Our study highlights that [18F]FAPI adds value to [18F]FDG PET/CT for diagnosing lymph node metastases in stage I-IIIA non-small cell lung cancer. Among patients undergoing [18F]FDG PET/CT, the presence of one to three risk factors identified an equivocal diagnostic zone. For this condition, the SUVmax of [18F]FAPI provided pivotal information leading to an accurate diagnosis. LNs with an [18F]FAPI SUVmax<6.2 were diagnosed as benign, while LNs with an [18F]FAPI SUVmax≥6.2 without calcification or high-attenuation, were diagnosed as LN metastasis. Finally, we provide a complete diagnostic pathway for diagnosing LNs metastases.

Among patients with NSCLC, [18F]FDG PET/CT is widely used in preoperative staging since it is sensitive in detecting LN metastasis [17]. In several recent large trials, primary tumor location, size, histology, and SUVmax were not associated with LN involvement [18, 19]. Therefore, the relationship between primary tumor characteristics and LN metastases was not investigated in this study.

In this study, we discovered 4 risk factors on [18F]FDG PET/CT (LN risk category, CHA, LN short-axis dimension, and LN [18F]FDG SUVmax) that were significantly associated with LN metastasis. Those were consistent with previous reports [5]. Patients with a high LN risk category, including ipsilateral peribronchial (Group 11), hilar (Group 10), lower paratracheal (Group 4), and subcarinal lymph nodes (Group 7), were more prone to LN metastasis. Shim et al [8] noted that nodes displaying calcification or higher attenuations than that of the surrounding great vessels, even with positive uptake at PET, are benign. These nodes show follicular hyperplasia in the cortex, anthracotic pigmentation, and macrophage infiltration with or without fibrotic micronodule formation in the medulla. These inflammatory changes of follicular hyperplasia and macrophage infiltration may increase glucose uptake. Similarly, our findings suggest that LNs with CHA are likely benign, whereas those with a longer short-axis dimension or higher [18F]FDG SUVmax tend to be malignant. Based on these 4 risk factors (high LN risk category, large LN short-axis size (≥ 1.0 cm), absence of LN CHA and higher LN FDG SUVmax (≥ 10.1)), we grouped LNs according to their metastatic potential. The coexistence of 4 risk factors was an excellent predictor of LN metastases (PPV 100%). The [18F]FDG negative group can effectively predict benign lesions (NPV 95.0%). Hence, evaluation of these LNs requires only an [18F]FDG PET/CT examination. However, the PPV for LNs with 1–3 risk factors was 21.7%, suggesting that the nature of these LNs remains ambiguous and may necessitate additional testing.

Several articles evaluated the role of FAPI PET/CT in lung cancer and suggested that FAPI was a promising tumor imaging agent for lung cancer [11,12,13]. Similar to previous studies [13], [18F]FAPI demonstrated high specificity in this study, and the specificity of [18F]FAPI was higher than that of [18F]FDG. The sensitivity of FAPI for detecting LN metastases (80%) in this study was in accordance with the literature (84%) [13]. Previous studies reported that the SUVmax of FAPI was significantly higher than that of FDG in metastatic LNs [11,12,13]. In our study, however, there was no difference between [18F]FAPI and [18F]FDG SUVmax in metastatic LNs. This may be caused by that larger lymph nodes (≥ 1.0 cm) had higher FAPI uptake than smaller LNs. The majority of patients had advanced lung cancer with larger LN metastases in prior studies [11,12,13]. However, the proportion of larger LNs (18.2%) in our study was relatively low. The AUC of [18F]FAPI PET/CT for detecting LNs metastases was higher than that of [18F]FDG PET/CT (0.67), indicating the better predictive value of [18F]FAPI. In addition, FAPI uptake was found to be an independent predictor for LN metastases in multivariable analysis. Finally, [18F]FAPI is more accurate than [18F]FDG in diagnosing LNs (P < 0.001).

Indeed, [18F]FAPI contributed additional valuable information after [18F]FDG PET/CT. In stage I-IIIA NSCLC, our study highlighted the utility of [18F]FAPI PET in cases of an inconclusive diagnosis after [18F]FDG PET/CT examination, such as LNs with 1–3 risk factors. Under this circumstance, LNs with a FAPI SUVmax of less than 6.2 were considered as benign (NPV 93.8%). Conversely, LNs with a FAPI SUVmax ≥6.2 and without CHA were identified as metastatic (PPV 87.5%). Our research helps to resolve the ambiguous diagnosis of the LN nature in stage I-IIIA NSCLC. Our study proposes a complete diagnostic flowchart that may allow better identification of the LN nature in clinical practice (Fig. 9).

Due to a high uptake of [18F]FDG in inflamed LNs in the mediastinum and bilateral hilum [20, 21], misdiagnosis of LN metastasis and the overestimation of N staging are frequently observed. The clinical consensus recommends that positive mediastinal LNs on [18F]FDG PET/CT should be histologically verified via fine-needle aspiration biopsy [22]. However, the invasiveness of this procedure is a notable drawback. Moreover, radical surgery cannot proceed if a misdiagnosis of N3 stage occurs. Our study found that most of these incorrect N stages could be corrected with [18F]FAPI. After integrating [18F]FDG and [18F]FAPI PET/CT, the overestimation of the N stage decreased dramatically, from 58.5% (FDG-based) to 3.8%. The accuracy of [18F]FDG and [18F]FAPI PET/CT combined was as high as 83.0%. Thus, the number of patients requiring LN biopsies was significantly reduced. Finally, 18 patients initially staged as N3 based on [18F]FDG imaging were corrected to N0-N2 stage after the combination of [18F]FDG and [18F]FAPI PET/CT, ultimately enabling them to undergo radical surgery.

Currently, no imaging test exists that can detect LN metastasis with high sensitivity and specificity. Therefore, anatomical lobectomy, along with systematic LN dissection, is considered to be the main surgical approach for early-stage NSCLC [23]. Nevertheless, for patients without LN metastasis, systematic LN dissection results in excessive LN dissection. Consequently, this approach extends the surgery duration and increases perioperative complications. In this study, compared to [18F]FDG and [18F]FAPI PET/CT separately, the combination of [18F]FDG and [18F]FAPI PET/CT exhibited the highest detection accuracy for N staging, particularly in patients with pathological N0 stage. For pathological N0 stage patients, FDG-based N staging was overestimated in 25 patients, whereas only one patient was overestimated after integrating [18F]FDG and [18F]FAPI PET/CT. Out of these 25 FDG-positive patients, 24 patients were correctly identified as N0 stage through the combination of [18F]FDG and [18F]FAPI PET/CT. Therefore, these 24 patients did not need LN dissection during surgery, thereby reducing the burden on patients and shortening the length and difficulty of the operation. Our study emphasizes that if preoperative [18F]FDG imaging fails to clarify the nature of LNs, further [18F]FAPI could contribute to identifying LN nature and making more precise clinical decisions.

Our study has several limitations. First, the relatively small number of participants and the small proportion of patients with LN metastases may lead to statistical uncertainty. Second, the lack of immunohistochemical staining of FAP prevented us from clarifying the histological cause of the inconformity between FAPI PET/CT and pathology. Third, the limitations include the study design itself, since negative nodes were never sampled, which could introduce bias to the sensitivity, specificity, PPV and NPV estimates.