Monitoring pembrolizumab response in patients with metastatic non-small cell lung cancer using circulating tumour DNA and circulating tumour cells

Introduction

Non-small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancer cases with most patients diagnosed with or developing metastatic disease (1). Prior to immunotherapy, patients with metastatic NSCLC had a poor prognosis, with only approximately 3% surviving 5 years (2). Palliative chemotherapy, radiotherapy, and targeted therapy were used for patients with advanced NSCLC (aNSCLC), with a modest prolongation of survival. In recent years, immune checkpoint inhibitors (ICIs) targeting molecules such as programmed cell death protein-1 (PD-1) or programmed death ligand-1 (PD-L1), provide more durable responses than traditional chemotherapy. ICIs can result in long survival in a minority of patients with 5-year survival rates of up to 19% for unselected patients, and up to 29% in patients with high (>50%) PD-L1-expression on tumour biopsy (3-5). PD-1 is a T-cell surface inhibitory receptor involved in the regulation of immune responses. The interaction of PD-1 with PD-L1, which is overexpressed on the surface of tumour cells, can lead to T-cell exhaustion and inhibition of the immune response (6).

Pembrolizumab is approved for metastatic NSCLC patients with PD-L1 expression with a ≥50% tumour-proportion score and no EGFR or ALK mutations, or in combination with chemotherapy regardless of PD-L1-expression in both the US and EU (7). Although pembrolizumab seems to benefit a proportion of patients, most fail to respond, suggesting that intrinsic or acquired resistance to treatment may be involved. PD-L1-expression in tumour biopsy by immunohistochemistry (IHC) has emerged as a biomarker that can predict which patients are more likely to respond to ICIs against PD-1/PD-L1. However, PD-L1 IHC has limitations, given that a minority of patients with PD-L1-negative tumours can still respond, whilst less than 50% of patients with high PD-L1 expression do so, indicating an urgent need for more reliable biomarkers (8). Additional biomarkers, such as tumour mutational burden (TMB), have been proposed to predict the efficacy of PD-1 inhibitors or other ICIs but have not yet been standardised. Recently, liquid biopsies (LB), including cell-free DNA (cfDNA), circulating tumour DNA (ctDNA), and circulating tumour cells (CTCs), have provided clinically valuable information for NSCLC (9-12). ctDNA harbours similar mutations to those in DNA extracted from solid tumour cells. Furthermore, ctDNA detection techniques are quantitative and changes in ctDNA-levels during treatment have been associated with tumour response or progression in several tumour types, including lung cancer (13-16). CTCs are released from various locations of primary or metastatic tumours into the blood circulation, potentially providing similar or sometimes even superior information compared with conventional tumour biopsies (17).

In contrast to solid tumour biopsies, LB is less invasive and captures more of the heterogeneity of tumours, allowing the detection of genetic alterations responsible for treatment resistance, which can be missed by tissue genotyping. Accordingly, a recent study in aNSCLC patients showed that ctDNA-profiling can capture heterogeneous oncogenic drivers that were not identified by tissue DNA sequencing (18). Given the growing number of targeted therapies for NSCLC (19), recent evidence has demonstrated that ctDNA-guided therapy in patients with advanced disease provides a longer overall survival (OS) benefit, revealing the clinical utility of ctDNA-matched targeted treatment in aNSCLC (18). Importantly, in aNSCLC, LB appears to be the only pragmatic option for real-time monitoring of response during treatment.

To define the usefulness of LB in predicting response to immunotherapy, we evaluated plasma-derived cfDNA levels of aNSCLC patients treated with pembrolizumab before and at early timepoints of treatment. Next, we performed molecular tag-based next-generation sequencing (NGS) of ctDNA and measured the allele-frequency of somatic mutations with great precision. Moreover, immunofluorescence-microscopy data evaluating PD-L1, and Ki67-expression in CTCs were generated at each timepoint studied. Results were then overlayed into a statistical data model aimed at investigating the predictive value of the three biomarkers, either separately or in combination, to enforce their predictive role on the response of NSCLC patients to pembrolizumab. This manuscript is written following REMARK reporting checklist (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2024-1095/rc).

Methods

We investigated the predictive value of LB by considering cfDNA, ctDNA somatic mutation frequencies, and CTC PD-L1 and Ki67 expression profiles in serial blood samples obtained by aNSCLC patients treated with the anti-PD-1 inhibitor pembrolizumab within a prospective switch maintenance pembrolizumab trial (NCT02705820). A schematic of the proposed methodology is shown in Figure 1.

Figure 1 Flowchart of the study design. cfDNA, cell-free DNA; CTCs, circulating tumour cells; ctDNA, circulating tumour DNA; NSCLC, non-small cell lung cancer; PD-L1, programmed death-ligand 1.

Study population

Forty-eight patients with aNSCLC were treated with pembrolizumab as part of the prospective SWIPE trial (NCT02705820). Of the original cohort, 46 patients were included in this study. Two patients were excluded from the analysis: one was excluded due to the absence of follow-up data beyond 6 weeks, as she passed away from septic shock not related to the study medication; the other was excluded due to disease progression prior to evaluating the first cycle of treatment. Plasma and peripheral blood mononuclear cell (PBMC) samples were collected at baseline (t0), 3 weeks (t1), 6 weeks (t2) and 9 weeks (t3) post-treatment (Suppl. Methods). Patient clinical outcomes were measured as progression-free survival (PFS) and OS. The treatment response was initially evaluated using RECIST v1.1 (20). Patients were also classified as having durable clinical benefit (DCB) if they presented with PFS ≥6 months or non-durable benefit (NDB) if they presented with PFS <6 months. Additionally, a healthy individual with no known medical conditions or history of illness was included as a control.

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Cyprus National Bioethics Committee (No. 2019/76) and informed consent was taken from all the patients.

Isolation of plasma and PBMCs and CTCs characterization

Plasma and PBMCs were isolated from 20 mL of whole blood and CTCs were phenotypically characterised based on cytokeratin, PD-L1 and Ki67, as previously described (21) (Appendix 1).

Isolation and quantification of cfDNA

Total cfDNA was extracted from 1–3 mL of plasma collected from 127 samples using the Norgen Plasma/Serum Cell-Free Circulating DNA Purification Midi Kit (Norgen Biotek Corp., ON, Canada) according to the manufacturer’s protocol. cfDNA was eluted in 50 µL nuclease-free water and quantified using a Qubit dsDNA HS-Assay Kit on a Qubit fluorometer 2.0 (Life Technologies, CA, USA).

