A Decision Analysis to Assess the Value of Prostate Cancer Screening: A Shift in Focus from Prostate Cancer-specific Mortality to Distant Metastasis-Free Survival

Reporter: Annemarie Fernandes, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: October 30, 2012

Presenter: Arie Dosoretz, MD
Presenter's Affiliation: Yale-New Haven Hospital


  • Prostate cancer is the most common cancer in men and the 2nd leading cause of cancer death. Prostate cancer screening has been performed using a non-invasive blood test, prostate specific antigen (PSA)
  • The precise magnitude of over-diagnosis associated with any screening and treatment program is difficult to determine, but estimates from the 2 largest trials (US Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial (PLCO) and European Randomized Study of Screening for Prostate Cancer (ERSPC)) suggest over-diagnosis rates of 17% to 50% for prostate cancer screening. These studies did not report on the effect of screening on the development of subsequent metastatic disease.
  • The US Preventive Services Task Force has recommended against PSA screening for prostate cancer, given the indolent nature of the disease and the potential for treatment related harm
  • The authors created a decision analytic model (Markov model) to calculate the quality-adjusted life expectancy (QALE) associated with screening men for prostate cancer with annual PSA testing in order to provide further guidance on screening and treatment decisions

Materials and Methods

  • A state transition Markov model was constructed to compare QALE in men with and without annual PSA screening
  • The model used US life-expectancy tables, sensitivity/specificity and true positive/negative values of PSA screening and rates of GI/GU toxicity reported in randomized control trials.
  • Men entered the model at age 65 and exited after 25 years
  • For the initial model, the authors assumed that patients who screened positive presented with clinically localized, low-risk prostate cancer and received radiation treatment with IMRT
  • In men not screened for prostate cancer that presented with localized disease, the authors also assumed that the patients presented with clinically localized, low-risk prostate cancer and received radiation treatment with IMRT
  • The model was then revised and enhanced to account for multiple health states, to run the model more often than yearly and to include intermediate and high risk patients
  • Probabilities associated with diagnosis and treatment responses as well as the utility values for different health states including living with distant metastatic disease and living with long-term treatment related adverse effects (genitourinary, gastrointestinal and sexual dysfunction) were derived from validated prospective studies


  • Annual PSA screening resulted in greater QALE compared to no PSA screening, producing 13.7 quality-adjusted life-years (QALYs) vs. 13.24 QALYs, respectively
  • The benefit of PSA screening on QALE decreased with increasing age
  • One-way sensitivity analysis revealed that the model is most sensitive to the probability of developing metastatic disease. If the 10-year probability of developing metastatic disease in the unscreened group was below 4.9%, no screening became the preferred strategy
  • As the risk of distant metastasis increases, screening has more benefit. However, the benefit of screening decreases as toxicity of treatment increases
  • Using the enhanced model, the authors found that patients screened would have a 3.1% risk of developing distant metastatic disease, while the unscreened patients would have a 8.5% risk of developing distant metastatic disease

Author's Conclusions

  • The authors conclude that annual PSA screening resulting in improved quality-adjusted life expectancy, with the exception of men with a very low risk of developing metastatic disease
  • The decision analysis suggests that the morbidity associated with a higher risk of developing metastatic disease outweighs the possibility of developing long-term treatment-related adverse effects when 65 year old men with average life expectancy and low-risk disease are treated with IMRT
  • The authors conclude that future decision-making analyses should account for individualized utilities and the economic costs associated with prostate cancer screening and treatment

Clinical Implications

  • In May of 2012, the USPSTF recommended against PSA screening, as a category D recommendation for all men, regardless of age, race, family history, etc. This general recommendation derived from 2 large screening trials (PLCO and ERSPC) and has been controversial.
  • Prostate cancer is the most common cancer in men and the second leading cause of cancer deaths. The screening mechanism for detecting prostate cancer is non-invasive, however these studies indicated that PSA screening results in unnecessary diagnosis-related (over-diagnosis) and treatment-related (over-treatment) morbidity due to the generally indolent nature of prostate cancer
  • Despite the frequent cases of over-diagnosis and over-treatment, prostate cancer remains the second leading cause of cancer death in men and is a very heterogenous disease. While the majority of patients present with indolent disease, there is a subset of patients (particularly African American males and men with family history of aggressive prostate cancer) who have aggressive disease. These patients will suffer from the USPSTF decision as their prostate cancer will only be detected after they become symptomatic from their disease
  • The authors present intriguing data from a decision analytic model that accounts for unnecessary treatment-related morbidity and ultimately demonstrates a benefit to PSA screening by decreasing the risk of developing distant metastatic disease
  • The previously published studies (PPLCO and ERSPC) did not evaluate the impact of PSA screening on the development of distant metastatic disease.
  • With the development of this analytic model, other questions can be evaluated and additional factors such as baseline sexual or urinary function can be taken into account to determine the value of PSA screening in patients with poor vs. adequate baseline function. Additionally, this model assumed that patients would be treated with IMRT. This model can be furthered to evaluate the impact of screening if a patient were to undergo other modalities of treatment such as surgery or brachytherapy.
  • While this model is hypothesis generating, the data is limited by the theoretical nature of the model. Information generated from analytic models is dependent on the accuracy and robustness of the data input. Varying the input can drastically change the outcomes of the model. Additionally, the models do not account for unmeasured confounders.
  • Nonetheless, this data provides some basis for the need to re-evaluate the role of PSA screening in men.
  • At this time, despite the prostate cancer nomograms and risk stratification models, we are unable to predict for aggressive disease and are unable to effectively identify patients at high risk for developing distant metastatic disease. Until predictive markers for developing distant metastatic disease are available, PSA is all we have to detect potentially aggressive prostate cancer. Decisions to discontinue screening should be made only after careful risk-benefit discussion between a patient and his physician and healthcare team.