Combining smoking information and molecular markers improves prognostication in patients with urothelial carcinoma of the bladder

Combining smoking information and molecular markers improves prognostication in patients with urothelial carcinoma of the bladder

Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440 Original article Combining smoking information and molecular markers impro...

172KB Sizes 0 Downloads 48 Views

Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

Original article

Combining smoking information and molecular markers improves prognostication in patients with urothelial carcinoma of the bladder Lily C. Wang, M.D., Ph.D.a,1, Evanguelos Xylinas, M.D., Ph.D.a,c,1, Matthew T. Kent, M.S.d, Luis A. Kluth, M.D.a,e, Michael Rink, M.D.a,e, Asha Jamzadeha, Malte Rieken, M.D.a, Bashir Al Hussein Al Awamlh, M.D.a, Quoc-Dien Trinh, M.D.f, Maxine Sun, Ph.D.f, Pierre I. Karakiewicz, M.D.f, Giacomo Novara, M.D.g, James Chrystala, Marc Zerbib, M.D.c, Douglas S. Scherr, M.D.a, Yair Lotan, M.D.h, Andrew Vickers, Ph.D.d, Shahrokh F. Shariat, M.D., Ph.D.a,b,* a

b

Department of Urology, Weill Cornell Medical College, New York-Presbyterian Hospital, New York, NY Division of Medical Oncology, Weill Cornell Medical College, New York-Presbyterian Hospital, New York, NY c Department of Urology, Cochin Hospital, APHP, Paris Descartes University, Paris, France d Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, New York, NY e Department of Urology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany f Department of Urology, University of Montreal, Montreal, Quebec, Canada g Department of Urology, University of Padua, Padua, Italy h Department of Urology, University of Texas Southwestern Medical Centre, Dallas, TX Received 23 September 2013; accepted 19 October 2013

Abstract Objectives: Tissue-based markers improve the accuracy of prediction models in urothelial carcinoma of the bladder (UCB). Current smoking status and cumulative exposure also affect outcomes. To evaluate whether the combination of molecular markers and smoking features further improved the prognostication of patients who underwent radical cystectomy (RC) for UCB. Materials and methods: A total of 588 patients underwent RC and bilateral lymphadenectomy for UCB from 1995 to 2005. Immunohistochemistry for p53, p21, pRB, p27, Ki-67, and survivin was performed on tissue microarrays from the RC specimen. Smoking features were routinely assessed at diagnosis. Multivariable Cox regression models assessed time to disease recurrence and cancer-specific mortality. Results: Of the 588 patients, 128 were never (22%), 283 former (48%), and 177 current smokers (30%). In total, 227 patients experienced disease recurrence, whereas 190 died of UCB. Smoking status was independently associated with both outcomes (hazard ratio [HR] ¼ 1.48 and 2.62, for former and current vs. never smokers, respectively, P o 0.001). All markers were significantly associated with both outcomes (P o 0.05) except for survivin. The combination of the 4 cell cycle markers p53, p21, pRB, and p27 increased the discrimination of clinicopathologic model for former and current vs. never smokers with c-indices 0.779 and 0.780, respectively (base model c-indices of 0.741 and 0.740 for former and current vs. never smokers, respectively). The further addition of smoking features and biomarker status improved the discrimination of the model (c-indices of 0.783 and 0.786 for former and current vs. never smokers, respectively). Conclusions: We confirmed that smoking information and tissue markers status improve prognostication of UCB outcomes after RC; the combination of both reaching the highest level of discrimination. r 2014 Elsevier Inc. All rights reserved. Keywords: Urothelial carcinoma of the bladder; Radical cystectomy; Prognostic; Survival; Biomarkers; Smoking

1. Introduction Corresponding author. Tel.: þ43-1-40400-2615; fax: þ43-1-404002332. E-mail address: [email protected] (S.F. Shariat). 1 The authors contributed equally to the manuscript. *

1078-1439/$ – see front matter r 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.urolonc.2013.10.015

Radical cystectomy (RC) with lymphadenectomy is the standard of care treatment for patients with muscle-invasive and some patients with high-risk non–muscle invasive urothelial carcinoma of the bladder (UCB) [1–4]. However,

