PD-L1 Expression May Be Associated with Pathological Complete Response in TNBC

September 19, 2020

“PD-L1 expression was significantly associated with pCR, which increased with higher PD-L1 expression on immune cells,” said Giampaolo Bianchini, an author of the NeoTRIPaPDL1 study.

Atezolizumab (Tecentriq) increased the pathological complete response (pCR) by 10% or more in “immune-rich” groups with high-risk and locally advanced triple-negative breast cancer (TNBC), according to data from the phase 3 NeoTRIPaPDL1 trial presented at the 2020 ESMO Virtual Meeting.1

Moreover, the agent also turned PD-L1 negative tumors positive in most immunotherapy-treated patients. However, atezolizumab did not improve pCR when added to carboplatin and nab-paclitaxel (Abraxane).

“PD-L1 expression was significantly associated with pCR, which increased with higher PD-L1 expression on immune cells,” said Giampaolo Bianchini, an author of the NeoTRIPaPDL1 study (NCT02620280) who presented the findings.

The pCR with atezolizumab was highest in the study’s “immune-rich” groups, defined as patients with PD-L1–positive immune cells (PDL1 IC+), or stromal tumor-infiltrating lymphocyte (sTILs) or intermediate/high intratumoral TILs (iTILs).1 Data showed that the pCR with the atezolizumab-containing regimen was 86.9% versus 72.0% with chemotherapy alone in patients with IC2/3 (PD-L1 expression of ≥ 5%; Δ + 14.96%) and 56.2% versus 44.0% in patients with IC1 (PD-L1 ≥ 1% and < 5%; Δ + 12.25%). A pCR benefit was not seen in patients with PD-L1 expression of less than 1% (IC0), where chemotherapy alone led to a higher pCR than the immune checkpoint inhibitor (ICI)-based combination (41.07% vs 35.09%) with a Delta that favored chemotherapy by 5.9%. The P value for this test for interaction, which measured log PD-L1 IC to correct skewness, was .02.

The overall pCR rates between treatment arms were 52.3% with atezolizumab and 47.7% with chemotherapy (P = .46; Δ 4.62%).1

The NeoTRIPaPDL1 trial tested the addition of atezolizumab to nab-paclitaxel in 280 women with high-risk (T1cN1; T2N1; and T3N0) and locally advanced TNBC versus carboplatin and nab-paclitaxel alone. After completing 8 cycles of either regimen, all patients in this intention-to-treat population underwent definite surgery. Investigators collected samples from all 260 patients evaluable for pCR response in the per-protocol population at baseline to assess sTILs, iTILs, and PD-L1 expression on ICs and tumor cells (TCs), and their dynamics and association with pCR. The objective of the analysis was to validate the presence of 40% or more sTILs following 1 cycle of treatment on day 1 of cycle 2 (d1c2), as predictive of pCR. Notably, d1c2 served as an early surrogate of pCR.

PD-L1 Characteristics and Shifts

PD-L1 expression on ICs and TCs was measured using Roche’s VENTANA PD-L1 (SP142) Assay. Baseline patient samples amassed from the 260 patients evaluable for pCR indicated that PD-L1 IC was balanced across all 3 expression groups. The percentages of patients with IC0, IC1, and IC2/3 in the atezolizumab and chemotherapy-only arms, respectively, were 44.53% versus 42.75%; 37.50% versus 38.17%, and 17.97% versus 19.08%.

However, both sTILs (Wilcoxon, P = .046) and iTILs (Wilcoxon, P = .005) were disproportionately higher in the arm that received carboplatin and nab-paclitaxel. This “significant imbalance” in sTILs and iTILs could be responsible for the smaller differences in pCR seen between arms, said Bianchini, head of the Breast Cancer Group in the department of Medical Oncology and head of the translational and immunotherapy research group at the IRCCS Ospedale San Raffaele in Milan, Italy.

Samples collected on d1c2 additionally demonstrated that in both treatment arms, sTILs observed on d1c2 were more informative than baseline sTILs and ΔTILs and were predictive of pCR. “The majority of patients demonstrated a robust increase of TILs after 1 cycle, which was similar among arms, suggesting that [this increase] is mostly driven by chemotherapy,” Bianchini said.

At d1c2, 33.7% of patients met or exceeded 40% sTILs in the general population. In the atezolizumab group, patients who had a 40% or higher percentage of sTILs (n = 21) had a pCR of 71.4% compared with 28.0% who did not meet this threshold (n = 56). The odds ratio (OR) was 6.83 (95% CI, 2.24-20.87; P = .0007).

Among patients receiving chemotherapy alone, a 40% or greater percentage of sTILs translated to a pCR of 63.16% in the 38 patients who met or surpassed this sTIL threshold and a pCR of 33.9% in the 59 patients who did not (OR, 3.34; 95% CI, 1.43-7.83; P = .0055).

Rates of tumor absence measured on d1c2 may also be predicative of pCR. “Following the biopsy on day 1, cycle 2, tumor cells were not found in almost 1 out of 3 samples in the atezolizumab arm, which was double the observed rate in the chemotherapy arm,” said Bianchini. Specifically, no tumor cells were observed in 32.6% of patients treated with atezolizumab versus 15.65% of patients treated with chemotherapy.

