Genetic profiling of tumors in patients with lung cancer provides the guidance of precise targeted therapy.1,2 Tissue biopsy is the gold standard for detecting tumor genetic alterations, yet its applications are quite limited in clinical practice, due to insufficient tissue samples for testing, unreachable lesions by conventional tumor biopsy, existing contraindications to tissue biopsy, etc.3,4 Previous studies reported that about 30% of tumor biopsy for next-generation sequencing (NGS) analyses were failed due to insufficient tissue.5 Liquid biopsy, such as plasma-based biopsy, can be an indispensable complementary modality to tissue biopsy via detecting genetic alterations of ctDNA (circulating tumor DNA), because of its advantages in safety, sparing unnecessary side effects, convenience, repeated sampling, and overcoming tumor heterogeneity.6 Although, the FDA has already approved the plasma test for the detection of EGFR mutations in lung cancer using PCR or NGS, the plasma-based NGS in detecting driver gene alterations had imperfect concordance rates of about 70.0%–80.0% with tissue-based NGS in non-small cell lung cancer (NSCLC) patients,7–12 which might affect the utility of plasma genotyping. Therefore, there is an urgent need to develop a reliable and convenient tool to predict the concordance between plasma and tissue biopsies to guide the appropriate utility of liquid biopsy for genotyping in advanced lung adenocarcinoma patients.
Previous studies showed that serum tumor markers, such as carcinoembryonic antigen (CEA), neuron-specific enolase (NSE), carbohydrate antigen (CA) 19-9, CA242, and cytokeratin fragment antigen 21-1 (CYFRA21-1), expressed in normal cells are abundantly synthesized and secreted in tumor cells.13 The levels of those serum tumor markers are frequently used to diagnose lung cancer, monitor therapeutic effects, and predict prognosis in clinical practice.13,14 In addition, the levels of serum tumor markers were reported to be related to the presence of driver gene mutations in tumor tissues15–20 and reflect the tumor burden of patients.7,13 However, there are few studies on the correlation between serum tumor markers and the concordance of plasma and tissue biopsies in detecting driver genes.
Herein, we evaluated the predictive value of serum tumor markers for the concordance between plasma and tissue biopsies in detecting driver mutations in advanced lung adenocarcinoma patients and constructed a predictive model for the concordance. We aimed to provide a feasible tool to accurately identify concordance rate between plasma and tissue biopsy based on NGS in lung adenocarcinoma patients based on serum tumor markers. For the patients who present higher concordance rate, the application of plasma biopsy could be prior to tissue biopsy and therefore avoid the operative risk of tissue biopsy. What is more, for the population with lower concordance rate, tissue biopsy is more recommendable than plasma biopsy, in order to reduce the economic burden caused by unnecessary plasma biopsy.
MATERIALS AND METHODS PatientsIn total, 1012 patients diagnosed with stage III or IV lung adenocarcinoma were retrospectively enrolled at Chongqing Daping Hospital, Hunan Cancer Hospital, and Guangzhou First People's Hospital in China (Figure S1). These patients underwent plasma biopsy for NGS between December 2015 and March 2021. Of them, 519 patients voluntarily underwent paired plasma and tissue biopsies at the same time.12,21 All 1012 patients were tested for serum tumor markers within 1 week of plasma biopsy and before a new round of treatment, including CEA, NSE, CYFRA21-1, and CA125. The clinical characteristics of patients including gender, age, smoking history, and treatment status were collected according to medical records system of the hospital. Clinical staging was based on the eighth edition of the TNM classification of lung cancer. This study followed the Standards for Reporting Diagnostic Accuracy (STARD) reporting guideline and was approved by the Research Ethics Committee of Daping Hospital. This study includes no data deposited in external repositories.
Tissue DNA and Cell-free DNA (cfDNA) were extracted using QIAamp DNA FFPE Tissue Kit (Qiagen) at the same time. Capture-based targeted sequencing was performed on tumor and plasma samples from 3 hospitals using a cancer gene panel spanning 237Kb of the human genome (Burning Rock Biotech), which covers the whole exons and selected introns of 168 genes, consisting of 68 lung cancer-related genes and 100 genes related to other cancers (full description was in Supplementary Methods).
Normalizing the concordance between plasma and tissue biopsy genomic profilesConcordance was defined as that a driver alteration could be detected in both plasma and tissue samples. Discordance was defined as that a driver alteration could be detected only in plasma or tissue samples.
