Introduction

Peripheral blood obtained from oncological patients often contains tumor fragments, such as circulating malignant cells, tissue-specific proteins and cancer-derived nucleic acids. The analysis of circulating tumor DNA (ctDNA) is particularly promising, given that somatic mutations are highly specific for malignant cells and that current methods are capable of detecting even single mutated molecules present within a huge excess of normal tissues [1,2,3]. There are intensive efforts to utilize so-called “liquid biopsy” for early diagnosis and monitoring of cancer disease [4, 5]. It is more or less established, that the concentration of ctDNA is generally proportional to the overall tumor mass [6,7,8,9]. Consequently, ctDNA is relatively easily detectable in patients with extensive tumor disease, and its content usually declines after successful treatment [10,11,12,13].

The reduction of tumor size, which is achieved with the use of cancer drugs, is attributed to several biological effects. Conventional chemotherapy and targeted agents have a cytostatic action, i.e. they prevent proliferation of cancer cells [14, 15]. In addition, both cytotoxic and targeted drugs may provoke tumor shrinkage by inducing programmed cell death [16,17,18]. The reduction of overall tumor volume, which is usually achieved after weeks of systemic therapy, is almost always accompanied by the decrease of the concentration of circulating markers, be it tumor-specific proteins or ctDNA [19, 20]. However, the immediate marker response to the cancer therapy is less studied, which is at least in part attributed to the difficulties in collecting multiple serial blood samples within short time intervals.

EGFR tyrosine kinase inhibitors (TKIs; gefitinib, erlotinib, afatinib, osimertinib, etc.) are highly efficient in non-small cell lung carcinomas (NSCLCs), which harbor activating mutation in exon 19 or 21 of the EGFR gene [21,22,23,24,25]. Administration of EGFR TKIs for the treatment of EGFR-mutated NSCLC is almost always accompanied by the objective tumor response or the disease stabilization. EGFR inhibitors are also characterized by the “Lazarus effect”, i.e. dramatic symptomatic relief occurring within first hours after the drug administration [26, 27]. It is well established that the reduction of tumor size, which is observed during regular patient check-ups, is paralleled by the decline of the amount of EGFR-mutated ctDNA. However, short-term effects of EGFR TKIs on the level of ctDNA have not been systematically analyzed. We aimed to investigate, how the administration of EGFR TKIs influences the concentration of plasma ctDNA within the first hours after the uptake of the drug, and whether these changes are predictive for the long-term effects of systemic therapy.

Materials and methods

Patients

The study considered consecutive treatment-naïve patients with EGFR-mutated locally advanced or metastatic NSCLC, who were referred to the St.-Petersburg City Cancer Center between August 2018 and March 2020. EGFR mutations testing in tumor tissue was performed as described in [28]. Briefly, EGFR deletions in exon 19 (19del) were analyzed using the primers 5ʹ-CTGTCATAGGGACTCTGGAT-3ʹ and 5ʹ-CAGCAAAGCAGAAACTCACAT-3ʹ; PCR products were electrophoretically separated in 10% polyacrylamide gel; a 127 bp fragment corresponded to the wild-type sequence, and an additional band of smaller size was observed in the case of deletion. Testing for EGFR L858R mutation in exon 21 was performed by allele-specific real-time PCR with the wild-type-specific primer 5ʹ-CACCCAGCAGTTTGGCCA-3’, mutation-specific primer 5ʹ-CACCCAGCAGTTTGGCCC-3ʹ, and common primer 5’-GCATGAACTACTTGGAGGAC-3.

All patients provided informed consent for the participation in the study. The study was approved by the local Ethics Committee (protocol #20 “Evaluation the clinical value of ctDNA testing in patients with EGFR-mutated NSCLC”; approval date November 23, 2017).

