Clinicopathological, Oncogenic, and 18F-FDG PET/CT Features of Primary Pulmonary Carcinoid in Resection Specimens

Objectives The metabolic parameters which included mean standardised uptake value (SUVmean), metabolic tumour volume (MTV), total lesion glycolysis (TLG), maximum standardised uptake lean body mass (SULmax), and maximum standardised uptake body surface area (SUVbsa) have rarely been investigated in pulmonary carcinoid (PC). This study aimed to retrospectively compare the 18F-FDG PET/CT features of PC subtypes and observe clinicopathological and oncogenic characteristics of PC. Methods We performed a retrospective review in 60 patients with PC, from January 2016 to November 2021, who underwent the 18F-FDG PET/CT scan. All the PC diagnoses were histopathologic confirmed by surgical samples. The metabolic and morphological features were obtained from 18F-FDG PET/CT images. The ratio of metabolic to morphological lesion volumes (MMVR) was calculated. Results Sixty patients with PC were consecutively identified, including 39 patients (65.0%) with typical carcinoids (TCs) and 21 (35.0%) with atypical carcinoids (ACs). One (1/21) patient had mutation in BRAF. The ACs have a larger size (P < 0.001), more metastatic lymph nodes (P = 0.011), higher Ki-67 expression (P < 0.001), higher SUVmax values (P = 0.003), higher SUVmean values (P = 0.006), higher SULmax values (P = 0.005), higher SUVbsa values (P = 0.001), higher MTV values (P = 0.033), and higher TLG values (P = 0.002). The multivariate analysis showed that MMVR (P = 0.020) was significantly associated with AC. For predicting AC, the optimal cut-off value of SUVmax, SUVmean, SULmax, SUVbsa, MTV, TLG, and the maximum diameter was 5.19, 3.18, 2.65, 1.47, 4.36, 18.44, and 3.0, respectively. The AUC values of above mentioned parameters was 0.756 (95%CI, 0.631–881; P = 0.001), 0.735 (95%CI, 0.602–868; P = 0.003), 0.736 (95%CI, 0.607–865; P = 0.003), 0.742 (95%CI, 0.612–873; P = 0.002), 0.593 (95%CI, 0.430–755; P = 0.239), 0.680 (95%CI, 0.531–829; P = 0.022), and 0.733 (95%CI, 0.598–868; P = 0.003), respectively. For predicting TC, the optimal cut-off value of the MMVR was 0.92, and the AUC value was 0.780 (95%CI, 0.647–0.913; P < 0.001). Conclusion 18F-FDG PET/CT can simultaneously reveal the metabolic and morphological characteristics of PC, which is important in the differentiation for histopathologic subtypes.


Introduction
Pulmonary carcinoid (PC) tumors are a rare subtype of neuroendocrine malignancies, accounting for less than 2% of all lung malignancies [1,2]. Environmental risk factors, including harmful maternal exposures, have yet to be identified for PC although tobacco smoking has been PC often manifests as a localized slowly growing mass, the majority of patients have no symptoms at all and are more likely to be diagnosed incidentally during investigation of a separate medical issue.
PCs are thought to originate from enterochromaffin (Kulchitsky) cells in the bronchial and bronchiolar mucosa. PC tumors can be subdivided into two categories according to the mitotic rate and the presence or absence of necrosis. Typical carcinoids (TCs) are well differentiated, of low-grade (less than two mitoses per 2 mm 2 and lack of necrosis), and less aggressive [5]. Atypical carcinoids (ACs) are poorly differentiated, of intermediate-grade (more than two mitoses per 2 mm 2 , and necrosis may be present), and more likely to exhibit regional lymph node or distant metastases [5]. Complete surgical resection is the therapeutic option that offers the best prognosis for patients. e surgical treatment regimen depends on the size, location, and histological subtype; the goal of surgery is to preserve as much lung tissue as possible while still eliminating the tumor. us, preoperative differentiation of TC and AC is critical for designing surgical plans and predicting prognosis.
Positron emission tomography (PET), using different tracers, has been used to differentiate between ACs and TCs. e maximum standardised uptake (SUV max ) of 18 -fluorodeoxyglucose ( 18 FDG) and 68 Ga-DOTA-labeled somatostatin analogues likely has limited value in the differential diagnosis of histological subtype [6][7][8][9][10][11]. 18 FDG SUV max values are higher in AC than in TC, whereas 68 Ga-DOTAlabeled somatostatin analogues SUV max values were higher in TC than in AC. ese studies also showed that the ratio of SUV max of 68 Ga-DOTA-labeled somatostatin analogues to that of 18 FDG provided the best diagnostic performance for predicting the histopathologic variety of PC. However, unfortunately, 68 Ga-DOTA-labeled somatostatin analogues are only available in a limited number of hospitals in China. Moreover, other metabolic parameters, including mean SUV (SUV mean ), metabolic tumor volume (MTV), total lesion glycolysis (TLG), maximum standardised uptake lean body mass (SUL max ), and maximum standardised uptake body surface area (SUVbsa) between PC subtypes have rarely been reported. us, the goals of this study were to retrospectively compare the 18 F-FDG PET/CT features of PC subtypes and observe clinicopathological and oncogenic characteristics of PC.