At d1c2 evaluation, 37 patients in the atezolizumab group had no tumor cells and 77 patients. had observed tumor cells. The resulting in a pCR rates were 78.3% and 38.9%, respectively (P = .0002). Eighteen patients in the chemotherapy arm showed no signs of a tumor whereas 97 did. The pCR rate was consequently higher in patients with absent tumor (77.7% vs 45.3%; P = .023).

PD-L1 dynamics from baseline to d1c2 were “strong and divergent by arm of therapy.” A key finding of this analysis was that atezolizumab turned most PD-L1–negative tumors positive. “Notably, 2 out of 3 PD-L1–negative tumors converted to [PD-L1–]positive tumors in the atezolizumab arm,” Bianchi said. In total, PD-L1 IC+ rose from 45.4% (IC1, 32.0% and IC2/3, 13.3%, respectively) to 74.6% (IC1, 9.3% and IHC2/3, 65.3%, respectively). At baseline, 54.7% of patients in the atezolizumab arm were IC0, compared with 25.3% at d1c2. “PD-L1 expression in immune cells strongly increased, suggesting that this is a factor activation driven by atezolizumab,” he added.

An overall decrease in PD-L1 positivity was observed in the chemotherapy-only arm, where PD-L1 IC+ fell from 52.1% (IC1, 38.4% and IC2/3, 13.7%, respectively) at baseline to 37.9% (IC1, 23.2% and IC2/3, 14.7%, respectively). At enrollment, 47.4% of patients were PD-L1–negative. This number increased to 62.1% at d1c2.

Clinical Implications

In early-stage TNBC, administering an ICI in combination with standard neoadjuvant chemotherapy is hypothesized to improve efficacy by decreasing tumor associated macrophages and increasing TILs. TNBC is an aggressive form of breast cancer associated with poor outcomes, due largely to the paucity

of effective targeted therapies. Immunotherapy-based combinations have been recognized for their ability to significantly increase pCR in TNBC, where checkpoint inhibitor-driven spikes in TILs are associated with elevated pCR.2

Although data from Bianchini et al’s analysis of TILs, PD-L1 expression, and the dynamics of their intersection in the NeoTRIPaPDL1 per-protocol population provide insight on the interaction of immunotherapy and chemotherapy in this high-risk setting, monoclonal antibody use in this space remains opaque. This is in part due to discordance in results from various neoadjuvant studies in TNBC, with conflicting results seen in NeoTRIPaPDL1, the phase 3 KEYNOTE-522 (NCT03036488) and IMpassion031 trials (NCT03197935), and the phase 2 GeparNUEVO study (NCT02685059).

To date, KEYNOTE-522 is the largest trial to evaluate immunotherapy-containing neoadjuvant chemotherapy in TNBC. Findings from the evaluation demonstrated a “significantly higher” pCR rate among patients receiving pembrolizumab (Keytruda) with pre-operative chemotherapy versus chemotherapy alone.3 This trend was also apparent in the IMpassion031 study, where atezolizumab and chemotherapy outperformed chemotherapy alone in patients with early TNBC. A statistically significant, clinically meaningful improvement in pCR was seen with the multi-class regimen, causing IMpassion031 to meet its primary end point.4

Conversely, in the GeparNUEVO study, patients with metastatic TNBC who were treated with neoadjuvant durvalumab (Imfinzi) and nab-paclitaxel achieved a higher, but not statistically significant pCR compared with patients who only received nab-paclitaxel.5 Hope S. Rugo, MD, FASCO, called the findings from these 4 investigations “perplexing” and posited that this observed discordance signifies that the field of TNBC must continue its efforts to “understand which patients benefit from immunotherapy” in her invited discussant session.6


Bianchini G. Tumour infiltrating lymphocytes (TILs), PD-L1 expression and their dynamics in the NeoTRIPaPDL1 trial. Presented at: 2020 ESMO Virtual Congress; September 19-21, 2020. LBA13.

Lee JS, Yost SE, Yuan Y. Neoadjuvant treatment for triple negative breast cancer: recent progresses and challenges. Cancers (Basel). 2020;12(6):1404. doi:10.3390/cancers12061404

Schmid P, Cortes J, Pusztai L, et al; KEYNOTE-522 Investigators. Pembrolizumab for early triple-negative breast cancer. N Engl J Med. 2020;382(9):810-821. doi:10.1056/NEJMoa1910549

Roche’s Tecentriq in combination with chemotherapy (including Abraxane) meets primary endpoint of improved pathological complete response, regardless of PD-L1 status, as initial treatment for people with early triple-negative breast cancer. News release. Roche. June 18, 2020. Accessed September 19, 2020. https://www.roche.com/media/releases/med-cor-2020-06-18.htm

Loibl S, Untch M, Burchardi N, et al. A randomised phase II study investigating durvalumab in addition to an anthracycline taxane-based neoadjuvant therapy in early triple-negative breast cancer: clinical results and biomarker analysis of GeparNuevo study. Ann Oncol. 2019;30(8):1279-1288. doi:10.1093/annonc/mdz158

Rugo HS. Predicting benefit from neoadjuvant therapy. Discussion presented at: 2020 European Society for Clinical Oncology Virtual Congress.; September 18-21, 2020; virtual. Presentation ID 4898.