RPS 2304 - Lung cancer imaging: characterisation and prognosis

Author Block: L. Biggemann, H. Bohnenberger, J. Vincke, P. Kraus, T. Overbeck, A. Hammerstein-Eqourd, J. Uhlig; Göttingen/DE
Purpose: Pulmonary adenocarcinoma (AC) can present with different pathological growth patterns. This study evaluates whether these growth patterns correlate with the radiological morphology of the tumor on computed tomography (CT).
Methods or Background: Patients with surgically resected pulmonary AC and preoperative CT imaging were retrospectively included. Cases were sampled to distribute growth patterns approximately evenly. Pathological growth patterns were assessed on a representative pathological slice. The predominant growth pattern was defined as >=60%. Radiological morphology was assessed using preoperative thoracic CT scans and compared across growth patterns using the chi-square test.
Results or Findings: A total of n=345 patients were included (43.8% female; median age 68 years). Pathological growth patterns were acinar (n=68), lepidic (n=51), micropapillary (n=58), papillary (n=50), mucinous (n=51), and solid (n=67).
While age was balanced across AC growth patterns, lepidic and micropapillary ACs were more common in women (56.9% / 51.7%); and solid and acinar ACs more likely in men (71.6% / 57.4%; overall p <0.05).
Pathological growth patterns demonstrated specific morphologies on CT regarding nodule type, margin, ground glass opacities, central low attenuation, air bronchogram, associated lymphadenopathy and location of distant metastases (each variable p<0.05).
For example, lepidic ACs commonly presented as ground-glass or partially solid nodules (13.7% / 35%); acinar ACs with associated ground-glass opacities (77.9%); solid ACs with contact to pulmonary fissures (50.7%) and central low attenuation (58.2%); and mucinous ACs with air bronchogram (41.2%).
Conclusion: Radiological morphology of pulmonary ACs on CT correlates well with pathological growth patterns, which could aid guiding diagnostic and treatment patterns.
Limitations: Limitations include that only a representative pathological slice of the pulmonary AC was assessed, whereas growth patterns might vary in the full tumor volume; and that CT-imaging was performed on different CT-scanners, introducing heterogeneity.
Funding for this study: This study has been supported by Siemens Healthineers.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ehtics committee of the University Medical Center Goettingen

Author Block: Y. Ozawa, D. Takenaka, H. Nagata, T. Ueda, M. Nomura, T. Yoshikawa, Y. Ohno; Toyoake/JP
Purpose: To determine the influence of radiation dose and reconstruction method on artificial intelligence (AI)-based computer-aided volumetry (CADv) for nodule component measurement and diagnostic performance to evaluate invasiveness in lung adenocarcinoma on high-definition CT (HDCT).
Methods or Background: 112 consecutive patients with 181 lung adenocarcinomas underwent thin-section HDCTs at standard-dose (SDCT: 9.0±1.8 mGy), reduced-dose (RDCT: 1.7±0.2 mGy) and ultra-low-dose (ULDCT: 0.8±0.1 mGy) levels. All HDCT data were reconstructed with hybrid-type iterative reconstruction (IR) and deep learning reconstruction (DLR). Then, standard references for solid and GGO components and consolidation-to-tumor ratio (CTR) were computationally determined with the simultaneous truth and performance level estimation (STAPLE) method from annotated CT data by three board-certified chest radiologists. Then, each component volume and CTR on all HDCT data were measured by AI-based CADv software. Each component volume and consolidation-to-tumor ratio (CTR) were correlated between CADv measurement on each CT data and standard reference. Then, Measurement differences of each index between standard reference and each CADv measurement were compared among all CT data by Tukey’s HSD test. Finally, diagnostic performance of invasiveness was compared among all CTR measurements by ROC analysis.
Results or Findings: There were significant correlations for each component and volume and CTR on all HDCTs (hybrid-type IR: 0.71≤r≤0.88, p<0.0001; DLR: 0.71≤r≤0.88, p<0.0001). Mean differences between each CADv measurement and standard reference had no significant differences among all HDCTs (p>0.05). Area under the curve (AUC) of each HDCT with DLR (SDCT: AUC=0.98, RDCT: AUC=0.98, ULDCT: AUC=0.97) was significantly larger than all HDCTs with hybrid-type IR (AUC=0.95, p<0.05).
Conclusion: Reconstruction method rather than radiation dose reduction had significantly affected to diagnostic performance of AI-based CADv for invasiveness evaluation in lung adenocarcinoma on HDCT.
Limitations: Number of subjects
Funding for this study: Reconstruction method is more important than radiation dose reduction for keeping diagnostic performance of AI-based CADv for invasiveness evaluation in lung ademocarcinoma on high-definition CT.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Research grants from Canon Medical Systems Corporation and Smoking Research Foundation

