Path to importance of PD-L1 status in breast cancer

Mary Jane Friedrich

June 2019—New data support testing patients for their PD-L1 immune cell status when they are diagnosed with metastatic or unresectable locally advanced triple-negative breast cancer to determine if they might benefit from a checkpoint inhibitor.

The phase three IMpassion130 study led to accelerated approval by the Food and Drug Administration for the use of atezolizumab in combination with the chemotherapy drug nab-paclitaxel for patients with metastatic triple-negative breast cancer whose tumors are infiltrated with immune cells that express PD-L1 (Schmid P, et al. N Engl J Med. 2018;​379​[22]:​2108–2121).

“It’s PD-L1 immune cells, not tumor cells. This is in contrast to what we do clinically for non-small cell lung carcinomas,” said Ashley Cimino-Mathews, MD, associate professor of pathology and oncology, Johns Hopkins University School of Medicine, and director of the breast pathology program, Johns Hopkins Hospital.

Dr. Cimino-Mathews co-presented a CAP TODAY webinar with an author of the IMpassion130 study, Leisha A. Emens, MD, PhD, director of translational immunotherapy for the Women’s Cancer Research Center and co-leader of cancer immunology and immunotherapy, UPMC Hillman Cancer Center/Magee Womens Hospital, Pittsburgh. The webinar was made possible by a promotional sponsorship from Genentech.

“It is any immune cell within the tumor area, including macrophages, neutrophils, and lymphocytes, because that’s what was scored in the study,” Dr. Cimino-Mathews said of the trial that led to the breakthrough for triple-negative breast cancer patients.

And the immune cells have to occupy greater than or equal to one percent of the tumor area. “This is not a percentage of immune cells that are positive,” she said, adding, “It harkens back to using a percentage stromal area to calculate tumor-infiltrating lymphocytes within a breast cancer.”

The big question now, she said, is which antibody or assay should be used to test PD-L1 status in breast cancer. The only FDA-approved assay now for triple-negative breast cancer (TNBC) is the Ventana SP142 assay, which is the one used in the IMpassion130 trial. Other assays are undergoing validation studies and may gain FDA approval for use in TNBC, she said.

The cellular milieu of the breast tumor microenvironment includes not only carcinoma cells but also stromal myofibroblasts and endothelial and immune cells, Dr. Cimino-Mathews noted. Immune cells can be dispersed individually or arranged in aggregates that roughly resemble the germinal centers in lymph nodes. The latter arrangement, called a tertiary lymphoid structure, is thought to play an important role in lymphocyte trafficking to and from the tumor microenvironment.

Immune cells play varying roles in the equilibrium of cancer immune surveillance and help shape the microenvironment to be pro or anti-tumorigenic. CD8-positive T cells, Th1 helper T cells, M1 macrophages, and innate immune cells such as neutrophils, if activated, can recognize neoantigens and mount an antitumor immune response that can lead to tumor elimination.

“In contrast,” she said, “a shift toward FoxP3-positive regulatory T cells, Th2 helper T cells, and M2 macrophages can actually shift this immune milieu to its pro tumorigenic environment.”

Some breast tumors contain brisk lymphocytic infiltrate throughout the stromal space. Others have little to no lymphocyte infiltration. Tumor-infiltrating lymphocytes, or TILs, are emerging as a therapeutic target.

An international TILs working group in 2015 proposed guidelines for how to measure TILs in breast cancer (Salgado R, et al. Ann Oncol. 2015;26[2]:259–271). The recommendation is to identify the actual tumor area and only score the lymphocytes or immune cells within the invasive margin of the tumor, omitting necrotic areas, and then to score the overall stromal cellularity, Dr. Cimino-Mathews said. “You don’t want to score inflammatory cells that are far outside the actual carcinoma itself.”

