Abstract
Aim
Since the importance of epidermal growth factor receptor (EGFR) and Kirsten rat sarcoma viral oncogene homolog gene (KRAS) mutation status in predicting treatment response in non-small cell lung cancer (NSCLC) patients is well known, we aimed to evaluate whether initial fluorine-18-fluorodeoxyglucose (18F-FDG) positron emission tomography-computed tomography (PET/CT) imaging could non-invasively predict EGFR or KRAS mutation states in this patient group.
Methods
This retrospective observational study examined patients with NSCLC who underwent 18F-FDG PET/CT for staging from August 2021 to January 2024. Age, sex, smoking status, pathological data, EGFR and KRAS mutation status, and metabolic and volumetric PET parameters were recorded. Groups were based on gene mutation status as follows: EGFR-mutations (mt) vs. EGFR wild-type (EGFR-wt) and KRAS-mt vs. KRAS-wt.
Results
Ninety-nine patients with a mean age of 62.96±9.66 (range: 37-87) were included. The EGFR-mt group had lower metabolic tumor volume (MTV) (p=0.015) and total lesion glycolysis (TLG) (p=0.017) values. MTV had an area under the receiver operating characteristic curve (AUC) of 0.667 [95% confidence interval (CI): 0.547-0.788, p=0.015], and with a ≤24.9 cut-off, yielded 60.87% sensitivity, 68.42% specificity, and 66.67% accuracy to detect EGFR-mt. For TLG, the AUC was 0.664 (95% CI: 0.540-0.788, p=0.017) and a ≤408.1 cut-off yielded 86.96% sensitivity, 43.42% specificity, 53.54% accuracy, and 91.67% NPV. KRAS-mt was detected in 34 (34.34%) patients, and there were no significant differences between the KRAS-mt and KRAS-wt groups in terms of PET parameters.
Conclusion
Primary tumor parameters derived from initial 18F-FDG PET/CT can predict EGFR mutation status but not KRAS mutation status. The high negative predictive value of TLG can be used to rule out EGFR-mt status, possibly preventing unnecessary treatments in patients without favorable genetic properties, especially when genetic analyses are not possible.
Introduction
Lung cancer is one of the leading causes of cancer-related deaths worldwide, with approximately 85% of patients suffering from non-small-cell lung cancer (NSCLC) (1). Tyrosine kinase inhibitors (TKIs) have been shown to improve NSCLC outcomes (2), and it has been established that a kinase domain mutation of the epidermal growth factor receptor (EGFR) is associated with a favorable response to TKIs. In fact, progression-free survival (PFS) is longer with the use of TKIs compared with chemotherapy in patients with EGFR mutations (EGFR-mt) (3). The Kirsten rat sarcoma viral oncogene homolog gene (KRAS) encodes a member of the small GTPase superfamily that has an impact on the EGFR pathway. Mutations in this gene (KRAS-mt) are associated with an unfavorable response to TKIs (4).
Fluorine-18-fluorodeoxyglucose (18F-FDG) positron emission tomography-computed tomography (PET/CT) a non-invasive molecular imaging method that is widely used for staging, response assessment, and recurrence detection in NSCLC (5). The EGFR mutation status plays an important role in the management of patients with advanced NSCLC; regardless of other factors, the molecular profiling of EGFR is essential to guide clinical treatment (6). However, obtaining high-quality tumor tissue for EGFR-mt testing is difficult in many cases, given the shortage of biopsy samples and the physical condition of patients. Therefore, a non-invasive and simple method to identify EGFR-mt is necessary to inform treatment decisions. Recent studies, the results of which are still controversial, have focused on whether EGFR-mt and KRAS-mt states can be associated with metabolic parameters obtained via 18F-FDG PET/CT, especially the maximum standardized uptake value (SUVmax) (5, 7). Few reports have explored the utility of metabolic tumor volume (MTV) and total lesion glycolysis (TLG) values in this context, and these studies have revealed inconsistent findings (8).
We hypothesized that initial 18F-FDG PET/CT parameters obtained from the primary lesion could be utilized to predict the presence or absence of EGFR-mt or KRAS-mt in patients with NSCLC.
Methods
Study Design and Patients
This study was approved by the Clinical Research Ethics Committee of University of Health Sciences Turkey, Basaksehir Cam and Sakura City Hospital (date: July 2023, approval no.: 2023.07.296). Informed consent was obtained from all participants included in the study. Diagnostic and therapeutic procedures were performed in accordance with national guidelines and the principles of the Declaration of Helsinki (1964). All patients received appropriate information and provided written informed consent.
