Impact of the Number of Prior Lines of Therapy and Prior Perioperative Chemotherapy in Patients Receiving Salvage Therapy for Advanced Urothelial Carcinoma: Implications for Trial Design

Accepted Manuscript Impact of number of prior lines of therapy and prior perioperative chemotherapy in patients receiving salvage therapy for advanced urothelial carcinoma: implications for trial design G.R. Pond , J. Bellmunt , J.E. Rosenberg , D.F. Bajorin , A.M. Regazzi , T.K. Choueiri , A.Q. Qu , G. Niegisch , P. Albers , A. Necchi , G. Di Lorenzo , R. Fougeray , Y.-N. Wong , S.S. Sridhar , Y.-J. Ko , M.I. Milowsky , M.D. Galsky , G. Sonpavde PII:

S1558-7673(14)00105-0

DOI:

10.1016/j.clgc.2014.06.004

Reference:

CLGC 275

To appear in:

Clinical Genitourinary Cancer

Received Date: 25 March 2014 Revised Date:

29 April 2014

Accepted Date: 3 June 2014

Please cite this article as: Pond G, Bellmunt J, Rosenberg J, Bajorin D, Regazzi A, Choueiri T, Qu A, Niegisch G, Albers P, Necchi A, Di Lorenzo G, Fougeray R, Wong Y-N, Sridhar S, Ko Y-J, Milowsky M, Galsky M, Sonpavde G, Impact of number of prior lines of therapy and prior perioperative chemotherapy in patients receiving salvage therapy for advanced urothelial carcinoma: implications for trial design, Clinical Genitourinary Cancer (2014), doi: 10.1016/j.clgc.2014.06.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Impact of number of prior lines of therapy and prior perioperative chemotherapy in patients receiving salvage therapy for advanced urothelial carcinoma: implications for trial design

1

2

3

3

3

4

4

5

1

RI PT

Pond GR , Bellmunt J , Rosenberg JE , Bajorin DF , Regazzi AM , Choueiri TK , Qu AQ , Niegisch G , 5 6 7 8 9 10 11 12 Albers P , Necchi A , Di Lorenzo G , Fougeray R , Wong Y-N , Sridhar SS , Ko Y-J , Milowsky MI , 13 14 Galsky MD , Sonpavde G 2

3

TE D

Corresponding Author Guru Sonpavde, MD UAB Comprehensive Cancer Center, 1802 6th Ave. S., NP2540B Birmingham, AL 35294 E-mail: [email protected] Phone: 205-975-2914, Fax: 205-975-3910 Total number of words in abstract: 244 Total number of words in text: 2466 Number of references: 28

EP

Number of tables plus figures: 6

AC C

Relevant disclosures: None

M AN U

Short title Salvage therapy for advanced urothelial carcinoma

SC

McMaster University, Ontario, Canada; University Hospital del Mar-IMIM, Barcelona, Spain; Memorial 4 Sloan Kettering Cancer Center, New York, NY; Dana-Farber Cancer Institute and Harvard Medical 5 6 School, Boston, MA; Heinrich Heine University, Dusseldorf, Germany; Fondazione IRCCS Istituto 7 8 Nazionale dei Tumori, Milan, Italy; University Federico II, Naples, Italy; Institut de Recherche Pierre 9 10 Fabre, Boulogne, France; Fox Chase Cancer Center, Philadelphia, PA; Princess Margaret Hospital, 11 12 Toronto, ON, Canada; Sunnybrook Odette Cancer Centre, Toronto, ON, Canada; University of North 13 Carolina, Chapel Hill, NC; Tisch Cancer Center Institute, Mount Sinai School of Medicine, New York, 14 NY; UAB Comprehensive Cancer Center, Birmingham, AL

ACCEPTED MANUSCRIPT 2

Microabstract This analysis of patients from prospective trials of salvage therapy for advanced urothelial carcinoma did not identify a prognostic for number of prior lines of therapy and prior perioperative chemotherapy.

RI PT

Performance status, hemoglobin, liver metastasis and time from prior chemotherapy were prognostic for outcomes. These findings allow trials to employ uniform eligibility criteria, which will enhance accrual and

AC C

EP

TE D

M AN U

SC

improve interpretability.

ACCEPTED MANUSCRIPT 3

Abstract Background: The differential impact of number of prior lines of therapy and setting of prior therapy (perioperative, metastatic) is unclear in advanced urothelial carcinoma (UC).

RI PT

Patients and methods: Ten phase II trials of salvage chemotherapy and/or biologic agent therapy

enrolling 731 patients were available. Data on the number of prior lines of therapy and setting of prior therapy were required in addition to known previously recognized prognostic factors: time from prior chemotherapy, hemoglobin, performance status and liver metastasis status. Cox proportional hazards

SC

regression was used to evaluate the association of number of prior lines and prior perioperative therapy with overall survival (OS) as the primary clinical endpoint. Trial was a stratification factor.

M AN U

Results: 711 patients were evaluable. The overall median PFS and OS were 2.7 and 6.8 months, respectively. The number of prior lines were 1 in 559 (78.6%), 2 in 111 (15.6%), 3 in 29 (4.1%), 4 in 10 (1.4%) and 5 in 2 (0.3%) patients. Prior perioperative chemotherapy was given to 277 (39.1%) and chemotherapy for metastatic disease to 454 (64.1%). The number of prior lines was not independently associated with OS (HR 0.99 [95% CI: 0.86-1.14]). Prior peri-operative chemotherapy was a favorable

TE D

factor for OS on univariable but not multivariable analysis.

