Advanced characterization and grading of invasive lung adenocarcinoma: integrative analysis with spectral CT and ¹⁸F-FDG PET/CT imaging

Introduction

Lung adenocarcinomas (LUADs) demonstrate histologically heterogeneous characteristics, with 80–90% comprising various combinations of patterns and proportions (1). This classification facilitates the stratification of histological grades in LUAD, thereby providing guidance for patient therapy and prognosis. According to the 2015 World Health Organization (WHO) classification, the architectural grading system was based on the predominant histologic pattern (2). Consequently, LUAD is categorized into three grades: low grade (grade 1, lepidic predominant), intermediate grade (grade 2, acinar or papillary predominant), and high grade (grade 3, solid or micropapillary predominant) (3-5). However, the 2020 International Association for the Study of Lung Cancer (IASLC) pathology committee proposed a revised grading system for LUAD. In this proposed IASLC grading system, based on a combination of predominant histologic and high-grade patterns, any tumor containing at least 20% high-grade histologic components (solid, micropapillary, and/or complex gland) is classified as high grade (grade 3) (6,7). Several studies have identified the presence of a high-grade subtype as a predictor of poor prognosis (8-10). Additionally, Rokutan-Kurata et al. (11) reported that the IASLC system has significant prognostic value for invasive LUADs.

Spectral computed tomography (CT) is a noninvasive imaging technique that employs an instantaneous kilovoltage (kVp)-switching method, alternating between 80 and 140 kVp, to quantitatively assess iodine concentration (IC) in material decomposition images and analyze spectral slope pitch in monochromatic images across energy levels ranging from 40 to 140 keV (12). Several studies have demonstrated that the application of spectral CT imaging parameters, such as iodine content, spectral curve, and effective atomic number (Zeff), provide supplementary information on tumor characteristics beyond morphology, potentially enhancing the accuracy of lung cancer diagnosis (13,14). ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG PET/CT) is extensively utilized in the diagnosis, treatment, and monitoring of lung cancer. This modality allows for the evaluation of tumor metabolic activity through quantitative metabolic parameters, including mean standardized uptake value (SUVmean), maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) (15). Thus, spectral CT and PET/CT have broad application prospects in the diagnosis and treatment of lung cancer (16-18).

In the present study, we describe the morphological CT, spectral CT, and ¹⁸F-FDG PET/CT findings in distinguishing high grade invasive non-mucinous adenocarcinoma (INMA) from low grade INMA, and we not only explore the utility of spectral CT and ¹⁸F-FDG PET/CT for predicting histological grades of LUAD preoperatively, but also compare the predictive performance of the spectral CT, PET/CT, and comprehensive model. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1735/rc).

Methods

Patients and study design

The present study retrospectively enrolled patients admitted to the Shanghai Chest Hospital from August 2021 to December 2022, pathologically diagnosed with LUAD, and subsequently underwent spectral CT and/or ¹⁸F-FDG PET/CT examinations. The study was approved by the Medical Ethical Committee of Shanghai Chest Hospital (No. IS23088). The requirement for informed consent was waived because of the retrospective nature of this study. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Initially, 1,360 patients suspected of lung cancer were included, and 135 patients with 144 lung nodules were finally enrolled in the current study. Histologic subtyping and grading of INMA were performed according to the IASLC system (6) as follows: grade 1, lepidic predominant with <20% of high-grade pattern (solid, micropapillary, and/or complex glandular patterns); grade 2, acinar or papillary predominant with <20% of high-grade pattern; grade 3, any tumor with ≥20% of high-grade patterns. The inclusion criteria for the study were as follows: (I) patients with pathologically confirmed INMA, available histologic subtypes and proportions; and (II) having undergone preoperative spectral CT examination, and/or ¹⁸F-FDG PET/CT scan. The exclusion criteria were as follows: (I) previous antitumor treatment; (II) lesion was not histologically confirmed as invasive LUAD; (III) lesion without histologic subtype; (IV) inadequate spectral CT and/or ¹⁸F-FDG PET/CT examination; and (V) coexistence with mucinous adenocarcinoma, and other variants of adenocarcinoma. The flowchart of the study participant screening is shown in Figure 1.

Figure 1 Study diagram of the patient selection process. CT, computed tomography; FDG, fluorodeoxyglucose; INMA, invasive non-mucinous adenocarcinoma; PET, positron emission tomography.

Spectral CT and PET/CT examinations

Spectral CT examinations were performed on a Discovery CT750HD scanner (GE Healthcare, Waukesha, WI, USA). Patients underwent non-contrast and dual-phase contrast-enhanced CT scanning. Non-contrast CT was performed first, and the scanning range included the entire chest from the thoracic inlet to the base of lung. Then, the patients were injected with a bolus of 50–70 mL (0.8 mL/kg of body weight) nonionic iodinated contrast medium (iopamidol, 370 mg I/mL) via the median cubital vein at a rate of 3.5 mL/s, followed by 30 mL of saline solution flush during contrast-enhanced scanning. The arterial phase (AP) and venous phase (VP) scan delay times in gemstone spectral imaging (GSI) mode were triggered at 28 and 46 seconds after contrast media injection, respectively. The scanning parameters were as follows: 80/140 kVp fast-tube voltage switching, automated mAs exposure, 0.5 seconds helical tube rotation time, 0.984 helical pitch, 1.25 mm reconstruction interval.

