ever, is that the qualitative hCG assay
with the greatest sensitivity may not
be necessary, or even desirable, in all
clinical situations (2). For example, in
the general population, 31% of pregnancies recognized by the most-sensitive assays of urine hCG will spontaneously abort before they become recognizable clinically (5). In essence, a
maximally
sensitive qualitative
hCG
assay, used as a pregnancy test, will
have a 31% clinical false-positive rate.
One solution to this problem is for the
laboratory to report very low hCG concentrations as “borderline,” which is
possible with the Tandem Icon H. This
assay incorporates a positive reference
zone, the intensity of color development in this reference zone being
equivalent to a sample containing
hCG at 50
units/L (Third International Standard or, equivalently, First
International Reference Preparation).
Thus, the users of the Tandem Icon H
may report results as “borderline” if
they see color development that is less
intense than that of the positive reference.
mt.
References
1. ChristensenH, ThyssenHH, Schebye0,
Berget A. Three highly sensitive “bedside”
serum and urine tests for pregnancy compared. Clin Chem 1990;36:1686-8.
2. EmancipatorK, CadoffEM, Burke MD.
Analytical versus clinical sensitivity and
specificity in pregnancy testing. Am J
Obstet Gynecol 1988;158:613-6.
3. Barnhart ER. Physician’s desk reference, 44th ed. Oradell, NJ: Medical Economics
Co., Inc., 1990:1770-2.
4. Zweig MH, Csako G, Spero M. Escape
from blockade of interfering heterophile
antibodies in a two-site immunoradiometnc assay for thyrotropin.
Clin Chem
1988;34:2589-91.
5. Wilcox AJ, Weinberg CR, O’ConnorJF,
et al. Incidence of early loss of pregnancy.
N Engl J Med 1988;319:189-94.
Kenneth Emancipator
Clin. Pat/wi. Dept.
Warren G. Magnuson Clinical
National Institutes of Health
Building 10, Room 2C407
Bethesda, MD 20892
Center
Blbllometrics and Clinical
Chemistry
To the Editor:
The term bibliometrics, proposed by
Pritchard (1) in 1969, was defined as
“the application of mathematics and
statistical methods to books and other
media of communication.” Bibliometrical techniques are being increasingly used to study scientific produc-
tion. Within the realm of analytical
bibliometry, appraisal of the dispersion of writings dealing with a certain
subject among different publications is
one of the most interesting applications and constitutes a necessary step
for the rational planning of specialized
libraries to ensure that the most suitable bibliographic sources are being
selected to gather information for a
given discipline or topic area.
Bradford, in 1934 (2) and later (3),
postulated the law that now bears his
name: If scientific journals
are arranged in order of decreasing productivity of articles on a given subject, they
may be divided into a nucleus of periodicals more particularly
devoted to
the subject and several groups or zones
containing the same number of articles
as the nucleus, when the numbers of
periodicals in the nucleus and succeeding zones will be as 1:n:n2....
We applied Bradford’s method to the
general topic of total quality in the
laboratory, presenting the results in a
meeting held in Madrid (“Total quality in laboratory tests,” May 28-29,
1990), with the intention of showing
the audience the infrastructure
of the
medical literature dealing with this
subject; some of our findings are discussed here. By “total quality,” we
mean the continuous process of (a)
defining the quality needed (i.e., setting analytical goals), (b) creating the
required quality (evaluation of methods), and (c) controlling routinely such
quality (quality-control procedures).
We explored the Medline database
through
a CD-ROM (Compact Disk,
Read-Only-Memory)
system. Medline,
produced by the U.S. National Library
of Medicine (NLM), includes references from >3000 journals worldwide,
incorporating >300 000 new referenceseach year (4). Our search, for the
years 1986-89, was performed according to a strategy in which we used the
medical subject headings terms “Diagnosis, laboratory” (A), “Quality control” (B), and “Reference Standards”
(C), along with subheading “Standards” (D). We retrieved all the citations located under “Diagnosis, laboratory” and all its descendants. (This
is what is called, among documentalists, a data explosion.) Afterwards, we
combined these terms, using the boolean logical operators AND (*) and OR
(+), with the searching equation being
A*(B+C+D).
