The validity of PSA as a screening test

Swedish research studies have long been on the “cutting edge” for understanding issues related to screening and early stage tretament for prostate cancer. It is possible that a new study by Holmström et al. may prove to be no exception to this historic experience..

In light of recent prostate cancer screening trial results, Holmström and colleagues have attempted to evaluate whether, in a recent screening population in Umeå, Sweden, the PSA test actually attains standards of sufficient validity for use as a prostate cancer screening tool.

The study population comprised 540 cases and 1,034 controls matched for age and date of blood draw. The validity of the PSA test as a means to predict subsequent prostate cancer diagnosis was assessed by comparing the patients’ PSA record to subsequent diagnosis of prostate cancer through cancer registry data

The study showed the following results:

  • On average, blood samples were drawn 7.1 ± 3.7 years before diagnosis.
  • The area under the curve for PSA was 0.84 (95% confidence interval 0.82 to 0.86).
  • At PSA cut-off values of 3, 4, and 5 ng/ml, sensitivity estimates were 59, 44, and 33 percent, and specificity estimates were 87, 92, and 95 percent, respectively.
  • The positive likelihood ratio commonly considered to “rule in disease” is 10; however, in this study the positive likelihood ratios were 4.5, 5.5, and 6.4 for PSA cut-off values of 3, 4, and 5 ng/ml.
  • The negative likelihood ratio commonly considered to “rule out disease” is 0.1; in this study the negative likelihood ratios were 0.47, 0.61, and 0.70 for PSA cut-off values of 3, 4, and 5 ng/ml. For a cut-off of 1.0 ng/ml, the negative likelihood ratio was 0.08.

The authors conclude that “No single cut-off value for [PSA] concentration attained likelihood ratios formally required for a screening test.”

Now this is a paper that requires a sophisticated appreciation of statistical methodology to be able to assess its importance. We do not pretend to have that degree of statisticalo sophistication … but, it would seem clear to us that PSA testing is (as Stamey suggested some years ago) no longer as useful as it may initially have been in predicting risk for clinically significant prostate cancer. Even though the specificity of the PSA test for prostate cancer still seems to be reasonably high at PSA levels between 3 and 5 ng/ml, the sensitivity of the test is poor, and PSA values below 1.0 ng/ml “virtually ruled out a prostate cancer diagnosis during the follow-up.”

This group of authors state very clearly that, “Additional biomarkers for early detection of prostate cancer are needed before population based screening for prostate cancer should be introduced.” This does not mean that we should not continue to use the PSA tests until we have better biomarkers, but it does suggest that as a screening tool for large numbers of otherwise healthy men at no specific risk for prostate cancer, the PSA test is far from accurate.

One Response

  1. Holmström et al. define a screening test as a test for which at least LR+ = 10 and LR- = 0.1. Since OR (odds ratio) = LR+/LR-, such a test has a degree of association of at least OR = 100. In a rapid response to the article, I justified the calculation that OR = 12.25 approximately. The divergence between the required OR = 100 and OR = 12.25 is so big that we can conclude safely that the PSA test is not a screening test.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: