Systematic Review of the Therapeutic Efficacy of Bladder-preserving Treatments for Non–muscle-invasive Bladder Cancer Following Intravesical Bacillus Calmette-Guérin

Abstract

Context:

There is a critical need for effective bladder-sparing therapies for bacillus Calmette-Guérin (BCG)-unresponsive non–muscle-invasive bladder cancer (NMIBC). Owing to the current lack of effective agents that can be used as a control, the US Food and Drug Administration began to accept single-arm trials for patients with carcinoma in situ (CIS), using complete response rate (CRR) and duration of response as the primary endpoints to support marketing applications. Despite the ensuing growth of clinical trials in this space, no consensus exists on a clinically relevant benchmark for CRR.

Objective:

To elucidate the CRR and recurrence-free rate (RFR) using bladder-sparing agents after BCG failure in order to provide a frame of reference for future clinical trial results.

Evidence acquisition:

We performed a systematic review of clinical trials utilizing bladder-sparing therapeutics for NMIBC recurring after intravesical BCG (PROSPERO CRD42019130553). The search was performed in MEDLINE, EMBASE, and Cochrane Library. Relevant studies identified from bibliography search and conference abstracts were searched to complement the systematic review. A total of 42 studies utilizing 24 treatment options and consisting of 2254 patients were included for final analysis.

Evidence synthesis:

Median CRRs in the treatment of CIS-containing tumors were 26% at 6 mo, 17% at 12 mo, and 8% at 24 mo after treatment. In comparison, median RFRs in the papillary-only studies were 67% at 6 mo, 44% at 12 mo, and 10% at 24 mo. Specifically in the BCG-unresponsive population, 6- and 12-mo CRRs in CIS-containing patients treated with Mycobacterium phlei cell wall-nucleic acid complex were 45% and 27%, respectively, and the median 6-, 12-, and 24-mo disease-free rates in the other studies were 43%, 35%, and 18%, respectively. The median progression-free rate was 91%: 95% in the CIS-containing studies and 89% in studies restricted to papillary-only recurrences. Toxicities of intravesical agents were generally mild, with very few dose limiting toxicities.

Conclusions:

We demonstrate that, to date, bladder-sparing therapies achieved modest efficacy in patients with NMIBC after BCG. Results from the current study will serve as a frame of reference for emerging trial results in the BCG-unresponsive space.

Patient summary:

In this study, we found that bladder-sparing therapies achieved modest efficacy in patients with non–muscle-invasive bladder cancer after bacillus Calmette-Guérin (BCG). These results will serve to inform future clinical trial results for salvage agents used to treat BCG-unresponsive bladder cancer.

1. Introduction

There is a critical unmet need for effective bladder-sparing therapies to treat recurrent non–muscle-invasive bladder cancer (NMIBC) not responsive to intravesical (i/ves) bacillus Calmette-Guérin (BCG). Although radical cystectomy is the recommended standard-of-care treatment in this setting, a substantial proportion of patients are either unfit for or unwilling to undergo surgery, thus raising the need for bladder-sparing alternatives.

Development of new agents in this disease space has been hampered by the heterogeneity in patient population, poor definition of disease states, a lack of appropriate control arms, and consensus on trial endpoints. To address these concerns, workshops cosponsored by the US Food and Drug Administration (FDA) and American Urological Association (AUA), as well as expert consensus guidelines, have helped orchestrate the general framework for future studies [1–3]. Many of the recommendations were subsequently incorporated into the FDA’s Guidance Document on developing drugs and biologics for the treatment of BCG-unresponsive bladder cancer. In this document, the FDA indicated that open-label, single-arm clinical trials for carcinoma in situ (CIS)-containing BCG-unresponsive NMIBC with complete response rate (CRR) and duration of response (DoR) as the primary endpoints would be accepted to support a marketing application [4]. In contrast, efficacy of salvage agents for BCG-unresponsive patients with only resected papillary disease should be tested using a randomized controlled trial design with a time-to-event endpoint such as recurrence-free survival.

The FDA’s decision to accept single-arm clinical trials for novel agents tested in BCG-unresponsive NMIBC patients with CIS means that reference rates will be needed to guide discussions evaluating the effectiveness of such agents. Although recommendations for clinically relevant CRRs have been proposed [2], they were not data driven. Accordingly, we conducted a systematic review of clinical trials involving bladder-sparing therapies used in the post-BCG setting to identify evidence-based CRRs and recurrence-free rates (RFRs) and to provide a frame of reference to guide interpretation of future clinical trial results.

