Recurrence prediction of lung adenocarcinoma using an immune gene expression and clinical data trained and validated support vector machine classifier.

Background: Immune microenvironment plays a critical role in cancer from onset to relapse. Machine learning (ML) algorithm can facilitate the analysis of lab and clinical data to predict lung cancer recurrence. Prompt detection and intervention are crucial for long-term survival in lung cancer relapse. Our study aimed to evaluate the clinical and genomic prognosticators for lung cancer recurrence by comparing the predictive accuracy of four ML models. Methods: A total of 41 early-stage lung cancer patients who underwent surgery between June 2007 and October 2014 at New York University Langone Medical Center were included (with recurrence, n=16; without recurrence, n=25). All patients had tumor tissue and buffy coat collected at the time of resection. The CIBERSORT algorithm quantified tumor-infiltrating immune cells (TIICs). Protein-protein interaction (PPI) network and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were conducted to unearth potential molecular drivers of tumor progression. The data was split into training (75%) and validation sets (25%). Ensemble linear kernel support vector machine (SVM) ML models were developed using optimized clinical and genomic features to predict tumor recurrence. Results: Activated natural killer (NK) cells, M0 macrophages, and M1 macrophages showed a positive correlation with progression. Conversely, T CD4+ memory resting cells were negatively correlated. In the PPI network, TNF and IL6 emerged as prominent hub genes. Prediction models integrating clinicopathological prognostic factors, tumor gene expression (45 genes), and buffy coat gene expression (47 genes) yielded varying receiver operating characteristic (ROC)-area under the curves (AUCs): 62.7%, 65.4%, and 59.7% in the training set, 58.3%, 83.3%, and 75.0% in the validation set, respectively. Notably, merging gene expression with clinical data in a linear SVM model led to a significant accuracy boost, with an AUC of 92.0% in training and 91.7% in validation. Conclusions: Using ML algorithm, immune gene expression data from tumor tissue and buffy coat may enhance the precision of lung cancer recurrence prediction.

Prognosis of lung cancer associated with cystic airspaces: A propensity score matching analysis.

OBJECTIVE: The association between the morphological characteristics and survival outcome of lung cancer associated with cystic airspaces (LCCAs) is unclear due to rarity of this disease. The current study attempted to compare the survival outcome between LCCAs and non-LCCAs and investigate the correlation between imaging features and prognosis of LCCA. METHOD: Of 10,835 patients diagnosed with non-small cell lung carcinoma (NSCLC) between January 2015 and December 2016, 123 patients with LCCA were included. The non-LCCA group comprised 3136 patients with primary solitary adenocarcinoma or squamous cell lung cancer. Propensity score matching (PSM) was performed for age, sex, tumor size, tumor stage, and lymph node involvement in a 1:1 ratio between the LCCAs and non-LCCAs, and the correlation between radiological features and recurrence-free survival (RFS) was analyzed. RESULT: The computed tomography (CT) lesion size was found to be higher in all LCCA subtypes, particularly in Type III (a cystic airspace with a mural nodule) and Type IV (mixed) LCCAs (3.09 and 3.65 cm, respectively), than in non-LCCAs (2 cm) after PSM. Three-year RFS in the LCCA group was higher than in the non-LCCA group (Type I- IV LCCAs: 100%, 84%, 77% and 83%, respectively vs. non-LCCAs: 77%). However, statistically significant difference was only found in comparison between LCCA Type I (thin-walled) and non-LCCA groups (P = 0.026). Type III lung cancer exhibited the worst survival among all four LCCA subtypes. CONCLUSIONS: The CT lesion size and pathologic tumor size varied significantly across LCCAs. Type I LCCAs exhibited better survival than non-LCCAs, whereas Type III LCCAs exhibited the worst survival rate among the four LCCA subtypes.

Cribriform growth pattern in lung adenocarcinoma: More aggressive and poorer prognosis than acinar growth pattern.

The predictive value of prognosis based on the histopathological subtype is a critical criterion in the new classification of lung adenocarcinoma published in 2011 by the International Association for the Study of Lung Cancer (IASLC), the American Thoracic Society (ATS), and the European Respiratory Society (ERS) (IASLC/ATS/ERS). In this new classification, the differences of histopathology and prognosis are two considerable parameters to classify the subtypes of lung adenocarcinoma. Cribriform growth pattern is regarded as a variant of acinar growth pattern in lung adenocarcinoma, however, more and more studies pointed out that cribriform growth pattern is associated with more aggressive histopathological structures, higher proportion of recurrence rates, and shorter postoperative survival than acinar growth pattern. These features are similar to solid or micropapillary predominant adenocarcinoma. In this review, we summarized the clinicopathological features, prognosis, and genetic variations of cribriform growth pattern of lung adenocarcinoma, and provided a novel insight into the diagnosis and treatment of cribriform lung adenocarcinoma.

Lung cancers associated with cystic airspaces: CT features and pathologic correlation.

OBJECTIVE: Lung cancer associated with cystic airspaces (LCCA) is a rare entity. The diagnosis and treatment is often delayed due to lack of comprehension of this disease. We aimed to elucidate LCCA's clinicopathological characteristics and investigate imaging features correlated with pathological invasiveness. METHOD: The preoperative computed tomographic (CT) scans of 10,835 patients diagnosed with NSCLC between January 2015 and December 2016 were reviewed by two thoracic radiologists for association with a cystic airspace. A clinicopathological and radiological feature analysis was done. RESULT: A total number of 123 LCCA patients were identified and four morphologic patterns were recognized: I, thin-walled type (n = 23, 18.7%); II, thick-walled type (n = 34, 27.6%); III, a cystic airspace with a mural nodule (CWN) type (n = 43, 35.0%); and IV, mixed type (n = 23, 18.7%). A solid component in the cyst wall predicted histological invasiveness in all four types of LCCA. The proportion of moderately/poorly (M/P)-differentiated subtype in type III (85.0%) was higher than in other three patterns (which were 50.0%, 50.0%, and 69.6%, respectively). Multivariate analysis revealed that type III pattern (odds ratio [OR], 6.5; 95% confidence interval [CI], 1.1-36.4; P = 0.035), part-solid/solid component in wall (part-solid: OR, 27.2; 95% CI, 5.6-3131.6; P < 0.001; solid: OR 614.6; 95% CI, 36.4-10,368.6; P < 0.001), and irregular inner surface of cyst (OR 7.0; 95% CI 1.9-26.2; P = 0.004) were independent risk factors for the M/P-differentiated subtype. EGFR mutations were the predominant genetic alterations in each type of LCCAs, but no significant difference was found among them. CONCLUSIONS: In LCCA, morphological patterns and wall components were two important predictors for determining pathological invasiveness.

The Association Between Imaging Features of TSCT and the Expression of PD-L1 in Patients With Surgical Resection of Lung Adenocarcinoma.

OBJECTIVES: Programmed death-ligand 1 (PD-L1) expression might serve as a predictive biomarker for immune checkpoint inhibitors in lung cancer. However, the relationship between PD-L1 expression and imaging features of lung cancer has not been fully understood. PATIENTS AND METHODS: A total of 350 patients with pathologically confirmed adenocarcinoma who received surgical treatment and had preoperative thin section computed tomography (CT) examination were included. Quantitative CT features including the mean CT value and tumor mass were measured on multiplanar reconstructed images. PD-L1-positive tumor was defined as the tumor proportion score > 5%. RESULTS: Seventy-four of 350 (21.1%) specimens were detected as PD-L1-positive tumors. PD-L1 expression was adversely associated with epidermal growth factor receptor mutation status (P < .001) and was significantly associated with invasive adenocarcinomas rather than preinvasive lesions and minimally invasive adenocarcinomas (P < .001). Multivariate analysis identified absence of surrounding ground glass opacity (P = .022), shape (P = .008), pleural indentation (P = .007), tumor mean CT value (P = .004), and the ratio of consolidation mass to tumor mass (P = .003) as being significantly associated with the expression of PD-L1. To improve the diagnostic accuracy, a joint model that combined 5 imaging traits was conducted. The area under the curve of the joint model was 0.783, with a sensitivity of 81.1% and specificity of 64.1%, respectively. CONCLUSION: PD-L1 expression was associated with pathologic invasiveness of adenocarcinomas and CT features, which suggested the possibility of predicting PD-L1 expression status via imaging features.

[Shanghai Pulmonary Hospital Experts Consensus on the Management of Ground-Glass Nodules Suspected as Lung Adenocarcinoma (Version 1)].

Background and objective As computed tomography (CT) screening for lung cancer becomes more common in China, so too does detection of pulmonary ground-glass nodules (GGNs). Although anumber of national or international guidelines about pulmonary GGNs have been published,most of these guidelines are produced by respiratory, oncology or radiology physicians, who might not fully understand the progress of modern minimal invasive thoracic surgery, and these current guidelines may overlook or underestimate the value of thoracic surgery in the management of pulmonary GGNs. In addition, the management for pre-invasive adenocarcinoma is still controversial. Based onthe available literature and experience from Shanghai Pulmonary Hospital, we composed this consensus about diagnosis and treatment of pulmonary GGNs. For lesions which are considered as adenocarcinoma in situ, chest thin layer CT scan follow-up is recommended and resection can only be adopt in some specific cases and excision should not exceed single segment resection. For lesions which are considered as minimal invasive adenocarcinoma, limited pulmonary resection or lobectomy is recommended. For lesions which are considered as early stage invasive adenocarcinoma, pulmonary resection is recommend and optimal surgical methods depend on whether ground glass component exist, location, volume and number of the lesions and physical status of patients. Principle of management of multiple pulmonary nodules is that primary lesions should be handled with priority, with secondary lesions taking into account.
.