This article is mentioned in:

Introduction

Radical cystectomy and pelvic lymphadenectomy are the standard treatments for muscle-invasive bladder cancer (BC) (1), and the presence of lymph node (LN) metastases is associated with poor survival outcomes (2). In the 8th American Joint Committee on Cancer/International Union Against Cancer tumor (T) node (N) metastasis (M) staging system, patients with positive LNs are stratified into three stages (N1, N2 and N3) based on the number of positive LNs and the site of metastatic LNs (3).

The number of positive LNs, LN density (defined as the number of positive LNs divided by the total number of removed LNs), maximum diameter of metastatic LNs and extranodal extension (ENE) are independent prognostic factors in BC (4–9). A previous multivariable model that included LN density, number of removed LNs, number of positive LNs and ENE, demonstrated that ENE is a strong independent prognostic factor (7).

The International Collaboration on Cancer Reporting currently recommends that the extranodal spread in regional LN status be included in the pathology reporting of the carcinoma of the bladder/cystectomy (10). Although ENE in metastatic LNs is associated with worse cancer-specific survival (CSS), an ENE assessment is not included in the current TNM staging system due to limited published data. Therefore, LN evaluation methods that include ENE are required.

In a previous study, the prognostic impact of cancer invasion levels in sentinel LNs was reported in melanoma (11). In this assessment, sentinel node invasion level (SNIL) was defined as follows: SNIL 1, tumor cells confined to lymphatic vessels and subcapsular sinus or transverse sinuses; SNIL 2, tumor cells infiltrating the cortex or paracortex; and SNIL 3, tumor cells infiltrating the medulla or capsule (11). The SNIL status stratified three independent risk groups of patients who were SN-positive. This is a simple histology-based assessment of metastasis using the anatomic localization of LNs without distance measurements with a microscope (12). Nevertheless, the prognostic impact of LN invasion levels in anatomically- and immunologically-defined substructures has not been assessed in BC.

In the present study, the clinical significance of LN invasion levels in BC was assessed and the prognostic accuracy of the generated model pathological (p)T stage and LN invasion level] and the conventional model (pT and pN stages) in patients who underwent radical cystectomy and pelvic lymphadenectomy was compared.

Materials and methods

Case selection

After receiving ethics approval from Institutional Review Board of Kansai Medical University Hospital (Osaka, Japan; approval no. 2021226), data from 131 patients with BC (≥pT1) who underwent radical cystectomy at Kansai Medical University Hospital between January 2006 and December 2017 were extracted from the institutional database. An opt-out approach was used to obtain informed consent on the hospital website. A total of 33 patients were excluded, including eight patients with nonurothelial carcinoma (small cell carcinoma, n=2; urachal carcinoma, n=2; and undifferentiated carcinoma, n=4), 14 patients with unremoved LNs, and 11 patients treated with neoadjuvant chemotherapy (NAC). Thus, 98 patients who underwent radical cystectomy and pelvic lymphadenectomy were selected for analysis. Patients with TNM clinical stage III or higher in the present study underwent an extended pelvic lymphadenectomy.

Histological evaluation

All surgical specimens were processed according to standard pathology procedures. Hematoxylin and eosin (H&E)-stained slides prepared for routine histological examination were re-evaluated by a urologic pathologist (CO) using the 2016 World Health Organization classification (13) and the 2017 Union for International Cancer Control TNM staging system (3). Clinicopathological characteristics such as pathological TNM stage, grade and histological subtypes of urothelial carcinoma, including divergent differentiation/subtypes, lymphovascular invasion and surgical margin, were reviewed. Furthermore, the association between inflammatory status and lymph node metastasis was assessed using histology-based tumor-associated immune cell status (TAICs), a methodology previously reported by our group (14).

The resected LNs of pelvic lymphadenectomy during radical cystectomy were evaluated using H&E-stained slides. Based on best practice guidelines for the routine pathology evaluation of the immune system (15), pN was assessed based on the TNM staging system (3): pN0, no cancer cells in any LNs; pN1, single regional LN metastasis in the true pelvis (perivesical, obturator, internal and external iliac, or sacral LN); pN2, multiple regional LN metastases in the true pelvis; and pN3, LN metastasis in the common iliac LNs.