2. Evidence acquisition

2.1. Systematic literature review

We performed a systematic review of prospective clinical trials utilizing bladder-sparing therapeutics for NMIBC after prior i/ves BCG therapy following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (PROSPERO CRD42019130553). Comprehensive search strategies were used to identify all relevant trials investigating the safety and/or efficacy of bladder-sparing therapeutics following persistent CIS and/or non–muscle-invasive papillary i/ves recurrence after BCG treatment. The search was performed in MEDLINE, EMBASE, and Cochrane Library using search terms defined by the population, interventions, comparisons, outcomes, and study design (PICOS) approach [5], using the terms “intravesical,” “non–muscle invasive bladder cancer” or “carcinoma” or “neoplasm” or “malignancy,” “refractory” or “recurring” or “failure” or “non-responsive” or “unresponsive,” and “BCG” or “bacillus Calmette-Guerin” (Supplementary material), with no limitations on publication year. Only English-language publications were considered. Observational studies, editorials, commentaries, review articles, and those not subject to peer review were excluded. Bibliographies of included studies were manually searched to ensure completeness. Conference abstracts of relevant medical societies up to and including the 2019 AUA Annual Meeting were searched to complement the systematic review. In cases where multiple reports were made on the same patient cohort, the most recent comprehensive publication was used for the analysis. To preserve fidelity of the analysis to outcomes in BCG-treated patients only, clinical trials including patients recurring after both BCG and non-BCG agents were excluded, unless a separate analysis was performed in the BCG recurrent subset. To accommodate for differences in study populations and disease descriptions before and after the definition of BCG-unresponsive NMIBC in 2015, all the trials accruing patients failing or recurring after induction or multiple courses of BCG, patients with both high-(HG) and low-grade (LG) recurrences, and patients with CIS only, papillary only, and any combinations thereof were captured. To assess specifically the efficacy of agents in the BCG-unresponsive setting, trials accruing patients satisfying the 2015 definition [2–4] were analyzed separately.

2.2. Study review methodology

Two authors (R.L. and D.S.) reviewed and selected studies independently; disagreements were resolved by discussion and consensus. Titles and abstracts were used to screen for initial study inclusion. Full texts of studies thought to meet or possibly meet the study inclusion were then reviewed. The same reviewers extracted relevant data independently using standardized data collection forms. Data retrieved from the reports include publication details (year of publication and authors), methodological components, and trial characteristics (sample size, therapeutic agent, and outcomes measures). The primary outcomes were stratified by tumor histology, using CRRs for studies limited to CIS-containing patients, RFRs for studies limited to patients with papillary disease, and disease-free rates (DFRs) for studies enrolling a combination of the two. CRR, RFR, and DFR were defined as the lack of a tumor or recurrence on cystoscopic evaluation with mandated/for-cause biopsies and negative urine cytology. Enhanced cystoscopy was not routinely used in any of the included studies. Secondary outcome measures included toxicity and progression-free rate (PFR) defined by progression to muscle-invasive or metastatic disease. In cases of comparative studies of more than one agent, outcomes in each treatment arm were separately extracted and analyzed.

2.3. Risk of bias assessment

A formal exclusion of studies due to risk of bias (RoB) assessment was not carried out as none of the existing RoB scales were felt to be appropriate for this systematic review. Many different factors could affect the outcome of the patients in the studies included in this systematic review. These and other limitations are presented in the Discussion. However, studies reported only in abstract form or not presenting complete outcome data were deemed to have a high RoB and excluded from the final analysis.

2.4. Data synthesis

Data synthesis was stratified by the histology of tumor after BCG, with CRRs reported for studies limited to CIS-containing patients, RFRs for studies limited to patients with papillary disease, and DFRs for studies enrolling a combination of the two. CRRs/RFRs/DFRs and PFRs reported at different time points in each study were collected and grouped by the post-treatment time intervals at which these were recorded. The median and range of the groups at 3, 6, 12, 18, and 24 mo following treatment were reported. Heterogeneity of study results was assessed by the inconsistency statistic (I²) and its connected chi-square test for heterogeneity. A meta-analysis was not feasible due to the extreme heterogeneity of the reported outcomes in each group. For illustrative purposes, the results were summarized using forest plots.

To further explore the efficacy of the bladder-sparing agents in different subgroups, separate analyses were carried out for studies accruing patients with BCG-unresponsive NMIBC [2–4]. Salvage agents were also broadly categorized as cytotoxic or immunogenic. CRRs/RFRs/DFRs of each drug type in the patient subsets were examined. No formal test was conducted for the purpose of subgroup comparisons and results were solely displayed in a descriptive manner.

