Subgroup Observations From Missed KEYNOTE-789 Data in NSCLC Provide Insight for the Future


James Chih-Hsin Yang, MD, PhD, discusses the implications of the phase 3 KEYNOTE-789 trial and what directions investigators can examine in the future to help fill the unmet need for patients with TKI-resistant, EGFR-mutated metastatic non–small cell lung cancer.

Lung cancer

Lung cancer

Although data from the phase 3 KEYNOTE-789 trial (NCT03515837) examining pemetrexed and platinum-based chemotherapy with or without pembrolizumab (Keytruda) for patients with TKI-resistant, EGFR-mutated metastatic non–small cell lung cancer (NSCLC) missed both primary end points, patients who received the pembrolizumab combination had numerically prolonged survival overall and in select subgroups, according to James Chih-Hsin Yang, MD, PhD.

“The standard of care [SOC] for EGFR-mutant NSCLC after failure of frontline EGFR TKI [treatment] is still combination chemotherapy, the most popular being pemetrexed plus carboplatin or cisplatin,” Yang said. “This study shows us that this remains the SOC but studies adding immune checkpoint inhibitors [ICIs] and even studies [examining] anti-angiogenesis inhibitors such as bevacizumab [Avastin] are warranted in the future.”

The final analysis of the KEYNOTE-789 study demonstrated that the median overall survival (OS) for patients who received pembrolizumab plus chemotherapy (n = 245) was 15.9 months (95% CI, 13.7-18.8) vs 14.7 months (95% CI, 12.7-17.1) in the placebo plus chemotherapy arm (n = 247; HR, 0.84; 95% CI, 0.69-1.02; P = .0362). At the final analysis, the duration of response (DOR) was 6.3 months (95% CI, 2.3-40.8+) vs 5.6 months (95% CI, 1.8+ to 40.6+), respectively. The median progression-free survival (PFS) at the second interim analysis was 5.6 months (95% CI, 5.5-5.8) vs 5.5 months (95% CI, 5.4-5.6), respectively (HR, 0.80; 95% CI, 0.65-0.97; P .0122).
In an interview with OncLive®, Yang, the Director of the Graduate Institute of Oncology at the National Taiwan University and director of the Department of Oncology at the National Taiwan University Hospital, discussed implications of the trial and what directions investigators can examine in the future to help fill the unmet need for this group of patients.

OncLive: What was the rationale for examining this treatment regimen in KEYNOTE-789?

Yang: For patients with EGFR mutations and stage IV NSCLC cancer, the SOC is to use EGFR TKIs in the frontline, but at the end of the day patients will progress, so the standard treatment afterward is combination chemotherapy. Combination chemotherapy provides some DOR and [improvements in] PFS, yet it is not satisfactory.

For [patients with] stage IV, wild-type EGFR-[positive] NSCLC, the SOC is immunotherapy plus chemotherapy because [data from] many studies in the past showed that in randomized phase 3 [trials] adding ICIs to combination chemotherapy vs chemotherapy [alone] improves OS. Most of these studies excluded patients with EGFR mutations because in earlier analyses, it was shown that patients with EGFR mutations have inferior responses to ICIs. Therefore, it is of importance to have this study evaluating the role of adding ICIs to SOC second line combination chemotherapy vs chemotherapy alone [to evaluate] OS benefit and PFS benefit.

What were the enrollment criteria for KEYNOTE-789?

KEYNOTE-789 is a well powered large, randomized phase 3 study. Patients must have [had] prior EGFR TKI treatment, RECIST 1.1–defined progression, common EGFR mutations, and good performance status. They were stratified according to PD-L1 expression (more than 50% [and] less than 50%), by region (East Asian vs non-East Asian), and by prior treatment (osimertinib [Tagrisso] vs no osimertinib).

Patients were randomly assigned 1:1 to [either the standard treatment arm or to] pembrolizumab plus pemetrexed with carboplatin or cisplatin for 4 cycles, followed by maintenance pemetrexed. Pembrolizumab was given for a total of 2 years. The standard treatment arm was the same chemotherapy agents using placebo to replace pembrolizumab and those who were randomly assigned to the standard treatment arm had the chance to cross over and use pembrolizumab for 2 years.

What efficacy findings from the study were presented at ASCO?

Three interim analyses were designed—one 6 months after the last patient had been randomly assigned, the second after 16 months, and the third after 29 months.