Conclusions: The number of prior lines of therapy and prior perioperative chemotherapy were not independently prognostic in UC patients receiving salvage therapy. Adoption of these data in salvage

EP

therapy trials should enhance accrual, the interpretability of results and drug development.

Keywords: Advanced urothelial cancer; Salvage therapy; Number of prior lines of therapy; Perioperative

AC C

chemotherapy; Prognosis

ACCEPTED MANUSCRIPT 4

Introduction The differential impact of number of prior lines of therapy and setting of prior therapy in patients receiving salvage systemic therapy for advanced urothelial carcinoma (UC) is unclear. Eligibility criteria for phase II

RI PT

trials conducted to identify the activity of new agents in the salvage setting have varied substantially. Some trials allow only one prior line of chemotherapy in any setting (perioperative or metastatic), while others require only one prior line for metastatic disease (and either disallow or permit prior perioperative

SC

chemotherapy). Few trials have allowed more than one prior line of therapy for metastatic disease.

Currently employed agents yield limited activity in the second-line setting of advanced UC and yield a

M AN U

median progression-free survival (PFS) of 2 to 4 months and median overall survival (OS) of 6 to 9 months [1]. Survival outcomes of patients receiving salvage therapy for advanced UC differs based on a 3-factor prognostic model derived from the sole large phase III trial evaluating vinflunine consisting of Eastern Cooperative Oncology Group (ECOG)-Performance Status (PS) >0, hemoglobin (Hb) <10 g/dL and liver metastasis (LM) [2, 3]. Another study reported that the addition of time from prior chemotherapy

TE D

(TFPC) to the prognostic model enhances the ability to prognosticate for progression-free survival (PFS) compared to the aforementioned 3-factor model [4]. However, TFPC has not been externally validated for association with overall survival (OS).

EP

A rationale may be offered to investigate the prognostic impact of the number of prior lines of therapy and setting of prior therapy (perioperative and/or metastatic) in patients receiving salvage systemic therapy for

AC C

advanced UC. Importantly, this information may facilitate the interpretation and conduct of phase II trials of salvage therapy. If these new variables impact outcomes, stratification to account for their impact will be necessary in randomized trials. Conversely, if these factors do not impact outcomes, uniform eligibility criteria may be advocated across trials and accrual may be enhanced with more inclusive criteria. Therefore, we retrospectively analyzed a large pooled dataset of prospective phase II trials to evaluate the impact of number of prior lines of therapy and setting of prior therapy independent of Hb, PS, LM and TFPC. We hypothesized that neither of these variables conferred an independent prognostic impact on OS.

ACCEPTED MANUSCRIPT 5

Patients and methods Patient population Ten prospective phase II trials of salvage systemic chemotherapy and/or biologic agent therapy for

RI PT

advanced UC were pooled [5-14]. These trials required previous pathological confirmation of UC and the presence of measurable metastatic disease. Trials conducted after the year 2000 were selected based on the availability of individual patient level data and willingness of the respective principal investigators to provide these data. Data regarding the number of prior lines of therapy and setting of prior therapy

SC

(perioperative and/or metastatic) were required in addition to TFPC (time from prior chemotherapy), Hb (hemoglobin), PS (performance status), and LM (liver metastasis) status. The data were deidentified and

M AN U

provided in an Excel spreadsheet by all investigators. The study was conducted after IRB approval at the University of Alabama, Birmingham (UAB) for retrospective analyses of such patients. The differences among these trials in patient eligibility criteria regarding number of prior lines and prior perioperative chemotherapy are shown in Table 1.

TE D

Statistical methods

Progression was defined as objective tumor progression or death from any cause, and calculated using the Kaplan-Meier method. OS was calculated from the date of study entry until death from any cause can calculated using the Kaplan-Meier method. Objective tumor assessment was performed by RECIST 1.0 in

EP

all trials except the trial evaluating pazopanib by Necchi et al, which used RECIST 1.1 [11, 15, 16]. Cox proportional hazards regression was used to evaluate the association of number of prior lines with OS,

AC C

the primary clinical endpoint, and progression-free survival (PFS), adjusted for the 3 validated prognostic factors along with TFPC. Trial was included as a stratification factor throughout. Patients in the Choueiri trial receiving docetaxel +/- vandetanib were included as one trial when stratifying for our analysis since there were no significant differences in OS and PFS between these arms [5]. Patients in the trial by Gallagher et al evaluated sunitinib administered in 2 doses and schedules (50 mg daily for 4 of every 6 weeks or 37.5 mg daily continuously) and the trial by Wong et al evaluated cetuximab alone or with paclitaxel as 2-non-comparative arms; for both of these trials, the different regimens were assigned a separate stratification [8, 10]. A sub-analysis examined the impact of prior perioperative chemotherapy.

ACCEPTED MANUSCRIPT 6

Internal validation was performed using bootstrap methods, with 95% bias-corrected and accelerated (BCa) confidence intervals, p-values and concordance statistics (c-index) calculated. All tests were two-

RI PT

sided and a p-value of 0.05 or less was considered statistically significant.

