A Phase 2 Study of Durvalumab for Bacillus Calmette-Guerin (BCG) Unresponsive Urothelial Carcinoma In Situ of the Bladder

Abstract

Purpose:

Immune checkpoint blockade holds promise for treating BCG-unresponsive non-muscle invasive bladder cancer. In this phase II study, we investigated the safety and efficacy of durvalumab, a human IgG1 monoclonal antibody, against BCG-unresponsive CIS.

Patients and Methods:

Patients with BCG-unresponsive CIS containing non-muscle invasive bladder cancer received durvalumab IV at 1500 mg every 4 weeks for up to 12 months. The primary endpoint was complete response rate at month 6, defined by negative cystoscopy, urine cytology and absence of high-grade recurrence on bladder mapping biopsy. The null hypothesis specified a complete response rate of 18% and alternative hypothesis 40%. According to the Simon 2-stage design, if ≤3/13 patients achieved complete response during stage 1, the trial is stopped due to futility.

Results:

Between 03/8/2017 and 1/24/2020, 17 patients were accrued while 4 withdrew from study treatment after bladder biopsy at month 3 was positive for CIS. 2 of 17 (12%) achieved a complete response at month 6, with duration of response of 10 and 18 months, respectively. A single Grade 3 lipase elevation was attributed to durvalumab, and immune related adverse events were observed in 7/17 (41%) patients. Only 1/17 patients had high PD-L1 expression pre-treatment. On RNAseq, complement activation genes were elevated post-treatment, along with enrichment of tumor associated macrophage signature.

Conclusions:

Durvalumab monotherapy conferred minimal efficacy in treating BCG-unresponsive CIS of the bladder, with 6mo complete response of 12%. Complement activation is a potential mechanism behind treatment resistance.

Statement of Translational Relevance

Current non-extirpative treatment options for BCG-unresponsive NMIBC are limited. This open-label, single-arm, Phase 2 study investigated the efficacy of durvalumab for patients with BCG-unresponsive CIS. PD-L1 Blockade by Durvalumab may alleviate immunosuppression by PD-L1 and allow for heightened activity of cytotoxic T cells. We report the safety, efficacy, and gene expression correlates of Durvalumab monotherapy. While efficacy was limited with only 2/17 patients achieving the primary endpoint of complete response at 6 months, upregulation of genes involved in complement-activation post-treatment suggest a mechanism of resistance to Durvalumab treatment.

INTRODUCTION

For more than 4 decades, intravesical Bacillus Calmette-Guérin (BCG) has been the most effective therapy for non-muscle invasive bladder cancer (NMIBC). Despite its success, recurrence rates range between 26% to 42% and progression rates between 8% to 13%.^1–3 In 2016, the term BCG-unresponsive NMIBC was coined by a consensus panel defined as persistent high grade NMIBC within 6–12 months despite adequate BCG therapy.⁴ For patients with BCG-unresponsive NMIBC⁴, radical cystectomy (RC) with pelvic lymph node dissection remains the standard-of-care.^5,6 Unfortunately, RC is fraught with high perioperative morbidity and is unfit for many patients who develop recurrence following BCG. Several bladder sparing agents have been tested in this setting with mixed results.⁷ Of these, pembrolizumab and nadofarogene firadenovec gained FDA approval in patients with BCG-unresponsive NMIBC with CIS based on observed 3-month CR rates of 41% and 53%, and median response duration of 16.2mo and 9.7mo, respectively.^8,9

Durvalumab, a human IgG1 monoclonal antibody that selectively binds to PD-L1, has been shown to antagonize the inhibitory effect of PD-L1 on primary human T cells, resulting in their restored proliferation and release of IFN-γ. Additionally, durvalumab has demonstrated antitumor activity in a xenograft model via T-lymphocyte dependent mechanisms.¹⁰ In a Phase 1/2 open-label study of 191 patients with locally advanced/metastatic urothelial cancer, confirmed ORR was 17.8%. Based on evidence showing adaptive immune resistance thwarting responsiveness to BCG¹¹ as well as the success achieved using pembrolizumab to modulate the PD-1/PD-L1 axis in metastatic bladder cancer, we posited that durvalumab could induce a durable clinical response in patients diagnosed with BCG-unresponsive CIS-containing NMIBC. In this study, we aimed to evaluate 6- and 24-mo CR rates following durvalumab treatment.

