Classification of Epidermal Growth Factor Receptor Gene Mutation Status Using Serum Proteomic Profiling Predicts Tumor Response in Patients with Stage IIIB or IV Non-Small-Cell Lung Cancer

Objectives Epidermal growth factor receptor (EGFR) gene mutations in tumors predict tumor response to EGFR tyrosine kinase inhibitors (EGFR-TKIs) in non-small-cell lung cancer (NSCLC). However, obtaining tumor tissue for mutation analysis is challenging. Here, we aimed to detect serum peptides/proteins associated with EGFR gene mutation status, and test whether a classification algorithm based on serum proteomic profiling could be developed to analyze EGFR gene mutation status to aid therapeutic decision-making. Patients and Methods Serum collected from 223 stage IIIB or IV NSCLC patients with known EGFR gene mutation status in their tumors prior to therapy was analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) and ClinProTools software. Differences in serum peptides/proteins between patients with EGFR gene TKI-sensitive mutations and wild-type EGFR genes were detected in a training group of 100 patients; based on this analysis, a serum proteomic classification algorithm was developed to classify EGFR gene mutation status and tested in an independent validation group of 123 patients. The correlation between EGFR gene mutation status, as identified with the serum proteomic classifier and response to EGFR-TKIs was analyzed. Results Nine peptide/protein peaks were significantly different between NSCLC patients with EGFR gene TKI-sensitive mutations and wild-type EGFR genes in the training group. A genetic algorithm model consisting of five peptides/proteins (m/z 4092.4, 4585.05, 1365.1, 4643.49 and 4438.43) was developed from the training group to separate patients with EGFR gene TKI-sensitive mutations and wild-type EGFR genes. The classifier exhibited a sensitivity of 84.6% and a specificity of 77.5% in the validation group. In the 81 patients from the validation group treated with EGFR-TKIs, 28 (59.6%) of 47 patients whose matched samples were labeled as “mutant” by the classifier and 3 (8.8%) of 34 patients whose matched samples were labeled as “wild” achieved an objective response (p<0.0001). Patients whose matched samples were labeled as “mutant” by the classifier had a significantly longer progression-free survival (PFS) than patients whose matched samples were labeled as “wild” (p=0.001). Conclusion Peptides/proteins related to EGFR gene mutation status were found in the serum. Classification of EGFR gene mutation status using the serum proteomic classifier established in the present study in patients with stage IIIB or IV NSCLC is feasible and may predict tumor response to EGFR-TKIs.


