The value of interlobar fissure semilunar sign based on multifactor joint analysis in predicting the invasiveness of ground glass nodules with interlobar fissure attachment in the lungs Authors: Wei Zhang, Weishu Hou, Mei Li, Puhe Zhu, Jialong Sun, Zongshan Wu, Bin Liu Published: 2024 ABSTRACT AbstractBackgroundThis study explores the value of interlobar fissure semilunar sign(IFSS) based on multifactor joint analysis in predicting the invasiveness of ground glass nodules(GGNs) with interlobar fissure attachment in the lungs.MethodsThis was a retrospective analysis of clinical data and CT images of 203 GGNs attached to the interlobar fissures confirmed by surgery and pathology. According to pathological results, those GGNs were divided into three groups: glandular precursor lesion (atypical adenomatous hyperplasia/adenocarcinoma in situ), minimally invasive adenocarcinoma (MIA), and invasive adenocarcinoma (IAC). Various quantitative and qualitative parameters were analyzed.ResultsPatient age, maximum diameter, mean size, maximum CT value, and mean CT value differed significantly among the three groups and between with the other group (P < 0.05). The types of GGNs, IFSS, lobulation, spiculation, cavity sign, air bronchogram sign, bronchial changes, and vascular changes had varying degrees of significance in the comparison of each group of lesions. Logistic regression analysis showed that IFSS is one of the important factors in predicting whether GGN is invasive. The regression model I was Logit (P) 1 = -3.578 + 0.272 × 2 + 2.253 × 5, with the area under curve (AUC) for diagnosis of MIA = 0.762. Model III was Logit (P) 3 = -4.494 + 0.376 × 2 + 2.363 × 5, with the AUC for diagnosis of MIA/IAC = 0.881. The sensitivity and specificity of IFSS in model III were 0.961 and 0.458, respectively.ConclusionsThe absence of IFSS in GGNs attached to the interlobar fissure suggests noninvasive lesions. The logistic regression model based on multi factor joint analysis IFSS and maximum diameter can better predict whether the GGN attached to the interlobar fissure pleura is invasive.Keywords: Lung adenocarcinoma, Ground-glass nodule, Interlobar fissure, CT, Pathology Background The incidence rate and mortality of lung cancer are high according to data released by the International Agency for Research on Cancer in 2020 [1], and the incidence rate is increasing annually. Lung adenocarcinoma is the most common pathological type of lung cancer [2], and early lung adenocarcinoma often presents as ground glass nodules (GGNs) on computed tomography (CT) [3]. GGNs can be divided into pure and mixed types, according to whether they contain solid components. Previously, the multidisciplinary classifications of lung adenocarcinoma were: adenocarcinoma in situ (AIS), atypical adenomatous hyperplasia (AAH), minimally invasive adenocarcinoma (MIA), and invasive adenocarcinoma (IAC) [4]. In the 2021 WHO classification of chest tumors, AIS and AAH were classified as precursor glandular lesions [5]. However, their diagnostic criteria remained unchanged, reflecting the progression of tumors from the same origin, from abnormal proliferation to carcinoma in situ and invasive lesions. There are differences in follow-up strategies, surgical plans, and prognosis among different pathological subtypes [6–9]. Patients with AAH and AIS can choose whether to continue follow-up. Surgical resection is often recommended for MIA and IAC, and segmental or wedge-shaped lung resection for MIA and AAH/AIS. Lobectomy as soon as possible is recommended for IAC, and patients with AIS/MIA can achieve a 10-year specific survival rate of 100% if they undergo surgical treatment to completely remove the lesion. Therefore, preoperative prediction of pathological subtypes is particularly important. Previous studies have focused on quantitative parameters and CT signs of GGNs [10–15], and the relationship between GGNs and the pleura is the main feature of studying the pathological subtypes of GGNs [11, 16]. In practice, we often encounter GGNs that are closely related to the pleura. Here, we divide the relationship between GGNs and the pleura into four types: (1) GGNs not related to the pleura (Fig. 1A); (2) GGNs