Research into the mechanisms that promote resistance to immune checkpoint blockade therapies suggests the need for new strategies for patients with immunorefractory disease.
Julie R. Brahmer, MD
Research into the mechanisms that promote resistance to immune checkpoint blockade therapies suggests the need for new strategies for patients with immunorefractory disease. Despite increased attention in recent years to tumor cell-extrinsic mechanisms and the roles that cells in the tumor microenvironment play, there are still many unanswered questions surrounding the causes of resistance, according to Julie R. Brahmer, MD, director of the thoracic oncology program and professor of oncology at The Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins University in Baltimore, MD. “Clarifying mechanisms of resistance to immune checkpoint inhibitors will help with development of strategies for patients who do not respond to or develop aquired resistance to immunotherapy,” Brahmer said in an interview with OncologyLive®.
Role of the Microenvironment
Resistance to immune checkpoint blockade is generally classified as primary resistance, referring to a lack of response at the start of therapy, or acquired resistance, which develops after an initial response to therapy.
Tumor-intrinsic mechanisms include oncogenic signaling through the MAPK pathway that reduces tumor-infiltrating lymphocytes (TILs),1 loss of phosphatase and tensin homolog (PTEN) expression,2 and stabilization of β-catenin/constitutive Wnt signaling.3 These signaling pathway aberrations may cause inhibition of T-cell recruitment and function, exclusion of T cells from the tumor microenvironment, or reduced expression of interferon-γ.
Many tumors are infiltrated with regulatory T cells, which contribute to maintaining self-tolerance and are thought to promote tumor progression and suppress effector T cells. A low ratio of regulatory to effector T cells was associated with improved response to anti—CTLA-4 therapy in murine models.4 However, a retrospective analysis of patients with melanoma treated with anti—CTLA-4 therapy in a phase II trial showed that high expression of regulatory T cells at baseline was associated with better clinical outcomes.5
biomar Myeloid-derived suppressor cells (MDSCs) are also key regulators of immune response, and higher frequencies of monocytic MDSCs in the tumor microenvironment were associated with lower efficacy of ipilimumab (Yervoy) in patients with melanoma.6
Tumors may also secrete certain chemokines, such as the ligands CCL5, CCL7, and CXCL8, that bind to receptors CCR1 and CXCR2 expressed on subtypes of MDSCs to recruit them into the tumor microenvironment. Disruption of CXCR2-mediated MDSC trafficking was therefore shown to enhance efficacy of PD-1 checkpoint blockade in a murine model of rhabdomyosarcoma.7 Tumor-associated macrophages (TAMs) in the tumor microenvironment also interfere with T-cell responses and have been implicated as a source of primary resistance to immune checkpoint therapy.8
Brahmer pointed out that some tumor cells have an absence or loss of HLA, β2 microglobulin, and/or major histocompatibility complex (MHC), thereby creating an immunologically “cold” tumor microenvironment because the tumor cells are unable to present their antigens to the immune system. She said that acquired resistance can also occur following genetic events that cause the tumor to decrease expression of tumor antigens previously recognized by antigen-specific T cells.
“It can be a complete immune desert, where there are no T cells in the tumor [microenvironment],” Brahmer said. “It’s such an immunosuppressive microenvironment that there’s no way checkpoint inhibitors can work.”
She added that treatments that a patient received prior to or concurrent with immunotherapy likely play a role in the mechanisms of acquired resistance, such as upregulation of VISTA, LAG-3, and TIM-3 immune checkpoint pathways, in the tumor microenvironment in response to anti—PD-1/PD-L1 and anti-CTLA-4 therapy.
Although the mechanisms for primary and acquired resistance are generally different, some mechanisms, such as increased infiltration of immunosuppressive cells into the tumor microenvironment through the upregulation of cytokines, likely play a role both types of resistance, Brahmer said.
Identifying these immunoresistant niches in the microenvironment and the respective therapies is the first step to developing personalized regimens. In a review published in the Lancet by Syn et al9, data collected from 2008 through 2017 was used to outline a framework to contextualize mechanisms of tumor escape and help form a basis for therapeutic efforts (Figure9,10). Six mechanisms were identified as playing a role in the development of both acquired and primary resistance.
Although much of the recent research has focused on the suboptimal tumor microenvironment as a cause of resistance to immune checkpoint inhibitors, Brahmer pointed out that environmental factors and patient comorbidities can also affect the function of the immune system and thus response to therapy.