3. Evidence synthesis

3.1. Literature search results

In total, 713 citations were retrieved from the databases and 12 additional abstracts were included from recent international conferences. After removing duplicates and screening 721 titles and abstracts, 619 citations were excluded from further analysis. The remaining 102 full texts were screened, of which 60 were excluded (Fig. 1). A total of 42 studies with 45 unique study arms (SAs) consisting of 2254 patients, with a median (interquartile range [IQR]) of 35 (18–47) patients were included for final analysis. Of these studies, three were randomized controlled trials, one was a nonrandomized comparative study, and 38 were single-arm studies. Ten of the 38 single-arm studies reported incomplete outcome data and/or were available in the abstract form only, and were deemed to have a high RoB. In total, 24 treatment options, including 19 i/ves, three oral, one intravenous, and one combination of oral and i/ves, were identified (Fig. 2). Therapeutic agents included i/ves BCG [6,7]; i/ves BCG + interferon alpha (IFNα) [8]; i/ves Mycobacterium phlei cell wall–nucleic acid complex (MCNA) [9]; i/ves oncolytic adenovirus CG0070 [10,11]; i/ves rAd-IFNα/Syn3 (Instiladrin) [12–14]; combination of i/ves ALT803 and BCG [15]; oral bropirimine [16]; intravenous pembrolizumab [17]; i/ves gemcitabine [6,7,18–25]; oral everolimus and i/ves gemcitabine [26]; combination of i/ves ALT-801 and gemcitabine [27]; i/ves mitomycin C (MMC) [19]; i/ves thermochemotherapy using MMC [28–30]; i/ves electromotive drug administration of MMC [31]; i/ves docetaxel [32]; combination of i/ves cabazitaxel, gemcitabine, and cisplatin [33]; i/ves nanoparticle albumin bound (nab-)paclitaxel [34,35]; i/ves paclitaxel-hyaluronic acid bioconjugate (Oncofid) [36,37]; i/ves valrubicin [38,39]; i/ves doxorubicin analog AD32 [40]; oral dovitinib [41]; i/ves oportuzumab monatox (Vicinium) [42–44]; oral sunitinib [45]; and i/ves photodynamic therapy (Fig. 2) [46,47].

Fig. 1 –

PRISMA flowchart of bladder-preserving treatments for recurrent non–muscle-invasive bladder cancer following intravesical BCG. BCG = bacillus Calmette-Guérin; PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

Fig. 2 –

Breakdown of study arms by different therapeutic agents and route administered. Cytotoxic agents are shown as red bars; immunogenic agents are shown as blue bars. BCG = bacillus Calmette-Guérin; EMDA = electromotive drug administration; IFNα = interferon alpha; IVe = intravesical; MMC = mitomycin C; PDT = photodynamic therapy.

3.2. Data extraction

Supplementary Table 1 lists the pertinent clinicopathological characteristics of the studies. The median/mean age ranged from 54 to 77 yr. A majority of the SAs (28/45) enrolled patients with papillary disease and/or CIS following BCG therapy, nine enrolled patients with CIS with/without papillary disease, one enrolled patients with CIS only, six enrolled patients with papillary disease only, and another did not specify (Table 1). All but nine SAs comprised only patients who had recurred after at least one course of induction BCG. Of the remaining nine SAs, six included a minority of BCG-intolerant patients (3–18%) [8,19,21,26,38] and three studies did not specify [27,36,45]. Many of the studies published prior to the introduction of the definition for BCG-unresponsive NMIBC did not specify the number of patients who satisfied this term. Instead, the percentage of patients who recurred after two or more courses of BCG, ranging from 31% to 100%, was reported in 27 SAs. Two SAs and one post hoc analysis included only patients satisfying the BCG-unresponsive NMIBC definition (Table 1) [12–14,48]. Of these, only the SA by Li et al [48] separately presented CRRs in the BCG-unresponsive patients with CIS-containing tumors.

Table 1 –

Treatment efficacy and progression-free rates in all studies included.

