Toward a rationale strategy for prostate cancer screening based on long term risk of prostate cancer metastases and death: data from a large, unscreened, population-based cohort followed for up to 30 years

Reporter: J. Nicholas Lukens, MD
The Abramson Cancer Center of the University of Pennsylvania
Last Modified: June 7, 2011

Presenter: H. Lilja, MD
Presenter's Institution: Memorial Sloan-Kettering Cancer Center, New York, NY


  • Routine screening for prostate cancer using prostate-specific antigen (PSA) testing remains controversial.
    • A European study demonstrated that PSA screening reduced the rate of death from prostate cancer by 20%; however, screening was associated with a high risk of overdiagnosis. For example, in this study, 1410 men would need to be screened and 48 additional cases of prostate cancer would need to be treated to prevent one death from prostate cancer.
    • A similar study conducted in the United States (PLCO) with shorter follow-up (7 years) did not demonstrate a reduction in the number of deaths from prostate cancer in patients that were screened, although this time frame is likely too short to detect deaths from prostate cancer.
  • The risks of PSA screening include overdiagnosis, resulting in over-treatment. In addition, the test has low specificity (there are many false positives), and a poor ability to discriminate between indolent and significant disease.
  • Current screening recommendations in the United States are based on few data, and the screening recommendations of various organizations differ.
  • The authors set out to determine if there is an age at which men with a PSA level below a certain threshold level can safely be exempt from further screening.
  • To answer this question, they looked to a cohort of Swedish men followed as part of the Malmö Preventive Project, which was initially designed to analyze cardiovascular risk factors, to characterize the relationship between PSA and the natural history of clinical prostate cancer. These men had blood drawn at the time of entry into the cohort at the age of 33-50, but they were not subject to screening. They were then followed over 30 years, and there is detailed information regarding their subsequent diagnoses and cause of death. This population has several advantages for the study of the relationship between PSA and the natural history of prostate cancer:
    • PSA could be measured in their archived blood plasma, which had been collected at or before the age when they would have been screened in the US,
    • They were not in fact screened for prostate cancer,
    • There is very long-term follow up to ascertain clinically significant prostate cancer, and
    • There are detailed records regarding prostate cancer metastases and death.
  • The study provides a "natural experiment" to understand the association between early PSA and long-term risk of prostate cancer morbidity and mortality.
  • The authors previously reported (Vickers, BMJ 2010) that the concentration of PSA at age 60 predicts lifetime risk of metastasis and death from prostate cancer, and that men aged 60 with a concentration below the median (?1 ng/ml) were unlikely to develop clinically relevant prostate cancer (0.5% risk of DM and and 0.2% risk of death from prostate cancer).
  • The objective of the current study was to evaluate the relationship between PSA measured at early middle age and the risk of metastases or death from prostate cancer 30 years later.


  • The study analyzed data from 21,277 male Malmö Preventive Project participants giving blood at age 33-50.
    • 4999 participants had given a second blood sample 6 years later
  • Design: This was a nested case-control study, using a 1:3 case-control match. Matching was based on:
    • Date of birth
    • Date of collection of blood samples
    • Controls needed to be alive at follow up when case died
  • The primary outcome was metastasis or death from prostate cancer.
  • Median follow up was 27 years
  • Death from prostate cancer was determined by case note review and death certificates
  • There were 1369 clinical prostate cancers identified, 241 cases of metastasis, and 163 prostate cancer deaths.
  • 252 men with evidence of prostate cancer metastasis or death were matched 3:1 with controls
  • Archived blood plasma was anti-coagulated and PSA levels were ascertained for cases and controls.


  • PSA was strongly associated with the risk of prostate cancer death or metastasis up to 30 years later (AUC 0.70, 0.76 and 0.90 at age 44-50, 51-55 and 60 respectively, p<0.005).
  • Long-term risk of prostate cancer death by PSA between age 45-49 was as follows:
    • Patients with a PSA in the top 10%, corresponding to a PSA ? 1.6 in this age group, predicts for 44% of prostate cancer deaths with a median follow up of 27 years.
      • These men would benefit most from close surveillance.
    • Can certain men in this age group be exempt from further screening? No, as there is no acceptable cutpoint that can be identified. For example, 30% of metastases were in men with a PSA < 0.7 (median PSA in this age group).
  • Long-term risk of prostate cancer death by PSA between age 51-55:
    • In this age range, the top 10% of PSA values corresponds to a PSA ? 2.4, which would predict 45% of deaths.
    • Again, the question is whether certain men in this age group can be exempt from further screening, and again the answer was No. In this age group, 18% of metastases were in men with PSA < 0.9 (median PSA).
  • The authors had previously reported that men aged 60 with a PSA below the median (<1.1 ng/ml) were unlikely to develop clinically relevant prostate cancer (0.5% risk of DM).
  • The risk of prostate cancer metastasis by 15 years for men with PSA at median was never greater than 0.3% at any age.
  • For the highest risk patients (in the top PSA decile), the time frame from screening PSA to a sharp increase in the probability of metastases appears to be > 12 years.

Authors' Conclusions

  • Early PSA discriminates men at low risk from men at elevated risk of death from prostate cancer many years later.
  • A single PSA at age 45-49 or 51-55 is not sufficient to rule out the risk of metastasis or death at 25-30 years of follow-up.
  • However, combing these data with their previously published data in men age 60, they propose that 3 tests (mid-to-late 40s, early 50s, and age 60) would likely be sufficient to eliminate the men unlikely to benefit from more intense PSA screening.
    • Close to half of all deaths could be prevented by intense surveillance of a small proportion of men with the highest PSA levels (the top quintile) at age 44-50.
    • For men with a low PSA in this early age group (below the median), repeat testing at age 51-55 and age 60 would be sufficient to capture the risk of prostate cancer metastases or death more than 10 years in advance. This strategy would allow 50% of men to have only three lifetime PSA tests.
  • They propose trying to replicate these observations in an independent data set, with a more genetically diverse group of men.

Clinical/Scientific Implications

  • This is an interesting study that exploits a "natural experiment" to attempt to derive rationale guidelines for PSA screening for prostate cancer, in an attempt to reduce over-diagnosis and over-treatment associated with heavy PSA screening.
  • There are several strengths to the study including its large, unique longitudinal cohort, rigorous clinical follow-up, and the fact that these men were unscreened for prostate cancer, which means that they were all clinically diagnosed.
  • There are several limitations of the study, including:
    • The genetic homogeneity of the Swedish cohort, which limits its applicability to other groups of patients, for example African American men in the United States.
    • Furthermore, the screening guidelines proposed here do not incorporate information from digital rectal examination (DRE).
    • The endpoint used for the study, clinically diagnosed prostate cancer, represents very advanced disease, with a median PSA of 17.8 at diagnosis in the previously reported data (Vickers BMJ 2010).
  • Unfortunately, it is unlikely that this data can be replicated in the US population because of the frequency of PSA screening in this country.
  • In summary, this is an interesting study that provides some insights into the relationship between PSA and the natural history of clinical prostate cancer, but it is unlikely to change screening practices in the United States at this time.