¹⁸F-FDG PET/CT was performed after fasting for at least 8 hours and 60 minutes after intravenous administration of ¹⁸F-FDG (5.0 MBq/kg). The blood glucose levels, measured just before tracer administration, were <8.0 mmol/L in all patients. The imaging was performed with a combined PET/CT device (Biograph mCT; Siemens, Erlangen, Germany). Immediately after non-contrast CT scan using a 64-slice helical CT (120 keV, 30–100 mA in AutomA mode, 5.0 mm slice), PET was performed with an acquisition time of 3 minutes per frame in 3-dimensional (3D) mode. CT data were used for attenuation correction, and images were iteratively reconstructed by a trueX + time of flight (TOF) algorithm (21 subsets, 3 iterations).

Spectral CT and PET/CT image analyses

The images were independently assessed by a radiologist with more than 5 years of experience and a nuclear medicine physician with more than 10 years of experience who were both blinded to others’ results and all clinical data. Disagreements were resolved by consensus and, when necessary, submitted to a third reviewer with more than 20 years of experience. The lung nodules were classified as pure ground-glass nodules (GGNs), part-solid nodules, or solid nodules. The spectral CT datasets were transferred to the AW4.7 workstation for retrospective analysis and post-processing of images by using GSI Viewer software package. The region of interest (ROI) was carefully placed in the slice with the maximum cross-sectional diameter of the lesion as large as possible and moved away from necrosis, vessels, bronchi, atelectasis, and calcification. ROIs were processed in duplicate in the upper and lower adjacent slices and the values were averaged to ensure measurement consistency. Post-processing was performed and spectral curve images, monochromatic images (energy levels of 40–140 keV), and iodine-based material decomposition images were generated. The iodine concentration of lesions (ICL) and the Zeff were measured both in the AP and VP. Simultaneously, the same ROI was placed in the thoracic aorta or subclavian artery on the same slice. The normalized IC (NIC) and the slope of spectral Hounsfield unit (HU) curve (λHU) were calculated using the following formula, respectively (19,20): NIC = IC_lesion / IC_aorta; λHU = [CT number (40 keV) – CT number (100 keV)] / (100–40) keV.

Axial, sagittal, and coronal PET/CT images were examined using Syngo software and then transferred in Digital Imaging and Communications in Medicine (DICOM) format to a Syngo workstation (Siemens). The ROIs of primary lesions were manually defined, and mean SUVmean, SUVmax, MTV, and TLG (TLG = SUVmean × MTV) were recorded.

Statistical analysis

Statistical analysis was performed using the software SPSS 27.0 (IBM Corp., Armonk, NY, USA) and R software (version 4.3.2, R Foundation for Statistical Computing, Vienna, Austria). Descriptive statistics were presented as mean ± standard deviation (SD) for normally distributed continuous variables, as median (interquartile range) for nonnormally distributed continuous variables, and as frequencies with proportions for categorical variables. To compare differences between two groups, the two independent-sample t-tests were employed to compare parameters consistent with normal distribution, whereas the two-sample Mann-Whitney test was applied to nonnormally distributed data. In multiparametric groups, one-way analysis of variance (ANOVA) was performed for parametric variables, and the Kruskal-Wallis test was used to assess the difference between groups for nonparametric variables. If a value of P<0.05 was found for multiple comparisons of variables, Bonferroni correction was applied (P values <0.017 were considered statistically significant). Categorical variables were analyzed using the chi-squared test or Fisher’s exact test. Multiparametric analysis was constructed using stepwise binary logistic regression analysis with high-resolution computed tomography (HRCT) signs, the spectral parameters, and the PET/CT parameters. The model was performed with forward stepwise selection, and the odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. The discriminative capability of significant parameters was evaluated by receiver operating characteristic (ROC) analysis. A nomogram was built based on the multivariate logistic regression model. A calibration curve was plotted to assess the model calibration, and the Hosmer-Lemeshow test was used to analyze the agreement between the nomogram-predicted grade 3 INMA and true grade from the calibration curve. Decision curve analysis (DCA) was carried out to assess the performance of the nomogram model by quantifying the net benefit and the corresponding threshold probability. Then, the area under the curve (AUC) values of different models were compared by DeLong’s test to assess the prediction performance. Further assessment of discrimination ability was performed using integrated discrimination improvement (IDI) analysis. Two-tailed P<0.05 was considered statistically significant between groups.

Results

Participant characteristics

A total of 135 participants (mean age, 62.4±9.5 years, range, 38–82 years, 55 male and 80 female) with 144 lesions were included in the present study. Of these, 13 nodules (9.0%) were categorized as grade 1, 88 nodules (61.1%) as grade 2, and 43 nodules (29.9%) were grade 3. Of the 135 patients, 55 participants (40.7%) underwent spectral CT and additional PET/CT. The specific characteristics of participants are shown in Table 1.

Spectral CT analyses across different grades of invasive LUADs

The morphological CT characteristics and spectral parameters of 144 lesions in 135 patients were extracted from spectral CT and compared between high-grade INMA (grade 3) and low-grade INMA (grade 1 and 2, non-grade 3) (Table 2). Among the common CT features, attenuation type was found to be associated with the higher invasion grade (P<0.05); however, no significant differences were found in the size and distribution of tumors, the incidence rate of lobulation, spiculation, air bronchogram, vacuolation sign, and pleural retraction (P=0.089–0.666). Besides, spectral parameters were found to be statistically different among adenocarcinoma patients with different pathological invasion grades (Figure 2). Specifically, iodine concentration of the lesion in arterial phase (ICL_a), normalized iodine concentration in arterial phase (NIC_a), slope of spectral Hounsfield unit curve in arterial phase (λHU_a), iodine concentration of the lesion in venous phase (ICL_v), normalized iodine concentration in venous phase (NIC_v), and slope of spectral Hounsfield unit curve in venous phase (λHU_v) of patients with grade 3 LUAD were statistically lower than those of patients with grades 1 and 2 (Bonferroni correction, all P<0.05; Figures 3,4). However, there was no significant difference in spectral parameters between grades 1 and 2.