Recent research on the gut microbiota has been particularly noteworthy; a recent study showed that patients with melanoma that responded to immune checkpoint blockade had higher α-diversity and a relative abundance of bacteria in the Ruminococcaceae family than did patients whose melanoma did not respond.11
Biomarkers of Response
A follow-up analysis presented at the American Association for Cancer Research (AACR) Annual Meeting 2019 showed that patients with melanoma who consumed a high-fiber diet had a more diverse gut microbiome and better response to anti—PD-1 therapy than patients who consumed a low-fiber diet.12
Because checkpoint inhibitor therapy is not effective in all patients, is costly, and has associated toxicities, identification of biomarkers that predict response has been a key focus of recent research. However, the complex and dynamic nature of the immune system has made effective identification of reliable biomarkers difficult.
Studies showing the relationship between high microsatellite instability (MSI-H)/ mismatch repair (MMR) deficiency (dMMR) and durable response to immune checkpoint inhibitors led to the first biomarker-based, tissue-agnostic FDA approval for pembrolizumab (Keytruda) in previously treated MSI-H/dMMR cancer regardless of primary tumor location.
However, MMR/MSI status does not identify all patients who could benefit from immune checkpoint blockade, which highlights the need for other biomarkers that could predict response. PD-L1 expression is the most commonly used biomarker for predicting response to anti—PD-1 therapy, but the variability in immunohistochemistry assays and cutoffs for expression have introduced the need for additional biomarkers, such as tumor mutational burden (TMB) with genomic analysis of specific mutations.
MSI-H/dMMR tumors have a much higher number of somatic mutations and potential neoantigens per tumor, as well as a denser infiltration of CD8-positive TILs, than MMR-proficient tumors. These characteristics contribute to their increased immunogenicity and better and more durable responses to checkpoint inhibitor therapy across multiple trials.13
The FDA approval of pembrolizumab for previously treated dMMR/MSI-H tumors was based on a pooled analysis of 149 patients (90 with colorectal cancer and 59 with 1 of 14 other cancer types) from 5 clinical trials showing an overall response rate of 39.6% and a durable (≥6 month) response in 78% of responders.14
However, the prevalence of MSI-H/dMMR in cancers ranges from 0% to 31% and is very low in many common types of cancer, including breast cancer, prostate cancer, and lung adenocarcinoma13; therefore, determining additional reliable biomarkers of response to immunotherapy is necessary.
Immunohistochemical expression of PD-L1 is the most commonly used biomarker to predict response to PD-1/PD-L1 checkpoint inhibitors. Clinical trials show an improved response to anti—PD-1 immunotherapy in multiple types of PD-L1-positive cancers, and follow-up analyses of the KEYNOTE-001 trial showed a 5-year overall survival rate of 29.6% in treatment-naïve patients with nonsmall cell lung cancer (NSCLC) and a PD-L1 tumor proportion score ≥50% following pembrolizumab monotherapy, whereas the historical 5-year survival rate is about 5.5% in this population.15
Furthermore, results from the KEYNOTE-042 trial showed a significant improvement in overall survival (OS) in patients with advancedstage NSCLC and tumor proportion score ≥1%, leading to an expanded indication for pembrolizumab monotherapy in the frontline setting for patients with stage III NSCLC whose disease is not metastatic and who are not candidates for surgical resection or definitive chemoradiation.16
However, multiple factors have complicated the interpretability of PD-L1 immunohistochemistry assays, including the wide variability in PD-L1—binding antibodies used for detection, the differences in criteria for determining PD-L1 positivity across studies, and the high spatial and temporal heterogeneity of PD-L1 expression in the tumor microenvironment.
Furthermore, Brahmer pointed out, many patients with PD-L1 expression in ≥50% of cells do not respond at all to anti—PD-1/PD-L1 checkpoint blockade: “We have some patients with high PD-L1 who have immediate progression, and there’s no obvious clear cause.”
Brahmer also noted that a small proportion of patients with low or no PD-L1 expression respond to anti—PD-1/PD-L1 therapy. Figuring out potential mechanisms for response in these patients should be a key area of focus in future research, she said. “I have some patients with low TMB and low PD-L1 [expression] who respond to immunotherapy,” Brahmer said. Although the reasons for response remain unclear, she added that “it just takes 1 T cell to recognize an abnormal protein that the cancer makes and expresses, and then they can recruit more T cells to attack that particular abnormality.”
Tumor Mutational Burden
High TMB leads to creation of neoantigens, which are thought to increase tumor immunogenicity and promote a favorable response to immune checkpoint inhibitors. For this reason, use of TMB as a biomarker for response to checkpoint blockade has been of interest.