The guidelines also recommend scoring only the percentage of the stromal area that’s occupied by TILs. Intratumoral TILs lying within the cancer cell mass are difficult to pick up on H&E alone and thus should be excluded. Stromal TILs are easier to detect and the recommendation is to score as a continuous variable from zero to 100 percent. “In reality,” she said, “we’ve probably been using 10 percent or five to 10 percent intervals.”

Evidence of improved survival with increasing degree of inflammation is seen most strongly in triple-negative and HER2-amplified breast carcinomas, Dr. Cimino-Mathews said. She described a study of 900 patients with TNBC that showed that an increasing degree of stromal TIL is associated with improvements in disease-free survival, distant-disease-free survival, and overall survival (Pruneri G, et al. Ann Oncol. 2016;27​[2]:​249–256).

Another study evaluated whether TIL in a tumor after treatment could predict prognosis in 300 patients with TNBC who were treated with neoadjuvant chemotherapy. Patients with a high level of TIL in residual disease after treatment—greater than 60 percent of stromal space occupied by TIL—were found to have improved metastasis-free survival and overall survival (Dieci MV, et al. Ann Oncol. 2014;25​[3]:​611–618).

“So the tumor-infiltrating lymphocytes appear to be important both before and after chemotherapy,” Dr. Cimino-Mathews said.

Proteins expressed by tumor and immune cells can stimulate or inhibit the antitumor response through a complicated interplay of protein-protein interactions at the interface of antigen-presenting cells and lymphocytes, she said. Those protein interactions that inhibit an immune response are called immune checkpoint signals. Checkpoint inhibitors can overcome a cancer cell’s evasive checkpoint maneuver and expose them to immune assault. “So what we’re trying to do with checkpoint inhibition is to apply therapies that put a brake on the brake—or that release the brake—and allow for antitumor T cell activation,” Dr. Cimino-Mathews said.

Several checkpoint inhibitors have been developed to block the PD-1/PD-L1 checkpoint pathway, including atezolizumab, which is a PD-L1 antagonist that selectively targets PD-L1 to prevent interaction with the receptor PD-1 and reverse immune cell suppression. These checkpoint inhibitors are being investigated in breast cancer along with checkpoint inhibitors that target CTLA4 (cytotoxic T-lymphocyte-associated antigen) and LAG-3 (lymphocyte activation gene-3), among others.

In breast carcinomas—and this varies among tumor types—PD-L1 can be expressed either by immune cells, including TIL, the macrophages, and neutrophils, or by the carcinoma cells themselves. The PD-1 receptor is expressed on the TIL.

Why would a tumor cell express PD-L1, which is part of an immune checkpoint inhibitor pathway? One mechanism, she said, is considered an adaptive pattern of expression and can be thought of purely as the tumor’s immune evasion strategy. “So in a tumor microenvironment that has a brisk inflammatory infiltrate, some of the cancer cells will try to stop this inflammatory, antitumor response by upregulating PD-L1 expression at the leading edge, or interface, with the inflammatory cells.”

The other pattern of expression is seen in a subset of tumors, such as a class of Hodgkin lymphoma, in which the neoplastic cells constitutively express PD-L1 on every cell. “And that’s due to amplifications in genes or gene mutations that lead to tumor cell PD-L1 expression,” whether or not inflammatory cells are present.

In breast carcinomas, PD-L1 labeling is seen on immune cells and carcinoma cells. In ductal carcinomas, this labeling occurs primarily according to an adaptive pattern of expression where PD-L1 is expressed on carcinoma cells that interface with inflammatory cells. In contrast to other tumor types, in breast carcinomas, tumor cell PD-L1 labeling is relatively rare compared with immune cell labeling.

So triple-negative and HER2-positive breast carcinomas not only have the highest degree of TIL but also are most likely to be PD-L1-positive, making them the most attractive candidates for immunotherapy.

Triple-negative breast cancer is the most likely of the three clinically relevant breast cancer types to be inflamed, or “hot,” with HER2-positive breast cancers just behind them, Dr. Leisha Emens said, noting they tend to have high numbers of CD3-positive T cells. At the other end of the spectrum are ER-positive luminal breast cancers, which tend to be devoid of T cells and are “cold,” she said.

