Stay Up to Date
Breaking News,
Updates, & More
Click Here to

Rationale for PD-L1 Expression as a Biomarker in Immuno-Oncology

Faculty Perspectives–Rationale for PD-L1 Expression as a Biomarker in Immuno-Oncology | Part 2 of a 4-Part Series

Therapy with immune checkpoint inhibitors (ICIs) heralds a new era in the management of many cancers. In particular, immunotherapy directed toward cytotoxic T-lymphocyte–associated antigen 4 and the programmed-cell death-1 (PD-1) receptor and its ligand, PD-L1, has demonstrated remarkable and durable responses in several types of cancer, including melanoma, non–small-cell lung cancer (NSCLC), renal-cell carcinoma (RCC), urothelial carcinoma (UC), and head and neck squamous-cell carcinoma (HNSCC). However, to optimize treatment with ICIs, there is a need to discover validated biomarkers that can guide treatment decisions and help identify patients who are more likely to respond to ICIs and less likely to experience immune-related adverse events (irAEs) or develop resistance to these agents.

PD-L1 expression is a rational biomarker to predict response to PD-1/PD-L1 ICI therapy, and has been studied extensively in clinical trials. A recurring theme emerging from available clinical data is that high levels of tumor cell membrane PD-L1 expression correlate with better outcomes with PD-1/PD-L1 blockade. However, it is also becoming apparent that patients with low or undetectable tumor PD-L1 expression may still gain benefit from checkpoint inhibition. This review summarizes the clinical evidence and relevance of PD-L1 tumor expression as a predictive biomarker for PD-L1–directed therapies across tumors, as well as the PD-L1 testing landscape, including the comparability of available PD-L1 immunohistochemistry (IHC) assays.

The PD-1/PD-L1 Immunomodulation Pathway

Under normal physiologic conditions, the negative immune checkpoint protein PD-1 is expressed primarily on the surface of activated T-cells.1 During an inflammatory response, PD-1 binds to its ligand PD-L1 that is expressed in the tissue, initiating inhibitory signals to limit a cytotoxic T-cell response and prevent autoimmunity (Figure).2 However, tumors exploit the PD-1/PD-L1–mediated negative feedback mechanism to evade immune detection and elimination.1,2


In the tumor microenvironment, overexpression of PD-L1 on the tumor cells leads to engagement of the PD-1/PD-L1 receptor–ligand system, which results in the suppression of both effector T-cell function and B-cell–mediated T-cell activation, and an increase in the conversion of the immunosuppressive regulatory T-cell population, causing a dysfunctional antitumor immune response. The pharmacologic use of PD-1/PD-L1 inhibitors prevents the PD-1/PD-L1 interaction, thus restoring T-cell–mediated antitumor immune response. However, it must be pointed out that PD-L1 is not only expressed on tumor cells but can also be expressed on tumor-infiltrating nonmalignant immune cells such as lymphocytes, mononuclear cells, and other immune cells, to suppress T-cell–mediated tumor cytotoxicity.1,3 Therefore, its predictive role is currently unclear and is being investigated in clinical trials.3

Clinical Evidence from PD-L1 Biomarker-Driven Studies Across Tumors

More than 1000 clinical trials have been performed to evaluate the clinical efficacy and safety of anti–PD-1/PD-L1 ICIs, including 5 that are currently approved by the US Food and Drug Administration (FDA) for the treatment of different types of cancer. These include the anti–PD-1 antibodies nivolumab (Opdivo®; Bristol-Myers Squibb) and pembrolizumab (Keytruda®; Merck & Co, Inc), and the anti–PD-L1 antibodies atezolizumab (Tecentriq®; Genentech/Roche), durvalumab (Imfinzi®; AstraZeneca), and avelumab (Bavencio®; EMD Serono/Pfizer) (Table 1).1,4-8 Evidence from clinical trials evaluating these approved ICIs across tumors has shown a positive correlation between tumor PD-L1 expression and the antitumor effect of ICIs.9 Early indications of this association were reported in 2012 by Topalian and colleagues. Results from their phase 1 clinical trial showed that among 42 evaluable patients with different types of solid tumors who received treatment with nivolumab, 36% of PD-L1–expressing tumors achieved a response, although none of the PD-L1–negative tumors responded to treatment.10


Subsequently, the association between PD-L1 expression and response to therapy has been described in numerous clinical trials, irrespective of the solid tumor type studied, the specific agent tested (anti–PD-1 or anti–PD-L1), the PD-L1 positivity threshold used, or tumor cell type scored for PD-L1 expression. In a pooled analysis of 1400 patients tested for PD-L1 expression in published studies of solid tumors, the weighted average overall response rate (ORR) was 48% in patients who were PD-L1–positive.9

Here, we present relevant data on the predictive role of PD-L1 expression for the approved anti–PD-1/PD-L1 agents in various types of cancer, with differential effects observed across tumor types and with different PD-1/PD-L1 ICIs.

Non–small-cell lung cancer

The largest predictive effect of PD-L1 expression for treatment response is documented in NSCLC. The international phase 1 KEYNOTE-001 trial validated the predictive value of tumor PD-L1 expression in 495 patients with advanced NSCLC treated with pembrolizumab. In this trial, patients treated with pembrolizumab (2 mg/kg or 10 mg/kg) were assigned to either a training group (N = 182) or a validation group (N = 313). Within the validation group, patients with PD-L1 expression in at least 50% of tumor cells—which was defined as the PD-L1 cutoff based on results from the training group—achieved an ORR of 45.2%, which translated into a median progression-free survival (PFS) of 6.3 months; median overall survival (OS) was not reached for all patients. Moreover, there was a direct correlation between higher PD-L1 expression levels and increased response rates, reported as a proportion score (PS; defined as the percentage of neoplastic cells with staining for membranous PD-L1). In the validation set, patients with a PS of 1% to 49% achieved an ORR of 16.5%, whereas those with a PS <1% achieved an ORR of only 10.7%.11

