Pathologic response in patients receiving neoadjuvant chemotherapy for muscle-invasive bladder cancer: Is therapeutic effect owing to chemotherapy or TURBT?

Pathologic response in patients receiving neoadjuvant chemotherapy for muscle-invasive bladder cancer: Is therapeutic effect owing to chemotherapy or TURBT?

Urologic Oncology: Seminars and Original Investigations ] (2016) ∎∎∎–∎∎∎ Original article Pathologic response in patients receiving neoadjuvant chem...

655KB Sizes 0 Downloads 60 Views

Urologic Oncology: Seminars and Original Investigations ] (2016) ∎∎∎–∎∎∎

Original article

Pathologic response in patients receiving neoadjuvant chemotherapy for muscle-invasive bladder cancer: Is therapeutic effect owing to chemotherapy or TURBT? Aaron Brant, B.S.a,*, Max Kates, M.D.a, Meera R. Chappidi, B.A., B.S.a, Hiten D. Patel, M.D., M.P.H.a, Nikolai A. Sopko, M.D., Ph.D.a, George J. Netto, M.D.a, Alex S. Baras, M.D., Ph.D.b, Noah M. Hahn, M.D.b,c, Phillip M. Pierorazio, M.D.a, Trinity J. Bivalacqua, M.D., Ph.D.a a

James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions, Baltimore, MD b Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, MD c Department of Oncology, Johns Hopkins Medical Institutions, Baltimore, MD Received 10 June 2016; received in revised form 20 July 2016; accepted 9 August 2016

Abstract Purpose: We estimated the proportion of patients who received neoadjuvant chemotherapy for muscle-invasive bladder cancer whose tumors were downstaged by transurethral resection. Materials and methods: We identified patients with cT2 N0 urothelial carcinoma who underwent cystectomy at our institution from 2005 to 2014—overall, 139 underwent transurethral resection without chemotherapy, and 146 underwent transurethral resection with chemotherapy. Pathologic response was defined as opT2 N0. We used a Poisson regression model to determine relative risk (RR) of pathologic response in nonneoadjuvant vs. neoadjuvant patients, adjusting for demographic and clinical covariates. This RR was used to estimate the response attributable to transurethral resection. Results: Neoadjuvant patients were younger than nonneoadjuvant patients (64.4 vs. 71.4 years, P o 0.01), with higher median body mass index (28.4 vs. 26.6 kg/m2, P o 0.01), lower prevalence of Charlson score Z3 (13.7% vs. 30.2%, P o 0.01), and lower prevalence of prior non–muscle-invasive cancer (7.5% vs. 20.9%, P o 0.01). More neoadjuvant patients achieved response compared with nonneoadjuvant patients (62.3% vs. 20.1%, RR ¼ 3.10, P o 0.01). Adjustment resulted in a RR of pathologic response in neoadjuvant vs. nonneoadjuvant patients of 2.60 (95% CI: 1.81–3.74, P o 0.01). This adjusted RR indicates that among patients who receive neoadjuvant chemotherapy and undergo transurethral resection, 38% (95% CI: 27%–55%) of pathologic response can be attributed to transurethral resection. Conclusions: We estimate that in a cohort of patients who receive chemotherapy and undergo transurethral resection before cystectomy, 38% of pathologic response can be attributed to transurethral resection. Understanding who responds to chemotherapy and who responds to transurethral resection is needed to measure the effectiveness of both interventions. r 2016 Elsevier Inc. All rights reserved. Keywords: Bladder cancer; Radical cystectomy; TURBT; Neoadjuvant chemotherapy

1. Introduction Survival in patients with muscle-invasive bladder cancer (MIBC) has been improved by the use of neoadjuvant chemotherapy (NAC) before radical cystectomy (RC), Funding: Greenberg Bladder Cancer Institute (Grant no 80035922) Research Funding. * Corresponding author. Tel./fax: þ1-443-413-4779. E-mail address: [email protected] (A. Brant). http://dx.doi.org/10.1016/j.urolonc.2016.08.005 1078-1439/r 2016 Elsevier Inc. All rights reserved.

which provides a 5% to 6.5% absolute benefit in 5-year survival compared with cystectomy alone [1–3]. Because of its effect on survival, guidelines from expert coalitions such as the National Comprehensive Cancer Network (NCCN) recommend cisplatin-based NAC for tumors that are at T2 clinical stage or greater [4]. The survival benefit of NAC is currently attributed to the treatment of microscopic metastatic disease, which is manifested clinically by tumor downstaging [5–7]. A recent