The T cells in the tumor microenvironment of triple-negative and HER2-positive breast tumors are poised to respond to immune checkpoint blockade. ER-positive breast tumors lacking T cells are not likely to respond to immunotherapy agents because “there’s no gas in the tank,” as she put it.

A large body of data suggests that breast tumors with poor prognostic factors—including those that are ER- and PR-negative, of high grade, or lymph-node–positive—tend to be associated with higher T cell infiltrates at diagnosis, Dr. Emens said. Studies also have found higher numbers of CD8-positive TILs and a higher ratio of “good” CD8-positive T cells to “bad” FOXP3-positive T cells predict better clinical outcomes with immunotherapy, including a complete pathologic response to neoadjuvant therapy and longer disease-free and overall survival, except for ER-positive breast cancers.

To give a sense of the difference in the prevalence of the PD-L1 biomarker between triple-negative and ER-positive breast cancer, Dr. Emens discussed results from two phase one studies with the anti-PD-1 checkpoint inhibitor pembrolizumab. Patients were selected for both studies based on PD-L1 expression within the tumor microenvironment.

In the TNBC study, 58 percent of the 111 patients screened had PD-L1-positive tumors. In the 27 patients who underwent treatment with pembrolizumab, a response rate of almost 19 percent was achieved, including one complete response and several partial responses (Nanda R, et al. J Clin Oncol. 2016;34[21]:2460–2467). In contrast, in the ER-positive breast cancer study, of the 248 tumors screened, only 19.4 percent were PD-L1-positive. And of the 20 patients who received pembrolizumab, there was a response rate of 12 percent, with no complete responses and three partial responses (Rugo HS, et al. San Antonio Breast Cancer Symposium, 2015).

“This illustrates the differences in immunobiology between triple-negative breast cancer and ER-positive breast cancer, but shows there is potential for response if the preexisting immunobiology is there with this agent.”

In another study, Dr. Emens and her colleagues tested atezolizumab in 115 patients with advanced metastatic TNBC. Most patients were heavily pretreated before receiving atezolizumab, with only 17 percent of patients having had no prior treatment for metastatic disease (Emens LA, et al. JAMA Oncol. 2019;5[1]:74–82). Sixty-three percent of patients enrolled in the phase 1A study were PD-L1-positive and 33 percent PD-L1-negative.

The median duration of response was 21 months. Median progression-free survival was 1.4 months by response evaluation criteria in solid tumors, or RECIST, and 1.9 months by immune-related RECIST—not terribly impressive, Dr. Emens said. The overall response rate by RECIST, which included complete and partial responses, was 10 percent, and increased to 13 percent by immune-related RECIST.

PD-L1-positive expression in tumor-infiltrating immune cells of at least one percent was associated with a better outcome. Patients with PD-L1-positive tumors had a median overall survival of 10.1 months compared with six months in those with PD-L1-negative tumors. One-year OS for PD-L1-positive patients was 44 percent compared with 32 percent for PD-L1-negative patients. OS rates for PD-L1-positive patients at two and three years were 25 percent and 21 percent respectively.

The small number of patients who received atezolizumab as a first-line treatment had better outcomes with regard to OS (median 17.6 months) than the groups of patients who received atezolizumab second-line or later (7.3 months). The one-year OS rate for first-line patients was 59 percent and for second-line or greater, 37 percent.

The data also show that patients who had a complete or partial response to single-agent atezolizumab tend to do quite well: 100 percent of patients who had an objective response by RECIST were alive at one and two years, and even out to three years if the objective response rate was by immune-related RECIST. “For these patients lucky enough to respond to this therapy, it can be quite transformative for them,” Dr. Emens said.

