Mm. Collins et al., EARLY DETECTION OF PROSTATE-CANCER - SERENDIPITY STRIKES AGAIN, JAMA, the journal of the American Medical Association, 278(18), 1997, pp. 1516-1519
An underappreciated characteristic of prostate cancer screening is tha
t it may detect some prostate cancers solely by serendipity or chance,
Serendipity, previously described in the detection of colonic neoplas
ms, could affect prostate cancer detection when a screening test resul
t is abnormal for reasons other than the presence of prostate cancer,
but prostate cancer is coincidentally detected during the subsequent e
valuation of the abnormal screening result. We reviewed published arti
cles about prostate cancer screening, searching for evidence of serend
ipity. We defined serendipity in digital rectal examination (DRE) scre
ening as the discovery of a prostate cancer by the random biopsy of an
area of the prostate gland other than the palpable suspicious area th
at prompted the biopsy. We defined serendipity in prostate-specific an
tigen (PSA) screening as the discovery of a prostate cancer by the ran
dom biopsy of a nonpalpable (stage T1c) prostate cancer less than 1.0
cm(3) in volume, since tumors less than 1.0 cm(3) are generally too sm
all to cause elevated PSA levels. We found that serendipity may be res
ponsible for the detection of more than one quarter of apparently DRE-
detected prostate cancers and up to one quarter of apparently PSA-dete
cted cancers. Additionally, serendipity played a larger role in the de
tection of smaller tumors that are common but of uncertain clinical si
gnificance. We conclude that serendipity-detected prostate cancers con
tribute to an overestimation of the true information value of DRE and
PSA screening. Whether serendipity is advantageous in prostate cancer
screening depends on the as yet uncertain outcomes for men with smalle
r prostate cancers. However, given our estimates of the potential magn
itude of the impact of serendipity, the currently popular DRE- and PSA
-based screening strategy may not be optimal. If smaller prostate canc
ers are important, then we are not finding enough; if they are unimpor
tant, then we are finding too many that we may feel compelled to treat
aggressively.