The significance of lymphovascular invasion in transurethral resection of bladder tumour and cystectomy specimens on the survival of patients with urothelial bladder cancer

Authors


Fadi N. Joudi, Department of Urology, University of Iowa, 200 Hawkins Drive, 3 RCP, Iowa City, IA 52242-1089, USA. e-mail: fadi-joudi@uiowa.edu

Abstract

OBJECTIVE

To test the hypothesis that patients with bladder cancer who had evidence of lymphovascular invasion (LVI) in their transurethral resection of bladder tumour (TURBT) and radical cystectomy (RC) specimens would have a worse prognosis and higher likelihood of clinical understaging, and to assess the effect of LVI discovered at RC on subsequent disease-related mortality, as the prognostic significance of LVI in TURBT or RC specimens of patients treated for urothelial carcinoma of the bladder is not completely established.

PATIENTS AND METHODS

We retrospectively reviewed the records of 163 patients with urothelial carcinoma of the bladder seen at our institution, and who had TURBT (69) or RC (94) between 1995 and 2005. We compared patients with LVI on TURBT and/or RC specimens to a group of controls who did not have LVI on TURBT (34) or RC (32).

RESULTS

Patients with LVI present in their TURBT specimen had a shorter disease-specific survival than those without LVI, with a 5-year survival of 33.6% vs 62.9% (log-rank test P = 0.027; hazard ratio 2.21). LVI at TURBT varied with clinical stage (P = 0.049). Patients with LVI and who were clinical stage I or II had lower survival than those without LVI (P = 0.049; hazard ratio 2.68). LVI did not affect survival among those with clinical stage III or IV (P = 0.29). There was a trend for patients with LVI at TURBT to be clinically understaged compared to those without LVI (75% vs 46%) but the difference was not significant (P = 0.086). Patients with LVI detected in their RC specimen were significantly more likely to have cancer recurrence than were those with no evidence of LVI (48% vs 19%, P = 0.006). For the RC group there was also a significant difference in survival distribution between patients with evidence of LVI vs those without (5-year survival 45.5% vs 78.4%, P = 0.017). Those with LVI were significantly more likely to die from the disease than those without LVI (P = 0.017; hazard ratio 2.92).

CONCLUSIONS

Our findings suggest that LVI is a histological feature that might be associated with a poorer prognosis in patients with urothelial carcinoma of the bladder. The presence of LVI in TURBT specimens predicts shorter survival for patients with stage I or II disease. The presence of LVI in RC specimens predicts recurrence of disease and shorter survival. Further studies are needed to determine whether this group of patients would benefit from early RC and/or perioperative chemotherapy to improve clinical outcomes.

Abbreviations
TURBT

transurethral resection of bladder tumour

LVI

lymphovascular invasion

RC

radical cystectomy

DSS

disease-specific survival

CIS

carcinoma in situ.

INTRODUCTION

Bladder cancer is the fourth most common cancer in men and the eighth most common in women [1]. The prognostic significance of the presence of lymphovascular invasion (LVI) on transurethral resection of bladder tumour (TURBT) and radical cystectomy (RC) specimens is still controversial, with some studies showing an independent prognostic effect [2–6] and others showing no effect [7–10]. Other studies have shown a correlation of LVI with shorter survival that is not independent of pathological stage [11]. In addition, LVI at TURBT has been shown to be associated with understaging and worsened disease-specific survival (DSS) after RC [12].

The purpose of the present study was to assess recurrence and DSS of patients with bladder cancer and evidence of LVI in their TURBT and RC specimens. We also assessed whether evidence of LVI predicted clinical understaging when compared to the pathological stage after RC. We hypothesized that patients who had evidence of LVI in their TURBT or RC specimens would have a poorer prognosis, be more likely to be clinically understaged, and have a higher risk of disease recurrence.

PATIENTS AND METHODS

On Institutional Review Board approval, we retrospectively reviewed the charts of consecutive patients with urothelial carcinoma of the bladder who had TURBT or RC at our institution between 1995 and 2005, and who had evidence of LVI. In all, 163 patients met our inclusion criteria (TURBT group, 69; RC group, 94) which included: (i) evidence of urothelial carcinoma of the bladder; (ii) had TURBT or RC with an available pathology report for review; and (iii) had documentation of the current status of disease. Individuals were excluded if they had evidence of non-urothelial carcinoma histology or evidence of distant metastasis (including regional nodal metastases) at diagnosis. The study population comprised two groups, i.e. patients with evidence of LVI at TURBT or RC. We compared those patients with LVI to a group patients who had TURBT or RC with no evidence of LVI.