PD-L1 and stromal TILs are likely to be imperfect biomarkers, she said, and the context in which they’re found matters. She discussed a phase two study (KEYNOTE-086) of single-agent pembrolizumab in patients with TNBC that consisted of two cohorts: cohort A enrolled patients previously treated for metastatic disease whose tumors expressed any level of PD-L1, while cohort B enrolled patients previously untreated for metastatic disease who were PD-L1-positive. The 52 first-line PD-L1-positive patients from cohort B had a response rate of 23 percent, which was similar to the response rate seen in the first-line setting with single-agent atezolizumab in the phase 1A study. In the previously treated patients in cohort A who had any level of PD-L1 expression, response rates ranged from four percent to 10 percent, depending on the level of expression, findings similar to those in the phase 1A study. Patients treated first line tended to have more TIL than patients in later lines.

While these studies provided preliminary evidence that therapy with immune checkpoint inhibitors can play a role in treating breast cancer, results from the randomized, double-blind phase three IMpassion130 trial “bring a potentially effective targeted therapy to the treatment armamentarium for this group of patients,” Dr. Emens said.

The IMpassion130 trial compared atezolizumab plus nab-paclitaxel with placebo plus nab-paclitaxel in patients with inoperable locally advanced or metastatic TNBC who had not received prior treatment for metastatic disease. Previous chemotherapy was allowed if it had been administered at least 12 months before this trial.

Patients were stratified by whether they had liver metastases, prior treatment with taxane, and PD-L1 status on immune cells. PD-L1 positivity on immune cells was defined as levels of one percent or more, with 41 percent of enrolled patients PD-L1-positive by this definition and 14 percent having PD-L1-positive expression at five percent or more. While PD-L1-positive expression wasn’t required for eligibility, the researchers analyzed the possible association of this biomarker with clinical benefit.

The findings revealed that the atezolizumab plus nab-paclitaxel combination led to significantly longer progression-free survival than was seen with placebo plus nab-paclitaxel in both the intention-to-treat population and the PD-L1-positive subgroup. In the PD-L1-positive subgroup, median PFS was longer by 2.5 months in the atezolizumab group than in the placebo group (7.5 months versus five months, respectively), and median OS was 10 months longer at the interim analysis (25 months versus 15.5 months).

No treatment effect was seen in progression-free or overall survival with the addition of atezolizumab to nab-paclitaxel in patients who were PD-L1-negative, Dr. Emens said. The addition of atezolizumab did not compromise the dose intensity of nab-paclitaxel.

Dr. Emens and colleagues analyzed other biomarkers that could be reflective of preexisting immunobiology and associated with clinical benefit from anti-PD-L1/PD-1 checkpoint inhibitors. They evaluated PD-L1 expression on tumor cells using the Ventana SP142 immunohistochemical assay, which was also used to evaluate immune cells, intratumoral CD8-positive T cells by IHC (Dako clone C8/144B) and stromal TIL using H&E, and BRCA1/2 mutation status using the FoundationOne assay.

“As long as patients had PD-L1 expression on immune cells at levels of one percent or more,” Dr. Emens said, “they derived clinical benefit for progression-free and overall survival. So the one percent cutoff appears to define a threshold above which patients can respond.”

Patients with CD8-positive tumors derived clinical benefit only if their tumors were also PD-L1 immune cell positive, she said, adding that the same is true for stromal TIL. And there was no association of benefit with the presence of a BRCA1 or 2 mutation unless the tumor was PD-L1 immune cell positive.

The immunobiology of patients seems to change dramatically after first-line therapy, Dr. Emens said. After that, patients are much less likely to be PD-L1-positive and their tumors are much less likely to contain TIL.

“All of this data taken together,” she said, “strongly support testing patients for their PD-L1 immune cell status when they are newly diagnosed with metastatic or unresectable locally advanced triple-negative breast cancer to determine if they might benefit from atezolizumab with nab-paclitaxel.” 

Mary Jane Friedrich is a writer in Lake Forest, Ill.