Roche’s PD-L1 Bladder Cancer Data Show Promise Of New Biomarkers

Use of The Cancer Genome Atlas subtyping and mutational load analysis complement PD-L1 expression, with potential to improve prediction of response to atezolizumab in the future, according to study in the Lancet.

New types of biomarkers are proving helpful in bladder cancer, playing a complementary role on top of PD-L1 expression, investigators for a pivotal trial of Roche’s PD-L1 inhibitor atezolizumab recently reported in The Lancet.

PD-L1 expression has been correlated with response to the family of PD-1/L1 checkpoint inhibitors and has emerged as the strongest biomarker to date for the cancer immunotherapy class. But it has also proven to be controversial, because patients testing negative can still benefit from treatment, which has spurred calls for more accurate ways to predict response (Also see "Should PD-L1 Expression Be Used As A Biomarker For Immunotherapy?" - Pink Sheet, 7 Jul, 2014.).

Data from the Phase II IMvigor 210 study support the correlation of PD-L1 expression in immune cells from bladder cancer specimens with response to atezolizumab therapy, but also suggest potential for new biomarkers, according to investigators who reported full data from the study in The Lancet.

Memorial Sloan Kettering genitourinary cancer specialist Jonathan Rosenberg and colleagues published full results from the single arm study in second-line bladder cancer in The Lancet online on March 4, following presentation of topline data in 2015 (Also see "PD-1/L1 Update: Roche’s Atezolizumab Advances, Merck’s Keytruda Strikes In New Tumors" - Pink Sheet, 28 Sep, 2015.).

Roche completed a rolling BLA in January for use in second-line bladder cancer or in first-line disease for those ineligible for cisplatin-based chemotherapy (Also see "Roche Gets Atezolizumab Applications Rolling" - Pink Sheet, 29 Jan, 2016.). Two Phase III studies of atezolizumab are ongoing in bladder cancer – IMvigor 211 in second-line disease and the IMvigor 010 study of atezolizumab in the adjuvant setting for PD-L1 positive, high-risk muscle invasive bladder cancer.

The company also is close to completing a rolling BLA for atezolizumab in non-small cell lung cancer. Full results from the POPLAR study, one of the trials supporting the filing, were published separately in The Lancet on March 9 (see sidebar).

The bladder cancer study tested atezolizumab every three weeks in 310 patients with second-line urothelial cancer and found a 15% objective response rate (with a 5% complete response rate) in all patients and a 26% ORR (11% CR) in those with medium-to-high expression of PD-L1 in tumor infiltrating immune cells, based on independent review with standard RECIST criteria. Researchers also looked at response judged by investigators using immune-mediated response criteria, and found similar ORRs.

Reporting exploratory analyses in the IMvigor 210 trial in The Lancet, researchers also noted the separate and complementary value of two other types of biomarkers -- classification based on The Cancer Genome Atlas project subtypes and analysis of mutations and mutation loads using technology from Foundation Medicine Inc.

For example, with the TCGA approach, specimens were classed into luminal and basal subtypes. Those with the “luminal cluster II subtype” were more likely to have activated T-effector cells and had significantly higher response rates.

TCGA analyses suggest that urothelial carcinoma has the highest mutation rate of all studied cancers and mutation load has been associated with response in a number of tumor types. In the IMvigor 210 study, an analysis using a Foundation Medicine panel of 315 cancer-related genes showed that median mutation load was significantly higher in responders vs. non-responders, and this was unrelated to TCGA subtype or the PD-L1 expression in immune cells.

“This report is the first to show the association of TCGA subtypes with response to immune checkpoint inhibition and to show the importance of mutation load as a biomarker of response to this class of agents in advanced urothelial carcinoma,” the IMvigor investigators reported.

Using all three measures significantly improved the association with response.

“Thus, disease subtype and mutation load do not simply recapitulate the information already provided by PD-L1 expression in immune cells, but rather, they provide independent and complementary information. Additional data and larger sample sizes are needed to allow the formal construction of a multibiomarker classifier, and continued consideration of all three biomarkers is warranted in next-generation companion diagnostics,” the article states.

The next generation biomarkers described in the paper are examples of emerging markers, which in the process of development are helping Genentech Inc. to better understand the data for PD-L1 expression. Looked at collectively, a variety of biomarkers will be useful in identifying the same underlying biology and preexisting immunity, in order to predict treatment outcomes with cancer immunotherapy, according to Dan Chen, global head of immunotherapy product development.

Mutational load, for example, is not a perfect reader of the biology, but it is a nice way to indirectly measure the likelihood that a tumor is visible to the immune system, the exec explained in an interview.

“The more of these mutations you have, the more a cancer cell is likely to look foreign to the immune system and so it’s more likely that the immune system can see the cancer and at least try to attack it, but it’s not perfect. It’s like a game of random chance, and the more mutations you have the more dice you get to roll, and depending on how those dice roll out, your immune system can better see the cancer,” Chen said.