Figure 2 Scatter plot showing ICL_a, NIC_a, λHU_a, ICL_v, NIC_v and λHU_v for patients with grade 1, grade 2 and grade 3 INMA. ICL_a, NIC_a, λHU_a, ICL_v, NIC_v, and λHU_v (A-F) of grade 3 lung adenocarcinoma were statistically lower than that of grade 1 and 2. The spectral parameters showed no significant differences between the grade 1 and 2. ns, P>0.05; **, P<0.01; ***, P<0.001. ICL_a, iodine concentration of the lesion in arterial phase; ICL_v, iodine concentration of the lesion in venous phase; INMA, invasive non-mucinous adenocarcinoma; NIC_a, normalized iodine concentration in arterial phase; NIC_v, normalized iodine concentration in venous phase; λHU_a, slope of spectral Hounsfield unit curve in arterial phase; λHU_v, slope of spectral Hounsfield unit curve in venous phase.

Figure 3 A 69-year-old man with pathologically confirmed grade 3 adenocarcinoma of the left lower lobe. (A) The iodine concentration of the lesion and thoracic aorta in AP were 1.276 and 12.341 mg/mL, respectively. (B) The slope of spectral curve was 1.524 in AP. (C) The iodine concentration of lesion and thoracic aorta in VP were 1.355 and 3.469 mg/mL, respectively. (D) The slope of spectral curve was 1.608 in VP. (E) PET/CT fusion image showed FDG uptake with SUVmax of 10.84, SUVmean of 6.31 and MTV of 7.20 cm³. AP, arterial phase; CT, computed tomography; FDG, fluorodeoxyglucose; MTV, metabolic tumor volume; PET, positron emission tomography; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; VP, venous phase.

Figure 4 A 75-year-old woman with pathologically confirmed grade 2 adenocarcinoma of the right lower lobe. (A) The iodine concentration of the lesion and thoracic aorta in AP were 1.810 and 9.193 mg/mL, respectively. (B) The slope of spectral curve was 2.140 in AP. (C) The iodine concentration of lesion and thoracic aorta in VP were 2.137 and 3.666 mg/mL, respectively. (D) The slope of spectral curve was 2.533 in VP. (E) PET/CT fusion image revealed FDG uptake with SUVmax of 13.68, SUVmean of 7.63 and MTV of 1.42 cm³. AP, arterial phase; CT, computed tomography; FDG, fluorodeoxyglucose; MTV, metabolic tumor volume; PET, positron emission tomography; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; VP, venous phase.

Based on the results of the univariate analysis, significant morphological CT characteristics (attenuation type) and spectral parameters (ICL_a, NIC_a, λHU_a, ICL_v, NIC_v, and λHU_v) were statistically significant for distinguishing high-grade (grade 3) invasive LUAD (P<0.05). The above variables were included in the multivariate logistic regression and ICL_a (OR =0.005, 95% CI: 0.001–0.027, P<0.001) and solid nodule (OR =26.757, 95% CI: 6.843–104.623, P<0.001) were found to be independent predictors of high-grade invasive LUAD based on the spectral CT (Table 3).

PET/CT analyses across different grades of invasive LUADs

Of 135 patients included, 55 with 60 lung nodules underwent additional PET/CT imaging. As presented in Table S1, attenuation type (solid) was significantly associated with a higher invasion grade (P<0.05). The grade 3 group had higher SUVmax and SUVmean than the grade 1 and 2 groups (P<0.05) (Figure 5). There was no significant difference in SUVmax and SUVmean between grades 1 and 2 (P>0.05). In these patients, spectral parameters (ICL_a, NIC_a, λHU_a, ICL_v, NIC_v, and λHU_v) also showed differences between grade 3 and others in the patients with LUADs. The above morphological CT characteristics (attenuation type) and PET/CT indicators (SUVmax, SUVmean, TLG) were included in the multivariate logistic regression based on the PET/CT. The SUVmax (OR =1.408, 95% CI: 1.132–1.751, P<0.01) and solid nodule (OR =5.575, 95% CI: 1.110–27.993, P<0.05) were shown to be independent predictors of high-grade invasive LUAD based on the PET/CT (Table S2).

Figure 5 Scatter plot showing SUVmax, SUVmean, MTV and TLG for patients with grade 1, grade 2 and grade 3 INMA. SUVmax (A) and SUVmean (B) of grade 3 lung adenocarcinoma were statistically higher than that of grade 1 and 2; but there were no significant differences between grade 1 and grade 2. MTV (C) and TLG (D) showed no significant differences among three grade groups. ns, P>0.05; *, P<0.05; **, P<0.01; ***, P<0.001. INMA, invasive non-mucinous adenocarcinoma; MTV, metabolic tumor volume; SUVmax, maximum standardized uptake value; SUVmean, mean standardized uptake value; TLG, total lesion glycolysis.