The LN invasion level in each case was evaluated in the most highly invasive LNs with reference to the previous methodology (11): Level 0, no cancer cell within the resected LNs; Level 1, cancer cells confined to intracapsular lymph vessels and subcapsular or transverse sinuses; Level 2, cancer cells infiltrating the cortex, paracortex, or medulla; and Level 3, cancer cells infiltrating or beyond the LN capsule, corresponding to the ENE (Fig. 1). Histological evaluations of LNs were performed by two independent pathologists (JI and CO) blinded to clinical outcomes, and discordant patterns were resolved by consensus. The LN invasion level was evaluated using the H&E-stained slides only.

Figure 1. Schematic diagram and representative images of H&E-stained slides of invasion levels in metastatic LNs. Level 1, cancer cells confined to intracapsular lymph vessels and subcapsular or transverse sinuses; Level 2, cancer cells infiltrating the cortex, paracortex or medulla; and Level 3, cancer cells infiltrating or beyond the LN capsule. Cancer metastatic areas are indicated by red dotted lines in the H&E-stained images. Scale bar, 100 µm. H&E, hematoxylin and eosin; LN, lymph node.

Statistical analysis

Continuous data are presented as median and interquartile range (IQR). Fisher's exact test and the Mann-Whitney U test were used for comparisons between two groups. Recurrence-free survival (RFS), CSS and overall survival (OS) were assessed using the Kaplan-Meier method and the log-rank test. Harrell's concordance index (c-index) was used to compare the predictive accuracy of the Cox models. Multivariate Cox proportional hazard models were assessed to determine hazard ratio (HR). All statistical analyses were performed using EZR version 1.55 (Saitama Medical Center, Jichi Medical University) (16). P<0.05 was considered to indicate a statistically significant difference.

Results

Patient characteristics

As shown in Table I, of the 98 patients, 16 (16.3%) were female and 82 (83.7%) were male, with a median age of 71 (IQR, 67.0-77.0) years. Pathology examination revealed the proportions of pT1, pT2, pT3 and pT4 as 9.2% (9/98), 34.7% (34/98), 42.9% (42/98), and 13.3% (13/98), respectively. Divergent differentiation/subtypes on urothelial histology were observed in 33 patients (33.7%). The number of removed LNs was 22.5 (IQR, 14.3-29.5). A total of 26 (26.5%) patients experienced recurrence and 22 (22.5%) patients died due to BC. The median follow-up time was 99.0 (IQR, 72.2-129.0) months. The number of metastatic lymph nodes at each lymph node invasion level is presented in Table SI. Furthermore, a total of 12 patients (12.24%) with intravesical Bacillus Calmette-Guérin therapy were included.

Table I. Patient characteristics (n=98).

Table I.

Patient characteristics (n=98).

Variable	Value
Age, years	71 (67.0–77.0)
Sex
Female	16 (16.3)
Male	82 (83.7)
Tumor grade
Low	0 (0.0)
High	98 (100.0)
Divergent differentiation/subtypea
Absent	65 (66.3)
Present	33 (33.7)
pT stage
1	9 (9.2)
2	34 (34.7)
3	42 (42.9)
4	13 (13.3)
pN stage
0	69 (70.4)
1	12 (12.2)
2	15 (15.3)
3	2 (2.0)
Surgical margin
Negative/X	91 (92.8)
Positive	7 (7.1)
Lymphovascular invasion
Absent	22 (22.5)
Present	76 (77.6)
Number of removed LNs	22.5 (14.3–29.5)
Adjuvant chemotherapy
No	64 (65.3)
Yes	34 (34.7)
Recurrence	26 (26.5)
Cancer-specific mortality	22 (22.5)
Overall mortality	40 (40.8)
Follow-up time, months	99 (72.2–129.0)

{ label (or @symbol) needed for fn[@id='tfn1-ol-28-5-14650'] } Data are presented as n (%) or median (interquartile range). pT, pathological tumor; pN, pathological node; LNs, lymph nodes.

a Divergent differentiation (squamous/glandular) and subtype (micropapillary/sarcomatoid).

Association of LN invasion level with pN stage, and TAICs with pN stage

The proportion of pN0, pN1, pN2 and pN3 was 70.4% (69/98), 12.2% (12/98), 15.3% (15/98) and 2.0% (2/98), respectively (Table I). Level 2 accounted for the majority of cases in pN1 and pN2, and Levels 2 and 3 accounted for 50% of the cases in pN3 (Fig. 2). TAICs was significantly associated with pN stages (P=0.03; Table SII).