				% Failed 1 prior BCG	% Failed 2 prior BCG	% BCG intolerant	CIS vs papillary	3-mo CRR/RFR/DFR (%)	6-mo CRR/RFR/DFR (%)	12-mo CRR/RFR/DFR (%)	18-mo CRR/RFR/DFR (%)	24-mo CRR/RFR/DFR (%)	PFR (%)
Authors	Year	Agent	n
Nseyo et al [47]	1998	PDT (C)	36	100	50		CIS ± papillary	58.3		30.6			77.8
Sarosdy et al [16]	1998	Bropirimine (I)	47	100	100	0	CIS ± papillary	14.9					89.5
Dalbagni et al [20]	2002	Gemcitabine (C)	18	100	56	0	Both	38.9					94.4
Bassi et al [18]	2005	Gemcitabine (C)	9	100		0	CIS	44.4	44.4	22.2	22.2	11.1	100
Dalbagni et al [21]	2006	Gemcitabine (C)	30	90	67	10	Both			10.0			93.3
Joudi et al [8]	2006	BCG + IFNα (I)	467	100		8	Both					45.0	93.1
Gacci et al [7]	2006	Gemcitabine (C)	9	100	100	0	Papillary	100	66.7	44.4	11.1	0.0	80.0
Gacci et al [7]	2006	BCG (I)	10	100	100	0	Papillary	80.0	60.0	10.0	10.0	10.0	60.0
McKiernan et al [32]	2006	Docetaxel (C)	18	100			Both	55.6					100
Gunelli et al [22]	2007	Gemcitabine (C)	40	100	100		Papillary		95.0	77.5	70.0	70.0	100
Mohanty et al [25]	2008	Gemcitabine (C)	35	100		0	Papillary				60.0		91.4
Ignatoff et al [40] a	2009	AD32 (doxorubicin analog) (C)	42	100			Both			19.0		16.7	78.6
Addeo et al [19]	2009	MMC (C)	55	82		18	Papillary						89.1
Addeo et al [19]	2009	Gemcitabine (C)	54	85		15	Papillary						88.9
Kowalski et al [42]	2010	Oportuzumab monatox (Vicinium; C)	64	97	55		Both	39.3					100
Di Lorenzo et al [6]	2010	Gemcitabine (C)	40	100			Both					19.0	82.5
Di Lorenzo et al [6]	2010	BCG (I)	40	100			Both					3.0	67.5
Perdona et al [23]	2010	Gemcitabine (C)	20	100	100		Both	75.0	65.0	50.0	40.0	35.0	75.0
McKiernan et al [34]	2011	Nab-paclitaxel (C)	18	100			Both	27.8					94.4
Bassi et al [36].a	2011	Paclitaxel-hyaluronic acid (Oncofid-P-B; C)	15				CIS ± papillary			60.0			100
Garcia et al [45].a	2011	Sunitinib (C)	13					61.5					100
Burke et al [10]	2012	CG0070 (I)	35	100	≥50	0	Both	48.6					100
Kowalski et al [43]	2012	Oportuzumab monatox (Vicinium; ()	45	100		0	CIS ± papillary	44.0	26.7	15.6			95.6
Dinney et al [12]	2013	rAd-IFNα/Syn3 (I)	17	94	88		Both	41.2	41.2	29.4	29.4	23.5	88.2
Dinney et al [38] (A9303)	2013	Valrubicin (C)	78	96	39	4	CIS ± papillary	35.9	17.9	10.0		4.0	100
Dinney et al [38]/Steinberg et al [39]	2013/2000	Valrubicin (C)	90	100	70		CIS ± papillary		17.8	13.0			97.8
Lee et al [46]	2013	PDT (C)	34	100			Both	100		90.9		64.4	91.2
Skinner et al [24]	2013	Gemcitabine (C)	47	100	100		Both	46.8		27.7		21.0	93.6
Inman et al [28]	2014	MMC @ 42.2C (C)	15	100			Both	73.3	66.7	66.7	40.0	26.7	100
McKiernan et al [35]	2014	Nab-paclitaxel (C)	28	100	75		Both	35.7		35.7		30.6	96.4
Morales et al [9]/Li et al [48]	2015/2017	MCNA (I)	129	100	75		Both	69.0	34.1	22.5		14.7	78.3
Sonpavde et al [27].a	2015	IV ALT-801 + gemcitabine (C + I)	9				Both	22.2	22.2	22.2			100
Navai et al [13]	2016	rAd-IFNα/Syn3 (I)	7	100	100		Both	28.6	14.3	0.0		0.0	85.7
Hahn et al [41]	2016	Dovitinib (C)	13	100	100		Both	38.5	7.7	7.7			76.9
Shore et al [14]	2017	rAd-IFNα/Syn3 (I)	40	100	95		Both	57.5	42.5	35.0			95.0
Dalbagni et al [26]	2017	PO everolimus, intravesical gemcitabine (C)	23	100	68	11	CIS ± papillary	57.9	26.3	21.1			94.6
Packiam et al [11]	2018	CG0070 (I)	45	100			Both		46.7	29.5			95.6
Dickstein et al [44]a	2018	Oportuzumab monatox (Vicinium; C)	111	100	100		Both	49.5					100
Racioppi et al [31]	2018	EMDA-MMC (C)	26	100	31	0	Both	80.8	73.1	69.2			84.6
Balar et al [17]a	2019	Pembrolizumab (I)	103	100	100		CIS ± papillary	38.8					100
DeCastro et al [33]a	2017	Cabazitaxel, gemcitabine, cisplatin (C)	18	100	100		Both	94.4					100
Tan et al [29]	2019	MMC @ 42.2C (C)	48	100	63		Both					35.0	83.3
Tan et al [30]a	2019	MMC @ 42.2C (C)	87	100	100		Both			55.2		48.3	95.4
Hurle et al [37]a	2019	Paclitaxel-hyaluronic acid (Oncofid-P-B; C)	21	100			CIS ± papillary	85.7	76.2				100
Chamie et al [15].a	2019	BCG + ALT803 (I)	24	100	100		Both	70.8					100

BCG = bacillus Calmette-Guérin; C = cytotoxic; CIS = carcinoma in situ; CRR = complete response rate; DFR = disease-free rate; EMDA = electromotive drug administration; I = immunogenic; IFNα = interferon α; IV = intravenous; MCNA = M. phiei cell wall–nucleic acid complex; MMC = mitomycin C; PDT = photodynamic therapy; PFR = progression-free rate; RFR = recurrence-free rate.