Detection of serum tumor markersWithin 1 week after plasma biopsy, 3 ml of fasting venous blood was collected in the morning. Then, serum was separated for testing the levels of serum tumor markers by electrochemiluminescence immunoassay. The range of normal serum tumor markers referred to CEA 0–5.0 ng/ml, NSE 0–16.3 ng/ml, CA125 0–35 U/ml, and CYFRA21-1 0–3.3 ng/ml.
Statistical analysisAll data were analyzed using the R software version 4.0.3. Continuous data conforming to the normal distribution were presented as mean ± standard deviation, and those nonconforming were presented as median (interquartile range). Categorical variables were presented as frequencies. Normality was evaluated using the Shapiro–Wilk test. In addition, Student's t-test was used for normally distributed continuous variables, and the Mann–Whitney U-test was performed for continuous non-normally distributed variables. Categorical variables were analyzed using the chi-square test. The correlation between two non-normally distributed continuous variables was assessed by determining Spearman's rank correlation coefficient. Receiver operating characteristic (ROC) curve analyses were performed, and the method of Delong was employed to calculate the standard error of the area under the curve (AUC).22 The least absolute shrinkage and selection operator (LASSO) method was applied to identify independent predictors of concordance between plasma and tissue biopsies. The prediction accuracy of the nomograms was evaluated by the C-index.23,24 The value of the C-index ranged from 0.0 (perfect discordance) to 1.0 (perfect concordance), and a value of 0.5 is equivalent to a random chance. Nomograms analysis was conducted for the prediction of the concordance rate between plasma and tissue samples.24 Bootstrapping techniques were used for internal validation of the prognostic models, and the calibration of nomograms was assessed graphically by plotting the actual probabilities versus the nomogram-predicted probabilities.23,24 The Cochran–Armitage trend and Wilcoxon signed-rank test were used to performing the trend test. Values of p <0.05 were considered statistical significance unless otherwise specified.
RESULTS Patient clinical and genetic characteristicsA total of 1012 patients with advanced lung adenocarcinoma were included in this study. The median age was 60 years (range 53–67) and 53.2% (538/1012) were males. The proportion of patient with distant metastases was 85.1% (861/1012). Pleura metastasis were the most frequent metastatic sites (based on the data in Table S1). Patients were divided into two groups according to the sample type: 519 patients for both plasma and tissue biopsies, and 493 patients for only plasma biopsy. Statistically significant differences between the two groups were observed in age and T stage (p < 0.05, Table S1), but not in sex, smoking history, distant metastasis, or clinical stages (Table S1). Nine driver genes and corresponding driver alterations recommended by the National Comprehensive Cancer Network (NCCN) guidelines with clinical therapeutic significance for lung cancer (details of driver genes are shown in Table S2) were selected to study whether the ctDNA mutational profiling based on NGS technology can accurately reflect the tumor genomic profile and overcome tumor heterogeneity.25–27 For plasma samples, the mutation rate of nine driver genes reached 61.9% (626/1012), and the top detected alterations were EGFR mutation (44.2%, 447/1012), KRAS mutation (8.3%, 84/1012), and ALK fusion (4.2%, 43/1012). For 519 tissue samples, the mutation rate of all driver genes was 79.6% (413/519). The top-three detected alterations were consistent with plasma-based biopsy with different ratios, that is, EGFR mutation (49.1%, 255/519), KRAS mutation (10.6%, 55/519), and ALK fusion (43/519, 8.3%). In addition, three cases (0.6%, 3/519) of NTRK fusion were detected in tissue samples but not in plasma samples (Figure 1).
