RESPONSE TO EDITORIAL COMMENTS

The critique of our work is appreciated, and theauthor makes some valid points, which I shall ad-dress. First, there is concern that we have changed theemphasis of the test from 95% specificity to 95% sensi-tivity. We chose to emphasize sensitivity (and, thus,cancer detection) for many of the reasons Dr. Littrupbrings up, e.g., increased mortality among AfricanAmerican men (AAM) from prostate cancer and thepossible increased biologic potential of prostate tu-mors in this cohort. He also points out that we cameup with a similar 95% sensitive value for Caucasianmen (CM). These values are shown in Table I of ourpaper.

The author states that the positive predictive value(PPV) would be a more interesting parameter, insofaras it is due to the higher incidence of prostate canceramong AAM. However, any cutoff values chosen forprostate specific antigen (PSA) will have a higher PPVamong AAM than CM (as Bayes’ Theorem predicts).Obviously, the PPV for any given value is a function ofthe number of true positives (TP) and false positives(FP). Of course, the values of TP and FP are in turndetermined by the sensitivity and specificity of a giventest (in this case, the cutoff values for PSA) and theincidence of the disease among the population tested.

As the author points out, focusing only on maxi-mizing the PPV would decrease screening costs (bydecreasing further testing), but at the cost of poten-tially increasing false negatives (missed cancers). Hethen proposes a compromise by using a low cutoffvalue for high-risk groups (which would emphasizesensitivity and cancer detection) and a higher cutoffvalue for lower risk groups (which would emphasize

specificity and decrease false positives). This we havedone by publishing age-specific ranges for AAM (ahigh-risk group) that emphasize sensitivity. Dr. Lit-trup is concerned that in calling these values age spe-cific we are misleading our readers, insofar as we areproposing test values that emphasize sensitivity ratherthan specificity.

His proposal that we use PSA values of 2 ng/ml orgreater for high-risk groups and 4 ng/ml or greater forlower risk groups, while disregarding age-related dif-ferences, has exactly the same potential of misleadingreaders who may not realize that the lower value em-phasizes sensitivity while the higher one emphasizesspecificity. We believe that our audience will under-stand that the word specific refers to age in this high-risk group, and that our values do not emphasizespecificity over sensitivity. Our published values havethe advantage of including age as a variable. The de-cision to screen for prostate cancer is still controversialand screening should be discussed on an individualbasis. We believe that physicians will appreciate age-specific values to tailor their screening methods asmuch as possible to the individual patient.

David G. McLeod, MD, JD, FACSChief, Urology ServiceDepartment of Surgery

Walter Reed Army Medical CenterWashington, DC

Uniformed Services University ofthe Health Sciences

Bethesda, MD

The Prostate 31:142 (1997)

© 1997 Wiley-Liss, Inc.