Pembrolizumab
The phase III KEYNOTE-024 trial comprises patients with advanced and strongly PD-L1-positive NSCLC [
15]. A total of 1942 patients were screened for enrolment; 1653 had evaluable samples and 500 (30.2%) patients had tumors with PD-L1 expression ≥ 50%. A total of 305 patients who met the inclusion criteria were randomized to pembrolizumab (200 mg every 3 weeks for up to 35 cycles or until documented progressive disease) versus four to six cycles of standard of care platinum-based chemotherapy (platinum/pemetrexed, platinum/gemcitabine, or carboplatin/paclitaxel) as first-line treatment. Pemetrexed maintenance therapy was received by 30% of patients with non-squamous histology. In addition, 43.7% of patients in the control arm crossed over per protocol to pembrolizumab upon disease progression. Patients were excluded from the trial if they were harboring
EGFR mutations or
ALK translocations, had an Eastern Cooperative Oncology Group (ECOG) performance status ≥ 2, had untreated brain metastasis, or were receiving any dose of oral steroids for an autoimmune disease. The primary endpoint of the trial was the median PFS. Compared with standard first-line platinum-based chemotherapy, pembrolizumab significantly improved the primary endpoint from 6.0 to 10.3 months (hazard ratio (HR), 0.50; 95% confidence interval (CI), 0.37–0.68;
P < 0.001). The RR according to the Response Evaluation Criteria In Solid Tumors (RECIST) (44.8% vs. 27.8%;
P < 0.001) and OS (not reached in both arms; HR, 0.60; 95% CI, 0.41–0.89;
P = 0.005) were also improved, with a 1-year OS of 70% versus 54% [
15]. The benefit of pembrolizumab with respect to PFS was evident in all subgroups examined according to sex, age, histology, smoking status, and brain metastases at baseline. However, the benefit was lower in female and never-smoker patients (probably related to the lower mutational load in this population [
16]), and the greatest benefit to PFS was observed in patients with squamous histology (HR, 0.35; 95% CI, 0.17–0.71). Grade 3, 4, or 5 treatment-related AEs also favored pembrolizumab (26.6% vs. 53.3%). The incidence of grade 3–4 immune-mediated AEs was 9.7% with pembrolizumab [
15]. Pembrolizumab had a clinically meaningful improvement in quality of life compared to platinum-based chemotherapy [
17] (Table
1).
Table 1
Immune checkpoint inhibitors in first-line treatment in advanced non-small cell lung cancer patients
KEYNOTE 024 [ 15] Pembrolizumab vs. CT | III | 305 PD-L1 ≥ 50%a
| 44.8 vs. 27.8
P < 0.001 | 10.3 vs. 6.0 HR, 0.50; P < 0.001 | HR, 0.60; P = 0.005 1-year OS: 70% vs. 54% | 26.6 vs. 53.3 |
CheckMate 026 [ 31] Nivolumab vs . CT | III | 423 PD-L1 ≥ 5%b
| 26.1 vs. 33.5 | 4.2 vs. 5.9 HR, 1.15; P = 0.251 | HR, 1.02 14.4 vs. 13.2 months | 17.6 vs. 50.6 |
KEYNOTE 021 [ 39] Pembrolizumab + CT vs. CT | II | 123 | 55 vs. 29
P = 0.0016 | 13.0 vs. 8.9 HR, 0.53; P = 0.0102 | HR, 0.90; P = 0.39 | 39 vs. 26 |
Nivolumab + Ipilimumab/12 weeks until disease progression or toxicity Nivolumab + Ipilimumab/6 weeks until disease progression or toxicity | I | 38 40 | 47 38 | 8.1 3.9 | Not calculated | 37 33 |
The magnitude of benefit in the control arm was consistent with historic controls [
18], suggesting that pembrolizumab efficacy is not overestimated for an ineffective control arm. However, it is unknown whether the survival benefit was due to pembrolizumab treatment being intrinsically more potent as a first-line treatment or because crossover was limited to < 50% of the patients in the control arm. Indeed, trials in patients with
EGFR-mutant or
ALK-rearranged NSCLC have had much higher rates of crossover from chemotherapy to personalized treatment after platinum-based chemotherapy progression (65% in
EGFR-mutant [
19] and 70% in
ALK-positive populations [
20]), leading to a lack of survival differences between treatment arms. The clear benefit for OS could also be due to a potentially lower efficacy of pembrolizumab in platinum-pretreated patients than in chemo-naïve patients, whereas, in the same settings, targeted therapies yield the same benefit [
21,
22].
