Refining maintenance strategies for metastatic nonsquamous non-small cell lung cancer: insights from the KEYLYNK-006 study

Introduction

The phase 3 KEYLYNK-006 study, recently published in the Journal of Thoracic Oncology (JTO), represents a significant exploration into the evolving landscape of maintenance therapy for metastatic nonsquamous non-small cell lung cancer (NSCLC). This randomized, open-label trial compared pembrolizumab plus olaparib with pembrolizumab plus pemetrexed as maintenance strategies in patients with previously untreated, metastatic nonsquamous NSCLC without actionable genetic alterations, who achieved complete response (CR), partial response (PR), or stable disease (SD) after induction therapy with pembrolizumab, pemetrexed, and platinum. Despite the promise of poly (adenosine diphosphate-ribose) polymerase inhibitors (PARPi) like olaparib in enhancing immune responses, the trial did not demonstrate a survival benefit for the combination of pembrolizumab and olaparib compared to the established pembrolizumab plus pemetrexed regimen (1).

Study rationale and design

The KEYLYNK-006 study (NCT03976323) aimed to test the hypothesis that combining pembrolizumab, a programmed death protein 1 (PD-1) inhibitor, with the PARPi olaparib could synergistically improve progression-free survival (PFS) and overall survival (OS) by enhancing antitumor immunity through increased programmed death ligand 1 protein (PD-L1) expression and modulation of the tumor microenvironment (TME). Preclinical studies have suggested that PARPi, including olaparib, can potentiate the effects of immune checkpoint inhibitors (ICIs) by inducing DNA damage, activating the stimulator of interferon genes (STING) pathway (2), and promoting interferon signaling (3), all of which may enhance immune recognition and tumor clearance. Additionally, PARPi have been shown to increase PD-L1 expression on tumor cells, thereby sensitizing them to immune attack (4). However, translating these preclinical observations into clinical benefit has proven challenging, as demonstrated by the KEYLYNK-006 trial.

This randomized, open-label phase 3 trial enrolled 1,003 patients with stage IV nonsquamous NSCLC lacking actionable alterations in the epidermal growth factor receptor (EGFR), anaplastic lymphoma kinase (ALK), or c-ros oncogene 1 (ROS1) gene and without prior systemic therapy for advanced or metastatic disease (Figure 1). However, patients who had completed adjuvant or neoadjuvant therapy at least 12 months before the onset of metastatic disease were eligible. Key exclusion criteria included active or unstable central nervous system metastases, pneumonitis, interstitial lung disease, symptomatic ascites or pleural effusion, active infection, and immunodeficiency. Patients received a 12-week induction phase consisting of four cycles (every three weeks) of pembrolizumab (200 mg), pemetrexed (500 mg/m²), and platinum-based chemotherapy (either carboplatin area under curve 5 or cisplatin 75 mg/m²). Of these, 672 patients achieving CR, PR, or SD after induction were randomized 1:1 to either pembrolizumab (200 mg every three weeks) plus olaparib (300 mg orally twice daily) (n=337) or pembrolizumab (200 mg every three weeks) plus pemetrexed (500 mg/m² every three weeks) (n=335). Maintenance therapy continued until disease progression per Response Evaluation Criteria in Solid Tumors (RECIST) 1.1, unacceptable toxicity, withdrawal of consent, prolonged treatment interruption, or completion of 31 cycles of pembrolizumab. Randomization was stratified by Eastern Cooperative Oncology Group (ECOG) performance status (0 vs. 1), PD-L1 tumor proportion score (TPS) (<50% vs. ≥50%), and response at randomization (CR/PR vs. SD). The dual primary endpoints were PFS per blinded independent central review (BICR) and OS. Secondary endpoints included safety and patient-reported outcomes (PROs), while exploratory endpoints included objective response rate (ORR) and duration of response (DOR). PD-L1 expression was centrally assessed from non-irradiated tumor samples. Tumor imaging was performed at baseline, at weeks 6 and 12 during induction, every six weeks for the first 60 weeks after randomization, and every nine weeks thereafter, with responses evaluated per RECIST 1.1 by BICR. Adverse events (AEs) were monitored from the start of induction until 30 days after treatment cessation. PROs were collected electronically at baseline, before each maintenance phase dosing, at the end of treatment, and at the 30-day safety follow-up using the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire C30 (EORTC QLQ-C30), Quality of Life Questionnaire Lung Cancer Module 13 (QLQ-LC13), and European Quality of Life 5 Dimensions, 5 Level version (EQ-5D-5L) questionnaires.