MATERIALS AND METHODS

Trial design and participants

This open-label, single-arm, Phase 2 study was performed at Moffitt Cancer Center, enrolling patients with BCG unresponsive CIS containing NMIBC of the bladder for treatment using Durvalumab to evaluate its antitumor efficacy. Eligible patients were aged 18 years or older at the time of written informed consent and met the definition of BCG-unresponsive CIS-containing NMIBC. This definition includes patients with persistent CIS-containing NMIBC at 6 months despite adequate BCG therapy, which is defined as ≥5 of 6 doses of an initial induction course plus ≥2 of 3 doses of maintenance therapy or ≥2 of 6 doses of a second induction course. It also includes those with recurrence of CIS containing NMIBC within 9 months of disease-free state after previous adequate BCG treatment as well as patients with T1HG with concomitant CIS following induction BCG according to the consensus BCG unresponsive definition.^4,12 All patients with concomitant papillary tumors had undergone complete TURBT, with those having T1 disease undergoing re-TURBT to confirm diagnosis. No intervening intravesical therapies from the time of most recent cystoscopy or TURBT to study initiation was allowed. Eligible patients had an Eastern Cooperative Oncology Group perform status of 0–1 and adequate organ function. Laboratory tests (i.e. hematology and comprehensive biochemistry panel) were required within 28 days prior to study enrolment. Patients with any history of muscle invasive urothelial carcinoma, evidence of upper urinary tract disease, lymphovascular invasion, or hydronephrosis due to tumor in the presence of T1 disease were excluded. Also excluded were patients who received investigational agent or anti-cancer therapy within 4 weeks before study drug initiation; previous treatment using a PD-1 or PD-L1 inhibitor; patients with unresolved adverse events from a previously administered agent; patient with a history of HIV infection, active hepatitis B or C infection; and any patients with immunodeficiency or with a history of pneumonitis.

The protocol and accrual timeline were designed by the trial investigators and the study done according to the Declaration of Helsinki, in compliance with Good Clinical Practice Guidelines. The study protocol was approved by the institutional review board at Moffitt before accrual was initiated. The study procedures and analyses were done per protocol and all protocol amendments were approved by the institutional review board before implementation.

Procedures

Screening procedures, including urine cytology with UroVysion test, CT scan, and chest X-ray were performed within 28 days from the start of treatment. All pathologic and cytologic specimens were evaluated by board certified genitourinary pathology specialists at Moffitt Cancer Center. Formalin-fixed, paraffin-embedded tumor samples pre- and post-treatment were submitted for PD-L1 assessment using the VENTANA PD-L1 (SP263) assay (Ventana Medical Systems, Tucson, AZ, USA).¹³ Cystoscopy with bladder mapping biopsy was performed under general anesthesia at baseline and month 6 and 24 for treatment efficacy assessment. Durvalumab was given at the FDA-approved dose of 1500 mg IV every 4 weeks for up to 12 months in patients without disease recurrence. All patients were evaluated for recurrence using urine cytology and cystoscopy with for-cause biopsy performed every 3 months during the first 12 months (treatment period) and planned for every 4 months in the second 12 months (surveillance period). Patients with persistent CIS at month 3 bladder biopsy were continued on treatment until month 6 bladder biopsy due to delayed response to immunotherapy.^14,15 Patients were taken off the study if any of the biopsies documented invasive (T1 or above) urothelial carcinoma at any time or if persistent or recurrent CIS was detected at or beyond 6 months. A positive urine cytology along with negative cystoscopic findings prompted further investigation of the upper urinary tract and random bladder biopsies at the discretion of the treating physician. Monitoring of disease status continued until recurrence, disease progression, cystectomy, initiation of a new anti-cancer therapy, withdrawal of consent, or death. All patients were followed for survival status until death or withdrawal of consent.

Safety assessments (monitoring of adverse events and immune-related adverse events; clinical laboratory evaluations including hematology, urinalysis, chemistry panel, thyroid function tests, amylase, and lipase; vital signs; physical examination; and symptoms review) were done on day 1, then every 4 weeks during the treatment phase, and every 4 months during the surveillance phase. Adverse events were graded according to National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.03. Immune-related adverse events were defined by the use of a list of terms specified by the principal investigator and included regardless of attribution to study treatment or immune relatedness. Any abnormalities and noteworthy changes from baseline were thoroughly investigated.