Thus, only those papers
included under heading A and also
under any of headings B, C, or D were
retrieved.
Using such a strategy, we retrieved
2297 papers published in 626 different
journals (by “paper” we refer to all
types of contributions to journals:
ticles, editorials,
correspondence,
aretc.).
We then used a flexible approach to
study the dispersion of these data,
trying to obtain a Bradford model with
minimal
nucleus, maximal number of
zones, and maximal homogeneity of
Bradford’s constants (the quotient of
the number of journals in any given
zone and the preceding one). This was
attained by distinguishing eight zones
in our literature sample, the first zone
(nucleus) being integrated only by one
journal,
Clinical Chemistry, with 266
papers on the subject, i.e., 11.58% of
the total. Given that Clinical Chemistry published 3055 papers in this period of time, those recovered by our
search strategy represent 8.7% of all
papers printed in this journal.
The second zone, with 252 papers
(10.97% of total), was integrated by
five specialized periodicals: Journal
of
Immunological
Methods (63 papers),
American Journal of Clinical Pathology (51), Acta Cytologica (49), Rinsho
Byori (46), and Annals of Clinical Biochemistry (43). Only in the third area,
with 248 papers and integrated by
eight journals, did the first nonspecialized general journal appear (The Lancet), with 41 papers in the defined
area. Bradford’s law was only partially obeyed, because of the high
number of papers published in a single
journal (Clinical Chemistry). Thus, the
first
Bradford’s constant was 5,
whereas the rest of them were more
similar:
1.6, 2, 1.75, 2.25, 2.31, and
2.45, with a mean of 2.06 and a CV of
16.3%. No other distribution in zones
gave a more homogeneous set of constants, with a smaller CV.
In conclusion, Bradford’s procedure
or any of its reformulations may be
conveniently used to study the dispersion of medical literature. In the area
that we defined and explored, a high
percentage of papers (11.58%) was
published in Clinical
Chemistry,
which equals by itself the output of the
next five specialized journals. Thus for
scientists interested in any special aspect of those areas included in what
we have delineated as “total quality,”
the primary place to scan for references may be this journal, wherein
they will find almost one-eighth of all
papers written on that topic indexed
in Medline. As many readers and potential readers have probably realized,
Clinical Chemistry also appears to be a
highly commendable journal to publish papers dealing with any of the
subjects that we have included in the
defined area of total quality.
CLINICALCHEMISTRY, Vol. 37, No. 2, 1991 303
clonal antibodies directed against tissue-nonspecific, intestinal, and placental ALP; by addition of neuraminidase and phosphatidylinositol-specific
phospholipase C; and by heating at
65 #{176}C.
We isolated high-Mi ALP (liver
plasma membrane fragments) (5) from
pooled human sera by gel filtration on
Sephacryl S-300 (Pharmacia, Uppsala, Sweden). Plasma obtained by
plasmapheresis was used to isolate intestinal variant ALP by chromatography on octyl-Sepharose. Intestinal
ALP was partially purified from human jejunal microvilli by pseudo-affinity chromatography on Reactive
Yellow 13-Sepharose.
Using the Resolve-ALP IEF system,
Griffiths and Black (3) initially described isoforms of tissue-nonspecific
ALP (at least seven bands with p1 3.0
to 4.45) and two placental ALP isoforms (pI 4.65 and 4.73), and subsequently identified the band at p1 4.65
to be of germ cell origin (6). We could
confirm the tissue-nonspecific nature
of the five to six anodally migrating
bands with p1 3.0 to 4.21. Intestinal
ALP, being the most cathodal, was
referred to as “band 12” (3). High-Mr
ALP (“fast liver” or “biliary” ALP),
the relevant fraction encountered in
cholestasis (5, 7), was not described,
limiting the clinical utility of that IEF
technique.
