PD-L1 assessment in breast cancer immunotherapy: a critical overview

The advent of immune checkpoint inhibitors (ICIs) has recently made a breakthrough in several hematological and solid tumors including, among others, non-small-cell lung cancer, renal cell carcinoma, melanoma, urothelial carcinoma, and hepatocellular carcinoma [1–3]. These agents are able to enhance antitumor activity, leading to an increase in the cytotoxicity of T cells and the blocking of downregulators of immunity such as programmed cell death protein 1 (PD-1) and its ligand PD-L1, cytotoxic T-lymphocyte antigen 4 (CTLA-4), and lymphocyte activating-3 (LAG-3) [4].

ICIs have also been recently assessed in breast cancer (BC), as monotherapy or in combination with other anticancer agents. First, monotherapy with ICIs has reported disappointing results in unselected triple-negative BC (TNBC), with approximately one-quarter of patients achieving response [5]; in fact, the KEYNOTE-086 and the KEYNOTE-119 trials evaluating pembrolizumab monotherapy highlighted response rates lower than 10%, and these findings have also been confirmed by clinical studies evaluating other immunotherapies, such as single-agent atezolizumab [6]. Thus, several combination treatments have been investigated, based on the synergistic effect of ICIs plus other anticancer agents with different mechanism of action.

Among these combinatorial strategies, and following the results of landmark trials, chemoimmunotherapy has entered into clinical practice as new front-line treatment in TNBC patients with metastatic disease and PD-L1 overexpression or elevated combined positive score (CPS) [7]. Moreover, a large number of phase I to III clinical trials are assessing immune-based combinations, with these studies having the potential to further shape the direction of first- and later-line therapy in this patient population. However, a high unmet need in BC immunotherapy remains the lack of biomarkers predictive of response to ICIs. In fact, if PD-L1 is considered the most reliable predictor, its assessment presents several limitations, and it is far from being standardized [8].

The expression of PD-L1 is typically detected on tumor cells (TC) or immune cells (IC), and in recent years, PD-L1 assessment has emerged as an important predictive biomarker of response to immunotherapy in several tumor types (e.g. non-small-cell lung cancer, head and neck cancer, and gastric cancer) [9,10]. As regards BC, PD-L1 status has been associated with a prognostic value, and high PD-L1 expression seems to predict worse clinical outcomes in triple-negative BC patients [11].

PD-L1 has been validated as a predictor of response to chemoimmunotherapy in metastatic BC and has entered into everyday clinical practice, following the results of recently published IMpassion130 and KEYNOTE-355 phase III clinical trials [12,13]. The IMpassion130 compared chemoimmunotherapy with atezolizumab–nab-paclitaxel versus placebo plus nab-paclitaxel as front-line treatment in TNBC patients with metastatic disease [12]; the coprimary endpoints were progression-free survival (PFS) and overall survival (OS) in the intention-to-treat (ITT) and in PD-L1-positive patients, with PD-L1 expression determined on IC and centrally evaluated per VENTANA SP142 immunohistochemistry assay (positive in the case of IC ≥1% and negative with IC <1%) [12]. The IMpassion130 highlighted a statistically superior and clinically meaningful benefit in terms of OS in patients with PD-L1 expression ≥1% receiving chemoimmunotherapy, leading to the approval of atezolizumab–nab-paclitaxel in this setting [12]. Similarly, the KEYNOTE-355 highlighted longer PFS and OS in metastatic TNBC patients with PD-L1 CPS ≥10 treated with pembrolizumab–chemotherapy versus placebo–chemotherapy [13]. In another study in the neoadjuvant setting, the KEYNOTE-522, the authors observed that in patients with early triple-negative breast cancer, neoadjuvant pembrolizumab plus chemotherapy, followed by adjuvant pembrolizumab after surgery, resulted in significantly longer event-free survival than neoadjuvant chemotherapy alone, regardless of PD-L1 status [14].

Despite these trials have recently had an important impact in BC clinical practice worldwide, some fundamental issues regarding PD-L1 assessment should be highlighted. First, recent studies have suggested differences in terms of PD-L1 status according to primary tumors and secondary lesions in BC patients [15]. For example, a retrospective trial conducted by Rozenblit et al. compared PD-L1 expression between primary tumors (n = 179) and metastatic sites (n = 161) [15]. Interestingly, the authors reported higher PD-L1 expression in primary cancers compared with secondary lesions (63.7% versus 42.2%, respectively) by using the SP142 antibody [15]; in addition, different positivity rates were observed across metastatic sites, with higher PD-L1 expression in cutaneous and hepatic metastases (23.8% and 17.4%, respectively) [15].

Second, and similarly to what observed in several tumor types, PD-L1 assessment presents some specific methodological issues, including the use of distinct antibodies, scoring systems, and platforms across different studies. Sensitivity and specificity of PD-L1 antibodies are non-superimposable, as also reported in the previously cited randomized controlled trials. Based on these premises, Rugo et al. recently conducted an exploratory post-hoc analysis with the aim of evaluating Dako PD-L1 immunohistochemistry (IHC) 22C3 assay (CPS ≥1) and VENTANA SP263 or SP142 assay (IC ≥1%) on 614 TNBC samples from IMpassion130 phase III trial [16]. PD-L1-positive expression widely differed according to assays and antibodies since it was higher for 22C3+ (80.9%) and SP263+ (74.9%) and lower for SP142+ (46.4%) [16]. Another interesting finding suggesting the non-interchangeable nature of antibodies has been highlighted in the KEYNOTE-119 trial [17]. In this study, assessing the PD-1 inhibitor pembrolizumab versus investigator-choice chemotherapy in previously treated metastatic TNBC, one out of 10 patients achieving response would have been classified as PD-L1 negative with other scoring systems, such as IC and TPS [17].

In summary, a wide range of challenges and issues are to be considered regarding the assessment of PD-L1 in BC, given the lack of interchangeability between different antibodies, assays, and scoring systems. Moreover, recent studies have also suggested a poor reproducibility among pathologists for IC scoring, and a standardization of these methodologies remains a high unmet need in BC immunotherapy [17,18]. The identification of reliable predictive biomarkers is a priority in this setting, and the BC medical community is called to focus its efforts toward this direction in the near future.

Declaration of interest

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Additional information

Funding

This paper was not funded.