2

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

meta-analysis found that 28.6% of patients who received NAC were downstaged to pT0 [8]. NAC, however, is not the only contributor to pathologic response. A small but significant portion of patients who do not receive NAC would also be downstaged [9,10]. The proposed explanation for pathologic response in these patients is a complete resection of muscle-invasive tumor through transurethral resection of bladder tumor (TURBT). TURBT is routinely performed in patients with suspected or newly diagnosed tumors to determine if muscle-invasion is present. Its efficacy as an organ-sparing monotherapy has also been demonstrated in a selected group of patients with MIBC [11–13]. Because some patients achieve pathologic response from TURBT without NAC, it follows that some patients are downstaged by radical TURBT before receiving chemotherapy. Using a large RC cohort containing patients who received NAC and those who did not receive NAC, we aimed to assess the relative contribution of TURBT to the pathologic responses observed in patients treated with NAC. 2. Methods 2.1. Study Design With institutional review board approval, we conducted a retrospective cohort study in 737 consecutive patients who underwent RC at Johns Hopkins Medical Institutions between January 2005 and December 2014. From this group, we identified 357 patients with biopsy-confirmed clinical stage T2 N0 M0 bladder cancer. Of 357 patients, 21 patients were excluded for having tumors of pure squamous, adenocarcinoma, or small-cell histology, 19 patients were excluded for receiving non–cisplatin-based NAC, 18 patients were excluded for TURBT reports indicating that complete resection was not attempted, 8 patients were excluded for clinical node-positive disease, and 6 patients were excluded because their body mass index (BMI) or pathologic tumor stage was unknown. Our cohort was divided into patients who underwent TURBT without neoadjuvant treatment (non-NAC patients) and those who underwent TURBT with cisplatin-based NAC (NAC patients). 2.2. Study Variables Clinicopathologic variables were collected from patient medical records, patient-completed surveys, the Johns Hopkins Cancer Registry, and the National Death Index. Collected variables include age at cystectomy, sex, race, smoking status, Charlson comorbidity score, BMI, presence of carcinoma in situ (cCIS) and lymphovascular invasion on TURBT specimen, tumor size at TURBT, American Joint Committee on Cancer TNM staging before NAC or RC (clinical stage), American Joint Committee on Cancer TNM

staging at time of RC (pathologic stage), history of prior non–muscle-invasive bladder cancer (NMIBC), days from last TURBT or last NAC dose to cystectomy, institution of TURBT, NAC status, NAC regimen, NAC cycles, NAC doses, recurrence status, survival status, days from cystectomy to recurrence/censoring, and days from cystectomy to death/censoring. Clinical stage was determined by histologic evaluation of TURBT specimens, physical examination, and computerized tomography or magnetic resonance imaging. Prior NMIBC indicates that tumor stage was initially T1, Tis, or Ta and then progressed to T2 on surveillance. Institution of TURBT was considered onsite if final TURBT before RC was performed at our institution and offsite if performed at any other institution. All TURBT specimens were independently reviewed by genitourinary pathologists at our institution. Pathologic stage was determined by histologic evaluation of cystectomy specimen. Beginning in 2010, it became standard practice to consider all patients with muscle-invasive disease for platinum-based NAC. Patients who did not receive NAC after this time were platinum ineligible (i.e., glomerular filtration rate o45 ml/min) or elected not to receive NAC after counseling. To determine if patients received adequate NAC, we used the previously established Johns Hopkins Hospital Dose Index (JHH-DI) [14], which takes number of cycles and dose reductions into account. Based on NCCN guidelines, a JHH-DI Z2 is considered adequate therapy. Follow-up was defined as time from cystectomy to event or last patient contact. The primary outcome measure was pathologic response, defined as pathologic stage T0, Ta, Tis, or T1 and N0. This was subclassified as partial response, defined as pathologic stage T1, Tis, or Ta, and complete response, defined as pathologic stage T0.

