Current Biomarker Testing for Immunotherapy

For cisplatin-ineligible bladder cancer, patients may receive atezolizumab or pembrolizumab, using 2 different PD-L1 biomarkers. Atezolizumab uses the SP142 antibody with a PD-L1 IHC cutoff of 5% for immune cells only. Pembrolizumab uses a different scoring system, the antibody 22C3, and calculates a PD-L1 combined score of ≥10% (percent of immune cells + percent of tumor cells, divided by percent of tumor cells).

Biomarker testing in lung cancer is simpler. For use of pembrolizumab in the first-line treatment of advanced EGFR/ALK-negative/unknown non–small cell lung cancer (NSCLC), the NCCN Guidelines for NSCLC dictate a PD-L1 IHC score >50%, and a score of 1% (tumor cells only for both) in the second line. “But these cutoffs and their relevance are debatable in the new era of chemotherapy plus immunotherapy in lung cancer,” she noted.

According to the guidelines, microsatellite instability/mismatch repair deficiency (MSI/MMR) testing should be used in colon and endometrial cancers. And in renal cell carcinoma and metastatic melanoma, all patients should receive immune checkpoint blockade, regardless of a biomarker, Dr. Naidoo said. “But this is a rapidly evolving space, and it's likely to change every couple of months in reality,” she added.

Inflamed Tumor Biomarkers

Immunotherapy biomarkers can generally be separated into 3 broad categories: inflamed tumor biomarkers (ie, PD-L1–high), immunogenic tumor biomarkers (related to how likely the tumor is to engender an immune response, such as high tumor mutational burden [TMB]), and factors of the host environment (ie, gut microbial diversity/composition). This last category is new and emerging, according to Dr. Naidoo, and examines characteristics of an individual, separate from the tumor, that can inform whether they are likely to respond to immunotherapy.

PD-L1 was originally identified as a biomarker as a result of a phase I study that revealed a correlation between lack of PD-L1 expression and lack of response to nivolumab in patients with multiple solid tumors, including lung cancer.³ This was followed by KEYNOTE-010, a large lung cancer study that validated PD-L1 expression cutoffs of ≥50% and ≥1% in tumor cells only using the antibody of 22C3.⁴ Findings from KEYNOTE-010 were confirmed in a subsequent phase III study of first-line pembrolizumab in NSCLC, which showed that patients whose tumor cells expressed PD-L1 at a rate of ≥50%, had an impressive benefit in progression-free survival, as well as an overall survival benefit.⁵

However, there are some key limitations to the use of PD-L1 as a biomarker. First, there may be spatial heterogeneity of PD-L1 assessment even within a lesion that is biopsied. Therefore, the results of PD-L1 testing will depend on the location of the biopsied tumor, the timing of biopsy relative to PD-L1 testing, and the tumor type being tested. Furthermore, depending on measurement of tumor cells, immune cells, or both, cell scoring may also be relevant in an assessment of PD-L1 status. Adding to the lack of standardization, different antibodies have been used with a variety of agents and tumor types, using different platforms and cutoff values (Figure 1). “Hopefully we will see a number of studies to try and harmonize these,” she said.

View Full Size

PD-L1 as a biomarker: key limitations.

Abbreviations: IC, immune cells; IO, immuno-oncology; N/A, not applicable; NSCLC, non–small cell lung cancer; RCC, renal cell carcinoma; TC, tumor cells.

Adapted from Drake et al, ASCO 2018.

Citation: Journal of the National Comprehensive Cancer Network J Natl Compr Canc Netw 17, 5.5; 10.6004/jnccn.2019.5021

Immunogenic Tumor Biomarkers

“We've all heard the term ‘TMB,’ but what does it mean and what does it reflect?” Dr. Naidoo asked. TMB is a measurement of the accumulated errors (ie, mutations) in tumor cell DNA. Its rationale as a biomarker for ICI therapy is based on the fact that with every mutational error that occurs in DNA, abnormal proteins or peptides form that can be expressed on the surface of immune cells, called neoantigens. These neoantigens result in tumors that appear more foreign, or less “self,” to the immune system and are thus more likely to elicit an immune response.

The relationship between TMB, neoantigen burden, and immunotherapy response has been demonstrated in many studies. A trial of 34 patients with lung cancer treated with single-agent pembrolizumab showed improved progression-free survival among those with a high TMB.⁶ When TMB was translated to neoantigen burden, the analysis revealed a similar pattern. Similarly, another study showed that patients with colorectal cancer and a high TMB from MSI also experienced improved responses to checkpoint blockade.⁷

“TMB really is where PD-L1 was several years ago in terms of the need to harmonize the different methods, the different cutoff values, and what they mean,” she said. “Some of the factors to standardize in TMB include sequencing depth, the mutations included, and the filtering process.”

Another important study from London assessed not only the counts of mutations in tumors and their neoantigen count but also their neoantigen heterogeneity. This study demonstrated that the neoantigens that elicit the strongest immune response originate from highly clonal mutations, or mutations that exist in many different tumor areas.⁸ “This is actually being looked at as a biomarker of response in lung cancer in ongoing studies,” she said.

Biomarkers of Nonresponse

The third type of biomarkers, those of nonresponse, inform when to avoid giving immunotherapy to select patients. Early studies in lung cancer showed that patients with EGFR-mutant NSCLC generally had lower TMB. “[These tumors] are not associated with smoking; it's just a phenotypically different cancer,” said Dr. Naidoo. “We observed that these patients can have response rates as low as 0% from immune checkpoint blockade, even if the tumors are high PD-L1 expressors.” A meta-analysis of 3 large studies in EGFR-mutant NSCLC showed no benefit with checkpoint blockade compared with chemotherapy.⁹

KRAS mutations, which are present in approximately 25% of lung adenocarcinomas and confer a poor prognosis, can co-occur with other mutations, and these genomic subsets have differential responses to immune checkpoint blockade. One study found that patients with a co-mutation between KRAS and STK11 had only a 0% to 7% response rate to checkpoint blockade, whereas those with a co-mutation between KRAS and p53 had a higher response rate of approximately 35%.¹⁰ “The take-home from this study was that co-mutations between KRAS and STK11 are bad actors and these patients shouldn't get checkpoint blockade,” she said.

Immunotherapy and the Microbiome

Microbiome is a term for the population of microbes present in a particular biologic niche, such as on the skin or in the gut; this can refer to bacteria, fungi, viruses, and any other organisms not classified as cells.¹¹ The relationship between the microbiome and immunotherapy is still being explored, but this particular host factor may augment responses and toxicity to ICIs.

“There is 100-fold more gene content in the microbiome than in the human genome, and it has been known to influence both health and disease,” said Dr. Naidoo. A seminal study showed that patients with metastatic melanoma who responded to anti–PD-1 therapy had higher gut microbiota diversity than nonresponders.¹² Another study showed that the microbiome is predictive not only of response but also of toxicity: the investigators demonstrated that the presence of a certain species of microbe may be protective against the development of colitis from anti–CTLA-4 therapy in melanoma.¹³

According to Dr. Naidoo, there will certainly be a series of studies examining what biomarkers can be combined and what may be relevant in different tumor types. “I think there's going to be considerable differences between tumor types and patient populations that fall out of these analyses,” she said. “This is a very crowded space. But, thankfully, many of us are studying it, and hopefully we will shed some light on these areas.”