Results Patient characteristics

Of 731 patients, data for all factors was available for 711 (Table 2), and 3 patients from the Stadler et al

SC

volasertib study did not have data regarding prior chemotherapy setting (i.e. for metastatic or

perioperative disease) and were therefore excluded from that analysis. Trials evaluated vinflunine

M AN U

(N=151), docetaxel +/- vandetanib (N=147), paclitaxel-gemcitabine (N=83), sunitinib (N=77), nabpaclitaxel (N=48), volasertib (N=46), everolimus (N=45), pazopanib (N=43), cetuximab +/-paclitaxel (N=39) and paclitaxel-cyclophosphamide (N=32). The number of prior lines of therapy including perioperative chemotherapy as a line were 1 in 559 (78.6%), 2 in 111 (15.6%), 3 in 29 (4.1%), 4 in 10 (1.4%) and 5 in 2 (0.3%) patients. Prior perioperative chemotherapy was given to 277 (39.1%),

TE D

chemotherapy for metastatic disease to 454 (64.1%) patients and prior chemotherapy in both perioperative and metastatic settings to 75 (27.1%) patients.

Impact of number of prior lines of therapy on PFS and OS

EP

The overall median PFS and OS were 2.7 and 6.8 months, respectively (Table 3). Outcomes with different regimens exhibited a range of outcomes, but were uniformly dismal (Table 3). TFPC, Hb, PS and

AC C

LM were significantly associated with OS and PFS on both univariate and multivariate analyses. However, the number of prior lines (by ordinal increase) was not associated with OS (HR=0.99 [95% CI: 0.86-1.14], p=0.89) or PFS (HR=0.92 [95% CI: 0.80-1.05], p=0.22) in either univariate or multivariate analyses (see Tables 4 and 5). A supportive analysis was performed which dichotomized the number of prior lines into 1 versus 2 or more, with similar results (see Figure 1A).

ACCEPTED MANUSCRIPT 7

Impact of prior perioperative chemotherapy on PFS and OS Prior peri-operative chemotherapy was a favorable factor for both OS and PFS on univariable but not multivariable analysis (HR=0.94 [95% CI: 0.77-1.14], p=0.51 for OS and HR=0.90 [95% CI: 0.75-1.08],

RI PT

p=0.25 for PFS), see Figure 1B and tables 4 and 5. A subgroup analysis which included only patients having one prior line of chemotherapy (n=559) showed similar results (data not shown). Further, no interaction between peri-operative chemotherapy and number of prior lines was observed for PFS

(p=0.88) or OS (p=0.40). Upon further analysis, it was observed that removing TFPC from the multivariate

SC

models rendered prior perioperative chemotherapy almost significantly associated with OS (HR=0.84 [95% CI: 0.70-1.01, p=0.059) and PFS (HR=0.84 [95% CI: 0.71-1.00], p=0.053). The removal of TFPC

M AN U

(and not the other 3 factors) from the model was considered, because TFPC is externally validated only for PFS and not OS (4).

Internal validation

Bootstrapping was performed to internally validate the results. The 95% BCa CI for the HR for the number

TE D

of prior lines was 0.83 to 1.16 for OS (p-value=0.87) and 0.79 to 1.09 for PFS (p-value=0.58). Bootstrapped 95% BCa CI for the HR for prior perioperative chemotherapy was 0.76 to 1.15 for OS (pvalue=0.61) and 0.72 to 1.08 for PFS (p-value=0.50). The addition of each factor into the multivariate

EP

model only minimally increased the c-index, with the upper bound of the 95% BCa CI for the increase in

AC C

c-index being <0.01 for both the number of primary lines and perioperative chemotherapy.

Discussion

In this retrospective analysis of individual patient level data from 10 prospective phase II trials of salvage therapy for advanced UC, we could not demonstrate an independent prognostic impact for number of prior lines of therapy (including perioperative chemotherapy as one line) and prior perioperative chemotherapy on both PFS and OS. Our analysis appears applicable across different classes of agents since it was stratified for trial, which provides a more granular analysis than stratifying for type of therapy used, i.e. cytotoxic vs. biologic vs. both. Additionally, we and others have shown that the prior platinum

ACCEPTED MANUSCRIPT 8

agent (cisplatin vs. carboplatin) does not confer an independent impact [17, 18]. The previously recognized prognostic factors, i.e. PS, Hb, ECOG-PS and TFPC continued to be independently significantly prognostic for both PFS and OS [4, 19]. In this context, it is also noteworthy that a previous

RI PT

study demonstrated that response to prior therapy, when administered for metastatic disease, also is not independently prognostic for either PFS or OS [20]. These data suggest that trials of salvage therapy could have more inclusive eligibility criteria and allow a larger number of previous lines of therapy.

Moreover, prior perioperative chemotherapy may be allowed as the only prior line of therapy with the

M AN U

to have been administered within one year (Table 1).

SC

caveat that many of the salvage trials that allowed such patients required the perioperative chemotherapy

The retrospective design of our study is accompanied by certain limitations, although we utilized individual patient level data from well conducted prospective phase II trials. However, the trials included represents those where individual level data for all requested variables were available and does not include all salvage trials ever conducted. There was some heterogeneity in the eligibility criteria and

TE D

patient characteristics in the different trials that were pooled (Tables 1 and 2), and there remains the possibility of confounding variables and ascertainment bias with respect to factors such as co-morbidities and molecular differences. Nevertheless, outcomes observed in these selected trials are quite similar to outcomes seen in other trials of salvage therapy. We employed OS as the primary clinical endpoint, since

EP

this is objectively measurable. Although PFS is affected by intervals between radiographic assessments, the patients in our pooled dataset were accrued in prospective phase II trials that performed imaging

AC C

every 6 to 12 weeks. The vast majority of patients in this dataset received up to 2 prior regimens; only 5.8% of patients had received >2 prior regimens and only 1.7% had received >3 prior regimens (Table 2). Hence, these data require external validation especially in patients receiving > 2 prior regimens. In the absence of external validation, an internal validation analysis using bootstrapping is considered the next best procedure [21, 22]. Although we were able to internally validate these results, since these dataset consisted of patients treated on well-conducted prospective phase II trials, with excellent clinical outcome and prior therapy annotation, differences may exist from what clinicians see in their general practice. Potentially, a greater number of prior lines may increase the risk of toxicities and compromise delivery of

ACCEPTED MANUSCRIPT 9

salvage therapy and reduce survival. Unfortunately, we did not have access to toxicity data, although the lack of significant independent impact of number of prior lines on clinical efficacy outcomes suggests that toxicities probably did not increase to an extent to compromise survival. Indeed, a previously published

RI PT

nomogram showed a marginal deleterious effect of number of previous chemotherapy regimens on severe toxicities attributable to molecularly targeted agents.[23]

Prior perioperative chemotherapy was a significantly favorable prognostic factor on univariable analyses

SC

for both PFS and OS, but not in multivariable analyses including PS, Hb, ECOG-PS and TFPC.

Interestingly, removing TFPC from the multivariate model rendered prior perioperative chemotherapy

M AN U

almost significant. Although TFPC was originally developed to be applicable to all patients receiving any line of salvage therapy, the majority of patients were receiving second-line therapy and the term ‘platinum-free interval’ may more appropriately describe this variable in those receiving second-line therapy [4]. These data suggest that including TFPC in the prognostic model to stratify patients in trials of salvage UC may eliminate the confounding impact of prior perioperative chemotherapy as a line of

TE D

therapy. Thus, although TFPC has not been externally validated for independent impact on survival, its inclusion in the prognostic model may allow broader eligibility criteria in trials of salvage UC. Notably, external validation of the impact of TFPC on OS may have been compromised by using the phase III vinflunine trial dataset, which only enrolled patients following a single prior line of platinum-based

EP

chemotherapy for metastatic disease, i.e. none of the patients had received prior perioperative chemotherapy. TFPC is likely most relevant following perioperative chemotherapy, when a wide range of

AC C

periods to progression may be expected, as opposed to following chemotherapy for metastatic disease, when patients generally progress after a brief period of time. It is therefore crucial that future studies evaluate and validate the effect of TFPC on OS in a salvage therapy dataset allowing prior chemotherapy for both perioperative and metastatic settings. An indirect benefit of allowing patients receiving prior perioperative chemotherapy and multiple prior lines of therapy on clinical trials is acceleration of trial accrual and drug development, which has generally been slow due to the poor performance status and comorbidities in advanced UC patients.

ACCEPTED MANUSCRIPT 10

It is noteworthy that in recently reported and ongoing second-line trials of other major malignancies, e.g. non-small cell lung cancer, colorectal cancer and breast cancer, prior perioperative chemotherapy alone within 6-12 months has generally been considered acceptable for eligibility [24-26]. Conversely, patients

RI PT

with these malignancies progressing beyond 6-12 months after perioperative chemotherapy have been typically eligible for first-line trials [27, 28]. Additionally, salvage trials in these malignancies have

generally allowed prior perioperative chemotherapy in addition to therapy for metastatic disease, i.e.

practices in trials of salvage therapy for advanced UC.

SC

patients with ≥2 lines of therapy [25]. Therefore, data from our analysis corroborate the adoption of these

M AN U

Despite the lack of a discernible significant independent impact of the number of prior lines of therapy on clinical outcomes, careful patient selection is still warranted based on specific prior agents administered. For example, trials evaluating new tubulin inhibitors probably should stratify for prior tubulin inhibitor administration (e.g. taxanes), given the possibility of cross resistance. If a specific molecular biomarker predictive for benefit from a new agent has been preclinically validated, a balanced distribution of this

TE D

biomarker is warranted in trials enrolling all unselected patients. Additionally, since heavily pretreated patients appear to have somewhat poorer tolerance of novel therapies, prudent patient selection is

Conclusion

EP

desirable for trials of agents which have the potential to cause substantial toxicity [23].

To summarize, the number of prior lines of therapy and prior perioperative chemotherapy were not

AC C

independently prognostic in advanced UC patients receiving salvage therapy. The inclusion of TFPC in the prognostic model appears to obviate the need to adjust for the favorable impact of prior perioperative chemotherapy. Uniform adoption of these data in trials of salvage therapy will probably enhance the interpretability of results and pace of patient accrual and drug development in this setting.

ACCEPTED MANUSCRIPT 11

Clinical practice points •

In this analysis of pooled data from 10 phase II trials of salvage chemotherapy and/or biologic agent therapy enrolling 731 patients, the number of prior lines was not independently associated

•

RI PT

with OS or PFS.

Prior perioperative chemotherapy was a favorable factor for both OS and PFS on univariable but not multivariable analysis.

•

Thus, patients may be enrolled in salvage therapy trials regardless of number of prior lines of

EP

TE D

results and drug development.

M AN U

Adoption of these data in salvage therapy trials should enhance accrual, the interpretability of

AC C

•

SC

therapy and whether prior perioperative therapy was administered.