Figure 1.

Study design of the randomized phase 3 KEYLYNK-006 trial. Patients with stage IV nonsquamous NSCLC received induction therapy with pembrolizumab, pemetrexed, and platinum chemotherapy (n=1,003). Those with CR, PR, or SD (n=672) were randomized 1:1 to maintenance pembrolizumab plus olaparib (Arm A) or pemetrexed (Arm B) for up to 31 cycles. Dual primary endpoints were PFS and OS. ALK, anaplastic lymphoma kinase; AUC, area under curve; BICR, blinded independent central review; BID, bis in die; CR, complete response; DOR, duration of response; ECOG, Eastern Cooperative Oncology Group; EGFR, epidermal growth factor receptor; IV, intravenous; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; PO, per os; PR, partial response; Q3W, quaque 3 weeks; QoL, quality of life; RECIST, Response Evaluation Criteria in Solid Tumors; ROS1, c-ros oncogene 1; SD, stable disease.

Key findings and clinical implications

The final analysis, conducted after a median follow-up of 39.9 months, showed that pembrolizumab plus olaparib did not significantly improve PFS or OS compared to pembrolizumab plus pemetrexed. The median PFS was 7.1 months in the pembrolizumab plus olaparib group vs. 8.3 months in the pembrolizumab plus pemetrexed group [hazard ratio (HR) 1.12, 95% confidence interval (CI): 0.92–1.36], while the median OS was 20.7 vs. 23.0 months, respectively (HR 1.04, 95% CI: 0.87–1.25), indicating no substantial survival benefit with the addition of olaparib. Patient demographics and baseline disease characteristics were generally balanced between the treatment arms, with similar PFS and OS outcomes across key subgroups. The ORR was 16.6% (95% CI: 12.8–21.0%) in the pembrolizumab plus olaparib group, compared to 20.0% (95% CI: 15.8–24.7%) in the pembrolizumab plus pemetrexed group. Among patients with a confirmed response, the median DOR was 20.4 months (range, 2.6–47.7 months) and 18.8 months (range, 1.4–38.7 months) in the olaparib and pemetrexed arms, respectively. The safety profile was consistent with prior studies, with grade 3–5 treatment-related AEs occurring in 26.1% of patients in the pembrolizumab plus olaparib group and 30.1% in the pembrolizumab plus pemetrexed group. Notably, both regimens were associated with significant hematologic toxicities, including anemia (13.6% vs. 10.2%), neutropenia (3.0% vs. 3.0%), and thrombocytopenia (1.5% vs. 3.3%), highlighting the need for careful patient selection and monitoring. The rate of treatment-related AEs leading to discontinuation was higher in the pemetrexed group (22.6% vs. 13.9%), likely reflecting the known toxicity profile of chemotherapy. PROs showed high rates of completion and compliance at baseline and at week 24 in both groups, with generally similar results between treatment arms.

Challenges and future directions

The disappointing outcomes of the KEYLYNK-006 study align with the findings from the KEYLYNK-008 trial in squamous NSCLC (5), highlighting the ongoing challenges of integrating PARPi into the NSCLC treatment landscape (Table 1). Despite compelling preclinical evidence suggesting that PARPi can enhance immune responses and potentially synergize with ICI, clinical trials have yet to demonstrate a clear survival benefit in unselected NSCLC populations. This discrepancy may be partly explained by the inherent heterogeneity of NSCLC, the variability of the TME, and the absence of reliable biomarkers to identify patients likely to benefit from PARPi-ICI combinations. The open-label design of the study, required due to the differing routes of administration (oral vs. intravenous), and the lack of biomarker stratification [e.g., tumor mutational burden (TMB), homologous recombination deficiency (HRD), breast cancer 1 and 2 genes (BRCA1/2) mutations] further limit the interpretation of these findings, as does the exclusion of patients with primary progressive disease, potentially restricting the applicability of the results to this subgroup.

