Clinical Implication of Monitoring Circulating Tumor DNA in Untreated Non-Small-Cell Lung Cancer Patients

Abstract

CtDNA has been utilized to monitor the clinical course of the patients of NSCLC who receive therapies targeting druggable mutations. However, despite providing valuable information on how NSCLC would naturally progress, the clinical utility of ctDNA for clinical-course monitoring and prediction of the treatment-naïve NSCLC patients without druggable mutations remain unknown. We longitudinally followed a total of 12 treatment-naïve NSCLC patients, who did not harbor EGFR and ALK mutations, by collecting clinical information, radiological data, and plasma samples. Changes in ctDNA levels and tumor burden (TB) were also compared with each other. Volume doubling time (VDT), and overall survival (OS) were analyzed regarding ctDNA detection at diagnosis. CtDNA was detected in the plasma of seven (58.3%) patients. Changes in ctDNA levels correlated with those in TB in a substantial fraction (57.1%) of patients and was also associated with brain metastasis, tumor necrosis, or pneumonia in other patients. In addition, the patients with ctDNA detection had shorter VDT (p = 0.039) and worse OS (p = 0.019) than those without ctDNA detection. The natural course of NSCLC progression can be monitored by measuring ctDNA levels. Detection of ctDNA at diagnosis can predict rapid tumor growth and poor survival of NSCLC patients.

Introduction

Easily obtainable through liquid biopsy, circulating tumor DNA (ctDNA) has been of great interest to those who seek cancer biomarkers ^1,2. In non-small cell lung cancer (NSCLC), however, the clinical utility of ctDNA has been limited to detecting mutations in the genes that encode druggable proteins such as Epidermal Growth Factor Receptor (EGFR) and Anaplastic Lymphoma Kinase (ALK). In addition, the current clinical-course monitoring of NSCLC by using ctDNA usually examines a single druggable driver gene in the patients with targeted therapy ³.

While the majority of NSCLC patients receive anti-cancer treatments, a substantial proportion of patients (~ 20%) do not ⁴. Elucidating the clinical course of such treatment-naïve patients, who currently are rarely followed up in most cases, would provide us with unprecedented, valuable information on how NSCLC naturally progresses in relation to patient survival ⁵. Genetic markers as carried by ctDNA would help reveal molecular paths that NSCLC would naturally follow in the absence of therapy. Notably, the two most commonly adopted druggable mutations in monitoring NSCLC using ctDNA, those in EGFR and ALK, are recoverable at most 20% of patients ⁶. Whether the remainder of NSCLC patients without the druggable mutations would benefit from using ctDNA for clinical-course prediction or monitoring remains unclear.

In this vein, we investigated whether ctDNA obtained from treatment-naïve NSCLC patients could provide us with critical information on natural clinical course when mutations affecting the druggable target genes are absent. We conducted longitudinal assessments of ctDNA levels, mutation maker fluctuations, and clinical features.

Results

Patient characteristics and ctDNA detection. Out of the 12 patients included in this study, eight (83.3%) were male and nine (75.0%) had smoking history (Fig. 1). While eight (83.3%) patients had adenocarcinoma, eight (83.3%) were diagnosed with stage IV NSCLC. Pneumonia was observed in two patients (P8 and P12) during follow-ups (Table 1). Necrosis of primary tumor was observed in two patients (P3 and P8). TB gradually increased in 11 (91.7%) patients except P8. In P8, the size of the solid tumor temporarily decreased slightly in proportion to the increase in necrotic area. The median duration of follow-ups was 298 days. Death was observed in 11 patients. The median survival of the patients was 235 days.

^aT1 to T4 denote a time point for plasma sampling and imaging tests.

ctDNA, circulating tumor DNA; VDT, volume doubling time

CtDNA was detected in the plasma of 7 (58.3%) patients (Supplementary Table S1 online). The median ctDNA level was 9.1 hGE/ml (range: 0.8–108.5) and the median variant allele frequency was 4.2% (range 0.1–21.9). Patients with detected ctDNA showed a trend to be male, have smoking history, have adenocarcinoma, and/or be diagnosed at advanced stages. The most common mutations detected were in TP53 (five patients [41.7%]), followed by those in CSMD3, FAT1, ARID1B, TULP4, DST, PCDH15, PKHD1L1, and LRP1B. The results of validation ddPCR were comparable to those of NGS in the patients with detected ctDNA (Supplementary Table S2 online).

Change in ctDNA levels and tumor burden. Four patients (57.1%) with detected ctDNA (P6, P9, P10, and P11) displayed a tendency of increasing ctDNA levels over time, which may correspond to increase in TB (Fig. 2). By contrast, changes of ctDNA levels in the other three patients (P1, P3, and P12) showed discordance with increase in TB. Several distinct clinical features were accompanied in these patients. In P1, although the primary tumor size gradually decreased, the total TB gradually increased as the size of the brain metastasis increased. In P3, the size of the primary tumor slightly increased whereas the area with accompanied tumor necrosis decreased. Pneumonia was occurred at the diagnosis and the third follow-up period in P12.