434

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

despite advances in surgical technique and medical oncology options, 10-year cancer-specific survival after RC has stagnated at approximately 50% [1,2]. Standard pathologic features such as tumor stage, grade, and lymph node metastasis are insufficient to discriminate individual clinical behavior and biomedical response to current therapies. Using molecular markers and known clinical risk factors such as smoking status, one may be able to the provide individualized prognostication desired to improve clinical decision making, guiding clinicians, and patients regarding optimal evidence-base management. Others and we have previously reported that alterations of proteins related to the cell cycle (i.e., p53, pRB, p21, and p27), proliferation (i.e., Ki-67) and apoptosis (i.e., survivin) improve the accuracy of algorithms based on standard clinicopathologic features for prediction of outcomes in patients with UCB [5–8]. Similarly, we have recently shown that cumulative smoking exposure and current smoking status predicts clinical outcomes in patients with non–muscle invasive UCB [9,10]. As smoking can lead to molecular alterations associated with UCB [11], we hypothesized that the combination of smoking information can enhance upon the already improved predictive accuracy obtained with established molecular markers. To test this hypothesis, we assessed smoking and molecular markers status in a large multi-institutional cohort of patients treated with RC for UCB. 2. Materials and methods 2.1. Patient population All studies were performed with the approval and oversight of the institutional review board at each institution, with all participating sites providing the necessary data-sharing agreements before initiation. The study cohort was composed of 668 potential patients who underwent RC and bilateral lymphadenectomy for UCB from 1995 to 2005 at 5 participating institutions. The study targeted 588 patients who did not receive neoadjuvant chemotherapy or radiotherapy (n ¼ 80). 2.2. Pathologic evaluation All surgical specimens were processed according to standard pathologic procedures at each institution. Tumors were staged according to the 2002 American Joint Committee on Cancer-Union Internationale Contre le Cancer TNM classification on the RC specimen [12]. Tumor grade was assessed according to 1988 World Health Organization/International Society of Urologic Pathology consensus classification [13]. Pelvic lymph nodes were examined grossly, and all lymphoid tissue was submitted for histologic examination [14]. Lymphovascular invasion

was defined as the unequivocal presence of tumor cells within an endothelium-lined space without underlying muscular walls [15]. Positive soft tissue surgical margin was defined as the presence of tumor at inked areas of soft tissue on the RC specimen [16]. Urethral or ureteral margin status was not considered soft tissue surgical margin in this analysis. 2.3. Immunohistochemistry and scoring Methods for staining and scoring p53, p21, pRB, p27, Ki-67, and survivin were performed as previously described [6,17–21]. Molecular marker staining was considered altered when samples demonstrated Z10% nuclear p53, undetectable or low nuclear p21, undetectable or 450% pRB, o30% nuclear p27, Z20% Ki-67, or 410% survivin staining. Tumors with no pRB expression and those with a strong homogeneous staining pattern were categorized as having altered pRB as overexpression of pRB in bladder cancer may be also indicative of dysfunctional pRB status. 2.4. Smoking information Self-reported smoking data were routinely assessed at the clinical visit before RC. Smoking characteristics analyzed include smoking status (never, former, or current smoker), duration of smoking (r10, 11–20, 21–30, or 430 y), quantity of smoking (1–10, 11–20, 21–30, or 430 cigarettes per day [CPD]), years since cessation (current, o10, Z10), cumulative smoking exposure (light short term r20 CPD for r20 y, heavy short term 420 CPD for r20 y, light long term r20 CPD for 420 y, and heavy long term 420 CPD for 420 y) [10,22–25]. 2.5. Follow-up Follow-up was performed according to institutional protocols. Patients were generally seen postoperatively at least every 3 to 4 months in the first year, every 6 months in the second year, and annually thereafter. Follow-up visits consisted of a physical examination and serum chemistry evaluation. Diagnostic imaging of the upper tract and chest radiography were performed annually or when clinically indicated. Disease recurrence was defined as tumor relapse in the operative field, regional lymph nodes, or distant organs. Cause of death was determined by treating physicians by chart review corroborated by death certificates or by death certificates alone [26]. 2.6. Statistical analysis Multivariable Cox regression models were used to assess the effect of marker and smoking status on time to recurrence and cancer-specific mortality after RC. Discrimination of these models was quantified with the Harrell

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

435

Table 1 Clinicopathologic characteristics of the 588 patients who underwent radical cystectomy for urothelial carcinoma of the bladder. All values are frequency (n, %) or median (interquartile range) Overall (n ¼ 588)

Smoking status Never (n ¼ 128)

Patient age (y)

65 (59, 72)

66 (59, 71)

Gender Male Female

494 (84%) 94 (16%)

Stage pTa–T1 pT2 pT3 pT4

93 168 233 94

Grade Low High

78 (13%) 510 (87%)

Presence of LVI Presence of concomitant Cis STSM Lymph node metastasis Altered Ki-67 Altered p21 Altered p27 Altered p53 Altered pRB Altered survivin