Figure 2. Lymph node invasion level distribution in each pN stage. pN, pathological node.

Comparison of prognostic significance between LN invasion level and pN stage

The Kaplan-Meier survival curves of RFS, CSS and OS in LN invasion Levels 1–3 and pN1-3 are presented in Fig. 3. In comparison with Level 1, LN invasion Levels 2 and 3 were associated with a higher risk of RFS (HR, 6.84; P=0.08 and HR, 20.80; P=0.01, respectively; Fig. 3A), CSS (HR, 5.72; P=0.11 and HR, 18.90; P=0.01, respectively; Fig. 3B) and OS (HR, 9.17; P=0.04 and HR, 29.38; P=0.002, respectively; Fig. 3C). In contrast, in comparison with pN1, pN2 and pN3 were associated with a higher risk of RFS (HR, 2.01; P=0.29 and HR, 3.12; P=0.20, respectively; Fig. 2D), CSS (HR, 1.55; P=0.53 and HR, 3.28; P=0.18, respectively; Fig. 3E) and OS (HR, 1.92; P=0.22 and HR, 2.17; P=0.35, respectively; Fig. 3F), but without statistical significance. Cases with low TAICs were associated with a significantly worse RFS rate (HR, 5.13; P<0.001; Fig. S1A), CSS rate (HR, 3.64; P=0.008; Fig. S1B) and OS rate (HR, 2.29; P=0.01; Fig. S1C) than those with high TAICs. Furthermore, 20/29 of patients with LN metastasis received adjuvant chemotherapy, which did not significantly improve CSS or OS (HR, 0.70; P=0.60 and HR, 0.52; P=0.20, respectively).

Figure 3. Comparison of prognostic significance between lymph node invasion levels and pN stages. Kaplan-Meier curve of (A) recurrence-free survival, (B) cancer-specific survival and (C) overall survival for lymph node invasion levels, and of (D) recurrence-free survival, (E) cancer-specific survival and (F) overall survival for pN stages. pN, pathological node; HR, hazard ratio; CI, confidence interval.

Prognostic significance of LN invasion level

The associations of clinicopathological factors with RFS, CSS and OS after radical cystectomy are presented in Table II, Table III, Table IV. In the univariate analysis, pT4, pN3 and LN invasion level 3 were significantly associated with RFS (P=0.026, P=0.009 and P<0.001, respectively), CSS (P=0.042, P=0.004 and P<0.001, respectively) and OS (P=0.023, P=0.017 and P<0.001, respectively). In multivariate analysis, two models were evaluated: Model 1 (conventional model) with pT and pN stages; and Model 2 (proposed model) with pT stage and LN invasion level. In comparison with Model 1, Model 2 demonstrated a higher accuracy in predicting RFS (c-index, 0.703 and 0.723, respectively), CSS (c-index, 0.694 and 0.710, respectively) and OS (c-index, 0.692 and 0.725, respectively) than Model 1.

Table II. Cox regression analysis of prognostic factors for predicting recurrence-free survival.

Table II.

Cox regression analysis of prognostic factors for predicting recurrence-free survival.

	Univariate		Model 1		Model 2
Variable	P-value	c-index	HR (95% CI)	P-value	HR (95% CI)	P-value
Age, years	0.418	0.542
Sex (female vs. male)	0.411	0.533
pT stage		0.687
1	-		1 (reference)		1 (reference)
2	0.800		1.16 (0.13–10.07)	0.892	1.29 (0.15–11.16)	0.815
3	0.270		2.11 (0.26–17.04)	0.486	2.25 (0.28–18.11)	0.446
4	0.026		7.53 (0.88–64.42)	0.065	5.73 (0.66–50.00)	0.115
pN stage		0.646
0	-		1 (reference)
1	0.159		2.00 (0.64–6.26)	0.235
2	0.004		2.34 (0.81–6.78)	0.116
3	0.009		6.90 (1.39–34.18)	0.018
Lymph node invasion level		0.672
0	-				1 (reference)
1	0.713				0.71 (0.09–5.47)	0.740
2	0.001				2.90 (1.06–7.96)	0.038
3	<0.001				7.25 (1.94–27.15)	0.003

[i] HR, hazard ratio; CI, confidence interval; c-index, Harrell's concordance index.

Table III. Cox regression analysis of prognostic factors for predicting cancer-specific survival.

