Catching the ‘tigers’: Retrospective cohort study identifying prognostic factors for prostate cancer progression using routine primary care data

Talk Code: 
Sam Merriel
Margaret May, Richard M Martin
Author institutions: 
University of Bristol


Prostate cancer is the most common cancer in males in the UK, and the second most common cause of cancer death amongst men. 5-year age-standardised survival for prostate cancer patients is 83.6%; therefore, many men live with the disease rather than dying from it. Some of these men with low grade disease are being actively monitored in shared care arrangements between primary and secondary care. Identifying clinically significant cases of prostate cancer with higher risk of progression remains a challenge. This study aimed to identify prognostic factors for prostate cancer progression in men using primary care data.


Retrospective cohort study using the Clinical Practice Research Datalink (CPRD) to identify men with a new diagnosis of prostate cancer between 01/01/1987 and 31/07/2016, with linked cancer registry and Office for National Statistics (ONS) mortality data. Men with a diagnosis outside of the study dates and men with metastatic disease at the time of diagnosis were excluded. Exposure to predetermined clinical, pharmacological, biochemical and lifestyle risk factors for prostate cancer mortality were extracted. Cox proportion hazard regression was utilised to determine factors associated with prostate cancer progression. The primary outcome was prostate cancer mortality. Secondary outcomes included all-cause mortality, developing locally invasive and/or metastatic disease, and commencing systemic therapy.


54,500 men (mean age 76.79 years [SD 9.31 years]) from CPRD with prostate cancer were analysed. Mean follow-up was 13.14 years (SD 6.58 years). 15,786 died during the follow-up period, 6,684 (42.34%) of whom died of prostate cancer. The most prevalent risk factors were BMI > 24.99 (n = 31746 [58.25%]), ischaemic heart disease (n = 9686 [17.77%]), and type 2 diabetes mellitus (n = 8109 [14.88%]). Aspirin (n = 2652 [4.87%]) was the most commonly prescribed medication in the 12 months prior to diagnosis. Raised acute phase reactants (CRP HR 2.15 95% CI 1.94, 2.38; Ferritin HR 1.65 95% CI 1.41, 1.93), low albumin (HR 2.72 95% CI 2.47, 3), anaemia (HR 2.61 95% CI 2.46, 2.78) and beta blockers were associated with increased prostate cancer mortality. Overweight (HR 0.82 95% CI 0.78, 0.86) and obese (HR 0.84 95% CI 0.78, 0.90) men had a reduced risk prostate cancer mortality.


A number of clinical prognostic factors for progression of prostate cancer have been identified. Identifying men with a higher risk of prostate cancer progression is important to inform treatment decision-making and reduce over-treatment. The factors associated with prostate cancer mortality in this study could inform the development and validation of a clinical risk prediction model for use in clinical practice.

Submitted by: 
Sam Merriel
Funding acknowledgement: 
SM is supported by an NIHR Academic Clinical Fellowship in General Practice; RM is supported by a Cancer Research UK programme grant, the Integrative Cancer Epidemiology Programme (C18281/A19169).