^aStudies having a high risk of bias.

CRR assessment and recurrence monitoring generally started 3 mo after the initiation of therapy, with a few studies expediting the first evaluation to immediately after the completion of induction therapy between 1 and 8 wk after initiation [3,8,18,20–22,27,36,46]. Random bladder biopsies (RBBs) were mandated as part of the surveillance protocol in 20 of the studies (Supplementary Table 1). While a majority of the studies defined recurrence as either a positive biopsy or a urine cytology, several published after the adoption of the BCG-unresponsive definition enumerated only HG recurrences on biopsy as an event.

3.3. Complete response/recurrence-free rates

CRRs/RFRs/DFRs were variably reported at 3 mo (31 SAs), 6 mo (19 SAs), 12 mo (26 SAs), 18 mo (eight SAs), and 24 mo (19 SAs) after study initiation (Table 1). After exclusion of SAs with a high RoB, seven SAs accrued patients with CIS with/without papillary disease, six SAs accrued patients with papillary disease only, and 22 accrued patients SAs with both (Fig. 3). The median (range) CRRs in the seven SAs with CIS-containing patients were 43% (15–58%, n = 6) at 3 mo, 26% (18–44%, n = 5) at 6 mo, 17% (9–31%, n = 6) at 12 mo, 22% (22%, n = 1) at 18 mo, and 8% (4–11%, n = 2) at 24 mo (Fig. 4A and Supplementary Table 3). RFRs in the six SAs with papillary disease studies were 88% (80–95%, n = 2) at 3 mo, 67% (60–95%, n = 3) at 6 mo, 44% (10–78%, n = 3) at 12 mo, 36% (10–70%, n = 4) at 18 mo, and 10% (5–70%, n = 3) at 24 mo (Fig. 4B). DFRs in the 22 SAs enrolling both CIS and papillary patients were 51% (28–99%, n = 14) at 3 mo, 43% (8–73%, n = 9) at 6 mo, 29% (6–88%, n = 13) at 12 mo, 40% (29–40%, n = 3) at 18 mo, and 27% (6–62%, n = 9) at 24 mo (Fig. 4C).

Fig. 3 –

Diagram illustrating the stratification and inclusion of the analyzed studies. Studies were first stratified according to the tumor histology of patients accrued (CIS-containing tumors, papillary tumors, and both); those at a high risk of bias were then excluded. Studies were finally classified by the mechanism of the therapeutic agent. CIS = carcinoma in situ.

Fig. 4 –

Forest plots of complete response rates of CIS-containing studies, recurrence-free rates (RFRs) of papillary-only studies; and disease-free rates (DFR) of studies accruing both CIS and papillary disease patients at 3, 6,12,18, and 24 mo following treatment initiation excluding studies with a high risk of bias. BCG = bacillus Calmette-Guérin; CR = complete response; EMDA = electromotive drug administration; IFNα = interferon alpha; i/ves = intravesical; MCNA = M. phlei cell wall–nucleic acid complex; MMC = mitomycin C.

Of the studies analyzed, immunomodulatory agents were utilized in 10 SAs [6–14,16]. Stratified by tumor histology, Sarosdyetal [16] reported a 3-mo CRR of 15% after treatment using oral bropirimine in 47 patients with CIS-containing recurrences after failing two or more courses of BCG. In comparison, Gacci et al [7] reported a 3-mo RFR of 88%, but a 12-mo RFR of only 10% in 10 patients with recurrent papillary disease after failing two or more courses of BCG, who were then salvaged with reinduction BCG. In eight other SAs enrolling 780 patients with CIS and/or papillary recurrences (one SA i/ves BCG + IFNα, two SAs i/ves CG0070, one SA i/ves BCG, three SAs i/ves Instiladrin, and one SA i/ves MCNA), median (range) DFRs were 49% (29–69%, n = 4) at 3 mo, 41% (14–47%, n = 5) at 6 mo, and 29% (6–35%, n = 5) at 12 mo. Another 25 SAs evaluated the efficacy of cytotoxic salvage agents [6,7,18–26,28,29,31, 32,34,35,38,39,41–43,46,47]. Of these, six SAs consisting of 281 patients with CIS-containing recurrences (two SAs i/ves gemcitabine ± PO everolimus, two SAs i/ves valrubicin, one SA i/ves oportuzumab monatox, and one SA i/ves photodynamic therapy) demonstrated median (range) CRRs of 44% (36–58%, n = 5) at 3 mo, 26% (18–44%, n = 5) at 6 mo, and 17% (9–31%, n = 6) at 12 mo [18,26,38,39,43,47]. Another five SAs evaluated cytotoxic agents in 193 patients with papillary recurrences (one SA i/ves mitomycin and four SA i/ves gemcitabine), yielding median (range) RFRs of 100% (100%, n = 1) at 3 mo, 81% (67–95%, n = 2) at 6 mo, and 61% (44–78%, n = 2) at 12 mo [7,19,22,25]. Fourteen SAs evaluated cytotoxic agents in 419 patients with CIS and/or papillary recurrences (five SAs i/ves gemcitabine, three SA i/ves thermochemotherapy, one SA i/ves oportuzumab monatox, one SA i/ves photodynamic therapy, one SA i/ves docetaxel, two SAs i/ves paclitaxel, and one SA PO dovitinib), yielding median (range) DFRs of 51% (28–99%, n = 10) at 3 mo, 66% (8–73%, n = 4) at 6 mo, and 41% (8–88%, n = 8) at 12 mo.