not in contact with the pleura, but the strip shadow at the edge stretches the pleura and causes indentation (Fig. 1B); (3) GGNs in contact with the pleura, but the pleural shape remains unchanged (Fig. 1C); and (4) GGNs in contact with the pleura, resulting in an arc-shaped indentation of the interlobar fissure, causing a semilunar shape in the GGNs (Fig. 1D). We refer to the last type as the interlobar fissure semilunar sign (IFSS), but IFSS is difficult to observe when GGNs come into contact with the chest wall pleura. After analyzing patients with GGNs who underwent pulmonary nodule surgery in our hospital, IFSS was not uncommon and can easily cause arc-shaped depressions with interlobar fissures. Therefore, in this study we analyzed the significance of IFSS in predicting the pathological subtypes of GGNs with interlobar fissure attachment. We aimed to provide a reference and basis for early diagnosis, follow-up, and treatment of such GGNs. Classification of the relationship between GGNs and the pleura. (A) GGNs had no relationship with pleura. (B) There was a strip shadow at the edge of the GGNs that pulled the pleura, and the local pleura appeared to have a triangular depression. (C) GGNs in contact with the pleura, but there was no significant change in the shape of the pleura. (D) GGNs in contact with the pleura, resulting in pleural indentation in interlobular fissures; GGNs changed to a semilunar shape to form IFSS Methods Patients This retrospective study was approved by our institutional Ethics Review Committee (Approval No. 2023LLKS008), and the requirement for informed consent was waived. We collected 2933 lung nodules from 2443 patients at our hospital, who underwent surgery and had complete clinical and imaging data of ≤ 3 cm. We excluded 1295 solid nodules, 203 benign GGNs, and 1222 GGNs without contact with the interlobar fissure. We included 203 GGNs with interlobar fissure attachment from 201 patients (two of whom had two GGNs). There were 24 cases of AAH/AIS, 74 of MIA, and 105 of IAC. One hundred and eighty-five GGNs had IFSS. We defined IFSS as the sign of GGN attached to the interlobar fissure causing interlobar fissure arc-shaped depression, and the GGN itself becoming semilunar. The screening process is shown in Fig. 2. There were 64 male and 137 female patients, aged 22–82 years, with an average of 57.98 ± 11.90 years. Inclusion criteria were: (1) GGNs with interlobar fissure attachment; (2) internal CT scan performed 1 month before surgery; (3) PACS memory contained complete preoperative CT images, with a thin-layer reconstruction layer thickness ≤ 1.25 mm; (4) maximum diameter of the selected lesions ≤ 3 cm; and (5) no puncture biopsy or antitumor treatment performed before CT examination. Exclusion criteria were: (1) solid nodules; (2) pathological diagnosis of benign GGNs; (3) GGNs without contact with the interlobar fissure; and (4) incomplete clinical and imaging data. Screening process for GGNs with interlobar fissure attachment Pathological diagnosis All specimens were obtained through thoracoscopy or thoracotomy biopsy. The surgical specimens were fixed with 10% neutral formalin, and all lesions were taken. Paraffin embedding, sectioning, and routine hematoxylin and eosin (HE) staining were performed. Diagnosis was made by two experienced pathologists. On the basis of light microscopy, immunohistochemical staining was performed to clarify the tumor classification for those with difficulty in diagnosis, and if necessary, elastic fiber staining was performed to determine the degree of tumor cell invasion of the pleura. If there were differences of opinion, a consensus was reached through discussion. CT protocol CT scans were performed using the Neusoft NeuViz 64In 64-slice, United Imaging uCT760 64-slice, Philips Brilliance 64-slice, GE Tevolution 256-slice or Optima 680Q 64-slice scanner. All patients were in the supine position, with the head advanced, arms raised, and the scanning range from the top to bottom of the lungs, including the chest wall and armpits on both sides. All parameters were obtained from CT plain scan images and reconstructed using standard algorithms. The scanning parameters were: tube voltage 120 kV, tube current automatically adjusted or 120–180 mA, layer thickness 5.0 mm, reconstruction layer thickness 0.625–1.25 mm, reconstruction interval the same as reconstruction layer thickness, pitch 0.9–1.375, and matrix 512 × 512. Image analysis Two doctors with 14 and 30 years’ experience in chest imaging diagnosis analyzed the images separately without knowing the histological examination results. If their opinions differed, they reached an agreement through discussion. Observations were conducted at the lung window (width 1500 HU, level −600 HU) and mediastinal window (width 350 HU, level 40 HU), using multiple plane reconstruction (MPR), maximum intensity projection and minimum intensity projection for multidirectional observation and optimization of bronchial and vascular display. Quantitative parameters (1) Maximum diameter (mm) of the lesion displayed on the MPR lung window image in multiple directions. (2) Mean size of the lesion (mm): maximum and perpendicular diameters of the largest layer of the lesion in the axial position, as well as the upper and lower diameters of the largest layer of the lesion in the coronal reconstruction image. (3) Maximum CT value (HU): we took 10 mm2 regions of interest (ROIs) and repeatedly measured the relatively high-density area of the lesion and obtained the maximum value. (4) Mean CT value (HU): after selecting standard axial, sagittal, and coronal lung window images, we used irregular graphics tools to draw the boundaries of the nodules as ROIs, and excluded blood vessels and bronchi from the ROIs as much as possible. Record the CT values and take the average value to represent the average CT value of the lesion. For mixed GGNs, the selected measurement plane was the image that displayed the largest nodule and included the largest image of the solid part as much as possible. Qualitative parameters Qualitative parameters included: lobulation; spiculation; vacuole and cavity signs; bronchial features [(air bronchogram sign, bronchial changes (truncation, stenosis, or dilation)]; tumor lung interface (clear or blurry); and vascular features [vascular passage, vascular changes (thickening, twisting, or aggregation)]. Statistical methods We used SPSS version 23.0 (IBM Corporation, Armonk, NY, USA) for statistical analysis. The Kolmogorov–Smirnov test was used to test the distribution normality of the quantitative parameters. If they conformed to a normal distribution, they were represented by mean ± standard deviation. Three groups were compared using one-way analysis of variance (ANOVA). The Tukey–Kramer test was used for intergroup pairwise comparison, and independent sample t-test was performed for comparison between the other two groups. The quantitative parameters data that did not conform to a normal distribution were represented by M (Q1, Q3). The Kruskal–Wallis H test was used for comparison among the three groups, and the Nemenyi method for multiple comparisons between the groups. The Mann–Whitney U test was used for comparison between the other two groups. Qualitative parameters using Pearson’s χ2 test and Fisher’s test (when the expected value was < 5), and perform multiple comparisons between groups. We performed intergroup binary logistic regression analysis (forward method) on quantitative and qualitative parameters that differed significantly (P < 0.05) between each group, and saved the predictive probability values to evaluate independent risk factors for GGN pathological subtypes. Using the predicted probability values and corresponding independent risk factors as test variables, we drew the receiver operating characteristic (ROC) curve, calculated the area under the curve (AUC), and obtained the diagnostic threshold, sensitivity, and specificity of each group of variables. Results Basic data analysis (Table 1) Table 1Basic dataGroupAAH/AIS (n = 24)MIA (n = 74)IAC (n = 105)P-valueaP-valuebP-valuecAge49.29 ± 13.2155.11 ± 11.44*62.00 (55.00 ~ 70.00)*< 0.001< 0.001< 0.001GenderMale6 (25.00%)24 (32.43%)34 (32.38%)0.7640.4640.786Female18 (75.00%)50 (67.57%)71 (67.62%)LocationRUL421330.7610.3670.885RML61217RLL91826LUL2610LLL31719Smoking history720250.8120.6710.542Basic lung