Genentech believes that within two years to five years, next generation biomarkers may be ready to use in clinical practice to guide cancer immunotherapy in a personalized manner. The goal is that a patient with cancer could be tested on a range of measures and receive therapy – a variety of combinations are possible – tailored to the particular biology of his or her tumor.

Staking Claim In Bladder Cancer

Roche has been at the lead of development among PD-1/L1 sponsors in bladder cancer, a disease with high unmet need, since the 2014 American Society of Clinical oncology meeting where the company released striking results in a Phase I study (Also see "Roche Breaks Into Bladder Cancer, But Merck Is Hot On Its Heels" - Pink Sheet, 16 Jun, 2014.).

In The Lancet paper, Rosenberg et al. noted that there have been no major improvements in 30 years for bladder cancer. First-line treatment revolves around platinum-based chemotherapy, but while response can be very good, it doesn’t last long. Nothing is approved for second-line treatment in the US and a range of chemotherapies, including docetaxel and paclitaxel, are given off label. The typical response rate is 10% and median overall survival is only from 5-7 months.

“New approaches are needed to break this therapeutic stalemate,” the IMvigor investigators said in The Lancet report.

Early on in development, some theorized that anti-PD-L1 agents would ultimately emerge with a better side effect profile compared to anti-PD-1 inhibitors, based on small differences in the mechanism of action. The current thinking is that both types of drugs look similar in terms of response and survival, Chen commented. In terms of safety, autoimmune nephritis has been reported for PD-1 but not for PD-L1, he added.

“There is still room for these molecules to end up ultimately being different, but how different we just don’t know yet. That will take much larger trials and potentially even head-to-head trials to really determine those differences – that is true not only in bladder cancer but in other tumor types,” Chen said.

The IMvigor 210 trial was done in advanced or metastatic urothelial cancer that progressed after platinum based chemotherapy. Atezolizumab was given at 1,200 mg every three weeks.

Positivity for the PD-L1 biomarker was measured in terms of the level of expression on immune cells in the tumor microenvironment using the SP142 assay from Ventana Medical Systems Inc.

Levels Of PD-L1 Expression In Immune Cells:

• IC0 (<1%)
• IC1 (≥1% but <5%)
• IC2/3 (≥5%)

PD-L1 expression in the immune cells was associated with response (see table), whereas PD-L1 in tumor cells was not. Genentech has stressed the value of testing both for expression on immune cells and tumor cells (Also see "Roche's Oncology Strategy: The Long Game Comes Into Focus" - Pink Sheet, 15 Jun, 2015.).

Furthermore, researchers noted ongoing responses in 38 (84% of responders) after follow up of 11.7 months. Durable responses were recorded in patients with metastases and poor prognostic features, although the ORR was lower compared to those with less disease burden.

IMvigor 210 Efficacy Results, By PD-L1 Expression In Immune Cells
PD-L1 status	Number of patients	Objective response rate	Complete response rate
RECIST version 1.1 criteria by independent review
IC2/3	100	26%	11%
IC1/2/3	207	18%	6%
All patients	310	15%	5%
IC1	107	10%	2%
IC0	103	8%	2%
Modified RECIST criteria by investigator review
IC2/3	100	27%	8%
IC1/2/3	207	22%	7%
All patients	310	19%	5%
IC1	107	17%	6%
IC0	103	13%	2%
IC0 is lowest level of PD-L1 expression (no detectable PD-L1) and IC3 is the highest (≥5%) Source: Jonathan Rosenberg, et al., published in The Lancet online March 4

Atezolizumab also had a good safety profile in the trial. The rate of Grade 3-4 adverse event was 16%, the most common of which was fatigue (2%). The rate of Grade 3-4 immune-mediated adverse event was 5% and included low rates of pneumonitis and rash.

No treatment-related deaths were reported.

“The low incidence of clinically relevant treatment-related adverse events makes atezolizumab widely applicable in these patients, who often have renal impairment and/or other comorbidities. This durable efficacy and tolerability is striking in comparison with outcomes recorded with presently available second-line chemotherapy for urothelial carcinoma,” Rosenberg and colleagues noted.

In an accompanying editorial in The Lancet, Johns Hopkins pathologist George Netto said that the trial findings are encouraging, especially for those with higher immune cell PD-L1 levels (IC2/3).

However, Netto also noted that in the overall cohort the “improvement in objective response remained incremental (15%, compared with the 10% historical control rate) and the median overall survival was a modest 7·9 months," which he said "highlights the importance of future efforts to identify, validate, and standardize biomarkers of response.”

The immunotherapy response rate is just the “tip of the iceberg” when it comes to PD-1/L1 therapy, because tumors may grow and then shrink again, Genentech’s Chen commented. When a physician and patient think about treatment options, they consider the response rate but also the stabilization of disease for a prolonged period of time and the toxicity profile of alternative therapies, he said. Patients with high levels of PD-L1 expression or no expression can both respond and experience highly durable responses, he added.

Response rate is “only part of the story,” he said.