Kantar Health Blog

CheckMate? The rules of the game may be changing

by Arnold DuBell | May 30, 2015
Arnold DuBell
Co-authored by Stephanie Hawthorne, Ph.D. 

Non-small cell lung cancer (NSCLC) is one of the tumor types with the highest unmet need, representing the most commonly diagnosed cancer globally1 and one that is frequently diagnosed at an advanced stage,2 where prognosis is most dire. As such, NSCLC has long been a focus of significant clinical development and currently is a poster child for personalized medicine; treatment varies significantly depending on a patient’s tumor histology and their biomarker status (EGFR mutation and ALK translocation are currently the key biomarkers guiding clinical decision making, although many others are under study).

The most recently approved agent for the treatment of NSCLC is Opdivo® (nivolumab, Bristol-Myers Squibb/Ono Pharmaceuticals), which was approved by the FDA in March 2015 for squamous metastatic NSCLC patients who have previously received a platinum-based therapy. This approval was based largely on a single-arm Phase II trial, supplemented by randomized Phase III data (the CheckMate-017 trial) that will be reported publicly for the first time at the American Society for Clinical Oncology (ASCO) annual meeting tomorrow. The approval in squamous histology is the first targeted therapy approved for this subset of patients, representing a major advance in their disease management. However, squamous NSCLC represents only a fraction (approximately 25%) of all NSCLC, leaving the majority of NSCLC patients who have non-squamous disease with an unknown level of benefit from this PD-1 inhibitor.

One of the most anticipated data from ASCO 2015 are the results of CheckMate-057, the Phase III trial that compares Opdivo versus docetaxel as second-line therapy for patients with non-squamous metastatic NSCLC.3 This trial randomized 582 patients to treatment with 3 mg/kg q2w Opdivo or 75 mg/m2 q3w docetaxel, both administered until disease progression or unacceptable toxicity. In the intent-to-treat (ITT) population, Opdivo significantly improved overall survival (OS) compared to docetaxel. OS is the gold standard by which to judge activity, and Opdivo resulted in a 27% reduction in the risk of death, which amounted to a 2.8-month difference at the median (12.2 months vs. 9.4 months, HR 0.73, p=0.0015). Importantly, the survival curve plateaued, with the one-year OS rates significantly differing between the two treatment arms (51% vs. 39%) and the curves continuing to separate even at later time points (18-month OS is approximately 40% vs. 20%, although there are still a large number of censored patients at this time point). No new safety signals were observed, and no single adverse event occurred in more than 1% of patients. However, the ITT population showed no progression-free survival (PFS) benefit (median PFS 2.3 months vs. 4.2 months, HR 0.92, p=0.3932). Interestingly, both the PFS and OS curves crossed, and the Opdivo arm eventually showed more patients progression-free at one year than those treated with docetaxel (19% vs. 8%). These results in the ITT population appear to be slightly less robust than the data for Opdivo in squamous patients, as reported for the CheckMate-017 trial to be presented later during ASCO.4

While it was exciting to see an OS benefit in the broad population of non-squamous NSCLC patients, the real story was the secondary endpoint of outcomes according to PD-L1 biomarker status. Patients were not required to overexpress the ligand for study entry but were required to have tissue available for secondary analysis and stratification. The majority (78%) of enrolled patients had tumor samples with quantifiable PD-L1 expression. When OS was evaluated by PD-L1 expression status, a clear and large benefit was associated with biomarker overexpression, and no OS benefit was seen in patients with less than 1% PD-L1 expression. Additionally, the magnitude of OS benefit (measured as the Hazard ratio; please refer to the table below) appeared to gain significance when higher thresholds of PD-L1 positivity were assessed. A similar association between activity and threshold of PD-L1 positivity was observed with regard to PFS and response rate outcomes.