Screening of cfDNA for somatic mutations with NGS

The Oncomine™ Lung cfDNA Assay (Thermo Fisher Scientific, CA, USA) was used as a selection method to perform NGS in 1–50 ng of total cfDNA isolated at t0 from 43 patients, at t1 from 18 patients, at t2 from 31 patients, and at t3 from 11 patients (n=103), according to the manufacturer’s instructions (Appendix 1). The amplified libraries were quantified using the Ion Library TaqMan Quantitation Kit (Thermo Fisher Scientific, CA, USA). The performance of the assay was initially evaluated using the Multiplex I cfDNA Reference Standard Set (Horizon Discovery, WBE, UK; Appendix 1, Table S1). Sequencing was performed on the Ion GeneStudio S5 platform (Thermo Fisher Scientific, CA, USA), and raw-data were analysed using Torrent Suite Software (v5.18) and Ion Reporter Software v5.12 (Thermo Fisher Scientific, CA, USA). The variant allele frequency (VAF), defined as the percentage of somatic mutation load over wild-type, and limit of detection (LOD) were calculated.

Statistical analysis

The Wilcoxon signed-rank test was employed to compare cfDNA, ctDNA, the number of CTCs, and the percentage of distinct CTC-phenotypes within patient groups (DCB and NDB) across-timepoints. The Mann-Whitney U test was used to compare these markers between patient groups within-timepoint. Fisher’s exact test assessed the presence of CTCs, the presence and total number of mutations between groups at each timepoint. Statistical tests were performed in R (RRID:SCR_001905).

Multivariate Cox-proportional hazards models were fitted to identify the prognostic ability of the three LB markers for PFS and OS while adjusting for age and sex. The results were presented as hazard ratios (HRs) and 95% confidence intervals (CIs). Kaplan-Meier PFS and OS curves were constructed, and significance was evaluated using the log-rank test (22,23).

Results

Clinicopathological characteristics of patients

Forty-six patients were included in the study. The median age was 66 years (range, 40–82 years), 38 patients were males (82.6%) and 8 were females (17.4%). Thirty-three patients had adenocarcinoma (71.7%), while 13 patients had squamous cell carcinoma (28.3%). The median PFS and OS were 3.95 months (range, 1.0–36.3 months) and 13.70 months (range, 1.7–48.5 months), respectively. Additional clinicopathological characteristics are summarised in Table S2.

cfDNA dynamics in patients with advanced NSCLC during treatment with pembrolizumab

To define the most suitable timepoint to measure changes in the amount of cfDNA, we initially selected 13 patients: seven NDB patients with rapid progression disease (PD) on the first computed tomography (CT) evaluation (median PFS: 2 months) and six patients with DCB (median PFS: 18.1 months). NDB patients showed a significant increase in cfDNA from t0 to t1 (P=0.03) and t2 (P=0.01; Figure S1A,S1B), whereas DCB patients showed no significant cfDNA changes at these timepoints (Figure S1A,S1C). A significant difference in cfDNA levels between NDB and DCB was observed at t2 only (P=0.01; Figure S1D).

Subsequently, we evaluated cfDNA-levels in 46 patients at t0, t1, t2 and t3 (28 NDB and 18 DCB). The amount of isolated cfDNA is presented in Table S3 along with the median values for each group and timepoint. NDB patients had higher cfDNA levels at t2 (P=0.06) and t3 (P=0.044; Figure 2A,2B). Although significance was observed at t3 in NDB, it was not chosen as the primary timepoint for downstream analyses as several patients had already progressed to PD by that time. On the contrary, in DCB patients, cfDNA-levels significantly decreased at t2 (P=0.03; Figure 2A,2C). Across-group comparison showed significantly higher levels of cfDNA in NDB (P=0.005; Figure 2D) at t2 only.

Figure 2 Predictive relevance according to the changes in the amount of cfDNA during treatment. (A) Line graph shows the changes in cfDNA from t0 to t3 for each patient individually. Patients of the DCB group are designated in light blue and patients of the NDB group in red. (B,C) Changes in cfDNA from t0 to early treatment timepoints (t1, t2 and t3) were evaluated for each patient group separately (NDB =28, DCB =18). The results showed an increase of cfDNA at t2 (P=0.06) and t3 (P=0.044) of treatment compared to t0 in NDB patients, whereas DCB patients showed a significant decrease in the amount of cfDNA at t2 (P=0.03). (D) Within-timepoint comparison of the amount of cfDNA between DCB and NDB showed a significant difference in cfDNA between DCB (n=12) and NDB (n=23) patients at t2 only (P=0.005). (E) Barplots indicate the percentage of patients that showed durable (DCB, n=12), or non-durable clinical benefit (NDB, n=23), according to their cfDNA changes. A greater percentage of patients with an increase in the amount of cfDNA from t0 to t2 had NDB (79.2%) compared to DCB (20.8%). For patients who presented a decrease in cfDNA from t0 to t2, 59.1% had DCB compared to 40.9% who had NDB. (F,G) Patients with increase in cfDNA from t0 to t2 showed shorter PFS and OS compared to those with a decrease in cfDNA (2.05 vs. 6.1 months, P=0.04 and, 8.35 vs. 20.0 months, P=0.004, respectively). *, P<0.05; **, P<0.01. t0, baseline; t1, 3 weeks post-treatment; t2, 6 weeks post-treatment; t3, 9 weeks post-treatment. cfDNA, cell-free DNA; DCB, durable clinical benefit; NDB, non-durable benefit; ns, not significant; OS, overall survival; PFS, progression-free survival.

We then investigated whether changes in cfDNA from t0 to t2 were associated with clinical outcomes. Interestingly, 19/24 (79.2%) of patients with cfDNA increase had NDB and 5/24 (20.8%) DCB (Figure 2E). However, among those who had a decrease in cfDNA, 9/22 (40.9%) experienced NDB and 13/22 (59.1%) DCB, suggesting that relying solely on the cfDNA-decrease does not provide definitive conclusions on a patient’s response. Patients with an increase in cfDNA from t0 to t2 presented worse PFS [hazard ratio (HR) =4.1, P=0.02] and OS (HR =4.1, P=0.002). Kaplan-Meier plots also confirmed a significant difference in PFS and OS between patients who showed an increase from t0 to t2, and patients with no increase (PFS: 2.05 vs. 6.1 months, P=0.04, Figure 2F; OS: 8.35 vs. 20.0 months, P=0.004, Figure 2G).