The confirmatory phase 3 KEYNOTE-010 study of pembrolizumab (2 mg/kg or 10 mg/kg) versus docetaxel enrolled patients with previously treated NSCLC and ≥1% PD-L1–positive staining (N = 1034). Survival benefits were more pronounced among patients with tumor PD-L1 staining ≥50% who were treated with pembrolizumab, with median survival of 14.9 months (hazard ratio [HR], 0.54) and 17.3 months (HR, 0.50) for the 2-mg/kg and 10-mg/kg cohorts, respectively. In addition, ORRs were higher in the cohort with PD-L1 staining ≥50% (30% and 29%, respectively) compared with the total population (18% for both doses). Subsequently, pembrolizumab was approved for the first-line treatment of advanced NSCLC in patients with tumor PD-L1 expression of at least 50%, as assessed by the companion diagnostic PD-L1 IHC 22C3 pharmDx assay.12,13

At the 2018 American Society of Clinical Oncology annual meeting, investigators presented results from the phase 3 open-label KEYNOTE-042 study that compared pembrolizumab with chemotherapy (carboplatin plus paclitaxel or carboplatin plus pemetrexed) in 1274 patients with locally advanced or metastatic NSCLC and much lower PD-L1 expression (tumor proportion score [TPS] ≥1%).14 In the highest TPS group, the response rate with pembrolizumab was 39.5% versus 32.0% with chemotherapy. In the TPS ≥20% group, the response rates were 33.4% and 28.9%, respectively, and in the TPS ≥1% group, they were 27.3% and 26.5%, respectively. The duration of response was longer with pembrolizumab; in the full TPS ≥1% cohort, the median duration of response was 20.2 months compared with 8.3 months with chemotherapy. Pembrolizumab is now an option for patients who have advanced NSCLC and no EGFR mutations or ALK translocations, and who express PD-L1 at least at the 1% level.

However, the results from KEYNOTE-042 demonstrated a greater OS benefit from pembrolizumab monotherapy compared with platinum-based chemotherapy in patients with PD-L1 expression ≥50% versus those with PD-L1 expression 1% to 49%.14 The median OS in patients treated with pembrolizumab monotherapy versus chemotherapy was 20.0 months and 12.2 months, respectively, in those with PD-L1 ≥50% (P = .0003); 17.7 months versus 13.0 months, respectively, in those with PD-L1 ≥20% (P = .0020); and 16.7 months versus 12.1 months, respectively, in those with PD-L1 ≥1% (P = .0018).14 These results provide further evidence for the utility of higher levels of PD-L1 in selecting patients with NSCLC for ICI monotherapy.

In the case of nivolumab treatment, evidence indicates that a positive relationship between PD-L1 expression and outcome is dependent on histology subtype. In the multicohort, open-label phase 1 CheckMate 012 trial, in which nivolumab was used as first-line treatment for advanced NSCLC, increasing PD-L1 expression level was associated with greater benefit in the expanded nivolumab monotherapy cohort, with clinical activity also observed in patients with a low PD-L1 expression level or with no PD-L1 expression. In the expanded cohort, an ORR of 50% was achieved in the population with ≥50% PD-L1 expression and in 14% of those with <1% PD-L1 expression.15

In the phase 2 CheckMate 063 trial of patients with advanced refractory squamous NSCLC, nivolumab therapy also benefited both the PD-L1–positive and PD-L1–negative groups, although a greater percentage of objective responses was observed in patients with PD-L1–positive tumors (<5% vs ≥5%: 38% vs 52%, respectively).16 The 2 parallel phase 3 CheckMate 017 and CheckMate 057 studies comparing nivolumab with docetaxel in patients with previously treated squamous and nonsquamous NSCLC, respectively, indicated that overall response was similar in PD-L1–positive (≥1%) and PD-L1–negative (<1%) tumors in patients with squamous NSCLC who were treated with nivolumab (17% vs 17%), but a strong predictive association was evident in nonsquamous NSCLC across the tumor PD-L1 expression levels assessed (1%, 5%, 10%), suggesting a dependence on tumor histology.17,18 In CheckMate 057, ORR related to PD-L1 expression was 31% in the PD-L1 level ≥1% cohort versus 9% in the PD-L1 level <1% cohort; 36% in the PD-L1 level ≥5% cohort versus 10% in the PD-L1 level <5% cohort; and 37% in the PD-L1 level ≥10% cohort versus 11% in the PD-L1 level <10% cohort.17 These results indicate that both patients with squamous NSCLC and patients with nonsquamous NSCLC may derive survival benefit with second-line nivolumab therapy independent of PD-L1 expression level, and that PD-L1 testing results do not inform treatment decisions in these settings. The FDA does not mandate PD-L1 testing for the use of nivolumab in patients with NSCLC.

In the first-line setting, divergent outcomes were observed with pembrolizumab and nivolumab treatment in the KEYNOTE-024 and CheckMate 026 trials, respectively, in the PD-L1–positive cohorts; the reasons for the disparity may be multifactorial and currently remain unclear. The prospective KEYNOTE-024 trial established a role for pembrolizumab versus chemotherapy as first-line treatment in patients with NSCLC who had a PD-L1 expression level ≥50%, based on prolongation of PFS (median PFS: 10.3 vs 6.0 months; HR, 0.50; P <.001), improvement in OS (6-month OS rate: 80.2% vs 72.4%; HR, 0.60; P = .005), and higher response rate (44.8% vs 27.8%).19 The phase 3 CheckMate 026 trial failed to demonstrate survival benefit with nivolumab therapy versus chemotherapy among patients with previously untreated stage IV or recurrent NSCLC with a PD-L1 expression level ≥5%; ORR was 26% versus 33%, respectively.20 In an exploratory subset analysis of patients with a PD-L1 expression level of ≥50%, ORR was similar in the nivolumab and chemotherapy arms (34% and 39%, respectively). Limited conclusions may be drawn from these data because of several confounding factors, including a high frequency of subsequent nivolumab treatment in the chemotherapy group; PD-L1 expression ≥50% not being prospectively assessed; a lower proportion of patients with PD-L1 expression ≥50% and high tumor mutational burden; and patient selection criteria related to previous radiotherapy and glucocorticoid use.20