Author Block: A. Olesinski, R. Lederman, Y. Azraq, J. Sosna, L. Joskowicz; Jerusalem/IL
Purpose: Detection of enlarged mediastinal lymph nodes (LNs) in CECT chest scans is necessary for the assessment of oncology patients. It is performed by manually measuring the short axis length (SAL) of the LNs. We aimed to quantify the interobserver agreement and variability of the SAL measurements and to compare them to automatically computed SALs from volumetric LN delineations.
Methods or Background: We retrospectively analyzed 40 CECT chest scans from patients with mediastinal lymphadenopathy. Two radiologists with 25 years and 35 years of experience independently measured the SALs of 471 LNs by: identifying the scan slice in which the LN appears largest, drawing the line segment along its short axis; and classifying the LN as normal (SAL<10mm) or enlarged (SAL≥10mm). Four weeks later, one of the radiologists manually segmented the volume of the LN's along its contours. The LNs were automatically classified as normal/enlarged based on their SALs computed from the delineation (COMP). Confusion matrices were computed, as well as the differences between the SALs.
Results or Findings: The normal/enlarged LN overall agreement (371 normal LNs, 49 enlarged LNs) between both radiologists was 94.8% (420/471). For agreement/disagreement, the SAL differences (std) were 1.2(1.1)mm and 3.5(3.2)mm. Note that the SALs difference is nearly twice as large for disagreement as for agreement. The normal/enlarged overall agreement between the manual and the automatically computed SALs and both radiologists was 94.8%(421/471) and 92.4%(411/471).
Conclusion: Identification of enlarged mediastinum lymph nodes in chest CECT based on short axis measurements derived manually or computed from lymph nodes delineations has high accuracy. The agreement of the computed SALs from LN contour delineations is within the manual SALs interobserver variability. Accurate automatic LN segmentations can be obtained with existing methods.
Limitations: Single institution and single contour delineation.
Funding for this study: None.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: Yes, Helsinki Committee of the Hadassah University Medical Center

Author Block: N. Meng¹, X. Liu¹, J. Pan¹, X. Yu¹, Y. Wu¹, Y. Yang², Z. Wang³, M. Wang¹; ¹Zhengzhou/CN, ²Bei Jing/CN, ³Shanghai/CN
Purpose: To investigate the value of multiparametric 18F-FDG PET/MRI based on tri-compartmental restrictive spectrum imaging (RSI), amide proton transfer-weighted imaging (APTWI), and diffusion-weighted imaging (DWI) in the assessment of lymph node metastases (LNM) of non-small cell lung cancer (NSCLC)
Methods or Background: A total of 152 NSCLC patients were enrolled. 18F-FDG PET- derived parameter (SUVmax), RSI-derived parameters (f1, f2, and f3), APTWI-derived parameter (MTRasym(3.5ppm)), DWI-derived parameter (ADC), and were calculated. Logistic regression analysis was used to identify independent predictors, and combined diagnostics. AUC, calibration curves and decision curve analysis (DCA) were employed to assess the performance of the combined diagnostics.
Results or Findings: MTRasym(3.5ppm), SUVmax, f2, and f3 were higher and ADC and f1 were lower in LNM-positive group than in LNM-negative group (all P < 0.05). Maximum lesion diameter, f1, MTRasym(3.5ppm) , SUVmax, and ADC were independent predictors of LNM status in NSCLC patients, and the combination of them had an optimal diagnostic efficacy (AUC = 0.978; sensitivity = 95.35 %; specificity = 90.91 %), which was significantly higher than maximum lesion diameter, f1, MTRasym(3.5ppm), SUVmax, and ADC (AUC = 0.774, 0.810, 0.832, 0.834, and 0.783, respectively, and all P < 0.01). The combined diagnosis showed a good performance (AUC = 0.968) in the bootstrap (1000 samples)-based internal validation. Calibration curves and DCA demonstrated that the combined diagnosis not only provided better stability, but also resulted in a higher net benefit for the patients involved.
Conclusion: Multiparametric 18F-FDG PET/MRI based on RSI, APTWI, and DWI is beneficial for the non-invasive assessment of LNM status in NSCLC, and the combination of maximum diameter, f1, MTRasym(3.5ppm), SUVmax, and ADC may serve as a promising biomarker.
Limitations: This study was conducted at a single institution with a relatively small sample size
Funding for this study: The National Key R&D Program of China (2023YFC2414200), the National Natural Science Foundation of China (82371934), the Joint Fund of Henan Province Science and Technology R&D Program (225200810062).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The current study received approval from the local ethics committee, and all participants provided written informed consent for participation.

