O. Jung Kwon

Background: the objective of this study is to evaluate the predictive power of the survival model using deep learning of diffusion-weighted images (DWI) in patients with non-small-cell lung cancer (NSCLC). Methods: DWI at b-values of 0, 100, and 700 sec/mm² (DWI₀, DWI₁₀₀, DWI₇₀₀) were preoperatively obtained for 100 NSCLC patients who underwent curative surgery (57 men, 43 women; mean age, 62 years). The ADC_0-100 (perfusion-sensitive ADC), ADC_100-700 (perfusion-insensitive ADC), ADC_0-100-700, and demographic features were collected as input data and 5-year survival was collected as output data. Our survival model adopted transfer learning from a pre-trained VGG-16 network, whereby the softmax layer was replaced with the binary classification layer for the prediction of 5-year survival. Three channels of input data were selected in combination out of DWIs and ADC images and their accuracies and AUCs were compared for the best performance during 10-fold cross validation. Results: 66 patients survived, and 34 patients died. The predictive performance was the best in the following combination: DWI₀-ADC_0-100-ADC_0-100-700 (accuracy: 92%; AUC: 0.904). This was followed by DWI₀-DWI₇₀₀-ADC_0-100-700, DWI₀-DWI₁₀₀-DWI₇₀₀, and DWI₀-DWI₀-DWI₀ (accuracy: 91%, 81%, 76%; AUC: 0.889, 0.763, 0.711, respectively). Survival prediction models trained with ADC performed significantly better than the one trained with DWI only (p-values < 0.05). The survival prediction was improved when demographic features were added to the model with only DWIs, but the benefit of clinical information was not prominent when added to the best performing model using both DWI and ADC. Conclusions: Deep learning may play a role in the survival prediction of lung cancer. The performance of learning can be enhanced by inputting precedented, proven functional parameters of the ADC instead of the original data of DWIs only.

Kim SH, et al. J Korean Med Sci. 2022 Aug;37(32):e250. https://doi.org/10.3346/jkms.2022.37.e250

Rationale: Although a history of pulmonary tuberculosis (PTB) is a risk factor for developing both chronic obstructive pulmonary disease (COPD) and lung cancer, it remains unclear whether a history of PTB affects lung cancer development in patients with COPD. Objectives: To investigate whether a history of PTB is associated with an increased risk of lung cancer development in a population with COPD. Methods: This cohort study included a nationwide representative sample of 13,165 Korean men and women with COPD, aged between 50 and 84 years. In addition, to assess whether the relationship between PTB and lung cancer risk differs between participants with and without COPD, a matched cohort without COPD was included. Participants were matched 1:3 for age, sex, smoking history, and PTB status based on the index health screening examination of corresponding participants with COPD. The two cohorts were followed up for 13 years (January 1, 2003, to December 31, 2015). PTB was diagnosed on the basis of the results of chest radiography, and incident lung cancer was identified from hospitalization and outpatient visit claims (International Classification of Diseases, Tenth Revision diagnosis code C33 or C34). Results: During 370,617 person-years (PY) of follow-up (median follow-up, 7.7 yr) in the COPD group, we observed 430 incident cases of lung cancer in participants without a history of PTB (incidence rate, 524 per 100,000 PY) and 148 cases in those with a history of PTB (incidence rate, 931 per 100,000 PY). Compared with participants without a PTB history, the fully adjusted subdistribution hazard ratio (95% confidence interval [CI]) for lung cancer in those with a history of PTB was 1.24 (1.03–1.50). The association of PTB history and lung cancer development was more evident in never-smokers with COPD. In contrast, among participants without COPD, the corresponding hazard ratio (95% CI) was 0.98 (0.78–1.22). There was no interaction among PTB, smoking status, and COPD. Conclusions: A history of PTB was associated with an increased risk of developing lung cancer among patients with COPD in our country with an intermediate tuberculosis burden. Patients with COPD with a history of PTB, particularly never-smokers, might benefit from periodic screening or assessment for lung cancer development.

Objective This study aimed to investigate the prognostic significance of dynamic contrast-enhanced computed tomography in patients with stage IA non–small cell lung cancer (NSCLC). Methods We retrospectively enrolled 139 patients (77 men, 62 women; mean age, 59 years) with stage IA NSCLC who underwent dynamic contrast-enhanced computed tomography. Data on age, pathologic subtype, peak enhancement, and net enhancement of primary lung cancer were collected and correlated with 5-year survival. Results Peak enhancement had a significant correlation with overall survival in the univariable analysis (hazard ratio [HR], 1.18, confidence interval [CI], 1.01–1.38; P = 0.04) and in the multivariable analysis (HR, 1.19; CI, 1.01–1.39; P = 0.04). Patients with peak enhancement of 90 Hounsfield unit or higher had a significantly increased risk of death compared with patients with less enhancement after curative surgery (HR, 4.15; CI, 1.23–13.95; P = 0.02). Conclusions Our study confirmed the prognostic significance of peak enhancement as an indicator for the overall survival of stage IA NSCLC.

We aim to examine how season and temperature levels affect physical activity using a wearable device among patients scheduled to undergo surgical resection of lung cancer. Physical activity (PA) data from the wearable device were analyzed by seasons for 555 preoperative lung cancer patients from the CATCH-LUNG cohort study. The seasons were divided into spring, summer, autumn, and winter using the study enrollment date before surgery. The overall mean (SD) age was 61.1 (8.9) years, and the mean (SD) daily steps at each season were 11,438 (5922), 11,147 (5065), 10,404 (4403), and 8548 (4293), respectively. In the fully-adjusted models, patients in the winter season had 27.04% fewer daily steps (95% CI = −36.68%, −15.93%) and 35.22% less time spent performing moderate to vigorous physical activity (MVPA) compared to patients in the spring. The proportion of participants with over 8000 steps and duration of MVPA were significantly lower in the winter than the spring. In particular, daily steps had a negative linear association with wind chill temperature in patients who lived in Seoul. In conclusion, PA was significantly lower in the winter and it was more robust in patients who had a low cardiorespiratory function.

Ko RE, et al. Yonsei Med J. 2020 Aug;61(8):670-678. https://doi.org/10.3349/ymj.2020.61.8.670

Rationale: Patients with refractory Mycobacterium avium complex lung disease (MAC-LD) undergo long-term macrolide therapy, but macrolide resistance develops infrequently. Objectives: The aim of this study was to determine whether reinfection was a factor in the low incidence of macrolide resistance in patients with refractory MAC-LD. Methods: Among 481 patients with treatment-naive MAC-LD who started antibiotic treatment between January 2002 and December 2013, we identified 72 patients with refractory disease, characterized by persistently positive sputum cultures despite ≥12 months of treatment. Molecular analyses of the 23S rRNA gene responsible for macrolide resistance and serial mycobacterial genotyping were performed using stored MAC isolates. Results: The median duration of treatment was 32 months (interquartile range, 24-41 months) in 72 patients. After treatment for a median of 33 months (interquartile range, 21-44 months), macrolide resistance developed in 16 (22%) patients. Molecular analysis of isolates from 15 patients revealed that 80% (12/15) had a point mutation at position 2058 or 2059 of the 23S rRNA gene. Of the 49 patients who had stored pre- and post-treatment isolates, mycobacterial genotyping revealed that reinfection by new MAC strains occurred in 36 (73%) patients. New MAC strains were found in 24 (49%) patients, and mixed infections with original and new strains occurred in 12 (24%) patients. Only 13 (27%) patients had persistent infections with their original MAC strains. Conclusions: Refractory MAC-LD is commonly caused by reinfection with new strains rather than persistence of the original strain, which may explain the infrequent development of macrolide resistance in refractory MAC-LD.

Ko RE, et al. J Korean Med Sci. 2018 Jan;33(8):e65. https://doi.org/10.3346/jkms.2018.33.e65