In terms of the role of PD-L1 expression in predicting benefit from therapy with atezolizumab, increased clinical benefit was seen with increasing PD-L1 expression on tumor cells and tumor-infiltrating immune cells in the POPLAR and OAK trials.21,22 In both trials, patients were stratified on the basis of PD-L1 expression on their tumor cells and tumor-infiltrating immune cells. Tumor cell (TC)3 represented patients with a percentage of PD-L1–expressing tumor cells ≥50%, TC2 represented patients with PD-L1–expressing tumor cells ≥5% and <50%, TC1 included PD-L1–expressing tumor cells ≥1% and <5%, and TC0 included PD-L1–expressing tumor cells <1%. Tumor-infiltrating immune cell (IC)3 had PD-L1 expression ≥10%, IC2 had ≥5% and <10%, IC1 had ≥1% and <5%, and IC0 had <1%.

In the randomized phase 2 POPLAR trial, increased ORR benefit was seen with increasing PD-L1 expression, with an ORR of 38% achieved in the TC3 or IC3 expression subgroup, 7.7% in the TC2 or IC2 expression subgroup, 14% in the TC1 or IC1 expression subgroup, and 7.8% in the TC0 or IC0 expression subgroup. This translated to an OS benefit for the higher PD-L1 expression subgroups (TC3 or IC3: 15.5 months vs 11.1 months, HR, 0.49, P = .068; TC2 or IC2: 9.0 months vs 6.2 months, HR, 0.59; TC1 or IC1: 15.6 months vs 12.4 months, HR, 0.65; TC0 and IC0: 9.7 months vs 9.7 months, HR, 1.04, P = .871).21

The randomized, open-label, phase 3 OAK trial demonstrated clinically relevant improvement in OS with atezoliz­umab versus docetaxel in previously treated NSCLC regardless of PD-L1 expression, but with increasing magnitude of responses with increasing PD-L1 expression.22 ORR was 30.6% in the TC3 or IC3 population, 22.5% in the TC2/3 or IC2/3 population, 17.8% in the TC1/2/3 or IC1/2/3 population, and 7.8% in the TC0 or IC0 population.

In patients with locally advanced, stage III unresectable NSCLC, the phase 3 PACIFIC trial showed that durvalumab treatment led to a PFS benefit irrespective of PD-L1 expression before chemoradiotherapy (PD-L1 <25%: HR, 0.59; PD-L1 ≥25%: HR, 0.41).23

A predictive association between PD-L1 expression and benefit from ICI treatment targeting the PD-1/PD-L1 axis also extended to combination therapies, including combination ICIs and ICI plus chemotherapy or antivascular endothelial growth factor therapies; such an association has been reported with the use of nivolumab, pembrolizumab, and atezolizumab. In the multicohort CheckMate 012 trial, a higher percentage of overall responses were achieved in patients with higher PD-L1 expression in the nivolumab plus ipilimumab combination therapy cohorts: ORR was 92% in the PD-L1 ≥50% population and 18% in the PD-L1 <1% population.24 In the KEYNOTE-189 trial, survival benefit in the pembrolizumab/pemetrexed/platinum arm was observed in all PD-L1 PS categories; the subgroup with the highest PS derived the most benefit in terms of OS (ie, ≥50%: median OS, not reached [NR] vs 10.0 months in the placebo arm; HR, 0.42; 1% to 49%: median OS, NR vs 12.9 months in the placebo arm; HR, 0.55; <1%: median OS, 15.2 months vs 12.0 months in the placebo arm; HR, 0.59) and PFS (≥50%: median PFS, 9.4 months vs 4.7 months in the placebo arm; HR, 0.36; 1% to 49%: median PFS, 9.0 months vs 4.9 months in the placebo arm; HR, 0.55; <1%: median PFS, 6.1 months vs 5.1 months in the placebo arm; HR, 0.75).25

In the ongoing randomized, double-blind, phase 3 KEYNOTE-407 trial, researchers are evaluating pembrolizu­mab plus carboplatin-paclitaxel and nab-paclitaxel chemotherapy or chemotherapy alone as first-line therapy in patients with metastatic squamous NSCLC (NCT02775435).26 The first interim results of this trial showed that in the initial 204 patients (median follow-up of 7 months) the combination of pembrolizumab plus chemotherapy significantly improved OS and PFS compared with chemotherapy alone, reducing the risk for death by 36% (HR, 0.64; 95% confidence interval [CI], 0.49-0.85; P = .0008) and disease progression by 44% (HR, 0.56; 95% CI, 0.45-0.70; P <.0001). Prespecified exploratory analyses based on PD-L1 expression demonstrated an OS and PFS benefit with pembrolizumab combination therapy regardless of PD-L1 expression.

In the pembrolizumab combination group, patients whose tumors did not express PD-L1 showed a 39% reduced risk for death (HR, 0.61; 95% CI, 0.38-0.98), those with TPS of 1% to 49% showed a 43% reduced risk for death (HR, 0.57; 95% CI, 0.36-0.90), and those with TPS ≥50% (HR, 0.64; 95% CI, 0.37-1.10) showed a 36% reduced risk for death; reductions in risk for progression were 32% (HR, 0.68; 95% CI, 0.47-0.98), 44% (HR, 0.56; 95% CI, 0.39-0.80), and 63% (HR, 0.37; 95% CI, 0.24-0.58), respectively.