diseases0011.0001.0000.333Former history of lung cancer0550.5950.6110.911Former malignancy history (except lung cancer)1360.8981.0000.814Number of GGNsingle1850750.7470.6020.750more62430RUL, right upper lobe; RML, right middle lobe; RLL, right lower lobe; LUL, left upper lobe; LLL, left lower lobe*Significant differences between the current and previous groupaComparison of AAH/AIS, MIA and IACbComparison between AAH/AIS and MIA/IACcComparison between AAH/AIS/MIA and IAC Basic data RUL, right upper lobe; RML, right middle lobe; RLL, right lower lobe; LUL, left upper lobe; LLL, left lower lobe *Significant differences between the current and previous group aComparison of AAH/AIS, MIA and IAC bComparison between AAH/AIS and MIA/IAC cComparison between AAH/AIS/MIA and IAC We included 203 GGNs in contact with the interlobar fissure from 201 patients who underwent surgery at our Hospital, with complete clinical data. There were 24 cases of AAH/AIS (Fig. 3A–C), 74 cases of MIA (Fig. 4A–C), and 105 cases of IAC (Fig. 5A–C) showed statistically significant age differences (P < 0.001). The age differences between AAH/AIS and MIA, MIA and IAC, AAH/AIS and MIA/IAC, and AAH/AIS/MIA and IAC were also significant. There were no significant differences in gender, smoking history, history of underlying lung diseases (chronic obstructive pulmonary disease, pulmonary fibrosis, or tuberculosis), personal history of lung cancer, history of tumors other than lung cancer, and the GGN number of patients. (A–C) Pathologically confirmed AIS, with CT findings of pure GGN, clear boundaries, and deformation of the adjacent interlobar fissures. GGN presented with IFSS. Pathology (HE staining, 40×): tumor cells were seen growing close to the alveolar wall, limited to the alveoli. The morphology of the alveoli near the pleura was altered, with no invasion of the interlobular fissure pleura (indicated by arrow) (A–C) Pathologically confirmed as MIA, with CT findings of mixed GGN with clear boundaries. The adjacent interlobar fissures were deformed, and the GGN presented with IFSS. Pathology (HE staining, 40×): the tumor cells in the lesion mainly grew along the alveolar wall, forming an acinar structure in the focal area. The maximum diameter of the invasive range was < 5 mm. The subpleural alveoli were deformed, and the interstitial structure created tension to stretch the pleura. The pleura was not invaded (indicated by arrow) (A–C) Pathologically confirmed as IAC, with CT findings of mixed GGN, irregular in shape, and with visible lobular signs. The adjacent interlobar fissures were deformed, and the GGN presented with IFSS. Pathology (HE staining, 40×, small image in the bottom left corner is 100×): the tumor cells in the lesion mainly grew as acinar, papillary, and micropapillary types. The growth of tumor cells generated tension to stretch the pleura, and in this case, the pleura of the interlobular fissure was invaded (indicated by arrow) Quantitative and qualitative parameter analysis (Table 2) Table 2Quantitative and qualitative parameter analysisGroupAAH/AIS (n = 24)MIA (n = 74)IAC (n = 105)P-valueaP-valuebP-valuecNodule TypemGGN4 (16.67%)35 (47.30%)*81 (77.14%)*< 0.001< 0.001< 0.001PGGN20 (83.33%)39 (52.70%)24 (23.86%)IFSS13 (54.17%)69 (93.24%)*103 (98.10%)< 0.001< 0.001< 0.001Tumour-lung interfaceClear20 (83.33%)62 (83.78%)98 (93.33%)*0.0690.4880.030vague4 (16.67%)12 (16.22%)7 (6.67%)lobulation1 (4.17%)10 (13.51%)45 (42.86%)*< 0.0010.006< 0.001Spiculation0 (0.00%)3 (4.05%)36 (34.29%)*< 0.0010.006< 0.001Bubble lucency sign1 (4.17%)9 (12.16%)22 (20.95%)0.0820.1360.036Cavity sign0 (0.00%)0 (0.00%)12 (11.43%)*0.0020.3670.001BronchusAir Bronchogram sign1 (4.17%)1 (1.35%)29 (27.62%)*< 0.0010.136< 0.001Bronchial change0 (0.00%)0 (0.00%)28 (26.67%)*< 0.0010.212< 0.001vascularcrossing23 (95.83%)73 (98.65%)103 (98.10%)0.5830.3980.944Change4 (16.67%)20 (27.03%)71 (67.62%)*< 0.0010.002< 0.001Maximum diameter (mm)9.72 ± 2.2111.44(9.49 ~ 13.80)*19.41(15.27 ~ 25.44)*< 0.001< 0.001< 0.001Mean size (mm)8.52 ± 1.939.85(8.36 ~ 12.05)*16.26(13.43 ~ 20.88)*< 0.001< 0.001< 0.001Maximum CT value (HU)−500.65 ± 156.44−435.40(− 561.08 ~ − 255.85)*−117.70(− 280.25 ~ − 23.15)*< 0.001< 0.001< 