Atezolizumab: A PD-L1–Blocking Antibody for Bladder Cancer

The programmed cell death 1 (PD-1, CD279) receptor and its ligand, programmed cell death 1 ligand 1 (PD-L1, CD274, B7-H1), comprise one of the main immune checkpoint pathways that downregulates immune activity (1). PD-1 is a transmembrane protein and is found mainly on activated T cells, B cells, and macrophages. When PD-1 binds one of its two known ligands, PD-L1 and programmed cell death 1 ligand 2 (PD-L2, CD273), T-cell function is downregulated as manifested by decreased IL2 and IFNγ production. The PD-L1 ligand binds not only to PD-1 but also to B7.1 (CD80), another T-cell costimulatory molecule, but this interaction is less well understood. Although PD-L1 is widely expressed by antigen-presenting cells (e.g., macrophages, B cells, dendritic cells) as a way of fine-tuning peripheral immune activation and avoiding autoimmunity, its expression can also be induced on T cells and other cell types (such as epithelial and endothelial cells) with exposure to proinflammatory cytokines such as IFNγ (1).

Many tumor types, including bladder cancer, express PD-L1 either on the tumor cells themselves or on immune cells (IC) that are infiltrating the tumor. Furthermore, PD-L1 mRNA can be alternatively spliced to produce a soluble non-membrane–bound version of the protein, which is biologically active and might regulate local immunity in a paracrine fashion (2). Factors affecting tumor PD-L1 expression are numerous and include microenvironment cytokines, PI3K–Akt pathway activation, MAPK pathway activation, STAT1 signaling, and epigenetic control through a variety of miRNAs (3). Importantly, tumors expressing PD-1/PD-L1 have been shown in several studies to have worse survival outcomes, which is a major reason that PD-L1–targeted therapies have been developed clinically. In bladder cancer, PD-L1 expression is associated with higher tumor grade, stage progression, poor responsiveness to bacillus Calmette–Guérin (BCG) immunotherapy, and worsening survival (4–6).

Atezolizumab (Tecentriq, MPDL3280A; Genentech/Roche) is a fully humanized IgG1 mAb that blocks the interaction of PD-L1 with both PD-1 and B7.1, but not the interaction of PD-L2 with PD-1 (Fig. 1; ref. 7). The pharmacokinetics of atezolizumab were initially studied in cynomolgus monkeys and mice, where its volume of distribution was calculated to be approximately that of the plasma volume. The in vivo biodistribution of atezolizumab 24 hours after infusion is, in order of magnitude, the spleen, lungs, kidneys, liver, heart, and muscle. In tumor-bearing animals, the drug also accumulates intratumorally, initially at the pushing border of the tumor and progressing later to the tumor core, particularly if the tumor is necrotic (7). The pharmacokinetic curve of atezolizumab is dose dependent (nonlinear) because of target-mediated drug disposition (binding of drug to the PD-L1 ligand in the body). Saturation of PD-L1 receptors by atezolizumab on circulating CD4 and CD8 T cells occurs between 24 and 48 hours after dosing with serum concentrations >0.5 μg/mL.

PCD4989G (NCT01375842), a phase I dose-escalation trial in subjects with a variety of metastatic solid tumors, administered atezolizumab at doses ranging from 0.03 to 20 mg/kg (8). When given every 3 weeks, a dose of 15 mg/kg was sufficient to maintain a trough level of >6 μg/mL, which results in >95% saturation of intratumoral PD-L1 receptors in murine models with a good margin of error that allows for interpatient variability in drug delivery to the tumor. A dose of 1,200 mg was consequently proposed as the optimal dose for future human studies (assumes an 80 kg person), administered over 60 minutes for the first infusion cycle and, if well tolerated, over 30 minutes for subsequent cycles. The volume of distribution is estimated to be 6.9 L, the half-life 27 days, and steady state achieved by the third cycle (week 9). In mice and monkeys, antitherapeutic antibodies that potentially neutralize atezolizumab start to develop around day 7 and are universally present by day 14 posttreatment (7). In a sample of 275 subjects treated in phase I, 42% developed antitherapeutic antibodies at some point after starting atezolizumab, although these did not appear to affect the pharmacokinetics, efficacy, or safety of the drug. In addition, mild hepatic impairment and moderate renal dysfunction [glomerular filtration rate (GFR) 30–89 mL/minute] do not appear to affect atezolizumab dosing, although the pharmacokinetics in severe renal impairment (GFR <30 mL/minute) or moderate hepatic dysfunction (>1.5 ULN for liver function tests) are unknown. Similarly, the genotoxicity, carcinogenicity, and fertility effects of atezolizumab are unknown.

