Dr Liza Villaruz (Hillman Cancer Center, PA, USA) examines recent KEYNOTE trials investigating pembrolizumab for advanced non-small-cell lung cancer and the questions they raise regarding patient selection and combination therapy.
Pembrolizumab was approved in 2016 as a single agent in the first-line treatment of advanced non-small-cell lung cancer (NSCLC) patients with PD-L1 expression on at least 50% of their tumor cells (tumor proportion score (TPS)≥50%). In June 2018, the phase III clinical trial KEYNOTE-042 was presented in the Plenary Session of the 2018 Annual Meeting of the American Society of Clinical Oncology. This study demonstrated significant survival gains with pembrolizumab over platinum-based chemotherapy in untreated patients with advanced NSCLC and any level of PD-L1 positivity (TPS≥1%) [1]. At first glance, these data appear to show an extension of the benefits of pembrolizumab from the minority of patients with high PD-L1 expressing tumors to the majority of patients with any level of PD-L1 positivity. However, consideration of these data in greater detail indicates that rigorous selection of patients based on PD-L1 expression is still needed.
Setting the stage
Pembrolizumab made its breakthrough in the first-line setting with KEYNOTE-024, a landmark randomized, phase III clinical trial that compared pembrolizumab single-agent immunotherapy with an investigators’ choice of platinum-doublet chemotherapy in patients with untreated advanced NSCLC and a PD-L1 TPS≥50%. In this patient population, pembrolizumab was associated with unprecedented gains in overall survival (OS) with a hazard ratio (HR) of 0.60 (p=0.005) compared with chemotherapy [2]. What has been profoundly impactful is the endurance of these survival gains in long-term follow-up, despite high rates of crossover in the chemotherapy arm of the trial to second-line immunotherapy [3]. KEYNOTE-024 not only established the role of immunotherapy in the first line treatment of advanced NSCLC, it cemented the role of PD-L1 testing in the diagnostic evaluation, as a means of personalizing therapy.
KEYNOTE-042: Caution warranted
More recently, KEYNOTE-042 compared single-agent pembrolizumab with platinum-based chemotherapy in patients with untreated advanced NSCLC and a PD-L1 TPS≥1%. This study confirmed the survival benefit of pembrolizumab over chemotherapy in patients with advanced NSCLC and a PD-L1 TPS≥50%, and further demonstrated significant survival gains based on PD-L1 TPS levels of ≥20% and ≥1% [1]. These results, at face value, extend the benefits of pembrolizumab from the minority of patients with high PD-L1 expressing tumors (about one third) to the majority of patients with any level of PD-L1 positivity.
However, it is notable that the gains in survival are less pronounced as the bar for PD-L1 expression is lowered. The HR for OS is 0.69 in patients with PD-L1 TPS≥50%, which increases to 0.77 in patients with PD-L1 TPS≥20% and 0.81 in patients with PD-L1 TPS≥1%. When measured against the ASCO HR standard for clinically meaningful trial outcomes, the HR of 0.81 is not practice changing. In an exploratory analysis of OS in patients with PD-L1 TPS≥1–49%, the OS HR is an insignificant 0.92. These data suggest that the survival benefit in KEYNOTE-042 is driven almost entirely by patients with high PD-L1-expressing tumors.
While KEYNOTE-024 cemented the role of single-agent pembrolizumab in patients with PD-L1 overexpressing NSCLC, KEYNOTE-042 did little to advance our understanding of the role of immunotherapy in this patient population.
Chemo-immunotherapy in the PD-L1 TPS<50% population
In approaching the advanced NSCLC patient population with PD-L1 TPS<50%, the KEYNOTE-189 data have been hugely impactful. This phase III clinical trial randomly assigned untreated metastatic non-squamous NSCLC patients to receive carboplatin and pemetrexed plus pembrolizumab or carboplatin and pemetrexed alone. In the entire study population, the HR for OS was 0.49 (p<0.001) [4]. Remarkably, depth of the HR persisted across PD-L1 expression cut-offs, with a HR for OS of 0.59 in patients with no PD-L1 expression, 0.55 in patients with PD-L1 TPS≥1–49%, and 0.42 in patients with PD-L1 TPS≥50% – all profoundly statistically significant.
A similar study in patients with metastatic squamous NSCLC, KEYNOTE-407, randomly assigned patients to carboplatin and paclitaxel/nab-paclitaxel, with or without pembrolizumab. An interim analysis demonstrated a HR for OS of 0.64 in the entire population, which remained significant in patients with PD-L1 TPS<1% (HR=0.61), PD-L1 TPS≥1–49% (HR=0.57), and PD-L1 TPS≥50% (HR=0.64) [5]. Together, these KEYNOTE studies establish a chemo-immunotherapy approach in patients with advanced NSCLC and PD-L1 TPS<50%.
The question central to the interpretation of the KEYNOTE data is how PD-L1 should be used in personalizing immunotherapy treatment in the first-line setting
Chemo-immunotherapy in the PD-L1 TPS≥50% population
The larger issue relates to the potential added benefit of combining chemotherapy with immunotherapy over immunotherapy alone in the PD-L1-overexpressing population (PD-L1≥50%). In the KEYNOTE-189 patient population, the response rate to combined chemotherapy plus pembrolizumab was 61.4%, as opposed to 44.8% with pembrolizumab alone in KEYNOTE-024 [2,4]. These data suggest the combined approach is superior from the stand point of achieving tumor shrinkage, which is particularly useful in patients with large symptomatic tumor burdens.
However, when considering the data in terms of survival, the 12-month OS rate with chemotherapy plus pembrolizumab was 69.2%; the median OS was not reached after 10.5 months of follow-up. This is on par with pembrolizumab alone in the same patient population that achieved a 12-month OS rate of 70.3%, and a median OS that has not been reached after 19 months of follow-up.
Single-agent immunotherapy also spares the toxicities of platinum-based chemotherapy and preserves the possibility of adding or sequencing chemotherapy at the time of progression on single-agent immunotherapy. Adding or sequencing chemotherapy at the time of pembrolizumab failure will be studied in closer detail in the proposed ECOG 5163 clinical trial. Unless longer follow-up of KEYNOTE-189 demonstrates a higher plateau in overall survival than KEYNOTE-024, single-agent immunotherapy remains a standard of care in this patient population.
Conclusions
Three of the four randomized, phase III clinical trials discussed above have been presented in the last 6 months. This has resulted in unprecedented shifts in standard-of-care practice in a short timeframe. The question central to the interpretation of these data is how PD-L1 should be used in personalizing immunotherapy treatment in the first-line setting. While it may be tempting based on the KEYNOTE-042 data to prescribe single-agent pembrolizumab to patients with any level of PD-L1 positivity, this approach should be reserved for patients with high PD-L1-expressing tumors.
Liza Villaruz is a member of the Academic Thoracic Oncology Medical Investigator’s Consortium’ (ATOMIC)