We evaluated stage and grade, and other possible prognostic factors, e.g. the presence of carcinoma in situ (CIS), LVI and single vs multifocal disease. Clinical and pathological staging was determined according to the American Joint Committee on Cancer 2002 TNM staging system [13]. Understaging was defined as having clinically organ-confined disease (clinical stage of T1 or T2) and a subsequent pathological stage at RC that indicated extravesical disease (T3 or T4 or N+) for those who subsequently had a RC. Patients were assessed for both local and distant disease recurrence after treatment. Cause of death was determined by reviewing death certificates.

LVI status was determined by reviewing the pathology reports, with LVI defined as the presence of vascular and/or lymphatic invasion. If there was no mention of the LVI in the pathology report, slides were re-examined by a pathologist (D.M.M.) to clarify the status.

The chi-square test was used to compare trends between different clinical variables. Kaplan-Meier analysis was used to determine the 5-year DSS. For multivariate analyses we used a Cox proportional hazards regression model. A two tailed P < 0.05 was taken to indicate statistical significance for all analyses.

RESULTS

Table 1 shows the demographic and clinical characteristics of the TURBT patients; the mean patient age for the LVI group was 64 years, while in the control group with no LVI it was 67 years, with the male-to-female ratio being 4:1 in both groups. Most patients in both groups had a history of smoking (85% of cases and 68% of controls) with a median of 40 pack years.

Table 1.  The demographic and clinical variables of the TURBT and RC groups
VariableTURBTRC
LVI presentNo LVIPLVI presentNo LVIP
  1. n/a, not applicable as not all patients had a RC or information was missing. *Chi-squared test; †Fisher’s exact test; t-test; §Wilcoxon rank-sum test.

No. of patients6934 9432 
n (%)
Gender, male56 (81)27 (79)0.833*74 (79)23 (72)0.427*
Ethnicity, Caucasian59 (86)31 (91)0.53782 (87)30 (94)0.516
Mean (sd, range):
age at diagnosis, years64.1 (11.7, 35–88)67.3 (12.1, 44–87)0.20864.8 (10.1, 35–85)64.0 (11.6, 44–87)0.696
Smoking history, n (%) (yes)58 (85)
68 (1 missing)
23 (68)0.038*75 (80)23 (72)0.352*
Smoking, pack-years66 (3 missing)33 (1 missing) 89 (5 missing)30 (2 missing) 
Median (25–75th %ile)40 (20–60)40 (0–50)0.171§35 (10–50)31.5 (0–50)0.646§
range 0–123 0–75  0–126 0–85 
n (%):
Length of follow-up, years to last clinic visit, median (25–75th %ile) and range    1.79 (0.92–3.31) 1.66 (0.9–3.30)0.966§
 0.04–10.57 0.25–10.16 
Clinical stage      
 1 2 (3) 4 (12)0.038§ 4 (4)1 (3)0.717§
 243 (62)24 (71) 74 (80)27 (84) 
 3 5 (7) 0  6 (6)2 (6) 
 419 (28) 6 (18)  9 (10)2 (6) 
RC pathological stage(47, 22 n/a)(20, 14 n/a)    
 0 5 (11) 6 (30)0.005§ 5 (5)6 (19)0.002§
 1 5 (11) 3 (15)  5 (5)4 (12) 
 2 5 (11) 4 (20)  9 (10)4 (12) 
 310 (21) 4 (20) 24 (26)9 (28) 
 422 (47) 3 (15) 51 (54)9 (28) 
Understaged (stage 1–2)32 (22 missing)13 (11 missing)    
RC pathology, stage (3–4)24 (75) 6 (46)0.086   
CIS (present)   92 (2 missing)  
11 (16) 4 (12) 0.572* 2 (6)14 (15) 0.237*
Single vs multifocal (single)50 (19 missing)24 (10 missing) 53 (41 missing)21 (11 missing) 
39 (78)19 (79)0.909*12 (23)2 (10)0.324*
Chemotherapy      
 None24 (35)17 (50)0.08650 (53)25 (78)0.089
 Neoadjuvant 8 (12)0  9 (10)1 (3) 
 Adjuvant10 (14)3 (9) 29 (31)6 (19) 
 Both 4 (6)0  6 (6)0 
 Not applicable23 (33)14 (41)  
Neoadjuvant or adjuvant, neoadjuvant/adjuvant/both   44 (47)7 (22)0.013
Cystectomy No22 (32)14 (41)0.726   
 Partial 2 (4) 2 (6)    
 Complete42 (61)17 (50)    
 Aborted 2 (3) 1 (3)    
Margin status (+ve)44 (25 n/a)20 (14 n/a) 92 (2 missing)  
 4 (9) 1 (5)1.016 (17) 2 (6)0.153
Recurrence after RC45 (24 n/a)19 (15 n/a)0.33688 (6 missing)31 (1 missing) 
 No26 (58)15 (79) 46 (52)25 (81)0.021*
 Yes, local 5 (11) 1 (5) 12 (14) 2 (6) 
 Yes, distant14 (31) 3 (16) 30 (34) 4 (13) 
 Yes, local/distant   42 (48) 6 (19)0.006*