Author Block: C. Mingels¹, M. H. Madani¹, F. Sen¹, H. Nalbant¹, Y. G. Abdelhafez¹, M. Guindani², R. Badawi¹, B. A. Spencer¹, L. Nardo¹; ¹Sacramento/US, ²Los Angeles, CA/US
Purpose: To characterize diagnostic accuracy for nodal (N)-staging with [18F]FDG Total-Body (TB) and short-axial field-of-view (SAFOV) PET/CT in non-small cell lung cancer (NSCLC) patients.
Methods or Background: In this prospective, randomized, single center head-to-head comparative study 48 patients underwent TB and SAFOV PET/CT. 700 nodal levels (1R/L, 2R/L, 3a/p, 4R/L, 5, 6, 7, 8R/L, 9R/L, 10-14R/L) of 28 patients could be associated to histopathological findings, imaging after localized or systemic treatment, which allowed calculation of sensitivity, specificity, positive (PPV) and negative predictive value (NPV). Thresholds for maximum standardized uptake value (SUVmax), tumor-to-background ratio (TBR), metabolic tumor volume (MTV) and total-lesion glycolysis (TLG) were calculated.
Results or Findings: TB and SAFOV PET/CT showed high diagnostic accuracy indices for patient-based N-staging. Sensitivity and specificity were 86.0% (CI: 77.0-95.0%) and 98.3% (CI: 97.3-99.3%) for TB; 77.2% (CI: 66.3-88.1%) and 97.4% (CI: 96.1-98.6%) for SAFOV PET. PPV was higher for TB (81.7%, CI: 71.9-91.5%) compared to SAFOV PET (72.1%, CI: 60.9-83.4%), however, this finding was not statistically significant (p=0.08). NPV for TB (98.6%, CI: 97.9-99.6%) and SAFOV PET/CT (98.0%, CI: 96.9-99.1%) were comparable (p=0.22). Overall, NSCLC N-staging was affected in six cases on SAFOV and only in one case on TB PET/CT. Semi-quantitative analysis revealed a SUVmax-threshold of 3.0 to detect TP lesions on both scanners. TBR, MTV and TLG thresholds were lower on TB compared to SAFOV PET (TBR: 1.2 vs. 1.7, MTV: 0.5 ml vs. 1.0 ml and TLG: 1.0 ml vs. 3.0 ml).
Conclusion: TB and SAFOV PET/CT showed high diagnostic accuracy for N-staging in NSCLC. Sensitivity and PPV on TB PET/CT were slightly higher compared to SAFOV PET/CT. TB PET/CT showed lower rate of incorrect N-staging and lower semi-quantitative thresholds.
Limitations: Small sample size, composite reference standard with imaging
Funding for this study: Research reported in this publication was supported by the National Institutes of Health under award number R01CA249422. The work was also supported by the In Vivo Translational Imaging Shared Resources with funds from NCI P30CA093373 and by the Fred and Julia Rusch Foundation for Nuclear Medicine Research and Education. Hande Nalbant’s funding is partially provided by United Imaging Health’s UIH Fellowship Gift.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the UC Davis institutional review board (IRB 1506448). Written informed consent for inclusion was obtained. The study was performed in accordance with the Declaration of Helsinki.