Thirty patients were recruited to the investigation; their characteristics are described in Table 1 [see also Tables 1S and 2S in the electronic supplementary material (ESM)] for the description of individual patient data and response to TKI treatment). Serial plasma samples were collected before intake of the first tablet (at base-line) and at 0.5, 1, 2, 3, 6, 12, 24, 36 and 48 h after the “starting point” (Fig. 1). In addition, all patients were invited to donate blood after 14 and 28 days of the treatment. EGFR TKIs were given at regular daily doses (gefitinib: 250 mg; erlotinib: 150 mg; afatinib: 40 mg; osimertinib: 80 mg). Early response evaluations were performed with spiral computed tomography (CT) on the week 4, and the routine check-ups occurred within weeks 8–12 after the beginning of the therapy. Chest scans were performed before and after the administration of the contrast agent (100 mL of non-ionic iodinated contrast with a 100 mL saline chaser at 4.5–5 ml/s). All nodules with the size of more than 10 mm were measured. The images were based on venous-phase scans. Image reconstructions were performed on a CT workstation (Vitrea). Tumor burden was evaluated using the slice-by-slice pathology volume measurement with the slice thickness of 1.0 mm and assistance of the RadiAnt DICOM Viewer V.4.5.9.18463 software. Tumor responses RECIST v.1.1 criteria and progression-free survival (PFS) were evaluated according to standard guidelines by the study investigators [29].

Table 1 Clinical characteristics of NSCLC patients included in the study
Full size table
Fig. 1
figure 1

Work-flow of the study

Full size image

ctDNA analysis

Blood samples (10 mL) were collected in cf-DNA/cf-RNA Preservative Tubes (Norgen) and the plasma was separated from the rest of the specimen by the two-step centrifugation protocol (400 g for 10 min at room temperature followed by 14,400 g for 10 min at 4 °C). Cell-free DNA was extracted with the QIAamp Circulating Nucleic Acid kit from 3–5 ml of plasma according to the manufacturer's instructions and dissolved in 50 μl of water.

The fractions of EGFR mutant alleles (exon 19 deletions or L858R substitutions) were measured by droplet digital PCR (ddPCR) using the QX100 Bio-Rad System [30]. ddPCR reactions were performed in triplicate and contained 2X ddPCR Supermix for Probes (no UTP, Bio-Rad), mutation-specific oligonucleotides (see Table 3S in the ESM) and 2–3 μl of the template DNA in a total reaction volume of 22–23 μl. Data analyses were performed with the QuantaSoft Software version 1.7.4 as recommended by the manufacturer. All ddPCR reactions, which yielded 10 or more droplets with the target DNA molecule, were considered informative.

The absolute number of tumor-derived “mutated” DNA copies in 1 mL of plasma (Cmut) was calculated according to the formula:

$$N \, mut{\text{ copies}}/1{\text{mL plasma }} = \, \frac{{{\text{Concentration}}\left( {\frac{{{\text{copies}}}}{{\mu {\text{L}}}}cf{\text{DNA}}} \right) \times V_{{{\text{template}}}} \times V_{{{\text{dilution}}}} }}{{V_{{{\text{plasma}}}} }}$$

where: Concentration—number of «mutated» droplets per 1 μL of ddPCR reaction. V template—volume of ctDNA aliquot taken into ddPCR, μL. V dilution—total volume of diluted ctDNA sample collected from the plasma, μL. V plasma—volume of processed plasma, mL.

Statistics

Quantitative data were present as a median values/range or means ± 95% confidence interval (1.960σx̄). Non-parametric Wilcoxon Signed Rank Test and Mann–Whitney U Test were utilized to compare the medians. p value of < 0.05 was considered statistically significant. All calculations were performed using IBM SPSS v.23 software package.

Results

Clinical responses to EGFR TKI therapy

All 30 included patients attended CT examination after 4 weeks of TKI treatment. 25 subjects demonstrated partial tumor response, 3 had stable disease and 2 progressed during the treatment (See Table 2 in the ESM). Twenty-nine patients continued TKI therapy (28 cases with the disease control (objective tumor response or stable disease) and 1 case beyond progression) after the first check-up. Twenty-five patients managed to attend the second CT examination, which was performed within 8–12 weeks after the beginning of the treatment; among 5 missing subjects, 3 patients underwent cytoreductive surgery, 1 subject refused examination due to COVID-19 epidemic precautions, and 1 patient died on the 6th week of treatment. The death of the patient occurred after sudden and rapid symptomatic deterioration; the cause of the death was unknown as the family of this subject refused an autopsy.

