Clinical concerns have been raised about making the most of specific and accurate differential diagnoses of lung cancer to reduce the false-positive rate and develop individualized treatment plans. In this study, we found that both static metabolic parameters (SUVmax), and dynamic metabolic parameters (Ki) have good diagnostic value in the differential diagnosis of lung cancer. However, the specificity can be improved when the dynamic metabolic parameter Ki is added. Another finding was that among AC patients, Ki values were lower in EGFR (+) patients than in EGFR (-) patients, and for some patients with non-small cell lung cancer (NSCLC) where EGFR testing is not available, Ki is improved discriminability.
Since FDG is not a tumor-specific imaging agent, not only malignant tumors but also granulomatous diseases such as sarcoidosis, and concurrent infectious and inflammatory diseases (tuberculosis, pneumonia, and interstitial lung disease)can exhibit FDG-avid [3–4]. As a result, uncertain PET imaging signatures could lead to unnecessary biopsies or thoracotomies for some benign pulmonary lesions with high FDG metabolism. Deppen et al. [7] concluded that, in regions with endemic infectious lung disease, the specificity of FDG PET/CT for the differential diagnosis of lung cancer was overstated (specificity of 61% [49%-72%]). In our study, 23 patients were confirmed as benign lesions (SUV range of 1.2–9.0) by surgical or puncture pathology results. Therefore, it is crucial to improve diagnostic specificity, thus allowing to operate early on malignant lesions and avoid unnecessary surgery in patients with benign lesions.
The compartmental model is considered the gold standard to quantify FDG uptake. In contrast to static imaging, dynamic acquisition was fitted to obtain quantitative information on FDG metabolism. These metabolic parameters enable a better description of the different FDG metabolism stages, thus reflecting the pathophysiological mechanisms. Huang et al. [16] concluded that in a small group of patients (N = 34), Ki can better identify benign and malignant solitary pulmonary nodules (0.004 vs 0.023 ml/g/min, P = 0.0034) in areas (Taiwan) with a high prevalence of granulomatous disease. Aleksander et al. [17] revealed that the lung malignancy group has higher Ki values than the benign group (0.0230 ± 0.0155 vs. 0.0057 ± 0.0071 ml/g/min) and could better distinguish benign from malignant (P = 0.0311). Consistent with this research, we found that both static metabolic parameters (SUVmax) and dynamic metabolic parameters (including k2, k3, and Ki) have good diagnostic value in the differential diagnosis of lung cancer. Parameter Ki was lower in the benign lesions than in the malignant lesions (0.0102 vs 0.0267 ml/g/min, P < 0.001). However, the sample sizes in the studies mentioned above were small and did not provide cut-off values for Ki.
The ROC curve analysis revealed that both the static metabolic parameter SUVmax and the dynamic metabolic parameter Ki had good diagnostic values (AUC of 0.819 and 0.830). Compared with SUVmax, the specificity of Ki has been further improved (0.870 vs. 0.999). In our study, 23 patients with SUVmax ranging from 1.2-9.0 had pathologically confirmed benign lesions after FDG PET/CT scan, and these patients had Ki values ranging from 0.0002–0.0246 ml/g/min (Fig. 5 and Fig. 6). Therefore, the specificity of the differential diagnosis can be improved when the cut-off value of Ki was 0.0250 ml/g/min, especially in FDG-avid patients. This may reduce unnecessary invasive tests/treatments.
The previous study concluded that, in lung cancer, SUVmax and Ki values of AC were lower than those of SCC (9.14 ± 1.48 vs 5.58 ± 0.62 and 0.052 ± 0.009 vs 0.029 ± 0.004 min− 1, P <0.05) [18]. Tineke et al. concluded that AC had lower k3 values than SCC in lung cancer [19]. In this study, we found that the SUVmax, k3, and Ki values in the AC group were lower than those in the SCC group, similar to previous reports.
EGFR can mediate oncogenic signals involved in the proliferation and survival of tumor cells and is expressed and activated in a variety of epithelial malignancies [20]. EGFR status has become a major prognosis factor. Previous studies have shown that treatment of patients with EGFR activating and sensitizing mutation-driven NSCLC with EGFR tyrosine kinase inhibitors (TKIs) achieved a response rate (RR) of 60–80% with a median progression-free survival (PFS) of 8–13 months [21–23]. Improved quality of life in EGFR (+) patients treated with gefitinib compared to standard chemotherapy [22–24]. In clinical practice, EGFR testing is not available for some patients or high-quality genetic testing of tumor tissue is challenging due to a variety of factors. Therefore, it is crucial to identify reliable metabolic parameters for non-invasive prediction of EGFR status based on FDG PET/CT imaging.
Previous studies related to the prediction of the status of EGFR based on SUVmax are numerous but yielded unsatisfactory results. Huang et al. [25] concluded that higher SUVmax values in lung adenocarcinoma patients are more likely to develop EGFR mutations. Subsequently, it has also been concluded that lower SUVmax values are associated with EGFR mutations in patients with NSCLC [26, 27]. Carlos Caicedo et al. [28] concluded that the presence of EGFR mutations was not correlated with FDG uptake. In our study, in the AC group, the SUVmax did not show such a difference between EGFR (+) and EGFR (-) groups. However, we found that Ki values were lower in the EGFR (+) group than in the EGFR (-) group (0.0279 vs. 0.0405 ml/g/min) and the differences were statistically significant (P = 0.032). For ROC analysis, Ki had a cut-off value of 0.0350 ml/g/min for predicting EGFR status, a sensitivity of 0.710, a specificity of 0.588, and an AUC of 0.674 (Fig. 7 and Fig. 8). Therefore, the addition of the dynamic metabolic parameter Ki provides more imaging and metabolic information and is expected to be a means of non-invasive de-prediction of the status of EGFR. In particular, patients in clinical work who are unable or unavailable for EGFR testing are likely to benefit from it.
Our study had several limitations. First, in this study, we have a small percentage of patients in the benign and SC groups, so the main studies have focused on the AC group. In the future, we will expand the sample size to continue related studies. Second, motion correction was not considered in this study. It is known that motion in the chest region can affect not only the SUV but also the kinetic parameters quantification [29–32]. Dedicated quality control and motion correction process may be required to obtain an accurate quantification before proceeding with the evaluation. Third, SUVmax rather than SUVmean was used in this study as we thought SUVmax was more stable and less affected by the partial volume effects [33–35].