Detection of low copy somatic mutations in plasma ctDNA

Forty-three of the 46 patients and one healthy individual, used as control (CTRL), were further analysed for somatic mutations by NGS at baseline and early during treatment (t1 and/or t2 and/or t3). Three patients were excluded due to inadequate cfDNA-levels (IDs: 2, 24, 33). Mutations were identified in 25/43 (58.14%) patients at any of the four timepoints analysed. Of the 74 somatic mutations identified, 36 (48.65%) were on TP53, 19 (25.68%) on KRAS, 16 (21.62%) on EGFR, 1 (1.35%) on ALK, 1 (1.35%) on PI3KCA and 1 (1.35%) on MAP2K1 (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-1.xlsx). Detailed NGS results are showed in table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-1.xlsx and summaries of the median read and median molecular coverage for each sample and timepoint analysed in NGS in table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-2.xlsx. Molecular analysis was performed on available tissue samples prior to treatment initiation with pembrolizumab, and the detailed results of these tests can be found in Table S4. Unfortunately, due to the metastatic nature of the cases, suitable tissue samples for NGS were not available.

We then examined whether the presence or number of somatic mutations was associated with response or resistance to pembrolizumab. Our results showed no significant association with neither PFS nor OS (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx, Table S5). Subsequently, we investigated the presence and number of specific mutations (TP53, EGFR, KRAS). We found that patients harbouring KRAS mutations (KRASmut) at t2 but not t0 or t1, displayed a 2.96 worse OS than those without KRASmut (HR =2.96, P=0.03; table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx, Table S5). Kaplan-Meier analysis also confirmed a significant difference in OS, between patients with KRASmut at t2 and those without (8.8 vs. 32.2 months, P=0.02; Figure S2A,S2B). In addition, the number of KRASmut at t2 negatively affected OS, with the same calculated HR at 2.96 (P=0.03; table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx). The presence of KRAS G12/G13 mutations was also associated to a 3.2 times higher mortality risk (HR =3.2, P=0.03) compared to those with no such mutations (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx).

Monitoring of the ctDNA somatic mutation frequency dynamics of NSCLC patients during treatment

ctDNA somatic mutation frequency dynamics were investigated as changes in the maximum VAF (%) over the course of treatment, to assess whether ctDNA mutations could be used as a potential pharmacodynamic marker of resistance or response to pembrolizumab (Figure 3A). Patients who presented a greater than 50% (>50%) decrease or clearance of ctDNA [ctDNA-molecular response (ctDNA-MR)] at an early phase during treatment presented longer PFS and OS, with a 7.2- and 3.34-times lower risk for progression (P=0.03) and better OS (P=0.03), respectively (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx). Kaplan-Meier plots also confirmed a significant difference in PFS (7.9 vs. 2 months, P=0.03) and OS (15.6 vs. 11.3 months, P=0.01), between patients with ctDNA-MR and patients without (<50%-decrease or increase in ctDNA) (Figure 3B,3C). Interestingly, evaluation of ctDNA-dynamics from baseline to only one timepoint early during treatment (t1 or t2, separately), failed to provide significant results (Figure S3), indicating the importance of monitoring ctDNA at additional sequential timepoints.

Figure 3 Dynamics of ctDNA and clinical outcomes. (A) Swimmer’s plot illustrating the detection of ctDNA-derived somatic mutations for each patient at the four timepoints analysed: t0, t1, t2, and t3 post-treatment initiation. The size of each circle corresponds to the mutation load, represented as the VAF (%). The length of the black swimmers’ bars indicate the OS for each patient. Kaplan-Meier plots show the difference in (B) PFS and (C) OS between patients presenting >50% decrease or clearance in ctDNA at an early phase during treatment (N=12), compared to those who showed <50% decrease or an increase in ctDNA (N=13). t0, baseline; t1, 3 weeks post-treatment; t2, 6 weeks post-treatment; t3, 9 weeks post-treatment. ctDNA, circulating tumour DNA; OS, overall survival; PD, progression disease; PFS, progression-free survival; PR, partial response; SD, stable disease; VAF, variant allele frequency.

Monitoring CTC status according to PD-L1 and Ki67

CTC-status [total number of CTCs, PD-L1 positive (PD-L1+) and Ki67-positive (Ki67+) CTCs] was investigated in 46 patients at t0 and 33 at t2 (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-4.xlsx). CTCs were detected in 43/46 (93.5%) patients at t0 and 21/33 (63.6%) at t2. PD-L1+ CTCs were identified in 37/43 (86%) patients at t0 and 18/21 (85.7%) at t2, with no significant differences between NDB and DCB patients. For Ki67, NDB patients more frequently presented Ki67+ CTCs than DCB patients: 20/27 (74%) vs. 4/16 (25%) at t0 (P=0.004), and 11/15 (73%) NDB patients vs. 2/6 (33%) DCB at t2 (P=0.15; Figure 4A). Within-timepoint comparison also showed that the percentage of Ki67+ CTCs was significantly higher in NDB than in DCB patients at both timepoints (median at t0: 33.3% vs. 0%, P=0.01, median at t2: 17.35% vs. 0%, P=0.03; Figure 4B).

Figure 4 Prognostic relevance of Ki67 positive (Ki67+) CTCs. (A) Percentage (%) of patients presenting Ki67+ CTCs before treatment initiation (t0: NDB, n=28 and DCB, n=18, 74% vs. 25%, P=0.004) and after the 2^nd-cycle (t2: NDB, n=21 and DCB, n=12, 73% vs. 33%, P=0.15). (B) The percentage of Ki67+ CTCs in NDB and DCB patients at t0 and t2. Each dot displays the percentage of Ki67+ CTCs per patient in NDB (red dots) and DCB patients (blue dots). The bold horizontal line in each plot indicates the median value, the lower and upper horizontal lines represent the 25^th and 75^th percentiles of the data. The median percentage of Ki67+ CTCs was significantly higher in NDB compared to DCB at both timepoints t0 and t2 (t0: 33.3% vs. 0%, P=0.01 and t2: 17.4% vs. 0%, P=0.02). (C,D) Prognostic relevance according to the percentage of Ki67+ CTCs in PD-L1+ patients at t2 of pembrolizumab treatment. Patients presenting a high Ki67 CTC index (>30% Ki67+ CTCs) at t2 had a significantly shorter PFS (2 vs. 4.45 months, P=0.03), and OS (5 vs. 27.2 months, P=0.003) compared to those with low Ki67-index (<30% Ki67+ CTCs). *, P<0.05; **, P<0.01; ns, not significant. t0, baseline; t1, 3 weeks post-treatment; t2, 6 weeks post-treatment; t3, 9 weeks post-treatment. CTCs, circulating tumour cells; ctDNA, circulating tumour DNA; DCB, durable clinical benefit; NDB, non-durable benefit; OS, overall survival; PFS, progression-free survival; PD-L1+, programmed death-ligand 1 positive.

