How do deep-learning models generalize across populations? Cross-ethnicity generalization of COPD detection.

OBJECTIVES: To evaluate the performance and potential biases of deep-learning models in detecting chronic obstructive pulmonary disease (COPD) on chest CT scans across different ethnic groups, specifically non-Hispanic White (NHW) and African American (AA) populations. MATERIALS AND METHODS: Inspiratory chest CT and clinical data from 7549 Genetic epidemiology of COPD individuals (mean age 62 years old, 56-69 interquartile range), including 5240 NHW and 2309 AA individuals, were retrospectively analyzed. Several factors influencing COPD binary classification performance on different ethnic populations were examined: (1) effects of training population: NHW-only, AA-only, balanced set (half NHW, half AA) and the entire set (NHW + AA all); (2) learning strategy: three supervised learning (SL) vs. three self-supervised learning (SSL) methods. Distribution shifts across ethnicity were further assessed for the top-performing methods. RESULTS: The learning strategy significantly influenced model performance, with SSL methods achieving higher performances compared to SL methods (p < 0.001), across all training configurations. Training on balanced datasets containing NHW and AA individuals resulted in improved model performance compared to population-specific datasets. Distribution shifts were found between ethnicities for the same health status, particularly when models were trained on nearest-neighbor contrastive SSL. Training on a balanced dataset resulted in fewer distribution shifts across ethnicity and health status, highlighting its efficacy in reducing biases. CONCLUSION: Our findings demonstrate that utilizing SSL methods and training on large and balanced datasets can enhance COPD detection model performance and reduce biases across diverse ethnic populations. These findings emphasize the importance of equitable AI-driven healthcare solutions for COPD diagnosis. CRITICAL RELEVANCE STATEMENT: Self-supervised learning coupled with balanced datasets significantly improves COPD detection model performance, addressing biases across diverse ethnic populations and emphasizing the crucial role of equitable AI-driven healthcare solutions. KEY POINTS: Self-supervised learning methods outperform supervised learning methods, showing higher AUC values (p < 0.001). Balanced datasets with non-Hispanic White and African American individuals improve model performance. Training on diverse datasets enhances COPD detection accuracy. Ethnically diverse datasets reduce bias in COPD detection models. SimCLR models mitigate biases in COPD detection across ethnicities.

Key summary of German national guideline for adult patients with nosocomial pneumonia- Update 2024 Funding number at the Federal Joint Committee (G-BA): 01VSF22007.

PURPOSE: This executive summary of a German national guideline aims to provide the most relevant evidence-based recommendations on the diagnosis and treatment of nosocomial pneumonia. METHODS: The guideline made use of a systematic assessment and decision process using evidence to decision framework (GRADE). Recommendations were consented by an interdisciplinary panel. Evidence analysis and interpretation was supported by the German innovation fund providing extensive literature searches and (meta-) analyses by an independent methodologist. For this executive summary, selected key recommendations are presented including the quality of evidence and rationale for the level of recommendation. RESULTS: The original guideline contains 26 recommendations for the diagnosis and treatment of adults with nosocomial pneumonia, thirteen of which are based on systematic review and/or meta-analysis, while the other 13 represent consensus expert opinion. For this key summary, we present 11 most relevant for everyday clinical practice key recommendations with evidence overview and rationale, of which two are expert consensus and 9 evidence-based (4 strong, 5 weak and 2 open recommendations). For the management of nosocomial pneumonia patients should be divided in those with and without risk factors for multidrug-resistant pathogens and/or Pseudomonas aeruginosa. Bacterial multiplex-polymerase chain reaction (PCR) should not be used routinely. Bronchoscopic diagnosis is not considered superior to´non-bronchoscopic sampling in terms of main outcomes. Only patients with septic shock and the presence of an additional risk factor for multidrug-resistant pathogens (MDRP) should receive empiric combination therapy. In clinically stabilized patients, antibiotic therapy should be de-escalated and focused. In critically ill patients, prolonged application of suitable beta-lactam antibiotics should be preferred. Therapy duration is suggested for 7-8 days. Procalcitonin (PCT) based algorithm might be used to shorten the duration of antibiotic treatment. Patients on the intensive care unit (ICU) are at risk for invasive pulmonary aspergillosis (IPA). Diagnostics for Aspergillus should be performed with an antigen test from bronchial lavage fluid. CONCLUSION: The current guideline focuses on German epidemiology and standards of care. It should be a guide for the current treatment and management of nosocomial pneumonia in Germany.

Efficacy and safety of simultaneous combined Zephyr and Spiration valve therapy in patients with advanced pulmonary emphysema.