244 245 8 168 266 246 277 266 278 267

(41%) (41.6%) (1.4%) (29%) (45%) (42%) (47%) (45%) (47%) (45%)

56 54 0 41 64 49 58 55 55 52

(44%) (42.2%) (0) (32%) (50%) (38%) (45%) (43%) (43%) (41%)

82 158 176 131 41

(14%) (27%) (30%) (22%) (7%)

15 42 42 25 4

(12%) (33%) (33%) (19%) (3%)

Number of cigarettes (daily) 1–10 11–20 21–30 430

96 146 149 69

(21%) (32%) (32%) (15%)

Smoking duration (y) o20 Z20

110 (86%) 18 (14%)

(16%) (28%) (40%) (16%)

18 38 58 14

(14%) (30%) (45%) (11%)

19 (15%) 109 (85%)

Former (n ¼ 283) 65 (58, 73) 236 (83%) 47 (17%) 44 80 110 49

(16%) (28%) (39%) (17%)

36 (13%) 247 (87%)

65 (59, 72) 148 (84%) 29 (16%) 31 50 65 31

(17%) (28%) (37%) (18%)

23 (13%) 154 (87%)

(42%) (38.7%) (2.2%) (27%) (39%) (40%) (47%) (43%) (46%) (46%)

69 81 2 52 93 83 87 88 93 85

(39%) (45.8%) (1.1%) (29%) (53%) (47%) (49%) (50%) (53%) (48%)

48 73 79 64 19

(17%) (26%) (28%) (22%) (7%)

19 43 55 42 18

(11%) (24%) (31%) (24%) (10%)

– – – –

50 105 87 41

(18%) (37%) (31%) (14%)

46 41 62 28

(26%) (23%) (35%) (16%)

224 (49%) 236 (51%)

– –

141 (50%) 142 (50%)

Smoking cessation (y) o10 Z10

108 (38%) 175 (62%)

– –

108 (38%) 175 (62%)

Cumulative exposure Light short term Light long term Heavy short term Heavy long term

n ¼ 460 115 (25%) 127 (27%) 109 (24%) 109 (24%)

– – – –

No. of altered markers (between p53, p21, pRB, and p27) 0 1 2 3

119 110 6 75 109 114 132 123 130 130

Current (n ¼ 177)

77 78 64 64

(27%) (27%) (23%) (23%)

83 (47%) 94 (53%) – – 38 49 45 45

(21%) (27%) (25%) (25%)

LVI ¼ lymphovascular invasion; Cis ¼ carcinoma in situ; STSM ¼ soft tissue surgical margins.

concordance index. As smoking is associated with increased mortality from health problems other than cancer, competing risk analyses were used to determine the significance of each smoking characteristic after accounting for other cause mortality. Multivariable Cox regression was then performed on the cohort of current and former smokers to examine the

role of each marker status in the setting of smoking history. To prevent overfitting, 10-fold cross-validation was used to obtain concordance indexes. All reported P-values were 2-sided with statistical significance set at 0.05; all statistical tests were performed using Stata 12.0 (StataCorp, College Station, TX).

436

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

3. Results 3.1. Association of smoking status with oncologic outcomes Table 1 summarizes the clinicopathologic characteristics of the patients. Of the 588 patients, 128 were never smokers (22%), 283 former smokers (48%), and 177 current smokers (30%). In total, 227 patients experienced disease recurrence whereas 190 died of UCB and 90 died of other causes. In former smokers, within a median follow-up of 40 months, 103 patients experienced disease recurrence and 84 died of their disease. In current smokers, within a median follow-up of 48 months, 85 patients experienced disease recurrence and 76 died of their disease. In multivariable analyses, smoking status (never, former, and current smokers) was independently associated with disease recurrence and cancer-specific death (HR ¼ 1.48 and 2.62, for former and current smokers vs. never smokers, respectively; both P o 0.001), when included in a model adjusting for the effects of standard clinicopathologic features such as age, tumor stage and grade, presence of lymph node metastasis, and presence of lymphovascular invasion. When performing a competing risk analysis, smoking status remained an independent prognosticator of cancer-specific death (HR ¼ 1.45 and 2.84, for former and current smokers vs. never smokers, respectively; both P o 0.001). As the results of the Cox regression model and the competing risk regression model were similar, competing risk regression was not used to further model cancer-specific death.

