Establishment and verification of a prediction model based on clinical characteristics and positron emission tomography/computed tomography (PET/CT) parameters for distinguishing malignant from benign ground-glass nodules

Clinicopathological and computed tomography features associated with recurrence-free survival of patients with small-sized peripheral invasive lung adenocarcinoma after sublobectomy

Background

Sublobar resection is gradually becoming a standard treatment for small-sized (≤2 cm) peripheral non-small cell lung cancer (NSCLC), with lung adenocarcinoma (LADC) being the most frequent histologic subtype. However, the prognostic predictors for preoperatively determining whether sublobectomy is feasible for patients with early LADC have not yet been well identified. Therefore, this study aimed to investigate the clinicopathological and computed tomography (CT) features associated with the recurrence-free survival (RFS) of patients with small-sized invasive LADC (SILADC) after sublobar resection.

Methods

This retrospective cohort study analyzed 107 patients with SILADC who underwent preoperative chest CT scan and sublobar resection from December 2012 to March 2019. The Kaplan-Meier survival was used to analyze the relationship between clinicopathological characteristics, preoperative chest CT findings, and RFS. The Cox proportional hazards regression was used to identify independent prognostic factors of poor RFS.

Results

For clinicopathological characteristics, RFS was shorter in patients aged ≥70 years, smokers, and those with micropapillary/solid-predominant adenocarcinomas (all P values <0.05). For preoperative CT features, RFS was shorter in patients with tumor size ≥1.4 cm, solid component size ≥1.1 cm, proportion of solid component ≥72%, solid density, spiculation, vascular convergence sign, peripheral fibrosis, and type II pleural tag (all P values <0.05). Multivariate analysis showed proportion of solid component ≥72% [hazard ratio (HR): 5.920; P=0.006; 95% confidence interval (CI): 1.686-20.794], spiculation (HR: 5.026; P=0.001; 95% CI: 2.008-12.581), and type II pleural tag (HR: 4.638; P=0.002; 95% CI: 1.773-12.136) were independent risk factors for poor prognosis in patients with SILADC after sub-lobectomy.

Conclusions

Clinicopathological and CT characteristics are helpful for predicting the RFS of patients with SILADC after sublobar resection and can be used as an auxiliary tool for thoracic surgeons to choose the best surgical mode.

Artificial intelligence system-based histogram analysis of computed tomography features to predict tumor invasiveness of ground-glass nodules

Background

The use of an artificial intelligence (AI)-based diagnostic system can significantly aid in analyzing the histogram of pulmonary nodules. The aim of our study was to evaluate the value of computed tomography (CT) histogram indicators analyzed by AI in predicting the tumor invasiveness of ground-glass nodules (GGNs) and to determine the added value of contrast-enhanced CT (CECT) compared with nonenhanced CT (NECT) in this prediction.

Methods

This study enrolled patients with persistent GGNs who underwent preoperative NECT and CECT scanning. AI-based histogram analysis was performed for pathologically confirmed GGNs, which was followed by screening invasiveness-related factors via univariable analysis. Multivariable logistic models were developed based on candidate CT histogram indicators measured on either NECT or CECT. Receiver operating characteristic (ROC) curve and precision-recall (PR) curve were used to evaluate the models' performance.

Results

A total of 116 patients comprising 121 GGNs were included and divided into the precancerous lesion and adenocarcinoma groups based on invasiveness. In the AI-based histogram analysis, the mean CT value [NECT: odds ratio (OR) =1.009; 95% confidence interval (CI): 1.004-1.013; P<0.001] and solid component volume (NECT: OR =1.005; 95% CI: 1.000-1.010; P=0.032) were associated with the adenocarcinoma and used for multivariable logistic modeling. The area under ROC curve (AUC) and PR curve (AUPR) were not significantly different between the NECT model (AUC =0.765, 95% CI: 0.679-0.837; AUPR =0.907, 95% CI: 0.825-0.953) and the optimal CECT model (delayed phase: AUC =0.772, 95% CI: 0.687-0.843; AUPR =0.895, 95% CI: 0.812-0.944). No significantly different metrics were observed between the NECT and CECT models (precision: 0.707 vs. 0.742; P=0.616).