When electrofocusing samples with
high activities of high-Me ALP, we
observed a “smudged” area around p1
4.6 to 5.1, caused by the lack of focusing of the high-M,. material in the
region where the sample was applied
(arectangularspaceof8
x l2mmin
the plastic applicator strip) (Figure
la). Treatment with phospholipase C
References
1. Pritchard A. Statistical bibliographyor
bibliometnics?J Doc 1969;25:348-9.
2. Bradford SC. Sources of information on
specific subjects.
Engineering 1934;137:
85-6.
3. Bradford SC. Documentation. London:
Crosby Lockwood, 1948.
4. Wertz RK. CD-ROM. A new advancein
medical information retrieval. J Am Med
Assoc 1986;256:3376-8.
Concepcion Campos
Medical Library
Francisco L. Redondo
Dept. of Clin. Chem.
Hospital Central de Cruz Roja
28003 Madrid,
Spain
Identification of Intestinal, Intestinal
Variant, and High-Me Alkaline
Phosphatase with the Resolve-ALP
isoelectric Focusing System
To the Editor:
Zonal electrophoresis of alkaline
phosphatase (ALP, EC 3.1.3.1) is characterized by major overlap between
the fractions. In contrast, isoelectric
focusing (IEF) separates ALP into 10
to 20 distinct bands (1, 2); not all of
them, however, have been unequivocally identified.
We performed IEF of ALP with the
Resolve-ALP system (Isolab, Akron,
OH) (3) on samples with a known
isoenzyme pattern, obtained by agaroseelectrophoresis with the Isopal system (Beckman, Analis, Namur, Be!gium) (4). The fractions were identified as described previously (4), by
treatment with polyclonal and mono-
(EC 3.4.1.3) removed the high-Me
smear almost completely, with a concomitant increase in the two most
acidic fractions (p13.0 and 3.1) (Figure
la). This was confirmed with an assay
of isolated high-Mr ALP (Figure ib).
Treatment with phospholipase C apparently released dimeric liver ALP
from the high-Me particles (8), which
are formed by “shedding” of the sinusoidal liver membrane (5). The two
acidic fractions probably correspond to
the ALP isoform described by Griffiths
as “band 1/biliary canaliculus band”
in patients with cholestasis (9).
We found three bands with intestinal ALP activity (p1 4.77, 4.85, and
5.04). Pre-incubation with phospholipase C did not alter their migration.
The two most anodal intestinal ALP
bands migrated close to placental ALP
(p1 4.65 and 4.73) (3). Correct identification, therefore, required use of
monoclonal antibodies and heat treatment at 65#{176}C.
We previously reported
that intestinal ALP was present in
human serum as a larger, hydrophobic
form (binding on octyl-Sepharose),
which we called intestinal variant
ALP, and as a smaller soluble form (4,
10). Isolated intestinal ALP migrated
as a single band (p1 4.85); isolated
intestinal variant ALP migrated as
two bands (p1 4.77 and 5.04).
On the basis of these observations,
we propose a scheme of the ALP migration pattern obtained with the Resolve-ALP system, complementary to
the one published by Griffiths and
Black (3) (Figure lc).
Of particular clininal interest is
high-Mr ALP, which in this system
may mask other relevant ALP fractions, particularly placental and intes-
N1)
C
0
p1 o
(.4
(‘a
+
ace
aio
3.65
01
388
TU-ALP
4.10
2R
&21
L4H
±4.6
I
bi
4.65
1PLAP*
4.73
4 77
2
4.86
IAP*
685
504
IAP+
}lAPVar
±5.11
J
IHIGH-Mr
ALP
Fig. 1. Isoelectricfocusingwiththe Resolve-ALPsystem:high-?4 ALP fromhumanserum(a), and isolatedhigh-M,ALP (b) (lane 1,untreated
sample; lane 2, sampletreatedwithphospholipaseC); (C) p1schemeof the migration of tissue-nonspecific ALP (TU-ALP), intestinalALP (lAP),
and high-M,ALP with the Resolve-ALPsystem
migration according to Griffiths
304
and Black(3, ;
lAP-Var, intestinal variant ALP; PLAP,placental ALP
CLINICAL CHEMISTRY, Vol. 37, No. 2, 1991