0.001Mean CT value (HU)−642.63 ± 90.70−621.18 ± 97.33−465.12 ± 147.34*< 0.001< 0.001< 0.001*Significant differences between the current and previous groupaComparison of AAH/AIS, MIA and IACbComparison between AAH/AIS and MIA/IACcComparison between AAH/AIS/MIA and IAC Quantitative and qualitative parameter analysis *Significant differences between the current and previous group aComparison of AAH/AIS, MIA and IAC bComparison between AAH/AIS and MIA/IAC cComparison between AAH/AIS/MIA and IAC Among the AAH/AIS, MIA, and IAC groups, there were significant differences in GGN type, IFSS, lobulation, spiculation, cavity sign, air bronchogram sign, bronchial changes, vascular changes, maximum diameter, mean diameter, maximum CT value, and mean CT value (P < 0.05). There was no significant difference in tumor–lung interface, vacuolar sign, and vascular crossing sign. The percentage of CT signs with significant differences among the three groups is shown in Fig. 6; among them, vascular changes, mixed GGN, and IFSS had the highest proportion in each group of lesions. There were significant differences between the AAH/AIS and MIA groups for GGN type, IFSS, maximum diameter, mean diameter, and maximum CT value (P < 0.05), but not for the other parameters. There were significant differences between the MIA and IAC groups in terms of GGN type, tumor–lung interface, lobulation, spiculation, cavity sign, air bronchogram sign, bronchial changes, and vascular changes (P < 0.05). The maximum diameter, mean diameter, maximum CT value, and mean CT value showed no significant differences. There were significant differences between the AAH/AIS and MIA/IAC groups in terms of GGN type, IFSS, tumor–lung interface, lobulation, spiculation, and vascular changes (P < 0.05). The maximum diameter, mean diameter, maximum CT value, and mean CT value showed no significant differences. There were significant differences between the AAH/AIS/MIA and IAC groups in terms of GGN type, IFSS, tumor–lung interface, lobulation, spiculation, cavity sign, air bronchogram sign, bronchial changes, vascular changes, maximum diameter, mean diameter, maximum CT value, and mean CT value (P < 0.05), but not for the other parameters. Comparison of CT signs (with statistical differences) and their respective percentages among AAH/AIS, MIA, and IAC groups Logistic regression analysis and ROC curve analysis between groups Logistic regression analysis with attached values: average size X1, maximum diameter X2, mean CT value X3, maximum CT value X4, IFSS X5, tumor–lung interface X6, where X5 and X6 have positive attached values of 1 and negative attached values of 0. Comparison between AAH/AIS and MIA We performed binary logistic regression analysis on statistically significant quantitative and qualitative parameters from the AAH/AIS and MIA groups, and saved the predicted probability values. IFSS X5 (P = 0.001) and maximum diameter X2 (P < 0.019) were independent risk factors for determining the grouping of AAH/AIS and MIA. Regression model I obtained was: Logit (P) 1 = −3.578 + 0.272 × 2 + 2.253 × 5. This model was tested for likelihood ratio (P < 0.001), indicating statistical significance. The predicted probability value, IFSS, and maximum diameter were used as test variables to draw the ROC curve (Fig. 7). AUC obtained by discriminating MIA was 0.762, 0.650, and 0.704, respectively (Table 3), with sensitivity of 0.986, 0.932, and 0.568, specificity of 0.458, 0.458, and 0.75, and the best cutoff values for predicting probability and maximum diameter were 0.396 and 11.035. Graph shows ROC curves predicted AAH/AIS and MIA value, IFSS, maximum diameter, and predicted probability, and the areas under the curve for the identification of MIA were 0.650, 0.704 and 0.762, respectively Table 3ROC analysis between groupsVariableAUCSensitivitySpecificityThreshold AAH/AIS and MIA IFSS0.6500.9320.458Yes/NoMaximum diameter0.7040.5680.7511.035Predicted probability0.7620.9860.4580.396 MIA and IAC Tumour-lung interface0.5480.9330.162Clear/vagueMaximum diameter0.8940.790.87815.055Mean CT value0.8000.6670.851−528.6Predicted probability0.9320.7810.9190.648 AAH/AIS and MIA/ IAC IFSS0.7100.9610.458Yes/NoMaximum