Of 519 patients, 381 patients had positive results of 9 driver genes in both plasma and tissue samples, reaching an overall concordance rate of 73.4% (Table 1). For the rest patients, driver gene mutation was only detected in plasma samples of 21 patients (4.1%, 21/519) (group A), of which 8 cases (38.1%, 8/21) were carrying EGFR T790M, and driver gene mutations were exclusively detected in tissue samples of 111 patients (111/519, 21.4%) (group B). Further univariate analysis showed that the concordance rate was significantly higher in males, M1 stage disease, and stage IV disease than in females, M0 stage disease, and stage III disease, respectively (p < 0.05, Table 1). Patients with bone metastases had the highest concordance rate among patients with distant metastases (p < 0.05, Table 1). The levels of CEA (14.68 ng/ml vs. 3.88 ng/ml, p < 0.001), CYFRA21-1 (5.56 ng/ml vs. 3.25 ng/ml, p < 0.001), NSE (15.61 ng/ml vs. 12.03 ng/ml, p < 0.001), and CA125 (58.73 U/ml vs. 29.94 U/ml, p < 0.001) in the concordance group were significantly higher than in the discordance group (Table 1; Figure S2). Interestingly, the levels of CEA (24.50 ng/ml vs. 3.90 ng/ml, p < 0.001) and CYFRA21-1 (7.22 ng/ml vs. 2.97 ng/ml, p < 0.001) in group A were significantly higher than in group B (Figure S3). We further conducted deciles analyses of four tumor markers in the paired samples and showed that the increased concordance rate was significantly associated with the increased level of serum tumor markers (Cochran–Armitage test for trend. CEA, p < 2.200 × 10−16; CYFRA21-1, p < 2.200 × 10−16; CA125, p = 3.316 × 10−8; NSE, p = 2.951 × 10−9; Figure 2). Compared with CA125 and NSE, the increase trend of concordance rate in CEA and CYFRA21-1 was more pronounced. In the subgroups of CEA ≥15.45 ng/ml and CEA ≥65.72 ng/ml, the concordance rates were more than 81% and 96%, respectively. In the subgroup of CYFRA21-1 ≥ 5.88 ng/ml, the concordance rate was more than 81% (Figure 2).
TABLE 1 Characteristics of samples paired with plasma and tissue biopsies
| Characteristic | Paired tissue and plasma samples (N = 519) | Concordance (N = 381) | Discordance (N = 138) | p |
| Age (years), median (IQR) | 60.0 (51.0–66.0) | 60.0 (51.0–66.0) | 60.5 (51.8–66.0) | 0.433 |
| Gender, n (%) | 0.025 | |||
| Male | 283 (54.5) | 219 (57.5) | 64 (46.4) | |
| Female | 236 (45.5) | 162 (42.5) | 74 (53.6) | |
| Smoking status | 0.153 | |||
| Never | 312 (60.1) | 222 (58.3) | 90 (65.2) | |
| Current or former | 207 (39.9) | 159 (41.7) | 48 (34.8) | |
| T stage, n (%) | 0.405 | |||
| T1 | 71 (13.7) | 47 (12.3) | 24 (17.4) | |
| T2 | 126 (24.3) | 92 (24.1) | 34 (24.6) | |
| T3 | 92 (17.7) | 68 (17.8) | 24 (17.4) | |
| T4 | 170 (32.8) | 125 (32.8) | 45 (32.6) | |
| Unknown | 60 (11.6) | 49 (12.9) | 11 (8.0) | |
| N stage, n (%) | 0.074 | |||
| N0 | 68 (13.1) | 43 (11.3) | 25 (18.1) | |
| N1 | 38 (7.3) | 24 (6.2) | 14 (10.1) | |
| N2 | 182 (35.1) | 133 (34.9) | 49 (35.5) | |
| N3 | 221 (42.6) | 174 (45.7) | 47 (34.1) | |
| Unknown | 10 (1.9) | 7 (1.8) | 3 (2.1) | |
| M stage, n (%) | 0.000 | |||
| M0 | 88 (17.0) | 48 (12.6) | 40 (29.0) | |
| M1 | 431 (83.0) | 333 (87.4) | 98 (71.0) | |
| Stage, n (%) | 0.000 | |||