We also evaluated the Ki67-index (iKi67), as the percentage of Ki67+ CTCs over the total CTC number per patient. We previously showed that a high-iKi67 (>30% Ki67+ CTCs) in patients with metastatic NSCLC harbouring PD-L1+ CTCs prior to pembrolizumab, was associated with shorter PFS (21). In this study, we investigated whether CTC-status could have a predictive value at t2 in the same patient cohort. We found that patients presenting PD-L1+ CTCs and a high-iKi67 at t2 had shorter PFS (2 vs. 4.45 months, P=0.03 and HR =10.13, P=0.03) and OS (5 vs. 27.2 months, P=0.003; HR =6.1, P=0.01) than those showing a low iKi67 (<30% Ki67+ CTCs) (Figure 4C,4D, table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx).

Circulating DNA and tumour cells as a combinatorial prognostic biomarker

Expanding on our previous findings on the iKi67, we further examined whether the combination of cfDNA or ctDNA and the iKi67 could serve as an alternative monitoring biomarker of response to pembrolizumab. We first analysed the distribution of biomarker detection status across treatment timepoints (Figure S4), which illustrates the percentage of patients with detectable ctDNA, CTCs, both markers, or neither of the two markers at each timepoint analysed. We then explored how ctDNA dynamics and CTC high-iKi67 correlate with treatment responses. Our results demonstrate that patients with a high-iKi67 and/or an increase in ctDNA or a decrease of less than 50% at an early phase of treatment had significantly worse PFS and OS compared to patients not presenting these markers (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx). Kaplan-Meier analyses also showed that patients presenting either or both biomarkers had a shorter PFS and OS than those who were not (PFS: 2 vs. 9.9 months, P<0.001 for t0; 1.95 vs. 6 months, P=0.02 for t2, OS: 11.2 vs. 20.6 months, P=0.003 for t0; 6.3 vs. 41.7 months, P=0.002 for t2, Figure 5A-5D). For t1, the results were not significant (PFS: P=0.27; OS: P=0.90); however, they may have been affected by the small sample size, necessitating further investigation.

Figure 5 Evaluation of the predictive value of ctDNA and Ki67-index of CTCs. (A-D) Kaplan-Meier survival analysis indicated a significant difference in (A,B) PFS and (C,D) OS between patients, based on the two biomarkers: <50% decrease or increase in ctDNA and/or high Ki67-index (>30% Ki67+ CTCs). (E,F) Venn diagrams were constructed showing the number and percentage of PD patients detected by either biomarker or both for each timepoint analysed (0 week to t0: n=17, 6 weeks to t2: n=10). The purple circle indicates the percentage of SD or PR patients (t0: n=23, t2: n=9) who presented either one or both biomarkers and were therefore classified as PD, (false positives). t0, baseline; t2, 6 weeks post-treatment. CTCs, circulating tumour cells; ctDNA, circulating tumour DNA; OS, overall survival; PD, progression disease; PFS, progression-free survival; PR, partial response; SD, stable disease.

We then layered ctDNA-dynamics over the assessment of iKi67 to establish a sensitive and robust combinatorial risk classification approach for detecting patients with PD with greater accuracy. Patients who were not examined for CTCs or did not have detectable CTCs or were not evaluated in NGS were excluded (sample used: n=40 for t0, n=15 for t1, n=19 for t2). A high-iKi67 was detected in 12/17 (70.6%) and 5/10 (50.0%) of PD patients at t0 and t2, respectively. In stable disease (SD)/partial response (PR) patients, a high-iKi67 was detected in 4/23 (17.4%) and 1/9 (11.1%) at t0 and t2, respectively, confirming the high specificity of this marker in identifying patients with PD. Similarly, increase in ctDNA or decrease of less than 50% was detected in 9/17 (53%) and 4/10 (40%) of PD patients at t0 and t2, respectively, and 4/23 (17%) and 0/9 (0%) of SD/PR patients at t0 and t2, respectively. The sensitivity of detecting PD patients increased to 88.2% at t0 and 70% at t2, when both markers were combined, which was higher than the sensitivity of either biomarker alone at any timepoint (Figure 5E). Importantly, the two assays could be combined without substantially affecting the false-positive rate, which remained low at 30.4% (7/23) for t0 and 11.1% (1/9) for t2. The corresponding analysis for t1, provided 86.7% sensitivity and 37.5% specificity. However, the small sample size (PD: n=7, SD: n=8) may have affected the validity of these results.

The results of the corresponding analysis for cfDNA-dynamics and high-iKi67 are presented in Figure S5.

Discussion

In the present study, we investigated the predictive value of LB in the form of cfDNA, ctDNA, and CTCs in blood samples obtained during a clinical trial involving aNSCLC patients receiving pembrolizumab. Our cohort consisted of 46 patients, predominantly males (n=38, 82.6%) and fewer females (n=8, 17.4%). This gender disparity mirrors broader trends observed in NSCLC, where males typically have a higher incidence of the disease. To account for this imbalance, gender was included as a covariate in our multivariate analyses. Nevertheless, given that NSCLC’s biological and clinical manifestations can differ between genders, future studies are needed to explore these differences more comprehensively.

In NSCLC, cfDNA quantification has been proposed for monitoring and follow-up of patients undergoing radiation therapy and chemotherapy (24). However, few studies have investigated the role of cfDNA as a predictive factor of immunotherapy response (25,26). In aNSCLC patients treated with pembrolizumab as first-line therapy, a recent study showed that high baseline cfDNA or an increase in cfDNA-levels from baseline to 12-weeks post-treatment were associated with shorter PFS (27). In our study, we found that patients who showed an increase in cfDNA at t2 (6-week post-treatment) had worse PFS and OS compared to those who showed a decrease, suggesting that dynamic monitoring of cfDNA in aNSCLC patients treated with pembrolizumab as maintenance-treatment after chemotherapy could also be predictive of their outcome. Although the dynamics of cfDNA appear to be associated with the response of patients to treatment, it is noted that cfDNA can also include non-tumour DNA and its presence can be influenced by various factors in cancer patients, including inflammation, hydration, the timing of the blood draw and the chemotherapy treatment that patients received prior to pembrolizumab. Further prospective studies are needed to validate cfDNA as a biomarker, and to determine how early (e.g., 3- or 6- or 9-week post-treatment), it should be used to guide decision making.