Table 2 shows the associations between the different smoking features and disease recurrence or cancer-specific mortality when added to the base model. Smoking duration (P ¼ 0.06 and P ¼ 0.3) and cigarette quantity (P ¼ 0.08 and P ¼ 0.1) were not significantly associated with disease recurrence and cancer-specific mortality. However, both cumulative smoking exposure (both P ¼ 0.01) and smoking cessation time (both P o 0.001) were significantly associated with disease recurrence and cancer-specific mortality. Including cumulative smoking exposure and smoking cessation time in the base model improved discrimination. In all patients (n ¼ 588), adding cumulative smoking exposure to the base model improved its discrimination with cindexes of 0.749 (vs. 0.740) and 0.748 (vs. 0.741) for predicting disease recurrence and cancer-specific mortality, respectively. In former and current smokers (n ¼ 460), the c-index of the base model for disease recurrence was 0.745. The addition of cumulative exposure and cessation time increased the c-indices to 0.762 and 0.769, respectively. Similar results were obtained for cancer-specific mortality (c-indices of 0.744, 0.770, and 0.771, for the base model, base model þ cumulative exposure, and base model þ cessation time, respectively). 3.2. Association of combined altered molecular markers with oncologic outcomes Tables 3 and 4 show the associations of marker alterations with disease recurrence and cancer-specific mortality in former and current smokers when added to the base

Table 2 Multivariable Cox regression models of disease recurrence and cancer-specific mortality for each smoking characteristic in former and current smokers (n ¼ 460), after adjustment for standard clinicopathologic characteristics. c-Indexes are shown for each variable when added to the base model Disease recurrence

Cancer-specific mortality

HR (95% CI)

P-value

c-Index

Smoking duration (y) 1–10 11–20 21–30 430

0.06 Ref. – – –

0.758

Ref. 0.79 (0.49–1.3) 1.4 (0.89–2.2) 1.2 (0.74–1.8)

Cigarette quantity 1–10 11–20 21–30 430

0.08 Ref. – – –

0.757

Ref. 0.98 (0.63–1.5) 1.5 (1.0, 2.2) 1.3 (0.80, 2.1)

Cumulative exposure Light short term Light long term Heavy short term Heavy long term

0.01 Ref. – – –

0.762

Ref. 1.5 (0.96–2.5) 1.6 (0.96–2.5) 2.3 (1.4–3.7)

Cessation time Z10 Y o10 Y Current

o0.001 Ref. – –

0.769

Ref. 2.8 (1.7–4.4) 3.4 (2.2–5.5)

HR (95% CI)

P-value

c-Index

0.3 Ref. – – –

0.766

Ref. 0.87 (0.51–1.5) 1.3 (0.80–2.2) 1.3 (0.78–2.0)

0.1 Ref. – – –

0.765

Ref. 1.0 (0.62–1.6) 1.5 (0.92–2.3) 1.6 (0.94–2.6)

0.01 Ref. – – –

0.770

Ref. 1.3 (0.79–2.2) 1.4 (0.81–2.4) 2.2 (1.3–3.6)

o0.001 Ref. – –

0.771

Ref. 2.3 (1.4–3.9) 3.2 (1.9–5.2)

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

437

Table 3 Multivariable Cox regression models of disease recurrence and cancer-specific mortality for each molecular marker in all patients (n ¼ 588), after adjustment for standard clinicopathologic characteristics. c-Indexes are for each variable when added to the base model Disease recurrence

Ki-67 Survivin p53 p21 pRB p27 Number of altered markers (between p53, p21, pRB, and p27) 0 1 2 3 4

Cancer-specific mortality

HR (95% CI)

P-value

c-Index

HR (95% CI)

P-value

c-Index

1.6 1.3 2.4 1.8 1.4 1.9

0.002 0.03 o0.001 o0.001 0.01 o0.001 o0.001

0.757 0.755 0.772 0.760 0.757 0.763 0.790

1.5 1.4 2.6 1.8 1.6 1.9

0.004 0.03 o0.001 o0.001 0.001 o0.001 o0.001

0.763 0.761 0.781 0.766 0.763 0.768 0.799

(1.2, (1.0, (1.8, (1.4, (1.1, (1.5,

2.0) 1.8) 3.2) 2.4) 1.8) 2.6)

Ref. 3.6 (1.4, 8.6 (3.4, 11 (4.4, 13 (4.9,

9.4) 21) 28) 33)

Ref. – – – –

model. All markers were significantly associated with both outcomes in all patients; however, in former and current smokers, survivin was not associated with outcomes (both P = 0.06). The combination of the 4 cell cycle-related markers p53, p21, pRB, and p27, added to the base model increased its discrimination for both disease recurrence and cancer-specific mortality in all patients (n = 588) (c-indexes 0.779 and 0.780, respectively) as well as in former and current smokers (n = 460) (c-indexes 0.791 and 0.797, respectively). As this study is testing the proof of concept that smoking and molecular markers have a cumulative value in prognosticating outcomes after RC, we did not attempt to identify the optimal marker combination. Thus, for further analyses we focused on the added value of cellcycle related markers.