diameter0.8560.6420.95812.945Predicted probability0.8810.620.9580.578 MIA and IAC Tumour-lung interface0.5480.9330.163Clear/vagueMaximum diameter0.9110.790.90815.055Mean CT value0.8110.6480.878−511.95Predicted probability0.9430.8860.8370.387 ROC analysis between groups Comparison between MIA and IAC We performed binary logistic regression analysis on statistically significant quantitative and qualitative parameters from the MIA and IAC groups, and saved the predicted probability values. Tumor–lung interface X6 (P = 0.013), maximum diameter X2 (P < 0.001), and mean CT value X4 (P < 0.001) were independent risk factors for determining the grouping of MIA and IAC. Regression model II obtained was: Logit (P) 2 = −3.583 + 0.465 × 2 + 0.008 × 4 + 1.987 × 6. This model was tested for likelihood ratio (P < 0.001) and showed statistical significance. The predicted probability value, maximum diameter, mean CT value, and tumor–lung interface were used as test variables to draw the ROC curve (Fig. 8). The AUC obtained by discriminating IAC was 0.932, 0.894, 0.800, and 0.548, respectively (Table 3), with sensitivity of 0.781, 0.790, 0.667, and 0.933, specificity of 0.919, 0.878, 0.851, and 0.162, respectively. The best cutoff values for predicting probability, maximum diameter, and mean CT value were 0.648, 15.055, and −528.60. Graph shows ROC curves predicted MIA and IAC value, tumor–lung interface, maximum diameter, mean CT value, and predicted probability, and the areas under the curve for the identification of IAC were 0.548, 0.894, 0.800 and 0.932, respectively Comparison between AAH/AIS and MIA/IAC We performed binary logistic regression analysis on statistically significant quantitative and qualitative parameters from the AAH/AIS and MIA/IAC groups, and saved the predicted probability values. IFSS X5 (P < 0.001) and maximum diameter X2 (P < 0.001) were independent risk factors for determining the grouping of AAH/AIS and MIA/IAC. Regression model III obtained was: Logit (P) 3 = −4.494 + 0.376 × 2 + 2.363 × 5. This model was tested for likelihood ratio (P < 0.001), indicating statistical significance. The predicted probability value, maximum diameter, and IFSS were used as test variables to draw the ROC curve (Fig. 9). The AUC obtained by discriminating MIA/IAC was 0.881, 0.856, and 0.710, respectively (Table 3), with sensitivity of 0.62, 0.642, and 0.961, and specificity of 0.958, 0.958, and 0.458, respectively. The best cutoff values for predicting probability and maximum diameter were 0.578 and 12.954. Graph shows ROC curves predicted AAH/AIS and MIA/IAC value, IFSS, maximum diameter, and predicted probability, and the areas under the curve for the identification of MIA were 0.710, 0.856 and 0.881, respectively Comparison between AAH/AIS/MIA and IAC We performed binary logistic regression analysis on statistically significant quantitative and qualitative parameters from the AAH/AIS/MIA and IAC groups, and saved the predicted probability values. Tumor–lung interface X6 (P = 0.011), maximum diameter X2 (P < 0.001), and average CT value X4 (P < 0.001) were independent risk factors for determining the grouping of AAH/AIS/MIA and IAC. The obtained regression model IV was: Logit (P) 4 = −4.169 + 0.5 × 2 + 0.008 × 4 + 2.012 × 6. This model was tested for likelihood ratio (P < 0.001) and showed statistical significance. The predicted probability value, maximum diameter, mean CT value, and tumor–lung interface were used as test variables to draw the ROC curve (Fig. 10). The AUC obtained by discriminating IAC was 0.943, 0.911, 0.811, and 0.548, respectively (Table 3), with sensitivity of 0.886, 0.79, 0.648, and 0.933, and specificity of 0.837, 0.908, 0.878, and 0.163, respectively. The best cutoff values for predicting probability, maximum diameter, and maximum CT value were 0.387, 15.055, and −511.95. Graph shows ROC curves predicted AAH/AIS/MIA and IAC value, tumor–lung interface, maximum diameter, mean CT value, and predicted probability, and the areas under the curve for the identification of IAC were 0.548, 0.911, 0.811 and 0.943, respectively Imaging and corresponding pathological findings of IFSS CT images showed that 185 of 203 GGNs attached to interlobar fissures