Study Population.
We performed a retrospective review of patients who underwent PET/CT by searching our hospital's database for relevant cases occurring between January 2016 and November 2021( Figure 1).
is study was approved by the institutional review board of our hospital. Sixty patients with PC were enrolled based on the following criteria: (1) newly diagnosed with PC and histopathological results confirmed via surgical pathological examination, (2) no previous history of malignancy, (3) no antitumor treatment before PET/CT scanning and surgery, (4) an interval between PET/CT scanning and surgery of less than one month, and (5) lesions were measurable and were clearly delineated. Clinicopathological characteristics of each patient were retrospectively obtained from electronic medical records. 18 F-FDG PET/CT imaging was performed using a PET/CT scanner (Biograph Mct64, Siemens, Erlangen, Germany). Patients with both a fasting (4-6 h) and serum glucose level <11.1 nmol/L were injected with 0.10-0.15 mCi/kg of 18 F-FDG. All patients rested for an hour and then underwent whole-body PET/CT scanning (six or seven bed positions from upper thighs to forehead) according to the manufacturer's protocol. All patients also underwent an additional thin-section, chest CT scan at full inspiration with a slice thickness of 1 mm. Two independent experienced nuclear medicine physicians (Z.L., L.Q.), blinded to the clinicopathological information, retrospectively performed metabolic parameter measurements and a morphological feature assessment of the primary tumor. Metabolic parameters included SUV max , SUV mean , SUL max , SUVbsa, MTV, and TLG (TLG = MTV × SUVmean). Using a threshold of 40% of the SUVmax by an SUV-based automated contouring program, the contour of the primary tumor was defined on attenuation-corrected 18 F-FDG PET/CT images. e ratio of metabolic to morphological lesion volumes (MMVR) was calculated (MMVR � MTV/morphological tumour volume). Morphological tumour volume was calculated based on the modified ellipsoidal formula on CT images. e morphological features included (1) size (maximum diameter of the primary lesion), (2) location (central or peripheral), (3) mass attenuation by CT (HU), (4) marginal characteristics (lobulated border and spiculation), (5) pleural characteristics (pleural indentation and pleural effusion), (6) bronchiectasis, (7) atelectasis, (8) calcifications, (9) airway involvement, and (10) opacity around the lesion.

Statistics.
Patient characteristics are descriptively summarized using the mean ± SD or frequencies (percentages) for continuous and categorical variables, respectively. Pathological features as well as metabolic and morphological parameters are also described, as appropriate. Features were compared between patients with AC versus TC using either a χ 2 test or Mann-Whitney U test. To predict the presence of AC or TC, receiver operating characteristic (ROC) analysis was employed to calculate optimal cut-off values; the sensitivity, specificity, positive-predictive value, negative-predictive value, accuracy, and the area under the curve (AUC) were also calculated for each variable. Differences were considered statistically significant when P < 0.05. Data in this study were statistically analyzed using a software program (version 21.0; IBM, Armonk, NY, USA).

Clinicopathological Findings.
Clinical characteristics of all patients are summarized in Table 1. A total of 60 patients were included, of which 25 (41.7%) were males and 35 were females (58.3%). e mean age of all patients was 54.1 ± 13.3 years (22-82); the mean age of patients with TC was 54.7 ± 11.1 years and the mean age of patients with AC was 53.1 ± 16.8 years . Twenty-one (35.0%) smokers and 39 (65.0%) nonsmokers were included in this study. Twentyone (35.0%) patients with AC and 39 (65.0%) patients with TC underwent surgical resection of their tumor. Twentynine patients (48.3%) with PC presented with symptoms, of which 25 patients complained of cough; of the 25 patients with cough, ten presented with haemoptysis. Two patients complained of dyspnea, and two patients exhibited fever. In addition, AC patients had a significantly greater number of metastatic lymph nodes than patients with TC (8/72.7% vs 3/ 27.3%, P � 0.011).