(1.2, (1.0, (1.9, (1.4, (1.2, (1.4,

2.1) 1.8) 3.6) 2.4) 2.2) 2.5)

Ref. 2.9 (1.0, 8.6 (3.1, 10 (3.6, 14 (5.0,

8.4) 24) 28) 41)

Ref. – – – –

3.3. Association of combined altered molecular markers and smoking with oncologic outcomes in all patients (n = 588) Table 5 shows the final model for all patients (n ¼ 588), including molecular markers and smoking information. The base model including standard clinical and pathologic features had c-indexes of 0.741 and 0.740 for predicting disease recurrence and cancer-specific mortality, respectively. Adding smoking cumulative exposure to the base model improved its discrimination (c-indexes of 0.749 and 0.748 for predicting disease recurrence and cancer-specific mortality, respectively). The full model (Table 5) including cumulative smoking exposure and the number of altered biomarkers exhibited the better discrimination (c-indexes of 0.783 and 0.786 for predicting disease recurrence and cancer-specific

Table 4 Multivariable Cox regression models of disease recurrence and cancer-specific mortality for each molecular marker in former and current smokers (n ¼ 460), after adjustment for standard clinicopathologic characteristics. c-Indexes are for each variable when added to the base model Disease recurrence

Ki-67 Survivin p53 p21 pRB p27 Number of altered markers (among p53, p21, pRB, and p27) 0 1 2 3 4

Cancer-specific mortality

HR (95% CI)

P-value

c-Index

HR (95% CI)

P-value

c-Index

1.5 1.3 2.6 1.8 1.4 1.8

0.01 0.06 o0.001 o0.001 0.04 o0.001 o0.001

0.759 0.760 0.776 0.766 0.760 0.765 0.791

1.6 1.4 2.8 1.8 1.6 1.7

0.01 0.06 o0.001 o0.001 0.007 0.002 o0.001

0.768 0.767 0.787 0.772 0.767 0.770 0.797

(1.1–2.0) (0.98–1.8) (1.9–3.5) (1.4–2.5) (1.0–1.8) (1.3–2.4)

Ref. 6.8 (1.6–29) 15 (3.5–60) 21 (5.1–87) 21 (4.9–90)

Ref. – – – –

(1.2–2.2) (0.99–1.9) (2.0–4.0) (1.3–2.5) (1.1–2.2) (1.2–2.4)

Ref. 8.3 (1.1–62) 23 (3.1–166) 29 (4.0–211) 34 (4.5–249)

Ref. – – – –

438

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

Table 5 Multivariable Cox regression models for disease recurrence and cancer-specific mortality in all patients (n ¼ 588) Recurrence

Disease-specific death

HR

95% CI

P-value

HR

95% CI

P-value

Age (continuous)

1.0

0.9, 1.2

0.6

1.1

1.0, 1.3

0.1

Pathological stage Ta–T1 T2 T3 T4

Ref. 1.1 2.6 4.3

Ref. 0.6, 2.0 1.5, 4.7 2.3, 8.0

o0.001 – – – –

Ref. 1.0 2.3 4.3

Ref. 0.5, 1.9 1.2, 4.4 2.2, 8.4

o0.001 – – – –

Higher tumor grade

1.1

0.7, 1.7

0.6

1.0

0.5, 1.9

0.9

Presence of LVI

1.5

1.1, 2.0

0.01

1.6

1.2, 2.2

0.004

Lymph node metastasis

1.7

1.2, 2.2

o0.001

1.6

1.2, 2.2

0.004

Cumulative exposure Light short term Light long term Heavy short term Heavy long term