ACCEPTED MANUSCRIPT 12

References

AC C

EP

TE D

M AN U

SC

RI PT

[1] Sonpavde G, Sternberg CN, Rosenberg JE, Hahn NM, Galsky MD, Vogelzang NJ. Second-line systemic therapy and emerging drugs for metastatic transitional-cell carcinoma of the urothelium. The lancet oncology. 2010 Sep: 11:861-70 [2] Bellmunt J, Choueiri TK, Fougeray R, et al. Prognostic factors in patients with advanced transitional cell carcinoma of the urothelial tract experiencing treatment failure with platinum-containing regimens. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Apr 10: 28:1850-5 [3] Bellmunt J, Theodore C, Demkov T, et al. Phase III trial of vinflunine plus best supportive care compared with best supportive care alone after a platinum-containing regimen in patients with advanced transitional cell carcinoma of the urothelial tract. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009 Sep 20: 27:4454-61 [4] Sonpavde G, Pond GR, Fougeray R, et al. Time from prior chemotherapy enhances prognostic risk grouping in the second-line setting of advanced urothelial carcinoma: a retrospective analysis of pooled, prospective phase 2 trials. Eur Urol. 2013 Apr: 63:717-23 [5] Choueiri TK, Ross RW, Jacobus S, et al. Double-blind, randomized trial of docetaxel plus vandetanib versus docetaxel plus placebo in platinum-pretreated metastatic urothelial cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Feb 10: 30:507-12 [6] Vaughn DJ, Srinivas S, Stadler WM, et al. Vinflunine in platinum-pretreated patients with locally advanced or metastatic urothelial carcinoma: results of a large phase 2 study. Cancer. 2009 Sep 15: 115:4110-7 [7] Albers P, Park SI, Niegisch G, et al. Randomized phase III trial of 2nd line gemcitabine and paclitaxel chemotherapy in patients with advanced bladder cancer: short-term versus prolonged treatment [German Association of Urological Oncology (AUO) trial AB 20/99]. Ann Oncol. Feb: 22:288-94 [8] Wong YN, Litwin S, Vaughn D, et al. Phase II Trial of Cetuximab With or Without Paclitaxel in Patients With Advanced Urothelial Tract Carcinoma. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. Oct 1: 30:3545-51 [9] Ko YJ, Canil CM, Mukherjee SD, et al. Nanoparticle albumin-bound paclitaxel for second-line treatment of metastatic urothelial carcinoma: a single group, multicentre, phase 2 study. The lancet oncology. 2013 Jul: 14:769-76 [10] Gallagher DJ, Milowsky MI, Gerst SR, et al. Phase II study of sunitinib in patients with metastatic urothelial cancer. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2010 Mar 10: 28:1373-9 [11] Necchi A, Mariani L, Zaffaroni N, et al. Pazopanib in advanced and platinum-resistant urothelial cancer: an open-label, single group, phase 2 trial. The lancet oncology. 2012 Aug: 13:810-6 [12] Milowsky MI, Iyer G, Regazzi AM, et al. Phase II study of everolimus in metastatic urothelial cancer. BJU Int. 2013 Aug: 112:462-70 [13] Di Lorenzo G, Montesarchio V, Autorino R, et al. Phase 1/2 study of intravenous paclitaxel and oral cyclophosphamide in pretreated metastatic urothelial bladder cancer patients. Cancer. 2009 Feb 1: 115:517-23 [14] Stadler WM, Vaughn DJ, Sonpavde G, et al. An open-label, single-arm, phase 2 trial of the pololike kinase inhibitor volasertib (BI 6727) in patients with locally advanced or metastatic urothelial cancer. Cancer. 2013 Dec 11: [15] Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). European journal of cancer. 2009 Jan: 45:228-47 [16] Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. Journal of the National Cancer Institute. 2000 Feb 2: 92:205-16 [17] Harshman LC, Fougeray R, Choueiri TK, et al. The impact of prior platinum therapy on survival in patients with metastatic urothelial cancer receiving vinflunine. Br J Cancer. 2013 Nov 12: 109:2548-53 [18] Sonpavde G, Bellmunt J, Rosenberg JE, et al. Impact of prior platinum agent and site of primary in patients with advanced urothelial carcinoma (UC) receiving salvage therapy. J Clin Oncol 32, 2014 (suppl 4; abstr 336).

ACCEPTED MANUSCRIPT 13

AC C

EP

TE D

M AN U

SC

RI PT

[19] Pond GR, Agarwal N, Bellmunt J, et al. A nomogram including baseline prognostic factors to estimate the activity of second-line therapy for advanced urothelial carcinoma. BJU Int. 2013 Nov 13: [20] Pond GR, Bellmunt J, Fougeray R, et al. Impact of Response to Prior Chemotherapy in Patients With Advanced Urothelial Carcinoma Receiving Second-Line Therapy: Implications for Trial Design. Clin Genitourin Cancer. 2013 Dec: 11:495-500 [21] Steyerberg EW, Harrell FE, Jr., Borsboom GJ, Eijkemans MJ, Vergouwe Y, Habbema JD. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. Journal of clinical epidemiology. 2001 Aug: 54:774-81 [22] Brunelli A, Rocco G. Internal validation of risk models in lung resection surgery: bootstrap versus training-and-test sampling. The Journal of thoracic and cardiovascular surgery. 2006 Jun: 131:1243-7 [23] Pond GR, Siu LL, Moore M, et al. Nomograms to predict serious adverse events in phase II clinical trials of molecularly targeted agents. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008 Mar 10: 26:1324-30 [24] Li N, Ou W, Yang H, et al. A randomized phase 2 trial of erlotinib versus pemetrexed as secondline therapy in the treatment of patients with advanced EGFR wild-type and EGFR FISH-positive lung adenocarcinoma. Cancer. 2014 Jan 30: [25] Cortes J, O'Shaughnessy J, Loesch D, et al. Eribulin monotherapy versus treatment of physician's choice in patients with metastatic breast cancer (EMBRACE): a phase 3 open-label randomised study. Lancet. 2011 Mar 12: 377:914-23 [26] Van Cutsem E, Tabernero J, Lakomy R, et al. Addition of aflibercept to fluorouracil, leucovorin, and irinotecan improves survival in a phase III randomized trial in patients with metastatic colorectal cancer previously treated with an oxaliplatin-based regimen. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2012 Oct 1: 30:3499-506 [27] Schmoll HJ, Cunningham D, Sobrero A, et al. Cediranib with mFOLFOX6 versus bevacizumab with mFOLFOX6 as first-line treatment for patients with advanced colorectal cancer: a double-blind, randomized phase III study (HORIZON III). Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2012 Oct 10: 30:3588-95 [28] Baselga J, Cortes J, Kim SB, et al. Pertuzumab plus trastuzumab plus docetaxel for metastatic breast cancer. N Engl J Med. 2012 Jan 12: 366:109-19