| III | 88 (17.0) | 48 (12.6) | 40 (29.0) | |
| IV | 431 (83.0) | 333 (87.4) | 98 (71.0) | |
| Metastatic site, n (%) | 0.036 | |||
| Contralateral lung | 179 (34.5) | 139 (36.5) | 40 (29.0) | |
| Pleura | 187 (36.0) | 144 (37.8) | 43 (31.2) | |
| Bone | 188 (36.2) | 167 (43.8) | 21 (15.2) | |
| Brain | 103 (19.9) | 85 (22.3) | 18 (13.0) | |
| Liver | 54 (10.4) | 45 (11.8) | 9 (6.5) | |
| Pericardium | 50 (9.6) | 44 (11.5) | 6 (4.3) | |
| Adrenal gland | 51 (9.8) | 45 (11.8) | 6 (4.3) | |
| Others | 20 (3.9) | 17 (4.5) | 3 (2.2) | |
| Therapy, n (%) | 0.155 | |||
| 1st line | 490 (94.0) | 363 (95.3) | 127 (92.0) | |
| ≥2nd line | 29 (6.0) | 18 (4.7) | 11 (8.0) | |
| CEA (ng/ml) | 9.01 (3.34–47.89) | 14.46 (4.79–68.13) | 3.88 (1.74–8.99) | <0.001 |
| NSE (ng/ml) | 14.57 (8.21–21.67) | 15.64 (8.93–22.91) | 11.99 (5.02–17.54) | <0.001 |
| CYFRA21-1 (ng/ml) | 4.75 (2.99–9.28) | 5.56 (3.59–11.15) | 3.25 (2.30–4.62) | <0.001 |
| CA125 (U/ml) | 45.92 (17.89–129.05) | 58.68 (20.79–142.05) | 29.70 (14.22–70.85) | <0.001 |
In order to identify patients who had a high concordance or discordance rate between plasma and tissue biopsies, thresholds of serum tumor markers in patients with a high PPV (positive predictive value reflecting the concordance rate in positive serum tumor marker screening test results) or NPV (negative predictive value for the discordance rate in the negative serum tumor marker screening test results). We found that CEA thresholds were 15.01 ng/ml and 51.15 ng/ml, and CYFRA21-1 thresholds were 6.21 ng/ml and 23.72 ng/ml in patients with the PPV of 90% and 95%, respectively (Table 2; Figure S4A,B). Furthermore, for patients with NPV of 50%, 45%, and 40%, the thresholds of CEA were 3.26 ng/ml, 5.19 ng/ml, and 7.10 ng/ml, and the CYFRA21-1 thresholds were 2.98 ng/ml, 3.97 ng/ml, and 4.81 ng/ml, respectively (Table 2; Figure S4C,D). However, different from CEA, the NPV decreased with the decrease of CYFRA21-1 level when it reached 50%.
TABLE 2 The predictive threshold value with high PPV and high NPV of concordance
| Serum tumor marker | Cut-off of concordance (%) | Threshold value (ng/ml) | Proportion of patients (%) | Sensitivity (%) (95% CI) | Specificity (%) (95% CI) |
| CEA | PPV | ||||
| 90.00 | 15.01 | 40.67 | 49.61 (44.52–54.71) | 84.78 (77.73–90.35) | |
| 95.00 | 51.15 | 24.47 | 31.50 (26.91–36.46) | 95.65 (90.85–98.47) | |
| 100.00 | 256.70 | 6.74 | 8.92 (6.30–12.20) | 100.00 (97.41–100.00) | |
| NPV | |||||
| 40.00 | 7.10 | 44.12 | 63.78 (58.74–68.62) | 66.67 (58.13–74.52) | |
| 45.00 | 5.19 | 34.87 | 73.75 (69.05–78.13) | 59.42 (50.71–67.72) | |
| 50.00 | 3.26 | 24.08 | 83.73 (79.59–87.28) | 45.65 (37.24–54.31) | |
| CYFRA21-1 | PPV | ||||
| 90.00 | 6.21 | 37.76 | 46.19 (41.11–51.27) | 86.96 (80.23–92.11) | |
| 95.00 | 23.72 | 6.36 | 7.87 (5.46–11.05) | 98.55 (94.91–99.79) | |
| 100.00 | 40.40 | 3.85 | 4.99 (3.02–7.72) | 100.00 (97.36–100.00) | |
| NPV | |||||
| 40.00 | 4.81 | 50.48 | 58.79 (53.72–63.82) | 76.81 (68.92–83.64) | |
| 45.00 | 3.97 | 38.73 | 70.87 (66.01–75.43) | 65.94 (57.43–73.81) | |
| 50.00 | 2.98 | 24.66 | 83.20 (79.15–86.85) | 47.10 (38.67–55.83) | |
Abbreviations: NPV, negative predictive value; PPV, positive predictive value.