Detection of genetic alterations in ctDNA has been used to guide treatment in aNSCLC patients (28). Several studies, including ours, have examined whether the detection and type of ctDNA somatic mutations could act as predictive biomarkers for immunotherapy response. There is evidence that some p53 mutations (p.G245S, p.G245D, p.Y220C) may be associated with immunotherapy response (29). Patients in our study with such mutations demonstrated PR, SD and longer PFS (IDs: 10, 25); while patient 46 despite carrying p.Y220C displayed NDB, likely due to co-occurring p53 mutations (p53mut). Six NDB patients (IDs: 8, 15, 18, 26, 36, 46) carried p53mut; the p.R248Q, p.R248W, p.R175H, p.R282W, p.R273H, p.R273C, p.V157F, p.R158L, p.H179R, p.Y205C and p.M237I, and a novel mutation p.R283H. For DCB, three patients (IDs: 38, 43, 47) carried the p.R282W, p.R248Q, p.R248W, and p.R273H substitutions (PFS: 14.5, 9.7 and 6.1 months, respectively), and all harboured co-occurring mutations, while patient 38 also presented a novel TP53 variant (c.375G>A), leading to a synonymous amino acid, p.T125.

Studies on the effect of KRASmut on immunotherapy outcomes remain conflicting. While some studies have shown no significant differences in outcomes based on KRAS-status (30,31), others suggested that specific mutations, like p.Q61K showed sensitivity to immunotherapy, whereas p.G12/G13 substitutions were linked to poorer outcomes (31-36). Our findings show that both the presence and the number of KRASmut in patients’ ctDNA at t2 were associated with poor response to immunotherapy (Figure S2A,S2B). The absence of detectable mutations at baseline could be due to limitations of the assay, particularly regarding mutations with low LOD or the presence of KRAS mutation at t2 may reflect clonal evolution of the tumour, with negative predictive value indicating disease progression not detectable at t0 or resistance via KRAS signalling. Furthermore, distinct KRASmut were associated to patients’ clinical outcomes; a patient with p.Q61L at t2, showed DCB (PFS: 11.5 months, OS: 15.5 months), whereas carriers of p.G12/G13 substitutions at t2 had NDB, with a 3.2 times higher risk for progression (table available at https://cdn.amegroups.cn/static/public/tlcr-2024-1095-3.xlsx), and a median PFS of 2 months. Additional investigation in independent and larger cohorts of aNSCLC patients is required to evaluate this hypothesis, including the detection of additional mutations in genes such as STK11 and KEAP1, which have been previously linked to resistance to ICIs in KRAS-mutant adenocarcinoma (37-39).

Growing evidence supports that NSCLC-tumours bearing common EGFR mutations (EGFRmut), such as deletions in exon 19 (Ex19Del) and p.L858R, exhibit low response rates to ICIs (40-44), while uncommon EGFRmut (p.E709K, p.G719X, p.S768I, p.L861Q, exon20-insertions) are associated with longer PFS when treated with ICIs (44,45). In agreement with these observations, we showed that five patients with p.E709K alone (IDs: 16, 19, 29, 38, 41) had DCB (median PFS =13.3 months); however, three NDB patients with the same mutation (IDs: 6, 13, 48) had a median PFS of 2 months. One patient with Ex19Del alone (ID: 7) had PFS of 1.8 months, while another (ID: 18) with both Ex19Del and p.E709K, along with p53 and ALK mutations, also had a PFS of 1.8 months. Another patient harboured the Ex19Del co-occurring with p53mut (ID: 46), with PFS of 4 months. Importantly, the Ex19Del was not identified in any of the DCB patients. These findings suggest that the identification of specific TP53, KRAS, and EGFR mutations may be associated with heterogeneous behaviour, hence may have dissimilar predictive significance in aNSCLC patients treated with immunotherapy.

Regarding the variation of ctDNA mutation load in terms of changes in VAF during immunotherapy, we used a cut-off of >50%-decrease in ctDNA (14,46) to further investigate whether reduction or clearance in ctDNA correlates with a benefitial response to pembrolizumab. Our results demonstrate that ctDNA-MR at an early phase during treatment is associated with longer PFS and OS (Figure 3), further confirming the findings of previous studies (14,46). Raja et al. also showed that a reduction in VAF at 6 weeks of durvalumab was associated with a reduction in tumour volume, and longer PFS and OS in patients with lung and bladder cancer (47). Anagnostou et al. showed that clearance of ctDNA maximal VAF at t2 or t3 was indicative of molecular response (MR), and was associated with longer PFS and OS in aNSCLC patients receiving pembrolizumab (48), while Stensgaard et al. showed that patients with NSCLC who presented an increase in ctDNA levels after immunotherapy initiation had inferior PFS and OS compared to those who did not (49). Importantly, our findings highlight that the correlation between ctDNA-dynamics and favourable clinical outcomes is significant when considering changes from baseline at sequential early treatment timepoints, but not when analysing changes from baseline to a single timepoint (t1/t2) alone, underlining the importance of ctDNA monitoring at an early phase of treatment. Corroborating these findings, Zhang et al. demonstrated that higher pre-treatment ctDNA VAF correlated with poorer OS in patients with advanced-stage tumours receiving durvalumab, while reduction of on-treatment VAF was associated with longer PFS and OS, suggesting that ctDNA dynamics are predictive of ICI response (50).

Previous evidence has suggested that CTCs may play a role in NSCLC diagnosis, biological characterization, and disease monitoring (51). Moreover, we previously demonstrated in the same cohort that a high-iKi67 in patients presenting PD-L1+ CTCs prior to immunotherapy was a predictive factor for poor response (21). Expanding on our previous findings, we found that NDB patients harboured a higher percentage of Ki67+ CTCs than DCB patients, and more importantly, patients showing a high-iKi67 at t2 had worse PFS and OS, suggesting that Ki67 might be used as a potential prognosticator of immunotherapy efficacy even after treatment initiation.

We subsequently examined whether the combination of the three circulating markers could predict disease outcomes. Such multifactorial analyses could provide superior prognostic information as more aspects of the disease are consolidated into a score. Interestingly, we found that the combination of ctDNA-dynamics and iKi67 could further improve sensitivity in the detection of NDB patients, while also maintaining a high specificity. Notably, the calculated sensitivity and specificity values suggest that the combination of these markers has the potential to inform patient stratification to treatment and improve decisions in clinical practice. Previous reports have also examined the value of combining different circulating biomarkers, such as cfDNA, ctDNA-derived somatic mutations and CTCs, to enhance predictive accuracy and better distinguish patients who will benefit from immunotherapy (25,26). To our knowledge, this is the first study to investigate ctDNA-dynamics in combination with the iKi67 in aNSCLC patients as combined biomarkers to monitor the response to pembrolizumab and potentially predict the effectiveness of ICI therapy at early timepoints through a non-invasive blood test.