ACCEPTED MANUSCRIPT 14

Figure 1. Overall survival based on number of prior lines (A) and prior perioperative chemotherapy (B) Legend: The number of prior lines was not associated with OS in univariate or multivariate analyses; prior

AC C

EP

TE D

M AN U

SC

RI PT

peri-operative chemotherapy was a favorable factor for OS on univariable but not multivariable analysis

ACCEPTED MANUSCRIPT 15

Table 1. Eligibility criteria for number of prior lines and prior perioperative chemotherapy in included trials Regimen

Total

Number

Evaluable for analysis

Allowed prior peri-operative chemotherapy as only prior therapy

Period allowed between prior peri-operative chemotherapy and secondline therapy

Required

only 1 prior

RI PT

Author

regimen in

Required prior

therapy for metastatic disease

any setting

Paclitaxel† Gemcitabine

98

83

*

Yes

Ko[9]

Nab-paclitaxel

48

48

Yes

Choueiri [5]

Docetaxel + † placebo/vandetanib

148

147

*

π

Yes

No

1 year

No

No

Yes

2 years

No

No

Yes

Unlimited

Yes

No

Yes

1 year

Yes

No

Yes

2 years

Yes

No

SC

Unlimited

M AN U

Albers[7]

Wong [8]

Cetuximab +/π Paclitaxel

39

39

Vaughn[6]

Vinflunine

151

151

Stadler[14]

Volasertib

50

46

Milowsky [12]

Everolimus

45

45

Yes

Unlimited

No

No

77

77

Yes

Unlimited

No

No

Sunitinib

π

TE D

Gallagher [10]

*

Pazopanib

43

43

No

Unlimited

No

Yes

Di Lorenzo [13]

PaclitaxelCyclophosphamide

32

32

No

Unlimited

No

Yes

731

711

†

AC C

Total

EP

Necchi [11]

Multiple

Randomized phase II trials with no significant differences in outcomes between arms, *either some or all patients were inevaluable from these trials due to lack of baseline Hb and/or LM status and/or setting of π prior chemotherapy, Patients in the Choueiri trial receiving docetaxel +/- vandetanib were included as one trial when stratifying for our analysis since there were no significant differences in OS and PFS between these arms. Patients in the trial by Gallagher et al evaluated sunitinib administered in 2 doses and schedules (50 mg daily for 4 of every 6 weeks or 37.5 mg daily continuously) and the trial by Wong et al evaluated cetuximab alone or with paclitaxel as 2-non-comparative arms; for both of these trials, the different regimens were assigned a separate stratification.

ACCEPTED MANUSCRIPT 16

Table 2. Patient characteristics by regimen

2 3 4

Cetuximab

PaclitaxelCetuximab

711

45

32

45

46

151

11

28

533 (75.0) 64.9 (9.6) 335 (47.1) 238 (33.5) 105 (14.8) 4.4 (0103.4)

32 (71.1) 62.9 (10.1) 27 (60.0) 17 (37.8) 2 (4.4) 7.8 (1.038.6)

21 (65.6) 66.8 (11.3) 13 (40.6) 12 (37.5) 3 (9.4) 3.2 (0.928.2)

31 (68.9) 63.6 (10.2) 22 (48.9) 18 (40.0) 7 (15.6) 4.3 (1.080.2)

36 (78.3) 68.9 (7.3) 21 (45.7) 20 (43.5) 6 (13.0) 6.7 (0.932.6)

121 (80.1) 64.5 (9.7) 48 (31.8) 49 (32.5) 14 (9.3) 4.3 (0.422.2)

9 (81.8) 68.5 (10.2) 4 (36.4) 3 (27.3) 3 (27.3) 4.7 (0.814.0)

25 (89.3) 67.3 (8.2) 19 (67.9) 3 (10.7) 8 (28.6) 2.9 (0.7-21.5)

450 (63.3)

31 (68.9)

16 (50.0)

29 (64.4)

36 (78.3)

94 (62.3)

7 (63.6)

559 (78.6) 111 (15.6) 29 (4.1) 10 (1.4)

29 (64.4) 12 (26.7) 3 (6.7) 1 (2.2)

24 (75.0) 6 (18.8) 1 (3.1) 1 (3.1)

151 (100.0) 0 (0.0) 0 (0.0) 0 (0.0)

11 (100.0) 0 (0.0) 0 (0.0) 0 (0.0)

29 (64.4) 13 (28.9) 3 (6.7) 0 (0.0)

40 (87.0) 5 (10.9) 1 (2.2) 0 (0.0)

Docetaxel +/vandetanib 147

Paclitaxel Cyclopho sphamide

Pazopanib

Gemcitab inepaclitaxel

Nab-Paclit -axel

32

43

83

48

101 (68.7) 64.6 (9.7) 72 (49.0) 49 (33.3) 25 (17.0) 3.9 (0.435.3)