Table 1. Summary of selected clinical trials evaluating PARPi-based strategies in lung cancer.

Trial	Population	Treatment arms	Primary endpoints	Key findings
KEYLYNK-006 (1) (phase 3)	Stage IV nonsquamous NSCLC	Maintenance after induction:	PFS, OS	No survival benefit with olaparib vs. pemetrexed
		Arm A: pembrolizumab + olaparib
		Arm B: pembrolizumab + pemetrexed
KEYLYNK-008 (5) (phase 3)	Stage IV squamous NSCLC	Maintenance after induction:	PFS, OS	No survival benefit with olaparib
		Arm A: pembrolizumab + olaparib
		Arm B: pembrolizumab alone
		Arm C: observation
KEYLYNK-012 (6) (phase 3)	Unresectable stage III NSCLC	Experimental arm: pembrolizumab + cCRT → pembrolizumab ± olaparib	PFS, OS	Ongoing
		Comparator Arm: cCRT → durvalumab
KEYLYNK-013 (7) (phase 3)	Limited-stage SCLC	Experimental arm: pembrolizumab + cCRT → pembrolizumab ± olaparib	PFS, OS	Ongoing
		Comparator arm: cCRT alone
GOAL (8) (phase 1B/2)	EGFR-mutant NSCLC	Gefitinib ± olaparib	PFS	Improved PFS for Olaparib in patients with high BRCA1 mRNA expression (12.9 vs. 9.2 months)
HUDSON (9) (phase 2)	Pretreated NSCLC with HRR gene alterations	Olaparib ± durvalumab	ORR	Modest clinical activity (ORR 9.5%) in patients with HRR gene alterations (e.g., BRCA1/2, ATM)
JASPER (10) (phase 2)	Stage IV NSCLC, treatment-naive	Niraparib ± pembrolizumab	ORR	Promising ORR (~56%) suggest potential synergy between PARPi and ICI in PD-L1-high tumors
		Cohort 1: PD-L1 TPS ≥50%
		Cohort 2: PD-L1 TPS 1–49%
PRIO (NCT04728230) (phase 1/2)	Extensive-stage SCLC	Olaparib + durvalumab + chemotherapy and/or radiotherapy	DLT	Ongoing

This table summarizes key trials investigating PARPi as monotherapy or in combination with immune checkpoint inhibitors, chemotherapy, or targeted therapies in lung cancer. Included are pivotal randomized phase 3 trials (e.g., KEYLYNK series), biomarker-driven studies (e.g., HUDSON), and early-phase exploratory trials (e.g., JASPER, GOAL, PRIO). The table highlights study populations, treatment regimens, and major outcomes to contextualize the role of PARPi across different clinical and molecular lung cancer subtypes. ATM, ataxia telangiectasia mutated; BRCA1/2, breast cancer gene 1/2; cCRT, concurrent chemoradiotherapy; DLT, dose-limiting toxicity; EGFR, epidermal growth factor receptor; HRR, homologous recombination repair; ICI, immune checkpoint inhibitor or immunotherapy; mRNA, messenger RNA; NSCLC, non-small cell lung cancer; ORR, objective response rate; OS, overall survival; PARPi, poly (adenosine diphosphate-ribose) polymerase inhibitors; PD-L1, programmed death ligand 1 protein; PFS, progression-free survival; SCLC, small cell lung cancer; TPS, tumor proportion score.