BACKGROUND: Endoscopic lung volume reduction with valves is a minimally invasive treatment strategy for patients with severe pulmonary emphysema. Two valve systems are currently available: Zephyr and Spiration valves. As these can be implanted simultaneously in the same procedure, the question arose as to the effect on lung function, exercise capacity and subjective disease perception after combined valve treatment. METHODS: We conducted a retrospective analysis of 108 patients with combined, simultaneous treatment of Zephyr and Spiration valves. The decision on which and how many valves to implant was based on the individual patient anatomy. Effects on lung function, exercise capacity and atelectasis formation as well as complications were evaluated 90- and 180-days post-treatment (90d-FU and 180d-FU). RESULTS: At 90d-FU (n = 90), the mean change was 86.7 ± 183.7 mL for FEV1 and -645.3 ± 1276.5 mL for RV, with responder rates of 39.8 % and 46.5 %, respectively. Complete atelectasis occurred in 16.7 % and partial atelectasis in 25.5 % of patients. Six-minute walking distance increased by 27.00 m [-1.50 - 68.50m]. The rates of pneumothorax (10.2 %) 6 months after treatment were not higher than in randomized controlled trials (RCTs). Likely due to the inclusion of high-risk patients, there was a higher incidence of severe COPD exacerbation (21.3 %) and pneumonia (12.0 %) compared to RCTs. CONCLUSIONS: The combined implantation of Zephyr and Spiration valves resulted in significant clinical and functional improvements with an acceptable risk profile. Therefore, the ability to combine both valve types in severe emphysema could be a promising option in endoscopic lung volume reduction.

Delta-radiomics features of ADC maps as early predictors of treatment response in lung cancer.

OBJECTIVE: Investigate the feasibility of detecting early treatment-induced tumor tissue changes in patients with advanced lung adenocarcinoma using diffusion-weighted MRI-derived radiomics features. METHODS: This prospective observational study included 144 patients receiving either tyrosine kinase inhibitors (TKI, n = 64) or platinum-based chemotherapy (PBC, n = 80) for the treatment of pulmonary adenocarcinoma. Patients underwent diffusion-weighted MRI the day prior to therapy (baseline, all patients), as well as either + 1 (PBC) or + 7 and + 14 (TKI) days after treatment initiation. One hundred ninety-seven radiomics features were extracted from manually delineated tumor volumes. Feature changes over time were analyzed for correlation with treatment response (TR) according to CT-derived RECIST after 2 months and progression-free survival (PFS). RESULTS: Out of 14 selected delta-radiomics features, 6 showed significant correlations with PFS or TR. Most significant correlations were found after 14 days. Features quantifying ROI heterogeneity, such as short-run emphasis (p = 0.04(pfs)/0.005(tr)), gradient short-run emphasis (p = 0.06(pfs)/0.01(tr)), and zone percentage (p = 0.02(pfs)/0.01(tr)) increased in patients with overall better TR whereas patients with worse overall response showed an increase in features quantifying ROI homogeneity, such as normalized inverse difference (p = 0.01(pfs)/0.04(tr)). Clustering of these features allows stratification of patients into groups of longer and shorter survival. CONCLUSION: Two weeks after initiation of treatment, diffusion MRI of lung adenocarcinoma reveals quantifiable tissue-level insights that correlate well with future treatment (non-)response. Diffusion MRI-derived radiomics thus shows promise as an early, radiation-free decision-support to predict efficacy and potentially alter the treatment course early. CRITICAL RELEVANCE STATEMENT: Delta-Radiomics texture features derived from diffusion-weighted MRI of lung adenocarcinoma, acquired as early as 2 weeks after initiation of treatment, are significantly correlated with RECIST TR and PFS as obtained through later morphological imaging. KEY POINTS: Morphological imaging takes time to detect TR in lung cancer, diffusion-weighted MRI might identify response earlier. Several radiomics features are significantly correlated with TR and PFS. Radiomics of diffusion-weighted MRI may facilitate patient stratification and management.

Midregional Proatrial Natriuretic Peptide (MRproANP) is associated with vertebral fractures and low bone density in patients with chronic obstructive pulmonary disease (COPD).