The PFS end point was determined at the second interim analysis…the HR for PFS was 0.80 [with a] P value of .0122. The statistical boundary that we set was 0.0117 [for PFS], so it just missed the end point.

[Median] OS was analyzed at the final analysis which had a median follow-up of 42 months…this is a very mature analysis. The HR was 0.84 and the P value was equal to 0.0364 [which] again missed the end point. This study was considered negative because both the PFS and OS end points were missed, yet both PFS and OS [data] were better for the pembrolizumab plus chemotherapy arm.

What does the safety profile of this combination look like?

Adding pembrolizumab to combination chemotherapy has been used widely, [such as] in the KEYNOTE-189 study [NCT02578680]. This is the same regimen, and we already know that there were no added toxicities to that, so the safety profile [in this trial] was everything that we have known. Very few patients discontinued due to toxicity and there was a higher incidence of immune-related adverse effects [AEs] where approximately 5% [of patients experienced] hyperthyroidism and 5% hypothyroidism. A few patients had very rare immune AEs such as myelitis, myocarditis, and hypophysitis, but that [was in] only 1 patient [each]. The AEs were what we expected and were quite manageable.

Even though this was considered a negative study, what are some of the future implications of this research?

We did see some difference in OS, and this is very uncommon in patients who have EGFR mutations. In the final analysis we analyzed the subsets that may affect the efficacy of pembrolizumab. One of the biomarkers that we focused on is PD-L1 expression. Patients who had more than 50% PD-L1 expression had a HR of 0.84, those who had 1% to 49% expression had a HR of 0.76, and those with absolutely no PD-L1 expression had a HR of 0.91.

We [generated a] Kaplan-Meier curve for patients who had PD-L1 expression more than 1% vs those who had no expression at all. For those who had no expression, the HR was 0.91 and the Kaplan-Meier curve completely overlapped. For those who had PD-L1 [expression of] 1% or more, the HR was 0.77 and the Kaplan-Meier curve overlapped in the first 4 to 5 months, but then separated until 3 to 4 years. This might be the right population to [treat with this combination] but in this study we have no power to detect the difference. Future studies in this population are warranted.

Were there any data presented at ASCO that caught your eye?

We see many new compounds coming—many of them are similar compounds to those that we have, for example, exon 20 insertion inhibitors, other new ICIs, [and] many of those for the targeted therapies have more known positions [in the treatment landscape because] many have shown early activity. For ICIs, unfortunately, many have not been able to demonstrate good activity [alone], but in the future, adding some combinations together perhaps will change our cancer treatment field.

Perioperative immunotherapy plus chemotherapy is a trend. There have been several studies showing that adding immunotherapy to chemotherapy in early-stage NSCLC can improve pathological complete responses [as well as] major pathological responses and may eventually improve event-free survival. This is something that we are following, and hopefully in the future will become SOC and cure more patients with early-stage NSCLC.

The largest [presentation at ASCO is] the use of osimertinib in the adjuvant treatment setting in the ADAURA study [NCT02511106]. In the ADAURA study, patients with early-stage NSCLC cancer after surgery were randomized to osimertinib for 3 years vs placebo. The PFS time was much better, but this year a plenary session demonstrated that OS was also superior in the treatment arm. Patients had stage IB-IIIA disease [but] even after recurrence still had some salvage treatment [options such as] an EGFR TKI [available which would] probably be very effective. The follow-up was still considered short…but seeing the difference in OS certainly has already made this treatment a SOC.


Yang JCH, Lee DH, Lee JS, et al. Pemetrexed and platinum with or without pembrolizumab for tyrosine kinase inhibitor (TKI)-resistant, EGFR-mutant, metastatic nonsquamous NSCLC: phase 3 KEYNOTE-789 study. J Clin Oncol. 2023; 41(suppl 17):LBA9000. doi:10.1200/JCO.2023.41.17_suppl.LBA9000

Related Videos
A panel of 5 experts on lung cancer seated at a long table
Joshua K. Sabari, MD, and Marina Chiara Garassino, MD, experts on non-small cell lung cancer
A panel of 5 experts on lung cancer
A panel of 5 experts on lung cancer
Catherine Ann Shu, MD
Christine Bestvina, MD
Charu Aggarwal, MD, MPH, and Melina E. Marmarelis, MD, experts on lung cancer
Charu Aggarwal, MD, MPH, and Melina E. Marmarelis, MD, experts on lung cancer
Lyudmila A. Bazhenova, MD
Benjamin Philip Levy, MD