Table III.

Cox regression analysis of prognostic factors for predicting cancer-specific survival.

	Univariate		Model 1		Model 2
Variable	P-value	c-index	HR (95% CI)	P-value	HR (95% CI)	P-value
Age, years	0.100	0.600
Sex (female vs. male)	0.145	0.565
pT stage		0.675
1	-		1 (reference)		1 (reference)
2	0.815		1.06 (0.12–9.24)	0.958	1.23 (0.14–10.65)	0.850
3	0.423		1.29 (0.15–11.08)	0.816	1.38 (0.16–11.73)	0.769
4	0.042		5.90 (0.66–52.68)	0.112	4.08 (0.44–37.53)	0.215
pN stage		0.665
0	-		1 (reference)
1	0.070		3.01 (0.92–9.88)	0.070
2	0.007		2.35 (0.72–7.66)	0.157
3	0.004		10.81 (2.05–57.08)	0.005
Lymph node invasion level		0.688
0	-				1 (ref.)
1	0.878				0.93 (0.12–7.35)	0.946
2	0.002				3.37 (1.09–10.43)	0.035
3	<0.001				13.04 (3.31–51.37)	<0.001

[i] HR, hazard ratio; CI, confidence interval; c-index, Harrell's concordance index.

Table IV. Cox regression analysis of prognostic factors for predicting overall survival.

Table IV.

Cox regression analysis of prognostic factors for predicting overall survival.

	Univariate		Model 1		Model 2
Variable	P-value	c-index	HR (95% CI)	P-value	HR (95% CI)	P-value
Age, years	0.019	0.610
Sex (female vs. male)	0.110	0.542
pT stage		0.660
1	-		1 (reference)		1 (reference)
2	0.784		1.04 (0.23–4.79)	0.962	1.21 (0.26–5.55)	0.804
3	0.275		1.45 (0.32–6.55)	0.632	1.34 (0.29–6.14)	0.708
4	0.023		3.36 (0.66–17.00)	0.143	2.50 (0.48–12.94)	0.273
pN stage		0.661
0	-		1 (reference)
1	0.071		2.20 (0.87–5.59)	0.100
2	<0.001		3.41 (1.45–8.05)	0.005
3	0.017		5.60 (1.21–25.82)	0.027
Lymph node invasion level
0	-	0.702			1 (reference)
1	0.450				0.47 (0.06–3.54)	0.465
2	<0.001				3.81 (1.60–9.06)	0.003
3	<0.001				14.81 (5.13–42.73)	<0.001

[i] HR, hazard ratio; CI, confidence interval; c-index, Harrell's concordance index.

Discussion

The present study demonstrated that a histological assessment of LN invasion levels in BC better stratified patient outcome using Levels 0, 1, 2 and 3 in comparison with the pN stage of the TNM staging system. In addition, the LN invasion level better predicted RFS, CSS and OS compared with the pN stage, and multivariate analysis revealed that the predictive accuracy of the proposed model was greater than that of the conventional model. Thus, LN invasion levels may provide a better prognostic prediction for patients with BC after radical cystectomy and lymphadenectomy. Furthermore, LN invasion level may be a useful criterion for implementing adjuvant therapy.

Recently, inflammation and nodal state have attracted attention in muscle invasive BC, and new biomarkers have been identified (17). Although the number of positive LNs and the site of metastatic LNs are the standard methods of evaluation in the TNM staging system (3), several LN parameters, such as the number of positive LNs, LN density, the maximum diameter of metastatic LNs and ENE, are also prognostic indicators (4–9). Consistent with previous studies (7), Level 3 corresponding to the ENE was significantly associated with a worse prognosis compared with other levels in RFS, CSS and OS in the present study. In addition to the prognostic impact of ENE the present study demonstrated that LN invasion levels could be used to classify patients into three independent risk groups according to the degree of cancer infiltration in metastatic LNs.

To the best of our knowledge, LN invasion levels have not been previously assessed in BC. Although a similar evaluation method was first proposed for metastatic sentinel nodes in melanoma (11), the definition of Levels 2 and 3 in the present study also integrated the ENE. Notably, the present study demonstrated that LN invasion levels were more significantly associated with prognosis compared with pN stages. By using LN invasion levels, the present study evaluated the metastatic infiltration in their anatomically and immunologically defined substructures. Thus, the results of the present study are consistent with the assessment that the natural route of metastatic spread in LNs reflects a more accurate prognosis compared with the current pN stage (11).