Ref. 1.4 1.4 2.0

Ref. 0.9, 2.2 0.9, 2.2 1.3, 3.0

0.02 – – – –

Ref. 1.5 1.6 2.2

Ref. 1.0, 2.4 1.0, 2.7 1.4, 3.5

0.01 – – – –

Number of altered markers 0 1 2 3 4

Ref. 3.8 8.6 11 13

Ref. 1.5, 9.8 3.5, 23 4.5, 28 4.8, 33

o0.001 – – – – –

Ref. 2.9 8.6 9.9 13

Ref. 1.0, 8.5 3.1, 24 3.6, 27 4.4, 37

o0.001 – – – – –

mortality, respectively) even when compared with markers alone (c-indexes of 0.779 and 0.780 for predicting disease recurrence and cancer-specific mortality, respectively). 3.4. Association of combined altered molecular markers and smoking with oncologic outcomes in former and current smokers (n ¼ 460) To better evaluate the added value of combining smoking features and altered markers, we performed the same analyses in former and current smokers (n ¼ 460) (Table 6). Cumulative smoking exposure, time since cessation, and the number of molecular markers present along with the standard clinical predictors were included. Overall, the inclusion of the smoking features and the number of altered markers achieved discrimination of 0.786 for disease recurrence and of 0.763 for cancer-specific mortality. However, this full model achieved a lower discrimination compared with the combination of the 4 cell cycle-related markers alone, which provided the greatest improvement in discrimination with c-indices of 0.791 for disease recurrence and 0.797 for cancer-specific mortality, in this subgroup of former and current smokers. 4. Discussion The combination of smoking features and marker status improved the discrimination of algorithms for prediction of

UCB regarding disease recurrence and cancer-specific mortality after RC. We confirmed that cumulative smoking exposure is significantly associated with disease recurrence and cancer-specific mortality and show for the first time that its addition to a model based on standard clinicopathologic characteristics improved its discrimination from 0.741 to 0.749 and from 0.740 to 0.748 for disease recurrence and cancer-specific mortality, respectively. The biomarkers p53, pRB, p21, p27, and Ki-67 but not survivin were significantly associated with UCB recurrence and mortality and the combination of the 4 cell cycle-related markers—p53, pRB, p21, and p27—improved the discrimination of the model with c-indices of 0.779 and 0.780 for disease recurrence and cancer-specific mortality, respectively. The addition of the markers to the smoking characteristics further improved the model (c-indexes of 0.783 and 0.786, respectively) indicating that together the use of biomarker status and smoking characteristics improve the discrimination of standard prediction models. These results emphasize the fact that smoking features and markers taken alone are not sufficient to help in predicting outcomes in patients with UCB who underwent RC. However, the full model (including smoking features and altered biomarkers) did not exhibit an improved discrimination when compared with altered markers alone in current and former smokers. This study included patients with heterogeneous tumor stage from pTa to pT4 disease (44% of pTa-T2 and 56% of pT3-T4) with and without lymph node metastasis (29% and 71%, respectively). Adjuvant chemotherapy has been

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

439

Table 6 Multivariable Cox regression models for disease recurrence and cancer-specific mortality in current and former smokers (n ¼ 460) Recurrence

Disease-specific death

HR

95% CI

P-value

HR

95% CI

P-value

Age (continuous)

1.01

0.88, 1.2

0.9

1.1

0.93, 1.3

0.3

Pathological stage Ta–T1 T2 T3 T4

Ref. 1.4 3.2 4.4

Ref. 0.69, 3.0 1.6, 6.4 2.1, 9.0

o0.001 – – – –

Ref. 1.2 2.8 4.5

Ref. 0.53, 2.6 1.3, 5.7 2.1, 9.8

o0.001 – – – –

Higher tumor grade

1.2

0.71, 1.9

0.5

0.95

0.55, 1.6

0.8

Presence of LVI

1.5

1.0, 2.0

0.028

1.6

1.1, 2.2

0.011

Lymph node metastasis

2.0

1.4, 2.8

o0.001

1.7

1.2, 2.5

0.002

Cumulative exposure Light short term Light long term Heavy short term Heavy long term

Ref. 1.6 1.6 2.1

Ref. 0.98, 2.5 1.0, 2.7 1.3, 3.3

0.03 – – – –

Ref. 1.3 1.5 1.9

Ref. 0.80, 2.3 0.85, 2.5 1.2, 3.2

0.07 – – – –

Cessation time Z10 y o10 y Current

Ref. 1.8 2.3

Ref. 1.1, 3.0 1.4, 3.8

0.003 – – –

Ref. 1.5 2.0

Ref. 0.88, 2.5 1.2, 3.3

0.02 – – –

Number of altered markers 0 1 2 3 4

Ref. 6.9 13 19 18

Ref. 1.6, 29 3.2, 55 4.5, 79 4.1, 78

o0.001 – – – – –

Ref. 8.2 21 26 27

Ref. 1.1, 62 2.8, 151 3.5, 188 3.6, 206

o0.001 – – – – –

considered for patients at high risk of disease recurrence and cancer-specific mortality (pT3 or greater and any T pNþ disease). Incorporating marker status and smoking information in combination with standard clinical and pathologic features could help further select the population that may better benefit from adjuvant and multimodal therapeutic strategy. This additional information could help clinicians refine clinical decision making and thus spare certain patients from unnecessary adjuvant chemotherapy and its side effects, while strongly recommend other patients who are at high risk of disease recurrence to undergo chemotherapy. We have shown in the past that the combination of molecular markers improved the discrimination of predictive models for UCB recurrence and cancer-specific mortality [5,7]. In addition, we have shown that current smoking status and high cumulative smoking exposure are also associated with disease recurrence and mortality [25]. Here, we show that the combination of marker status and smoking characteristics further improve UCB recurrence and mortality predictive models suggesting that the use of a model incorporating clinicopathologic features, marker status, and smoking information would improve the ability of clinicians to advise patients regarding adjuvant chemotherapy. UCB is a multistep process involving multiple alterations in molecular pathways and is affected significantly by environmental