Moreover, this study was designed with an ambitious HR of 0.65 to demonstrate superiority, which now appears overly optimistic given meta-analytic data suggesting more modest effect sizes, including an HR of 0.85 for OS and 0.93 for PFS for the PARPi arm versus control (11). This disconnect raises the question of whether a non-inferiority design might have been more appropriate, particularly given the established role of pemetrexed as a potent maintenance therapy in NSCLC (12). Additionally, while preclinical studies have suggested that alterations in tumor suppressor genes such as serine/threonine kinase 11 (STK11) may confer enhanced sensitivity to PARPi (13,14), the clinical reality appears more complex. In the recent phase 2 umbrella HUDSON study by Besse et al. (2024), the ORR with durvalumab-olaparib was only 4.6%, including 9.5% in the homologous recombination repair-mutated (HRRm) and 4.8% in the STK11 biomarker-matched cohorts, and 0% in primary resistance cohorts (9). These results underscore the limited clinical utility of targeting these pathways, despite promising preclinical data.

Despite overall negative results, PARPi-based combination therapies may still represent a promising option for selected subgroups of NSCLC patients. Recent meta-analyses have demonstrated a significant impact on OS rather than PFS, suggesting that the benefits of PARPi may be more long-term and attributable to complex biological interactions within the TME and specific genomic features-distinct from the immediate cytotoxic effects of chemotherapy on cell proliferation and viability (11,15). As such, PARPi-based regimens could be particularly suitable for patients with indolent or SD following induction chemotherapy. Additionally, combinations of PARPi with EGFR tyrosine kinase inhibitors (TKIs) have been associated with improved survival outcomes. Notably, high BRCA1 messenger RNA (mRNA) expression identified a subgroup of NSCLC patients who benefited from the combination of gefitinib and olaparib in the phase 2 GOAL trial (16). Furthermore, PARPi has shown potential as a radiosensitizer. In EGFR-mutant NSCLC models, PARPi not only enhanced the effects of radiation in suppressing tumor growth but also promoted anti-tumor immune responses when used in combination with radiotherapy (17). Taken together, PARPi may provide meaningful clinical benefit in biologically distinct NSCLC subtypes when used in rationally designed combination strategies.

To overcome these challenges, future clinical trials should implement more refined patient selection strategies, ideally incorporating comprehensive genomic and transcriptomic profiling to identify individuals with deficiencies in the DNA damage repair (DDR) pathway, such as BRCA1/2 mutations, HRD, or high TMB, as well as alterations in STK11, Kelch-like ECH-associated protein 1 (KEAP1), or AT-rich interactive domain-containing protein 1A (ARID1A) genes. These molecular features may help pinpoint patients who are more likely to benefit from PARPi-based combination therapies. Additionally, retrospective analyses of these biomarkers could further support the identification of responsive subgroups and guide future treatment decisions. In parallel, the exploration of novel therapeutic strategies, such as combining PARPi with other immunomodulatory agents, DNA damage response enhancers, or targeted therapies, may offer a more promising and personalized approach (18,19). Nonetheless, it is important to recognize that the KEYLYNK-006 study design may have primarily captured the efficacy of pemetrexed, a well-established agent in NSCLC maintenance therapy, potentially confounding the interpretation of olaparib’s true impact in this setting. Ongoing phase 3 trials using olaparib in patients with stage III NSCLC (KEYLYNK-012, NCT04380636) (6) or limited-stage SCLC (KEYLYNK-013, NCT04624204) (7) will hopefully add further information to the real efficacy of PARPi in the setting of lung cancer (Table 1).

Conclusions

While the KEYLYNK-006 study did not meet its primary endpoints, it provides valuable insights into the complexities of combining targeted and immune-based therapies in NSCLC. Moving forward, a more personalized approach, integrating comprehensive genomic and immune profiling, may be necessary to fully unlock the potential of PARPi in this challenging patient population. The ongoing KEYLYNK-012 and KEYLYNK-013 studies may further elucidate the role of these combinations in different lung cancer subgroups, potentially reshaping the landscape of precision oncology for lung cancer.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.