The median (range) follow-up for those with LVI was 1.4 (0.01–10.5) years and was 1.5 (0.04–10.16) years for the controls. Patients with LVI at TURBT tended to have a higher clinical stage than those without LVI (28% stage 4, vs 18%, P = 0.038) and a higher pathological stage at RC than those without LVI (47% stage 4 vs 15%, P = 0.005). There was no significant difference between the two groups in terms of the presence of CIS, multifocality, or margin status at RC and receipt of adjuvant/neoadjuvant chemotherapy. Of patients with LVI on TURBT, 22% had no subsequent RC due to treatment with intravesical immunotherapy for non-muscle-invasive disease or presentation with advanced disease. For those who had RC there was no statistical difference in disease recurrence (local vs distant), although there were few patients in each group (P = 0.336).

Of the 69 patients with LVI on TURBT, 41 died from their disease; in the control group, nine of the 34 patients died from disease. The Kaplan-Meier survival analysis is shown in Fig. 1A. Patients with LVI had shorter survival than those without LVI, with a 5-year survival of 33.6% vs 62.9% (log-rank test P = 0.027; hazard ratio 2.21). The median survival of patients with LVI was 2.2 years from initial diagnosis.

Figure 1.

The DSS distribution of those with and without LVI among patients who had TURBT (A) or RC (B).

There was also a trend for patients with LVI to be clinically understaged compared to patients without LVI (75% vs 46%) but the difference was not statistically significant (P = 0.086).

Using Cox regression analysis, LVI on TURBT was directly related to increasing clinical stage (P = 0.049). Patients with LVI who were clinical stage I or II had worse survival than same-stage patients without LVI (P = 0.049; hazard ratio 2.68). There was no significant effect of LVI on survival rate among those with clinical stage III or IV (P = 0.29; hazard ratio 0.54).

Particular attention was given to whether or not there was concordance between specimens from those patients who had TURBT and subsequent RC. Forty-two patients had LVI in TURBT specimens, of whom 33 had LVI in their RC specimens, corresponding to a sensitivity of 79% (95% CI 73–90%). Thirty-four patients had no LVI in their TURBT specimens, of whom 21 had no LVI in their RC specimens (specificity of 62%, 44–78%). The negative predictive value of LVI in TURBT specimens was 62%, while the positive predictive value was 72%.

The demographic and clinical variables of both RC groups are also shown in Table 1. The mean (range) patient age at diagnosis for those with LVI was 64.8 (35–85) years, while that of the control group was 64 (44–87) years, and the male-to-female ratio was 3:1 for both groups. Most patients also had a history of smoking (80%) with a median of 35 pack-years for cases and 31.5 pack-years for controls.

The median (range) follow-up for the RC group with LVI was 1.8 (0.04–10.6) years and that for the control group was 1.7 (0.25–10.2) years. There was no significant difference between the RC group with LVI and the controls for the presence of CIS, multifocality or margin status at RC. More patients in the LVI group received chemotherapy (adjuvant, neoadjuvant, or both) than in the control group. Although there was no significant difference in clinical stage between the groups, patients with LVI had a higher pathological stage at RC than the control group.

