CHICAGO (GenomeWeb) – Multiple biomarkers are likely needed to more accurately predict patient response to anti-PD-1 immunotherapies, according to data presented at the American Society of Clinical Oncology meeting this week.
The theme was peppered throughout the meeting and especially emphasized at a clinical science symposium on Sunday, which touched on findings from several ongoing checkpoint blockade immunotherapy trials.
Results from such studies underscored the ongoing debate over which tumor and immune features need to be considered when prioritizing patients for such checkpoint blockade treatments and determining whether to offer immunotherapy alone or in combination with other treatment approaches.
A number of commercial tests that have been approved under the US Food and Drug Administration's recently established complementary diagnostics category target PD-L1 expression, for example, though questions remain about its reliability and potential expression dynamics over time.
And the immunohistochemistry-based PD-L1 IHC 22C3 pharmDx test from Dako, a subsidiary of Agilent Technologies, has been approved as a companion diagnostic, specifically for use with Keytruda (pembrolizumab) in certain advanced non-small cell lung cancer patients.
As results from immunotherapy trials come to fruition, other biomarkers are being proposed, including the inflammatory cell features, gene expression profiles, tumor mutational burden, and the presence of DNA repair defects that can ramp up that burden.
For his part, Johns Hopkins University oncology researcher Luis Diaz argued that mismatch repair deficiency — which can be caused by germline or somatic mutations to mismatch repair genes or regulatory alterations suppressing their activity — may be a method for pegging tumors with potential anti-PD-1 antibody response.
Diaz presented data on response to Merck's anti-PD-1 drug Keytruda (pembrolizumab) in an expanded trial of individuals with mismatch repair-deficient or -proficient colorectal cancer who had undergone at least two other therapies in the past.
The work builds on research that he and his team presented at ASCO last year and subsequently published in the New England Journal of Medicine on use of the anti-PD-1 antibody drug for mismatch-repair deficient colon cancers, which are typically marked by microsatellite instability and elevated mutation rates.
In patients treated for two weeks with pembrolizumab alone in the now-expanded study, the researchers saw response in more than half of the 28 individuals with MMR-deficient colorectal cancers and complete response in three of those individuals. On the other hand, they saw no response to the drug in the 25 patients with tumors with a functioning MMR system.
While median overall survival time was just six months or so for the MMR-proficient tumor group, it was not reached for those MMR-deficient tumors, Diaz noted. More than 60 percent of the MMR-deficient cases were marked by progression-free and overall survival after two years.
Based on his team's findings, Diaz argued that it may be time to start considering the possibility of using mismatch repair status — specifically MMR deficiency — as a tissue-independent marker for response to immune checkpoint blockade drugs such as pembrolizumab.
Using archival tumor samples from participants in the Imvigor210 trial of Roche's Tecentriq (atezolizumab) in metastatic urothelial carcinoma, Memorial Sloan-Kettering Cancer Center's Jonathan Rosenberg and his colleagues searched for their own clues to checkpoint blockade response.
Focusing on the cohort of patients with prior platinum-based chemotherapy treatment, that team looked at whether response to atezolizumab coincided with PD-L1 expression in tumors, as measured using Ventana's SP142 immunohistochemistry assay, a complementary diagnostic for atezolizumab that was approved by the US Food and Drug Administration last month.
The researchers gauged gene expression in nearly 200 archival samples from the trial with RNA sequencing and used Foundation Medicine's 315-gene FoundationOne cancer gene panel to estimate mutational burden in 150 tumors.
With the expression data, the team recapitulated luminal and basal subtypes described in bladder cancer by the Cancer Genome Atlas (TCGA), uncovering enhanced PD-L1 expression in basal subtype tumors and the tumor infiltrating cells associated with them.
Along with PD-L1 expression, atezolizumab response and progression-free survival tended to correspond to higher overall tumor mutational loads and with specific expression profiles, including the expression of genes related to interferon-gamma, which regulates PD-L1 expression.
But they also uncovered enhanced treatment response in individuals with luminal II TCGA subtype tumors, Rosenberg said, suggesting that PD-L1 expression, tumor mutation burden, and TCGA subtype can act as independent predictors of atezolizumab response.
He noted that the pursuit of multifaceted biomarkers to anti-PD-1 and anti-PD-L1 therapies will likely be enhanced by more extensive studies on interactions between tumors, immunologic factors, and other aspects of the tumor microenvironment.
