LETTER TO THE EDITOR
An Extraordinarily Inaccurate Assay for Free
Testosterone Is Still with Us
To the editor:
Perusing the November 2000 issue of JCE&M, I noted an article in
which one of the outcome measures was serum free testosterone (1). The
method used to measure serum free testosterone, a direct RIA that
purports to, but does not, measure the free testosterone concentration is
one whose use I decried in a letter to the editor published more than 3
yr ago (2). Not long after, Vermeulen et al. (3) published a side-by-side
experimental comparison of methods for determining the concentration
of free testosterone in serum. Those experiments conclusively show that
the direct RIA of free testosterone is seriously inaccurate, underestimating
its concentration by many-fold. To my knowledge, there are no
data that contradict these conclusions. Hence, I was surprised to find that
the direct assay method not only was still being used for investigative
purposes, but also was being published in what is among the foremost
endocrine journals in the world.
To see how pervasive this situation was, I undertook to examine the
frequency with which those publishing in the journal were using this
methodology.
Methods. I conducted a full text, online search, of the JCE&M (January
1998-November 2000) using the term “free testosterone.” Then, I evaluated
each retrieved citation containing the term and ascertained
whether it revealed the use of a direct RIA for free testosterone, an
alternate method for measuring free testosterone, or was not applicable
(e.g. free testosterone mentioned in an editorial, a comment, a discussion,
a bibliography, etc.). In addition to the foregoing, there were a number
of publications in which no reference was given.
Results. A total of 116 citations were retrieved: 49 were not applicable;
11 contained no reference for the method cited; 24 used a method other
than direct RIA; and 32 used the direct RIA. Of the 116 citations, 67 were
suitable to address the question at hand. Thus, 48% (32 of 67) of the
applicable papers used a seriously inaccurate method for estimating free
testosterone and, almost as serious, 16% (11 of 67) cited no method at all.
Why would anyone choose to use this methodology? Perhaps the
answer can be found in the technical bulletins of one of the companies
that manufacture and sell kits that use this method. The relevant citations
reveal that kits made by two companies, Diagnostic Products and
Diagnostic Systems Laboratories, Inc. (DSL), account for almost all the
inappropriate measurements of free testosterone in papers published in
the journal during the period in question. A DSL technical bulletin
(
cPanel®. dslabs.com/techlit/4900tb2.doc) advertises: “Historically,
free testosterone levels were determined by a method known as
equilibrium dialysis. . . . The method is cumbersome, time-consuming,
and equipment intensive.” Conversely, we are told, the DSL direct
method is simple and rapid. The only difficulty is that “the equilibrium
dialysis method gave values approximately 4 times higher than did the
DSL kit.” As if to compensate for the inexcusable inaccuracy, the technical
bulletin and accompanying X-Y plot of RIA vs. equilibrium dialysis
makes the point that the correlation coefficient is 0.92. Even this is
deceiving. The points at the upper end of the DSL method are the major
contributors to the fitted line. Visual examination of the plot indicates
that a line through the lower points (17 of 21 points below the fitted line
and 2 points above it) would have a substantially different slope than
that indicated by the published fitted line. Thus, in addition to being
inaccurate, this observation indicates that, compared with equilibrium
dialysis, the assay is not linear. Yet, the kit remains on the market
because it is easy to use.
Almost one half of the publications dealing with free testosterone, in
the period under consideration, used an inaccurate assay for its measurement.
Even if the (somewhat more time-consuming) procedure of
equilibrium dialysis were the only alternative, the literature of science
ought not to use a method so grossly inaccurate when better ones exist.
We all know that there are numerous assays for hormones in serum that
are method specific. However, I know of no other that has been demonstrated
to be so egregiously incorrect. The journal might choose to
return manuscripts that use it without further evaluation to discourage
its use.
William Rosner
Department of Endocrinology
St. Luke’s-Roosevelt Hospital Center
New York, New York 10019References
1. Brown GA, Vukovich MD, Martini ER, et al. 2000 Endocrine responses to
chronic androstenedione intake in 30- to 56-year-old men. J Clin Endocrinol
Metab. 85:4074–4080.
2. Rosner W. 1997 Errors in the measurement of plasma free testosterone. J Clin
Endocrinol Metab. 82:2014–2015 (Letter).
3. Vermeulen A, Verdonck L, Kaufman JM. 1999 A critical evaluation of simple
methods for the estimation of free testosterone in serum. J Clin Endocrinol
Metab. 84:3666 –3672.
Received November 29, 2000. Address correspondence to: William
Rosner, M.D., Department of Endocrinology, St. Luke’s-Roosevelt Hospital
Center, 1000 Tenth Avenue, New York, New York 10019.
0021-972X/01/$03.00/0 Vol. 86, No. 6
The Journal of Clinical Endocrinology & Metabolism Printed in U.S.A.
Copyright © 2001 by The Endocrine Society
2903
Downloaded from jcem.endojournals.org by on January 6, 2007