In patients with stage IV nonsquamous NSCLC, the 3-arm, open-label, randomized, phase 3 IMpower150 study demonstrated that first-line atezolizumab plus bevacizumab/carboplatin/paclitaxel significantly improved PFS compared with bevacizumab plus carboplatin and paclitaxel (median, 8.3 months vs 6.8 months; stratified HR, 0.61; 95% CI, 0.52-0.72); the PFS benefit with atezolizumab combination therapy was observed across varying levels of PD-L1 expression, including those with no or low PD-L1 expression (TC0/1/2 or IC0/1/2, 8.0 months vs 6.8 months; unstratified HR, 0.68) and those with high PD-L1 expression (TC3 or IC3, 12.6 months vs 6.8 months; unstratified HR, 0.39).27

Similarly, in the advanced squamous NSCLC population, the phase 3 IMpower131 study randomized patients to receive atezolizumab plus carboplatin and paclitaxel (Arm A), atezoliz­umab plus carboplatin and nab-paclitaxel (Arm B), or carboplatin plus nab-paclitaxel (Arm C). At a minimum follow-up of 9.8 months, the primary analysis of PFS for Arm B (N = 343) versus Arm C (N = 340) showed that PFS benefit was enriched in all PD-L1–posiitve IHC subgroups, but was most pronounced in TC3 or IC3. The 12-month PFS was 24.7% in Arm B compared with 12.0% in Arm C, with a median duration of response of 6.6 months versus 4.4 months, respectively. Ongoing responses were observed in 27% of patients in Arm B and 15% of those in Arm C. The occurrence of adverse events was not significantly different among treatment arms.28

Renal-cell carcinoma

In RCC, current data from the large international, randomized, open-label phase 3 CheckMate 025 study indicate that OS benefit can be achieved with nivolumab therapy irrespective of PD-L1 expression (≥1% vs <1%; 21.8 months vs 27.4 months).29 However, in the phase 2 IMmotion150 trial of patients with untreated metastatic RCC, the immunotherapy-based combination regimen of atezolizumab plus bevaciz­umab demonstrated enhanced antitumor activity in the PD-L1–positive population (defined as ≥1% of tumor and tumor-infiltrating immune cells expressing PD-L1), with an ORR of 46% in the combination arm, 28% in the atezolizu­mab arm, and 27% in the sunitinib arm; median PFS was 14.7 months, 5.5 months, and 7.8 months, respectively.30


Current evidence indicates an imperfect correlation of PD-L1 expression and ICI efficacy in melanoma, with considerable responses achieved by patients with tumors expressing low or undetectable levels of PD-L1. For example, in the randomized phase 3 CheckMate 067 trial, nivolumab combined with ipilimumab resulted in a higher ORR than the nivolumab monotherapy group at each PD-L1 expression level tested.31 However, in the combination arm, an ORR of 85% was achieved in the population demonstrating ≥10% PD-L1 expression versus 55% in those demonstrating <10% expression, whereas in the monotherapy arm, ORRs were 58% and 44%, respectively. Of note, patients determined to be PD-L1–negative (tumor PD-L1 expression <1%) also achieved objective responses in both nivolumab arms, with an ORR of 54% in the combination arm and an ORR of 35% in the monotherapy arm. Moreover, in time-dependent receiver operating characteristic curves generated for OS at 3 years that were used as a measure of discriminatory ability of tumor PD-L1 expression, area under the curve (AUC) values were more aligned with the line of no discrimination (AUC, 0.50) for both the combination group (AUC, 0.56; P = .09) and the nivolumab monotherapy group (AUC, 0.57; P = .04), indicating that the level of tumor PD-L1 expression alone is a poor predictive biomarker of OS in melanoma treated with nivolumab.31 Similar findings were documented in the CheckMate 066 trial, where 33.1% of the patient population responded to nivolumab therapy despite having negative or indeterminate tumor PD-L1 expression levels; ORR was 52.7% in the PD-L1–positive subgroup.32 Thus, the predictive value of PD-L1 is imperfect in this patient population, since overexpression of PD-L1 can be predictive of a response to nivolumab, although patients with no expression of PD-L1 still respond to this agent.

In the phase 3 KEYNOTE-006 trial of patients with advanced melanoma, the PFS benefit with pembrolizumab (10 mg/kg every 2 weeks or every 3 weeks) was observed in both PD-L1–positive (staining in ≥1% of tumor cells) and PD-L1–negative subgroups compared with the ipilimumab group (4 doses of 3 mg/kg every 3 weeks).33

Urothelial carcinoma

Currently, 5 PD-1/PD-L1 inhibitors (atezolizumab, pembrolizumab, nivolumab, durvalumab, and avelumab) are approved for the treatment of UC.34 In the metastatic UC (mUC) population refractory to cisplatin-based chemotherapy, the single-arm phase 2 IMvigor 210 trial (Cohort 2) reported higher ORRs with atezolizumab in PD-L1–positive patients (≥5%) versus those with <1% PD-L1 expression (26% vs 8%).35 However, these results were not corroborated by the phase 3 IMvigor 211 trial, which evaluated atezolizu­mab versus chemotherapy in the PD-L1 IC2/3 population (≥5% of tumor-infiltrating immune cells), where no statistical improvement in mean OS was reported with atezolizumab in the PD-L1–enriched population (HR, 0.87; OS, 11.1 months vs 10.6 months; P = .41) and ORRs were similar in the 2 treatment groups; this may be partly attributed to a small sample size.36