factors such as cigarette smoking. Combining clinicopathologic features, marker status, and smoking history in a predictive model allows clinicians to take into account multiple aspects of this complex disease. This study has several potential limitations. First of all, this study is retrospective in its nature and therefore subject to the limitations of retrospective data collection, leading to possible bias such as differences in surgical management, pathologic specimens processing, and reading. Another potential limitation is the reliability of immunohistochemistry. We used tissue microarrays, which may lead to sampling effects. Immunohistochemistry is semiquantitative and highly dependent on a range of variables, such as choice of antibody, antibody concentration, fixation techniques, variability in the interpretation and stratification criteria, and inconsistency in specimen handling and technical procedures. We have optimized immunohistochemistry protocols in test tissue microarrays as well as in full-tissue sections of bladder biopsies and transurethral resection of bladder tumor specimen and could confirm that the protocol in the current study is robust and reproducible. Another bias may have resulted from the exclusion of other tobacco products (e.g., cigars, pipes, and tobacco chewing) and different forms of tobacco exposure (e.g., second-hand smoking and occupational exposure). Finally, smoking history was self-reported

440

L.C. Wang et al. / Urologic Oncology: Seminars and Original Investigations 32 (2014) 433–440

and therefore subject to recall bias. Smoking status was not verified by biochemical verification. Patients who did not have smoking data available were excluded from the study, which possibly, but not likely, resulted in a selection bias. Underreporting of tobacco use (i.e., underestimation of the true number of current smokers and overestimation of former smokers) might have biased our data. These limitations clearly indicate that our findings need to be confirmed in robust, prospective studies. 5. Conclusions UCB is a heterogeneous disease with high risk of disease recurrence and cancer-specific mortality despite surgical advances in RC and chemotherapeutic options. We confirmed that smoking information and tissue markers status improve prognostication of UCB outcomes after RC. More importantly, the combination of markers and smoking features reached the highest level of discrimination. References [1] Stein JP, Lieskovsky G, Cote R, Groshen S, Feng AC, Boyd S, et al. Radical cystectomy in the treatment of invasive bladder cancer: longterm results in 1,054 patients. J Clin Oncol 2001;19:666–75. [2] Shariat SF, Karakiewicz PI, Palapattu GS, Lotan Y, Rogers CG, Amiel GE, et al. Outcomes of radical cystectomy for transitional cell carcinoma of the bladder: a contemporary series from the Bladder Cancer Research Consortium. J Urol 2006;176:2414–22[discussion 2422]. [3] Stenzl A, Cowan NC, De Santis M, Kuczyk MA, Merseburger AS, Ribal MJ, et al. Treatment of muscle-invasive and metastatic bladder cancer: update of the EAU guidelines. Eur Urol 2011;59:1009–18. [4] Hautmann RE, de Petriconi RC, Pfeiffer C, Volkmer BG. Radical cystectomy for urothelial carcinoma of the bladder without neoadjuvant or adjuvant therapy: long-term results in 1100 patients. Eur Urol 2012;61:1039–47. [5] Shariat SF, Chade DC, Karakiewicz PI, Ashfaq R, Isbarn H, Fradet Y, et al. Combination of multiple molecular markers can improve prognostication in patients with locally advanced and lymph node positive bladder cancer. J Urol 2010;183:68–75. [6] Shariat SF, Karakiewicz PI, Godoy G, Karam JA, Ashfaq R, Fradet Y, et al. Survivin as a prognostic marker for urothelial carcinoma of the bladder: a multicenter external validation study. Clin Cancer Res 2009;15:7012–9. [7] Shariat SF, Chromecki TF, Cha EK, Karakiewicz PI, Sun M, Fradet Y, et al. Risk stratification of organ confined bladder cancer after radical cystectomy using cell cycle related biomarkers. J Urol 2012;187:457–62. [8] Margulis V, Lotan Y, Karakiewicz PI, Fradet Y, Ashfaq R, Capitanio U, et al. Multi-institutional validation of the predictive value of Ki-67 labeling index in patients with urinary bladder cancer. J Natl Cancer Inst 2009;101:114–9. [9] Rink M, Xylinas E, Babjuk M, Pycha A, Karakiewicz PI, Novara G, et al. Smoking reduces the efficacy of intravesical bacillus Calmette-Guerin immunotherapy in non-muscle-invasive bladder cancer. Eur Urol 2012;62:1204–6.