BACKGROUND: Patients with COPD are often affected by loss of bone mineral density (BMD) and osteoporotic fractures. Natriuretic peptides (NP) are known as cardiac markers, but have also been linked to fragility-associated fractures in the elderly. As their functions include regulation of fluid and mineral balance, they also might affect bone metabolism, particularly in systemic disorders such as COPD. RESEARCH QUESTION: We investigated the association between NP serum levels, vertebral fractures and BMD assessed by chest computed tomography (CT) in patients with COPD. METHODS: Participants of the COSYCONET cohort with CT scans were included. Mean vertebral bone density on CT (BMD-CT) as a risk factor for osteoporosis was assessed at the level of TH12 (AI-Rad Companion), and vertebral compression fractures were visually quantified by two readers. Their relationship with N-terminal pro-B-type natriuretic peptide (NT-proBNP), Mid-regional pro-atrial natriuretic peptide (MRproANP) and Midregional pro-adrenomedullin (MRproADM) was determined using group comparisons and multivariable analyses. RESULTS: Among 418 participants (58% male, median age 64 years, FEV1 59.6% predicted), vertebral fractures in TH12 were found in 76 patients (18.1%). Compared to patients without fractures, these had elevated serum levels (p ≤ 0.005) of MRproANP and MRproADM. Using optimal cut-off values in multiple logistic regression analyses, MRproANP levels ≥ 65 nmol/l (OR 2.34; p = 0.011) and age (p = 0.009) were the only significant predictors of fractures after adjustment for sex, BMI, smoking status, FEV1% predicted, SGRQ Activity score, daily physical activity, oral corticosteroids, the diagnosis of cardiac disease, and renal impairment. Correspondingly, MRproANP (p < 0.001), age (p = 0.055), SGRQ Activity score (p = 0.061) and active smoking (p = 0.025) were associated with TH12 vertebral density. INTERPRETATION: MRproANP was a marker for osteoporotic vertebral fractures in our COPD patients from the COSYCONET cohort. Its association with reduced vertebral BMD on CT and its known modulating effects on fluid and ion balance are suggestive of direct effects on bone mineralization. TRIAL REGISTRATION: ClinicalTrials.gov NCT01245933, Date of registration: 18 November 2010.

[Differential diagnosis of cystic and nodular lung diseases]. / Differenzialdiagnose zystischer und nodulärer Lungenkrankheiten.

BACKGROUND: Cystic and nodular lung diseases encompass a broad spectrum of diseases with different etiologies and clinicoradiological presentations. Their differentiation is crucial for patient management but can be complex due to diseases with features of both categories and overlapping radiological patterns. OBJECTIVE: This study aims to describe the imaging features of cystic and nodular lung diseases in high-resolution computed tomography (CT) in detail-primarily based on their etiology-in order to allow a more accurate differential diagnosis of these diseases. MATERIALS AND METHODS: A narrative review based on current literature on the topic was conducted from a clinicoradiological perspective. RESULTS: This paper systematically categorizes the differential diagnosis of cystic and nodular lung disease and provides insights into their radiological patterns and etiologies. It highlights the role of CT in the diagnosis of these diseases and emphasizes the importance of multidisciplinary panels combining expertise from radiology, pulmonology, rheumatology, and pathology. CONCLUSION: Reliable differential diagnosis of cystic and nodular lung diseases, particularly based on their radiological features alone, remains difficult due to their overlapping and dynamic nature. Multidisciplinary boards should be the clinical standard for accurate work-up of these diseases, as they combine the medical history, symptoms, radiological findings, and, if necessary, histopathological examinations, thus providing a more robust framework for diagnosis and management.

Diagnostic Potential of Supplemental Static and Dynamic 68Ga-FAPI-46 PET for Primary 18F-FDG-Negative Pulmonary Lesions.

PET using 68Ga-labeled fibroblast activation protein (FAP) inhibitors (FAPIs) holds high potential for diagnostic imaging of various malignancies, including lung cancer (LC). However, 18F-FDG PET is still the clinical gold standard for LC imaging. Several subtypes of LC, especially lepidic LC, are frequently 18F-FDG PET-negative, which markedly hampers the assessment of single pulmonary lesions suggestive of LC. Here, we evaluated the diagnostic potential of static and dynamic 68Ga-FAPI-46 PET in the 18F-FDG-negative pulmonary lesions of 19 patients who underwent surgery or biopsy for histologic diagnosis after PET imaging. For target validation, FAP expression in lepidic LC was confirmed by FAP immunohistochemistry. Methods: Hematoxylin and eosin staining and FAP immunohistochemistry of 24 tissue sections of lepidic LC from the local tissue bank were performed and analyzed visually. Clinically, 19 patients underwent static and dynamic 68Ga-FAPI-46 PET in addition to 18F-FDG PET based on individual clinical indications. Static PET data of both examinations were analyzed by determining SUVmax, SUVmean, and tumor-to-background ratio (TBR) against the blood pool, as well as relative parameters (68Ga-FAPI-46 in relation to18F-FDG), of histologically confirmed LC and benign lesions. Time-activity curves and dynamic parameters (time to peak, slope, k 1, k 2, k 3, and k 4) were extracted from dynamic 68Ga-FAPI-46 PET data. The sensitivity and specificity of all parameters were analyzed by calculating receiver-operating-characteristic curves. Results: FAP immunohistochemistry confirmed the presence of strongly FAP-positive cancer-associated fibroblasts in lepidic LC. LC showed markedly elevated 68Ga-FAPI-46 uptake, higher TBRs, and higher 68Ga-FAPI-46-to-18F-FDG ratios for all parameters than did benign pulmonary lesions. Dynamic imaging analysis revealed differential time-activity curves for LC and benign pulmonary lesions: initially increasing time-activity curves with a decent slope were typical of LC, and steadily decreasing time-activity curve indicated benign pulmonary lesions, as was reflected by a significantly increased time to peak and significantly smaller absolute values of the slope for LC. Relative 68Ga-FAPI-46-to-18F-FDG ratios regarding SUVmax and TBR showed the highest sensitivity and specificity for the discrimination of LC from benign pulmonary lesions. Conclusion: 68Ga-FAPI-46 PET is a powerful new tool for the assessment of single 18F-FDG-negative pulmonary lesions and may optimize patient stratification in this clinical setting.