Ten SAs enrolled patients having failed two or more courses of BCG [7,13,14,16,22–24,41,48]. Of those, three SAs were restricted to patients satisfying the BCG-unresponsive definition [13,14,48]. Another five SAs enrolled patients with LG recurrences after treatment with adequate BCG [7,22–24]. Baseline clinicopathological information and treatment efficacy results are summarized in Table 2. RFRs/DFRs at 12 mo ranged from 0% to 35% in the BCG-unresponsive trials. In a separate analysis of the CIS-containing cohort in the MCNA trial, CRRs were found to be 45% at 6 mo and 27% at 12 mo after treatment [48].

Table 2 –

Treatment efficacy in BCG-unresponsive trials (top) and all patients following adequate BCG (bottom).

Authors	Year	Phase	Agent	n	Age	M (%)/F (%)	CIS vs papillary	Tumor stage			Tumor grade			Biopsy required	Recurrence definition	3-mo CRR/RFR/DFR (%)	6-mo CRR/RFR/DFR (%)	12-mo CRR/RFR/DFR (%)	18-mo CRR/RFR/DFR (%)	24-mo CRR/RFR/DFR (%)
								Ta (%)	T1 (%)	CIS (%)	Low (%)	Int (%)	High (%)
Li et al [48]	2017	II	MCNA	94	67.9	69/31	Both	42	4	72	0	0	100	Yes	HG recurrences		48.9	34.8		28.3
Li et al [48] (CIS cohort)				68						100					HG recurrences		44.8	26.5		16.6
Navai et al [13]	2016	Ib	Instiladrin	7	76.2	100/0	Both	57	14	71	0	0	100	No	Positive biopsy or cytology	28.6	14.3	0.0		0.0
Shore et al [14]	2017	II	Instiladrin	40	70.5	83/17	Both	20	28	75	0	0	100	Yes	HG recurrences	57.5	42.5	35.0
Gacci et al [7]	2006	II/III	Gemcitabine	9	75	78/22	Papillary	44	56	0	11	33	56	No	Positive biopsy or cytology	100.0	66.7	44.4	11.1	0.0
Gacci et al [7]	2006	II/III	BCG	10	75.5	80/20	Papillary	70	30	0	10	60	30	No	Positive biopsy or cytology	80.0	60.0	10.0	10.0	10.0
Gunelli et al [22]	2007	II	Gemcitabine	40	66	95/5	Papillary	10	90	0	0	53	43	No	Positive biopsy or cytology		95.0	77.5	70.0	70.0
Perdona et al [23]	2010	II	Gemcitabine	20	68.3	65/35	Both	20	80	35	25	0	75	No	Positive biopsy	75.0	65.0	50.0	40.0	35.0
Skinner et al [24]	2013	II	Gemcitabine	47	70	70/30	Both	36	4	60	11	0	89	Yes	Positive biopsy or cytology	46.8		27.7		21.0
Dinney et al [12]	2013	I	Instiladrin	17	72	94/6	Both	59	6	65	12	0	88	Yes	Positive biopsy or cytology	41.2	41.2	29.4	29.4	23.5

BCG = bacillus Calmette-Guérin; CIS = carcinoma in situ; CRR = complete response rate; DFR = disease-free rate; HG = high grade; Int = intermediate; MCNA = M. phlei cell wall–nucleic acid complex; M/F = male/female; RFR = recurrence-free rate.

3.4. Progression-free rate

Overall, median follow-up in all studies excluding those with a high RoB was 17 mo (IQR 12–28.1 mo), with the median (range) PFR of 91% (60–99%; Fig. 5). Stratified by tumor histology, median PFRs were 96% (78–99%, n = 7) in SAs enrolling patients with recurrent CIS with/without papillary disease at a median follow-up of 12 mo (IQR 12 mo), 89% (60–99%, n = 6) in those enrolling patients with recurrent papillary disease only at a median follow-up of 19.9 mo (IQR 16.8–30 mo), and 94% (68–99%, n = 22) in studies enrolling both CIS and papillary patients at a median follow-up of 17 mo (IQR 9–29.6 mo).

