The Effects of Radioisotopes Used in Nuclear Medicine
on Diagnostic Radioimmunoassay Testing
Is There Any Significant Interference?
JOHN A. RICCIO, M.D., DIANE MATURANI, B.S., MT(ASCP), JOHN WRIGHT, PH.D.,
AND MILDRED K. FLEETWOOD, PH.D.
The administration of radioisotopes for diagnostic nuclear medicine scans and therapeutic procedures is quite prevalent today.
A period of interference with the counting of a radioimmunoassay
[RIA| test may occur with the serum of a patient receiving an in
vivo radionuclide that decays by gamma emission. Because the
logistics of precounting all specimens may be cumbersome and
prohibitive, it is important to determine the degree of this interference. In this study, the authors evaluate the potential interference of the most commonly used radioisotopes with RIA
studies. For two months (March and August 1988) 10,650 patient
serum specimens were counted for significant background gamma
radiation before RIA testing. Forty-three patients, on whom 105
RIA tests were performed, were identified as having preassay
gamma radiation in their serum. With the use of selective energy
windows for each different interfering radionuclide, proportional
determinations were made as to the amount of interfering gamma
radiation spilling into the iodine 125 test marker window. It was
shown that initial whole serum pretest gamma counts as high as
111,000 counts/minute did not significantly affect the results of
the RIA. Because of the meticulous washing and decanting procedures required in modern RIA and the monoclonal nature of
most antibodies used currently, it appears the degree of nonspecific binding of this potentially interfering radiation is minuscule.
The energy level of cobalt 57, however, and many of the other
commonly used radioisotopes, overlaps so closely that it is difficult to window for this interference. It is possible, therefore,
that this distinction cannot be made and folate and vitamin B12
test systems using cobalt 57 markers may have to be routinely
prescreened. The authors conclude that the requirement for
prescreening of all RIA test samples for interfering gamma radiation is unnecessary (1987 CAP Commission on Laboratory
Accreditation Inspection Checklist, Section VII, Question
07.0290). (Key words: Radioimmunoassay testing; Nuclear
medicine; Radioisotopes; Interference) Am J Clin Pathol
1990;94:618-623
IN VITRO DIAGNOSTIC radioimmunoassay (RIA) has
been an accurate and reliable method of determining even
picogram serum levels of many clinically important substances. Since the development of adequately purified,
Received October 24, 1989; received revised manuscript and accepted
for publication April 6, 1990.
Address reprint requests to Dr. Riccio: Department of Pathology and
Laboratory Medicine (01-31), Geisinger Medical Center, North Academy
Avenue, Danville, Pennsylvania 17822.
Departments of Pathology and Laboratory Medicine and
Nuclear Medicine, Geisinger Medical Center,
Danville, Pennsylvania
radioactively labeled proteins, and the technical development of large-capacity, computerized counting equip-,
ment during the 1960s, broad clinical application has been
accomplished.7 This method has occasionally been reported to be susceptible to interference, usually resulting
from antibodies to animal immunoglobulins (acquired
heterophilic antibodies) 1,4 ' 910 and less commonly from
other causes such as the nephrotic syndrome8 and immunoreactive proteins mimicking substances of clinical
significance, such as prostatic acid phosphatase in one
recent report.3 Another source of potential interference
that has been postulated since the early days of RIA is
the presence of radionuclides exogenously administered
for therapeutic and/or diagnostic nuclear medicine procedures. This extraneous radioactivity could be encountered in serum drawn for radioassay, for varying periods
of time, depending on physical and biologic half-lives
of these radiopharmaceuticals. Logically, substantial
amounts of this potentially interfering radioactivity may
give rise to erroneous results, because in RIA quantitation
of the analyte is a function of radiation counts. One recent
ongoing study determined that extraneous isotopes did
not significantly affect RIA results.6 We present a study
examining the frequency of detection of radioactive patient serum samples and whether this radioactivity significantly affects the radioassay results.