Capturing COPD heterogeneity: anomaly detection and parametric response mapping comparison for phenotyping on chest computed tomography.

Background: Chronic obstructive pulmonary disease (COPD) poses a substantial global health burden, demanding advanced diagnostic tools for early detection and accurate phenotyping. In this line, this study seeks to enhance COPD characterization on chest computed tomography (CT) by comparing the spatial and quantitative relationships between traditional parametric response mapping (PRM) and a novel self-supervised anomaly detection approach, and to unveil potential additional insights into the dynamic transitional stages of COPD. Methods: Non-contrast inspiratory and expiratory CT of 1,310 never-smoker and GOLD 0 individuals and COPD patients (GOLD 1-4) from the COPDGene dataset were retrospectively evaluated. A novel self-supervised anomaly detection approach was applied to quantify lung abnormalities associated with COPD, as regional deviations. These regional anomaly scores were qualitatively and quantitatively compared, per GOLD class, to PRM volumes (emphysema: PRMEmph, functional small-airway disease: PRMfSAD) and to a Principal Component Analysis (PCA) and Clustering, applied on the self-supervised latent space. Its relationships to pulmonary function tests (PFTs) were also evaluated. Results: Initial t-Distributed Stochastic Neighbor Embedding (t-SNE) visualization of the self-supervised latent space highlighted distinct spatial patterns, revealing clear separations between regions with and without emphysema and air trapping. Four stable clusters were identified among this latent space by the PCA and Cluster Analysis. As the GOLD stage increased, PRMEmph, PRMfSAD, anomaly score, and Cluster 3 volumes exhibited escalating trends, contrasting with a decline in Cluster 2. The patient-wise anomaly scores significantly differed across GOLD stages (p < 0.01), except for never-smokers and GOLD 0 patients. In contrast, PRMEmph, PRMfSAD, and cluster classes showed fewer significant differences. Pearson correlation coefficients revealed moderate anomaly score correlations to PFTs (0.41-0.68), except for the functional residual capacity and smoking duration. The anomaly score was correlated with PRMEmph (r = 0.66, p < 0.01) and PRMfSAD (r = 0.61, p < 0.01). Anomaly scores significantly improved fitting of PRM-adjusted multivariate models for predicting clinical parameters (p < 0.001). Bland-Altman plots revealed that volume agreement between PRM-derived volumes and clusters was not constant across the range of measurements. Conclusion: Our study highlights the synergistic utility of the anomaly detection approach and traditional PRM in capturing the nuanced heterogeneity of COPD. The observed disparities in spatial patterns, cluster dynamics, and correlations with PFTs underscore the distinct - yet complementary - strengths of these methods. Integrating anomaly detection and PRM offers a promising avenue for understanding of COPD pathophysiology, potentially informing more tailored diagnostic and intervention approaches to improve patient outcomes.

Respiratory involvement in connective tissue diseases.

Pulmonary involvement is doubtless one the most fatal organ manifestations of the autoimmune rheumatic diseases (ARD) and involves the parenchyma, the vessels, the respiratory system itself, but also the muscles and the pleura. Close and regular screening assessments, identification of risk factors, clinical signs associated with the existence of lung disease should alarm the involved physicians treating these patients. The accurate classification is essential, as different treatment options are nowadays available. Pulmonary manifestations of ARD will be analyzed in this review article with special emphasis on interstitial lung disease and pulmonary hypertension.

CYFRA 21-1, CA 125 and CEA provide additional prognostic value in NSCLC patients with stable disease at first CT scan.