Outcomes

The primary endpoint was the proportion of patients achieving complete response at month 6, defined by negative cystoscopy, urine cytology and absence of high-grade disease as assessed by bladder mapping biopsies. Secondary endpoints included the proportion of patients with durable complete response at year 2 based on findings on surveillance cytology, cystoscopies, and mapping biopsy examination (if clinically indicated or mandated) as well as on urine cytology, cystoscopy and end-of-study bladder mapping biopsies; duration of response for high-grade disease; progression-free survival to muscle-invasive or metastatic disease or death (defined as time from enrolment to muscle-invasive or metastatic disease or death from any cause, whichever occurred first); overall survival (defined as time from enrolment to death from any cause); and safety in all treated patients. Exploratory endpoints included the correlation between PD-L1 expression levels (as assessed using the VENTANA PD-L! SP263 Assay Roche, catalog no. 790–4905, RRID: AB_2819099), chromosomal abnormalities found on UroVysion test, as well as tumor mutational burden and alterations in DNA repair genes in the pre- and post-treatment samples with treatment response.

RNA/DNA Sequencing

RNA-sequencing libraries were prepared using the Illumina TruSeq RNA Exome Library Preparation Kit (llumina Inc., San Diego, CA). Following library quantitation, the libraries were sequenced on the NextSeq 2000 to generate 25–30 million pairs of 100-base reads per sample. Transcriptomic analysis was performed by aligning reads to hg19 using STAR-2.5.3a¹⁶. Gene-level quantification was determined using HTSeq v0.6.1 by summation of raw counts of reads aligned to the region associated with each gene. Differential expression analyses were performed using DESeq2_1.6.3.

Whole-exome sequencing was performed in order to identify somatic mutations in the coding regions of the human genome. Per sample, 200ng of DNA was used as input into the Agilent SureSelect XT Clinical Research Exome kit, with increased coverage at 5000 disease-associated targets (Agilent Technologies, Santa Clara, CA). Approximately 75 million 100-base paired-end sequences per sample were generated on an Illumina NextSeq 2000 sequencer. For whole-exome DNA-Sequencing, sequence reads were adaptor trimmed and aligned to the reference human genome hg19 with the Burrows-Wheeler Aligner (BWA v0.7.7)¹⁷ and duplicate identification, insertion/deletion realignment, quality score recalibration, and variant identification were performed with PICARD and the Genome Analysis ToolKit (GATK) v2.2¹⁸. All genotypes were determined across all samples at variant positions using GATK. Various quality control measures were applied to determine depth of coverage in each sample across the targeted genes. Sequence variants were annotated using ANNOVAR.

Statistical analysis

The null hypothesis is that the historical 18% CR rate at month 6 observed using valrubicin¹⁹ will be tested against a one-sided alternative hypothesis of ≥40% CR rate using Durvalumab. Simon’s two-stage optimal design²⁰ was used for this Phase II single arm clinical trial. The maximum sample size is 34, with a sample size of 13 patients in the first stage. If 3 or fewer responses are observed during stage 1, the trial was stopped early for futility. This design yielded a one-sided type I error rate of 0.047 and power of 80% when the true response rate is 40%. Discrete variables were summarized using proportions along with 95% confidence intervals by the Pearson-Clopper method. The Kaplan-Meier method was used to summarize time-to-event outcomes and median survival time with 95% confidence intervals. Median follow-up time was calculated using the reversed Kaplan-Meier method.

Data Availability Statement

The data generated in this study are available within the article and its supplementary data files. Raw data for this study were generated at the Molecular Genomics Core at Moffitt Cancer Center. Derived data supporting the findings of this study are available from the corresponding author upon request.

RESULTS

Patient Characteristics and Efficacy Outcomes

Between Mar. 8, 2017, and Jan. 24, 2020, 17 patients were enrolled and received at least one dose of durvalumab. Of these patients, 4 withdrew from study treatment after month 3 bladder biopsy demonstrating persistent CIS, leaving 13 evaluable patients. The data cutoff date was 4/18/2022. All patients were included in both the efficacy and safety analyses. At enrolment, median age was 77 (IQR 69–79). The patients were predominantly male (88%); had a median of 22.4-month (IQR 16.4–35.7) history of bladder cancer; and received a median of two courses of intravesical BCG (Table 1). All patients were diagnosed with BCG unresponsive CIS, with one having concomitant T1HG papillary disease.