Methods
All patient serum samples received in the RIA laboratory of the Geisinger Medical Center for a two-month
period (March and August 1988) were counted for gamma
radiation before RIA analysis. This encompassed 10,650
patient serum samples, among which 43 patients (on
whom 105 RIA tests were performed) were identified as
618
Vol. 94 • No. 5
DIAGNOSTIC RADIOISOTOPE EFFECT ON RIA TESTING
619
Table 1. Characteristics of Frequently Used Radioisotopes
Test Isotopes
Major Range of
Gamma Emission
(keV)
Major Peak
Gamma Emission
(keV)
Half-Life
of Isotope
Study Isotope
Window
(keV)
Iodine 131
Gallium 67
Thallium 201
Technetium 99
Cobalt 57
Iodine 125
80.2-637
93.3-388
135-167
140
114-136
27-36
364.5
93.3
167
140
122
35.5
8.05 days
78.3 hours
73.1 hours
6.0 hours
270 days
60 days
125-200
90-300
130-170
130-150
110-150
15-80
having preassay gamma radiation present. This group of
patients constituted the study group. All gamma counts
were performed on a GENESYS®, 5000 series, dualchannel, 25-well sodium iodine crystal scintillation counter (Laboratory Technology, Inc., Schaumburg, IL). These
study samples were considered to be "contaminated" with
gamma radiation if initial serum sample counts were
greater than 100 counts/minute, regardless of volume, on
the iodine 125 test marker energy window (15-80 keV).
We then determined through the ordering clinician, nuclear medicine department, or medical record review
which radionuclides were responsible for this contamination. Five different radionuclides were found to be contaminants and encompassed the most commonly used
radioisotopes in nuclear medicine. These radioisotopes
and their characteristics are noted in Table 1.
The second differential energy windows used to quantitate the contaminating radioisotopes were determined
by examining published spectral analyses of each radioisotope in question." In nature, radionuclides give specific
unique gamma emission spectra with little or no overlap
between different radioisotopes. With the use of scintillation gamma counting, however, these finite spectral lines
are detected as pulse height spectral curves with a defined
portion of photon energy represented usually at lower energy levels, theoretically causing interfering spurious
gamma counts (Fig. 1).
To quantitate these potential interfering counts, we determined the predictable proportional relationship between the two separate energy windows. As noted in Figure 2, if element A is a contaminating radionuclide with
a peak occurring in a higher energy window (El), there
would be a proportionately predictable back scatter into
the lower energy window (E2). Energy window E2 is used
to detect the labeling radionuclide in the RIA test. The
peak energy at point P presents back scatter and potential
interference into a lower energy level. When two radionuclides are present in an assay system, a curve such as
A+B exists with a theoretically spuriously higher peak in
the E2 window represented by point R. Q represents the
peak energy level of the interfering radioisotope. Mathematically then, this relationship is expressed as follows,
using this generic formula for correcting for any interfering
radioisotope:
True counts of I125 = R - [Q(P/Q)]
Once this proportion is determined for a given radionuclide, using the same differential energy windows, it should
remain constant. To determine this ratio, a pure radioisotope test was performed. Known quantities of the contaminating radioisotopes were counted in the prescribed
windows and then in the iodine 125 window as noted in
Table 2.
|lodln» - 1 3 1 |
83%
Percent
01
!
Gamma
Emlealon
Counts
par
KEV
Emission Spectrum (Top) vs Gamma Scintillation
Counted Total Absorption Peak (Bottom)
FlG. 1. Spectral analysis of pure Iodine 131 (upper) represented as
finite spectral lines. Stylized gamma scintillation counting of the same
radioisotope yields a spectral curve, as noted in bottom corresponding
figure. (Redrawn with permission, from Wagner.")
RICCIO ET AL.
