Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom

Hilary A Robbins; Karine Alcala; Anthony J Swerdlow; Minouk J Schoemaker; Nick Wareham; Ruth C Travis; Philip A J Crosbie; Matthew Callister; David R Baldwin; Rebecca Landy; Mattias Johansson

doi:10.1038/s41416-021-01278-0

Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom

Hilary A Robbins, Karine Alcala, Anthony J Swerdlow, Minouk J Schoemaker, Nick Wareham, Ruth C Travis, Philip A J Crosbie, Matthew Callister, David R Baldwin, Rebecca Landy, Mattias Johansson

Division of Immunology, Immunity to Infection and Respiratory Medicine

Research output: Contribution to journal › Article › peer-review

Abstract

BACKGROUND: The National Health Service England (NHS) classifies individuals as eligible for lung cancer screening using two risk prediction models, PLCOm2012 and Liverpool Lung Project-v2 (LLPv2). However, no study has compared the performance of lung cancer risk models in the UK.

METHODS: We analysed current and former smokers aged 40-80 years in the UK Biobank (N = 217,199), EPIC-UK (N = 30,813), and Generations Study (N = 25,777). We quantified model calibration (ratio of expected to observed cases, E/O) and discrimination (AUC).

RESULTS: Risk discrimination in UK Biobank was best for the Lung Cancer Death Risk Assessment Tool (LCDRAT, AUC = 0.82, 95% CI = 0.81-0.84), followed by the LCRAT (AUC = 0.81, 95% CI = 0.79-0.82) and the Bach model (AUC = 0.80, 95% CI = 0.79-0.81). Results were similar in EPIC-UK and the Generations Study. All models overestimated risk in all cohorts, with E/O in UK Biobank ranging from 1.20 for LLPv3 (95% CI = 1.14-1.27) to 2.16 for LLPv2 (95% CI = 2.05-2.28). Overestimation increased with area-level socioeconomic status. In the combined cohorts, USPSTF 2013 criteria classified 50.7% of future cases as screening eligible. The LCDRAT and LCRAT identified 60.9%, followed by PLCOm2012 (58.3%), Bach (58.0%), LLPv3 (56.6%), and LLPv2 (53.7%).

CONCLUSION: In UK cohorts, the ability of risk prediction models to classify future lung cancer cases as eligible for screening was best for LCDRAT/LCRAT, very good for PLCOm2012, and lowest for LLPv2. Our results highlight the importance of validating prediction tools in specific countries.

Original language	English
Pages (from-to)	2026-2034
Number of pages	9
Journal	British Journal of Cancer
Volume	124
Issue number	12
Early online date	12 Apr 2021
DOIs	https://doi.org/10.1038/s41416-021-01278-0
Publication status	Published - 8 Jun 2021

Keywords

Adult
Aged
Calibration
Cohort Studies
Early Detection of Cancer/methods
Female
Humans
Lung Neoplasms/diagnosis
Male
Middle Aged
Models, Statistical
Patient Selection
Predictive Value of Tests
Risk Assessment
Risk Factors
Social Class
State Medicine
United Kingdom/epidemiology

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1038/s41416-021-01278-0

Cite this

Robbins, H. A., Alcala, K., Swerdlow, A. J., Schoemaker, M. J., Wareham, N., Travis, R. C., Crosbie, P. A. J., Callister, M., Baldwin, D. R., Landy, R., & Johansson, M. (2021). Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom. British Journal of Cancer, 124(12), 2026-2034. https://doi.org/10.1038/s41416-021-01278-0