A key feature of atezolizumab is that it is FcγR-binding deficient due to an asparagine-to-alanine substitution at position 298 of the CH2 domains of each heavy chain, which means that it cannot bind to Fc receptors on phagocytes and, therefore, does not cause antibody-dependent cell-mediated cytotoxicity (ADCC; ref.8). This is important because PD-L1 is heavily expressed by T cells and other leukocytes, and binding of a mAb to their cell membrane could result in ADCC-mediated depletion of tumor-specific T cells, an event that could worsen antitumor immunity instead of improving it.

The immunologic effects of atezolizumab treatment have been only preliminarily reported from subjects on clinical trials. Cytokine changes that have been observed include transient increases in IL18, IFNγ, and CXCL11 and a transient decrease in IL6 (8, 9). Cellular changes include increases in proliferating CD8⁺ T cells (8, 9).

Atezolizumab was first evaluated in bladder cancer in an expansion cohort of the PCD4989G trial (9). Patients with bladder cancer were initially selected by PD-L1 status (see “Biomarkers” section), but after 21 patients were enrolled, the trial was opened to all subjects with bladder cancer. Sixty-seven patients were evaluable for efficacy, and PD-L1 expression was high in 10 (15%), moderate in 20 (30%), low in 23 (34%), absent in 12 (18%), and unknown in two (3%). Subjects were heavily pretreated, with 62 (93%) having previous platinum-based chemotherapy and 48 (72%) having at least two prior systemic therapy regimens. Poor prognostic features were also prevalent, including 39 (59%) with an Easter Cooperative Oncology Group (ECOG) performance status of 1, 26 (42%) with a time from previous chemotherapy of <3 months, and 50 (75%) with visceral metastases. The objective response rates (ORR), observed at a median of 42 days, were 43% for patients with high/moderate PD-L1 group and 11% for those with low/absent PD-L1 group. Complete responses (CR) were noted in 7% of subjects, and all had high/moderate PD-L1 expression. At the time of data lock, 16 of the 17 initial responders had not progressed and the median duration of response had not been reached. On the basis of these results, atezolizumab received breakthrough therapy designation status by the FDA in June 2014.

At the 2015 American Society of Clinical Oncology (ASCO) Annual Meeting, bladder cancer survival data from the PCD4989G trial were presented (10). Of 87 evaluable subjects, the ORR was 50% for the high/moderate PD-L1 group and 17% in the low/absent PD-L1 group. Median duration of response had not been reached, and 67% of responders had ongoing responses. Median progression-free survival (PFS) was 6 months in the high/moderate PD-L1 group and 1 month in the low/absent PD-L1 group, with 1-year PFS rates of 39% and 10%, respectively. Median overall survival (OS) was not reached in the high/moderate PD-L1 group and was 7.6 months in the low/absent PD-L1 group, with 1-year OS rates of 57% and 38%, respectively.

IMvigor210 (NCT02108652) is a multicenter, single-arm, two-cohort phase II trial of atezolizumab treatment in bladder cancer, and cohort 2 of this trial consisted of subjects with inoperable locally advanced or metastatic bladder cancer progressing after prior platinum-based chemotherapy (11). Of 310 evaluable subjects, 100 (32%) had high/moderate PD-L1 expression, 107 (37%) low PD-L1, and 103 (33%) absent PD-L1. Using RECIST v1.1, an ORR was noted in 15%, including a CR rate of 5%. The ORR was 26% in the high/moderate PD-L1 group, 11% in the low PD-L1 group, and 8% in the absent PD-L1 group. With a median follow-up of 11.7 months, median duration of response was not reached, although median PFS was 2.1 months and median OS was 7.9 months. On the basis of these results, the FDA approved atezolizumab for the treatment of bladder cancer in patients with locally advanced or metastatic urothelial carcinoma who have progressed (i) during/after platinum-based chemotherapy or (ii) within 12 months of neoadjuvant or adjuvant treatment with platinum-based chemotherapy (12).