Introduction
Lung cancer is the leading cause of cancer-related death worldwide [1]. Non-small-cell lung cancer (NSCLC) is the most common histologic type of the disease and accounts for approximately 80% of lung cancers [2]. Because more than 70% of patients with lung cancer are diagnosed with advanced-stage disease [3], systemic treatment plays an important role in clinical management. Chemotherapy has been the cornerstone of treatment for NSCLC for many years. However, epidermal growth factor receptor tyrosine kinase inhibitors (EGFR-TKIs), such as erlotinib, gefitinib and icotinib, have been shown to greatly improve clinical outcomes and safety when compared with chemotherapy in some patients with advanced NSCLC [4][5][6][7][8]. EGFR-TKI sensitivity has been associated with activating mutations in the kinase domain of the EGFR gene, especially an exon 19 deletion and mutations in exon 21(L858R) and exon 18 (G719X) [9][10][11]. All EGFR gene TKI-sensitive mutations result in activation of the EGFR tyrosine kinase domain, which is the target of EGFR-TKIs. Therefore, patients with these EGFR gene TKI-sensitive mutations have a significantly better response to EGFR-TKIs, whereas those with wild-type EGFR genes exhibit a worse tumor response. Assessment of EGFR gene mutation status is critically important for therapeutic decision-making.
National comprehensive cancer network (NCCN) guidelines state that DNA mutational analysis in tumor cells is the preferred method to assess EGFR gene mutation status. However, in some cases, tumor tissue either is inadequate for molecular testing because of its small quantity or very low tumor content or is not readily available [3]. Several groups have detected EGFR gene mutations in DNA isolated from plasma [3,[12][13][14][15][16] or serum samples [17,18], which serve as substitutes for tumor tissue; some groups have demonstrated a correlation between mutation status in the plasma/serum and tumor tissue [3,12,13,[15][16][17][18]. Furthermore, EGFR gene mutations detected in plasma or serum may be predictive of the response to EGFR-TKIs [3,13,14,16,18]. However, the methods used to assess EGFR gene mutation status in plasma or serum samples are not approved by the current guidelines. Thus, other sensitive and noninvasive approaches for evaluating EGFR gene mutation status using surrogate tumor tissues to predict EGFR-TKI efficacy are still needed.
Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) is a sensitive, rapid, inexpensive, and simple technique for proteomic analysis of complex biological samples, such as tissue, urine and blood [19][20][21][22][23][24][25][26]. Peaks in the mass spectrum correspond to ions formed from relatively abundant species in the sample, predominantly peptides and proteins. Recently, peptide mass fingerprinting based on MALDI-TOF-MS has been widely used to detect diagnostic, prognostic, and predictive proteomic biomarkers. In recently published studies, peptide mass fingerprinting has been successfully applied to analyze serum from patients and healthy controls to detect differences in peptides/proteins; these differences were used to develop classification algorithms for disease diagnosis [22][23][24][25]. In addition, peptide mass fingerprinting can detect differences in serum/plasma peptides/proteins between subgroups of patients with same type of disease. Taguchi [26] and Wu [27] used MAL-DI-TOF-MS to analyze serum and plasma from NSCLC patients; they observed subtle differences in serum/plasma peptides/proteins between two subgroups that experienced significantly different EGFR-TKI efficacies and developed classification algorithms using differential peptides/proteins to predict the efficacy of EGFR-TKI in NSCLC patients. Because the efficacy of EGFR-TKI has been associated with EGFR gene mutation status, the constituting peptides/proteins of the serum/plasma classification algorithms developed by Taguchi and Wu to predict EGFR-TKI efficacy may be associated with EGFR gene mutation status [27,28].
In this study, we aimed to detect serum peptides/proteins associated with EGFR gene mutation status and test whether a classification algorithm based on serum proteomic profiling could be developed for analysis of EGFR gene mutation status to assist in therapeutic decisionmaking. To accomplish this, we applied peptide mass fingerprinting using MALDI-TOF-MS coupled with ClinProTools software to analyze serum from 223 NSCLC patients with a known EGFR gene mutation status (i.e., determined by amplification refractory mutation system [ARMS] in tumor tissue) and detect differences in serum peptides/proteins between NSCLC patients with EGFR gene TKI-sensitive mutations and NSCLC patients with wild-type EGFR genes. We developed a serum proteomic classifier to evaluate EGFR gene mutation status and tested the classifier on an independent validation group. We also analyzed correlations between EGFR gene mutation status as identified by the serum proteomic classifier and response to EGFR-TKIs to test the potential utility of EGFR gene mutation status identified by the serum proteomic classifier for predicting clinical responses to EGFR-TKI treatment.

Patients and samples
To be eligible for the study, patients were required to have pathologically confirmed stage IIIB or IV NSCLC, an Eastern Cooperative Oncology Group performance status of 0 to 2, predefined EGFR gene mutation status in tumor tissues based on ARMS (scorpions amplification refractory mutation system, Qiagen, Germany) prior to therapy, and available serum. Only patients treated at 307 Hospital of PLA from May 2011 to April 2013 were enrolled. This study was performed according to protocols approved by the local ethical committee (the Ethics Committees of 307 Hospital, PLA), and all the patients provided written informed consent to participate in this study and gave permission for the use of their blood samples. For the tumor response assessment, we evaluated objective responses after 8 weeks of treatment on the basis of computed tomography (CT) scans. Tumor response was determined according to RECIST 1.0. Overall survival (OS) was defined as the time from the date of lung cancer diagnosis to the date of death. Progression free survival (PFS) was defined as the time from the start of EGFR-TKI treatment to the date of disease progression or death from any cause. The cutoff date for follow-up was November 10, 2014. Smoking status was based on records from the patients' first clinic visits, and people who had smoked more than 100 cigarettes in their lifetime were considered smokers. Laboratory data were obtained and recorded independently by investigators who were blinded to the clinical data until the analyses were completed by a biostatistician.
Fifty patients were randomly selected from patients with EGFR gene TKI-sensitive mutations and wild-type EGFR genes respectively (a total of 100 patients) to form the training group for the detection of differences in serum peptides/proteins between NSCLC patients with EGFR gene TKI-sensitive mutations and NSCLC patients with wild-type EGFR genes, and the generation of the classification model, and the remaining patients formed the validation group to test the model.
The patients fasted overnight. All blood samples were collected before the patients received first-line treatment. Blood samples were collected in vacuum blood collection tubes containing coagulant and separation gel and centrifuged at 3000 rpm for 10 min at 4°C to separate the serum. The supernatant was divided into 100-μl aliquots and stored at −80°C until processing.