Volume doubling time and overall survival regarding ctDNA detection at diagnosis. The median value of VDT in the patients whose ctDNA was detected at diagnosis was 73 days. The patients whose ctDNA was not detected displayed a longer median VDT, 179 days (p = 0.039) (Fig. 3). Of note, the patients with detected ctDNA had a significantly worse OS than those without detected ctDNA (the median OS of 153 days versus 501 days, log-rank p = 0.019) (Fig. 4). Moreover, patients with high-level ctDNA had worse OS compared with those with low-level ctDNA (the median OS of 121 days versus 235 days, log-rank p = 0.048).

Discussion

We have here shown that changes in ctDNA levels correlated with TB in a substantial fraction of treatment-naïve NSCLC patients without the druggable mutations, but it could be affected by various clinical factors. We also noted that the natural course of NSCLC progression could be predicted by detecting ctDNA. We found that patients with ctDNA detected at diagnosis had shorter VDT and worse OS during follow-ups.

The ctDNA levels of patients as measured in hGE in this study was lower than those of the patients receiving chemotherapy in other studies ^7,8. This indicates that a fraction of treatment-naïve NSCLC patients could already harbor a considerable amount of ctDNA, which could be further increased by treatment-induced cancer cell death. Previous studies suggested that changes in ctDNA levels can be positively correlated with TB of the patients who receive chemotherapy ^7,9. Our results showed that alteration in ctDNA levels correlates with that in TB in some treatment-naïve patients whereas others do not. We speculate that the relatively low amount of ctDNA detected in the treatment-naïve patients could be easily affected by clinical factors. Increased size of brain metastasis may account for the increased overall TB while the primary tumor size shrinking (P1). As the blood-brain barrier could prevent ctDNA from entering the circulation, the increased size of brain metastasis may not be reflected in ctDNA ¹⁰. The fall of ctDNA level could be also explained by alteration of area with tumor necrosis (P3) ³ or pneumonia (P12) ¹¹. Our results showed that the patients with ctDNA detection would develop accelerated tumor growth. The short VDT of the patients with ctDNA detection could reflect their high tumor proliferation rate ¹². The high-level tumor proliferation observed in the patients with ctDNA detection may account for the poor prognosis of these patients.

Our results should be interpreted with caution. First, this study included a relatively small number of samples. However, finding a sizable number of appropriate treatment-naïve patients, who also received follow-ups with imaging studies and ctDNA isolation, required a huge patient population. In our cohort, only 12 patients (0.3%) met all the criteria among 3,740 patients diagnosed as NSCLC. Second, somatic mutations recovered from the ctDNA were not compared with those from the corresponding tumor specimens. However, we did every effort to filter out false signals in this study (see the Supplementary material).

To summarize, this study reveals that natural course could be monitored using ctDNA levels in NSCLC patients but various clinical factors such as brain metastasis, tumor necrosis or infection can affect the ctDNA levels. Our results also imply that detection of ctDNA at diagnosis can predict rapid tumor growth and poor survival of the NSCLC patients. This study indicates that natural course could be monitored or predicted with individually harbored somatic mutations in ctDNA of NSCLC patients.

Methods

Patients and radiological evaluation. 3,740 patients who had been histologically diagnosed as NSCLC between January 2008 and March 2017 were identified in the Inha Lung Cancer Cohort (Inha University Hospital, Incheon, South Korea) and initially considered for this study (Fig. 5). Among them, 153 patients did not take any anti-cancer treatment. The treatment refusal was based on (1) either severe comorbid diseases or more than three markings in the Eastern Cooperative Oncology Group performance status (n = 87); (2) patients’ wish not to take treatment (n = 39); (3) poor economic status (n = 18); and (4) unknown reasons (n = 9). Finally, a total of 12 patients (P1 to P12) were included in this study.

All clinical information regarding age, gender, smoking history, histology, and stage was prospectively acquired. Radiological evaluations were performed at the time of diagnosis and the ensuing follow-ups with the average of three-month intervals (range: 2–5). The stage was estimated using the 8th edition of the Tumor-Node-Metastasis classification ¹³. Tumor burden (TB) was measured with the sum of greatest diameters of target lesions with the standard of RECIST v1.1 ¹⁴. The study protocol was approved by the Institutional Review Board of Inha University Hospital and informed consents were obtained from patients. All methods were performed in accordance with the relevant guidelines and regulations.

Plasma samples and DNA extraction. 37 serial plasma samples of the patients included in the study were drawn at the same time as the radiological evaluation and subsequently stored at -80^oC until use. Cell-free DNA (cfDNA) was extracted from the plasma samples using the QIAamp Circulating Nucleic Acid Kit (QIAGEN, Hilden, Germany) and eluted in 30㎕ of Buffer AVE according to the manufacturer's instruction (Supplementary materials).