This retrospective observational cohort study examined patients with NSCLC who underwent 18F-FDG PET/CT imaging for staging purposes at baseline between August 2021 and January 2024. The inclusion criteria were as follows: pathologically confirmed diagnosis of NSCLC, being treatment-naïve before 18F-FDG PET/CT imaging, confirmation of EGFR-mt and KRAS-mt status within 1 month from the time of imaging, and availability of all relevant clinical and imaging data. The exclusion criteria were having received any antitumor therapy before 18F-FDG PET/CT, having a history of other malignancies, having comorbidities affecting the metabolic parameters examined by 18F-FDG PET/CT (such as diabetes mellitus), and having pneumonia or other infections that might confound the analyses (Figure 1).
Age, sex, smoking characteristics (pack years), histopathology, clinical stage, primary tumor site, nodal involvement, metastasis, operation history, molecular analysis results concerning EGFR and KRAS (mutations), and metabolic and volumetric parameters derived from 18F-FDG PET/CT were recorded. The cancer stage was determined according to the 8th TNM classification for lung and pleural tumors (9).
Patients were classified according to mutation status as EGFR-mt and EGFR wild type (EGFR-wt) or KRAS-mt and KRAS wild type (EGFR-wt). The specimens were defined as EGFR-mt if mutations were identified in exons 18, 19, 20, and 21. The presence of KRAS-mt was determined by detecting mutations in KRAS codons 12, 13, and 61.
18F-FDG PET/CT Procedure
Imaging was performed after at least 6 h of fasting and the presence of a glucose level of <150 mg/dL. After 18F-FDG injection, patients were left to rest for 50 min before imaging with the Ingenuity TF 64 scanner (Philips Medical Systems, OH, USA). Low-dose CT was performed with the following settings: 113 mAs, 120 kV, and 4-mm section thickness. The PET images were recorded in the caudocranial axis on the identical transverse field of view for a duration of 3 min per bed, and corrections for attenuation were based on the initial CT images. All obtained images (PET, CT, corrected, and uncorrected) were assessed on maximum intensity projection images as well as transaxial, coronal, and sagittal cross-sectional images. Reconstructions were performed according to the EANM procedure guidelines for tumor imaging (version 2.015) (10).
Imaging Assessment
For the evaluation of images, a volume of interest (VOI) including all relevant tissue was delineated on attenuation-corrected 18F-FDG PET/CT images of the primary tumor (automated contouring and manual correction) in the axial, sagittal, and coronal planes (Figure 2).
SUVmax was defined as the highest SUV measured from any voxel within the VOI. The mean SUV (SUVmean) was defined as the SUV inside the VOI. The MTV was recorded with a threshold of 40% (of the SUVmax) within the VOI, and TLG was calculated by multiplying the MTV by the SUVmean. To normalize FDG uptake, three spherical regions of interest (ROIs) with a diameter of 3 cm were placed in sites with homogenous FDG uptake inside the right lobe of the liver, and the mean value was calculated (liver-SUVmean). The lung-to-liver ratio (SUVmax/liver-SUVmean) was then calculated to generate normalized SUV data. All evaluations and calculations were performed by two nuclear medicine physicians who were blinded to the study data.
Pathologic Evaluation
Genomic DNA was extracted from formalin-fixed, paraffin-embedded non-small-cell lung carcinoma tissues, and EGFR-mt were detected using real-time PCR. Tissues were sectioned to 5 µm, deparaffinized, and then subjected to genomic DNA analysis using a DNA sample preparation kit according to the manufacturer’s instructions. Following quantification of genomic DNA, real-time PCR was carried out to amplify the target area and detect the targeted mutations in exon 18 (G719A, G719C, and G719S), 19 (deletions and complex mutations), 20 (S768I, T790M, and insertions), and 21 (L858R and L861Q) using the EGFR mutation test (v2) on the Cobas® z480 analyzer, which were automatically analyzed and output by Cobas® 4800 software. For KRAS, we also used real-time PCR to detect targeted mutations. Additionally, we detected mutations in codons 12 and 13 of exon 2 and 61 of exon 3 in the KRAS gene using the KRAS mutation test on Cobas® z480 analyzer, which were again automatically analyzed and collected using the same software.
Statistical Analysis
All analyses were performed on IBM SPSS Statistics for Windows, Version 25.0 (IBM Corp., Armonk, NY, USA). For the normality check, the Kolmogorov-Smirnov test was used. Descriptive statistics were presented by using mean ± standard deviation for normally distributed continuous variables, median (25th percentile-75th percentile) for non-normally distributed continuous variables, and frequency (percentage) for categorical variables. Normally distributed variables were analyzed with the Student’s t-test. Non-normally distributed variables were analyzed using the Mann-Whitney U test. Categorical variables were analyzed using the chi-square tests, the Fisher-Freeman-Halton test, or Fisher’s exact test. Prediction performances were assessed using receiver operating characteristic (ROC) curve analysis. The optimal cut-off points were determined using the Youden index. Logistic regression analyses were performed to identify factors independently associated with EGFR and KRAS mutations. Variables were initially analyzed by univariate regression analysis, and those showing significance were included in the multivariate model. Detection of p<0.05 values was accepted to show statistically significant.