Multivariate analysis predicts concordance between plasma biopsy and tissue biopsyThe LASSO regression analysis showed that several independent factors were associated with the concordance, including male (odds ratio [OR] (95% confidence interval [CI]): 1.85 (1.18, 2.93), p = 0.008), M1 stage disease (OR (95% CI): 1.65 (0.94, 2.88), p = 0.079), bone metastasis (OR (95% CI): 3.10 (1.78, 5.57), p < 0.001), increased CEA level (every 10 ng/ml increment, OR (95% CI): 1.53 (1.28, 1.93), p < 0.001), increased CYFRA21-1 level (every 10 ng/ml increment, OR (95% CI): 2.62 (1.55, 4.91), p = 0.001), increased NSE level (every 10 ng/ml increment, OR (95% CI): 1.29 (1.01, 1.66), p = 0.049) (Figure S5). We then analyzed the predictive efficacy of serum tumor markers and their association with multiple independent factors through ROC curves. In 519 paired plasma-and-tissue samples, the area under ROC curve (AUC) of CEA, CYFRA21-1, NSE, and CA125 was 0.741, 0.727, 0.621, and 0.611, respectively (Figure 3A). The individual model of CEA or CYFRA21-1 had a larger AUC than the other two serum tumor markers (p < 0.001, Figure 3B). The AUC of CEA combined with CYFRA21-1 was equal to that of the combination of 4 serum tumor markers and smaller than that of LASSO regression model, but there were no statistical differences in the predictive efficacy among them (p > 0.05, Figure 3A,B). The efficacy of CEA combined with CYFRA21-1 in predicting the concordance was higher than that of any individual serum tumor marker (p < 0.05, Figure 3B). Collectively, the combination of four serum tumor markers and LASSO regression model are the most efficient in predicting concordance between plasma and paired tissue samples. Next, a nomogram model based on the four serum tumor markers was constructed to predict the concordance rate between plasma and tissue samples (Figure S6), which revealed a comparable performance in predicting the concordance to CEA combined with CYFRA21-1. Therefore, only CEA combined with CYFRA21-1 were included in the optimized the nomogram (Figure 3C). Internal validation and decision curve analysis of clinical benefit shows that this model had clinical utility along with stable diagnostic efficiency (Figure S7).
CEA level was correlated with the mutation detection rate of EGFR in plasma (r = 0.39, p < 0.05, Figure 4A), and no more associations of driver gene mutations with serum tumor markers were observed (Figure 4A; Figure S8). Both CEA (r = 0.47, p < 0.01) and CYFRA21-1 level (r = 0.45, p < 0.05, Figure 4A) were significantly correlated with the maximum variant allele frequency (maxVAF) of ctDNA. In addition, the AUC of maxVAF in predicting concordance was 0.836 (Figure S9). The relationship between the CEA level and mutation rate of EGFR in total plasma samples, paired-plasma samples, and plasma-only samples was subsequently explored. The CEA level was positively correlated with EGFR mutation rate in each group (p < 0.001, Figure S10A). EGFR mutation rate in the plasma sample of CEA-positive patients with the CEA level >5 ng/ml was significantly higher than that in CEA-negative patients (54% vs. 24%, p < 0.001, Figure S10B). Then, the CEA level was stratified in equal increasing multiples in plasma-paired and plasma-only sample groups, the EGFR mutation rate increased gradually with the increase of CEA level in both groups (p for trend <0.001, Figure 4B). The Wilcoxon signed-rank test showed that the change trend of EGFR mutation rate was comparable between the two groups, and there was no significant difference in EGFR mutation rates in each same CEA range (p = 0.168, Figure 4B).
Tissue-based biopsy is regarded as the gold standard for the detection of genetic alterations in clinical applications. However, the limited quantity and quality of tumor tissue hindered its clinical operations. The utility of liquid biopsy appropriately could complement the shortage of qualified tumor samples in clinic. In this study, we analyzed the correlation of four serum tumor markers and the concordance rates of plasma and tumor biopsies. The level of CEA 15.01 ng/ml or CYFRA21-1 6.21 ng/ml could be the threshold of 90% PPV of the concordance rate, and CEA 3.26 ng/ml or CYFRA21-1 2.98 ng/ml could be the threshold of 50% NPV of the discordance rate. We constructed a predictive model based on the level of both CEA and CYFRA21-1 to evaluate the utility of plasma samples to diagnose genetic alterations of advanced lung cancer based on the high concordance or discordance of plasma and tumor biopsies.
In clinical applications, the tumor tissue might be unavailable due to the low compliance or the difficulties in operation. Even if the sample was obtained, the quality was another major concern. In the Checkmate 227 study, the tissue sample was collected in 94.8% (1679/1739) patients. However, only 57.7% (1004/1739) of the tumor sample was left for tumor mutational burden study.28 Also other studies reported that about 70% of tumor samples were qualified for NGS.7 With our nomogram model (Figure 3C), the clinical physicians could calculate a predictive score based on the level of CEA and CYFRA21-1, which were part of the routine exam for lung cancer patients. For example, the total score was about 24 in a patient with CEA of 50 ng/ml and CYFRA21-1 of 15 ng/ml, who was then predicted to have a concordance rate higher than 90% based on the nomogram model. For this patient, plasma could be used prior to tissue samples for tumor genotyping.