Model building and comparison

In these patients who underwent both PET/CT and spectral CT scans, indices with differences in the above multivariate logistic regression analysis were included in regression models. The PET/CT model takes into account both morphological CT characteristics and PET/CT parameters, including SUVmax (OR =1.408, 95% CI: 1.132–1.751, P<0.01) and solid nodule (OR =5.575, 95% CI: 1.110–27.993, P<0.05). The spectral model takes into account both morphological CT characteristics and spectral parameters, including ICL_a (OR =0.005, 95% CI: 0–0.083, P<0.001) and solid nodule (OR =27.691, 95% CI: 4.113–186.444, P<0.001). The comprehensive model constructed based on the spectral parameters, PET/CT parameters, and morphological CT characteristics, including SUVmax (OR =1.306, 95% CI: 1.002–1.704, P<0.05), ICL_a (OR =0.007, 95% CI: 0–0.134, P<0.001) and solid nodule (OR =11.182, 95% CI: 1.442–86.700, P<0.05) (Table 4).

The DeLong test was conducted to compare the comprehensive model with the spectral model (Z=0.819, P=0.413) and the PET/CT model (Z=1.959, P=0.050). Additionally, the DeLong test was employed to compare the spectral model with the PET/CT model (Z=1.116, P=0.265). The IDI analysis revealed no statistically significant difference when comparing the comprehensive model with the spectral model (IDI =0.042, 95% CI: −0.015–0.099, P=0.149). However, the IDI indicated that both the comprehensive model and the spectral model outperformed the PET/CT model (IDI =0.190, 95% CI: 0.085–0.295, P<0.001; IDI =0.148, 95% CI: 0.016–0.280, P<0.05, respectively). When compared to the comprehensive model, the spectral model demonstrated satisfactory predictive performance in preoperatively differentiating between grade 3 tumors and non-grade 3 tumors, with an AUC of 0.930 vs. 0.949, sensitivity of 96.4% vs. 85.7%, specificity of 81.2% vs. 93.7%, and accuracy of 88.3% vs. 90.0%, respectively (Table 5).

Nomogram development

The comprehensive nomogram incorporating morphological CT characteristics, spectral and PET/CT parameters was constructed based on the comprehensive model of the data of 60 lesions in 55 patients (Figure 6A), and the comprehensive model discrimination showed proper predictive ability (AUC =0.949, 95% CI: 0.899–0.999) (Figure 6B). Calibration curves were used to assess the calibration, and the Hosmer-Lemeshow test significance level was 0.375, which indicated good calibration power (Figure 6C). The decision curves of the nomogram prediction model presented a good net clinical benefit (Figure 6D). As for promising diagnostic performance of the spectral CT, the spectral nomogram was constructed based on the spectral model of the data of 144 lesions in 135 patients, which integrated ICL_a and solid nodule for differentiating high-grade (grade 3) from lower-grade invasive LUAD (Figure 7A), and the model had good discrimination (AUC =0.917, 95% CI: 0.864–0.969) (Figure 7B). The calibration and decision curves (Figure 7C,7D) demonstrated good calibration power (Hosmer-Lemeshow test: 0.414) and a strong net clinical benefit for the nomogram prediction model.

Figure 6 The comprehensive nomogram (A), ROC curves for the spectral model, PET/CT model and comprehensive model (B), calibration curve of the nomogram model (C) and decision curve of the nomogram model (D) for predicting grade 3 INMA in the 55 patients with both PET/CT and spectral CT scans. AUC, area under the curve; CI, confidence interval; CT, computed tomography; FPR, false positive rate; ICL_a, iodine concentration of the lesion in arterial phase; INMA, invasive non-mucinous adenocarcinoma; PET, positron emission tomography; ROC, receiver operating characteristic; SUVmax, maximum standardized uptake value; TPR, true positive rate.

Figure 7 The spectral nomogram (A), ROC curve for the nomogram (B), calibration curve (C) and decision curve (D) for predicting grade 3 INMA in the 135 patients with spectral CT scans. AUC, area under the curve; CI, confidence interval; CT, computed tomography; FPR, false positive rate; ICLa, iodine concentration of the lesion in arterial phase; INMA, invasive non-mucinous adenocarcinoma; ROC, receiver operating characteristic; TPR, true positive rate.

Discussion

We conducted an in-depth analysis of spectral CT and ¹⁸F-FDG PET/CT features and developed multiple predictive models to preoperatively distinguish high-grade (grade 3) from non-high-grade INMA. The results demonstrated that the spectral CT-based model, incorporating spectral parameters (ICL_a, NIC_a, λHU_a, etc.) and CT morphological characteristics (e.g., solid nodule), exhibited excellent performance in predicting grade 3 adenocarcinoma, achieving an AUC of 0.930, with a sensitivity of 96.4% and specificity of 81.2%. Although the comprehensive model, which combined spectral CT, PET/CT parameters, and CT features, achieved a slightly higher AUC (AUC =0.949), the difference was not statistically significant (DeLong test, Z=0.819, P=0.413; IDI =0.042, 95% CI: −0.015–0.099, P=0.149) between the spectral CT and the comprehensive model, underscoring the independent value of spectral CT. In contrast, the PET/CT model, which incorporated attenuation type and SUVmax, showed inferior performance of the comprehensive model and the spectral model (AUC: 0.864 vs. 0.949, IDI: P<0.001; AUC: 0.864 vs. 0.930, IDI: P<0.05, respectively). Decision curve and nomogram analyses confirmed the clinical utility of the spectral and comprehensive models, supporting their role in optimizing treatment strategies and reducing unnecessary interventions.