Peptidome isolation
Serum samples were thawed on ice and fractionated with weak cation exchange magnetic beads (MB-WCX, National Center of Biomedical Analysis, China). The samples were processed following three steps: binding, washing and elution. For each analysis, 5 μl of beads washed three times in 50 μl of binding solution (National Center of Biomedical Analysis, China), 20 μl of binding solution and 5 μl of sample were added into an Eppendorf tube and incubated for 10 min at room temperature. The tube was placed on a magnetic bead separation device to isolate the peptidome. The supernatant was removed, and the beads were washed three times with 100 μl of washing solution (National Center of Biomedical Analysis, China) to discard unbound proteins. Finally, the beads were washed with 20 μl of eluting solution (National Center of Biomedical Analysis, China) to acquire bound proteins for MALDI-TOF-MS analysis.
Mass spectrometry analyses were performed on an Ultraflex III MALDI-TOF-MS (Bruker Daltonics, Germany). The operating conditions were as follows: linear positive ion mode; repetition rate, 200 Hz; ion source voltages, 25 and 23.50 kV; lens voltage, 6.5 kV; pulsed ion extraction time, 100 ns. For matrix suppression, we used a high gating factor with signal suppression of up to 300 m/z. For each spectrum, 3000 shots were acquired manually from six random positions over the surface of the spot (i.e., 500 shots per position). Data acquisition was carried out at 43% of the maximum laser energy. Each spectrum was externally calibrated. Peaks in the m/z range of 800-10,000 Da were recorded with the FlexControl acquisition software v3.4 (Bruker Daltonics, Germany).

Bioinformatics
Spectral processing. ClinProTools software v2.1 (Bruker Daltonics, Germany) was used to automatically process MALDI-TOFMS spectra data using data preparation settings according to the following standard workflow: Each raw spectrum was normalized to its total ion current; all the spectra were recalibrated using the prominent, common m/z values; baseline subtraction, smoothing, and peak detection were performed; and peak areas for each spectrum were calculated. The signal-to-noise ratio was set at 5 for peak detection. Peak areas were calculated using zero level integration type. Spectra were also ''top hat" baseline subtracted with the minimum baseline width set to 10%, smoothed and processed in the 800-10,000 Da range.
Training and classification model establishment in the training group. Only spectra from the training group were used. Differences in peptide peaks between patients with EGFR gene TKI-sensitive mutations and patients with wild-type EGFR genes were selected using peak areas on the basis of statistical differences. Built-in mathematical models in ClinProTools 2.1 (i.e., genetic algorithm (GA), supervised neural network (SNN) algorithm and quick classifier (QC) algorithm) were then used to select peptide peaks and set up classification models to determine the optimal separation planes between samples from patients with EGFR gene TKIsensitive mutations and wild-type EGFR genes. After each model was generated, a random cross-validation process was carried out with the software, and the percent to leave out and number of iterations were set at 20 and 10, respectively.
To determine the accuracy of the class prediction model, the software quantifies crossvalidation and recognition capability. Cross-validation is a measure of the reliability of a model and can be used to predict how a model will behave in the future. This method is used for evaluating the performance of an algorithm for a given data set and under a given parameterization. Recognition capability describes the performance of an algorithm, i.e., the proper classification of a given data set.
Blind test of the classification model that most efficiently separated samples from patients with EGFR gene TKI-sensitive mutations from samples from patients with wild-type EGFR genes in the validation group. This validation was performed in a blinded manner in that MALDI-TOF-MS analysis was performed and samples were classified before the clinical outcome data were made available to the investigators.
For each patient from the validation groups, a corresponding spectrum was presented to the selected classification model (named classifier), which then returned a label, either "mutant" (i.e., classification to class consisting of samples from patients with EGFR gene TKI-sensitive mutations) or "wild" (i.e., classification to class consisting of samples from patients with wildtype EGFR gene), or output a message that the spectrum was unclassifiable. The results from the selected classification model were compared with findings from ARMS in tumors to estimate the separation efficiency of the model.