20 (62.5) 61.3 (8.6) 16 (50.0) 16 (50.0) 10 (31.3) 4.0 (1.0-9.0)

36 (83.7) 63.5 (11.1) 26 (60.5) 22 (51.2) 6 (14.0) 3.0 (1.024.0)

61 (73.5) 65.0 (8.0) 35 (42.2) 13 (15.7) 11 (13.3) 7.1 (0-103.4)

40 (83.3) 65.9 (9.8) 32 (66.7) 16 (33.3) 10 (20.8) 4.5 (0.5-48.8)

14 (50.0)

92 (62.6)

18 (56.3)

20 (46.5)

63 (75.9)

30 (62.5)

28 (100.0) 0 (0.0) 0 (0.0) 0 (0.0)

84 (57.1) 44 (29.9) 14 (9.5) 4 (2.7)

24 (75.0) 8 (25.0) 0 (0.0) 0 (0.0)

19 (44.2) 12 (27.9) 7 (16.3) 4 (9.3)

76 (91.6) 7 (8.4) 0 (0.0) 0 (0.0)

44 (91.7) 4 (8.3) 0 (0.0) 0 (0.0)

RI PT

Vinflunine

SC

Volasertib

M AN U

Median (range) months from last chemotherapy ≥3 Months from last chemotherapy Number of Prior Lines 1

Everolimus

TE D

N (%) ECOG PS 1 or higher N (%) liver metastases N (%) anemia

Sunitinib (37.5 mg daily)

EP

Mean (sd) age

Sunitinib (50 mg 4/2)

AC C

N Included in Analysis N (%) male

All Trials

ACCEPTED MANUSCRIPT 17

0 (0.0) 23 (51.1)

0 (0.0) 36 (78.3)

0 (0.0) 22 (14.6)

0 (0.0) 5 (45.5)

0 (0.0) 15 (53.6)

454 (64.1)

29 (64.4)

20 (62.5)

28 (62.2)

7/43 (16.3)

97 (64.2)

6 (54.6)

13 (46.4)

1 (0.7) 57 (38.8)

0 (0.0) 8 (25.0)

1 (2.3) 8 (18.6)

0 (0.0) 41 (49.4)

0 (0.0) 18 (37.5)

104 (70.8)

32 (100.0)

43 (100.0)

42 (50.6)

33 (68.8)

RI PT

0 (0.0) 16 (50.0)

TE D

M AN U

SC

0 (0.0) 28 (62.2)

EP

N (%) PeriOperative Chemotherapy (n=708) N (%) chemo for metastatic disease (n=708)

2 (0.3) 277 (39.1)

AC C

5

ACCEPTED MANUSCRIPT 18

Table 3. Clinical outcomes by regimen Sunitinib (50 mg 4/2)

Sunitini b (37.5 mg daily)

Everolimus

Volasertib

Vinflunine

Cetuximab

Paclitaxel -Cetuximab

Docetaxel +/vandetanib

Pazopanib

Gemcita binepaclitaxe l

Nab-Paclit -axel

32

Paclitax elCyclop hospha mide 43

N in Analysis

711

45

32

45

46

151

11

147

83

28

48

N (%) Censore d Med (CI)

140 (19.7)

11 (24.4)

4 (12.5)

7 (15.6)

11 (23.9)

65 (43.1)

9 (6.1)

0 (0.0)

2 (4.7)

3 (3.6)

6 (21.4)

22 (45.8)

3-month (95% CI)

6.8 (6.2-7.3) 82.1 (79.0-84.7)

6-month (95% CI)

54.2 (50.3-57.8)

1-year (95% CI)

27.3 (23.9-30.9)

7.5 (4.510.2) 82.2 (67.690.7) 60.0 (44.372.6) 30.0 (16.844.3)

5.6(4.36.8) 87.5 (70.095.1) 43.8 (26.559.8) 21.4 (9.236.9)

8.3 (5.310.5) 86.7 (72.793.8) 60.0 (44.372.6) 32.8 (19.746.6)

6.2 (3.69.8) 72.8 (57.183.6) 51.3 (35.565.0) 30.9 (17.545.2)

7.9 (6.79.5) 85.0 (78.189.9) 62.6 (53.770.2) 28.8 (20.138.0)

5.3 (4.46.5) 74.2 (66.380.5) 43.5 (35.451.4) 20.6 (14.427.5)

6.5 (5.07.0) 100.0 (-)

7.8 (5.811.7) 80.7 (70.587.7) 57.8 (46.567.6) 37.3 (27.147.6)

9.7 (6.413.5) 85.7 (66.394.4) 75.0 (54.687.2) 41.3 (22.958.7)

10.8 (5.9-16.8) 84.3 (69.9-92.2)

50.0 (31.965.7) 3.1 (0.213.8)

4.9 (4.27.5) 86.0 (71.693.5) 41.9 (27.155.9) 11.6 (4.323.1)

N (%) Censore d Med (CI)

53 (7.5)

3 (6.7)

2 (6.3)

6 (13.3)

1 (2.2)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

0 (0.0)

13 (27.1)

2.7 (2.4-2.8) 42.6 (38.9-46.3) 19.0 (16.122.1)

2.4 (1.6-3.6) 37.3 (23.251.4) 16.3 (7.228.7)

1.8 (1.42.7) 27.0 (12.843.4) 3.4 (0.314.6)

3.0 (1.9-3.5) 50.2 (34.064.4) 7.5 (1.918.3)

