Comparison of rates of radiological imaging in control arm of PLCO vs UK population
The PLCO study assessed the effect of CXR screening on lung cancer mortality in a North American population. CXR screening resulted in only a small increase in the proportion of patients presenting with stage I disease (27.1% in control population compared to 31.7% in screened population) and no effect on lung cancer mortality. In comparison, NLST reported a larger increase in stage I disease seen with CT screening (50.0% in CT screened population vs 31.1% in CXR screened population) and a 20% reduction in lung cancer mortality. Given that some UK early diagnosis initiatives are seeking to increase referral for CXR even in the absence of respiratory symptoms, the relevance of the PLCO findings to the UK population needs to be established.
The proportion of patients diagnosed with lung cancer who present with early stage disease is much smaller in the UK population than in the control arm of PLCO (13.5% with stage I disease in UK ICBP vs 27.1% in PLCO control). The reasons for this difference are unclear, but may relate to different background levels of radiological investigation in the population at large (CXR or CT), and are likely to lead to worse survival in UK patients compared to those in the control arm of PLCO. Leeds Teaching Hospitals (LTH) serves the city of Leeds (North England, population 770,000) and sees the largest number of new lung cancer diagnoses of any single hospital in the UK (610 in 2012). A comprehensive database has been established of both lung cancer patient characteristics and outcomes, and of CXR usage in the general population of the city. The UK National Lung Cancer Audit records information about an estimated 98% of cases of lung cancer diagnosed per year. Similarly the recently established NHS Diagnostic Imaging Dataset records information on all radiological tests undertaken including chest X-ray.
This proposal seeks to assess whether the difference in stage distribution between lung cancers diagnosed in the PLCO control population and the UK (nationally and in Leeds) reflects different background rates of CXR usage in these populations. Secondly it will compare outcome between lung cancers diagnosed in these groups. The proposal is to analyse data from three population cohorts – the control population in PLCO (n=77,456), the population of England and Wales aged 55-75 (n=12.6 million), and the population of Leeds aged 55-75 (n=130,000). Rate of CXR per patient per year will be compared between these three groups for the duration of 1993-2001, 2012-2013, and 2008-2012 respectively. Further analysis will focus on patients diagnosed with lung cancer in these three cohort populations in the years considered (PLCO control n=1,620, England and Wales n=20,800, Leeds n=1,076). The number and timing of CXRs performed prior to lung cancer diagnosis will be calculated, and expressed as CXRs per year per 1,000 population. Lung cancer survival (and other clinical parameters available from PLCO dataset) will be compared between these patients diagnosed with lung cancer.
Dr Paul Pinsky
Acting Chief
Early Detection Research Group
Division of Cancer Prevention
National Cancer Institute
9609 Medical Center Drive, Room 5E444
Rockville, MD 20850