METASTATIC TRIPLE-NEGATIVE BREAST CANCER MAY BENEFIT FROM ANTI-PD-L1 MAINTENANCE
GS3-02. Durvalumab compared to maintenance chemotherapy in patients with metastatic breast cancer: Results from phase II randomized trial SAFIR02-IMMUNO
Florence Dalenc, PhD (Institut Claudius Regaud, Toulouse, France) presented findings from the phase II SAFIR02-IMMUNO study of patients with HER2-negative metastatic breast cancer. After 6 to 8 cycles of chemotherapy, patients without a targetable molecular alteration were randomized 2:1 to durvalumab or maintenance chemotherapy (paclitaxel, capecitabine, or FEC). The primary endpoints of the study at two years were progression-free survival (PFS) and overall survival (OS).
A total of 191 patients have been randomized in the study, meeting the enrollment goal, of whom 45.5% had triple negative breast cancer (TNBC) and 53.4% were hormone receptor positive (HR+). There was a slight imbalance in breast cancer subtype between the arms with more patients in the chemotherapy maintenance arm having TNBC (52.2% vs 37.6%) and more patients in the durvalumab arm having HR+ breast cancer (60.8% vs 47.8%), although this was not significant (p =.09). Patients in both arms were matched by age, functional status, line of chemotherapy and tumor burden. Of the 133 patients who were tested, approximately 1/3 of patients in both arms were PD-L1 positive per the SP142 assay with PD-L1 positivity defined as IC >= 1%.
There were no differences in PFS or OS in the overall population between arms. Exploratory analysis demonstrated that the median OS in patients with TNBC was statistically improved for patients who received durvalumab maintenance rather than chemotherapy maintenance (21 vs 14 months, HR 0.54, p = 0.0377) and approached, but did not reach, significance in patients with PD-L1-positive tumors (26 vs 12 months, HR 0.42, p = 0.0552). Exploratory analyses also showed that anti-PD-L1 as a single agent was less effective than maintenance chemotherapy in HR+ breast cancer (HR 2.08, p = 0.0025). Safety analysis demonstrated a similar rate of treatment-related grade 3 and 4 adverse events (AEs; 13.2% in durvalumab arm vs 15.9% in chemotherapy maintenance arm) and of AEs leading to discontinuation of treatment (6.2% vs 9.5%, respectively), but a higher rate of treatment-related severe AEs in the durvalumab arm (18.6% vs 1.6%, respectively).
PATHOLOGIC COMPLETE RESPONSES ENHANCED WITH NEOADJUVANT PEMBROLIZUMAB + CHEMOTHERAPY
GS3-03. Keynote-522 study of pembrolizumab + chemotherapy vs placebo + chemotherapy as neoadjuvant treatment, followed by pembrolizumab vs placebo as adjuvant treatment for early triple-negative breast cancer: Pathologic complete response in key subgroups
Pathologic response results from KEYNOTE-522 were presented by Peter Schmid, PhD (Barts Cancer Institute, Queen Mary University of London, London, UK). Patients with early stage TNBC were randomized 2:1 to pembrolizumab every 2 weeks or placebo plus standard-of-care neoadjuvant chemotherapy with carboplatin + paclitaxel followed by doxorubicin (or epirubicin) + cyclophosphamide (CP > AC). Following surgery, patients continued on adjuvant therapy with pembrolizumab or placebo for an additional year. KEYNOTE-522 had dual primary endpoints of pathologic complete response (pCR) rate and event free survival (EFS).
One thousand, seventy four patients enrolled on trial, with 784 receiving pembrolizumab along with standard of care treatment. Patients in the two arms were balanced by age, functional status, tumor size and nodal involvement. Approximately 80% of patients in both arms were PD-L1 positive using the 22C3 pharmDx assay with PD-L1 positive defined as CPS >= 1.
pCR rates were significantly higher in patients who received neoadjuvant pembrolizumab + chemotherapy compared to placebo + chemotherapy (64.8% vs 51.2%, p = 0.00055). The greatest improvement in pCR was seen in patients with Stage 3 breast cancer, where the addition of neoadjuvant pembrolizumab to chemotherapy improved pCR rates by approximately 25%, as well as in patients with positive lymph nodes, where pCR rates were improved by 20% over placebo + chemotherapy. Furthermore, pCR rates were improved in patients receiving neoadjuvant pembrolizumab + chemotherapy compared to placebo + chemotherapy who were PD-L1 negative (CPS < 1) as well as in patients who were PD-L1 positive (CPS >= 1). The first pre-planned interim analysis also demonstrated improved EFS in patients who received pembrolizumab compared to placebo at 15.5 months median follow-up (7.4% events in pembrolizumab vs 11.8% in placebo; HR 0.62). Safety analysis showed immune-mediated adverse events that were consistent with the known profiles of each regimen and did not uncover any new safety concerns.
ADDITION OF ATEZOLIZUMAB DOES NOT APPEAR TO ENHANCE PCR RATES IN EARLY BREAST CANCER
GS3-04. Pathologic complete response (pCR) to neoadjuvant treatment with or without atezolizumab in triple negative, early high-risk and locally advanced breast cancer. NeoTRIPaPDL1 Michelangelo randomized study
The NeoTRIPaPDL1 study of patients with early stage, high risk (T1cN1, T2N1, or T3N0) or locally advanced TNBC was discussed by Luca Gianni, MD (Fondazione Michelangelo, Milano, Italy). Prior to surgery, patients received either chemotherapy (carboplatin AUC2 + nab-paclitaxel) for 8 cycles +/- atezolizumab, followed by an adjuvant anthracycline regimen. Event-free survival at 5 years was the primary endpoint of this study, with pathologic complete response as a key secondary goal, the results of which were presented by Dr. Gianni.
Two hundred and eighty patients enrolled in the study with half (n = 138) receiving atezolizumab + chemotherapy. Patients in the two arms were matched on age, T stage, nodal status, disease stage and PD-L1 positive status. Approximately 50% of patients in both arms were PD-L1 positive per the Ventana SP142 assay with PD-L1 positivity defined as IC >= 1.
In the intent-to-treat population, there was no difference in pCR rate between those who received atezolizumab and those who received placebo (43.5% vs 40.8%, respectively). In multivariate analysis, the presence of PD-L1 positivity was the most significant factor influencing pCR (OR 2.08) but in this study, the addition of atezolizumab to chemotherapy did not improve pCR rates in patients with PD-L1 positive tumors compared to those who received placebo (51.9% in atezolizumab arm vs 48.0% in placebo arm). Patients with PD-L1 positive tumors had higher pCR rates compared to those with PD-L1 negative tumors, regardless of the regimen used (chemotherapy + atezolizumab or placebo). Continuous follow-up for EFS is ongoing at this time and was not presented. Safety analysis demonstrated side effect profiles consistent with the agents except for a higher overall incidence of SAEs (18.1% vs 5.7%, respectively) and liver transaminase abnormalities with atezolizumab.
DISCUSSANT KEVIN KALINSKY, MD OF COLUMBIA UNIVERSITY MEDICAL CENTER
Dr. Kalinsky provided a thorough discussion of the conflicting results of the KEYNOTE-522 and NeoTRIPaPDL1 trials. Notably he highlighted that while higher pCR rates have been used as a surrogate for improved clinical outcomes, not all agents that have led to higher pCR rates have resulted in improved EFS in breast cancer (e.g., neoadjuvant bevacizumab or neoadjuvant carboplatin). Consistent with the ISPY2 trial, KEYNOTE-522 found an improved pCR rate with the addition of the anti-PD-1 antibody pembrolizumab to neoadjuvant chemotherapy involving a taxane and anthracycline. However, the NeoTRIPaPDL1 trial was consistent with previously published results of GeparNeuvo, where the addition of an anti-PD-L1 antibody to neoadjuvant chemotherapy did not improve pCR rates. Dr. Kalinsky highlighted that, in early breast cancer, the choice of anti-PD-1 versus anti-PD-L1 may make a difference in clinical outcomes citing that an anti-PD1 antibody blocks both PD-L1 and PD-L2 where as anti-PD-L1 only blocks PD-L1 and allows PD-L2 inhibitory signaling. Another key difference between KEYNOTE-522 and NeoTRIPaPDL1 was the clinicopathologic features of the patient populations with more node positive, higher-stage patients in NeoTRIPaPDL1 (Node positive 87% in NeoTRIP vs 51% in KEYNOTE; Stage 3 49% in NeoTRIP vs 25% in KEYNOTE). Finally, the chemotherapy backbones were also different between the two trials with the anthracycline being administered in the adjuvant setting in the NeoTRIPaPDL1 trial. Given that anthracyclines have positive immune effects like MDSC depletion, increased type I IFN and immunogenic cell death, the omission of the anthracycline in the neoadjuvant setting in these patients with higher volume/stage disease may also have negatively affected the pCR rates.