Author Block: K. Zhu, J. Zhou; Lanzhou/CN
Purpose: To evaluate and compare the performances of whole-lesion iodine map histogram analysis to those of single-slice spectral CT parameters in discriminating of visceral pleural invasion in NSCLC.
Methods or Background: A total of 99 NSCLC patients underwent preoperative spectral CT and were divided into two groups: VPI and non-VPI.There were 65 men and 34 women with a mean age of 59.33 ± 8.62 (standard deviation) years(range:37-79 years) .The whole-lesion iodine map histogram parameterswere measured for each NSCLC patient. By placing regions of interest at representative levels of the tumor and normalizing them, spectral CT parameters IC and NIC were obtained. Discriminating capabilities of spectral CT and histogram parameters were assessed and compared using area under the ROC curve (AUC)and logistic regression models.
Results or Findings: The SD, Variance and CV of the iodine map histogram analysis,and iodine concentration and normalized iodine concentration of single-slice spectral CT parameters were significantly different of visceral pleural invasion in NSCLC (P < 0.001 to P = 0.03). The CV of histogram parameters (AUC=0.65; 95% [CI]: 0.54-0.76) and normalized iodine concentration (AUC=0.75; 95% CI: 0.64-0.85) from spectral CT parameters had the best performance for distinguishing whether visceral pleural invasion occurred in NSCLC. At ROC curve analysis no significant differences in AUC were found between histogram parameters (AUC = 0.84; 95% CI: 0.76-0.91) and spectral CT parameters (AUC = 0.66; 95% CI: 0.55-0.77) (P = 0.24).
Conclusion: Both whole-lesion iodine map histogram analysis and single-slice spectral CT parameters help discriminate whether of visceral pleural invasion in NSCLC, and the single-slice spectral CT parameters performed better in terms of diagnostic efficacy.
Limitations: This finding has not been validated in an independent population, limiting their generalizability. Future prospective studieswith larger and external patient cohorts are warranted.
Funding for this study: This research has been supported by the National Natural Science Foundation of China (82371914).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the Institutional Ethics Committee (2021A-498) and exempted from patient informed consent.

Author Block: Y. Ozawa, H. Nagata, T. Ueda, M. Nomura, T. Yoshikawa, D. Takenaka, Y. Ohno; Toyoake/JP
Purpose: To compare the prediction capability for postoperative recurrence among FDG-PET/CT and MRI-based molecular information from chemical exchange saturation transfer (CEST) imaging and diffusion-weighted imaging (DWI) in stage I non-small cell lung cancer (NSCLC) patients.
Methods or Background: 79 pathologically diagnosed and surgically treated NSCLC patients who underwent CEST imaging, DWI and FDG-PET/CT, follow-up and pathological examinations were included in this study. According to the follow-up and pathological examination results, all patients were divided as recurrence (n=13) and non-recurrence (n=66) groups. In each lesion, magnetization transfer ratio asymmetry at 3.5ppm (MTRasym), apparent diffusion coefficient (ADC) and SUVmax of each nodule were assessed by ROI measurements. To compare all indexes between two groups, Student’s t-test was performed. To determine the significant predictors, multiple logistic regression analysis was performed. Then, ROC analysis was performed to compare distinguishing two groups among all indexes and combined significant predictors. Finally, sensitivity (SE), specificity (SP) and accuracy (AC) were compared among all methods by McNemar’s test.
Results or Findings: There was significant difference of each index between two groups (p<0.05). Multiple logistic regression analyses determined MTRasym (Odds ratio [OR]: 1.31, p=0.03) and ADC (OR: 0.002, p=0.008) as significant predictors. When applied each threshold value, SPs and ACs of MTRasym (SP: 81.8%, AC: 82.2%), ADC (SP: 87.9%, AC: 86.1%) and combined predictors (SP: 89.4%, AC: 89.9%) were significantly higher than those of SUVmax (SP: 69.7%, p<0.05; AC: 72. %, p<0.05). Moreover, AC of combined predictors was significantly higher than that of MTRasym (p=0.03).
Conclusion: MRI-based molecular information has better prediction capability for postoperative recurrence than FDG-PET/CT in stage I NSCLC patients.
Limitations: Limited study cohort number and follow-up periods in some patients are considered as limitations in his study.
Funding for this study: Canon Medical Systems Corporation
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Fujita Health University Hospital