2.3. Statistical Analysis We compared patient characteristics using WilcoxonMann-Whitney test for continuous variables and chi-squared test for categorical variables. Comparisons were made between non-NAC and NAC patients to identify baseline characteristics that differed significantly between groups. Comparisons were also made between pathologic responders and nonresponders among non-NAC and NAC patients to identify predictors of pathologic response in each group. We defined probability of response to TURBT as PT and probability of response to NAC as PN. Assuming no interaction between TURBT and NAC, the prevalence of pathologic response in non-NAC patients is equal to PT, whereas prevalence of pathologic response in NAC patients is equal to PT þ PN. If we assume homogeneity between our 2 patient groups, the proportion of pathologic response attributable to TURBT among patients who receive both TURBT and NAC can be determined by dividing the prevalence of patients who respond to TURBT (PT) by

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

the prevalence of patients who respond to either TURBT or NAC (PT þ PN). This equation, PT/(PT þ PN), is equal to the relative risk (RR) of pathologic response in non-NAC vs. NAC patients (prevalence of pathologic response in non-NAC patients divided by the prevalence of pathologic response in NAC patients). However, we assume the population is not homogenous, requiring a modified multivariable Poisson regression with robust error variance to estimate RR of pathologic response with 95% CIs. This approach has been validated in data sets with sample sizes as small as 100 [15] and used in various settings to determine RR [16–18]. Our multivariate regression model adjusted for age, BMI, race, sex, Charlson score, smoking status, history of prior NMIBC, institution of TURBT, cCIS, lymphovascular invasion, cystectomy delay 490 days from TURBT or last NAC dose, and NAC status. All patients underwent TURBT, thus an interaction term between TURBT and NAC was not calculated. KaplanMeier curves with log-rank test were created to compare

3

unadjusted overall survival and recurrence-free survival rates, stratified by NAC status and pathologic response. 3. Results 3.1. Baseline Characteristics Our cohort consisted of 139 (48.8%) non-NAC patients and 146 (51.2%) NAC patients. Non-NAC patients were significantly older than NAC patients (71.4 vs. 64.4 years, P o 0.01), with lower median BMI (26.6 vs. 28.4 kg/m2, P o 0.01), higher frequency of Charlson score Z3 (30.2% vs. 13.7%, P o 0.01), and higher frequency of prior NMIBC (20.9% vs. 7.5%, P o 0.01) (Table 1). 3.2. Pathologic Response The prevalence of pathologic response was significantly lower in non-NAC patients compared with NAC patients

Table 1 Baseline characteristics of the entire cohort, non-NAC patients, and NAC patients Variable

All patients

Non-NAC

NAC

P value

Age at cystectomy BMI Sex Male Female

67.1 (40.1–86.6) 27.3 (17.6–44.6)

71.4 (40.1–86.6) 26.6 (19.5–42.0)

64.4 (41.9–84.0) 28.4 (17.6–44.6)

o0.0001* 0.0057* 0.8714

243 (85.3) 42 (14.7)

119 (85.6) 20 (14.4)

124 (84.9) 22 (15.1)

Race White Non-White

263 (92.3) 22 (7.7)

129 (92.8) 10 (7.2)

134 (91.8) 12 (8.2)

Somkigs status Current somker Former somker Never somker

55(19.3) 144 (50.5) 86 (30.2)

24 (17.3) 77 (55.4) 38 (27.3)

31 (21.2) 67 (45.9) 48 (32.9)

Charlson range o3 Z3

223 (78.3) 62 (21.8)

97 (69.8) 42 (30.2)

126 (86.3) 20 (13.7)

Prior NMIBC Yes No

40 (14.0) 245 (86.0)

29 (20.9) 110 (79.1)

11 (7.5) 135 (92.5)

cCIS Yes No

75 (26.3) 210 (73.7)

33 (23.7) 106 (76.3)

42 (28.8) 104 (71.2)

cLVI Yes No

25 (8.8) 260 (91.2)

8 (5.8) 131 (94.2)

17 (11.6) 129 (88.4)

Institution of TURBT Onsite Offsite

78 (27.4) 207 (72.6)

38 (27.3) 101 (72.7)

40 (27.4) 106 (72.6)

Days from TURBT/NAC to cystectomy 490 r90

69 (24.2) 216 (75.8)

35 (25.2) 104 (74.8)

34 (23.3) 112 (76.7)

*Indicates a p-value o0.05

0.7459

0.2756

0.0007*

0.0012*

0.3355

0.0790

0.9911

0.7093

4

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

Because both partial and complete responders demonstrated improved recurrence-free and overall survival rates, the remaining analyses were conducted evaluating pathologic response as a whole. 3.3. Regression Model

Fig. 1. Prevalence of pathologic response in non-NAC patients vs. NAC patients. (Color version of figure is available online.)