Our findings underscore the value of LB biomarkers, such as ctDNA and CTCs, as early indicators of response to pembrolizumab in aNSCLC, and their potential in refining and personalising immunotherapy strategies. In this study, we show that a ctDNA-MR, defined as a >50% reduction or clearance in ctDNA VAF at early treatment timepoints, is associated with longer PFS and OS. Our results align with previous research, which also showed that reduction or clearance of ctDNA VAF at 6- or 9-weeks post-treatment initiation can correlate with reductions in tumour volume and improved survival outcomes in patients undergoing immunotherapy (47,48). Additional studies have demonstrated that a reduction in ctDNA VAF of more than 50% is associated with radiographic response and longer PFS and OS (14,46). These findings collectively highlight the importance of dynamic ctDNA monitoring at multiple early timepoints during immunotherapy, thereby enhancing the clinical relevance of our findings in aNSCLC. Regarding CTCs, our study indicates that patients with high iKi67 levels (>30% Ki67+ CTCs) at t2 experienced poorer PFS and OS. This observation aligns with and extends our previous findings, which highlighted that high Ki67 expression at baseline was a negative prognostic factor of pembrolizumab response in the same cohort of aNSCLC (21). Previous studies, including a detailed analysis by Tabata et al. (52), have also underscored the prognostic significance of Ki67 in NSCLC. They found that the highest score (HS) of Ki67 staining was an independent prognostic factor for OS in patients with NSCLC. Additionally, Hommura et al. (53) explored the role of Ki67 alongside p27KIP1 protein expression in surgically resected NSCLC and showed that a high iKi67 (>30%) was significantly associated with shorter survival. These studies support the clinical relevance of Ki67 as a robust marker for predicting outcomes in NSCLC patients treated with immunotherapy, further validating the utility of monitoring Ki67 expression on CTCs in our study. The multifactorial analysis that combines ctDNA dynamics with iKi67 levels has been shown to improve sensitivity in detecting NDB patients, reflecting the utility of integrating multiple circulating biomarkers to enhance predictive accuracy. This multifactorial assessment approach is supported by previous studies (25,26), such as the study of Alama et al. who found that a combined assessment of ctDNA and CTCs could predict OS with higher accuracy than when evaluating each biomarker individually. The clinical applicability of this multi-collective biomarker monitoring approach requires further validation in carefully designed and implemented clinical trials before drawing firm conclusions. Nevertheless, the logic of presenting a systematic overlay of multiple disease features as a means to monitor patients’ response to therapy essentially captures the complexity presented by each individual patient, while also considering their exposure to previous regimens, such as chemotherapy. For instance, this approach could be extremely useful in distinguishing against pseudo-progression in patients with MR or offering treatment intensification within clinical trials for patients without MR, as per the S1800 trial (54).

Conclusions

Our work suggests that an increase in cfDNA at an early stage of treatment with pembrolizumab could be predictive of lack of response of aNSCLC patients, suggesting a potential role for cfDNA-kinetics in real-time monitoring of immunotherapy. Furthermore, our study highlights the significance of monitoring ctDNA-derived mutational frequency dynamics. A ctDNA-MR during the early phases of treatment seems to serve as a promising real-time monitoring and predictive marker for patients’ response to pembrolizumab. Moreover, our findings demonstrate that the combination of ctDNA-dynamics with iKi67 provides superior predictive value compared to each one alone and supports the value of multifactorial analyses in the prediction of PD. Despite the limitation of the small sample size, our results provide new perspectives for monitoring immunotherapy response in aNSCLC patients, which could be further validated in larger cohorts of NSCLC patients receiving immunotherapy.

Acknowledgments

The authors sincerely thank all patients who participated in this project. Preliminary results were presented at the European Lung Cancer Congress 2024, and the submitted abstract was published in ESMO Open (DOI: 10.1016/j.esmoop.2024.102655).

Footnote

Reporting Checklist: The authors have completed the REMARK reporting checklist. Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2024-1095/rc

Data Sharing Statement: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2024-1095/dss

Peer Review File: Available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2024-1095/prf

Funding: The present work was financed by the European Regional Development Fund and the Republic of Cyprus through the Cyprus Research and Innovation Foundation as part of project RESTART 2016-2020 (EXCELLENCE/0918/0358) (to A.I.C. and C.D.). This study was partly supported by the EU’s HORIZON 2020 Research and Innovation Program, CY-Biobank, under Grant Agreement No. 857122, the Republic of Cyprus, and the University of Cyprus for supporting the biobank.cy Center of Excellence in Biobanking and Biomedical Research (to C.D.). Funding was also provided by MSD for the clinical conduct of the prospective SWIPE study (NCT02705820) and work on circulating tumour cells.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tlcr.amegroups.com/article/view/10.21037/tlcr-2024-1095/coif). A.I.C. acted as the coordinator of a competitive grant by the Cyprus Research and Innovation Foundation provided to realize this study. H.C. declares research institutional funding from MSD as well as advisory board participation with MSD, Novartis, Pfizer, and Ipsen, with all fees collected by his institution. H.C. also declares travel expenses in relation to oncology meetings covered by MSD, AstraZeneca, Roche, and Pfizer in the last 3 years. C.D. acted as the coordinator of a competitive grant by the Cyprus Research and Innovation Foundation provided to realize this study, and acts as the coordinator of an European Commission grant that partly supported research activities implemented to produce this manuscript. The other authors have no conflicts of interest to declare.