Table 1:

Baseline Patient Characteristics

		Full Cohort (N=17)
Age, (IQR)		77 (58–91)
Sex, No. (%)	Male	15 (88.2)
	Female	2 (11.8)
Race, No. (%)	White	17 (100)
BMI, (IQR)		28.3 (25.9–32.3)
Number of Previous Courses of BCG (%)	2	9 (52.9)
	3	3 (17.6)
	4	2 (11.8)
	5+	3 (17.6)
Smoking History No. (%)	No	8 (47.1)
	Yes	9 (52.9)
Stage at Baseline	Carcinoma in Situ	16 (94.1)
	Carcinoma in Situ with T1	1 (5.9)
PD-L1 Status	Combined Positive Score < 25	16 (94.1)
	Combined Positive Score ≥ 25	1 (5.8)

Median follow-up from the time of enrolment to data cutoff was 36.4 months (IQR 20.5–50.9). Overall, 2/17 (12%; 95% CI 1.5%−36.4%) patients achieved a complete response at month 6 as assessed by cystoscopy, urine cytology and bladder mapping biopsy, with duration of response of 10 and 18 months, respectively (Fig. 1a, b). Overall, 7/17 (41%; 95% CI 18.4%−67.1%) patients were found to be disease-free on cystoscopy and urine cytology at month 3, all but 2 of whom had recurrent high grade NMIBC at month 6. Another patient had a positive urine cytology and negative cystoscopy at month 3 and was subsequently confirmed to have persistent CIS on month 6 mapping biopsy. One patient was taken off the study after finding T1HG NMIBC at 3-month evaluation. All other patients (8/17, 47%) had persistent Ta HG or CIS at the 3-month evaluation.

Figure 1:

(A) Swimmer plot showing Time following the Start of Durvalumab Treatment (B) High grade recurrence free survival, defined as the length of time after treatment until high grade bladder cancer recurrence or death, is shown.

A total of 9 patients underwent radical cystectomy, 6 had residual pTis, 1 upstaged to pT1, and 2 progressed to muscle invasive disease (≥pT2), one of whom also harbored concomitant lymph node metastases (pN2). An additional patient was found to have lymph node metastases approximately 6 months following treatment failure within the bladder. Altogether, 4/17 (24%; 95% CI 6.81%−49.9%) patients had progression to muscle invasive disease, metastasis, or death event, with a median time-to-progression of 14 months (95% CI 9-NR). A total of 3 deaths occurred over the study period, making 24-month overall survival to be 82.3% (95% CI 57.7%−100%) in all patients. Of these, 2 were deemed to be related to bladder cancer.

Safety Outcomes

The median duration of durvalumab treatment was 4.5 months (IQR 3.9–6.0), with a median of 7 administrations (IQR 6–7) per patient. Two of 17 patients (12%, 95% CI 14.5%−36.4%) completed the 12-month treatment course. Treatment discontinuation in 15/17 patients were due to persistent disease found at either 3- or 6-month evaluation. No patient discontinued treatment due to toxicities.

Treatment-emergent adverse events (TEAEs) were observed in all 17 patients. Summary of Grade 1–2 TEAEs that occurred more than twice and all Grade 3–4 TEAEs were summarized in Table 2. Interruptions in drug administration due to TEAEs occurred in two patients: one related to asymptomatic elevation of amylase and lipase, and the other related to supratherapeutic INR while on warfarin. Five patients experienced immune-related adverse events, the most common of which were fatigue (n=7, 41.2%), diarrhea (n=5, 29.4%) and arthralgia (n=4, 23.5%). Grade 3 elevated lipase level was the only grade 3 or above immune related adverse event. No patient required systemic corticosteroids to treat immune related AEs. The Grade 3 lipase elevation was self-limiting after one week off treatment and did not recur after resuming treatment.

Table 2.

Summary of grade 1–2 treatment emergent adverse events (TEAEs) that occurred for more than 2 occasions and all the grade 3–4 TEAEs.

	Grade 1–2	Grade 3–4	Attribution
Fatigue	7 (41.2%)	0	related
Diarrhea	5 (29.4%)	0	related
Arthralgia	4 (23.5%)	0	related
Urinary tract infection	4 (23.5%)	0	unrelated
Cough	3 (17.6%)	0	unrelated
Hypothyroidism	3 (17.6%)	0	related
Amylase increased*	3 (17.6%)	0	related
Hypertension	0	2 (11.8%)	unrelated
INR increased**	2 (11.8%)	1 (5.88%)	unrelated
Hematuria**	0	1 (5.88%)	unrelated
Leukocytosis	0	1 (5.88%)	unrelated
Anemia**	0	1 (5.88%)	unrelated
Lipase increased*	0	1 (5.88%)	related

^*These AEs occurred in the same individual with no clinical or CT evidence of pancreatitis.