In this study, we first used the cancer-gene NSG panel to analyze 9 lung cancer driver genes in 519 patients with paired plasma and tissue samples and found the concordance of plasma and tissue biopsies was 73.4%, similar to previous studies.8–12,29,30 We also found that the higher concordance rate of plasma and tissue biopsies were correlated with more malignant cancer stages and males. Patients with bone metastases showed the highest concordance rate in the distant metastases, implying a preference for plasma biopsy in clinic.
Serum tumor markers are abnormally expressed by malignant tumor cells, or derived from the host response to tumor, which can reflect the occurrence and development of tumor and monitor tumor response to treatment.31 In our study, we found that the level of serum tumor markers, CEA, CYFRA21-1, NSE, and CA125, was all associated with the concordance rate of plasma and tissue biopsies. This funding is consistent with previous studies on the correlation between the level of CEA, CA125, and CYFRA21-1 and the stages of lung adenocarcinoma. All three tumor markers were correlated with TNM stage of lung adenocarcinoma32,33 and serum tumor markers are associated with tumor burden,32–34 implying the correction of serum tumor marker with both tissue and plasma samples.
The further exploration of the relationship between the levels of linear changes in serum tumor markers and concordance rates indicates that the concordance rate increased with the increased level of serum tumor marker, especially CEA and CYFRA21-1. The thresholds of CEA and CYFRA21-1 in predicting concordance rates of 90% to 95% and discordance of 40% to 50%. We found that CEA and CYFRA21-1 threshold was 15.01 ng/ml and 7.43 ng/ml in patients having a PPV of 90%, respectively. We also found that a predictive model of CEA combined with CYFRA21-1 outperformed individuals of CEA, CYFRA21-1, NES, and CA125, showing similar predictive efficacy to the combination of the four tumor markers and the LASSO model. CEA combined with CYFRA21-1 was also recommended for assisting the diagnosis of lung adenocarcinoma,32 while some studies reported that CEA was expressed higher in squamous cell cancer.35 Similarly, a previous study on paired tissue and plasma biopsy study of 185 patients showed that a scoring system based on the serum tumor markers, CYFRA21-1, and CA19-9 predicted more than 90% concordance between paired tissue and plasma samples,36 indicting the use of liquid biopsy. Although CEA was also included in this study,36 different experiment designs could be the main reason for the differences between two models, as the author pre-classified lung cancer driver genes and 8 different driver genes were analyzed primarily in their study.
Furthermore, we found that the EGFR mutation rate in plasma increased with the increment of CEA level, consistent with previous research results in tissues.37,38 However, no correlation was observed between other serum tumor markers and mutation rates of other driver genes. EGFR mutation rate in the group of plasma-only samples was comparable with that in the group of paired-plasma samples at the same CEA level, suggesting that CEA has similar efficacy in predicting concordance in plasma-only and paired-plasma samples. Therefore, patients with high levels of CEA should be recommended for detecting EGFR mutation in plasma.
Through a multicenter clinical study with large samples, we found that serum tumor markers, especially CEA combined with CYFRA21-1, could effectively predict the concordance rates between plasma and tissue biopsies, and provided a predictive model of serum tumor markers for clinicians to quickly and accurately identify patients who are suitable for plasma biopsy for genotyping first. This study had some limitations. First, it was a retrospective real-world study. Second, only four serum tumor markers, which were used to monitor diseases in the clinical diagnosis and treatment in all three hospitals, were included. Other markers associated with lung cancer were not included in this study such as tissue polypeptide antigen (TPA). Moreover, changes in the predictive efficacy of serum tumor marker during the course of treatment were not further explored. A well-designed, prospective study is needed to stratify patients according to tumor stage, histological subtype, and treatment line to validate our findings in future work.
CONCLUSIONSIn summary, CEA combined with CYFRA21-1 can effectively predict the concordance rate of plasma and tumor biopsies in advanced lung adenocarcinoma patients. Elevated CEA levels also could predict the presence of EGFR mutations in plasma. Our study provides a nomogram model based on CEA and CYFRA21-1 as a convenient tool for individualized prediction of the concordance of plasma and tumor biopsies, to optimize the utility of plasma biopsy in genotyping for advanced lung adenocarcinoma patients.
