Development and Validation of a Risk Prediction Tool for Second Primary Lung Cancer

Software Application Profile: dynamicLM-a tool for performing dynamic risk prediction using a landmark supermodel for survival data under competing risks

MOTIVATION

Providing a dynamic assessment of prognosis is essential for improved personalized medicine. The landmark model for survival data provides a potentially powerful solution to the dynamic prediction of disease progression. However, a general framework and a flexible implementation of the model that incorporates various outcomes, such as competing events, have been lacking. We present an R package, dynamicLM, a user-friendly tool for the landmark model for the dynamic prediction of survival data under competing risks, which includes various functions for data preparation, model development, prediction and evaluation of predictive performance.

IMPLEMENTATION

dynamicLM as an R package.

GENERAL FEATURES

The package includes options for incorporating time-varying covariates, capturing time-dependent effects of predictors and fitting a cause-specific landmark model for time-to-event data with or without competing risks. Tools for evaluating the prediction performance include time-dependent area under the ROC curve, Brier Score and calibration.

AVAILABILITY

Available on GitHub [https://github.com/thehanlab/dynamicLM].

Second Primary Lung Cancer Among Lung Cancer Survivors Who Never Smoked

Importance

Lung cancer among never-smokers accounts for 25% of all lung cancers in the US; recent therapeutic advances have improved survival among patients with initial primary lung cancer (IPLC), who are now at high risk of developing second primary lung cancer (SPLC). As smoking rates continue to decline in the US, it is critical to examine more closely the epidemiology of lung cancer among patients who never smoked, including their risk for SPLC.

Objective

To estimate and compare the cumulative SPLC incidence among lung cancer survivors who have never smoked vs those who have ever smoked.

Design, Setting, and Participants

This population-based prospective cohort study used data from the Multiethnic Cohort Study (MEC), which enrolled participants between April 18, 1993, and December 31, 1996, with follow-up through July 1, 2017. Eligible individuals for this study were aged 45 to 75 years and had complete smoking data at baseline. These participants were followed up for IPLC and further SPLC development through the Surveillance, Epidemiology, and End Results registry. The data were analyzed from July 1, 2022, to January 31, 2023.

Exposures

Never-smoking vs ever-smoking exposure at MEC enrollment.

Main Outcomes and Measures

The study had 2 primary outcomes: (1) 10-year cumulative incidence of IPLC in the entire study cohort and 10-year cumulative incidence of SPLC among patients with IPLC and (2) standardized incidence ratio (SIR) (calculated as the SPLC incidence divided by the IPLC incidence) by smoking history.

Results

Among 211 414 MEC participants, 7161 (3.96%) developed IPLC over 4 038 007 person-years, and 163 (2.28%) developed SPLC over 16 470 person-years. Of the participants with IPLC, the mean (SD) age at cohort enrollment was 63.6 (7.7) years, 4031 (56.3%) were male, and 3131 (43.7%) were female. The 10-year cumulative IPLC incidence was 2.40% (95% CI, 2.31%-2.49%) among ever-smokers, which was 7 times higher than never-smokers (0.34%; 95% CI, 0.30%-0.37%). However, the 10-year cumulative SPLC incidence following IPLC was as high among never-smokers (2.84%; 95% CI, 1.50%-4.18%) as ever-smokers (2.72%; 95% CI, 2.24%-3.20%), which led to a substantially higher SIR for never-smokers (14.50; 95% CI, 8.73-22.65) vs ever-smokers (3.50; 95% CI, 2.95-4.12).

Conclusions and Relevance

The findings indicate that SPLC risk among lung cancer survivors who never smoked is as high as among those with IPLC who ever-smoked, highlighting the need to identify risk factors for SPLC among patients who never smoked and to develop a targeted surveillance strategy.

Combining Asian and European genome-wide association studies of colorectal cancer improves risk prediction across racial and ethnic populations

Polygenic risk scores (PRS) have great potential to guide precision colorectal cancer (CRC) prevention by identifying those at higher risk to undertake targeted screening. However, current PRS using European ancestry data have sub-optimal performance in non-European ancestry populations, limiting their utility among these populations. Towards addressing this deficiency, we expand PRS development for CRC by incorporating Asian ancestry data (21,731 cases; 47,444 controls) into European ancestry training datasets (78,473 cases; 107,143 controls). The AUC estimates (95% CI) of PRS are 0.63(0.62-0.64), 0.59(0.57-0.61), 0.62(0.60-0.63), and 0.65(0.63-0.66) in independent datasets including 1681-3651 cases and 8696-115,105 controls of Asian, Black/African American, Latinx/Hispanic, and non-Hispanic White, respectively. They are significantly better than the European-centric PRS in all four major US racial and ethnic groups (p-values < 0.05). Further inclusion of non-European ancestry populations, especially Black/African American and Latinx/Hispanic, is needed to improve the risk prediction and enhance equity in applying PRS in clinical practice.