BACKGROUND: Serum tumor markers (STM) may complement imaging and provide additional clinical information for patients with non-small cell lung cancer (NSCLC). OBJECTIVE: To determine whether STMs can predict outcomes in patients with stable disease (SD) after initial treatment. METHODS: This single-center, prospective, observational trial enrolled 395 patients with stage III/IV treatment-naïve NSCLC; of which 263 patients were included in this analysis. Computed Tomography (CT) scans were performed and STMs measured before and after initial treatment (two cycles of chemotherapy and/or an immune checkpoint inhibitor or tyrosine kinase inhibitor); analyses were based on CT and STM measurements obtained at first CT performed after cycle 2 only PFS and OS were analyzed by Kaplan-Meier curves and Cox-proportional hazard models. RESULTS: When patients with SD (n = 100) were split into high- and low-risk groups based on CYFRA 21-1, CEA and CA 125 measurements using an optimized cut-off, a 4-fold increase risk of progression or death was estimated for high- vs low-risk SD patients (PFS, HR 4.17; OS, 3.99; both p < 0.0001). Outcomes were similar between patients with high-risk SD or progressive disease (n = 35) (OS, HR 1.17) and between patients with low-risk SD or partial response (n = 128) (PFS, HR 0.98; OS, 1.14). CONCLUSIONS: STMs can provide further guidance in patients with indeterminate CT responses by separating them into high- and low-risk groups for future PFS and OS events.

Prognostic value of tumor markers ProGRP, NSE and CYFRA 21-1 in patients with small cell lung cancer and chemotherapy-induced remission.

BACKGROUND: Despite successful response to first line therapy, patients with small-cell lung cancer (SCLC) often suffer from early relapses and disease progression. OBJECTIVE: To investigate the relevance of serum tumor markers for estimation of prognosis at several time points during the course of disease. METHODS: In a prospective, single-center study, serial assessments of progastrin-releasing peptide (ProGRP), neuron-specific enolase (NSE), cytokeratin-19 fragments (CYFRA 21-1) and carcino-embryogenic antigen (CEA) were performed during and after chemotherapy in 232 SCLC patients, and correlated with therapy response and overall survival (OS). RESULTS: ProGRP, NSE and CYFRA 21-1 levels decreased quickly after the first chemotherapy cycle and correlated well with the radiological response. Either as single markers or in combination they provided valuable prognostic information regarding OS at all timepoints investigated: prior to first-line therapy, after two treatment cycles in patients with successful response to first-line therapy, and prior to the start of second-line therapy. Furthermore, they were useful for continuous monitoring during and after therapy and often indicated progressive disease several months ahead of radiological changes. CONCLUSIONS: The results indicate the great potential of ProGRP, NSE and CYFRA 21-1 for estimating prognosis and monitoring of SCLC patients throughout the course of the disease.

Prediction of disease severity in COPD: a deep learning approach for anomaly-based quantitative assessment of chest CT.

OBJECTIVES: To quantify regional manifestations related to COPD as anomalies from a modeled distribution of normal-appearing lung on chest CT using a deep learning (DL) approach, and to assess its potential to predict disease severity. MATERIALS AND METHODS: Paired inspiratory/expiratory CT and clinical data from COPDGene and COSYCONET cohort studies were included. COPDGene data served as training/validation/test data sets (N = 3144/786/1310) and COSYCONET as external test set (N = 446). To differentiate low-risk (healthy/minimal disease, [GOLD 0]) from COPD patients (GOLD 1-4), the self-supervised DL model learned semantic information from 50 × 50 × 50 voxel samples from segmented intact lungs. An anomaly detection approach was trained to quantify lung abnormalities related to COPD, as regional deviations. Four supervised DL models were run for comparison. The clinical and radiological predictive power of the proposed anomaly score was assessed using linear mixed effects models (LMM). RESULTS: The proposed approach achieved an area under the curve of 84.3 ± 0.3 (p < 0.001) for COPDGene and 76.3 ± 0.6 (p < 0.001) for COSYCONET, outperforming supervised models even when including only inspiratory CT. Anomaly scores significantly improved fitting of LMM for predicting lung function, health status, and quantitative CT features (emphysema/air trapping; p < 0.001). Higher anomaly scores were significantly associated with exacerbations for both cohorts (p < 0.001) and greater dyspnea scores for COPDGene (p < 0.001). CONCLUSION: Quantifying heterogeneous COPD manifestations as anomaly offers advantages over supervised methods and was found to be predictive for lung function impairment and morphology deterioration. CLINICAL RELEVANCE STATEMENT: Using deep learning, lung manifestations of COPD can be identified as deviations from normal-appearing chest CT and attributed an anomaly score which is consistent with decreased pulmonary function, emphysema, and air trapping. KEY POINTS: • A self-supervised DL anomaly detection method discriminated low-risk individuals and COPD subjects, outperforming classic DL methods on two datasets (COPDGene AUC = 84.3%, COSYCONET AUC = 76.3%). • Our contrastive task exhibits robust performance even without the inclusion of expiratory images, while voxel-based methods demonstrate significant performance enhancement when incorporating expiratory images, in the COPDGene dataset. • Anomaly scores improved the fitting of linear mixed effects models in predicting clinical parameters and imaging alterations (p < 0.001) and were directly associated with clinical outcomes (p < 0.001).