Immunohistochemistry and Gene Expression Findings .
Immunohistochemistry was used to assess the expression of CD56, TTF-1, Syn, CgA, INSM1, Ki-67, and S-100 in resected tumor tissue samples; detailed information is presented in Table 2. In this study, proliferation marker Ki-67 expression was higher in patients with AC than that in patients with TC (P < 0.001). TTF-1 tended to be positive in peripheral tumors but negative in central tumors (51.6% vs. 14.3%, P � 0.003), and the result was not shown. Mutations in EGFR, EML4-ALK, and ROS1 were evaluated in 36 patients; however, no mutations were detected. e mutation status of KRAS and BRAF was evaluated in 21 patients; one patient had mutation in BRAF.

18 F-FDG PET/CT Findings.
Metabolic and morphological features of tumor tissue from patients with PC were assessed using 18 F-FDG PET/CT scans and are summarized in Table 3. e mean maximum diameter of PC tumors was 2.2 ± 1.3 cm (ranging, 0.6-7.0 cm), while the median maximum diameter of TC tumors was 1.8 cm (interquartile range, 1.3-2.8 cm). AC tumors thus had larger diameters than TC tumors (P < 0.001). No other radiological differences were identified between AC and TC tumors.

Discussion
Pulmonary carcinoids are a rare and heterogeneous group, comprising TC and AC tumor types; prognosis varies with different histological subtypes. However, there is no clear clinical picture that can differentiate between TC and AC tumor types as patients with either type can present asymptomatically or with nonspecific symptoms [4]. us, there is a lack of professional consensus regarding diagnostic assessment and treatment options for PC. In this study, use of 18 F-FDG PET/CT scans revealed a statistically significant difference in several metabolic parameters between TC and AC tumor types.
Clinical findings in patients with PC investigated in this study were consistent with the previously published studies [2,5,12,13]. In our cohort, the mean age was 54.1 years, which also agrees with data previously reported. Petursdottir