Statistical analysis
The clinical and disease characteristics between different arms, the objective response rate (ORR) and disease control rate (DCR) between patients whose matched samples were labeled as "mutant" and "wild" were compared using a χ 2 or Fisher's exact test. The concordance between ARMS in tumors and the serum proteomic classifier in evaluating EGFR gene mutation status was assessed using a Kappa test. Survival curves were estimated by the Kaplan-Meier method, and differences between curves were evaluated by the log-rank test. Statistical analyses were performed with SPSS software, v19.0 (SPSS Inc., USA). A p-value less than 0.05 was considered statistically significant.

Patient Characteristics
A total of 223 patients met the enrollment criteria and were enrolled in this study. Based on the criterion of ARMS in tumors, there were 102 patients with EGFR gene TKI-sensitive mutations and 121 patients with wild-type EGFR genes. Fifty patients were randomly selected from those with EGFR gene TKI-sensitive mutations and from those with wild-type EGFR genes (i.e., a total of 100 patients) to form the training group, and the remaining 123 patients (i.e., 52 patients with EGFR gene TKI-sensitive mutations and 71 with wild-type EGFR genes) formed the validation group. The clinical and disease characteristics of all the patients are listed in Table 1. The patients were balanced between the training group and the validation group (Table 2). In the training group, there were no significant differences between patients with EGFR gene TKI-sensitive mutations and wild-type EGFR genes with respect to age, histologic type, or disease stage, but differences in sex and smoking history were observed between these two arms, with more females and more non-smokers in patients with EGFR gene TKI-sensitive mutations ( Table 3).
Differences of peaks in serum between patients with EGFR gene TKIsensitive mutations and patients with wild-type EGFR genes in the training group A total of 129 peptide peaks were identified in the spectra of the training group data set generated by MALDI-TOF-MS, and 9 peaks were significantly different (p<0.05) between the patients with EGFR gene TKI-sensitive mutations and patients with wild-type EGFR genes (      with wild-type EGFR genes (p<0.00001). Therefore, these two peaks (m/z 4092.4, x axis; m/z 4585.05, y axis) were plotted in a 2D peak distribution view (Fig 1).

Classification model establishment
Three algorithms, GA (optimized by adjusting the number of neighbors for a k-nearest neighbor classification), SNN and QC, were applied for classification model construction using spectral data from the training group generated by MALDI-TOF-MS. The recognition capability and cross-validation of the models are presented in Table 5, and the Model GA-7 (named classifier), which was composed of five peptide peaks with m/z 4092.4, 4585.05, 1365.1, 4643.49 and 4438.43, exhibited the best efficiency in separating samples from patients with EGFR gene TKI-sensitive mutations and samples from patients with wild-type EGFR genes, with a recognition capability of 93.32% and a cross-validation of 81.23% (Fig 2).

Blinded test of the classifier in the validation group
The classifier was then validated in an independent validation group of 123 NSCLC patients in a blinded test ( Table 6). Three of the 123 samples yielded unclassifiable spectra (i.e., one sample was from a patient with EGFR gene TKI-sensitive mutation, and two were from patients with wild-type EGFR genes, as confirmed by ARMS in tumors). Among the 52 samples from patients with EGFR gene TKI-sensitive mutations confirmed by ARMS in tumors, 44 (84.6%) were labeled as "mutant" by the serum proteomic classifier; and among the 71 samples from patients with wild-type EGFR genes confirmed by ARMS in tumors, 55 (77.5%) were labeled as "wild" by the serum proteomic classifier, achieving an overall accuracy of 80.5%, with a sensitivity of 84.6% and a specificity of 77.5%, which indicated a high consistency between ARMS in tumors and the serum proteomic classifier in evaluating EGFR gene mutation status (P<0.001;

