prognosis in patients with multifocal tumors, our study did find differences in bladder recurrence rates based on tumor distribution in the multivariate analysis. Previous studies appeared to indicate that the bladder recurrence rate was higher in multifocal tumors compared to single tumors 14,15,18,19 . Through the subdivision of multifocal tumors, our analysis showed that the bladder recurrence rate of synchronous tumors was significantly higher than that of multiple RPTs. The results of Fradet et al. 17 and our previous study 5 partially supported our findings by demonstrating no significant difference in the BRFS rate between patients with multiple lesions and those with single lesions in multivariate analyses. In summary, we believe that within multifocal tumors, synchronous lesions have a more profound impact on intravesical recurrence compared with multiple RPTs.
There are two main theories for the development of multifocal UTUC. The first is field cancerization, in which carcinogens cause independent genetic alterations in different sites of the urothelium 22 . Although all included patients had multifocal lesions, the urinary tract involvement in patients with synchronous tumors appeared to be more extensive than in patients with multiple RPTs because they had skip lesions affecting the renal pelvis and ureter. The hypothesis is also supported by the higher incidence of urothelial CIS in synchronous tumors ( Table 1, p = 0.028), because CIS is usually multifocal and can be a manifestation of extensive filed cancerization. Therefore, we consider that patients with synchronous tumors may have a wider range of cancerous changes, which are more likely to further develop bladder recurrence.  www.nature.com/scientificreports/ Tumor seeding is another hypothesis of multifocal UTUC, indicating that monoclonal cancer cells from the first tumor flow antegrade and redeposit in the urinary tract to grow new lesions 23,24 . We assumed that patients with multiple RPTs may be less likely to experience remote seeding compared to patients with synchronous tumors characterized by long-distance spread (renal pelvis to ureter). Therefore, patients with synchronous tumors had a higher risk of recurrence at the distal end (bladder).
While comprehensive molecular or genetic evidence is still needed, the findings of this study have clinical relevance. A single dose of intravesical chemotherapy after RNU has been suggested for high-risk patients by the authors of several studies and in the latest guideline to prevent bladder cancer development 2,25,26 . Our results indicated that those with synchronous tumors had the highest risk of bladder recurrence and may benefit from multiple intravesical instillations with stricter follow-up regimens.
Another finding of our study was the association of gout with bladder cancer recurrence in patients with UTUC. Previous studies have shown a higher risk of cancer development in patients with gout, especially the development of bladder cancer 27,28 . Based on our analysis, patients with gout were three times more likely to experience intravesical recurrence after RNU. Therefore, we suggest that these patients should also be closely monitored after RNU and prophylactic intravesical therapy.
Several limitations existed in this study because of its retrospective nature. First, among the multifocal tumors, only the pathological features of the dominant lesion were collected, preventing detailed analyses of individual lesions. Additionally, a centralized pathological review was difficult to conduct because this was a multi-institutional study. Furthermore, surgical practice may have changed over time throughout the three decades and, therefore, could have biased the disease outcomes. Nonetheless, this study provided insight into tumor multifocality and identified that tumor distribution had limited association with cancer-specific outcomes; however, the specific population with synchronous tumors had an increased risk for bladder recurrence.
In conclusion, multifocal UTUC represents a heterogeneous population in which tumor distribution affects disease outcomes to varying degrees. We demonstrated no difference in the CSS and DFS rates among patients with multiple RPTs, multiple UTs and synchronous tumors. However, synchronous tumors portended a higher bladder recurrence rate than multiple RPTs, which should be considered in clinical assessment. We believe that synchronous tumors may be a manifestation of widespread malignant involvement of the entire urinary tract, but have limited significance in the cancer-specific outcomes of multifocal UTUC.