ctDNA analysis at base-line

Thirty patients were subjected to the ctDNA analysis at base-line. EGFR-mutated DNA was detected in 25/30 (83%) subjects (Table 2, Fig. 1). As expected, the sum volume of the tumor lesions was evidently higher in patients with detectable mutated ctDNA level as compared with “plasma-negative” patients, but the difference did not reach the statistical significance (29,463 mm3 vs. 9963 mm3, p = 0.552, Mann–Whitney U test). The probability of detecting ctDNA at base-line did not correlate with the patient age or gender, number of metastatic sites or EGFR mutation type (Table 2). The first CT evaluation of tumor response at 4th week after the beginning of anti-EGFR therapy documented a trend towards more pronounced tumor volume decrease in the “ctDNA-positive” group as compared with “ctDNA-negative” patients (− 61% vs. − 18.5%, p = 0.208, Mann–Whitney U test). This tendency was not maintained after 8–12 weeks of treatment (Table 2). Patients with detectable EGFR-mutated DNA at base-line had shorter PFS than “ctDNA-negative” cases, however this difference was also below the level of statistical significance [11.4 months vs. 21.0 months, p = 0.238, Breslow (generalized Wilcoxon) test for comparison of Kaplan–Meier curves].

Table 2 Clinical characteristics of NSCLC patients, tumor response to TKI treatment and changes in ctDNA content
Full size table

Changes of ctDNA concentration during the first hours of TKI exposure

None of 5 patients, who were negative for plasma EGFR-mutated DNA at base-line, showed the presence of ctDNA (at least 5 mutation-specific signals per reaction) in the serial samples, which were taken in the first hours after the beginning of the treatment. The remaining 25 subjects demonstrated some changes in the amount ctDNA (Table 3, Fig. 2, see also Fig. 1S in the ESM). One of these subjects, patient #Pt22, experienced in the 1st day of treatment the femur fracture at the site of the metastatic lesion; the trauma was accompanied by the increase of the concentration of EGFR mutation signals in the plasma; this patient was considered not informative for further analysis.

Table 3 Serial measurement of ctDNA content in plasma of NSCLC patients in first hours after EGFR-TKI treatment initiation (the individual graphics are presented on Fig. 1S in the ESM, for the detailed clinicopathological features of patients see Table 2S in the ESM)
Full size table
Fig. 2
figure 2

Spider plots of changes in ctDNA concentrations occurring in the plasma obtained from the NSCLC patients within the first 48 h of anti-EGFR treatment

Full size image

The analysis of changes in ctDNA concentrations occurring within the first 48 h of treatment revealed a few patterns (Table 3, Fig. 2, see also Fig. 1S in the ESM). Some patients demonstrated more or less consistent decline of the ctDNA content during first two days of therapy (#Pt7, #Pt8, #Pt10). A minority of cases showed a trend towards continuous increase in the number of circulating EGFR mutant copies (#Pt21, #Pt24). There were instances of relatively steady level of ctDNA over the first 48 h (#Pt23). The majority of patients showed less consistent variations in the ctDNA contents, with a number of spikes and drops (#Pt3, #Pt4, #Pt25, #Pt30, etc.).

We further attempted to correlate, which of the ctDNA measurements provides the best correlation with the tumor response and PFS. We have conditionally chosen 25% difference between the numbers of EGFR-mutated signals as a threshold. This cut-off was evaluated by the analysis of intrapatient differences in ctDNA concentration in 8 paired blood samples obtained at 0.5 h before the treatment start and at the time of TKI administration (see Table 4S in the ESM); none of the these pairs showed difference exceeding 25%. The samples were classified for 3 groups according to change of the ctDNA content between the base-line and a given time point; accordingly, there were groups with the increased, decreased and stable concentration of tumor DNA in plasma. The measurements made at 0.5, 1, 2, 3, 6, 12, 24 and 36 h did not produce statistically significant correlations with the disease outcome (p-values (Breslow test) for comparison of Kaplan–Meier curves in different subgroups: 0.714, 0.841, 0.206, 0.798, 0.255, 0.276, 0.161 and 0.737, respectively; p-values (Fisher exact test) for tumor response rates: 0.697, 0.697, 0.283, 0.657, 0.444, 0.978, 0.408 and 0.319, respectively). However, there were clinical correlations with the change of ctDNA level registered at 48 h after the start of the treatment (see Fig. 2S in the ESM).