Results
We included 99 patients (77 males and 22 females) in our study; the mean age was 62.96±9.66 (range 37-87) years. Seventy-six (84.44%) patients had a smoking history. Most cases (71.72%) involved adenocarcinoma. Fifty-one (51.52%) patients were stage T4, 45 (45.45%) were stage N3, 54 (54.55%) had metastasis, and 54 (54.55%) were clinical stage IV. The right lung (63.64%) and upper lobe (53.54%) were the most common sites of primary lesions. The bone was the most common site of metastasis (31.31%). Ten (10.10%) patients underwent surgical treatment. EGFR and KRAS mutations (EGFR-mt and KRAS-mt) were detected in 23 (23.23%) and 34 (34.34%) patients, respectively (Table 1).
Association between EGFR status, clinical features, and PET/CT results
Female frequency (p<0.001) was significantly higher in the EGFR-mt group than in the EGFR-wt group. Smoking frequency (p<0.001) and pack years (p=0.003) were significantly lower in the EGFR-mt group than in the EGFR-wt group. We found no significant differences between the EGFR mutation groups in terms of age, histopathology, T stage, N stage, M stage, clinical stage, side, lobe, and surgery percentages (Table 2).
MTV (p=0.015) and TLG (p=0.017) were significantly lower in the EGFR-mt group than in the EGFR-wt group. We found no significant differences between the EGFR mutation groups in terms of SUVmean, SUVmax, liver-SUVmean, and normalized SUV (Table 2).
When we evaluated the EGFR-mt prediction performance of the PET parameters, MTV had 60.87% sensitivity, 68.42% specificity, 66.67% accuracy, 36.84% positive predictive value (PPV), and 85.25% negative predictive value (NPV) for a cut-off value of 24.9 (equal or lower values represent the presence of EGFR mutation). The area under the ROC curve (AUC) was 0.667 [95% confidence interval (CI): 0.547-0.788, p=0.015]. TLG had 86.96% sensitivity, 43.42% specificity, 53.54% accuracy, 31.75% PPV, and 91.67% NPV for a cut-off value of 408.1 (equal or lower values represent the presence of EGFR mutation). The AUC was 0.664 (95% CI: 0.540-0.788, p=0.017) (Figure 3).
The SUVmean, SUVmax, and normalized SUV were non-significant in distinguishing patients with or without EGFR-mt (Table 3).
According to the multivariable logistic regression analysis, female sex was the only factor independently associated with EGFR-mt (OR: 12.882, 95% CI: 2.922-56.796, p=0.001) (Table 4).
The baseline PET/CT images of a patient with EGFR-mt infection are presented in Figure 4, and those of a patient with EGFR-wt infection are presented in Figure 5.
Association between KRAS status, clinical features, and PET/CT results
Adenocarcinoma (p=0.001) percentage was significantly higher in the KRAS-mt group than in the KRAS-wt group. We found no significant differences between the KRAS mutation groups in terms of age, sex, smoking status, pack years, T stage, N stage, M stage, clinical stage, side, lobe, or surgical treatment. In addition, there were no significant differences between the KRAS mutation groups in terms of SUVmean, SUVmax, MTV, TLV, liver-SUVmean, and normalized SUV (Table 5).
Logistic regression analysis revealed that adenocarcinoma was the only factor associated with the presence of KRAS-mt (OR: 10.667, 95% CI: 2.351-48.396, p=0.002).
Discussion
In light of the advances and diversity in the treatment of NSCLC, such as TKIs, it is crucial to identify favorable mutations in the early stages of the disease. Although genetic analysis is undoubtedly the best approach, the use of molecular profiling may be limited due to various factors, including difficulty in obtaining sufficient tumor tissue, unavailability of genetic analyses, and the invasiveness of the procedure (4, 6). The current study reported that it may be possible to obtain information regarding the presence/absence of relevant mutations by evaluating metabolic parameters obtained from 18F-FDG PET/CT imaging performed at baseline. MTV and TLG were found to have statistical significance in distinguishing patients with and without EGFR-mt, but not KRAS-mt. Although the overall accuracy values were not excellent for distinguishing the presence of EGFR mutations, TLG had exceptional sensitivity and NPV, indicating notable utility in detecting and ruling out the presence of favorable EGFR mutations among patients with NSCLC.