(20.1% vs. 62.3%, P o 0.01) (Fig. 1). Among non-NAC patients, 13 (9.4%) achieved complete response, 15 (10.8%) achieved partial response, and 111 (79.9%) did not respond. Of the 111 patients who did not respond, 70 (63.1%) progressed to T3 or T4. Among NAC patients, 41 (28.1%) achieved complete response, 50 (34.2%) achieved partial response, and 55 (37.7%) did not respond. Of the 55 patients who did not respond, 35 (63.6%) progressed to T3 or T4. When looking at both non-NAC and NAC patients together, there was no significant difference in overall survival between partial and complete responders (P ¼ 0.67) (Fig. 2A). Additionally, partial and complete responders each experienced improved survival compared with nonresponders (P o 0.01 and P o 0.01). Likewise, there was no significant difference in recurrence-free survival between partial and complete responders (P ¼ 0.65) (Fig. 2B), and partial and complete responders each experienced improved recurrence-free survival when compared with nonresponders (P o 0.01 and P o 0.01).

Log-rank: No response vs. complete response p<.0001 No response vs. partial response p<.0001 Complete response vs. partial response p=0.6718

A

Among non-NAC patients, history of NMIBC before muscle-invasive diagnosis was significantly less prevalent in patients who achieved pathologic response compared with those who did not respond (3.6% vs. 25.2%, P ¼ 0.01) (Table 2). Among NAC patients, the prevalence of cCIS was greater in patients who achieved pathologic response than in nonresponders (37.4% vs. 14.5%, P o 0.01), as was the prevalence of TURBTs performed at outside institutions (80.2% vs. 60.0%, P o 0.01), the prevalence of patients who received cystectomy r90 days from last NAC dose (82.4% vs. 67.3%, P ¼ 0.04), and the prevalence of patients who received adequate NAC, defined as JHH-DI Z 2 (83.4% vs. 70.0%, P o 0.01). The difference in cCIS prevalence was no longer significant when looking at complete responders vs. partial and nonresponders (26.8% vs. 29.5%, P ¼ 0.75). Median age at cystectomy was significantly lower (62.3 vs. 67.0 y, P o 0.02) and median BMI was significantly greater (28.9 vs. 26.6 kg/m2, P o 0.01) in NAC patients who achieved pathologic response compared with those who did not respond. These differences in age and BMI were no longer significant when looking only at patients who received adequate NAC (62.7 vs. 65.9 years, P ¼ 0.25 and 28.8 vs. 26.9 kg/m2, P ¼ 0.13). Complete information was not present on tumor size at time of TURBT, but we conducted a subanalysis on the 139 patients with available information, in which no association between tumor size and pathologic response was found. Among non-NAC patients, the prevalence of tumor

Log-rank: No response vs. complete response p<.0001 No response vs. partial response p<.0001 Complete response vs. partial response p=0.6463

B

Fig. 2. Kaplan-Meier curves with log-rank test for overall survival (A) and recurrence (B) stratified by degree of pathologic response. (Color version of figure is available online.)

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

5

Table 2 Baseline characteristics stratified by NAC status and pathologic response Variable

Age at cystectomy BMI Sex Male Female

Nonneoadjuvant

Neoadjuvant

Downstage

No Downstage

P value

Downstage

No downstage

P value

67.3 (40.1–84.1) 26.2 (20.2–32.2)

71.6 (49.4–86.6) 26.9 (19.5–42.0)

0.2942 0.6058 0.9862

62.3 (41.9–81.8) 28.9 (19.5–44.6)

67.0 (42.1–84.0) 26.6 (17.6–43.3)

0.0177* 0.0106* 0.8908

24 (85.7) 4 (14.3)

95 (85.6) 16 (14.4)

Race White Non-White

27 (96.4) 1 (3.6)

102 (91.9) 9 (8.1)

Smoking status Current smoker Former smoker Never smoker

4 (14.3) 16 (57.1) 8 (28.6)

20 (18.0) 61 (55.0) 30 (27.0)

Charlson score o3 Z3

20 (71.4) 8 (28.6)

77 (69.4) 34 (30.6)

Prior NMIBC Yes No

1 (3.6) 27 (96.4)

28 (25.2) 83 (74.8)

cCIS Yes No

5 (17.9) 23 (82.1)

28 (25.2) 83 (74.8)

cLVI Yes No

0 (0.0) 28 (100.0)

8 (7.2) 103 (92.8)

Institution of TURBT Onsite Offsite

5 (17.9) 23 (82.1)