INMA exhibits heterogeneity, and the three-tier grading system has been implemented in accordance with the IASLC grading criteria. Several studies have emphasized the clinical significance of predicting histological grades of lung cancer for determining therapeutic strategies and assessing prognosis (21). Wang et al. (22) reported that the prognosis for grade 3 INMA was significantly poorer compared to non-grade 3 INMA. In our study, we identified several imaging features from CT, spectral CT, and PET/CT that offer valuable insights into high-grade INMA. Our findings indicate that solid nodules occur more frequently in grade 3 than they do in grades 1 and 2, whereas the occurrence of ground glass opacity (GGO) features (pure GGN or part-solid nodule) was more frequent in grades 1 and 2 than in grade 3 (Table 2). Jeon et al. (23) conducted a retrospective analysis of 429 patients with stage IA invasive adenocarcinoma, classified according to the new grading system, and reported that GGO features (pure GGN or part-solid nodule) were more frequently observed in grade 1 (83.8%) compared to grades 2 (57.0%) and 3 (34.2%), which aligns with our results. Travis et al. (24) and Fujikawa et al. (25) revealed that solid components on CT images correlate with the invasiveness of LUAD, as tumors with solid/micropapillary patterns show dense structures and aggressive growth, whereas lepidic/acinar/papillary patterns have looser structures and richer stroma. Thus, we speculate that the presence of these CT features (solid density) may reflect aggressive and destructive growth patterns in high-grade malignancies.

In our study, spectral CT analysis revealed that the spectral parameters—IC, NIC, and λHU—in both the AP and VP were significant indicators of grade 3 invasive LUAD (Table 2, with all P<0.05). This finding aligns with the observations of Lin et al. (21) and Iwano et al. (26), who reported that high-grade lung tumors exhibit significantly lower iodine volume, NIC, or λHU compared to low-grade tumors. Furthermore, prior research by Li et al. (27) demonstrated that the Zeff on unenhanced CT, as well as the IC in both AP and VP, was significantly higher in tumors with lepidic, acinar, or papillary predominant patterns compared to those with solid or micropapillary predominant patterns. Tumors with solid or micropapillary predominant patterns tend to have denser parenchyma and internal structures, whereas those with lepidic, acinar, or papillary patterns are characterized by a richer stroma, more abundant blood supply, and often include glandular structures. The IC is indicative of a tumor’s blood supply (28). Our multivariate analysis of the spectral parameters identified ICL_a as an independent predictor of grade 3 tumors, which can be of clinical value in judging the degree of malignancy of LUAD. Li et al. (28) reported that the IC in the VP of the well-differentiated lung cancer was statistically significantly higher than that of the poorly differentiated lung cancer, and thought that the NIC affected by the change in the IC of both the lesion and the aorta might deviate from the actual IC of the lesion. Therefore, the IC may be a better indicator reflecting tumor blood supply than the NIC.

Our study further elucidated that metabolic parameters, specifically SUVmax, SUVmean, and TLG, are instrumental in determining the malignancy level of invasive LUAD. Notably, SUVmax values were significantly elevated in the grade 3 group (median SUVmax, 7.505) compared to the non-grade 3 group (2.045) in Table S1. These findings are consistent with those of Jeon et al. (23), who conducted a retrospective analysis of 429 patients with stage IA, invasive adenocarcinoma classified according to the new grading system. Their study revealed that SUVmax was higher in grade 3 (median SUVmax, 5.30) than in grades 1 (1.47) and 2 (2.24). In addition, Sun et al. (29) reported that SUVmax was higher in the high-grade LUAD group (mean SUVmax, 12.53) than in the low-grade group (1.21) and the intermediate-grade group (6.62) using the 2011 IASLC/ATS/ERS classification.

Importantly, our study also assessed the discriminatory capabilities of various imaging modalities and their combinations in identifying high-grade INMA. The diagnostic performance of the spectral CT model, PET/CT model and the comprehensive model, evaluated in 55 patients with 60 lung nodules, yielded efficacy rates of 0.930, 0.864, and 0.949, respectively, as illustrated in Figure 6B. The spectral CT model demonstrated predictive performance comparable to the comprehensive model (IDI =0.042, 95% CI: −0.015–0.099, P=0.149) and outperformed the PET/CT model (IDI =0.148, 95% CI: 0.016–0.280, P<0.05). Previous studies have shown that spectral CT has good differential diagnostic value in the histological classification and grades of lung cancer. Deng et al. (30) revealed that the diagnostic efficacy of the spectral parameter model (AUC =0.93, sensitivity =0.94, specificity =0.83) was better than that of the perfusion parameter model (AUC =0.81, sensitivity =0.99, specificity =0.57) in differentiating the pathological types of non-small cell lung cancer (NSCLC), and the spectral CT can replace perfusion CT to indirectly assess hemodynamic changes in NSCLC. Mu et al. (31) constructed a diagnostic model combining the spectral and morphological parameters to predict the pathological grades of LUAD, with an AUC of 0.916, sensitivity of 96.4%, and specificity of 82.1%. Their study revealed that the diagnostic efficiency of the combining model was better than that of spectral parameters in the AP for preoperative grading of LUAD. In addition, Liu et al. (32) reported that the integrated model combining traditional CT and spectral parameters outperformed the traditional model using traditional CT features significantly for spread through air spaces prediction in LUAD. These findings suggest that spectral parameters offer additional value in predicting the pathological characteristics of lung cancer, aligning with the results of our current study. In this study, we developed a nomogram based on the spectral model, which serves as a graphical tool for prediction. Furthermore, the nomogram, which incorporates independent predictors such as ICL_a and solid nodule, effectively predicts grade 3 INMA preoperatively, with an excellent concordance index (C-index) of 0.915. The DCA indicated that within a threshold probability range of 0.050 to 0.920, the net benefit of the nomogram model surpasses that of assuming all patients have grade 3 INMA, demonstrating robust predictive performance.