Author Block: T. Hino, T. Akamine, T. Hida, K. Sagiyama, Y. Yamasaki, K. Tabata, K. Ishigami; Fukuoka/JP
Purpose: Interstitial lung abnormalities (ILA) are known to be associated with increased mortality; however, the impact of ILA on post-operative prognosis of non-small cell lung cancer (NSCLC) remains unclear. The aim of study is to assess the association between mortality in post-operative clinical stage Ⅰ NSCLC patients and the presence of ILA.
Methods or Background: We retrospectively evaluated the patients who underwent chest CT, followed by radical resection for clinical stage Ⅰ NSCLC from 2006 to 2018. The presence of ILA was evaluated on high-resolution CT images by two radiologists and one thoracic surgeon. Five-year overall survival (OS), recurrence-free survival (RFS), and cumulative incidence of other cause of death stratified by the presence of ILA were assessed.
Results or Findings: Among 709 patients included in this study, 80 patients had ILA (11.2%). Five-year OS and RFS were significantly lower in patients with ILA than those without ILA (43.6% vs. 90.1 %, log-rank test p<0.001; 44.7% vs. 80.6 %, log-rank test p<0.001). Multivariable analysis demonstrated that the presence of ILA was an independent poor prognostic factor in both OS and RFS (HR: 1.45, 95% CI: 1.14–1.84, p=0.003; HR: 1.37, 95% CI: 1.08–1.74, p=0.010, respectively). Five-year cumulative incidence of other cause of death was significantly higher in patients with ILA than those with non-ILA (21.1% vs. 5.4%, Gray’s test p<0.001).
Conclusion: The presence of ILA affected the cause of other death and was an independent poor prognostic factor in clinical stage Ⅰ NSCLC.
Limitations: All the participants were retrospectively collected. ILA was not conformed with pathological specimens.
Funding for this study: N/A
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Kyushu University Hospital
IRB No. 23240-00

Author Block: G. Lin, W. Chen, J. Ji; Lishui/CN
Purpose: To explore the value of intratumoral and peritumoral radiomics in preoperative prediction of anaplastic lymphoma kinase (ALK) mutation status and survival in patients with lung adenocarcinoma.
Methods or Background: We retrospectively collected data from 505 eligible patients with lung adenocarcinoma from four hospitals (training and external validation sets 1–3). The CT-based radiomics features were extracted separately from the gross tumor volume (GTV) and GTV incorporating peritumoral 3-, 6-, 9-, 12-, and 15-mm regions (GPTV3, GPTV6, GPTV9, GPTV12, and GPTV15), and screened the most relevant features to construct radiomics models to predict ALK (+). The combined model incorporated radiomics scores (Rad-scores) of the best radiomics model and clinical predictors was constructed. Performance was evaluated using receiver operating characteristic (ROC) analysis. Survival outcomes were examined using the Cox proportional hazards model.
Results or Findings: The GPTV3 radiomics model using a support vector machine (SVM) algorithm achieved the best predictive performance, with the highest average AUC of 0.811 in the validation sets. Clinical TNM stage and pleural indentation were independent predictors. The combined model incorporating the GPTV3-Rad-score and clinical predictors achieved higher performance than the clinical model alone in predicting ALK (+) in three validation sets (AUC: 0.855 vs. 0.648, 0.882 vs. 0.634, 0.810 vs. 0.663). The prediction score of the combined model could stratify survival outcome in patients receiving ALK-TKI therapy (P=0.026) and immunotherapy (P=0.012).
Conclusion: The presented combined model based on GPTV3 effectively mined tumor features to predict ALK mutation status and stratify survival in patients with lung adenocarcinoma.
Limitations: This is a retrospective study and may have varying degrees of selection bias.
Funding for this study: This research was funded by the National Natural Science Foundation of China (Grant No.82072026 to Jiansong Ji), Key Project of Joint Construction by Provincial and Ministerial Authorities (Grant No.WKJ-ZJ-2452 to Minjiang Chen), Medical and Health General Project of Zhejiang Province (Grant No. 2024KY568 to Weiyue Chen, and No. 2023KY425 to Guihan Lin).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This study was approved by the Institutional Review Boards, and the requirement for informed consent was waived due to retrospective nature.