33 (29.7) 78 (70.3)

Days from TURBT/NAC to cystectomy r90 490

23 (82.1) 5 (17.9)

81 (73.0) 30 (27.0)

JHH-DI o2 Z2

77 (84.6) 14 (15.4)

47 (85.5) 8 (14.5)

84 (92.3) 7 (7.7)

50 (90.9) 5 (9.1)

24 (26.4) 41 (45.1) 26 (28.6)

7 (12.7) 26 (47.3) 22 (40.0)

79 (86.8) 12 (13.2)

47 (85.5) 8 (14.6)

5 (5.5) 86 (94.5)

6 (10.9) 49 (89.1)

34 (37.4) 57 (62.6)

8 (14.5) 47 (85.5)

10 (11.0) 81 (89.0)

7 (12.7) 48 (87.3)

18 (19.8) 73 (80.2)

22 (40.0) 33 (60.0)

75 (82.4) 16 (17.6)

37 (67.3) 18 (32.7)

10 (11.6) 76 (83.4)

15 (30.0) 35 (70.0)

0.4064

0.7656

0.8962

0.1105

0.8321

0.817

0.0117*

0.2297

0.0032*

0.4129

0.1434

0.751

0.0079*

0.2078

0.0359*

0.3178

0.0077* NA NA

NA NA

cLVI ¼ lymphovascular invasion.

size 45 cm was 50.0% in patients who achieved pathologic response compared with 46.4% of patients who did not respond (P ¼ 0.97). Among NAC patients, the prevalence of tumor size 45 cm was 52.2% in patients who achieved pathologic response when compared with 55.6% of patients who did not respond (P ¼ 0.94). The results of our univariate and multivariate analyses are listed in Table 3. With multivariate regression, higher age at cystectomy (RR ¼ 0.98; 95% CI: 0.98–1.00; P ¼ 0.02), onsite TURBT (RR ¼ 0.66; 95% CI: 0.47–0.92; P ¼ 0.01), and cystectomy delay 490 days from TURBT or NAC (RR ¼ 0.66; 95% CI: 0.47–0.94; P ¼ 0.02) were significant predictors of the absence of pathologic response.

The adjusted RR of pathologic response in non-NAC vs. NAC patients was 0.38 (95% CI: 0.270.55, P o 0.01). This indicates that in a cohort of NAC patients with cT2 disease, 38% (95% CI: 27%–55%) of pathologic response can be attributed to TURBT. 3.4. Survival and Recurrence Overall survival was significantly worse in non-NAC patients compared with NAC patients (P ¼ 0.03) (Fig. 3A). Both non-NAC and NAC patients who achieved pathologic response showed improved survival when compared with those who did not respond (P o 0.01) (Fig. 3B).

6

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

Table 3 Results from univariate and multivariate Poisson regression with robust error variance for relative risk of pathologic response Variable

Univariate regression Relative risk

Multivariate regression 95% CI

P value

0.97 1.04

0.96–0.98 1.01–1.07

o0.0001 0.0038*

Ref 1.03

0.71–1.51

Race White Nonwhite

Ref 0.84

Smoking status Never smoker Ever smoker

Ref 1.08

Age at cystectomy, Y BMI, kg/m2 Sex Male Female

Charlson score o3 Ref Z3 0.73 Prior NMIBC No Ref Yes 0.33 cCIS No Ref Yes 1.37 cLVI No Ref Yes 0.95 Institution of TURBT Offsite Ref Onsite 0.64 Days from TURBT/NAC to cystectomy r90 Ref 490 0.67 NAC No 0.32 Yes Ref