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. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by Cyprus National Bioethics Committee (No. 2019/76) and informed consent was taken from all the patients.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Gridelli C, Rossi A, Carbone DP, et al. Non-small-cell lung cancer. Nat Rev Dis Primers 2015;1:15009. [Crossref] [PubMed]
2. Reck M, Rodríguez-Abreu D, Robinson AG, et al. Five-Year Outcomes With Pembrolizumab Versus Chemotherapy for Metastatic Non-Small-Cell Lung Cancer With PD-L1 Tumor Proportion Score ≥ 50. J Clin Oncol 2021;39:2339-49. [Crossref] [PubMed]
3. Garassino MC, Gadgeel S, Speranza G, et al. Pembrolizumab Plus Pemetrexed and Platinum in Nonsquamous Non–Small-Cell Lung Cancer: 5-Year Outcomes From the Phase 3 KEYNOTE-189 Study. J Clin Oncol 2023;41:1992-8. [Crossref] [PubMed]
4. Novello S, Kowalski DM, Luft A, et al. Pembrolizumab Plus Chemotherapy in Squamous Non–Small-Cell Lung Cancer: 5-Year Update of the Phase III KEYNOTE-407 Study. J Clin Oncol 2023;41:1999-2006. [Crossref] [PubMed]
5. Lung cancer survival statistics | Cancer Research UK. Available online: https://www.cancerresearchuk.org/health-professional/cancer-statistics/statistics-by-cancer-type/lung-cancer/survival
6. Ahmadzadeh M, Johnson LA, Heemskerk B, et al. Tumor antigen-specific CD8 T cells infiltrating the tumor express high levels of PD-1 and are functionally impaired. Blood 2009;114:1537-44. [Crossref] [PubMed]
7. Summary of product characteristics. Available online: https://www.ema.europa.eu/en/documents/product-information/keytruda-epar-product-information_en.pdf
8. Patel SP, Kurzrock R. PD-L1 Expression as a Predictive Biomarker in Cancer Immunotherapy. Mol Cancer Ther 2015;14:847-56. [Crossref] [PubMed]
9. Krebs MG, Malapelle U, André F, et al. Practical Considerations for the Use of Circulating Tumor DNA in the Treatment of Patients With Cancer: A Narrative Review. JAMA Oncol 2022;8:1830-9. [Crossref] [PubMed]
10. Spagnolo CC, Pepe F, Ciappina G, et al. Circulating biomarkers as predictors of response to immune checkpoint inhibitors in NSCLC: Are we on the right path? Crit Rev Oncol Hematol 2024;197:104332. [Crossref] [PubMed]
11. Ricciuti B, Jones G, Severgnini M, et al. Early plasma circulating tumor DNA (ctDNA) changes predict response to first-line pembrolizumab-based therapy in non-small cell lung cancer (NSCLC). J Immunother Cancer 2021;9:e001504. [Crossref] [PubMed]
12. Ignatiadis M, Sledge GW, Jeffrey SS. Liquid biopsy enters the clinic - implementation issues and future challenges. Nat Rev Clin Oncol 2021;18:297-312. [Crossref] [PubMed]
13. Cabel L, Riva F, Servois V, et al. Circulating tumor DNA changes for early monitoring of anti-PD1 immunotherapy: a proof-of-concept study. Ann Oncol 2017;28:1996-2001. [Crossref] [PubMed]
14. Goldberg SB, Narayan A, Kole AJ, et al. Early Assessment of Lung Cancer Immunotherapy Response via Circulating Tumor DNA. Clin Cancer Res 2018;24:1872-80. [Crossref] [PubMed]
15. Iijima Y, Hirotsu Y, Amemiya K, et al. Very early response of circulating tumour-derived DNA in plasma predicts efficacy of nivolumab treatment in patients with non-small cell lung cancer. Eur J Cancer 2017;86:349-57. [Crossref] [PubMed]
16. Cabel L, Proudhon C, Romano E, et al. Clinical potential of circulating tumour DNA in patients receiving anticancer immunotherapy. Nat Rev Clin Oncol 2018;15:639-50. [Crossref] [PubMed]
17. Vasseur A, Kiavue N, Bidard FC, et al. Clinical utility of circulating tumor cells: an update. Mol Oncol 2021;15:1647-66. [Crossref] [PubMed]
18. Jee J, Lebow ES, Yeh R, et al. Overall survival with circulating tumor DNA-guided therapy in advanced non-small-cell lung cancer. Nat Med 2022;28:2353-63. [Crossref] [PubMed]
19. Kris MG, Johnson BE, Berry LD, et al. Using multiplexed assays of oncogenic drivers in lung cancers to select targeted drugs. JAMA 2014;311:1998-2006. [Crossref] [PubMed]
20. Schwartz LH, Litière S, de Vries E, et al. RECIST 1.1-Update and clarification: From the RECIST committee. Eur J Cancer 2016;62:132-7. [Crossref] [PubMed]
21. Spiliotaki M, Neophytou CM, Vogazianos P, et al. Dynamic monitoring of PD-L1 and Ki67 in circulating tumor cells of metastatic non-small cell lung cancer patients treated with pembrolizumab. Mol Oncol 2023;17:792-809. [Crossref] [PubMed]
22. Therneau TM. A package for survival analysis in R. 2022. Comprehensive R Archive Network (CRAN).
23. Therneau TM, Grambsch PM. Modeling Survival Data: Extending the Cox Model. Springer, Berlin; 2000. doi: 10.1007/978-1-4757-3294-8.
24. Gautschi O, Bigosch C, Huegli B, et al. Circulating deoxyribonucleic Acid as prognostic marker in non-small-cell lung cancer patients undergoing chemotherapy. J Clin Oncol 2004;22:4157-64. [Crossref] [PubMed]
25. Alama A, Coco S, Genova C, et al. Prognostic Relevance of Circulating Tumor Cells and Circulating Cell-Free DNA Association in Metastatic Non-Small Cell Lung Cancer Treated with Nivolumab. J Clin Med 2019;8:1011. [Crossref] [PubMed]
26. Passiglia F, Galvano A, Castiglia M, et al. Monitoring blood biomarkers to predict nivolumab effectiveness in NSCLC patients. Ther Adv Med Oncol 2019;11:1758835919839928. [Crossref] [PubMed]
27. Mondelo-Macía P, García-González J, León-Mateos L, et al. Clinical potential of circulating free DNA and circulating tumour cells in patients with metastatic non-small-cell lung cancer treated with pembrolizumab. Mol Oncol 2021;15:2923-40. [Crossref] [PubMed]
28. Rolfo C, Mack P, Scagliotti GV, et al. Liquid Biopsy for Advanced NSCLC: A Consensus Statement From the International Association for the Study of Lung Cancer. J Thorac Oncol 2021;16:1647-62. [Crossref] [PubMed]