AUTHOR CONTRIBUTIONSChenlong Zhao: Conceptualization (equal); formal analysis (equal); methodology (equal); visualization (equal); writing – original draft (lead). JiangHua Li: Conceptualization (equal); formal analysis (equal); investigation (equal); software (equal); writing – original draft (equal). Yongchang Zhang: Data curation (equal); project administration (equal). Rui Han: Data curation (equal); formal analysis (equal); resources (equal). Yubo Wang: Data curation (equal). Li Li: Data curation (equal). Yimin Zhang: Data curation (equal); investigation (equal). Mengxiao Zhu: Methodology (equal); validation (equal); writing – original draft (supporting). Jie Zheng: Data curation (equal). Haiwei Du: Methodology (equal); writing – review and editing (equal). Chen Hu: Data curation (equal). Nong Yang: Data curation (equal); project administration (equal). Chengzhi Zhou: Data curation (equal); project administration (equal). Shangli Cai: Writing – review and editing (equal). Yong He: Conceptualization (equal); methodology (equal); project administration (equal); resources (equal); writing – review and editing (equal).
FUNDING INFORMATIONThis work was supported by the National Natural Science Foundation of China (81972189, 81802293, 82172623), the Chongqing Science Technology Commission (cstc2021jcyj-msxmX0014), Key Technology Project for Prevention and Control of Major Diseases in Chongqing (2019ZX002), Clinical medical technology innovation ability training program (2019CXLCA003), Clinical Medical Research Talent Training Program of Army Medical University (2018XLC1015).
CONFLICT OF INTERESTThe authors declare that they have no competing interests.
DATA AVAILABILITY STATEMENTAll data generated or analyzed during this study are included in this published article (and its supplementary information files).
ETHICS STATEMENTTissue and plasma samples of patient were obtained with informed consent, under the protocol approved by the Research Ethics Committee of Daping Hospital.
CONSENT FOR PUBLICATIONAll patients involved in our study obtained written consent for publication.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2023. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
Background
Plasma and tissue biopsy have both used for targeting actionable driver gene mutations in lung cancer, whose concordance is imperfect. A reliable method to predict the concordance is urgently needed to ease clinical application.
Methods
A total of 1012 plasma samples, including 519 with paired-tissue biopsy samples, derived from lung adenocarcinoma patients were retrospectively enrolled. We assessed the associations of several clinicopathological characteristics and serum tumor markers with the concordance between plasma and tissue biopsies.
Results
When carcinoembryonic antigen (CEA) levels were higher than thresholds of 15.01 ng/ml and 51.15 ng/ml, the positive predictive value of concordance reached 90% and 95%, respectively. When CEA levels were lower than thresholds of 5.19 ng/ml and 3.26 ng/mL, the negative predictive value of concordance reached 45% and 50%. The performance of CYFRA21-1 in predicting concordance was similar but inferior to CEA (AUC: 0.727 vs. 0.741,
Conclusions
CEA combined with CYFRA21-1 could effectively predict the concordance between plasma and tissue biopsies, which could be used for evaluating the priority of plasma and tissue biopsies for gene testing to timely guide clinical applications in advanced lung adenocarcinoma patients.
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
Details
; Han, Rui 1 ; Wang, Yubo 1 ; Li, Li 1 ; Zhang, Yimin 1 ; Zhu, Mengxiao 1 ; Zheng, Jie 1 ; Du, Haiwei 4 ; Hu, Chen 1 ; Zhou, Chengzhi 5
; Yang, Nong 3 ; Cai, Shangli 4 ; He, Yong 1
1 Department of Respiratory Disease, Daping Hospital, Army Medical University, Chongqing, China
2 Department of Intensive care unit, Daping Hospital, Army Medical University, Chongqing, China
3 Department of Medical Oncology, Lung Cancer and Gastrointestinal Unit, Hunan Cancer Hospital/The Affiliated Cancer Hospital of Xiangya School of Medicine, Central South University, Changsha, China
4 Burning Rock Biotech, Guangzhou, China
5 Respiratory Medicine Department, State Key Laboratory of Respiratory Disease, National Clinical Research Center for Respiratory Disease, Guangzhou Institute of Respiratory Health, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, China