had IFSS. Twenty-four glandular precursor lesions showed tumor cells growing along the alveolar wall, limited to the alveoli. Among these, 13 cases of GGNs with IFSS were closer to the pleura, and the growth of tumor cells created tension, causing the alveoli to deform and concave changes in the pleura (Fig. 2C). According to the degree of tumor invasion of the pleura (TNM 8th edition) [17], all those 24 cases of glandular precursor lesions belong to PL0. The 74 cases of MIA showed that tumor cells in the lesion mainly grew along the alveolar wall, with the maximum diameter of tumor cell growth and invasion being < 5 mm. Among these cases, 69 GGNs with IFSS had a close relationship with the pleura. The growth of tumor cells created tension, causing changes in alveolar morphology, and the tension extended to the pleura, and the pleura showed concave transformation, with no invasion in the pleura (Fig. 3C). All those 74 cases of MIA belong to PL0. The 105 cases of IAC showed that the tumor cells in the lesion mainly grew as acinar, papillary, and micropapillary types (Fig. 3C), and the tumor cell growth stretched the interstitial structure, pleura, or invaded the pleura, resulting in a concave change in the pleura. Seventy-one out of 105 cases were classified as PL0, 30 as PL1, and 4 as PL2. Discussion We retrospectively analyzed the imaging features of patients with GGNs with interlobar fissure attachment confirmed by pathology in our hospital. We also explored the value of IFSS based on multifactor joint analysis in predicting the invasiveness of GGNs with interlobar fissure attachment in the lungs. There has been controversy in previous studies regarding the differences in age in group comparisons [13, 14, 18]. Our results showed that as age increased, the pathological grading of GGNs also increased accordingly. However, there was no significant difference in gender, smoking history, basic lung diseases, and previous history of lung cancer or other malignancy, which is similar to previous studies [15, 18]. We also analyzed and compared single and multiple GGNs in the same patient, and found that the number of lesions had no significance in determining the pathological subtypes of GGNs. The morphological characteristics of GGNs have always been an important reference for predicting their pathological subtypes. Previous studies have mostly analyzed them from the perspective of pleural traction depression [3, 12, 18, 19]. However, IFSS in GGNs with interlobar fissure contact in this study was not separately analyzed in previous studies. This study showed that IFSS was significant in determining whether there was invasion of GGNs attached to interlobar fissures. Pathologically, in GGNs with IFSS, tumor cell growth created tension, causing alveolar deformation, stretching or invading the pleura, resulting in concave changes in the pleura. Nevertheless, IFSS has no clear significance in determining the degree of invasion. According to the degree of tumor invasion of the pleura (TNM 8th edition), all precursor glandular lesions, MIA, and most IAC were in PL0, with only a few IAC cases belonging to PL1 or PL2. The IFSS investigated in this study did not find a clear correlation with the degree of pleural invasion. Previous studies [12, 15, 18–21] have shown that conventional CT features (such as tumor–lung interface, lobulation sign, spicule sign, etc.) have varying value in predicting pathological grading of GGNs, which is consistent with the results of this study. If the tumor–lung interface of GGNs is clear or has other features (such as lobulation sign, spicule sign, or vascular changes, etc.), the likelihood of GGNs being MIA or IAC is higher. However, we found no significant difference in the characteristics of blood vessels passing through the GGNs in the pathological classification, which is consistent with the study of Zhang et al. [22]. In addition, the types of GGNs (pure or mixed) are not only related to pathological subtypes but also to the size of solid components and degree of infiltration [21]. Because most of the glandular precursor lesions collected in this study were pGGNs, and only four cases were mGGNs, the size of solid components