Changes of ctDNA concentration at 48 h are predictive for TKI clinical efficacy

Twelve (50%) out of 24 informative patients showed > 25% reduction of the plasma ctDNA concentration (median decrease: − 85%; range: from − 100% to − 49%) at 48 h after the start of treatment. All these patients demonstrated disease control after 4 and 8–12 weeks of therapy (at 4 weeks: 11 PR and 1 SD; at 8–12 weeks: 8 PR and 2 SD; two patients underwent surgery and were not evaluable by RECIST) (Tables 2, 3). One of two patients, who underwent surgery between 1st and 2nd assessments, demonstrated complete pathologic tumor response.

The remaining 12 individuals showed either stable content of circulating EGFR-mutated DNA (n = 5) or the elevation of ctDNA concentration (n = 7) at 48 h after the start of the therapy (Table 2, 3). The median increase of the ctDNA level in the latter group was 95% (range: from 36 to 276%). 10 of 12 patients with elevated or stable ctDNA level achieved an objective response at 4 weeks, but only 5 of 10 evaluable patients still demonstrated disease control at 8–12 weeks of the treatment (Fisher exact test: p = 0.032, when compared to the group with ctDNA decrease). Progressive disease (PD) at weeks 8–12 was documented in 5/10 (50%) patients, who showed increased or stable ctDNA content at 48 h after the treatment; one additional patient died before the second assessment (Tables 2, 3).

The decline of concentration of EGFR-mutated DNA in plasma, which was observed at 48 h after the start of the TKI treatment, predicted for longer PFS as compared with patients with increasing or stable level of ctDNA (14.7 months vs. 8.5 months, p = 0.013, Kaplan–Meier method; Table 2; Fig. 3).

Fig. 3
figure 3

Probability of survival in NSCLC patients with different patterns of early ctDNA dynamics during the first 48 h of anti-EGFR treatment

Full size image

Discussion

EGFR TKIs are characterized by a relatively rapid absorption, with peak plasma concentrations achieved within a few hours after the uptake of the tablet [31, 32]. The therapeutic doses of the EGFR TKIs are significantly higher than the minimal drug concentrations, which exert some antitumor effect [33,34,35]. Consequently, it is explainable, that some patients experience evident symptomatic relief within the first hours after the beginning of the treatment [26, 27, 36].

Studies of EGFR inhibition in cell lines revealed, that the administration of anti-EGFR drugs results in immediate biological consequences. Exposure to TKI causes the decrease of EGFR autophosphorylation, followed by down-regulation of ERK, AKT, STAT3 and other signaling proteins; all these events are observed within 10–30 min after the addition of TKI to the cell culture medium [37,38,39]. Activation of the apoptotic signaling cascade can also be observed within first 10 min of TKI exposure [40, 41]. The reduction of the tumor mass upon EGFR TKI therapy is likely to be attributed both to the cessation of cell proliferation and to the induction of programmed cell death [17, 42, 43]. Some data indicate that immune-related mechanisms may also contribute to the tumor shrinkage [44, 45]. Use of MRI in animal experiments revealed evidence for tumor regression occurring already within 1–7 days after TKI administration [46,47,48]. Consequently, the mere fact of the existence of rapid plasma ctDNA response to TKI treatment is in agreement with preclinical observations.

The numerical data obtained in our study correspond well to the results observed in similarly designed investigations. In particular, we were able to detect EGFR mutations in plasma in 25/30 (83%) patients at base-line, which is close to the observations made in other studies [49,50,51,52,53]. Our data also validate previous findings, which suggest that the absence of detectable EGFR-mutated copies in the plasma at base-line correlates with improved PFS [13, 51, 54]. Similarly to the reports of Riediger et al. [55] and Phallen et al. [56], we observed temporary increase of the level of ctDNA during the first hours of treatment in some although not all cases. It is unclear whether these changes are attributed to the massive tumor cell shedding in response to the drug, or caused by other reasons.