In the KEYNOTE-024 trial, 11.7% of patients in the pembrolizumab arm had previously treated brain metastases at baseline; the PFS benefit in this subgroup was similar to that in patients without such metastases at baseline (HR, 0.55 vs. HR, 0.50). The efficacy of pembrolizumab in patients with PD-L1-positive (>1%) NSCLC with untreated or progressive asymptomatic brain metastases measuring between 5 and 20 mm in diameter has also recently been tested in a phase II trial that reported a cerebral response rate of 33% [
23]; the median duration of confirmed brain responses was 6 months. Approximately 17% of NSCLC patients have brain metastases at baseline [
24]. In our opinion, supra-tentorial asymptomatic brain metastases should not be considered exclusion criteria for immune checkpoint inhibitor treatment. The risk of brain metastases increases over time due to the prolonged survival of patients with advanced NSCLC [
25]. Therefore, further investigations are needed to determine optimal treatment combinations with brain radiotherapy, sequences of treatment, and safety [
26].
Globally, pembrolizumab results from KEYNOTE-024 [
15] were consistent with the efficacy observed in the KEYNOTE-001 study [
27] in the subgroup of chemo-naïve patients. The FDA approved pembrolizumab in the first-line setting in this population on October 24, 2016, and on December 15, 2016, the EMA Committee for Medicinal Products for Human Use also approved pembrolizumab as monotherapy in the first-line setting of metastatic NSCLC in adults whose tumors express PD-L1 with a tumor proportion score (TPS) ≥ 50% and who have no
EGFR- or
ALK-positive tumor mutations. The efficacy of pembrolizumab as a first-line treatment in NSCLC patients with PD-L1 expression < 50% remains unknown. The ongoing phase III KEYNOTE-042 study (NCT02220894) will assess the survival benefit of pembrolizumab over standard first-line platinum-based chemotherapy in treatment-naïve patients who have tumors with ≥ 1% PD-L1 positivity. Stratification according to PD-L1 expression (strong (≥ 50%) vs. weak (1–49%)) will be performed in the study.
Among the 30% of patients whose tumors express PD-L1 with a TPS ≥ 50%, other clinical exclusion criteria limit the extended use of pembrolizumab in the first-line setting; for example, exclusion of patients considered unfit or with poor performance status (representing almost 34% of NSCLC patients in contemporary cohorts [
28]), patients with
EGFR-mutant and
ALK-rearranged tumors (approximately 17% of adenocarcinoma lung cancers in Caucasian populations [
29]), and the absence of steroids or autoimmune disorders (13.5% of lung cancer patients [
30]). As such, the pool of patients eligible for upfront pembrolizumab is certainly not 30% of all chemo-naïve patients with NSCLC (which represents the percentage of frontline patients whose tumors express PD-L1 with a TPS ≥ 50%), but probably closer to 10%; this pool clearly needs to be enlarged.
Moreover, the turnaround time from patient selection to treatment, based on PD-L1 expression, is not reported in KEYNOTE-024 but is expected to be considerablylonger than 1 month. There is a high probability that patients with relatively indolent disease were favored for inclusion in the study, adding another bias compared to routine practice. Patients with a poorer prognosis need to be explored such as in the ongoing phase II trial NCT02879617 evaluating first-line durvalumab in performance status 2 patients with advanced NSCLC.