CHANGES IN MOLECULAR BREAST CANCER SUBTYPES OBSERVED WITH PROGRESSION
GS3-08. Multiplatform analysis of matched primary and metastatic breast tumors from the AURORA US Network
Results of the retrospective phase of the AURORA US study were presented by Tari A. King, MD, FACS (Dana Farber Cancer Institute, Boston, MA). This investigation compared paired primary and metastatic tissue samples from 55 breast cancer patients, integrating, among other factors, genomic, epigenetic, and structural data. Samples from metastatic sites were collected via biopsy (n = 35) as well as during autopsy (n = 20).
Most patients had intraductal carcinoma (84%) and were diagnosed with early stage breast cancer (Stage 1: 16%, Stage 2: 44%). Median age at diagnosis was 49 years old with 16% of patients less than 40 years old at the time of their diagnosis. Half of patients had a family history of breast cancer. The molecular subtype distribution of the primaries was well-distributed, including 31% basal-like, 31% Luminal A, 13% normal-like, 9% HER2-enriched, and 1 Luminal B tumor. Most patients received adjuvant chemotherapy (76%) and had a median disease-free survival before metastatic diagnosis of 2.5 years. Half of patients were diagnosed with a single site of metastatic disease (lung > liver > bone) and survived a median of 1.5 years after developing metastatic disease. Several trends were noted when comparing intra-patient primaries and metastases. Basal-like primaries generally led to basal-like metastases with only 10% of cases evolving into another molecular subtype, while metastases from luminal primary tumors evolved into a different subtype in approximately 30% of metastatic cases. Considering all metastatic samples, increases in metabolism pathways and proliferation were observed, while the immune infiltrate and differentiation signatures were decreased in metastases. Across the majority of patients, CpG island hypermethylation persisted across primary and metastatic samples; in particular, promoter CpG hypermethylation was noted in some metastases at JAM3, which is critical for cellular adhesion. Dr. King highlighted the genomic heterogeneity of a patient with a HR+/HER2- primary breast cancer that developed multiple clones over the course of treatment and that these clones induced downstream genomic changes including the transition to a HER2+ subtype. This study highlighted the role for sequential biopsies of metastatic sites to reassess the molecular subtype, as well as other malleable factors of the tumor microenvironment of breast cancer throughout treatment.
NIVOLUMAB, PACLITAXEL, AND BEVACIZUMAB MAY BENEFIT HR+ BREAST CANCERS
PD1-03. A multicenter phase II study evaluating the efficacy of nivolumab plus paclitaxel plus bevacizumab triple-combination therapy as a first-line treatment in patients with HER2-negative metastatic breast cancer: WJOG9917B NEWBEAT trial
Yukinori Ozaki, MD (Toranomon Hospital, Tokyo, Japan) presented a phase II study of nivolumab, paclitaxel, and bevacizumab combination therapy as a first-line treatment for HER2-negative breast cancer. This single-arm study enrolled 57 evaluable patients (39 HR+HER2-, 18 TNBC) in order to measure the objective response rate as the primary endpoint.
With at least a year of follow-up, the overall response rate across all patients was 70%, with all responding patients experiencing a partial response. The rates of response differed slightly between HR+ cancers and TNBC, with ORRs of 74% and 59%, respectively. These high overall response rates led to a median progression-free survival of 14 months in the entire cohort and the median overall survival was not reached. In a subgroup analysis, there was a trend toward higher PFS rates in HR+HER2- patients over TNBC (mPFS 19.1 vs. 8.1 months). PD-L1 expression was evaluated using both the 28-8 and SP142 IHC assays; however, no correlation of PD-L1 positivity with outcomes was observed in this study. All treated patients experienced some level of adverse event, with 75% of patients with an immune-related reaction; however, most of these were grade 1 or 2. Given the positive outcomes found through this study, a phase III trial is being developed for this promising combination.