Relative risk

95% CI

0.98 1.02

0.98–1.00 0.99–1.04

0.0220* 0.1539

0.8743

Ref 1.12

0.80–1.59

0.4818

0.59–1.17

0.3006

Ref 0.76

0.46–1.15

0.2850

0.79–1.47

0.6214

Ref 1.28

0.96–1.70

0.0877

0.49–1.07

0.1080

Ref 0.99

0.70–1.41

0.9758

0.15–0.69

0.0033*

Ref 0.53

0.27–1.06

0.0731

1.03–1.80

0.0280

Ref 1.23

0.97–1.57

0.0886

0.58–1.58

0.9105

Ref 0.77

0.52–1.16

0.2194

0.44–0.92

0.0174*

Ref 0.66

0.47–0.92

0.0134*

0.46–0.99

0.0424*

Ref 0.66

0.47–0.94

0.0221*

0.23–0.46

o0.0001*

0.27–0.55

o0.0001*

There was no significant difference in overall survival between non-NAC and NAC patients who achieved pathologic response (P ¼ 0.98) and between non-NAC and NAC patients who did not achieve pathologic response (P ¼ 0.35). The difference in recurrence-free survival between nonNAC and NAC patients did not meet traditional levels of significance (P ¼ 0.23, Fig. 4A). Both Non-NAC and NAC patients who achieved pathologic response showed improved recurrence-free survival compared with nonNAC and NAC patients who did not respond (P ¼ 0.02 and P o 0.01). There was no significant difference in recurrence-free survival between non-NAC and NAC patients who achieved pathologic response (P ¼ 0.23). However, recurrence-free survival was significantly improved in non-NAC patients who did not achieve pathologic response compared with NAC patients who did not respond (P ¼ 0.01).

*

0.38 Ref

P value

4. Discussion NAC provides a survival benefit to patients who achieve pathologic response. However, a significant number of patients are downstaged by radical TURBT alone. To determine the relative contribution of TURBT to pathologic response among NAC patients, we used a Poisson regression model with robust error variance. Our model found that the RR of pathologic response in non-NAC vs. NAC patients was 0.38. This RR indicates that among patients who receive NAC and underwent TURBT, 38% of pathologic response is attributable to TURBT. Several randomized controlled trials have demonstrated the efficacy of NAC in patients with MIBC. The landmark trial by the Southwest Oncology Group found that 15% of non-NAC patients achieved complete response when compared with 38% of NAC patients [9]. Assuming that

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

7

Log-rank: NAC responder vs. non-NAC responder p=0.9779 NAC non-responder vs. non-NAC non-responder p=0.3506 NAC responder vs. NAC non-responder p<0.0001 Non-NAC responder vs. non-NAC non-responder p=0.0005

Log-rank: NAC vs. Non-NAC p=0.0306

A

B

Fig. 3. Kaplan-Meier curves with log-rank test for overall survival comparing NAC vs. non-NAC patients (A) and comparing NAC vs. non-NAC patients stratified by pathologic response (B). (Color version of figure is available online.)

the non-NAC and NAC patients were well matched, the percentage of complete response attributable to TURBT can be calculated without adjustment: 0.15/0.38 ¼ 0.39 or 39%. This result is similar to the 38% determined by our regression model, though their study included patients with clinical stages T3 and T4, looked at only methotrexate, vinblastine, doxorubicin, and cisplatin chemotherapy, and used complete response as the primary outcome. If we had included patients with clinical stages T3 and T4, we anticipate that the percentage of NAC patients downstaged by TURBT would be lower, as pathologic response to TURBT is less likely in extravesicular disease. Several other studies have examined the efficacy of NAC in patients with MIBC. A recent multicenter assessment of cisplatin and non–cisplatin-based regimens in patients with cT2 disease reported that 24.7% of patients who received

NAC achieved complete response and 19% achieved partial response [19]. This is somewhat lower than our result of 27.0% of NAC patients who achieved complete response and 34.2% who achieved partial response, which may be explained by differences in physician practice, patient demographics, and mix of chemotherapy regimens. Along with our study, this study found that both partial and complete pathologic response were predictive of improved survival [19], which has been reported in other cohorts as well [6,7,20]. Patients who achieve pathologic response after TURBT also experience improved survival [21,22]. A study of patients with muscle-invasive disease who received TURBT without NAC reported that 7.2% of patients achieved complete response and 14.6% achieved partial response [23]. This is similar to our finding of 9.3% of non-NAC

Log-rank: NAC responder vs. non-NAC responder p=0.2330 NAC non-responder vs. non-NAC non-responder p=0.0140 NAC responder vs. NAC non-responder p<0.0001 Non-NAC responder vs. non-NAC non-responder p=0.0195

Log-rank: NAC vs. Non-NAC p=0.2324

A

B

Fig. 4. Kaplan-Meier curves with log-rank test for overall recurrence comparing NAC vs. non-NAC patients (A) and comparing NAC vs. non-NAC patients stratified by pathologic response (B). (Color version of figure is available online.)