Spectral CT appears to be a promising technique for simultaneously evaluating morphology and function, addressing shortcomings of conventional CT. Zhang et al. (33) found that IC has better performance than CT enhancement numbers in differentiating benign and malignant solitary lung nodules. Spectral CT from a single CT scan, without prolonging the patient’s examination time, can not only acquire routine CT images, but also provides supplementary functional information through post-processing multiparameter measurements. Ito et al. (34) reported that the diagnostic performance of dual-energy CT may be comparable to that of FDG-PET/CT in predicting the histopathological invasiveness of NSCLCs. Compared with PET/CT, spectral CT has a shorter examination time and lower cost. In addition, some research reported that the overall radiation dose from spectral detector CT is equivalent to, or often lower than, that of conventional scanning (35,36). However, as far as we know, there are few studies on the accuracy and cost-effectiveness of spectral CT compared with conventional CT. We will pay attention to these aspects of spectral CT in the future and conduct corresponding investigations.

Our research has some limitations. First, the retrospective cohort was a relatively small sample, and selection bias is inevitable. Second, this study of patients from a single institution may constrain the generalizability of the results. In the future, the prospective studies with larger samples and multiple centers would help validate the present findings. Third, only a subset of patients of this study underwent the PET/CT scans. We plan to expand patient enrollment to include larger cohorts with both spectral CT and PET/CT scans in future studies. Fourth, this study does not include a comparison of cost-effectiveness and radiation dose between spectral CT and routine CT in clinical practice. Finally, the focus of this study is to distinguish the histological grades of INMA. This study is preliminary, and future investigations with more cases across different pathological types and invasion levels of lung cancer are needed to draw broader conclusions.

Conclusions

The diagnostic ability of the spectral CT model was comparable to the comprehensive model and significantly better than the PET/CT model for high-grade INMA, which takes nodule attenuation type and IC in AP derived from spectral CT into account. Our economical single-modality spectral CT model shows great promise in clinical application and histopathological high-grade diagnosis of INMA in the future, which warrants further in-depth and broad research.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1735/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1735/dss