Several studies monitored ctDNA concentration in the beginning of the treatment by EGFR inhibitors. Lee et al. [51] analyzed EGFR-mutated ctDNA at 8 weeks after the TKI administration, and observed the decline of its concentration in all 40 patients analyzed. Subjects with complete clearance of ctDNA at 8 weeks had significantly longer PFS as compared to patients with residual amounts of EGFR-mutated copies in the bloodstream. Molina-Vila et al. [54] examined 74 patients at 6 weeks after the start of therapy, and observed the presence of EGFR-mutated DNA only in 3 (4%) subjects. Monitoring of ctDNA at earlier time points is significantly more complicated. Riediger et al. [55] obtained serial blood samples at 1-day intervals during the first week of therapy from a patient receiving afatinib. They observed an increase of ctDNA concentration at 26 h, and subsequent continuous decline of the number of EGFR-mutated copies starting from 48 h time point. Husain et al. [57] monitored the amount of ex19del, L858R and T790M mutation in patients, who acquired resistance to erlotinib or afatinib, and started to receive osimertinib. Serial urinary samples were obtained at 1-day intervals during the first week of therapy. Some of these patients showed temporary spikes of urine tumor DNA concentration within days 1–7, however the subsequent observation revealed a consistent decline of the amount of EGFR-mutated DNA by the end of the second week. Our study has a novelty as compared to the mentioned above investigations, as it included a relatively large number of patients and involved a serial blood-takes performed within first hours after TKI administration. The justification of this effort was based on published observations describing a very rapid treatment response in a subset of patients [26, 27, 36] as well as on the data obtained in preclinical experiments [37,38,39]. Our results suggest that good responders to TKIs can be identified already at 48 h after the start of the EGFR-targeted therapy. Some NSCLCs treated by first- or second-generation EGFR TKIs demonstrate emergence of EGFR T790M mutations before clinical disease progression [58, 59]. However, treatment-naïve tumors usually do not contain EGFR T790M mutation as base-line [60] therefore our study considered only monitoring of ex19del- and L858R-mutated ctDNA.

This study has some limitations. Blood-take at 48 h after the beginning of the treatment was the latest time point in the early ctDNA response analysis. This was due to convenience reasons, as the patients started to receive TKI while been in a hospital, and the 2 days was a period between the first tablet and the hospital discharge. While 48 h was the only informative point for clinical prediction within the range 0.5–48 h, one could expect that the analysis of ctDNA at somewhat longer time intervals could have even better predictive value. The design of our study initially considered blood-takes at 2 and 4 weeks after the beginning of TKI treatment, however the compliance of patients was incomplete and the obtained data did not provide additional information (Table 3). It is also desirable to validate the obtained findings in larger studies involving serial blood-takes from NSCLCs patients undergoing EGFR TKI therapy.

Early monitoring of ctDNA after the start of treatment allows to evaluate whether the tumor will indeed consistently respond to TKI. It appears that although the majority of EGFR-mutated tumors demonstrate some initial disease control upon TKI administration, they can broadly be divided into two categories. In approximately a half of tumors the majority of cells constituting neoplastic lumps are vulnerable to TKI exposure, and these tumors demonstrate prolonged tumor response. Another half of EGFR-mutated NSCLCs is characterized by some intratumoral heterogeneity caused by compromised access to the drug for some tumor cells or by various in-built signaling mechanisms for TKI resistance. In these NSCLCs only a fraction of cells composing the tumor mass respond to TKI, while the remaining malignant clones facilitate rapid disease progression after initial short-term disease control.

Conclusions

The present study demonstrates that the clinical response to TKI can be predicted by the analysis of changes in the plasma ctDNA concentration at 48 h after the start of EGFR-targeted therapy. It is not obvious whether the results of this plasma test could call to some action. While the prediction for good response clearly supports the continuation of the treatment, it is unclear what options can be offered to potential poor responders. There are studies demonstrating promising results of the combined use of EGFR TKI inhibitors and antiangiogenic or cytotoxic drugs [61, 62]. Perhaps, lack of rapid response to a single-agent EGFR TKI may justify the addition of another antitumor compound to the front-line therapy. Current clinical trials often involve liquid biopsy; therefore testing of this concept is compatible with the available medical resources.