A recent literature search and analysis of 27 tumor types or subtypes showed a significant correlation between high TMB and objective response rate to anti—PD-1/PD-L1 monotherapy.17 However, the authors of the analysis noted that some tumor types demonstrated better or worse responses than were predicted by TMB. For example, Merkel cell carcinoma can have a high TMB and be virus negative or it can be virus associated, often with a low TMB, yet both subtypes had relatively high objective response rates (44% and 62%, respectively), presumably because the antigens expressed by oncogenic viruses also represent T-cell targets.18
Recent results presented at the International Association for the Study of Lung Cancer 2019 World Conference on Lung Cancer showed that loss-of-function mutations in STK11, KEAP1, and PTEN, as well as ERBB2 exon 20 insertion mutations, were negative predictors of response to pembrolizumab in metastatic NSCLC, and a combined prediction method that included these mutations along with TMB improved prediction of progression-free survival (HR, 0.18; 95% CI, 0.08-0.41) and overall survival (HR, 0.27; 95% CI, 0.1-0.73) over TMB alone.19
Although more data are needed to support adding genetic analyses of these genes to TMB, Brahmer suggested that examination of TMB status could be used in the future to help a physician decide whether to start a patient on PD-1/PD-L1 checkpoint inhibitor therapy. “I can tell a patient…‘You have these mutations, so you’re less likely to do well with this combination and maybe we should consider a clinical trial,’” she said.
Strategies to Improve Response
Given the modest overall response rates to immune checkpoint inhibitors, investigating ways to sensitize tumors to immunotherapy and overcome primary and acquired resistance has been the focus of many preclinical and clinical trials.
Specifically, dual checkpoint blockade and combining immune checkpoint inhibitors with molecularly targeted therapies have been investigated for transforming an immunologically cold tumor into a “hot” tumor that would respond to checkpoint inhibitor therapy.
Combination therapy with the CTLA-4 inhibitor ipilimumab and the PD-1 inhibitor nivolumab (Opdivo) has been shown to yield better response rates and survival in patients with metastatic melanoma than either checkpoint inhibitor alone,20 and CTLA-4 inhibition may facilitate the conversion from a cold to a hot tumor microenvironment through selective depletion of regulatory T cells.21
Another study showed that blockade of the colony-stimulating factor 1 receptor (CSF1R), which is expressed by monocytes, monocytic MDSCs, and TAMs, reprogrammed macrophage responses to enhance antigen presentation and improve responses to antiPD-L1 and CTLA-4 checkpoint therapy in a murine model of pancreatic cancer.22 An ongoing phase Ib/II trial (NCT02880371) is investigating the CSF1R inhibitor ARRY-382 in combination with pembrolizumab for patients with advanced solid tumors.
BRAF-targeted therapy in melanoma is associated with increased antigen and HLA expression, increased T-cell infiltration, and reduced levels of immunosuppressive cytokines in the tumor microenvironment, potentially enhancing responses to anti—PD-1/ PD-L1 therapy.23,24 A phase Ib dose-escalation study found an unconfirmed response rate of 85.3% in the first 34 patients with BRAF-mutated melanoma who received vemurafenib (Zelboraf), cobimetinib (Cotellic), and the anti—PD-L1 immunotherapy agent atezolizumab (Tecentriq).25 The randomized phase II component of the KEYNOTE-022 trial showed that 60% of patients who received pembrolizumab, dabrafenib (Tafinlar), and trametinib (Mekinist) had responses lasting longer than 18 months versus 28% of patients who received placebo, dabrafenib, and trametinib.26
The recent discovery of the link between the gut microbiome and response to anti—PD-1 therapy prompted studies involving optimization of the gut microbiome, including a phase I clinical trial that will investigate responses to fecal microbial transplantation from a healthy donor in addition to pembrolizumab or nivolumab in patients with unresectable or metastatic melanoma (NCT03772899).
In addition to showing a relationship between a high-fiber diet and improved response to anti—PD-1 therapy, the preliminary results presented at AACR Annual Meeting 2019 showed that over-the-counter probiotics and antibiotic usage were associated with a decreased chance of response to immunotherapy in patients with melanoma,12 suggesting that modification of diet and supplement intake to promote high gut microbial diversity during treatment with immune checkpoint inhibitors may increase likelihood of response.
“The hope is [that] in the future, we will be able to identify different ways to get around that resistance [to immunotherapy],” said Brahmer.