Nivolumab
The phase III CheckMate 026 trial tested the efficacy of nivolumab compared to standard first-line chemotherapy (platinum/pemetrexed, platinum/gemcitabine, or carboplatin/paclitaxel) in 423 patients with PD-L1-positive (≥ 5% of expression by 28-8 clone) advanced NSCLC [
31]. Patients harboring
EGFR mutations or
ALK translocations were ineligible. Patients with adequately treated brain metastases were allowed. No imbalances were reported in either arm regarding brain metastases (~12%), histology (~24% of squamous), ECOG performance status (~30% PS0), or current smokers (~20% in both arms). A higher proportion of females was included in the chemotherapy arm (45.2% vs. 32.1%). Maintenance treatment was prescribed in 38% of patients [
31]. No benefit was seen with nivolumab compared to chemotherapy in terms of the primary endpoints PFS (4.2 vs. 5.9 months; HR, 1.15; 95% CI, 0.91–1.45;
P = 0.251), OS (14.4 vs. 13.2 months; HR, 1.02; 95% CI, 0.80–1.30), or RR (26.1% vs. 33.5%). However, the toxicity profile favored nivolumab, with 17.6% of patients having grade 3–4 AEs compared to 50.6% in the chemotherapy arm (Table
1). Of note, patients with NSCLC with strong PD-L1 expression (TPS ≥ 50%) did not derive a greater benefit from nivolumab than those with weaker expression. Nivolumab was the post-discontinuation treatment in 60% of patients in the chemotherapy arm. The lack of survival benefit could be related to various hypothetical factors. First, there was a higher proportion of tumors with strong PD-L1 expression (TPS ≥ 50%) in the control arm compared to the nivolumab arm (74.1% vs. 53.2%). Second, only 44% of patients in the nivolumab arm received second-line treatment, mostly platinum-based chemotherapy, suggesting that a certain subgroup of patients was untreated [
31]. This could be a consequence of hyper-progressive diseases on immunotherapy, as recently reported by Champiat et al. [
32]. Overall, results from CheckMate 026 in the whole population and for those with strongly positive PD-L1-expressing tumors are inconsistent with first-line nivolumab performance in phase I/II trials [
33].
Although the reason for the contrasting results between the KEYNOTE-024 [
15] and Checkmate 026 [
31] trials remains unclear, we should consider the nivolumab trial as negative, and we believe that differences in patient selection are the primary cause of this discrepancy. Differences in biomarker tests and in PD-L1 expression cut-off point (22C3 and 50% with pembrolizumab vs. 28-8 clone and 5% with nivolumab) could have contributed to the discordant results between the trials. Thus, patients with strong PD-L1 positivity in the KEYNOTE-024 trial may not be similar to patients with strong PD-L1 positivity in the CheckMate 026 trial since the sensitivity of the relevant clones used to define PD-L1 status is potentially different. Additionally, PD-L1 testing was performed after metastatic diagnosis in the pembrolizumab trial, whereas in the nivolumab trial, it was performed in archival tissue biopsy specimens taken within 6 months prior to randomization. However, in the KEYNOTE-010 trial, survival benefit with pembrolizumab as a second-line treatment was independent of whether the PD-L1 test was performed in an archival or in a new tissue biopsy specimen [
11]. Further, the efficacy of immune checkpoint inhibitors is higher among smokers [
16]; a higher percentage of never-smoker patients was included in the nivolumab trial than in the pembrolizumab trial (11% vs. 3%) and such patients have lower mutational loads that negatively correlate with the success of immune checkpoint-targeting therapies [
34]. Another major difference between the trials was the percentage of patients who received radiotherapy prior to enrollment; this percentage was abnormally high (37.6%) for patients enrolled in the CheckMate 026 trial [
31], whereas in the KEYNOTE-024 trial [
15], prior radiation therapy of > 30 Gy within 6 months of the first dose of trial treatment constituted an exclusion criterion. Therefore, sites that were involved in both trials may have operationally favored enrollment of all previously irradiated patients into CheckMate 026. It is clear that previous radiotherapy can have major consequences on the tumor microenvironment [
35] and potentially lead to decreased activity of immune checkpoint inhibitors in previously irradiated areas. At this stage, it is unknown which areas (mediastinum, others) were previously irradiated in patients enrolled in the CheckMate 026 trial [
31].
Other first-line randomized phase III clinical trials are testing anti-PD-1 monotherapy, such as nivolumab in CheckMate 227 (NCT02477826), or anti-PD-L1 monotherapies such as atezolizumab in IMpower 110 (NCT02409342) and avelumab in the ongoing JAVELIN Lung 100 trial (NCT02576574). These trials may validate immune checkpoint inhibitors as a first-line treatment in patients with PD-L1-positive NSCLC.