In our study, similar to the literature, female sex, low smoking percentage, as well as PET parameters such as MTV and TLG obtained from the primary lesion, were significantly lower in patients with EGFR-mt compared with those without (7, 11, 12). Additionally, the MTV and TLG values obtained from the initial imaging of the primary lesions were significantly lower in the EGFR-mt group.
This relationship is likely due to the effect of EGFR on glucose transporters in the cell membrane via downstream pathways that may affect tumor glucose metabolism. In previous studies, different results were presented when examining the role of SUVmax in predicting EGFR mutation in patients with NSCLC.
Most studies have reported higher frequencies of EGFR mutations in patients with lower SUV values (12, 13). However, a study consisting of Asian patients with advanced lung adenocarcinoma reported that patients with higher SUVmax values were more likely to carry EGFR-mt (14). Wang et al. (15) also reported significantly higher SUVmax values in patients with NSCLC with EGFR mutations compared with those without mutations, which they attributed to increased glucose uptake. These varying results suggest that SUVmax values in the present study may be affected by a multitude of factors, such as patient characteristics and imaging technique. As such, we believe that incorporating parameters like MTV and TLG could be beneficial for the evaluation and management of this patient group. Additionally, ethnic differences in patient groups and other uncontrolled genetic mutations in the EGFR-wt group may also cause these differences. Liu et al. (16), supporting this perspective and similar to our study, found that MTV values were lower in the EGFR-mt. However, the authors did not find a correlation between SUVmax and mutation status, thereby providing credibility to the perceived impact of similar underlying reasons. The authors suggested that this may be due to the semiquantitative nature of SUV data, which might vary depending on the PET scanner, fasting time, plasma glucose level, and selected ROIs (16).
In contrast to our findings, Minamimoto et al. (8) found that SUVmax values of primary lesions were predictive of EGFR-mt, whereas MTV and TLG were not, suggesting that gene mutations were unassociated with tumor size or volume. However, most of the patients included in their study (70.2%) had clinical stages of IA or IB, whereas 91% of the patients in our study had stage III and IV. The difference in the severity of patients is an important factor that could explain the contrasting results (8).
Our study demonstrated no correlation between 18F-FDG uptake and KRAS-mt status, which is consistent with several other studies previously reported in the literature (8, 17). Interestingly, Caicedo et al. (18), in their study enrolling patients with stage III and IV NSCLC, reported that SUV values were significantly higher in the KRAS-mt group than in the KRAS-wt group. Further studies with larger cohorts are required to clarify the associations between PET parameters and KRAS-mt states.
Since there are many variables affecting SUVmax values, such as body size and amount of tracer injected, standardization of SUVmax by assessing liver SUV data, as well as other approaches, is known to reduce variability. It has been reported to improve prognosis prediction and treatment response assessment in NSCLC (19). Mak et al. (20) reported that normalized SUVmax values of the primary tumor (normalized for SUV of blood in the pulmonary artery) were predictive of EGFR mutation. In our study, similar to the SUV values, there was no significant difference between the mutation groups when we used normalized SUV values. Since the utility of this approach has not been validated in this particular patient group, more studies are needed to determine whether normalized SUV values differ from other metabolic parameters in predicting genetic mutations in NSCLC.
In our study, based on the exceptionally high NPV of TLG in predicting EGFR-mt (91.6% for a cut-off point of 408.1), it may be feasible to suggest that early 18F-FDG PET/CT imaging data can be used to primarily predict the absence of EGFR mutations, thereby facilitating the noninvasive identification of patients that will not respond to TKI in the early stages of the disease. With this approach, treatment options suitable for the appropriate patient group can be determined, and unnecessary treatments can be prevented when genetic analyses are unavailable or cannot be performed. Prospective studies with a larger number of patients are needed to confirm this finding.
Study Limitations
One of the limitations of our study was its retrospective design, which included a relatively small number of patients. Second, we could not report the prognostic value of 18F-FDG PET/CT because the prognosis of the patients was not yet determined. Despite these limitations, our study has identified a non-invasive 18F-FDG PET/CT parameter that can predict patients with NSCLC who are unlikely to respond to treatment.
Conclusion
Our study revealed that it is possible to noninvasively predict the presence/absence of EGFR mutations in the early period of the disease using the MTV and TLG values extracted from baseline 18F-FDG PET/CT imaging. Since the sensitivity and NPV of TLG were exceptionally high, it may be possible to rule out the presence of EGFR mutations using this approach. It is evident that these results are relevant in cases in which genetic detection is not feasible and could prevent unnecessary treatments by identifying patients without EGFR-mt who do not demonstrate favorable response to TKIs.