[10] Rink M, Furberg H, Zabor EC, Xylinas E, Babjuk M, Pycha A, et al. Impact of smoking and smoking cessation on oncologic outcomes in primary non-muscle-invasive bladder cancer. Eur Urol 2012;63:724–32. [11] Burger M, Catto JW, Dalbagni G, Grossman HB, Herr H, Karakiewicz P, et al. Epidemiology and risk factors of urothelial bladder cancer. Eur Urol 2013;63:234–41. [12] Greene FL: AJCC cancer staging manual 6th ed. Greene Frederick L. et al. editors. New York; London: Springer; 2002. [13] Epstein JI, Amin MB, Reuter VR, Mostofi FK. The World Health Organization/International Society of Urological Pathology consensus classification of urothelial (transitional cell) neoplasms of the urinary bladder. Bladder Consensus Conference Committee. Am J Surg Pathol 1998;22:1435–48. [14] Fajkovic H, Cha EK, Jeldres C, Robinson BD, Rink M, Xylinas E, et al. Extranodal extension is a powerful prognostic factor in bladder cancer patients with lymph node metastasis. Eur Urol 2012;64:837–45. [15] Shariat SF, Svatek RS, Tilki D, Skinner E, Karakiewicz PI, Capitanio U, et al. International validation of the prognostic value of lymphovascular invasion in patients treated with radical cystectomy. BJU Int 2010;105:1402–12. [16] Novara G, Svatek RS, Karakiewicz PI, Skinner E, Ficarra V, Fradet Y, et al. Soft tissue surgical margin status is a powerful predictor of outcomes after radical cystectomy: a multicenter study of more than 4,400 patients. J Urol 2010;183:2165–70. [17] Benedict WF, Lerner SP, Zhou J, Shen X, Tokunaga H, Czerniak B. Level of retinoblastoma protein expression correlates with p16 (MTS1/INK4A/CDKN2) status in bladder cancer. Oncogene 1999;18: 1197–203. [18] Shariat SF, Tokunaga H, Zhou J, Kim J, Ayala GE, Benedict WF, et al. p53, p21, pRB, and p16 expression predict clinical outcome in cystectomy with bladder cancer. J Clin Oncol 2004;22:1014–24. [19] Shariat SF, Ashfaq R, Sagalowsky AI, Lotan Y. Correlation of cyclin D1 and E1 expression with bladder cancer presence, invasion, progression, and metastasis. Hum Pathol 2006;37:1568–76. [20] Margulis V, Shariat SF, Ashfaq R, Sagalowsky AI, Lotan Y. Ki-67 is an independent predictor of bladder cancer outcome in patients treated with radical cystectomy for organ-confined disease. Clin Cancer Res 2006;12:7369–73. [21] Shariat SF, Zlotta AR, Ashfaq R, Sagalowsky AI, Lotan Y. Cooperative effect of cell-cycle regulators expression on bladder cancer development and biologic aggressiveness. Mod Pathol 2007; 20:445–59. [22] Rink M, Xylinas E, Babjuk M, Hansen J, Pycha A, Comploj E, et al. Impact of smoking on outcomes of patients with a history of recurrent nonmuscle invasive bladder cancer. J Urol 2012;188:2120–8. [23] Rink M, Xylinas E, Babjuk M, Pycha A, Karakiewicz PI, Novara G, et al. Smoking reduces the efficacy of intravesical Bacillus CalmetteGuerin immunotherapy in non-muscle-invasive bladder cancer. Eur Urol 2012;62:1204–6. [24] Rink M, Xylinas E, Margulis V, Cha EK, Ehdaie B, Raman JD, et al. Impact of smoking on oncologic outcomes of upper tract urothelial carcinoma after radical nephroureterectomy. Eur Urol 2012;63:1082–90. [25] Rink M, Zabor EC, Furberg H, Xylinas E, Ehdaie B, Novara G, et al. Impact of smoking and smoking cessation on outcomes in bladder cancer patients treated with radical cystectomy. Eur Urol 2012; 64:456–64. [26] Rink M, Fajkovic H, Cha EK, Gupta A, Karakiewicz PI, Chun FK, et al. Death certificates are valid for the determination of cause of death in patients with upper and lower tract urothelial carcinoma. Eur Urol 2012;61:854–5.