Conclusions

The AI diagnostic system can help in the diagnosis of GGNs. The system displayed decent performance in GGN detection and alert to malignancy. Mean CT value and solid component volume were independent predictors of tumor invasiveness. CECT provided no additional improvement in diagnostic performance as compared with NECT.

Metabolic phenotypes, serum tumor markers, and histopathological subtypes in predicting bone metastasis: analysis of 695 patients with lung cancer in China

Background

Patients with lung cancer who develop bone metastasis (BM) generally have an adverse prognosis. Although several clinical models have been used to predict BM in patients with lung cancer, the results are unsatisfactory. In this retrospective study, we investigated the role of ¹⁸F-2-fluoro-2-deoxyglucose (FDG) metabolic activity, serum tumor markers, and histopathological subtypes in predicting BM in patients with lung cancer.

Methods

This study included 695 consecutive patients with lung cancer who underwent ¹⁸F-FDG positron emission tomography/computed tomography (PET/CT) and in whom serum tumor markers were detected prior to treatment. The maximum standardized uptake value of primary tumors (pSUV_max), metastatic lymph nodes (nSUV_max) and distant metastases (mSUV_max), 8 serum tumor markers [carcinoembryonic antigen (CEA), neuron-specific enolase (NSE), squamous cell carcinoma-related antigen (SCCA), cytokeratin 19 fragment (CYFRA21-1), carbohydrate antigen (CA) 125, CA50, CA72-4, and ferritin], and histopathological subtypes were compared between patients with and without BM. Receiver operating characteristic (ROC) curve and multiple logistic regression analyses were performed to identify predictors of BM in patients with lung cancer.

Results

BM was identified in 133 (19.1%) patients and not in 562 (80.9%). Patients with BM had significantly higher pSUV_max, nSUV_max, and mSUV_max than did those without BM. High concentrations of 6 serum tumor markers (i.e., CEA, ferritin, NSE, CA50, CA125, and CYFRA21-1) were significantly associated with BM. There were significant differences in the proportion of histopathological subtypes between patients with and without BM (χ²=32.35; P<0.001). The area under ROC-derived curve based on metabolic parameters was 0.737 (95% CI: 0.644-0.829) and 0.884 (95% CI: 0.825-0.943) when combined with the 6 serum tumor markers and histopathological subtypes, respectively.

Conclusions

High pSUV_max, nSUV_max, and mSUV_max favor the presence of BM in patients with lung cancer, and serum tumor markers and histopathological subtypes are important factors for predicting BM in these patients.

Development, Validation, and Comparison of Image-Based, Clinical Feature-Based and Fusion Artificial Intelligence Diagnostic Models in Differentiating Benign and Malignant Pulmonary Ground-Glass Nodules

Objective

This study aimed to develop effective artificial intelligence (AI) diagnostic models based on CT images of pulmonary nodules only, on descriptional and quantitative clinical or image features, or on a combination of both to differentiate benign and malignant ground-glass nodules (GGNs) to assist in the determination of surgical intervention.

Methods

Our study included a total of 867 nodules (benign nodules: 112; malignant nodules: 755) with postoperative pathological diagnoses from two centers. For the diagnostic models to discriminate between benign and malignant GGNs, we adopted three different artificial intelligence (AI) approaches: a) an image-based deep learning approach to build a deep neural network (DNN); b) a clinical feature-based machine learning approach based on the clinical and image features of nodules; c) a fusion diagnostic model integrating the original images and the clinical and image features. The performance of the models was evaluated on an internal test dataset (the “Changzheng Dataset”) and an independent test dataset collected from an external institute (the “Longyan Dataset”). In addition, the performance of automatic diagnostic models was compared with that of manual evaluations by two radiologists on the ‘Longyan dataset’.

Results

The image-based deep learning model achieved an appealing diagnostic performance, yielding AUC values of 0.75 (95% confidence interval [CI]: 0.62, 0.89) and 0.76 (95% CI: 0.61, 0.90), respectively, on both the Changzheng and Longyan datasets. The clinical feature-based machine learning model performed well on the Changzheng dataset (AUC, 0.80 [95% CI: 0.64, 0.96]), whereas it performed poorly on the Longyan dataset (AUC, 0.62 [95% CI: 0.42, 0.83]). The fusion diagnostic model achieved the best performance on both the Changzheng dataset (AUC, 0.82 [95% CI: 0.71-0.93]) and the Longyan dataset (AUC, 0.83 [95% CI: 0.70-0.96]), and it achieved a better specificity (0.69) than the radiologists (0.33-0.44) on the Longyan dataset.

Conclusion

The deep learning models, including both the image-based deep learning model and the fusion model, have the ability to assist radiologists in differentiating between benign and malignant nodules for the precise management of patients with GGNs.

Risk analysis in peripheral clinical T1 non-small cell lung cancer correlations between tumor-to-blood standardized uptake ratio on 18F-FDG PET-CT and primary tumor pathological invasiveness: a real-world observational study

BACKGROUND

Sublobar resection is not suitable for patients with pathological invasiveness [including lymph node metastasis (LNM), visceral pleural invasion (VPI), and lymphovascular invasion (LVI)] of peripheral clinical T1 (cT1) non-small cell lung cancer (NSCLC), while primary tumor maximum standardized uptake value (SUV_max) on ¹⁸F-FDG PET-CT is related to pathological invasiveness, the significance differed among different institutions is still challenging. This study explored the relationship between the tumor-to-blood standardized uptake ratio (SUR) of ¹⁸F-FDG PET-CT and primary tumor pathological invasiveness in peripheral cT1 NSCLC patients.

METHODS

This retrospective study included 174 patients with suspected lung neoplasms who underwent preoperative ¹⁸F-FDG PET-CT. We compared the differences of the clinicopathological variables, metabolic and morphological parameters in the pathological invasiveness and less-invasiveness group. We performed a trend test for these parameters based on the tertiles of SUR. The relationship between SUR and pathological invasiveness was evaluated by univariate and multivariate logistics regression models (included unadjusted, simple adjusted, and fully adjusted models), odds ratios (ORs), and 95% confidence intervals (95% CIs) were calculated. A smooth fitting curve between SUR and pathological invasiveness was produced by the generalized additive model (GAM).

RESULTS

Thirty-eight point five percent of patients had pathological invasiveness and tended to have a higher SUR value than the less-invasiveness group [6.50 (4.82-11.16) vs. 4.12 (2.04-6.61), P<0.001]. The trend of SUV_max, mean standardized uptake value (SUV_mean), metabolic tumor volume (MTV), total lesion glycolysis (TLG), mean CT value (CT_mean), size of the primary tumor, neuron-specific enolase (NSE), the incidence of LNM, adenocarcinoma (AC), and poor differentiation in the tertiles of SUR value were statistically significant (P were <0.001, <0.001, 0.010, <0.001, <0.001, 0.002, 0.033, <0.001, 0.002, and <0.001, respectively). Univariate analysis showed that the risk of pathological invasiveness increased significantly with increasing SUR [OR: 1.13 (95% CI: 1.06-1.21), P<0.001], and multivariate analysis demonstrated SUR, as a continuous variable, was still significantly related to pathological invasiveness [OR: 1.09 (95% CI: 1.01-1.18), P=0.032] after adjusting for confounding covariates. GAM revealed that SUR tended to be linearly and positively associated with pathological invasiveness and E-value analysis suggested robustness to unmeasured confounding.

CONCLUSIONS

SUR is linearly and positively associated with primary tumor pathological invasiveness independent of confounding covariates in peripheral cT1 NSCLC patients and could be used as a supplementary risk maker to assess the risk of pathological invasiveness.