PD-L1 status did not predict response to pembrolizumab therapy in the phase 3 KEYNOTE-045 trial versus standard-of-care chemotherapy in patients with mUC; a combined positive score (CPS) ≥10% was considered PD-L1–positive and represented the percentage of PD-L1–expressing tumor and infiltrating immune cells relative to the total number of tumor cells. ORRs were similar in both the CPS ≥10% and <10% cohorts (22% in both cases).37

In CheckMate 275, nivolumab monotherapy provided clinical benefit regardless of PD-L1 expression in patients with mUC after platinum therapy; ORR was 28.4% in the ≥5% PD-L1 subgroup, 23.8% in the ≥1% PD-L1 subgroup, and 16.1% in the low or undetectable (<1%) PD-L1 subgroup.38 Based on available data, the predictive value of PD-L1 status for clinical benefit with second-line durvalumab or avelumab therapy was more pronounced in post-platinum mUC. In an open-label phase 1/2 trial in patients with locally advanced UC or mUC, durvalumab therapy resulted in an ORR of 27.6% in patients with high PD-L1 expression and 5.1% in patients with low or no PD-L1 expression.39 With avelumab therapy, an ORR of 54% was achieved in the PD-L1–positive group (defined as ≥5% PD-L1 expression) compared with 4% in the PD-L1–negative group.40

Cohort 1 of the IMvigor 210 trial and KEYNOTE-052 explored the predictive value of PD-L1 expression in platinum-ineligible patients with mUC.41,42 In the IMvigor 210 trial, atezolizumab administered as first-line treatment in cisplatin-ineligible patients with locally advanced and metastatic UC provided clinical benefit regardless of PD-L1 expression; ORR was 26% in the cohort with <1% PD-L1 expression and 31% in the ≥5% PD-L1 subgroup.41 In the single-arm phase 2 KEYNOTE-052 trial of first-line treatment with pembrolizumab, the highest ORR was achieved in the PD-L1 cohort with a CPS ≥10% (39%). However, responses were also evident in the other 2 cohorts: the PD-L1 cohort with a CPS of ≥1% to <10% had an ORR of 20% and the PD-L1 cohort with a CPS <1% had an ORR of 11%.42 Based on these findings, it may be premature to assign platinum-ineligible patients with mUC to therapy with atezolizumab or pembroliz­umab solely on the basis of PD-L1 expression.

Head and neck cancers

In the CheckMate 141 study, the clinical benefit with nivolumab therapy was higher in patients with HNSCC and PD-L1 expression ≥1% versus the <1% PD-L1–negative cohort (ORR, 15% vs 9%) but was not further increased in the ≥5% or ≥10% subgroups (ORR, 12% in both cases). Patients with tumor PD-L1 expression ≥1% showed improved OS with nivolumab therapy versus the control group, which received standard single-agent therapy (methotrexate, docetaxel, or cetuximab) of the investigator’s choice (median OS, 8.7 months vs 4.6 months; HR, 0.55); however, patients with tumor PD-L1 expression <1% had similar OS with nivolumab and standard treatment (median OS, 5.7 months vs 5.8 months; HR, 0.89).43

In the expansion cohort of the open-label, single-arm phase 1b KEYNOTE-012 trial, HNSCC patients with tumor PD-L1 expression ≥1% achieved an ORR of 19%, while those with <1% PD-L1 expression achieved an ORR of 16% (P = .348); however, when measuring PD-L1 expression from both tumor cells and immune cells, there was a statistically significant increase in the probability of response among patients with PD-L1–positive tumors compared with patients with PD-L1–negative tumors (22% vs 4%; P = .021).44 Although the small number of events observed in this study preclude broad interpretation of these findings, the preliminary data suggest that PD-L1 expression on tumor-infiltrating immune cells may also be an important determinant of clinical outcome in HNSCC.

The Testing Landscape of PD-L1

As part of the clinical development of approved ICIs, tissue-based diagnostic PD-L1 assays were developed in parallel with the therapeutic agents as potential companion biomarkers. This has led to considerable variability in PD-L1 testing, with unique PD-L1 diagnostic systems developed for each anti–PD-1/PD-L1 agent, including a primary antibody clone, detection reagents, a staining platform, quantitative staining cutoffs for positivity, and a software protocol agent. This poses a challenge to clinicians in everyday practice with respect to clinical application and treatment decision-making due to limited availability of tissue samples, the number of tissue-based diagnostic tests required in the management of a patient, and the complexity of testing and interpretation of results.

For NSCLC, 4 PD-L1 assay systems are currently FDA approved as companion or complementary diagnostics to anti–PD-1/PD-L1 agents (Table 2).4-7,45 It must be noted that a companion diagnostic test is required for the approved use of the drug and is specified in the FDA prescribing information for the drug. In contrast, results of complementary diagnostic tests are considered predictive, but are not required for prescribing the drug.


Each PD-L1 IHC assay for the approved agents was developed with a unique anti–PD-L1 antibody clone (ie, 28-8 [Dako] for nivolumab; 22C3 [Dako] for pembrolizumab; SP263 [Ventana] for durvalumab; and SP142 [Ventana] for atezolizumab; Table 2).4-7,45 In addition, different immuno­staining protocols based on different antigen retrieval conditions and staining platforms are used; the 28-8 and 22C3 anti–PD-L1 antibody clones are optimized with the Dako Autostainer Link 48 staining platform,46 whereas the SP142 and SP263 clones use the Ventana BenchMark ULTRA platform.47 Both platforms are automated IHC systems, including instrumentation and reagent packages that allow high-throughput deparaffinization and staining of individual slides.46,47 Whereas the cell components for PD-L1 expression scoring are tumor cell membranes for all 4 diagnostic systems, the atezolizumab and pembrolizumab testing systems also include tumor-infiltrating PD-L1–positive immune cells as part of the scoring.45

Given such variability among the PD-L1 testing assays, several studies have sought to compare the 4 PD-L1 IHC assays analytically and clinically, to ultimately assess their validity and interchangeability.48-51 To that end, the 2-phase Blueprint PD-L1 IHC Assay Comparison Project was initiated as an industrial–academic collaborative effort including pharmaceutical companies (Genentech/Roche, Bristol-Myers Squibb, Merck & Co Inc, and AstraZeneca), diagnostic companies (Dako/Agilent, Personal Genome Diagnostics, and Ventana/Roche), and national societies (International Association for the Study of Lung Cancer and the American Association for Cancer Research).48

The primary goals of the phase 1 part of this feasibility study in 39 NSCLC tumors were to compare the analytical staining factors reported as percentages of stained tumor and immune cells, as well as to assess clinical diagnostic performance using validated scoring algorithms developed for each assay that classify patients above and below a selected expression cutoff. When stained with the 4 PD-L1 IHC assays (22C3, 28-8, SP142, and SP263) and assessed by expert pathologists from the diagnostic companies, the percentage of PD-L1–stained tumor cells were found to be comparable for 22C3, 28-8, and SP263 assays, whereas the SP142 assay showed fewer stained tumor cells.48

Clinical diagnostic performance comparison using an alternative assay combination of the validated assay and its associated scoring algorithm found that 19 of the 38 evaluable cases (50%) were classified above the cutoffs for all assays, indicating that the clinical positivity was concordant regardless of the assay used; 5 (13%) cases were classified below the selected cutoffs of all assays, meaning that they were clinically negative for PD-L1 expression if an alternative similar assay was used. However, 14 (37%) cases were discordant in terms of being classified as above or below the cutoff, indicating that a different PD-L1 classification would be made depending on the assay/scoring system used. The Blueprint study was not statistically powered to detect differences among the anti–PD-L1 antibodies and did not address specificity and sensitivity of the assays, or performance impact of differences in specimen type or preanalytical factors across laboratories (such as fixatives and tissue processing methods).48

A prospective multi-institutional National Comprehensive Cancer Network (NCCN) study involving 13 pathologists from 7 academic sites also compared the 22C3, 28-8, SP142, and E1L3N antibodies (N = 90 cases).49 Consistent with the Blueprint study, researchers in the NCCN study found that 22C3, 28-8, and E1L3N antibodies were mostly similar, and that the SP142 antibody detected significantly fewer PD-L1–positive tumor and immune cells. The Ratcliffe-AstraZeneca NSCLC Concordance Study also showed strong overall agreement between 22C3, 28-8, and SP263 at different PD-L1 expression cutoffs, including 1%, 10%, 25%, and 50%.50

Based on this evidence, it would be prudent to avoid substitution of diagnostic PD-L1 assays for the approved anti–PD-1/PD-L1 agents. However, given the concordance between 3 of the PD-L1 antibody clones, there is promise in the future for interchangeability following more rigorous studies with larger data sets and real-world samples.

Future Directions and Conclusions

As the field of immunotherapy continues to evolve, selection of patients who will respond best to immunotherapy directed toward the PD-1/PD-L1 axis remains a challenge, underscoring the need for identification of predictive biomarkers. Clinical research efforts have focused on evaluating the role of PD-L1 expression status on tumor cells and tumor-infiltrating immune cells to predict clinical benefit from anti–PD-1/PD-L1 antibodies in many cancers. As outlined in this review, there are limitations to the capability of PD-L1 expression in predicting response to ICIs. PD-L1 expression is variable and appears to depend on multiple factors, including tumor histology, the assay antibody used to measure expression, the cell component assessed (tumor cells and/or tumor-infiltrating immune cells), and the staining cutoff applied (eg, 1%, 5%, 10%, 50%). Based on the available evidence, although higher levels of PD-L1 expression may indicate a greater likelihood of response to anti–PD-1/PD-L1 ICIs, it is an imperfect biomarker of response and may not be suitable as a definitive biomarker to select patients for PD-1/PD-L1 blockade across tumor types and in response to different ICIs.

This conclusion is primarily based on evidence suggesting that patients with tumors expressing low or undetectable levels of PD-L1 may also derive benefit from anti–PD-1/PD-L1 therapy, which is in direct variance with the ideal characteristics of a clinically useful predictive biomarker.17,24,31,32,38,43 As an ideal biomarker, PD-L1 expression on both the tumor and tumor-infiltrating immune cells would be expected to identify potential responders or nonresponders based on a positive or negative PD-L1 test result, facilitating the selection of patients who are likely to respond to anti–PD-1/PD-L1 antibody therapy, while sparing nonresponders from unnecessary treatment and potentially harmful irAEs.34 Currently, we are not at that juncture with PD-1/PD-L1 as a predictive biomarker for ICI therapy.

An important consideration that likely contributes to the limitation of PD-L1 expression as a predictive biomarker is that it is a heterogeneous and dynamic biomarker, exhibiting differences in regions within each tumor and among different tumors, as well as among primary sites and metastases.34,45 Moreover, PD-L1 expression may be modulated by prior exposure to other treatments such as radiation therapy, chemotherapy, and by changes in the tumor microenvironment (eg, tissue hypoxia), as well as during development of treatment resistance.34,45,52 Another confounder for defining PD-L1 positivity is that PD-L1 can be expressed by both tumor and inflammatory cells within the tumor microenvironment; however, their roles in predicting treatment response are presently unclear.45

Given the dynamic nature of PD-L1 expression, use of archival tissue may not be optimal for PD-L1 testing because it may not reflect the true PD-L1 status, which may be affected by variations in tissue fixation, storage, and antigen retrieval, all of which can impact the PD-L1 expression readout.34,45 Moreover, standardization of the staining and scoring methods in IHC analysis is needed for further optimization of PD-L1 assays and adoption into clinical practice.