Fig. 5 –

Comparison of 3-, 6-, 12-, 18-, and 24-mo progression-free rates (PFRs) in all study arms excluding those with a high risk of bias. BCG = bacillus Calmette-Guérin; IFNα = interferon alpha; i/ves = intravesical; MCNA = M. phlei cell wall–nucleic acid complex; MMC = mitomycin C.

3.5. Toxicity

Toxicity of i/ves treatments were relatively mild, with grade 3 or higher adverse events (AEs) ranging from 0% to 25%. In total, 23 dose limiting toxicities (DLTs) occurred in 2046 evaluable patients. Most of the AEs due to i/ves therapy were related to urinary bother, including dysuria, frequency, and hematuria. Notably, three episodes of lymphopenia were observed in patients treated with i/ves gemcitabine. Expectedly, severe toxicities were more frequently seen in trials using systemic therapies. A total of 18 DLTs occurred in 208 patients. Toxicities in these trials included hepatotoxicity, metabolic disorders, and death due to severe colitis.

3.6. Discussion

Despite the well-documented unmet need for bladder-sparing therapies in NMIBC patients recurring after i/ves BCG, innovation and progress have been hampered by the lack of consensus over trial design and primary study endpoints. Owing to the complex scheduling of BCG infusions, nuances in the timing, grade, and histology of recurrent tumors may reflect biological heterogeneity, leading to different prognosis and levels of response to salvage agents. To facilitate accurate interpretation of trial results, the term BCG-unresponsive NMIBC was developed to more precisely define the patient cohort refractory to i/ves BCG therapy in whom alternate treatment modalities should be sought [2,3]. Although subsequently incorporated into the FDA investigational guidelines for developing drugs and biologics for the treatment of BCG-unresponsive NMIBC [4], many questions still remain.

Foremost, there exists no data-driven efficacy benchmark for salvage bladder-sparing therapy. The only FDA-approved agent valrubicin was felt to provide inadequate cancer control, demonstrating a 12-mo CRR of only 13% [38]. Lacking standards, the expert panel at the public workshop cohosted by the AUA and FDA felt that “an initial response rate of 40–50% at 6 mo and a durable response rate of at least 30% for 18–24 mo with the lower bound of 95% confidence interval excluding 20% should be clinically meaningful” [1]. Our study demonstrated that, to date, bladder-sparing therapies achieved modest efficacy in NMIBC patients having recurred after BCG. Median CRRs in the treatment of CIS-containing tumors were 26% at 6 mo, 17% at 12 mo, and 8% at 24 mo after treatment. In comparison, median RFRs in the papillary-only studies were 67% at 6 mo, 44% at 12 mo, and 10% at 24 mo. Specifically in the BCG-unresponsive population, 6- and 12-mo CRRs in the CIS-containing patients from the MCNA trial were 45% and 27%, respectively. Median PFRs were 91% at a median follow-up of 17 mo, 95% in the CIS-containing studies (median 12-mo follow-up), and 89% in studies restricted to papillary-only recurrences (median 19.9-mo follow-up).

Although formal statistical comparison was not feasible due to clinical and study design heterogeneity, an interesting finding emerging from the current study was that patients with papillary-only recurrences were more effectively treated with adjuvant bladder-sparing therapy than patients with CIS-containing tumors. This pattern was observed across studies using both immunogenic and cytotoxic agents (Supplementary Table 3). The FDA specified that the primary endpoints of CRR and DoR in trials to support marketing applications should be assessed only in patients with active disease at the time of trial enrollment (ie, CIS-containing patients) [4]. Although pretreatment concomitant CIS has been demonstrated to predict poor progression-free survival and cancer-specific survival following induction BCG [49,50], its relevance in treating BCG recurrent NMIBC is relatively unknown. The differences seen in the salvage rates of recurrent papillary versus CIS-containing tumors point to possible intrinsic biological differences between these disease subtypes. On the contrary, differences seen in the early treatment response may merely reflect the ability to eradicate visible papillary tumor by transurethral resection. Either way, results herein support the FDA’s recommendation to stratify the analysis of trial results based on the type of disease at trial enrollment [4]. The relative inefficacy in treating CIS-containing tumors should also be considered in the discussions surrounding marketing applications of novel agents.

