Interpreting The
Results Of PSA Testing (March 2004)
(Data assistance from Dr. David Corwin, Director, Clinical Division,
Dynacare Laboratory, Swedish Hospital, Seattle)
Like Heisenberg's
electrons, PSA test results are subject to a bit of uncertainty. A
clinician's desire is that the test consistently and accurately
represent the serum PSA level proportional to the aggregate mass of
prostatic tissue, benign and malignant, so that he can make valid
comparisons among sequential tests. Consider a case in point:
postoperative values of .1 ng/mL on 9/03; .3 on 12/03; .3 on 1/04 and .6
on 2/04. Is the accuracy of the test in this low range sufficiently
accurate to permit the calculation of a reliable PSA doubling time
(actual calculation: 2.13 months)? The Dynacare Laboratory at Swedish
Hospital, using the Beckman Access II test (traceable to the Hybritech
method), reports an analytic imprecision (% coefficient of variation at
the 95% level) of approximately 8% for PSA values in the range of .07 to
1.0 ng/mL, and <5% for values in the range 1.0 to 10 ng/mL. As the PSA
values become considerably higher the figure increases from 5% to 8%.
Using a coefficient of
variation of 8%, the fourth test report above could be reported .6 ng/mL
(with a .05 to .07 ng/mL 95% confidence interval) This is sufficiently
accurate for a reliable PSA doubling time calculation. Considering that
an important clinical decision might be made based on this calculation
of PSA doubling time, and also considering that the .6 ng/mL value is
the major element affecting the rapid DT figure, a re-measurement of
this crucial value would be prudent.
The circadian variation
of PSA is minimal to the point of being clinically insignificant so
intra day fluctuation doesn't complicate interpretation. One
investigator found little change in the PSA values of 14 men tested q 4
hours over 32 hours. (Serum testosterone, however, has a marked
circadian rhythm: highest in the early AM and lowest in the late
evening, but this is muted in older age.) When a clinician compares two
PSA values separated by many days, a compound variation representing
both analytic and biologic variations must be taken into account,
and this combined variance at minimum is in the range of 11 to 12%.
However, the biologic fluctuation over days varies considerably. The
Price review (see reference below) gives an 18% figure for the total
variance, analytical plus biological, based on an average of 5 studies
of men without cancer. Other factors that affect the PSA value include
the effect of DRE, which causes a probably insignificant variation of .4
ng/mL. or less, and ejaculation, prostate instrumentation and, of
course, prostatitis. Considering that the half life of PSA is
approximately 2 to 3 days (2.3 days) and the half life of free PSA, 110
minutes, sufficient time should be allowed for the perturbations to
return to baseline. It is suggested that 6 weeks should be allowed
before measuring the PSA following a prostate biopsy or TURP.
Eastham et al.
presented cautionary information in the article, "Variation of Serum
Prostate-Specific Antigen Levels: An Evaluation of Year-to-Year
Fluctuations in a Well Defined Cohort of Men", Am J Urologic Review, Jan
2004. They analyzed the yearly PSA values of 972 apparently healthy men
(median age 62) participating in a colonic polyp evaluation study. Each
man was tested yearly for more than five years and the test results were
referenced against five different standard criteria of PSA abnormality,
ones which often provide guidance as to the need for a biopsy: PSA > 4;
PSA > 2.5; age-specific PSA; and free PSA ratio, < 25%; and PSA
velocity, >.75 ng/mL/year. Result: during the course of the study 361
men (37%) had an abnormal test with regard to one of these criterion and
of this group, regardless of which PSA criterion was exceeded,
approximately 30% had a normal value for that marker at the next test (1
year). The mean PSA values for PSA in the age group 50 - 59 was 1.5; 60
- 69, 2.3; and 70-79, 2.8 ng/mL.
The establishment of
age-specific PSA ranges is a important contribution to a more focused
interpretation of the test. The most commonly quoted data comes from
Oesterling et al. (JAMA 1993 Aug 18) based on 537 randomly chosen
apparently healthy men seen at the Mayo Clinic and evaluated by a
detailed clinical examination that included a PSA test, a DRE, and a
prostate U/S. They wrote, "For a healthy 60 year old man with no
evidence of prostate cancer, the serum PSA increased by approximately
3.2% per year (0.04 ng/mL per year)". Their well known guidelines for
the upper limits of the "normal" PSA range for various age groupings of
white men are: age 40 to 49, < 2.5; 50 to 59 < 3.5; 60 to 69 < 4,5; 70
to 79, < 6.5. However, certainty in this area remains a rare commodity.
Setting the upper limits for an age group involves an arbitrary decision
as what sensitivity to employ. A 99% sensitivity will embrace all but 1%
of the "normals". But by reducing the sensitivity to the 97.5th centile
the "normal" upper levels are different: age 40-50, <1.81; 50-60, <3.36;
60-70, <6.16; and 70-80, <4.33. At the 95% percentile of normal
reference values, the upper limit values are 1.3, 1.6, 2.6, 5.6 for the
same age "bins". The age "bin" can be changed, i.e. for age group 45-55
the upper limit is <2.93. The average PSA values are, of course,
considerably lower: for age 41-50, 1.2; 51-60, 1.7; 61-70, 2.2; and for
71-80, 3.8. Partin (J Urol 1996 Apr) pointed out that by adherence to
the suggested upper limits of age-specific PSA ranges, prostate cancer
detection for men older than 60 years was decreaed by 22%.
This entire topic is
extensively examined in the review article by Price et al., "Pre- and
post analytical factors that may influence use of serum prostate
specific antigen and its isoforms in a screening program for prostate
cancer", Ann Clin Biochem 2001: 38: 188-216.
Bottom Line:
To successfully accommodate the uncertainties associated with the
interpretation of the PSA test a clinician needs sufficient information,
... and a good measure of clinical
judgment.
«
Back to Article List |