Updated data from IMvigor210 were reported at the 2016 ASCO Annual Meeting and included new data on cohort 1, subjects who had bladder cancer metastases, were chemotherapy naïve, and were cisplatin ineligible (13). The most common reason for cisplatin ineligibility was renal dysfunction (71%), and 66% had visceral metastasis. A total of 119 patients were evaluable for response, and the ORR was 24%, including 7% CRs. Median duration of response was not reached, with 75% (21/28) of responses ongoing. Interestingly, the ORR for the high/moderate PD-L1 group was 28%, and for subjects with upper tract (renal pelvis or ureter) primary tumors, the ORR was 42%. With a median follow-up of 14.4 months, median OS was 14.8 months and 1-year OS rate was 57%.

In summary, evidence from phase I and II trials suggests durable activity of atezolizumab in advanced bladder cancer that has progressed during or after platinum-based chemotherapy. A combined analysis of 10 phase II trials of second-line therapies for metastatic bladder cancer showed an average 1-year OS of 20% (14). Given that the 1-year OS of cohort 2 of IMvigor210 of 36% is nearly double that, a randomized phase III trial comparing atezolizumab with standard chemotherapy in subjects with platinum-refractory metastatic bladder cancer is underway (IMvigor211, NCT02302807). For cisplatin-ineligible subjects, although the ORR of cohort 1 of the IMvigor210 (24%) was not better than the historic control of gemcitabine/carboplatin chemotherapy (36%), the median OS (14.8 months) does appear better than chemotherapy (9.3 months), suggesting that a randomized comparison is warranted (15).

In general, atezolizumab has been well tolerated across all studies in bladder cancer, with most adverse events (AE) being mild to moderate in grade. This is important as patients with bladder cancer are often elderly and have multiple comorbidities and, consequently, many may not be chemotherapy candidates (16). In IMvigor210, 310 patients were evaluable for safety, and AEs of any grade were reported in 97% of patients, with 55% having grade 3 or 4 AEs (11). The most common AEs were fatigue (49%), nausea (26%), decreased appetite (27%), pyrexia (22%), and diarrhea (20%). The most frequent high-grade (grade ≥3) AEs were anemia (9%), fatigue (6%), dyspnea (4%), nausea (2%), and hypertension (2%). Importantly, many AEs were not treatment related, and treatment-related AEs (trAE) occurred in 69% of patients and only 16% of subjects had high-grade trAEs. The most common trAEs were fatigue (30%), nausea (14%), decreased appetite (12%), pruritus (10%), pyrexia (9%), diarrhea (8%), rash (7%), arthralgia (7%), and vomiting (6%). There were no treatment-related deaths, but atezolizumab therapy was interrupted in 20% of subjects and discontinued in 4% due to AEs.

Atezolizumab's immunotherapeutic mechanism of action has led to special attention regarding immune-related AEs (irAE). In IMvigor210, irAEs occurred in 7% of subjects, with pneumonitis, increased liver enzymes, rash, and dyspnea being the most commonly observed (1% each; ref.11). Systemic corticosteroid treatment (for any AE, not just irAEs) was required in 22% of patients.

Detailed reviews of the management of immune checkpoint inhibitor–induced toxicity are available elsewhere (17, 18). However, depending on the severity of the irAE, withholding atezolizumab and providing temporary immunosuppression with topical/systemic corticosteroids, TNFα antagonists, mycophenolate mofetil, or other agents is highly effective. For many moderate irAEs, atezolizumab treatment may be withheld for a few weeks and then the therapy restarted once the irAE has resolved. More serious and potentially life-threatening irAEs, such as pneumonitis, hepatitis, colitis, endocrinopathies, neurologic and ocular toxicities, and pancreatitis, usually preclude further treatment with atezolizumab.

Attempts have been made to develop biomarkers that predict which patients will respond to atezolizumab and which will not, and principle among these is its target PD-L1. PD-L1 protein can be measured as a biomarker on the cell membrane or in soluble form in biofluids (2). Currently, it is unclear which form of PD-L1 (membrane bound or soluble) or which cell type with cell membrane expression (tumor cells, tumor-infiltrating leukocytes, endothelial cells in the tumor vasculature) is important for predicting disease outcome and response to therapy. Numerous factors affect measurement of the PD-L1 biomarker, including the specific mAb clones used in the assays, the cutoff points used to define positivity, and the method used for detecting the presence of PD-L1 (IHC, quantitative PCR, ELISA, etc.). Furthermore, the immune system is dynamic, and measuring any immunologic biomarker at a single time point in time or space may be insufficient to fully characterize the immune response in the tumor microenvironment. A blueprint proposal for standardizing PD-L1 measurement was made at the FDA-AACR-ASCO workshop titled “Complexities in Personalized Medicine” (March 24, 2015; Washington, DC). This proposal, put forth by pharmaceutical and diagnostic companies, hopes to lay the groundwork for developing companion diagnostic biomarkers for PD-1/PD-L1–targeted immunotherapies.