Contrast Media & Molecular Imaging
Previous studies have indicated that central tumors often present with symptoms related to segmental or larger airway obstruction, while peripheral tumors generally present asymptomatically or are only incidentally detected upon chest imaging [14]. ese differences might account for the discrepancy between the results of our study and those of Petursdottir et al.
Our study also found that lymph node metastases were more frequent in patients with AC tumors than in patients with TC tumors (8/72.7% vs 3/27.3%, P = 0.011), which was consistent with previous study results [2,5,12,13]. Georgakopoulou et al. reported that patients with surgicallyresected pulmonary carcinoids without distant metastases had increased survival [12]. akur et al. also reported that the presence and location of lymph node metastases were significant prognostic factors [5].
us, it is critical to evaluate preoperative lymph node status when making decisions regarding the extent of surgery.
To the best of our knowledge, cancer-related gene mutations are not common in patients with PC. e low frequency mutations of EGFR, EML4-ALK, BRAF, KRAS, SMAD4, and PIK3CA have previously been reported for PC tumors [15][16][17][18]. We only identified a single patient with a BRAF mutation in this study. Chen et al. reported that a 72year-old man with pulmonary atypical carcinoid harbored the EGFR L858R mutation [17]. After receiving combination chemotherapy consisting of irinotecan and icotinib plus cisplatin, the patient exhibited a partial response prior to resection. With the advent of precision medicine and the rising attention given to molecular pathology, mutation detection has gradually become the standard of care in clinical practice. It can be expected that patients with rare tumors will also benefit from more personalized treatment. Very little information is known about the genetic background of PCs. Further studies are necessary to validate these results and enhance understanding of their molecular characteristics.
Our immunohistochemical results were similar to those reported by others [19][20][21]. PC exhibited strong protein expression of neuroendocrine markers such as CD56, Syn, CgA, and INSM1. In addition, TTF-1 positivity was observed at greater frequency in peripheral tumors as compared to central tumors. Our study also revealed that PC tumors with a Ki-67 index ≥5% were more likely to be of the AC subtype. e finding showed that AC tumours exhibited a stronger proliferation activity than TC tumours, which was in line with the former being clinically more aggressive forms. And, the proliferative activity of AC tumors also likely reflects increased mitotic activity. Boland et al. reported that a Ki-67 index ≥3.5% predicted atypical histology for both biopsy and resection [22]. us, it may be useful to integrate the Ki-67 index into differential diagnosis of AC tumors and TC tumors. Previous studies have reported that the Ki-67 index is also associated with recurrence and prognosis, but that the cut-off value differs in these studies [19][20][21]. Further research regarding gene expression in PC is required in order to improve preoperative determination of histological subtype, risk stratification, and prediction of prognosis.  Figure 2: e Mann-Whitney U test showed the maximum diameter and most of metabolic parameters of 18 FDG PET/CT were significantly associated with histological subtypes of pulmonary carcinoid. 6 Contrast Media & Molecular Imaging Preoperative evaluation of PC is preferably evaluated via a CT scan. Previous studies have reported that PC often presents with no specific radiographic features, while a lobulated border and direct and indirect findings related to bronchial obstruction are more common [23,24]. e imaging heterogeneity has also not been established between    AC and TC tumors, which have similar imaging characteristics. In this study, AC patients exhibited larger tumor sizes; however, this was not sufficient to correctly distinguish AC from TC tumors. Recent developments in radiomic methods, machine learning, and artificial intelligence have revolutionized radiology. In order to appropriately distinguish AC and TC tumors, future studies may benefit from these advanced methods. At present, published studies have reported that 18 F-FDG and 68 Ga-DOTA-labeled somatostatin analogues used in PET/CT scanning might have limited value in distinguishing the PC subtype [6][7][8][9][10][11]. SUV max values of 18 FDG were higher in AC tumors, while SUV max values of 68 Ga-DOTA-labeled somatostatin analogues were higher in TC tumors. And several studies showed that 18 F-FDG PET metabolic parameters were not able to distinguish TC from AC [25]. Furthermore, previous studies reported that the SUV max ratio of 68 Ga-DOTA-labeled somatostatin analogues and 18 F-FDG is more accurate for the prediction of histological subtype in PC as compared with the SUV max of 68 Ga-DOTA-labeled somatostatin analogues or 18 FDG alone [7][8][9][10][11]. According to a meta-analysis, Jiang et al. reported that the SUV max ratio had 89.3% sensitivity and 100% specificity in distinguishing TC from AC [7]. However, 68 Ga-DOTA-labeled somatostatin analogs are only available in a limited number of hospitals in China. By comparing metabolic parameters, we found that the TLG had the highest specificity (84.6%), while SUL max had the highest sensitivity (85.7%) and negative-predictive value (87.0%). On the other hand, values of MMVR in AC were significantly lower than those in TC. is is the most probably attributed more necrotic area in AC tumours, which can cause low MTV on 18 FDG PET/CT images. e MMVR had a specificity of 89.7% and an accuracy of 80.0%. And the multivariate analysis showed that only an MMVR was significantly associated with AC. e MMVR may be better than other metabolic parameters for distinguishing TC from AC.
e study showed that AC tumours with stronger proliferation activity have higher relative metabolic activity on 18 FDG PET/CT examination. However, metabolic parameters obtained via 18 F-FDG were not as good at differentiating histopathologic variety of PC, individually. When 68 Ga-DOTA-labeled somatostatin analogues are unavailable, 18 FDG PET/CT scanning can provide valuable additional information in for preoperative differentiation of AC and TC tumors. In recent years, there have been an increasing number of studies related to texture features in the analysis of PET/CT data. In this regard, further studies are needed to improve predictive performance.
is study has several limitations. First, it is an observational, single-center, retrospective study. Second, only patients who underwent surgical resection were included in this study. Both of these limitations can result in selection bias. Finally, the present study did not include contrast enhancement features of CT images in patients with PC because only some patients underwent enhanced CT examination. Moreover, in our hospital, enhanced CT examination were not performed on the same device. e lack of standardization resulted in a lack of uniformity in the data, making them difficult to compare. More rigorous studies are warranted.

Conclusion
In conclusion, our findings revealed that metabolic parameters of 18 FDG PET/CT were significantly associated with histopathological subtypes of pulmonary carcinoid. AC tumors tend to be larger and exhibit higher proliferative  Epidermal growth factor receptor KRAS: Kirsten rat sarcoma viral oncogene homolog BRAF: B-type Raf kinase EML4-ALK: Echinoderm microtubule-associated protein-like 4-anaplastic lymphoma kinase ROS1: C-ros oncogene 1.

Data Availability
e datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Disclosure
Yun Chen and Yun Dong are co-first authors.

Conflicts of Interest
e authors declare that there are no conflicts of interest regarding the publication of this paper.