ATEZOLIZUMAB COMBINATIONS SHOW PROMISE IN HER2-POSITIVE BREAST CANCER
PD1-05. Atezolizumab in combination with trastuzumab emtansine or with trastuzumab and pertuzumab in patients with HER2-positive breast cancer and atezolizumab with doxorubicin and cyclophosphamide in HER2-negative breast cancer: Safety and biomarker outcomes from a multi-cohort Phase Ib study
Combination treatments incorporating atezolizumab for treatment of early and advanced breast cancer were discussed by Erika P. Hamilton, MD (Sarah Cannon Research Institute and Tennessee Oncology, Nashville, TN). Patients with HER2+ metastatic or unresectable, locally advanced breast cancer were allotted to three different regimens: atezolizumab + ado-trastuzumab emtansine (T-DM1, cohort 1B: n=6; expansion cohort 2C: n=14); atezolizumab + trastuzumab + pertuzumab (cohort 1A: n=6; expansion cohort 2D: n=1); or atezolizumab + trastuzumab + pertuzumab + docetaxel (cohort 1F: n=6). HER2+ operable or inflammatory early breast cancer patients could receive two neoadjuvant regimens through a window-of-opportunity study: atezolizumab + trastuzumab + pertuzumab (cohort 2A: n=20) or atezolizumab + T-DM1 (cohort 2B: n=20). Finally, HER2- patients were eligible for treatment with atezolizumab + doxorubicin + cyclophosphamide + pegfilgrastim or G-CSF (cohort 1E: n=3), although this cohort closed early due to poor enrollment. Across all cohorts, safety and tolerability were the main endpoints, while exploratory goals included ORR, DOR, and PFS in metastatic breast cancer, and pCR in operable cancers, and biomarker exploration.
The median duration of safety follow-up ranged from 3.8 months (cohort 1A) to 17.5 months (cohort 2D), at which point the majority of patients experienced at least one adverse event. However, no new safety signals were identified with the combinations. Among all the drugs employed in this study, atezolizumab had the most adverse events attributable to it across all cohorts. Overall response rates were 35, 20, 0, and 100% for patients in cohorts 1B + 2C, 1A + 2D, 1E, and 1F, respectively. Although the patient numbers were small, trends were noted that indicated the adaptive immune response was boosted in patients receiving atezolizumab in combination with ADC or ADCC agents. Specifically, increases in PD-L1 expression and CD8+ T cells were noted with atezolizumab + T-DM1 (cohorts 1B and 2C) and atezolizumab + trastuzumab + pertuzumab (cohorts 1A and 2D) but these were not consistently maintained or associated with a clinical response. Furthermore, cytolytic, immune checkpoint and antigen-processing signatures increased significantly with the addition of atezolizumab to T-DM1 or trastuzumab/pertuzumab. Combining checkpoint inhibitors with standard-of-care treatments may therefore enhance outcomes for breast cancer patients.
DISCORDANCE AMONG PD-L1 ASSAYS SUGGESTS THEY ARE NOT INTERCHANGEABLE
PD1-07. Exploratory analytical harmonization of PD-L1 immunohistochemistry assays in advanced triple-negative breast cancer: A retrospective substudy of IMpassion130
Hope Rugo, MD, FASCO (UCSF Helen Diller Family Comprehensive Cancer Center, San Francisco, CA) discussed a comparison of three commonly-used PD-L1 immunohistochemistry assays: VENTANA SP142, Dako 22C3, and VENTANA SP263. Tissues from the IMpassion130 trial (n = 614 patients) comparing first-line atezolizumab + nab-paclitaxel to placebo + nab-paclitaxel were retrospectively analyzed using the three assays, as the SP142 assay was employed in the actual study and was able to select patients who may benefit from this regimen. The aim of the study was to define optimal cutoffs for the 22C3 and SP263 assays that maximize the overall percentage agreement (OPA) with the SP142 IC 1% cutoff used in the IMpassion130 study.
Across all biomarker-evaluable patients, the intraclass correlation indices between SP142 IC and 22C3 CPS or SP263 IC were 0.57 and 0.69, respectively. The highest OPA values (75%) with IC >= 1% in the SP142 assay could be obtained with non-standard cutoffs for the two additional assays: CPS 10 for 22C3 and IC 4% for SP263, compared to the standard values of CPS 1 and IC 1%. These alternative cutoffs also led to increases in the negative percentage agreement values, and decreases in the positive percentage agreement values as well, which indicates that the three assays may be measuring different cell populations. Using the model-derived cutoffs and the standard SP142 IC value, 36% of patients were SP142+/22C3+ and 34% were SP142+/SP263+. However, some discordance was also noted, as 10% of patients were SP142+/22C3-, 17% SP142-/22C3+, and 12% each were SP142+/SP263- or SP142-/SP263+. These cutoffs were, to widely varying degrees, only moderately able to distinguish responders from non-responders. This study therefore indicates that these three assays may measure different biology, and their interchangeability and differences should be further investigated.