^**These AEs occur in an individual who is on warfarin for rate controlled atrial fibrillation.

Biomarker Analyses

PD-L1 expression level has been linked to immune checkpoint blockade (ICB) treatment response in the neoadjuvant²¹, adjuvant²², and metastatic settings²³ for urothelial carcinoma, but has not been correlated to response in the treatment of BCG unresponsive NMIBC. Of the 17 pre-treatment samples, only one (6%, CI 0.1%−28.7%) was found to be PD-L1 high (at least 25% of tumor cells or immune cells staining for PD-L1 at any intensity). Following durvalumab treatment, this patient was found to have persistent multifocal CIS at 3-month evaluation. In contrast, 3/8 (37.5%, 95% CI 8.5%−75.5%) of the post-treatment samples were found to have high PD-L1 expression, suggestive of developing adaptive immune resistance with treatment.

Tumor mutational burden (TMB) was estimated using DNA extracted from 12 pairs of pre- and post-treatment samples. Median TMB in the pre-treatment samples was 8.0 mut/Mb, and remained the same following treatment. On transcriptomic analysis, expression levels of genes related to complement activation were markedly increased following treatment (Fig. 2a). Cellular deconvolution using EPIC²⁴ demonstrated enrichment of the macrophage signature within the tumor microenvironment (TME) following treatment (Fig. 2b). Accordingly, CCL2, an important monocyte recruitment chemokine, was elevated in the post-treatment samples (Fig. 2c). In contrast, the 12-CK score, which has previously been linked to the formation of tertiary lymphoid structures and associated with favourable response to ICB did not increase. Finally, IL-10 levels were increased following treatment, suggestive of macrophage mediated suppression of CD8+ T cell mediated anti-tumor response (Fig. 2d).^25,26

Figure 2:

(A) Heatmap showing clustering of genes by expression level by (left) pre and (right) post-treatment. Line plots noting upregulation in gene expression for (B) Macrophages, (C) CCL2, and (D) IL-10

DISCUSSION

Despite renewed enthusiasm following alignment between the FDA and the bladder cancer research community on the definition, endpoints and trial design for drug testing against BCG Unresponsive NMIBC²⁷, pembrolizumab and nadofaragene firadenovec remain the only approved agent in this space over the last quarter century. Notwithstanding its approval, adoption of pembrolizumab for the treatment of this disease remains guarded due to its modest efficacy and significant toxicity. For CIS containing BCG unresponsive NMIBC, 6-month pooled average CR rates from 3 separate clinical trials was 38%⁷, falling short of the aspirational response rate of 50% proposed by the International Bladder Cancer Group (IBCG).⁴ Even against this backdrop, the 6mo CR rate of 12% achieved by durvalumab treatment in the current study was disappointing, lower than rates previously described using pembrolizumab²³ and atezolizumab.²⁸ Furthermore, durability of response was poor, with no patient maintaining CR at 24mo. Alarmingly, progression rate of 24% observed in the current trial was higher than reported in previous BCG unresponsive trials.^8,9

One explanation could be the low rate of PD-L1 positivity seen in the pre-treatment samples. Albeit measured by different assays, PD-L1 levels have been consistently linked to treatment response in clinical trials using immune checkpoint blockade to treat bladder cancer in various clinical settings.^21–23 This was especially true in trials involving durvalumab, in which the disparity in ORR was as high as 22.5% between the PD-L1-positive vs. negative patients. Conflicting results regarding the effect of BCG treatment on PD-L1 status have emerged from different series, with some showing increased expression²⁹ vs. others showing no change.¹¹ These results may be reconciled due to different BCG regimens and/or methods of PD-L1 assays. Alternatively, Rouanne et al³⁰ described distinct mechanisms of immune escape following BCG, characterized by tumors with different HLA-I and immune exhaustion marker expression. Our results demonstrate lower than expected PD-L1 positivity, pointing to a possible over-representation by the HLA-I deficient tumors in our cohort, which may have led to the poor rate of response seen with durvalumab.