8

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9

patients who achieved complete response and 10.6% who achieved partial response. We also found no survival difference between non-NAC and NAC patients who achieved pathologic response, which has been corroborated by other studies [5,19,24]. This finding suggests that patients who are truly downstaged by TURBT may not require NAC. If this is the case, accurate predictors of response to TURBT could improve clinical decision making. We identified the absence of prior NMBIC as a potential predictor of response to TURBT. This is especially compelling considering there may a bias for patients with prior NMIBC to have smaller tumors, given that they are under closer surveillance than patients with nascent muscleinvasive disease. The discovery of more predictors would likely require larger datasets and more precise identification of molecular subtyping of the naïve urothelial cancers to identify those tumors that have indolent biological behaviors. NAC is the standard of care for MIBC because of its positive effect on pathologic response and survival. We found that patients who received NAC were 2.63 times more likely to achieve pathologic response and that this conferred improved survival. However, we also found that NAC patients who were not downstaged experienced worse survival than non-NAC patients who were not downstaged, which demonstrates the need to identify predictors of pathologic response to NAC. In an optimal setting, patients with profiles predictive of NAC response and TURBT nonresponse would receive aggressive NAC, whereas those with profiles predictive of NAC nonresponse and TURBT response would proceed straight to surgery. Thus far, studies have identified ATM, RB1, FANCC, ERBB2, GDPD3, and SPRED1 as genetic loci involved in predicting NAC response [25–27]. Although their results are promising, these studies did not account for patients downstaged by TURBT rather than NAC. To accurately determine who would benefit most from chemotherapy, researchers must take patients downstaged by TURBT into account. Our study's limitations include its retrospective nature and the fact that all information was collected from a single institution with unique referral and practice patterns. We were not able to obtain complete information on tumor size, which is a potential predictor of pathologic response. Additionally, some TURBTs were performed at outside institutions with physicians who differ in their approach to resection. However, we accounted for offsite TURBTs in our regression model, excluded patients for whom complete TURBTs were not attempted, and genitourinary pathologists at our institution reviewed all TURBT specimens to verify the patient's stage. Another potential limitation is the assumption that TURBT and NAC do not influence each other in any way. Because all patients received TURBT, we could not add an interaction term between TURBT and NAC into our model. Although not substantiated by any published reports, a potential for TURBT to render the primary tumor either more or less susceptible

to NAC is conceivable. To confirm the validity of our results, further investigation in a multi-institutional setting is warranted.

5. Conclusions In prior research and in clinical practice, chemotherapy is credited as the curative treatment in patients who receive NAC and subsequently experience pathologic response. Our study indicates that in a cohort of patients who undergo and receive both TURBT and NAC, a substantial portion of the pathological responses observed can be attributed to TURBT. This finding demonstrates the need to accurately classify responders of TURBT and NAC. Identifying patients who are considered responders to TURBT with comparable long-term recurrence-free and overall survival would allow physicians to reevaluate the decision to initiate NAC. Identifying the true responders to NAC would allow researchers to more accurately measure chemotherapy effectiveness, thus enhancing their ability to identify predictors of chemotherapy response. Acknowledgments The authors would like to acknowledge Jiangxia Wang, Department of Biostatistics, Bloomberg School of Public Health. References [1] Advanced Bladder Cancer Meta-analysis Collaboration. Neoadjuvant chemotherapy in invasive bladder cancer: a systematic review and meta-analysis. Lancet 2003;361:1927. [2] Advanced Bladder Cancer (ABC) Meta-analysis Collaboration. Neoadjuvant chemotherapy in invasive bladder cancer: update of a systematic review and meta-analysis of individual patient data advanced bladder cancer (ABC) meta-analysis collaboration. Eur Urol 2005;48:202. [3] Advanced Bladder Cancer Overview Collaboration. Neoadjuvant chemotherapy for invasive bladder cancer. Cochrane Database Syst Rev 2005:CD005246. [4] Clark PE, Agarwal N, Biagioli MC, et al. Bladder cancer. J Natl Compr Canc Netw 2013;11:446. [5] Lavery HJ, Stensland KD, Niegisch G, et al. Pathological T0 following radical cystectomy with or without neoadjuvant chemotherapy: a useful surrogate. J Urol 2014;191:898. [6] Rosenblatt R, Sherif A, Rintala E, et al. Pathologic downstaging is a surrogate marker for efficacy and increased survival following neoadjuvant chemotherapy and radical cystectomy for muscle-invasive urothelial bladder cancer. Eur Urol 2012;61:1229. [7] Sonpavde G, Goldman BH, Speights VO, et al. Quality of pathologic response and surgery correlate with survival for patients with completely resected bladder cancer after neoadjuvant chemotherapy. Cancer 2009;115:4104. [8] Petrelli F, Coinu A, Cabiddu M, et al. Correlation of pathologic complete response with survival after neoadjuvant chemotherapy in bladder cancer treated with cystectomy: a meta-analysis. Eur Urol 2014;65:350.