Recurrence After Complete Resection for Non-Small Cell Lung Cancer in the National Lung Screening Trial

BACKGROUND

The objective of this study was to evaluate patterns, predictors, and long-term outcomes of recurrent disease after complete resection for early-stage non-small cell lung cancer (NSCLC) using the National Lung Screening Trial (NLST).

METHODS

The frequency of recurrence in patients with pathologic stage I-II NSCLC who underwent complete resection (lobectomy or bilobectomy) in the NLST was evaluated. Predictors of increased risk of recurrence were assessed by Fine-Gray competing risks regression.

RESULTS

Of the 497 patients meeting study inclusion criteria, 94 experienced a recurrence-a rate of 4.9 (95% CI, 4.0-6.0) per 100 person-years. The 5-year cumulative incidence of recurrence was 20.1% (95% CI, 16.5%-23.9%). Most patients experienced recurrences at distant sites alone (n = 47 [50.0%]) or at both locoregional and distant sites (n = 30 [31.9%]). The median time from resection to recurrence was 18.8 (10.6-30.7) months. The incidence rate of recurrence was significantly lower among patients with lung cancer detected by low-dose computed tomography screening during one of the three screening rounds of the NLST when compared with patients with lung cancer detected by chest radiography screening and patients with lung cancer not detected by any form of screening (ie, those diagnosed after a negative or missed screening exam and those diagnosed during follow-up after the three screening rounds of the NLST were completed) (P < .001). Median survival (from the date of recurrence) of patients with pathologic stage I and stage II disease who had recurrences at locoregional, distant, or both sites was 63.0, 23.1, and 9.8 months and 28.9, 8.7, and 10.2 months, respectively.

CONCLUSIONS

In this analysis of NLST participants with completely resected stage I-II NSCLC, the 5-year cumulative incidence of recurrence was 20%. Nearly 82% of recurrences were at distant sites and associated with poor survival.

Optimizing Lung Cancer Screening With Risk Prediction: Current Challenges and the Emerging Role of Biomarkers

The Oncology Grand Rounds series is designed to place original reports published in the Journal into clinical context. A case presentation is followed by a description of diagnostic and management challenges, a review of the relevant literature, and a summary of the authors' suggested management approaches. The goal of this series is to help readers better understand how to apply the results of key studies, including those published in Journal of Clinical Oncology, to patients seen in their own clinical practice.Lung cancer screening has been demonstrated to reduce lung cancer mortality, but its benefits must be weighed against the potential harms of unnecessary procedures, false-positive radiological findings, and overdiagnosis. Individuals at highest risk of lung cancer are more likely to maximize benefits while minimizing harm from screening. Although current lung cancer screening guidelines recommended by the US Preventive Services Task Force (USPSTF) only consider age and smoking history for screening eligibility, National Comprehensive Cancer Network and other society guidelines recommend screening on the basis of individualized risk assessment including family history, environmental exposures, and presence of chronic lung disease. Risk prediction models have been developed to integrate various risk factors into an individualized risk prediction score. Previous evidence showed that risk prediction model-based screening eligibility could improve sensitivity for detecting lung cancer cases without reducing specificity. Furthermore, recent advances in lung cancer biomarkers have enhanced the performance of risk prediction in identifying lung cancer cases relative to the USPSTF criteria. These risk prediction models can be used to guide shared decision-making discussions before proceeding with lung cancer screening. This study aims to provide a concise overview of these prediction models and the emerging role of biomarker testing in risk prediction to facilitate conversations with patients. The goal was to assist clinicians in assessing individual patient risk, leading to more informed decision making.

Developing a machine learning algorithm to predict probability of retear and functional outcomes in patients undergoing rotator cuff repair surgery: protocol for a retrospective, multicentre study

INTRODUCTION

The effectiveness of rotator cuff tear repair surgery is influenced by multiple patient-related, pathology-centred and technical factors, which is thought to contribute to the reported retear rates between 17% and 94%. Adequate patient selection is thought to be essential in reaching satisfactory results. However, no clear consensus has been reached on which factors are most predictive of successful surgery. A clinical decision tool that encompassed all aspects is still to be made. Artificial intelligence (AI) and machine learning algorithms use complex self-learning models that can be used to make patient-specific decision-making tools. The aim of this study is to develop and train an algorithm that can be used as an online available clinical prediction tool, to predict the risk of retear in patients undergoing rotator cuff repair.