Multicenter Standardization of Phase-Resolved Functional Lung MRI in Patients With Suspected Chronic Thromboembolic Pulmonary Hypertension.

BACKGROUND: Detection of pulmonary perfusion defects is the recommended approach for diagnosing chronic thromboembolic pulmonary hypertension (CTEPH). This is currently achieved in a clinical setting using scintigraphy. Phase-resolved functional lung (PREFUL) magnetic resonance imaging (MRI) is an alternative technique for evaluating regional ventilation and perfusion without the use of ionizing radiation or contrast media. PURPOSE: To assess the feasibility and image quality of PREFUL-MRI in a multicenter setting in suspected CTEPH. STUDY TYPE: This is a prospective cohort sub-study. POPULATION: Forty-five patients (64 ± 16 years old) with suspected CTEPH from nine study centers. FIELD STRENGTH/SEQUENCE: 1.5 T and 3 T/2D spoiled gradient echo/bSSFP/T2 HASTE/3D MR angiography (TWIST). ASSESSMENT: Lung signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were compared between study centers with different MRI machines. The contrast between normally and poorly perfused lung areas was examined on PREFUL images. The perfusion defect percentage calculated using PREFUL-MRI (QDPPREFUL ) was compared to QDP from the established dynamic contrast-enhanced MRI technique (QDPDCE ). Furthermore, QDPPREFUL was compared between a patient subgroup with confirmed CTEPH or chronic thromboembolic disease (CTED) to other clinical subgroups. STATISTICAL TESTS: t-Test, one-way analysis of variance (ANOVA), Pearson's correlation. Significance level was 5%. RESULTS: Significant differences in lung SNR and CNR were present between study centers. However, PREFUL perfusion images showed a significant contrast between normally and poorly perfused lung areas (mean delta of normalized perfusion -4.2% SD 3.3) with no differences between study sites (ANOVA: P = 0.065). QDPPREFUL was significantly correlated with QDPDCE (r = 0.66), and was significantly higher in 18 patients with confirmed CTEPH or CTED (57.9 ± 12.2%) compared to subgroups with other causes of PH or with excluded PH (in total 27 patients with mean ± SD QDPPREFUL = 33.9 ± 17.2%). DATA CONCLUSION: PREFUL-MRI could be considered as a non-invasive method for imaging regional lung perfusion in multicenter studies. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 1.

Potential of ChatGPT and GPT-4 for Data Mining of Free-Text CT Reports on Lung Cancer.

Background The latest large language models (LLMs) solve unseen problems via user-defined text prompts without the need for retraining, offering potentially more efficient information extraction from free-text medical records than manual annotation. Purpose To compare the performance of the LLMs ChatGPT and GPT-4 in data mining and labeling oncologic phenotypes from free-text CT reports on lung cancer by using user-defined prompts. Materials and Methods This retrospective study included patients who underwent lung cancer follow-up CT between September 2021 and March 2023. A subset of 25 reports was reserved for prompt engineering to instruct the LLMs in extracting lesion diameters, labeling metastatic disease, and assessing oncologic progression. This output was fed into a rule-based natural language processing pipeline to match ground truth annotations from four radiologists and derive performance metrics. The oncologic reasoning of LLMs was rated on a five-point Likert scale for factual correctness and accuracy. The occurrence of confabulations was recorded. Statistical analyses included Wilcoxon signed rank and McNemar tests. Results On 424 CT reports from 424 patients (mean age, 65 years ± 11 [SD]; 265 male), GPT-4 outperformed ChatGPT in extracting lesion parameters (98.6% vs 84.0%, P < .001), resulting in 96% correctly mined reports (vs 67% for ChatGPT, P < .001). GPT-4 achieved higher accuracy in identification of metastatic disease (98.1% [95% CI: 97.7, 98.5] vs 90.3% [95% CI: 89.4, 91.0]) and higher performance in generating correct labels for oncologic progression (F1 score, 0.96 [95% CI: 0.94, 0.98] vs 0.91 [95% CI: 0.89, 0.94]) (both P < .001). In oncologic reasoning, GPT-4 had higher Likert scale scores for factual correctness (4.3 vs 3.9) and accuracy (4.4 vs 3.3), with a lower rate of confabulation (1.7% vs 13.7%) than ChatGPT (all P < .001). Conclusion When using user-defined prompts, GPT-4 outperformed ChatGPT in extracting oncologic phenotypes from free-text CT reports on lung cancer and demonstrated better oncologic reasoning with fewer confabulations. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Hafezi-Nejad and Trivedi in this issue.

Analysis of rare fusions in NSCLC: Genomic architecture and clinical implications.

OBJECTIVES: Molecular diagnosis for targeted therapies has been improved significantly in non-small-cell lung cancer (NSCLC) patients in recent years. Here we report on the prevalence of rare fusions in NSCLC and dissect their genomic architecture and potential clinical implications. MATERIALS AND METHODS: Overall, n = 5554 NSCLC patients underwent next-generation sequencing (NGS) for combined detection of oncogenic mutations and fusions either at primary diagnosis (n = 5246) or after therapy resistance (n = 308). Panels of different sizes were employed with closed amplicon-based, or open assays, i.e. anchored multiplex PCR (AMP) and hybrid capture-based, for detection of translocations, including "rare" fusions, defined as those beyond ALK, ROS1, RET and <0.5 % frequency in NSCLC. RESULTS: Rare fusions involving EGFR, MET, HER2, BRAF and other potentially actionable oncogenes were detected in 0.5% (n = 26) of therapy-naive and 2% (n = 6) TKI-treated tumors. Detection was increased using open assays and/or larger panels, especially those covering >25 genes, by approximately 1-2% (p = 0.001 for both). Patient characteristics (age, gender, smoking, TP53 co-mutations (56%), or mean tumor mutational burden (TMB) (4.8 mut/Mb)) showed no association with presence of rare fusions. Non-functional alterations, i.e. out-of-frame or lacking kinase domains, comprised one-third of detected rare fusions and were significantly associated with simultaneous presence of classical oncogenic drivers, e.g. EGFR or KRAS mutations (p < 0.001), or use of larger panels (frequency of non-functional among the detected rare fusions 57% for 25+ gene- vs. 12% for smaller panels, p < 0.001). As many rare fusions were identified before availability of targeted therapy, mean survival for therapy-naïve patients was 23.8 months, comparable with wild-type tumors. CONCLUSION: Approximately 1-2% of advanced NSCLC harbor rare fusions, which are potentially actionable and may support diagnosis. Routine adoption of broad NGS assays capable to identify exact fusion points and potentially retained protein domains can increase the yield of therapeutically relevant molecular information in advanced NSCLC.

GOLD stage-specific phenotyping of emphysema and airway disease using quantitative computed tomography.

Background: In chronic obstructive pulmonary disease (COPD) abnormal lung function is related to emphysema and airway obstruction, but their relative contribution in each GOLD-stage is not fully understood. In this study, we used quantitative computed tomography (QCT) parameters for phenotyping of emphysema and airway abnormalities, and to investigate the relative contribution of QCT emphysema and airway parameters to airflow limitation specifically in each GOLD stage. Methods: Non-contrast computed tomography (CT) of 492 patients with COPD former GOLD 0 COPD and COPD stages GOLD 1-4 were evaluated using fully automated software for quantitative CT. Total lung volume (TLV), emphysema index (EI), mean lung density (MLD), and airway wall thickness (WT), total diameter (TD), lumen area (LA), and wall percentage (WP) were calculated for the entire lung, as well as for all lung lobes separately. Results from the 3rd-8th airway generation were aggregated (WT3-8, TD3-8, LA3-8, WP3-8). All subjects underwent whole-body plethysmography (FEV1%pred, VC, RV, TLC). Results: EI was higher with increasing GOLD stages with 1.0 ± 1.8% in GOLD 0, 4.5 ± 9.9% in GOLD 1, 19.4 ± 15.8% in GOLD 2, 32.7 ± 13.4% in GOLD 3 and 41.4 ± 10.0% in GOLD 4 subjects (p < 0.001). WP3-8 showed no essential differences between GOLD 0 and GOLD 1, tended to be higher in GOLD 2 with 52.4 ± 7.2%, and was lower in GOLD 4 with 50.6 ± 5.9% (p = 0.010 - p = 0.960). In the upper lobes WP3-8 showed no significant differences between the GOLD stages (p = 0.824), while in the lower lobes the lowest WP3-8 was found in GOLD 0/1 with 49.9 ± 6.5%, while higher values were detected in GOLD 2 with 51.9 ± 6.4% and in GOLD 3/4 with 51.0 ± 6.0% (p < 0.05). In a multilinear regression analysis, the dependent variable FEV1%pred can be predicted by a combination of both the independent variables EI (p < 0.001) and WP3-8 (p < 0.001). Conclusion: QCT parameters showed a significant increase of emphysema from GOLD 0-4 COPD. Airway changes showed a different spatial pattern with higher values of relative wall thickness in the lower lobes until GOLD 2 and subsequent lower values in GOLD3/4, whereas there were no significant differences in the upper lobes. Both, EI and WP5-8 are independently correlated with lung function decline.