Early results have suggested that PD-L1–positive bladder tumors respond better to atezolizumab than negative ones (9, 11). However, in bladder cancer and other tumors, clinical activity is still observed with PD-1/PD-L1 immune checkpoint inhibitors in tumors that are classified as PD-L1 negative (19). In both the PCD4989G and IMvigor210 trials, tumor PD-L1 expression was analyzed by Ventana SP142 IHC assay that measures tumor-infiltrating IC PD-L1 expression (8, 9, 11), which is categorized as high (IC3, ≥10%), moderate (IC2, 5%–10%), low (IC1, 1%–5%), or absent (IC0, <1%). The Ventana SP142 assay is FDA approved as a companion diagnostic for atezolizumab. Although many studies have found tumor cell membrane PD-L1 expression to be prognostic of bladder cancer outcome (4, 6, 20), the IMvigor210 trial did not confirm this finding. Instead, only tumor-infiltrating IC PD-L1 expression predicted treatment response and outcome (9). It is important to realize that tumor cell PD-L1 expression is strongly correlated with the presence of tumor-infiltrating ICs, which in turn predict the response of checkpoint inhibitors (21). In essence, there is an interplay between tumor and the immune system in the tumor microenvironment that determines PD-L1 expression on various cell types in a dynamic manner. Indeed, serial biopsies of tumors undergoing atezolizumab treatment demonstrate that PD-L1 expression increases in subjects responding to treatment (8).

Other potential predictive biomarkers of bladder cancer response to atezolizumab include The Cancer Genome Atlas molecular subtype and the tumor mutation load (11). However, in other tumor types, a number of other predictive biomarkers have been associated with atezolizumab response, including a dominant pretreatment effector T-cell signature (22), IFNγ gene expression (8, 23), indoleamine 2,3-dioxygenase expression (8), JAK3 activation (24), and the expression of other immune checkpoint molecules, to list a few (8). Many other biomarkers exist for other drugs targeting the PD-1/PD-L1 axis (25). Most biomarkers useful for determining the likelihood of response to atezolizumab and other immunotherapies have not been well validated.

Atezolizumab's FDA approval was granted on the basis of good ORRs and durable responses in the IMvigor210 trial (11). A number of other trials are underway (Table 1). To date, there is no survival-prolonging second-line therapy in advanced bladder cancer following progression during/after platinum-based chemotherapy (16). Although data from IMVigor210 are promising, definite proof of a survival benefit with atezolizumab treatment in the second-line metastatic bladder cancer setting is pending the results of IMvigor211, which has completed its planned accrual of 932 subjects. Another trial (IMvigor130) is examining OS differences between atezolizumab–gemcitabine–carboplatin and carboplatin–gemcitabine in untreated cisplatin-ineligible metastatic bladder cancer patients.

The benefit of atezolizumab is also being examined in localized bladder cancer disease stages. Currently, cisplatin-based systemic chemotherapy is the only proven systemic therapy proven to extend survival in patients with muscle-invasive bladder cancer undergoing curative cystectomy (26). However, due to comorbidities, many cystectomy candidates never receive neoadjuvant or adjuvant cisplatin-based chemotherapy. IMvigor010 will determine whether adjuvant atezolizumab improves survival in subjects with residual muscle-invasive cancer at cystectomy. Finally, the GU123 trial will assess the role of atezolizumab in BCG-unresponsive non-muscle–invasive bladder cancer in a phase Ib/II clinical trial. The results of all these trials will inform greatly on the role of atezolizumab across the bladder cancer disease spectrum.

B.A. Inman reports receiving commercial research grants from Abbott Laboratories, Dendreon, Genentech, and GlaxoSmithKline and is a consultant/advisory board member for AstraZeneca, Combat Medical, Ferring, Genentech, and Taris Biomedical. M.R. Harrison reports receiving commercial research grants from Acerta and Genentech, speakers bureau honoraria from Genentech, and is a consultant/advisory board member for AstraZeneca and Genentech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: B.A. Inman, M.R. Harrison

Development of methodology: M.R. Harrison

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): B.A. Inman, M.R. Harrison

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): B.A. Inman, M.R. Harrison

Writing, review, and/or revision of the manuscript: B.A. Inman, T.A. Longo, S. Ramalingam, M.R. Harrison

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): B.A. Inman