Correlation between EGFR gene TKI-sensitive mutations identified by the classifier and the therapeutic effect of EGFR-TKIs in the validation group
In the validation group, three of the 123 samples yielded unclassifiable spectra, and the three corresponding patients were excluded from the analysis. Among the remaining 120 patients, 81 had measurable tumors and received EGFR-TKI treatment. The clinical and disease characteristics of these 81 patients are presented in Table 7, and the median follow-up time of these patients was 29.0 months (range, 7.0 to 40.0 months). Patients whose matched samples were labeled as "mutant" and "wild" by the classifier exhibited different tumor responses to EGFR-T-KIs; these responses are listed in Table 8. Twenty-eight (59.6%) of 47 patients whose matched samples were labeled as "mutant" by the classifier and 3 (8.8%) of 34 patients whose matched samples were labeled as "wild" by the classifier exhibited an objective response (p<0.0001). Disease control was noted in 41 (87.2%) of 47 patients whose matched samples were labeled as "mutant" by the classifier and 12 (35.3%) of 34 patients whose matched samples were labeled as "wild" by the classifier (p<0.0001). Kaplan-Meier survival plots of PFS and OS for patients whose matched samples were labeled as "mutant" and "wild" by the classifier are shown in Fig  3. The median PFS time for patients whose matched samples were labeled as "mutant" and "wild" by the classifier were 10.0 months (95% CI, 9.0 to 10.9) and 2.3 months (95% CI, 1.9 to 2.7), respectively. Patients whose matched samples were labeled as "mutant" by the classifier had a significantly longer PFS than patients whose matched samples were labeled as "wild" by the classifier (p = 0.001, log-rank test, Fig 3A). Patients whose matched samples were labeled as "mutant" by the classifier had an OS time of 29.0 months (95% CI, 25.2 to 32.8) compared with 28.0 months (95% CI, 17.7 to 38.3) for the patients whose matched samples were labeled as "wild-type" by the classifier. There was no significant difference in OS between the two groups (p = 0.441, log-rank test, Fig 3B).