Funding: This study was supported by the Interdisciplinary Program of Shanghai Jiaotong University (No. YG2025QNB12), Young Scientists Fund of the National Natural Science Foundation of China (No. 82302188), National Key R&D Program of China (No. 2021YFC2500700), Shanghai Health Research Foundation for Talents (No. 2022YQ060), Shanghai Science and Technology Innovation Action Plan (No. 22Y11911100), Shanghai Innovative Medical Product Application Demonstration Project (No. 24SF1904000), National Natural Science Foundation of China (No. 82272044), and the Natural Science Foundation of Shanghai (No. 21ZR1458900).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1735/coif). All authors report that this study was supported by the Interdisciplinary Program of Shanghai Jiaotong University (No. YG2025QNB12), Young Scientists Fund of the National Natural Science Foundation of China (No. 82302188), National Key R&D Program of China (No. 2021YFC2500700), Shanghai Health Research Foundation for Talents (No. 2022YQ060), Shanghai Science and Technology Innovation Action Plan (No. 22Y11911100), Shanghai Innovative Medical Product Application Demonstration Project (No. 24SF1904000), National Natural Science Foundation of China (No. 82272044), and the Natural Science Foundation of Shanghai (No. 21ZR1458900). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by the Medical Ethical Committee of Shanghai Chest Hospital (No. IS23088). Informed consent was waived because of the retrospective nature of this study. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Duhig EE, Dettrick A, Godbolt DB, Pauli J, van Zwieten A, Hansen AR, Yang IA, Fong KM, Clarke BE, Bowman RV. Mitosis trumps T stage and proposed international association for the study of lung cancer/american thoracic society/european respiratory society classification for prognostic value in resected stage 1 lung adenocarcinoma. J Thorac Oncol 2015;10:673-81. [Crossref] [PubMed]
2. Travis WD, Brambilla E, Nicholson AG, Yatabe Y, Austin JHM, Beasley MB, Chirieac LR, Dacic S, Duhig E, Flieder DB, Geisinger K, Hirsch FR, Ishikawa Y, Kerr KM, Noguchi M, Pelosi G, Powell CA, Tsao MS, Wistuba I, Panel WHO. The 2015 World Health Organization Classification of Lung Tumors: Impact of Genetic, Clinical and Radiologic Advances Since the 2004 Classification. J Thorac Oncol 2015;10:1243-60. [Crossref] [PubMed]
3. Campos-Parra AD, Avilés A, Contreras-Reyes S, Rojas-Marín CE, Sánchez-Reyes R, Borbolla-Escoboza RJ, Arrieta O. Relevance of the novel IASLC/ATS/ERS classification of lung adenocarcinoma in advanced disease. Eur Respir J 2014;43:1439-47. [Crossref] [PubMed]
4. Warth A, Muley T, Meister M, Stenzinger A, Thomas M, Schirmacher P, Schnabel PA, Budczies J, Hoffmann H, Weichert W. The novel histologic International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society classification system of lung adenocarcinoma is a stage-independent predictor of survival. J Clin Oncol 2012;30:1438-46. [Crossref] [PubMed]
5. Yoshizawa A, Sumiyoshi S, Sonobe M, Kobayashi M, Fujimoto M, Kawakami F, Tsuruyama T, Travis WD, Date H, Haga H. Validation of the IASLC/ATS/ERS lung adenocarcinoma classification for prognosis and association with EGFR and KRAS gene mutations: analysis of 440 Japanese patients. J Thorac Oncol 2013;8:52-61. [Crossref] [PubMed]
6. Moreira AL, Ocampo PSS, Xia Y, Zhong H, Russell PA, Minami Y, et al. A Grading System for Invasive Pulmonary Adenocarcinoma: A Proposal From the International Association for the Study of Lung Cancer Pathology Committee. J Thorac Oncol 2020;15:1599-610. [Crossref] [PubMed]
7. Nicholson AG, Tsao MS, Beasley MB, Borczuk AC, Brambilla E, Cooper WA, Dacic S, Jain D, Kerr KM, Lantuejoul S, Noguchi M, Papotti M, Rekhtman N, Scagliotti G, van Schil P, Sholl L, Yatabe Y, Yoshida A, Travis WD. The 2021 WHO Classification of Lung Tumors: Impact of Advances Since 2015. J Thorac Oncol 2022;17:362-87. [Crossref] [PubMed]
8. Tsubokawa N, Mimae T, Sasada S, Yoshiya T, Mimura T, Murakami S, Ito H, Miyata Y, Nakayama H, Okada M. Negative prognostic influence of micropapillary pattern in stage IA lung adenocarcinoma. Eur J Cardiothorac Surg 2016;49:293-9. [Crossref] [PubMed]
9. Zhang Y, Wang R, Cai D, Li Y, Pan Y, Hu H, Wang L, Li H, Ye T, Luo X, Zhang Y, Li B, Shen L, Sun Y, Chen H. A comprehensive investigation of molecular features and prognosis of lung adenocarcinoma with micropapillary component. J Thorac Oncol 2014;9:1772-8. [Crossref] [PubMed]
10. Hong JH, Park S, Kim H, Goo JM, Park IK, Kang CH, Kim YT, Yoon SH. Volume and Mass Doubling Time of Lung Adenocarcinoma according to WHO Histologic Classification. Korean J Radiol 2021;22:464-75. [Crossref] [PubMed]
11. Rokutan-Kurata M, Yoshizawa A, Ueno K, Nakajima N, Terada K, Hamaji M, Sonobe M, Menju T, Date H, Morita S, Haga H. Validation Study of the International Association for the Study of Lung Cancer Histologic Grading System of Invasive Lung Adenocarcinoma. J Thorac Oncol 2021;16:1753-8. [Crossref] [PubMed]
12. Yu Y, Wang X, Shi C, Hu S, Zhu H, Hu C. Spectral Computed Tomography Imaging in the Differential Diagnosis of Lung Cancer and Inflammatory Myofibroblastic Tumor. J Comput Assist Tomogr 2019;43:338-44. [Crossref] [PubMed]
13. Wu F, Zhou H, Li F, Wang JT, Ai T, Spectral CT. Imaging of Lung Cancer: Quantitative Analysis of Spectral Parameters and Their Correlation with Tumor Characteristics. Acad Radiol 2018;25:1398-404. [Crossref] [PubMed]
14. Fehrenbach U, Kahn J, Böning G, Feldhaus F, Merz K, Frost N, Maurer MH, Renz D, Hamm B, Streitparth F. Spectral CT and its specific values in the staging of patients with non-small cell lung cancer: technical possibilities and clinical impact. Clin Radiol 2019;74:456-66. [Crossref] [PubMed]