A. Brant et al. / Urologic Oncology: Seminars and Original Investigations ] (2016) 1–9 [9] Grossman HB, Natale RB, Tangen CM, et al. Neoadjuvant chemotherapy plus cystectomy compared with cystectomy alone for locally advanced bladder cancer. N Engl J Med 2003;349:859. [10] Beukers W, Meijer T, Vissers CJ, et al. Down-staging (opT2) of urothelial cancer at cystectomy after the diagnosis of detrusor muscle invasion (pT2) at diagnostic transurethral resection (TUR): is prediction possible? Virchows Arch 2012;461:149. [11] Solsona E, Iborra I, Collado A, et al. Feasibility of radical transurethral resection as monotherapy for selected patients with muscle invasive bladder cancer. J Urol 2010;184:475. [12] Leibovici D, Kassouf W, Pisters LL, et al. Organ preservation for muscle-invasive bladder cancer by transurethral resection. Urology 2007;70:473. [13] Herr HW. Transurethral resection of muscle-invasive bladder cancer: 10-year outcome. J Clin Oncol 2001;19:89. [14] Gandhi NM, Baras A, Munari E, et al. Gemcitabine and cisplatin neoadjuvant chemotherapy for muscle-invasive urothelial carcinoma: Predicting response and assessing outcomes. Urol Oncol 2015;33:204.e1. [15] Zou G. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004;159:702. [16] Kozuki N, Katz J, Christian P, et al. Comparison of US Birth Weight References and the International Fetal and Newborn Growth Consortium for the 21st Century Standard. JAMA Pediatr 2015;169:e151438. [17] Fedirko V, Bradshaw PT, Figueiredo JC, et al. Urinary Metabolites of Prostanoids and Risk of Recurrent Colorectal Adenomas in the Aspirin/Folate Polyp Prevention Study (AFPPS). Cancer Prev Res (Phila) 2015;8:1061. [18] Kirwa K, Eliot MN, Wang Y, et al. Residential proximity to major roadways and prevalent hypertension among postmenopausal women: results from the Women0 s Health Initiative San Diego Cohort. J Am Heart Assoc 2014;3:e000727.

9

[19] Zargar H, Espiritu PN, Fairey AS, et al. Multicenter assessment of neoadjuvant chemotherapy for muscle-invasive bladder cancer. Eur Urol 2015;67:241. [20] Splinter TA, Scher HI, Denis L, et al. The prognostic value of the pathological response to combination chemotherapy before cystectomy in patients with invasive bladder cancer. European Organization for Research on Treatment of Cancer—Genitourinary Group. J Urol 1992;147:606. [21] Nielsen ME, Bastian PJ, Palapattu GS, et al. Recurrence-free survival after radical cystectomy of patients downstaged by transurethral resection. Urology 2007;70:1091. [22] Tollefson MK, Boorjian SA, Farmer SA, et al. Downstaging to noninvasive urothelial carcinoma is associated with improved outcome following radical cystectomy for patients with cT2 disease. World J Urol 2012;30:795. [23] Shariat SF, Palapattu GS, Karakiewicz PI, et al. Discrepancy between clinical and pathologic stage: impact on prognosis after radical cystectomy. Eur Urol 2007;51:137. [24] Volkmer BG, Kuefer R, Bartsch G Jr., et al. Effect of a pT0 cystectomy specimen without neoadjuvant therapy on survival. Cancer 2005;104:2384. [25] Plimack ER, Dunbrack RL, Brennan TA, et al. Defects in DNA repair genes predict response to neoadjuvant cisplatin-based chemotherapy in muscle-invasive bladder cancer. Eur Urol 2015;68:959–957. [26] Groenendijk FH, de Jong J, Fransen van de Putte EE, et al. ERBB2 mutations characterize a subgroup of muscle-invasive bladder cancers with excellent response to neoadjuvant chemotherapy. Eur Urol 2015;69:384–8. [27] Baras AS, Gandhi N, Munari E, et al. Identification and validation of protein biomarkers of response to neoadjuvant platinum chemotherapy in muscle invasive urothelial carcinoma. PLoS One 2015;10:e0131245.