Anatomically, LNs are surrounded by capsules, and multiple lymph lobules are demarcated by lymph-filled sinuses. The superficial cortex is composed of follicles and the interfollicular cortex of all adjacent lobules, and the paracortex is formed by their deep cortical units. The medulla is composed of the paracortex and its medullary cords and sinuses (18). In the first step of LN metastasis, cancer cells enter the LNs through the subcapsular sinus; from there, they invade the cortex to the medulla and break the LN capsule (19). The leakage of tumor cells through holes in the LN capsule into nearby soft tissues results in the ENE (20–22). Due to the metastatic process in LNs, it was reasonable to expect that a higher LN invasion level was significantly associated with poor prognosis.

Univariate analysis revealed that there were no significant differences between invasion levels 0 and 1 (Table II, Table III, Table IV); cancer cells confined to intracapsular lymph vessels and subcapsular or transverse sinuses corresponded to no parenchymal metastasis in LNs. Contrary to the findings in the present study, the cause-specific survival rate, which is defined as the mortality rate owing to bladder cancer in patients with molecularly detected micrometastases, has been reported to be markedly lower than in those without micrometastases, independent of pathologically-positive LNs (23). In contrast, it has been reported that there is no notable difference in survival between the pN micro+ and pN0 groups of patients with immunohistochemically-detected micrometastases in LNs, nor between the pN micro+ and pN+ groups (24,25). The issue is controversial, and larger studies are needed to evaluate the clinical significance of micrometastases and parenchymal metastasis in LNs as the number of patients with Level 1 LN metastases remains too small for a definitive conclusion to be reached.

The present study has several limitations; i) The present study was a retrospective, single-center study with a small sample size. Therefore, validation analysis across multiple institutions is needed; ii) LN invasion levels were assessed in the most highly invasive LNs, not the sentinel LNs, using H&E-stained slides prepared for routine pathology examinations; iii) lymphadenectomy was performed by different surgeons, although a pelvic lymphadenectomy was performed and the mean number of resected LNs was 22.5, which was comparable to the number of dissected lymph nodes reported previously (26,27). If the dissection range is determined in advance and the procedure is performed reliably, there should be little difference between surgeons; iv) no patients treated with NAC were included in the present study. Lymph node status may change when NAC is administered, and the pN+ status after NAC may have a negative impact compared with a pN+ status without NAC (28). Therefore, the present study excluded patients who received NAC to adjust LN status. Del Bene et al (28) reported that NAC reduces pN+ status and improves prognosis; therefore, further investigation in to the association between LN invasion levels and prognosis after NAC treatment is needed; and v) sentinel LNs were not detected in Kansai Medical University Hospital. For a more detailed assessment of sentinel LNs, indocyanine green fluorescence should be considered (29). Despite these limitations, the present study demonstrated the clinical impact of a novel LN assessment in BC considering the natural route of metastatic spread in LNs.

In conclusion, the present study demonstrated the usefulness of the histological assessment of LN invasion levels, considering their anatomical and immunological substructures. The proposed model integrating pT and LN invasion levels accurately predicted the prognosis of patients with BC undergoing radical cystectomy and pelvic lymphadenectomy.

Supplementary Material

Supporting Data

Supporting Data

Acknowledgements

Not applicable.

Funding

Funding: No funding was received.

Availability of data and materials

The data generated in the present study may be requested from the corresponding author.

Authors' contributions

JI, CO and TY designed the present study. JI, CO, TY and TN performed data collection and pathological assessments. JI and CO confirm the authenticity of all the raw data. JI performed the statistical analyses. JI, CO, TY, RS, KT and HK interpreted the data. JI and CO drafted the manuscript. JI, CO, TY, TN, RS, KT and HK critically revised the manuscript for important intellectual content. All authors have read and approved the final manuscript.

Ethics approval and consent to participate

The present study was approved by the Institutional Review Board of Kansai Medical University Hospital (Osaka, Japan; approval no. 2021226). Kansai Medical University Hospital is one of the affiliated medical facilities of Kansai Medical University. Informed consent was obtained as an opt-out on the Kansai Medical University Hospital website (https://hp.kmu.ac.jp/about/research/diagnostic_pathology/). No patients expressed a refusal.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Glossary

Abbreviations

Abbreviations:

References