29. Deniger DC, Pasetto A, Robbins PF, et al. T-cell Responses to TP53 "Hotspot" Mutations and Unique Neoantigens Expressed by Human Ovarian Cancers. Clin Cancer Res 2018;24:5562-73. [Crossref] [PubMed]
30. Nakajima EC, Ren Y, Vallejo JJ, et al. Outcomes of first-line immune checkpoint inhibitors with or without chemotherapy according to KRAS mutational status and PD-L1 expression in patients with advanced NSCLC: FDA pooled analysis. J Clin Oncol 2022;40:9001.
31. Riudavets M, Auclin E, Mosteiro M, et al. Durvalumab consolidation in patients with unresectable stage III non-small cell lung cancer with driver genomic alterations. Eur J Cancer 2022;167:142-8. [Crossref] [PubMed]
32. Zulato E, Attili I, Pavan A, et al. Early assessment of KRAS mutation in cfDNA correlates with risk of progression and death in advanced non-small-cell lung cancer. Br J Cancer 2020;123:81-91. [Crossref] [PubMed]
33. Liu C, Zheng S, Wang Z, et al. KRAS-G12D mutation drives immune suppression and the primary resistance of anti-PD-1/PD-L1 immunotherapy in non-small cell lung cancer. Cancer Commun (Lond) 2022;42:828-47. [Crossref] [PubMed]
34. Michaelidou K, Koutoulaki C, Mavridis K, et al. Detection of KRAS G12/G13 Mutations in Cell Free-DNA by Droplet Digital PCR, Offers Prognostic Information for Patients with Advanced Non-Small Cell Lung Cancer. Cells 2020;9:2514. [Crossref] [PubMed]
35. Yang S, Yu X, Fan Y, et al. Clinicopathologic characteristics and survival outcome in patients with advanced lung adenocarcinoma and KRAS mutation. J Cancer 2018;9:2930-7. [Crossref] [PubMed]
36. Muthiah A, Chudasama R, Mingrino J, et al. Clinical characteristic and outcomes of KRAS G13 mutant non-small cell lung cancers. J Clin Oncol 2022;40:e21017.
37. Ricciuti B, Arbour KC, Lin JJ, et al. Diminished Efficacy of Programmed Death-(Ligand)1 Inhibition in STK11- and KEAP1-Mutant Lung Adenocarcinoma Is Affected by KRAS Mutation Status. J Thorac Oncol 2022;17:399-410. [Crossref] [PubMed]
38. Peters S, Cho BC, Luft A, et al. OA15.04 Association Between KRAS/STK11/KEAP1 Mutations and Outcomes in POSEIDON: Durvalumab ± Tremelimumab + Chemotherapy in mNSCLC. J Thorac Oncol 2022;17:S39-41.
39. Pavan A, Bragadin AB, Calvetti L, et al. Role of next generation sequencing-based liquid biopsy in advanced non-small cell lung cancer patients treated with immune checkpoint inhibitors: impact of STK11, KRAS and TP53 mutations and co-mutations on outcome. Transl Lung Cancer Res 2021;10:202-20. [Crossref] [PubMed]
40. Lisberg A, Cummings A, Goldman JW, et al. A Phase II Study of Pembrolizumab in EGFR-Mutant, PD-L1+, Tyrosine Kinase Inhibitor Naïve Patients With Advanced NSCLC. J Thorac Oncol 2018;13:1138-45. [Crossref] [PubMed]
41. Lin A, Wei T, Meng H, et al. Role of the dynamic tumor microenvironment in controversies regarding immune checkpoint inhibitors for the treatment of non-small cell lung cancer (NSCLC) with EGFR mutations. Mol Cancer 2019;18:139. [Crossref] [PubMed]
42. Lee CK, Man J, Lord S, et al. Checkpoint Inhibitors in Metastatic EGFR-Mutated Non-Small Cell Lung Cancer-A Meta-Analysis. J Thorac Oncol 2017;12:403-7. [Crossref] [PubMed]
43. Lei Y, Wang K, Liu Y, et al. Various Subtypes of EGFR Mutations in Patients With NSCLC Define Genetic, Immunologic Diversity and Possess Different Prognostic Biomarkers. Front Immunol 2022;13:811601. [Crossref] [PubMed]
44. Mazieres J, Drilon A, Lusque A, et al. Immune checkpoint inhibitors for patients with advanced lung cancer and oncogenic driver alterations: results from the IMMUNOTARGET registry. Ann Oncol 2019;30:1321-8. [Crossref] [PubMed]
45. Yoshida H, Kim YH, Ozasa H, et al. Nivolumab in non-small-cell lung cancer with EGFR mutation. Ann Oncol 2018;29:777-8. [Crossref] [PubMed]
46. Thompson JC, Carpenter EL, Silva BA, et al. Serial Monitoring of Circulating Tumor DNA by Next-Generation Gene Sequencing as a Biomarker of Response and Survival in Patients With Advanced NSCLC Receiving Pembrolizumab-Based Therapy. JCO Precis Oncol 2021;5:PO.20.00321.
47. Raja R, Kuziora M, Brohawn PZ, et al. Early Reduction in ctDNA Predicts Survival in Patients with Lung and Bladder Cancer Treated with Durvalumab. Clin Cancer Res 2018;24:6212-22. [Crossref] [PubMed]
48. Anagnostou V, Ho C, Nicholas G, et al. ctDNA response after pembrolizumab in non-small cell lung cancer: phase 2 adaptive trial results. Nat Med 2023;29:2559-69. [Crossref] [PubMed]
49. Stensgaard S, Thomsen A, Helstrup S, et al. Blood tumor mutational burden and dynamic changes in circulating tumor DNA predict response to pembrolizumab treatment in advanced non-small cell lung cancer. Transl Lung Cancer Res 2023;12:971-84. [Crossref] [PubMed]
50. Zhang Q, Luo J, Wu S, et al. Prognostic and Predictive Impact of Circulating Tumor DNA in Patients with Advanced Cancers Treated with Immune Checkpoint Blockade. Cancer Discov 2020;10:1842-53. [Crossref] [PubMed]
51. Tartarone A, Rossi E, Lerose R, et al. Possible applications of circulating tumor cells in patients with non small cell lung cancer. Lung Cancer 2017;107:59-64. [Crossref] [PubMed]
52. Tabata K, Tanaka T, Hayashi T, et al. Ki-67 is a strong prognostic marker of non-small cell lung cancer when tissue heterogeneity is considered. BMC Clin Pathol 2014;14:23. [Crossref] [PubMed]
53. Hommura F, Dosaka-Akita H, Mishina T, et al. Prognostic significance of p27KIP1 protein and ki-67 growth fraction in non-small cell lung cancers. Clin Cancer Res 2000;6:4073-81.
54. Reckamp KL, Redman MW, Dragnev KH, et al. Phase II Randomized Study of Ramucirumab and Pembrolizumab Versus Standard of Care in Advanced Non-Small-Cell Lung Cancer Previously Treated With Immunotherapy-Lung-MAP S1800A. J Clin Oncol 2022;40:2295-306. Erratum in: J Clin Oncol 2022;40:3002. [Crossref] [PubMed]