was not included in this study. But our results also showed that the type of those GGNs was related to pathological subtypes, and mixed GGNs were more likely to be MIA or IAC. The size of GGNs has always been considered an important factor in determining their pathological subtypes [18–21], and the correlation between CT values and GGNs pathological subtypes is controversial [19, 21, 23–25]. However, previous studies had limited analysis of GGN attachment to interlobular fissures. Our results showed that, except for average CT value, which showed no significant difference between AAH/AIS and MIA (possibly related to only analyzing the selective bias of GGNs attached to interlobar fissures), other quantitative parameters showed significant differences among each group. This indicated that, as the size and density of GGNs attached to interlobar fissures gradually increased, the likelihood of MIA or IAC increased. The comprehensive analysis of regression models in previous studies [12] has high diagnostic value for the pathological subtypes of GGNs. The GGNs analyzed in this study had specificity, as the growth of GGN tumor cells attached to the pleura of interlobar fissures stretched the surrounding interstitial structure, pleura, or invaded the pleura, leading to the formation of IFSS due to pleural indentation. This study found that IFSS is one of the important reference factors for evaluating the GGNs. The absence of IFSS in the GGNs suggests noninvasive lesions. Combined with other significantly different parameters, the efficacy of regression models I and III in diagnosing MIA and MIA/IAC was significantly improved. However, IFSS was not an independent risk factor for diagnosing the degree of GGN invasion (distinguishing MIA from IAC). Therefore, when the GGNs attached to the interlobular fissure have IFSS and are considered more invasive in combination with maximum diameter, we recommend short-term follow-up or surgical resection. There were some limitations to this study. (1) There is not yet a deeper understanding of IFSS, and further research is needed in subsequent pathological analysis. (2) We only collected GGNs with interlobar fissure attachment, and all data came from one institution, which may have led to selection bias. (3) When measuring quantitative parameters, blood vessels and bronchioles could not always be completely excluded, which may have led to differences in quantitative parameter measurement. Conclusion There is a high probability of the appearance of IFSS in GGNs with interlobar fissure attachment. The absence of IFSS in the GGNs suggests noninvasive lesions, but IFSS cannot predict the degree of invasion (i.e., distinguishing MIA from IAC). The logistic regression model based on multifactor joint analysis of IFSS and maximum diameter can better predict whether the GGNs attached to the interlobar fissure pleura are invasive. Acknowledgements We thank International Science Editing ( http://www.internationalscienceediting.com) for editing this manuscript. Abbreviations Interlobar fissure semilunar sign Ground glass nodule Atypical adenomatous hyperplasia Adenocarcinoma in situ Minimally invasive adenocarcinoma Invasive adenocarcinoma Area under curve Region of interest Author contributions WZ carried out the study, and wrote the first draft. BL designed the study, analysed data and revision. WH helped in writing-review & editing. PZ, JS, ZW helped in CT scanning and data Collection. ML performed all histological preparations and analyses. All authors read and approved the final manuscript. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Data availability The data supporting this study are not publicly available due to privacy or ethical restrictions but can be obtained from the corresponding author upon reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Medical School Committee of the Lu’an Hospital Affiliated to Anhui Medical University (Approval No. 2023LLKS008), and informed consent was waived. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data supporting this study are not publicly available due to privacy or ethical restrictions but can be obtained from the corresponding author upon reasonable request.