The complexity of tumor-directed immune responses may likely preclude the use of a single biomarker to determine treatment responses; composite assessment of several biomarkers may ultimately be needed to inform clinical treatment decisions. In this context, there is accumulating evidence to indicate that data from other biomarkers such as tumor mutational burden or microsatellite instability status, combined with PD-L1 tumor expression, may be more informative in predicting response to immunotherapy.34,45 Prospective validation of these concepts in the context of therapeutic clinical trials is warranted.


  1. Alsaab HO, Sau S, Alzhrani R, et al. PD-1 and PD-L1 checkpoint signaling inhibition for cancer immunotherapy: mechanism, combinations, and clinical outcome. Front Pharmacol. 2017;8:561.
  2. Smart Patients. Cancer immunotherapy: PD-1 and beyond. Accessed May 26, 2018.
  3. Bellmunt J, Mullane SA, Werner L, et al. Association of PD-L1 expression on tumor-infiltrating mononuclear cells and overall survival in patients with urothelial carcinoma. Ann Oncol. 2015;26:812-817.
  4. Opdivo (nivolumab) [prescribing information]. Princeton, NJ: Bristol-Myers Squibb; 2018.
  5. Keytruda (pembrolizumab) [prescribing information]. Whitehouse Station, NJ: Merck & Co, Inc; 2017.
  6. Tecentriq (atezolizumab) [prescribing information]. South San Francisco, CA: Genentech; 2018.
  7. Imfinzi (durvalumab) [prescribing information]. Cambridge, England: AstraZeneca, Inc; 2018.
  8. Bavencio (avelumab) [prescribing information]. Rockland, MA: EMD Serono, Inc; 2017.
  9. Sunshine J, Taube JM. PD-1/PD-L1 inhibitors. Curr Opin Pharmacol. 2015; 23:32-38.
  10. Topalian SL, Hodi FS, Brahmer JR, et al. Safety, activity, and immune correlates of anti–PD-1 antibody in cancer. N Engl J Med. 2012;366:2443-2454.
  11. Garon EB, Rizvi NA, Hui R, et al. Pembrolizumab for the treatment of non–small-cell lung cancer. N Engl J Med. 2015;372:2018-2028.
  12. Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial. Lancet. 2016;387:1540-1550.
  13. FDA Approved Drugs. Pembrolizumab (KEYTRUDA) Checkpoint Inhibitor. Accessed February 2, 2018.
  14. Lopes G, Wu Y-L, Kudaba I, et al. Pembrolizumab (pembro) versus platinum-based chemotherapy (chemo) as first-line therapy for advanced/metastatic NSCLC with a PD-L1 tumor proportion score (TPS) ≥ 1%: open-label, phase 3 KEYNOTE-042 study. Presented at: Annual Meeting of the American Society of Clinical Oncology; June 1-5, 2018; Chicago, IL. Abstract LBA4.
  15. Gettinger S, Rizvi NA, Chow LQ, et al. Nivolumab monotherapy for first-line treatment of advanced non–small-cell lung cancer. J Clin Oncol. 2016;34:2980-2987.
  16. Rizvi NA, Mazieres J, Planchard D, et al. Activity and safety of nivolumab, an anti-PD-1 immune checkpoint inhibitor, for patients with advanced, refractory squamous non-small-cell lung cancer (CheckMate 063): a phase 2, single-arm trial. Lancet Oncol. 2015;16:257-265.
  17. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus docetaxel in advanced squamous-cell non-small-cell lung cancer. N Engl J Med. 2015;373:123-135.
  18. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus docetaxel in advanced non-squamous non–small-cell lung cancer. N Engl J Med. 2015;373:1627-1639.
  19. Reck M, Rodríguez-Abreu D, Robinson AG, et al. Pembrolizumab versus chemotherapy for PD-L1–positive non–small-cell lung cancer. N Engl J Med. 2016;375:1823-1833.
  20. Carbone DP, Reck M, Paz-Ares B, et al. First-line nivolumab in stage IV or recurrent non–small-cell lung cancer. N Engl J Med. 2017;376:2415-2426.
  21. Fehrenbacher L, Spira A, Ballinger M, et al. Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial. Lancet. 2016;387:1837-1846.
  22. Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet. 2017;389:255-265.
  23. Antonia SJ, Villegas A, Daniel D, et al. Durvalumab after chemoradiotherapy in stage III non–small-cell lung cancer. N Engl J Med. 2017;377:1919-1929.
  24. Hellmann MD, Rizvi NA, Goldman JW, et al. Nivolumab plus ipilimumab as first-line treatment for advanced non-small-cell lung cancer (CheckMate 012): results of an open-label, phase 1, multicohort study. Lancet Oncol. 2017;18:31-41.
  25. Gandhi L, Rodríguez-Abreu D, Gadgeel S, et al; for the KEYNOTE-189 Investigators. Pembrolizumab plus chemotherapy in metastatic non-small-cell lung cancer. N Engl J Med. 2018;378:2078-2092.
  26. Paz-Ares LG, Luft A, Tafreshi A, et al. Phase 3 study of carboplatin-paclitaxel/nab-paclitaxel (Chemo) with or without pembrolizumab (Pembro) for patients (Pts) with metastatic squamous (Sq) non-small cell lung cancer (NSCLC). J Clin Oncol. 2018;36(15 suppl):105.
  27. Socinski MA, Jotte RM, Cappuzzo F, et al; for the IMpower150 Study Group. Atezolizumab for first-line treatment of metastatic nonsquamous NSCLC. N Engl J Med. 2018;378:2288-2301.
  28. Jotte RM, Cappuzzo F, Vynnychenko I, et al. IMpower131: primary PFS and safety analysis of a randomized phase III study of atezolizumab + carboplatin + paclitaxel or nab-paclitaxel vs carboplatin + nab-paclitaxel as 1L therapy in advanced squamous NSCLC. J Clin Oncol. 2018;36(18 suppl):LBA9000.