@article{67a836ac91e44be7974943a4065b2309,

title = "Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom",

abstract = "BACKGROUND: The National Health Service England (NHS) classifies individuals as eligible for lung cancer screening using two risk prediction models, PLCOm2012 and Liverpool Lung Project-v2 (LLPv2). However, no study has compared the performance of lung cancer risk models in the UK.METHODS: We analysed current and former smokers aged 40-80 years in the UK Biobank (N = 217,199), EPIC-UK (N = 30,813), and Generations Study (N = 25,777). We quantified model calibration (ratio of expected to observed cases, E/O) and discrimination (AUC).RESULTS: Risk discrimination in UK Biobank was best for the Lung Cancer Death Risk Assessment Tool (LCDRAT, AUC = 0.82, 95% CI = 0.81-0.84), followed by the LCRAT (AUC = 0.81, 95% CI = 0.79-0.82) and the Bach model (AUC = 0.80, 95% CI = 0.79-0.81). Results were similar in EPIC-UK and the Generations Study. All models overestimated risk in all cohorts, with E/O in UK Biobank ranging from 1.20 for LLPv3 (95% CI = 1.14-1.27) to 2.16 for LLPv2 (95% CI = 2.05-2.28). Overestimation increased with area-level socioeconomic status. In the combined cohorts, USPSTF 2013 criteria classified 50.7% of future cases as screening eligible. The LCDRAT and LCRAT identified 60.9%, followed by PLCOm2012 (58.3%), Bach (58.0%), LLPv3 (56.6%), and LLPv2 (53.7%).CONCLUSION: In UK cohorts, the ability of risk prediction models to classify future lung cancer cases as eligible for screening was best for LCDRAT/LCRAT, very good for PLCOm2012, and lowest for LLPv2. Our results highlight the importance of validating prediction tools in specific countries.",

keywords = "Adult, Aged, Calibration, Cohort Studies, Early Detection of Cancer/methods, Female, Humans, Lung Neoplasms/diagnosis, Male, Middle Aged, Models, Statistical, Patient Selection, Predictive Value of Tests, Risk Assessment, Risk Factors, Social Class, State Medicine, United Kingdom/epidemiology",

author = "Robbins, {Hilary A} and Karine Alcala and Swerdlow, {Anthony J} and Schoemaker, {Minouk J} and Nick Wareham and Travis, {Ruth C} and Crosbie, {Philip A J} and Matthew Callister and Baldwin, {David R} and Rebecca Landy and Mattias Johansson",

note = "Funding Information: Funding information This study was funded in part by the US National Cancer Institute (R03 CA245979 and U19 CA203654) and Cancer Research UK (C18281/ A19169). The Generations Study was funded by Breast Cancer Now and the Institute of Cancer Research. The ICR acknowledges NHS funding to the Royal Marsden/ICR NIHR Biomedical Research Centre. P.A.J.C. is supported by the NIHR Manchester Biomedical Research Centre. M.C. is co-chief investigator of the Yorkshire Lung Screening Trial, which is funded by Yorkshire Cancer Research. R.L. is supported by the Intramural Research Program of the US National Institutes of Health/National Cancer Institute. The funders had no role in the design, analysis, or reporting of the study. Publisher Copyright: {\textcopyright} 2021, World Health Organization.",

year = "2021",

month = jun,

day = "8",

doi = "10.1038/s41416-021-01278-0",

language = "English",

volume = "124",

pages = "2026--2034",

journal = "British Journal of Cancer",

issn = "0007-0920",

publisher = "Springer Nature",

number = "12",

}

TY - JOUR

T1 - Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom

AU - Robbins, Hilary A

AU - Alcala, Karine

AU - Swerdlow, Anthony J

AU - Schoemaker, Minouk J

AU - Wareham, Nick

AU - Travis, Ruth C

AU - Crosbie, Philip A J

AU - Callister, Matthew

AU - Baldwin, David R

AU - Landy, Rebecca

AU - Johansson, Mattias

N1 - Funding Information: Funding information This study was funded in part by the US National Cancer Institute (R03 CA245979 and U19 CA203654) and Cancer Research UK (C18281/ A19169). The Generations Study was funded by Breast Cancer Now and the Institute of Cancer Research. The ICR acknowledges NHS funding to the Royal Marsden/ICR NIHR Biomedical Research Centre. P.A.J.C. is supported by the NIHR Manchester Biomedical Research Centre. M.C. is co-chief investigator of the Yorkshire Lung Screening Trial, which is funded by Yorkshire Cancer Research. R.L. is supported by the Intramural Research Program of the US National Institutes of Health/National Cancer Institute. The funders had no role in the design, analysis, or reporting of the study. Publisher Copyright: © 2021, World Health Organization.