Meanwhile, investigators from Vanderbilt University, Foundation Medicine, Adaptive Biotechnologies, and the University of Texas, Southwestern presented on a study of how mutational load and gene expression profiles provided insight into anti-PD-1 therapy response in melanoma patients.
Employing a new algorithm, that team used FoundationOne panel testing as a proxy for mutational load in tumor samples, demonstrating a strong correlation between mutational burdens detected with the panel and those found by exome sequencing in 300 melanoma samples from TCGA.
Foundation Medicine said it plans to develop a CLIA-certified version of the tumor mutation burden analysis, which will be provided to physicians with all FoundationOne and FoundationOne Heme reports by the third quarter of 2016.
A representative for Foundation Medicine said the firm is working on developing a universal companion diagnostic test, but has not announced plans for developing specific biomarkers.
The Vanderbilt-led team's FoundationOne mutation analysis allowed it to classify tumor samples from dozens of checkpoint blockade-treated melanoma patients into high-, intermediate-, and low-mutational load groups. A subset of the samples was also analyzed using Adaptive Biotechnologies' ImmunoSeq method for sequencing T cell receptors.
In those samples, mutational load tended to be higher when specific gene mutations were present, including alterations affecting NF1. And patients with such mutations typically fared better on anti-PD-1 treatment. In contrast, mutations such as BRAF V600E turned up more often in treatment non-responders.
In that study, at least, clonal expansion levels in T cells did not seem to correlate with treatment response, though others have suggested that such immune cell features may be informative.
The panel of clinical science symposium presenters and commentators generally agreed that they would be hesitant to exclude patients from checkpoint blockade immunotherapy based on PD-L1 expression or other biomarkers explored so far, since a subset of individuals do respond even with seemingly unfavorable biomarkers such as low PD-L1 expression or modest mutation rates.
Still, they noted that biomarkers are needed to prioritize and tailor patient treatment so that combination therapy can be considered for patients with potentially poor responses, though the data available to date suggests more than one marker may be needed.
In a recent PD-L1 market research report by Diaceutics, the personalized medicine-focused consulting firm urged pharma companies to diversify their diagnostic approaches in the checkpoint blockade space to improve therapeutic outcomes.
But the choice of marker may vary between cancer types to some extent. During a session on immunotherapy for head and neck cancer at ASCO yesterday, data from the CheckMate 141 trial of Bristol-Myers Squibb's Opdivo (nivolumab) for metastatic squamous cell carcinoma of the head and neck showed increasing response with rising PD-L1 levels in the nivolumab-treated group. There was also an uptick in response in tumors positive for HPV 16.
On the other hand, a trial of Keytruda (pembrolizumab) in recurrent or metastatic head and neck squamous cell carcinoma with or without PD-L1 expression suggested response clues could come from testing for PD-L1 and/or PD-L2 expression, only when both tumor and inflammatory cell expression was included.
The recurrent or metastatic head and neck squamous cell carcinoma study investigators also observed ties between pembrolizumab response and a six-gene signature for interferon-gamma activity, which was identified by NanoString nCounter expression testing on RNA extracted from tumor FFPE samples.
"Inclusion of inflammatory cells in [immunohistochemistry] scoring improves the ability to predict response based on PD-L1 status compared to tumor cells alone in [patients with recurrent or metastatic head and neck squamous cell carcinoma]," Laura Chow and her colleagues noted. "Exploratory analyses suggest that PD-L2 and [interferon-gamma] signature may be associated with clinical response to pembro and may offer additional strategies to improve prediction of response."
The biomarker potential of PD-L2 was also assessed by researchers from Barts Cancer Institute, Genentech, and elsewhere, who focused on response to atezolizumab in four clinical trial cohorts.
As Barts Cancer Institute's Peter Schmid explained at his ASCO presentation today, the team used a Fluidigm gene expression platform or RNA sequencing to gauge PD-L1 and PD-L1 expression in tumors from individuals treated for melanoma, non-small cell lung cancer, metastatic urothelial carcinoma, and renal cell carcinoma.
Both PD-L1 and PD-L2 expression correlated with benefit from atezolizumab treatment, though this association was less robust in the renal cell carcinoma patient.
Schmid noted that potential mechanisms for PD-L2 contribution to atezolizumab response are still be explored. And while the clinical trial data his team tapped suggests PD-L2 expression is linked to tumor infiltrating cell activity and inflammatory tumor types, Schmid said he's not yet convinced that the inclusion of PD-L2 expression profiles would significantly improve prognostic information.