  29. Motzer RJ, Escudier B, McDermott DF, et al. Nivolumab versus everolimus in advanced renal-cell carcinoma. N Engl J Med. 2015;373:1803-1813.
  30. Atkins M, McDermott D, Powles T, et al. IMmotion150: a phase II trial in untreated metastatic renal cell carcinoma (mRCC) patients (pts) of atezolizumab (atezo) and bevacizumab (bev) vs and following atezo or sunitinib (sun). J Clin Oncol. 2017;35(15 suppl):4505.
  31. Wolchok JD, Chiarion-Sileni V, Gonzalez R, et al. Overall survival with combined nivolumab and ipilimumab in advanced melanoma. N Engl J Med. 2017;377:1345-1356.
  32. Robert C, Long GV, Brady B, et al. Nivolumab in previously untreated melanoma without BRAF mutation. N Engl J Med. 2015;372:320-330.
  33. Robert C, Schachter J, Long GV, et al. Pembrolizumab versus ipilimumab in advanced melanoma. N Engl J Med. 2015;372:2521-2532.
  34. Zhu J, Armstrong AJ, Friedlander TW, et al. Biomarkers of immunotherapy in urothelial and renal cell carcinoma: PD-L1, tumor mutational burden, and beyond. J Immunother Cancer. 2018;6:4.
  35. Rosenberg JE, Hoffman-Censits J, Powles T, et al. Atezolizumab in patients with locally advanced and metastatic urothelial carcinoma who have progressed following treatment with platinum-based chemotherapy: a single-arm, phase 2 trial. Lancet. 2016;387:1909-1920.
  36. Powles T, Duran I, van der Heijden MS, et al. Atezolizumab versus chemotherapy in patients with platinum-treated locally advanced or metastatic urothelial carcinoma (IMvigor211): a multicentre, open-label, phase 3 randomised controlled trial. Lancet. 2018;391:748-757.
  37. Bellmunt J, de Wit R, Vaughn DJ, et al. Pembrolizumab as second-line therapy for advanced urothelial carcinoma. N Engl J Med. 2017;376:1015-1026.
  38. Sharma P, Retz M, Siefker-Radtke A, et al. Nivolumab in metastatic urothelial carcinoma after platinum therapy (CheckMate 275): a multicentre, single-arm, phase 2 trial. Lancet Oncol. 2017;18:312-322.
  39. Powles T, O’Donnell PH, Massard C, et al. Efficacy and safety of durvalumab in locally advanced or metastatic urothelial carcinoma: updated results from a phase 1/2 open-label study. JAMA Oncol. 2017;3:e172411.
  40. Apolo AB, Infante JR, Balmanoukian A, et al. Avelumab, an anti–programmed death-ligand 1 antibody, in patients with refractory metastatic urothelial carcinoma: results from a multicenter, phase Ib study. J Clin Oncol. 2017;35: 2117-2124.
  41. Balar AV, Galsky MD, Rosenberg JE, et al. Atezolizumab as first-line treatment in cisplatin-ineligible patients with locally advanced and metastatic urothelial carcinoma: a single-arm, multicentre, phase 2 trial. Lancet. 2017;389:67-76.
  42. Balar AV, Castellano D, O’Donnell PH, et al. First-line pembrolizumab in cisplatin-ineligible patients with locally advanced and unresectable or metastatic urothelial cancer (KEYNOTE-052): a multicentre, single-arm, phase 2 study. Lancet Oncol. 2017;18:1483-1492.
  43. Ferris RL, Blumenschein G Jr, Fayette J, et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med. 2016;375:1856-1867.
  44. Chow LQM, Haddad R, Gupta S, et al. Antitumor activity of pembrolizumab in biomarker-unselected patients with recurrent and/or metastatic head and neck squamous cell carcinoma: results from the phase Ib KEYNOTE-012 Expansion Cohort. J Clin Oncol. 2016;34:3838-3845.
  45. Grigg C, Rizvi NA. PD-L1 biomarker testing for non-small cell lung cancer: truth or fiction? J Immunother Cancer. 2016;4:48.
  46. Colley EC, Stead RH. Optimized immunohistochemistry workflow facilitated by new Dako Autostainer Link 48 Software. Accessed May 20, 2018.
  47. Ventana introduces the BenchMark Special Stains platform [press release]. Tucson, AZ: Ventana Medical Systems, Inc; March 19, 2012. Accessed May 20, 2018.
  48. Hirsch FR, McElhinny A, Stanforth D, et al. PD-L1 immunohistochemistry assays for lung cancer: results from phase 1 of the Blueprint PD-L1 IHC Assay Comparison Project. J Thorac Oncol. 2017;12:208-222.
  49. Rimm DL, Han G, Taube JM, et al. A prospective, multi-institutional pathologist-based assessment of four immunohistochemistry assays for PD-L1 expression in non-small cell lung cancer. JAMA Oncol. 2017;3:1051-1058.
  50. Ratcliffe MJ, Sharpe A, Midha A, et al. AACR 2016. Agreement between programmed cell death ligand-1 diagnostic assays across multiple protein expression cutoffs in non-small cell lung cancer. Clin Cancer Res. 2017;23:3585-3591.
  51. Brunnström H, Johansson A, Westbom-Fremer S, et al. PD-L1 immunohistochemistry in clinical diagnostics of lung cancer: inter-pathologist variability is higher than assay variability. Mod Pathol. 2017;30:1411-1421.
  52. Shin SJ, Jeon YK, Cho YM, et al. The association between PD-L1 expression and the clinical outcomes to vascular endothelial growth factor-targeted therapy in patients with metastatic clear cell renal cell carcinoma. Oncologist. 2015;20:1253-1260.
Last modified: July 22, 2021