1.4 (1.3-2.3) 28.3 (16.341.6) 19.6 (9.732.0)

1.8 (1.5-2.4) 33.3 (25.941.0) 15.6 (10.322.0)

3.5 (3.0-5.0) 50.0 (31.965.7) 6.3 (1.118.1)

2.4 (1.7-2.9) 34.9 (21.248.9) 11.6 (4.323.1)

3.5 (2.7-5.7) 51.8 (40.661.9) 38.6 (28.248.8)

3.8 (1.9-5.6) 60.7 (40.476.0) 21.4 (8.737.8)

5.5 (3.0-8.4) 64.3 (49.076.1) 49.0 (34.162.3)

6-month (95% CI)

M AN U

3.9 (1.08.1) 72.7 (37.190.3) 27.3 (6.553.9) 18.2 (2.944.2) PFS 28 (18.5) 0 (0.0)

TE D

EP

AC C

3-month (95% CI)

0 (0.0)

SC

OS

RI PT

All Trials

2.8 (2.6-3.8) 47.5 (39.055.5) 14.7 (8.921.7)

1.7 (0.7-2.4) 0 (-) 0 (-)

64.9 (48.5-77.2) 47.9 (30.663.2)

ACCEPTED MANUSCRIPT 19

Table 4. Univariate and multivariate analyses of potential prognostic factors for overall survival Factor

Factor Type

Hazard Ratio (95% CI)

p-value

UNIVARIATE ANALYSIS Time since last chemotherapy

1.85 (1.54-2.21)

<0.001

ECOG 1+ versus ECOG 0

1.84 (1.54-2.19)

<0.001

Yes versus No

1.62 (1.35-1.94)

<0.001

<10 versus ≥10g/dL

2.53 (2.01-3.19)

<0.001

# of Prior Lines

Ordinal Increase

1.04 (0.90-1.20)

0.58

Peri-Operative Chemotherapy

Yes versus No

0.78 (0.65-0.94)

0.008

Liver metastases

M AN U

Anemia

SC

Performance Status

RI PT

<3 months versus 3+ months

MULTIVARIATE MODEL INCLUDING NUMBER OF PRIOR LINES (n=711) Time since last chemotherapy Performance Status

<3 months versus 3+ months

1.61 (1.34-1.94)

<0.001

ECOG 1+ versus ECOG 0

1.59 (1.33-1.91)

<0.001

Yes versus No

1.44 (1.19-1.73)

<0.001

<10 versus ≥10g/dL

1.91 (1.50-2.44)

<0.001

Ordinal Increase

0.99 (0.86-1.14)

0.89

TE D

Liver metastases Anemia # of Prior Lines

Time since last chemotherapy

Liver metastases Anemia

<3 months versus 3+ months

1.58 (1.31-1.92)

<0.001

ECOG 1+ versus ECOG 0

1.58 (1.31-1.89)

<0.001

Yes versus No

1.44 (1.19-1.73)

<0.001

<10 versus ≥10g/dL

1.90 (1.49-2.43)

<0.001

Yes versus No

0.94 (0.77-1.14)

0.51

AC C

Performance Status

EP

MULTIVARIATE MODEL INCLUDING PRIOR PERIOPERATIVE CHEMOTHERAPY (n=708)

Peri-Operative Chemotherapy

* Note: if an interaction term (for peri-operative chemotherapy with # of prior lines) was added to this model, it did not significantly add information (p=0.40).

ACCEPTED MANUSCRIPT 20

Table 5. Univariate and multivariate analyses of potential prognostic factors for progression-free survival Factor

Factor Type

Hazard Ratio (95% CI)

p-value

UNIVARIATE ANALYSIS Time since last chemotherapy

1.65 (1.39-1.95)

<0.001

ECOG 1+ versus ECOG 0

1.43 (1.21-1.68)

<0.001

Yes versus No

1.61 (1.35-1.92)

<0.001

<10 versus ≥10g/dL

1.97 (1.58-2.47)

<0.001

# of Prior Lines

Ordinal Increase

0.99 (0.86-1.13)

0.84

Peri-Operative Chemotherapy

Yes versus No

0.78 (0.66-0.93)

0.006

Liver metastases

M AN U

Anemia

SC

Performance Status

RI PT

<3 months versus 3+ months

MULTIVARIATE MODEL INCLUDING NUMBER OF PRIOR LINES (n=711) Time since last chemotherapy Performance Status

<3 months versus 3+ months

1.48 (1.24-1.76)

<0.001

ECOG 1+ versus ECOG 0

1.27 (1.08-1.51)

0.005

Yes versus No

1.48 (1.24-1.76)

<0.001

<10 versus ≥10g/dL

1.67 (1.32-2.12)

<0.001

Ordinal Increase

0.92 (0.80-1.05)

0.22

TE D

Liver metastases Anemia # of Prior Lines

Time since last chemotherapy

Liver metastases Anemia

<3 months versus 3+ months

1.42 (1.19-1.70)

<0.001

ECOG 1+ versus ECOG 0

1.27 (1.07-1.50)

0.006

Yes versus No

1.50 (1.25-1.79)

<0.001

<10 versus ≥10g/dL

1.60 (1.27-2.03)

<0.001

Yes versus No

0.90 (0.75-1.08)

0.25

AC C

Performance Status

EP

MULTIVARIATE MODEL INCLUDING PRIOR PERIOPERATIVE CHEMOTHERAPY (n=708)

Peri-Operative Chemotherapy

* Note: if an interaction term (for peri-operative chemotherapy with # of prior lines) was added to this model, it did not significantly add information (p=0.88).

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT