Use of deep learning to predict postoperative recurrence of lung adenocarcinoma from preoperative CT

PURPOSE

Although surgery is the primary treatment for lung cancer, some patients experience recurrence at a certain rate. If postoperative recurrence can be predicted early before treatment is initiated, it may be possible to provide individualized treatment for patients. Thus, in this study, we propose a computer-aided diagnosis (CAD) system that predicts postoperative recurrence from computed tomography (CT) images acquired before surgery in patients with lung adenocarcinoma using a deep convolutional neural network (DCNN).

METHODS

This retrospective study included 150 patients who underwent curative surgery for primary lung adenocarcinoma. To create original images, the tumor part was cropped from the preoperative contrast-enhanced CT images. The number of input images to the DCNN was increased to 3000 using data augmentation. We constructed a CAD system by transfer learning using a pretrained VGG19 model. Tenfold cross-validation was performed five times. Cases with an average identification rate of 0.5 or higher were determined to be a recurrence.

RESULTS

The median duration of follow-up was 73.2 months. The results of the performance evaluation showed that the sensitivity, specificity, and accuracy of the proposed method were 0.75, 0.87, and 0.82, respectively. The area under the receiver operating characteristic curve was 0.86.

CONCLUSION

We demonstrated the usefulness of DCNN in predicting postoperative recurrence of lung adenocarcinoma using preoperative CT images. Because our proposed method uses only CT images, we believe that it has the advantage of being able to assess postoperative recurrence on an individual patient basis, both preoperatively and noninvasively.

Postoperative Adverse Events are Associated with Oncologic Recurrence Following Curative-intent Resection for Lung Cancer

BACKGROUND

Up to 50% of patients suffer short-term postoperative adverse events (AEs) and metastatic recurrence in the long-term following curative-intent lung cancer resection. The association between AEs, particularly infectious in nature, and disease recurrence is controversial. We sought to evaluate the association of postoperative AEs on risk of developing recurrence and recurrence-free survival (RFS) following curative-intent lung resection surgery.

METHODS

All lung cancer resections at a single institution (January 2008-July 2015) were included, with prospective collection of AEs using the Thoracic Morbidity & Mortality System. Cox proportional hazards models were used to estimate the effect of AEs on recurrence, with results presented as hazard ratio (HR) with 95% confidence interval (CI). An a priori, clinically driven approach to predictor variable selection was used. Kaplan-Meier curves were used examine the relationship between AE and RFS. p < 0.05 was considered statistically significant.

RESULTS

892 patients underwent curative-intent resection. 342 (38.3%) patients experienced an AE; 69 (7.7%) patients developed infectious AEs. 17.6% (n = 157) of patients had disease recurrence after mean follow-up of 26.5 months. Severe (Grade IV) AEs were associated with increased risk of recurrence (3.40; 95% CI 1.56-7.41) and a trend to decreased RFS. Major infectious AEs were associated with increased risk of recurrence (HR 1.71; CI 1.05-2.8) and earlier time to recurrence (no infectious AE 66 months, minor infectious 41 months, major infectious 54 months; p = 0.02).

CONCLUSION

For patients undergoing curative-intent lung cancer resection, postoperative AEs associated with critical illness or major infection were associated with increased risk of oncologic recurrence.

Application of survival tree analysis for exploration of potential interactions between predictors of incident chronic kidney disease: a 15-year follow-up study

Background

Chronic kidney disease (CKD) is a growing public health challenges worldwide. Various studies have investigated risk factors of incident CKD; however, a very few studies examined interaction between these risk factors. In an attempt to clarify the potential interactions between risk factors of CKD, we performed survival tree analysis.

Methods

A total of 8238 participants (46.1% men) aged > 20 years without CKD at baseline [(1999–2001) and (2002–2005)], were followed until 2014. The first occurrence of CKD, defined as the estimated glomerular filtration rate (eGFR) < 60 ml/min/1.73 m², was set as the main outcome. Multivariable Cox proportional hazard (Cox PH) regression was used to identify significant independent predictors of CKD; moreover, survival tree analysis was performed to gain further insight into the potential interactions between predictors.

Results

The crude incidence rates of CKD were 20.2 and 35.2 per 1000 person-years in men and women, respectively. The Cox PH identified the main effect of significant predictors of CKD incidence in men and women. In addition, using a limited number of predictors, survival trees identified 12 and 10 subgroups among men and women, respectively, with different survival probability. Accordingly, a group of men with eGFR > 74 ml/min/1.73 m², age ≤ 46 years, low level of physical activity, waist circumference ≤ 100 cm and FPG ≤ 4.7 mmol/l had the lowest risk of CKD incidence; while men with eGFR ≤ 63.4 ml/min/1.73 m², age > 50 years had the highest risk for CKD compared to men in the lowest risk group [hazard ratio (HR), 70.68 (34.57–144.52)]. Also, a group of women aged ≤ 45 years and eGFR > 83.5 ml/min/1.73 m² had the lowest risk; while women with age > 48 years and eGFR ≤ 69 ml/min/1.73 m² had the highest risk compared to low risk group [HR 27.25 (19.88–37.34)].

Conclusion

In this post hoc analysis, we found the independent predictors of CKD using Cox PH; furthermore, by applying survival tree analysis we identified several numbers of homogeneous subgroups with different risk for incidence of CKD. Our study suggests that two methods can be used simultaneously to provide new insights for intervention programs and improve clinical decision making.

Prediction of distant recurrence in resected stage I and II lung adenocarcinoma

OBJECTIVES

Optimal procedures for adjuvant treatment and post-surgical surveillance of resected non-small-cell lung cancer remain under discussion. Pathological features are the main determinant of follow-up therapy but have limited ability to identify patients at risk of recurrence. Increasingly, molecular markers are incorporated into clinical decision-making, including measures of tumor growth. The CCP score is a quantitative, molecular measure of proliferation derived from the RNA expression of 31 cell cycle genes and a component of the molecular prognostic score (mPS). The mPS score is a linear combination of CCP score and pathological stage. CCP score and mPS are independent predictors of survival in resected lung adenocarcinoma.

MATERIALS AND METHODS

CCP scores were determined by RT-qPCR for 318 patients diagnosed with stage I-II lung adenocarcinoma. Association of mPS and CCP score with distant recurrence and lung-cancer specific survival was assessed in Cox proportional hazards regression models adjusted for age, gender, tumor size, pathological stage and pleural invasion. Distant recurrence-free survival and lung-cancer specific survival by mPS risk group were calculated by Kaplan-Meier survival analysis.

RESULTS

CCP scores were obtained for 205 stage I and 84 stage II patients. CCP score and mPS were independent markers of distant recurrence (CCP: HR 1.62, 95%CI 1.15-2.29, p=0.0055; mPS: HR 2.22, 95%CI 1.11-4.44, p=0.023). Patients with low mPS tumors were at significantly reduced risk of distant recurrence (log-rank p=4.2×10^-5). Among stage I patients, stratification by mPS identified a patient group with increased risk of distant recurrence (36%, 95%CI 28-46%, log-rank p=0.0011) CONCLUSIONS: The molecular prognostic score stratifies early-stage, resected lung cancer patients for risk of distant recurrence and could be useful to inform treatment and surveillance decisions.

Long-term Outcomes of Patients With Ground-Glass Opacities Detected Using CT Scanning

BACKGROUND

The long-term outcomes of follow-up care for ground-glass opacity (GGO) lesions need to be clarified.

METHODS

Between 2000 and 2005, a total of 226 patients with pure or mixed GGO lesions ≤ 3 cm in size were registered. The CT findings and changes in the findings during the follow-up period and the outcomes of the 226 patients were subsequently reviewed.

RESULTS

Overall, 124 patients underwent resections, 57 did not receive follow-up examinations after 68 months because of stable disease or disease reduction, and 45 are continuing to receive follow-up examinations. Thirty-nine patients exhibited tumor growth during the follow-up period. Among the patients with a ratio of the diameter of consolidation relative to the tumor diameter (CTR) > 0, all cases with tumor growth were identified within 3 years; meanwhile, > 3 years were required to identify tumor growth in 16% of the patients with a CTR of 0. Aggressive cancer occurred in 4% of patients with a CTR of 0 and in 70% of patients with a CTR > 25%. Aggressive cancer was observed in 46% of the patients whose CTR increased during the follow-up period and in 8% of the patients whose tumors increased in size.

CONCLUSIONS

A higher CTR and an increase in CTR during follow-up were associated with invasive cancer. A follow-up period of 3 years is considered to be adequate for judging tumor growth in patients with a CTR > 0, whereas a longer follow-up period might be needed for patients with a CTR of 0.

Factors affecting tumor recurrence after curative surgery for NSCLC: impacts of lymphovascular invasion on early tumor recurrence

BACKGROUND

Although surgery is potentially curative treatment for non-small cell lung cancer (NSCLC), the risk of postoperative disease recurrence is still high. This study was conducted to assess the factors associated with postoperative tumor recurrence in patients who underwent curative surgery for NSCLC.

METHODS

One hundred seventy-one patients who underwent curative surgery for NSCLC were included in this study. Clinicopathological factors of histologic type, pathologic TNM stage, T stage, N stage, lymphovascular invasion (LVI), perineural invasion (PNI), surgical procedure, adjuvant chemotherapy and adjuvant radiotherapy were investigated. Gender, age, and clinicopathologic factors were included in univariate and multivariate analyses using the Kaplan-Meier method and Cox proportional hazards model, respectively. Mann-Whitney U and Kruskal-Wallis tests were used to investigate the significance of differences in recurrence-free interval (RFI) according to clinicopathological factors.

RESULTS

Median RFI was 20 months. Univariate and multivariate analyses for overall recurrence identified T stage, N stage, and LVI as significant factors (P=0.045, 0.044, and <0.001, respectively). Pathologic stage (P=0.005) was the only factor that was significantly associated with locoregional recurrence. T stage (P=0.040) and LVI (P<0.001) were significantly associated with distant recurrence. The difference in 2-year freedom from recurrence between LVI positive and negative groups was significant (14.9% vs. 44.6%, P<0.001). LVI was the only factor that was significantly associated with a shortened mean RFI (P<0.001).

CONCLUSIONS

LVI had a significant effect on both overall and distant recurrence rates as well as on early tumor recurrence after curative surgery for NSCLC.