Pembrolizumab Alone or With Chemotherapy for 70+ Year-Old Lung Cancer Patients: A Retrospective Study.

OBJECTIVE: First-line pembrolizumab alone, as approved for PD-L1 ≥50%, or with chemotherapy was analyzed in older non-small-cell lung cancer (NSCLC) patients, for whom evidence is scarce. MATERIALS AND METHODS: A total of 156 consecutive ≥70 year-old patients treated between January 2016 and May 2021 were retrospectively analyzed. Tumor progression was verified through radiologic review, while toxicity was captured from records. RESULTS: Pembrolizumab plus chemotherapy (n = 95) caused higher rates of adverse events (91% vs. 51%, P < .001), treatment discontinuation (37% vs. 21%, P = .034), and hospitalization (56% vs. 23%, P < .001), but similar rates of immune-related adverse events (irAEs, mean 35%, P = .998) compared to pembrolizumab monotherapy (n = 61). Progression-free (PFS) and overall survival (OS) were similar between the 2 groups (7 vs. 8 months, and 16 vs. 14 months in median, P > .25). Occurrence of irAEs was associated with longer survival in a 12-week landmark analysis (median PFS 11 vs. 5 months, hazard ratio [HR] 0.51, P = .001; median OS 33 vs. 10 months, HR 0.46, P < .001), but occurrence of other AEs not (both P > .35). A worse ECOG performance status (PS) ≥2, presence of brain metastases at diagnosis, squamous histology and lack of tumor PD-L1 expression were independent predictors of shorter PFS and OS in multivariable analysis (HR 1.6-3.9 for PFS and OS, all P < .05). CONCLUSION: Chemoimmunotherapy increases the rate of adverse events and hospitalization without prolonging PFS or OS in newly diagnosed NSCLC patients aged 70 years or older compared to pembrolizumab monotherapy. ECOG PS 2, presence of brain metastases at diagnosis, squamous histology and PD-L1 negativity are associated with poor outcome.

Spatial profiling of the microenvironment reveals low intratumoral heterogeneity and STK11-associated immune evasion in therapy-naïve lung adenocarcinomas.

OBJECTIVE: Intratumoral heterogeneity was found to be a significant factor causing resistance to lung cancer therapies, including immune checkpoint blockade. Lesser is known about spatial heterogeneity of the tumor microenvironment (TME) and its association with genetic properties of the tumor, which is of particular interest in the therapy-naïve setting. MATERIALS AND METHODS: We performed multi-region sampling (2-4 samples per tumor; total of 55 samples) from a cohort of 19 untreated stage IA-IIIB lung adenocarcinomas (n = 11 KRAS mutant, n = 1 ERBB2 mutant, n = 7 KRAS wildtype). For each sample the expression of 770 immunooncology-related genes was analyzed using the nCounter platform, while the mutational status was determined by hybrid capture-based next-generation sequencing (NGS) using a large panel covering more than 500 genes. RESULTS: Global unsupervised analyses revealed clustering of the samples into two groups corresponding to a 'hot' or 'cold' immunologic tumor contexture based on the abundance of immune cell infiltrates. All analyzed specific immune cell signatures (ICsig) showed a significantly higher intertumoral than intratumoral heterogeneity (p < 0.02), as most of the analyzed cases (14/19) showed a very homogenous spatial immune cell profile. PD-L1 exhibited a significantly higher intertumoral than intratumoral heterogeneity (p = 1.03e-13). We found a specific association with 'cold' TME for STK11 (11/14, p < 0.07), but not KRAS, TP53, LRP1B, MTOR, U2AF1 co-mutations, and validated this finding using The Cancer Genome Atlas (TCGA) data. CONCLUSION: Early-stage lung adenocarcinomas show considerable intertumoral, but limited intratumoral heterogeneity, which is clinically highly relevant as assessment before neoadjuvant treatment is based on small biopsies. STK11 mutations are specifically associated with a 'cold' TME, which could affect the efficacy of perioperative immunotherapy.