Targeted next-generation sequencing. Targeted next-generation sequencing (NGS) was performed on all ctDNA samples using a panel of 113 genes (Supplementary Table S3 online). Several filtering steps were applied to sieve out putative germline and false variants. A total of 66 variants were identified (Supplementary Table S1 online). Among these variants and some putative germline variants, 8 variants with sufficient amounts of DNA were validated with digital droplet PCR (ddPCR) (Supplementary Table S2 online). CtDNA levels were expressed in haploid genome equivalents per milliliter of plasma (hGE/mL). Detailed information on the methods for the targeted NGS, preprocessing and variant analysis, and ddPCR is described in the Supplementary materials.

Statistical analysis. CtDNA levels were analyzed as continuous or dichotomized variables based on the median value. Volume doubling time (VDT) was calculated by using an equation based on the modified Schwartz formula ¹⁵. Overall survival (OS) was defined from the time of diagnosis until death as a result of all causes. If patients were still alive on the last follow-up date, they were censored on that day. Patients were classified into two groups with the presence of detected ctDNA. Changes in ctDNA levels were compared with those of TB. VDT was compared between groups using t-test. The median OS was estimated by the Kaplan-Meier method and compared by using log-rank test. To estimate statistical significance, two-tailed hypothetical testing was performed with rejecting the null hypothesis when p < 0.05. All statistical analyses were performed using the IBM SPSS statistical software package (version 19.0; SPSS Inc., Chicago, IL, USA).

Declarations

Acknowledgments

This work was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF-2017R1E1A1A01074863).

Author Contributions

(I) Conception and design: JH Lim, H-T Shin, S Oh, J-S Ryu; (II) Administrative support: J-S Ryu; (III) Provision of study materials or patients: JH Lim, J-S Ryu; (IV) Collection and assembly of data: JH Lim, H-T Shin, SJ Lee, J-S Ryu; (V) Data analysis and interpretation: JH Lim, H-T Shin, S Oh, J-S Ryu; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors.

Additional Information

Competing Interests 

The authors have declared no competing interests.

Data Availability

The data and analytic methods are available upon request.

References

1. Aggarwal, C. et al. Clinical Implications of Plasma-Based Genotyping With the Delivery of Personalized Therapy in Metastatic Non-Small Cell Lung Cancer. JAMA Oncol, doi:10.1001/jamaoncol.2018.4305 (2018).
2. de Bruin, E. C. et al. Spatial and temporal diversity in genomic instability processes defines lung cancer evolution. Science 346, 251-256, doi:10.1126/science.1253462 (2014).
3. Diaz, L. A., Jr. & Bardelli, A. Liquid biopsies: genotyping circulating tumor DNA. J Clin Oncol 32, 579-586, doi:10.1200/jco.2012.45.2011 (2014).
4. David, E. A. et al. Increasing Rates of No Treatment in Advanced-Stage Non-Small Cell Lung Cancer Patients: A Propensity-Matched Analysis. J Thorac Oncol 12, 437-445, doi:10.1016/j.jtho.2016.11.2221 (2017).
5. Detterbeck, F. C. & Gibson, C. J. Turning gray: the natural history of lung cancer over time. J Thorac Oncol 3, 781-792, doi:10.1097/JTO.0b013e31817c9230 (2008).
6. Campbell, J. D. et al. Distinct patterns of somatic genome alterations in lung adenocarcinomas and squamous cell carcinomas. Nat Genet 48, 607 (2016).
7. Chaudhuri, A. A. et al. Early Detection of Molecular Residual Disease in Localized Lung Cancer by Circulating Tumor DNA Profiling. Cancer Discov 7, 1394-1403, doi:10.1158/2159-8290.Cd-17-0716 (2017).
8. Cabel, L. et al. Circulating tumor DNA changes for early monitoring of anti-PD1 immunotherapy: a proof-of-concept study. Ann Oncol 28, 1996-2001 (2017).
9. Horn, L. et al. Monitoring therapeutic response and resistance: analysis of circulating tumor DNA in patients with ALK+ lung cancer. J Thorac Oncol 14, 1901-1911 (2019).
10. Bettegowda, C. et al. Detection of circulating tumor DNA in early- and late-stage human malignancies. Sci Transl Med 6, 224ra224, doi:10.1126/scitranslmed.3007094 (2014).
11. De Vlaminck, I. et al. Noninvasive monitoring of infection and rejection after lung transplantation. Proc Natl Acad Sci U S A 112, 13336-13341 (2015).
12. Riva, F. et al. Patient-specific circulating tumor DNA detection during neoadjuvant chemotherapy in triple-negative breast cancer. Clin Chem 63, 691-699 (2017).
13. Goldstraw, P. et al. The IASLC Lung Cancer Staging Project: Proposals for Revision of the TNM Stage Groupings in the Forthcoming (Eighth) Edition of the TNM Classification for Lung Cancer. J Thorac Oncol 11, 39-51, doi:10.1016/j.jtho.2015.09.009 (2016).
14. Eisenhauer, E. A. et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 45, 228-247 (2009).
15. Schwartz, M. A biomathematical approach to clinical tumor growth. Cancer 14, 1272-1294 (1961).