15. Tosi D, Pieropan S, Cattoni M, Bonitta G, Franzi S, Mendogni P, Imperatori A, Rotolo N, Castellani M, Cuzzocrea M, Schiorlin I, Casagrande S, De Palma D, Nosotti M, Dominioni L. Prognostic Value of 18F-FDG PET/CT Metabolic Parameters in Surgically Treated Stage I Lung Adenocarcinoma Patients. Clin Nucl Med 2021;46:621-6. [Crossref] [PubMed]
16. Kupik O, Metin Y, Eren G, Orhan Metin N, Arpa M. A comparison study of dual-energy spectral CT and 18F-FDG PET/CT in primary tumors and lymph nodes of lung cancer. Diagn Interv Radiol 2021;27:275-82. [Crossref] [PubMed]
17. Martin SS, Muscogiuri E, Burchett PF, van Assen M, Tessarin G, Vogl TJ, Schoepf UJ, De Cecco CN. Tumorous tissue characterization using integrated 18F-FDG PET/dual-energy CT in lung cancer: Combining iodine enhancement and glycolytic activity. Eur J Radiol 2022;150:110116. [Crossref] [PubMed]
18. Gehling K, Mokry T, Do TD, Giesel FL, Dietrich S, Haberkorn U, Kauczor HU, Weber TF. Dual-Layer Spectral Detector CT in Comparison with FDG-PET/CT for the Assessment of Lymphoma Activity. Rofo 2022;194:747-54. [Crossref] [PubMed]
19. Song H, Cui S, Zhang L, Lou H, Yang K, Yu H, Lin J. Preliminary exploration of the correlation between spectral computed tomography quantitative parameters and spread through air spaces in lung adenocarcinoma. Quant Imaging Med Surg 2024;14:386-96. [Crossref] [PubMed]
20. Wang T, Fan Z, Yue Y, Lu X, Deng X, Hou Y. Predictive value of spectral dual-detector computed tomography for PD-L1 expression in stage I lung adenocarcinoma: development and validation of a novel nomogram. Quant Imaging Med Surg 2024;14:5983-6001. [Crossref] [PubMed]
21. Lin LY, Zhang Y, Suo ST, Zhang F, Cheng JJ, Wu HW. Correlation between dual-energy spectral CT imaging parameters and pathological grades of non-small cell lung cancer. Clin Radiol 2018;73:412.e1-7. [Crossref] [PubMed]
22. Wang K, Liu X, Ding Y, Sun S, Li J, Geng H, Xu M, Wang M, Li X, Sun D. A pretreatment prediction model of grade 3 tumors classed by the IASLC grading system in lung adenocarcinoma. BMC Pulm Med 2023;23:377. [Crossref] [PubMed]
23. Jeon HW, Kim YD, Sim SB, Moon MH. Significant difference in recurrence according to the proportion of high grade patterns in stage IA lung adenocarcinoma. Thorac Cancer 2021;12:1952-8. [Crossref] [PubMed]
24. Travis WD, Asamura H, Bankier AA, Beasley MB, Detterbeck F, Flieder DB, Goo JM, MacMahon H, Naidich D, Nicholson AG, Powell CA, Prokop M, Rami-Porta R, Rusch V, van Schil P, Yatabe Y; International Association for the Study of Lung Cancer Staging and Prognostic Factors Committee and Advisory Board Members. The IASLC Lung Cancer Staging Project: Proposals for Coding T Categories for Subsolid Nodules and Assessment of Tumor Size in Part-Solid Tumors in the Forthcoming Eighth Edition of the TNM Classification of Lung Cancer. J Thorac Oncol 2016;11:1204-23.
25. Fujikawa R, Muraoka Y, Kashima J, Yoshida Y, Ito K, Watanabe H, Kusumoto M, Watanabe SI, Yatabe Y. Clinicopathologic and Genotypic Features of Lung Adenocarcinoma Characterized by the International Association for the Study of Lung Cancer Grading System. J Thorac Oncol 2022;17:700-7. [Crossref] [PubMed]
26. Iwano S, Ito R, Umakoshi H, Ito S, Naganawa S. Evaluation of lung cancer by enhanced dual-energy CT: association between three-dimensional iodine concentration and tumour differentiation. Br J Radiol 2015;88:20150224. [Crossref] [PubMed]
27. Li Q, Li X, Li XY, He XQ, Chu ZG, Luo TY. Histological subtypes of solid-dominant invasive lung adenocarcinoma: differentiation using dual-energy spectral CT. Clin Radiol 2021;76:77.e1-7. [Crossref] [PubMed]
28. Li Q, Li X, Li XY, Huo JW, Lv FJ, Luo TY. Spectral CT in Lung Cancer: Usefulness of Iodine Concentration for Evaluation of Tumor Angiogenesis and Prognosis. AJR Am J Roentgenol 2020;215:595-602. [Crossref] [PubMed]
29. Sun XY, Chen TX, Chang C, Teng HH, Xie C, Ruan MM, Lei B, Liu L, Wang LH, Yang YH, Xie WH. SUVmax of (18)FDG PET/CT Predicts Histological Grade of Lung Adenocarcinoma. Acad Radiol 2021;28:49-57. [Crossref] [PubMed]
30. Deng L, Yang J, Ren T, Jing M, Han T, Zhang B, Zhou J. Can spectral computed tomography (CT) replace perfusion CT to assess the histological classification of non-small cell lung cancer? Quant Imaging Med Surg 2023;13:4960-72. [Crossref] [PubMed]
31. Mu R, Meng Z, Guo Z, Qin X, Huang G, Yang X, Jin H, Yang P, Zhang X, Zhu X. Dual-layer spectral detector computed tomography parameters can improve diagnostic efficiency of lung adenocarcinoma grading. Quant Imaging Med Surg 2022;12:4601-11. [Crossref] [PubMed]
32. Liu BC, Ma HY, Huang J, Luo YW, Zhang WB, Deng WW, Liao YT, Xie CM, Li Q. Does dual-layer spectral detector CT provide added value in predicting spread through air spaces in lung adenocarcinoma? A preliminary study. Eur Radiol 2024;34:4176-86. [Crossref] [PubMed]
33. Zhang Y, Cheng J, Hua X, Yu M, Xu C, Zhang F, Xu J, Wu H. Can Spectral CT Imaging Improve the Differentiation between Malignant and Benign Solitary Pulmonary Nodules? PLoS One 2016;11:e0147537. [Crossref] [PubMed]
34. Ito R, Iwano S, Shimamoto H, Umakoshi H, Kawaguchi K, Ito S, Kato K, Naganawa S. A comparative analysis of dual-phase dual-energy CT and FDG-PET/CT for the prediction of histopathological invasiveness of non-small cell lung cancer. Eur J Radiol 2017;95:186-91. [Crossref] [PubMed]
35. Duan X, Ananthakrishnan L, Guild JB, Xi Y, Rajiah P. Radiation doses and image quality of abdominal CT scans at different patient sizes using spectral detector CT scanner: a phantom and clinical study. Abdom Radiol (NY) 2020;45:3361-8. [Crossref] [PubMed]
36. Wortman JR, Shyu JY, Dileo J, Uyeda JW, Sodickson AD. Dual-energy CT for routine imaging of the abdomen and pelvis: radiation dose and image quality. Emerg Radiol 2020;27:45-50. [Crossref] [PubMed]