Patients with LVI present at RC were significantly more likely to have disease recurrence than those with no evidence of LVI (48% vs 19%, P = 0.006). Of the 94 patients with evidence of LVI at RC, 46 (48.9%) died from their disease, compared with five of 32 (16%) in the control group. The Kaplan-Meier survival analysis is shown in Fig. 1B; there was a significant difference in survival distribution between patients with evidence of LVI on RC and those without (5-year survival 45.5% vs 78.4%, P = 0.017). Using Cox proportional hazard regression analysis, after adjusting for the effect of clinical stage and chemotherapy, those with LVI were significantly more likely to die from disease than those without LVI (P = 0.031; hazard ratio 2.82).

DISCUSSION

The present study provides further evidence suggesting that LVI is a pathological feature that might have a role as a prognostic indicator in patients with urothelial carcinoma. There are several studies that have found LVI to be an independent prognostic variable when examining RC specimens [2,5,6]. There is also evidence that LVI leads to a poorer prognosis in patients with upper tract disease [14,15]. In the present study LVI in TURBT and RC specimens was associated with a poor prognosis.

In testicular germ cell carcinoma, the presence of vascular invasion upstages the cancer in the TNM classification system. With accumulating evidence that LVI leads to a poorer prognosis in patients with urothelial carcinoma of the bladder, the TNM staging criteria for testicular germ cell carcinoma might set a precedent for the need to change the staging criteria in bladder cancer. Our results suggest that there is a trend for patients with LVI to be clinically understaged compared with the pathological stage after RC.

There are several implications from this study. First, it is important for pathologists to specifically comment on the LVI status of both TURBT and RC specimens. Using patients from the present study who had a TURBT followed by a subsequent RC, the results showed 79% sensitivity and a negative predictive value of 62% when LVI was found in the TURBT specimen. This supports the validity of finding LVI in TURBT specimens. Pathologists at our institution consistently examine surgical specimens for the presence or absence of LVI. Therefore, particular attention should be focused on determining the status of this histological marker during TURBT. In addition, LVI status should be considered when determining the aggressiveness of treatment for an individual patient. Specifically, the presence of LVI on TURBT might be considered as a high risk for extravesical disease and thereby the patients candidates for neoadjuvant chemotherapy. The presence of LVI in RC specimens might identify patients who would benefit from adjuvant chemotherapy or possibly more aggressive postoperative surveillance. Another implication is that the determination of the LVI status in TURBT specimens is especially important for patients with stage T1 or T2 clinical disease compared to those with T3 or T4. Those with lower-stage disease will derive the most benefit from more aggressive therapy according to LVI status, as those with higher stage would be treated aggressively regardless of LVI status. This is consistent with our findings, in that there was a significant difference in survival for patients with clinical stage T1 and T2 disease who had LVI on TURBT, compared to those without, but there was no difference in patients with stage T3 or T4 cancer.

These results are further evidence that LVI in TURBT and RC pathology specimens predict shorter survival in patients with urothelial carcinoma. There has been no consensus on the clinical management of patients with evidence of LVI, in which additional prospective research is needed.

There are some limitations to the present study, one being that it is retrospective. Another limitation was the short follow-up for some patients. Moreover, several pathologists, who were not dedicated to genitourinary pathology, evaluated the slides over 10 years. There were no standardized criteria to detect LVI among the different pathologists, and interobserver variability, as reported in other studies, could affect the results. False-positive results are possible and, as indicated by other authors, retraction artefact can mimic LVI of urothelial cancer [16,17]. This underscores the need to standardise the detection of LVI, perhaps with immunohistochemical studies such as CD31 or CD34 staining [16].

In conclusion, our study provides further evidence that LVI is an important prognostic factor for survival in patients with urothelial carcinoma of the bladder. The presence of LVI in RC specimens is also associated with an increased risk of disease recurrence. These findings suggest that LVI status detected at TURBT or RC should be considered when a treatment plan is devised for patients with urothelial cancer.

CONFLICT OF INTEREST

None declared.

Ancillary