Additionally, comparing bulk RNAseq data pre- and post-treatment clearly demonstrated activation of the complement pathway, highlighted by upregulation of expression of C1QC, C3AR1, and C5AR1 (Fig. 2a). However, the complement system is known to play a very complex role in anti-tumor immunity that is context dependent: whereas it may act to kill antibody-coated tumor cells, it may also support chronic inflammation, thus hampering antitumor T cell response leading to tumor progression.³¹ Complement effectors have been implicated in directly stimulating tumor proliferation³² and metastasis³³; promoting angiogenesis³⁴; and altering various aspects of the tumor immune contexture^35,36. Specifically, Bonavita et al³⁵ demonstrated complement induced carcinogenesis orchestrated by increased production of the monocyte attracting chemokine CCL2 and subsequent recruitment of TAM. From our data, both CCL2 and cellular deconvolution signature for TAM were elevated in the post-durvalumab treated samples, corroborating the proposed mechanism of complement associated disease progression. Several monocytes markers/receptors such as CD14, CD163, FCGR3A, and FPR1 were also found to be elevated post-treatment. Furthermore, the protumorigenic properties of TAMs were shown to be mediated through the suppression of CD8+ T cells with the production of IL-10^25,26, also corroborated in our transcriptomic data. Unfortunately, the paucity of the biopsy samples post-treatment precluded confirmatory proteomic analyses.

One possible mechanism of complement activation was proposed by Roumenia et al.³¹ Using immunohistochemistry, they described C1q production by TAMs along with other components of the C1 complex assembled by the tumor cells, together leading to the synthesis of the C1 complex. Upon binding to IgG deposits on the tumor cells serving as C1 ligands, the complement cascade was initiated. In the context of the current clinical trial, it is plausible that the Fc domain of certain unknown tumor directed antibodies may have triggered off-target complement activation.

Interference by complement activation on the anti-tumor effects of immune checkpoint blockade has previously been described in pre-clinical models.^37,38 These experiments have additionally pinpointed C3a/C3aR and C5a/C5aR as targetable axes for cancer immunotherapy. C5aR was found to be upregulated in patients with NSCLC who progressed after initial response to anti-PD-(L)1 therapy, justifying the launch of a phase I clinical trial (STELLAR-001) combining IPH5401 (fully human anti-C5aR1 antibody) and durvalumab in advanced solid tumors.³⁹ In the preliminary results reported in 2019, the toxicity profile of the combination appeared manageable with no DLTs encountered and the dose escalation is currently ongoing. Albeit requiring validation, our findings suggestive of complement activation and the elevation of C5aR1 in the post-treatment samples support further testing of this combination in BCG-unresponsive NMIBC.

Our study was not without limitations. The small sample size of the study makes mechanistic interrogations difficult. However, the futility of the treatment, underscored by the high recurrence and progression rates, made it unethical to proceed onto the second stage of the clinical trial. Sample size was further limited by the small biopsy specimens leading to even fewer available samples for correlative analysis. Nevertheless, several of the limitations were overcome and we provide one of the first genomic analyses of paired pre- and post-treatment samples from BCG-unresponsive patients. Secondly, due to limited tissue samples, no further validation studies were performed to corroborate the RNA-sequencing data. Additionally, as the trial was designed and implemented prior to the FDA guidance on BCG-Unresponsive NMIBC²⁷, the population treated may not conform to the current BCG-Unresponsive definition used for regulatory approval. Nevertheless, inclusion criteria per the consensus statements on BCG-unresponsive disease^4,40 were strictly adhered to and reflect in spirit the population of heavily pre-treated patients not responding to intravesical BCG.

In summary, durvalumab monotherapy conferred minimal efficacy in treating BCG-unresponsive CIS of the bladder, with 6mo CR of 12% and no patient with durable response at 24mo. PD-L1 positivity in BCG-unresponsive CIS was low, but increased following anti-PD-L1 inhibitor treatment, indicating adaptive immune resistance. Additionally, correlative data demonstrated complement activation that blunted the anti-tumor response. Our results build strong rationale for combination treatment using durvalumab and anti-C5aR1 antibody in the treatment of BCG-unresponsive CIS.

Abbreviations

BCG

Bacillus Calmette-Guérin

CIS

Carcinoma in situ

NMIBC

Non-muscle invasive bladder cancer

RC

Radical cystectomy

CR

Complete response

ORR

Objective response rate

PFS

Progression free survival

IQR

Interquartile range

ICB

Immune checkpoint blockade

CPS

Combined positive score

TMB

Tumor mutational burden

TME

Tumor microenvironment

12-CK score

12-chemokine score

FDA

Food and Drug Administration

IBCG

International Bladder Cancer Group

TAM

Tumor associated macrophage