Discussion
The assessment of EGFR gene mutation status in tumor tissue has important predictive value and can be used to select therapies for the treatment of NSCLC. Many patients with advanced Classification of EGFR in NSCLC and metastatic NSCLC are diagnosed with small biopsies or by fine needle aspiration of tumors, which often yields insufficient DNA for evaluating EGFR gene mutation status. Noninvasive approaches of evaluating EGFR gene mutation status using substitutes for tumor tissues would be of value for patients in whom sufficient tumor tissue is not available [16]. Table 9   Table 7. Clinical and disease characteristics of patients enrolled in the analysis of EGFR-TKI therapeutic effects in the validation group.
Characteristics Total (N = 81) Labeled as "mutant" by the classifier (N = 47) Labeled as "wild-type" by the classifier (N = 34) Age, y  Table 8. Tumor response in patients whose matched samples were labeled as "mutant" and "wild" by the classifier in the validation group. provides details on the various methods used in previous reports to detect EGFR gene mutations in serum/plasma samples [3,[12][13][14][15][16][17][18]. Depending on the technique, the concordance between EGFR gene mutation status in tumor and plasma/serum samples ranges from 66% to 100%, with the highest correlation index being reported for denaturing high-performance chromatography [3] and mutant-enriched PCR [13]. However, these methods of assessing EGFR gene mutation status in plasma or serum samples are not widely used to guide EGFR-TKI therapy in clinical practice because of their inferior sensitivity when compared with findings from tumor tissue or the fact that studies examining these methods have utilized small sample sizes. In this study, we found that among 123 patients in the validation group, assessments of EGFR gene mutation status using serum proteomic classifiers yielded results that were concordant with the results of ARMS in tumors in 80.5% of the cases, with a high sensitivity of 84.6%.  Classification of EGFR in NSCLC However, we did find EGFR gene mutations that were successfully identified by only one method (i.e., serum proteomic classifier or ARMS in tumors) in 17.1% of patients in the validation group (i.e., 11.4% [14/123] by the serum proteomic classifier only, 5.7% [7/123] by ARMS in tumors only). It is important to note that cases in which test results were inconsistent between the serum proteomic classifier and ARMS in tumors cannot be considered as conventional "false-negatives" or "false-positives." One possible explanation for this inconsistency in the determination of mutational status is heterogeneity of genetic abnormalities in the tumors. In such instances, tumor biopsy specimens might not carry the EGFR gene mutations identified by the serum proteomic classifier because these classifier-constituting peptides/proteins related to EGFR gene mutation status could be derived from different parts of the tumor. The lower tumor cell content in some of the tumors might also contribute to the lack of detectable mutations. Similarly, either there is little or no classifier-constituting peptides/proteins related to EGFR gene mutation status being shed in the blood in a given case, or the quantity of peptides/ proteins in the serum is affected by certain conditions, such as inflammation, the classification of EGFR gene mutation status based on serum proteomic profiling might be impeded despite the presence of mutations in tumors.
EGFR encoded by the wild-type EGFR gene is a transmembrane tyrosine kinase receptor with a molecular weight of 170 kDa. The difference between EGFR encoded by EGFR gene with TKI-sensitive mutations and EGFR encoded by wild-type EGFR gene is that the former harbors activating tyrosine kinase domain. These two EGFRs should have similar molecular weights. Due to the high molecular weight, it is important to note that the MALDI-TOF-MS described in this study is neither suited for directly detecting EGFR encoded by EGFR gene with TKI-sensitive mutations nor EGFR encoded by wild-type EGFR gene because the typical observable mass range is 800-10000 Da. Instead, we detected the differential peptide/protein profiles between EGFR encoded by EGFR gene with TKI-sensitive mutations and EGFR encoded by wild-type EGFR gene. The identities of the constituting peptides/proteins are unknown at present; it is possible that they are unknown co-expressed peptides/proteins with low molecular weights involved, or that we detected fragments of EGFR or other high molecular weight proteins, such as proteins from the EGFR signaling pathway [29]. It is well known that tumor cell dissemination and apoptotic processes in tumors and at tumor-tissue boundaries involve changes in the proteolytic activities of a series of different proteases that may lead to the formation of protein fragments, thus providing a strong correlation with tumor tissue, and that as well serve as a basis for tumor differentiation and prognosis [29][30][31][32]. In agreement with this assumption, the proteins that have been identified thus far from blood samples by MALDI--TOF-MS have largely been degradation products of larger proteins [29,[33][34][35][36].
We also analyzed the potential implications of EGFR gene mutation status, as identified by the serum proteomic classifier, for predicting clinical outcomes in patients with NSCLC who received EGFR-TKIs. Our findings of a correlation between EGFR gene mutations identified by the classifier and tumor response to EGFR-TKI treatment and such treatment's lack of impact on OS were also consistent with previous studies in which EGFR gene mutation status was tested in tumor tissue [4][5][6][7][8]. In patients treated with EGFR-TKIs in the validation group, 59.6% of the patients whose matched samples were labeled as "mutant" responded to EGFR-TKIs, whereas 8.8% of the patients whose matched samples were labeled as "wild" also responded. Although no difference in OS was observed between patients whose matched samples were labeled as "mutant" and "wild", patients whose matched samples were labeled as "mutant" had significantly longer PFS after EGFR-TKI treatment, which suggests that these patients might have benefitted from the treatment. It should be noted that our study was not specifically designed to test EGFR-TKI treatment and that many patients received other chemotherapeutic agents, which makes data interpretation difficult. Additional clinical studies with specifically defined treatment regimens and larger sample sizes are necessary.
Tumor-based assays require well-preserved biopsy material, are technically difficult, incur substantial costs, and have a slow turnaround time. By contrast, the MALDI-TOF-MS method that we have described can be performed using less than 1 μl of pretreatment serum. Additionally, this method is inexpensive and rapid, and it can easily be fully automated. In our study, the assessment of EGFR gene mutation status using the serum proteomic classifier produced results that were not completely consistent with those obtained with ARMS in tumors. However, the inability to obtain primary tumor tissues, particularly through repeated biopsies, from patients with advanced-stage lung cancer makes the use of a serum proteomic classifier for analysis of EGFR gene mutation status clinically important given the high sensitivity (84.6%) of the technique and the favorable response to EGFR-TKIs in patients whose matched samples were labeled as "mutant" by the serum proteomic classifier.
One limitation of our analysis is the inability of the serum proteomic classifier to precisely determine the type of EGFR gene TKI-sensitive mutation, such as exon 19 deletion (E19del [LREA deletion]) and exon 21 mutation (L858R). Several studies have demonstrated that patients with an exon 19 deletion experienced, on average, longer PFS and OS than those with an L858R mutation after first-line EGFR-TKI treatment for advanced non-small cell lung cancer [37,38], indicating the clinical significance of the type of EGFR gene TKI-sensitive mutation. Therefore, our serum proteomic classifier must be modified to enable it to determine the type of EGFR gene TKI-sensitive mutation. Another limitation is the unknown biology underlying the correlation of these features with EGFR gene mutation status. Identification and analysis of the informative peaks might lead to important insights into the mechanisms underlying the correlation, and these studies are underway.
In conclusion, in this study, we detected differences in serum peptides/proteins between patients with EGFR gene TKI-sensitive mutations and patients with wild-type EGFR genes; based on these differences, a classification algorithm was developed for the analysis of EGFR gene mutation status. Furthermore, EGFR gene mutation status, as determined by the serum proteomic classifier, may be predictive of the response to EGFR-TKIs. All of the above provide evidence to suggest that a serum proteomic classifier may be used instead of tumor tissue for analysis of EGFR gene mutation status in NSCLC. It will be important to validate these findings and determine the value of the assay in predicting patients' responses to TKIs in randomized trials with larger cohorts.