PY - 2021/6/8

Y1 - 2021/6/8

N2 - BACKGROUND: The National Health Service England (NHS) classifies individuals as eligible for lung cancer screening using two risk prediction models, PLCOm2012 and Liverpool Lung Project-v2 (LLPv2). However, no study has compared the performance of lung cancer risk models in the UK.METHODS: We analysed current and former smokers aged 40-80 years in the UK Biobank (N = 217,199), EPIC-UK (N = 30,813), and Generations Study (N = 25,777). We quantified model calibration (ratio of expected to observed cases, E/O) and discrimination (AUC).RESULTS: Risk discrimination in UK Biobank was best for the Lung Cancer Death Risk Assessment Tool (LCDRAT, AUC = 0.82, 95% CI = 0.81-0.84), followed by the LCRAT (AUC = 0.81, 95% CI = 0.79-0.82) and the Bach model (AUC = 0.80, 95% CI = 0.79-0.81). Results were similar in EPIC-UK and the Generations Study. All models overestimated risk in all cohorts, with E/O in UK Biobank ranging from 1.20 for LLPv3 (95% CI = 1.14-1.27) to 2.16 for LLPv2 (95% CI = 2.05-2.28). Overestimation increased with area-level socioeconomic status. In the combined cohorts, USPSTF 2013 criteria classified 50.7% of future cases as screening eligible. The LCDRAT and LCRAT identified 60.9%, followed by PLCOm2012 (58.3%), Bach (58.0%), LLPv3 (56.6%), and LLPv2 (53.7%).CONCLUSION: In UK cohorts, the ability of risk prediction models to classify future lung cancer cases as eligible for screening was best for LCDRAT/LCRAT, very good for PLCOm2012, and lowest for LLPv2. Our results highlight the importance of validating prediction tools in specific countries.

AB - BACKGROUND: The National Health Service England (NHS) classifies individuals as eligible for lung cancer screening using two risk prediction models, PLCOm2012 and Liverpool Lung Project-v2 (LLPv2). However, no study has compared the performance of lung cancer risk models in the UK.METHODS: We analysed current and former smokers aged 40-80 years in the UK Biobank (N = 217,199), EPIC-UK (N = 30,813), and Generations Study (N = 25,777). We quantified model calibration (ratio of expected to observed cases, E/O) and discrimination (AUC).RESULTS: Risk discrimination in UK Biobank was best for the Lung Cancer Death Risk Assessment Tool (LCDRAT, AUC = 0.82, 95% CI = 0.81-0.84), followed by the LCRAT (AUC = 0.81, 95% CI = 0.79-0.82) and the Bach model (AUC = 0.80, 95% CI = 0.79-0.81). Results were similar in EPIC-UK and the Generations Study. All models overestimated risk in all cohorts, with E/O in UK Biobank ranging from 1.20 for LLPv3 (95% CI = 1.14-1.27) to 2.16 for LLPv2 (95% CI = 2.05-2.28). Overestimation increased with area-level socioeconomic status. In the combined cohorts, USPSTF 2013 criteria classified 50.7% of future cases as screening eligible. The LCDRAT and LCRAT identified 60.9%, followed by PLCOm2012 (58.3%), Bach (58.0%), LLPv3 (56.6%), and LLPv2 (53.7%).CONCLUSION: In UK cohorts, the ability of risk prediction models to classify future lung cancer cases as eligible for screening was best for LCDRAT/LCRAT, very good for PLCOm2012, and lowest for LLPv2. Our results highlight the importance of validating prediction tools in specific countries.

KW - Adult

KW - Aged

KW - Calibration

KW - Cohort Studies

KW - Early Detection of Cancer/methods

KW - Female

KW - Humans

KW - Lung Neoplasms/diagnosis

KW - Male

KW - Middle Aged

KW - Models, Statistical

KW - Patient Selection

KW - Predictive Value of Tests

KW - Risk Assessment

KW - Risk Factors

KW - Social Class

KW - State Medicine

KW - United Kingdom/epidemiology

U2 - 10.1038/s41416-021-01278-0

DO - 10.1038/s41416-021-01278-0

M3 - Article

C2 - 33846525

SN - 0007-0920

VL - 124

SP - 2026

EP - 2034

JO - British Journal of Cancer

JF - British Journal of Cancer

IS - 12

ER -

Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom

Abstract

Keywords

UN SDGs

Access to Document

Fingerprint

Cite this