Another challenge in applying the BCG-unresponsive definition in clinical trials involves its restrictions on the timing of recurrence and the requirement for RBBs prior to enrollment. Despite the FDA’s efforts to relax the inclusion criteria, many patients fall out of the window for trial enrollment due to delays in referrals after being diagnosed with BCG-unresponsive disease. Kamat et al [51] described another clinical scenario where a BCG-unresponsive patient, having failed a course of salvage agent, will no longer qualify for trial enrollment due to the time elapsed since the receipt of the last BCG therapy. They suggested that these patients should qualify for clinical trials on the basis of previously being deemed BCG unresponsive. Moreover, mandatory bladder mapping biopsies prior to the initiation of salvage therapy in patients already diagnosed with CIS yield little additional prognostic information and can further delay trial enrollment. These inclusion criteria will likely need to be fine-tuned as additional BCG-unresponsive trials get underway.

A third factor that may introduce a bias into the interpretation of CRR pertains to the definition used for recurrence and the varying methods for its detection. Most commonly, studies have relied on a combination of cystoscopy and urine cytology to detect i/ves recurrences. However, as CIS and incipient papillary disease are notoriously elusive on cystoscopy and urine cytology [52,53], variations from trial to trial in the requirement of RBBs can lead to differences in CRRs/RFRs. In a recently completed trial using rAd-IFNα/Syn3, five patients with negative cystoscopy and cytology were found to have CIS on RBBs, underscoring the importance of incorporating RBBs into the surveillance monitoring protocol [54]. On the contrary, although specificity of urine cytology was as high as 86% [55], false positive results can occur and histological evidence of an HG tumor should be confirmed prior to enumerating cancer recurrence. Lastly, the use of enhanced cystoscopy has been demonstrated to improve detection rates of NMIBC [56]. The effects of its adoption in the assessment of treatment efficacy of bladder-sparing drugs in the BCG-unresponsive setting remain unclear.

Traditionally, patients with LG recurrences following BCG were included in bladder-sparing salvage trials. Compared with such trials encompassing patients with LG recurrences following adequate BCG, treatment efficacy was inferior in the bona fide BCG-unresponsive patients. This observation points to possible differences in the disease course following LG versus HG recurrences, corroborating with the FDA guidelines restricting BCG-unresponsive NMIBC to HG recurrences only. By contrast, there is emerging evidence that up to 14.4% of patients with LG recurrence will progress to muscle-invasive or metastatic disease [57]. As such, although LG recurrences do not constitute recurrences by the current guidelines, these patients still require close monitoring lest there is disease progression.

The study is not without limitations. First, in order to preserve the fidelity of our analysis to only patients having failed BCG, the scope of this systematic review was limited by excluding several clinical trials enrolling patients after failing both BCG and non-BCG agents. Second, due to the vast clinical and study design heterogeneity, a meta-analysis of the combined results was not feasible. As such, we used the narrative synthesis method recommended by the European Association of Urology to present the results of the systematic review [58]. Additionally, as some of the studies have been conducted over 2 decades ago, obtaining individual data for an individual patient data meta-analysis was not possible. Furthermore, as many of the studies were published prior to the formalization of the BCG-unresponsive NMIBC definition, patient enrollment and interpretation of results were affected as outlined in the Discussion. Relatively few studies were available that strictly conformed to the formal BCG-unresponsive definition, in which treatment efficacy was expectedly inferior to those in studies conducted with BCG recurrent patients overall. These results are in line with previous studies demonstrating the relative inefficacy of continued i/ves treatment in the BCG-unresponsive population [59]. Finally, response rates were reported at nonuniform time intervals after treatment initiation.

Nevertheless, results from the current study will serve as a frame of reference for emerging trial results in the BCG-unresponsive space. For instance, recently presented results from the phase 3 rAd-IFNα/Syn3 trial for BCG-unresponsive NMIBC demonstrated 3-, 6-, and 12-mo CRRs of 53.4%, 40.8%, and 24.3%, respectively, in 103 CIS-containing patients and RFRs of 72.9%, 58.3%, and 43.8%, respectively, in 48 patients with papillary disease [54], in line with results from previous trials conducted in the BCG-unresponsive population. In addition, 17% of 97 BCG-unresponsive CIS-containing patients treated with pembrolizumab had a durable complete response of ≥12 mo in the KEYNOTE-057 trial [60]. With the accumulation of efficacy data for bladder-sparing salvage agents tested in the precisely defined BCG-unresponsive population, more robust analytic methods can be applied to establish the efficacy benchmark to guide future trials.

4. Conclusions

We demonstrate that, to date, bladder-sparing therapies achieved modest efficacy in patients with NMIBC after BCG, with median 6-, 12-, and 24-mo CRRs of 26%, 17%, and 8%, respectively, in CIS-containing patients. Recurrent papillary disease was more effectively treated, with median 6-, 12-, and 24-mo RFRs of 67%, 44%, and 10%, respectively. The median PFR was 91%. Toxicities of i/ves agents were generally mild, with very few DLTs. Results from the current study will serve as a frame of reference for emerging trial results in the BCG-unresponsive space.