Overall Survival data from CheckMate-057

PD-L1 expression level

Median OS, months

Hazard ratio



<1% (n=159)

10.4 months

10.1 months

HR 0.90

≥1% (n=108)

17.2 months

9.0 months

HR 0.59

≥5% (n=79)

18.2 months

8.1 months

HR 0.43

≥10% (n=72)

19.4 months

8.0 months

HR 0.40


These results present a bit of a conundrum. In one sense, the trial met its primary endpoint in the ITT population, showing a significant OS benefit for pretreated non-squamous NSCLC patients regardless of biomarker status. This could support FDA approval in such a broad population, and indeed its current approval in squamous NSCLC is not limited by biomarker (we hope to learn at tomorrow’s presentation of CheckMate-017 whether a correlation exists with biomarker status in that patient population). Use of Opdivo in the ITT population without limitation by PD-L1 status was supported by Dr. Roy Herbst, who discussed the CheckMate-057 data. He argued that the biomarker isn’t ready for prime time, for a number of reasons. On the other hand, the lack of any benefit in patients with less than 1% PD-L1 expression makes a strong case for not using Opdivo in these patients, especially given the high cost of this therapy (in and of itself, but also when compared to generic docetaxel).

Two clear arguments emerge in favor of using Opdivo regardless of biomarker status. First, it is clearly better tolerated than docetaxel, with very few Grade 3/4 adverse events observed with Opdivo, and docetaxel is characteristically associated with myelosuppression. Therefore, even in a patient with no PD-L1 expression, Opdivo could be viewed as an equally effective and better tolerated treatment option. The second argument, and the one that is the subject of the most debate, is the reliability of the PD-L1 diagnostic assay and how and when it is evaluated. The majority of patients in CheckMate-057 had archival tumor tissue that was evaluated for the biomarker. Knowing that tumors evolve over time, how accurate is it to categorize a patient’s biomarker status for treatment in second-line based on an assay that was performed on tumor tissue collected potentially much earlier in the course of their disease? Additionally, tumors are notoriously heterogeneous, both within the same lesion as well as comparing primary and metastatic lesions. Can we be sure a patient is truly PD-L1-negative based on lack of overexpression in a single tumor sample? Although Dr. Herbst was particularly vocal about this point during his discussion, this same argument can be made about most biomarkers used in oncology today, and yet we don’t argue that EGFR wildtype NSCLC or KRAS-mutant colorectal cancer patients should be treated with Tarceva® (erlotinib, OSI/Astellas/Roche) or Erbitux® (cetuximab, Eli Lilly/Merck KGaA), respectively. Why should we treat the PD-L1 biomarker any differently? One reason to offer Opdivo to PD-L1-negative patients may be the potential lost long-term OS benefit if an active drug is withheld from a patient based on the results of an imperfect assay.

Clearly the issue of PD-L1 as a necessary biomarker in NSCLC continues to be a subject of debate. Data to be presented later at this conference for other checkpoint inhibitors will add to the body of data for this, but the final answer is unlikely to emerge before the conference concludes. However, the data will add complexity to the treatment paradigm and has the potential to change the rules of the game when it comes to PD-1 and PD-L1 inhibitors. The results of CheckMate-057 clearly establish Opdivo as a new standard of care for second-line NSCLC. What remains to be determined is for how many of those patients?


  1. Globocan 2012, available from globocan.iarc.fr; accessed May 30, 2015.
  2. Kantar Health, CancerMPact Patient Metrics, available from www.cancermpact.com; accessed May 30, 2015.
  3. Paz-Ares L, Horn L, Borghaei H, et al. Phase III, randomized trial (CheckMate 057) of nivolumab (NIVO) versus docetaxel (DOC) in advanced non-squamous cell (non-SQ) non-small cell lung cancer (NSCLC). J Clin Oncol. 2015;33(suppl; abstr LBA109).
  4. Spigel DR, Reckamp KL, Rizvi NA, et al. A phase III study (CheckMate 017) of nivolumab (NIVO; anti-programmed death-1 [PD-1]) vs docetaxel (DOC) in previously treated advanced or metastatic squamous (SQ) cell non-small cell lung cancer (NSCLC). J Clin Oncol. 2015;33: (suppl; abstr 8009).




Leave a comment