METHODS AND ANALYSIS

This is a retrospective, multicentre, cohort study using pooled individual patient data from multiple studies of patients who have undergone rotator cuff repair and were evaluated by advanced imaging for healing at a minimum of 6 months after surgery. This study consists of two parts. Part one: collecting all potential factors that might influence retear risks from retrospective multicentre data, aiming to include more than 1000 patients worldwide. Part two: combining all influencing factors into a model that can clinically be used as a prediction tool using machine learning.

ETHICS AND DISSEMINATION

For safe multicentre data exchange and analysis, our Machine Learning Consortium adheres to the WHO regulation 'Policy on Use and Sharing of Data Collected by WHO in Member States Outside the Context of Public Health Emergencies'. The study results will be disseminated through publication in a peer-reviewed journal. Institutional Review Board approval does not apply to the current study protocol.

Combining Asian-European Genome-Wide Association Studies of Colorectal Cancer Improves Risk Prediction Across Race and Ethnicity

Polygenic risk scores (PRS) have great potential to guide precision colorectal cancer (CRC) prevention by identifying those at higher risk to undertake targeted screening. However, current PRS using European ancestry data have sub-optimal performance in non-European ancestry populations, limiting their utility among these populations. Towards addressing this deficiency, we expanded PRS development for CRC by incorporating Asian ancestry data (21,731 cases; 47,444 controls) into European ancestry training datasets (78,473 cases; 107,143 controls). The AUC estimates (95% CI) of PRS were 0.63(0.62-0.64), 0.59(0.57-0.61), 0.62(0.60-0.63), and 0.65(0.63-0.66) in independent datasets including 1,681-3,651 cases and 8,696-115,105 controls of Asian, Black/African American, Latinx/Hispanic, and non-Hispanic White, respectively. They were significantly better than the European-centric PRS in all four major US racial and ethnic groups (p-values<0.05). Further inclusion of non-European ancestry populations, especially Black/African American and Latinx/Hispanic, is needed to improve the risk prediction and enhance equity in applying PRS in clinical practice.

Postoperative radiotherapy might be a risk factor for second primary lung cancer: A population-based study

BACKGROUND

Surgery is the main curative therapeutic strategy for patients with initial primary lung cancer (IPLC). Most international guidelines recommend regular follow-ups after discharge to monitor patients for tumor recurrence and metastasis. As the overall survival (OS) in patients with lung cancer improves, their risk of secondary primary lung cancer (SPLC) increases. Previous studies on such patients lack separate assessment of different survival outcomes and evaluation of high-risk factors for SPLC. Therefore, we aimed to determine the correlation between high-risk factors and causes of death in patients with SPLC, based on the Surveillance, Epidemiology, and End Results (SEER) database.

METHODS

We screened the SEER database for patients with IPLC and SPLC from 2004 to 2015 and included only patients who underwent surgery since the IPLC and in whom the cancer was pathologically verified of an International Classification of Diseases grade of 0-3 and to be non-small-cell lung cancer. The standardized incidence ratio (SIR) was calculated between variables and SPLC. Multivariable Cox proportional-hazards regression analyses were conducted to calculate the correlation of different variables with overall survival (OS) and cancer-specific survival (CSS). A competing-risk model was conducted for SPLC. The effect of baseline bias on survival outcomes by performing propensity score matching analysis in a 1: 6 ratio (SPLC: IPLC).

RESULTS

For patients aged 0-49 years, the overall SIR was higher in older patients, reaching a maximum of 27.74 in those aged 40-49 years, and at 11.63 in patients aged 50-59 years. The overall SIR was higher for patients who were more recently diagnosed with IPLC and increased with time after diagnosis. Male sex, SPLC (hazard ratio, 1.6173; 95% confidence interval, 1.5505-1.6869; P < 0.001), cancer grade III or IV, lower lobe of the lung, advanced stage and postoperative radiotherapy (PORT) were independently detrimental to OS. In terms of CSS, PORT was a high-risk factor.

CONCLUSIONS

Postoperative radiotherapy is a risk factor for second primary lung cancer and detrimental to overall and cancer-specific survival in patients who had initial primary lung cancer. These data support the need for life-long follow-up of patients who undergo treatment for IPLC to screen for SPLC.