620
/
A.J.C.P. • November 1990
Emrgy wlndo*
c i n t i r a d on
miln p u k s
— —' — Elaminl A • B
togithir
Gamma
Counts
A his p u k i t E y
B h n p u k i t E2
FIG. 2. Theoretic bimodal gamma energy
peaks present in a spectral curve when an
exogenous radioisotope contaminates an
RIA assay system. The lower energy isotope
is the Iodine 125 radiolabel present in the
RIA kit used as a marker.
Energy
Once the contaminated serum samples were identified,
they were run through the RIA test system according to
standard protocol. The serum test aliquot, which ranged
in size from 10 to 200 /xL, was counted in both windows
before performance of the RIA procedure and again after
completion of the assay. With the use of the mathematical
calculation above and the appropriate derived proportions, potential interfering gamma counts were calculated
(Table 3). During the data-reduction phase of the assay,
these calculated interfering gamma counts were subtracted
from the total radiolabel counts to determine whether
any significant alteration of the results would occur.
The RIA commercial test kits used in this study included the following: Gamma Coat® Total T4 and T3
Uptake Kits (Baxter/Dade Healthcare Corporation and
Travenol Diagnostic, Inc., Cambridge, MA); IRMACount TSH®, FSH Double Antibody®, LH Double Antibody®, Coat-A-Count® Total Testosterone, and CoatA-Count Free Testosterone Kits (Diagnostic Products
Corporation, Los Angeles, CA); Amerlex-M® Free T3 and
Amerlex-M Beta HCG RIA Kits (Amersham International, Arlington Heights, IL); Prolactin [l25I]® and
Simltrac-S® Solid Phase Vitamin B-12 and Folate RIA
Kits (Becton Dickinson Immunodiagnostics, Orangeburg,
NY); Allegro® Intact PTH (Nicholas Institute Diagnostics, San Juan Capistrano, CA).
We believe that these kits are representative of the type
used in modern RIA laboratories. In all cases, the bound
phase was counted in these tests. These phases consisted
of solid-phase matrices (coated tubes, latex beads, magnetic particles) with the exception of double-antibody
precipitated techniques or binding protein techniques in
which, once again, the bound phase was counted. The
numbers of tests performed are presented in Table 4.
Negative controls were run with the use of randomly
selected patient serum samples containing no preassay
contaminating radiation. Six hundred seventy RIA tests
were performed according to standard protocol and a final
dual determination of the counted phase was accomplished in both the iodine 125 marker window and the
various contaminating radioisotope windows. One
hundred thirty-five negative controls were tested for thallium 201, 113 for iodine 131, 80 for technetium 99, 92
for cobalt 57, and 250 for gallium 67. The counts for
these control tests were compared with background counts
of empty wells at each respective energy level.
Table 2. Pure Radioisotope Test
Test Isotope
(test dose)
Test Isotope
Window Counts
(cpm)
Iodine 125
Window Counts
(cpm)
Calculated Ratio
of Backscatter
Through ,25I
Window
Iodine 131 (1.3 MCi)
Gallium 67 (1.0 jiCi)
Thallium 201 (1.7 MCi)
Technicium 99 (1.5 /iCi)
699,920
929,801
354,242
668,941
216,079
62,528
779,755
122,680
0.31
0.07
2.20
0.18
Vol. 94 • No. 5
DIAGNOSTIC RADIOISOTOPE EFFECT ON RIA TESTING
621
Table 3. Results of Radioactively Contaminated RIA Tests
RIA
Tests
Isotope (n)t
(n)t
Iodine 131 (13)
Thallium 201 (17)
Technicium 99 (11)
Gallium 67 (2)
32
43
21
9
Contaminated
Whole Serum
Sample (cpm)*
31,013
3,280
11,405
498
±42,055
± 6,282
± 19,069
± 575
Test Aliquots Prior to RIA
(cpm)*
125
I Window
1,415
121
854
30
±2,875
±313
± 3,064
±33
Test Media after RIA
(cpm)*
,25
Other
2,867
60
305
682
± 7,028
±57
± 892
± 463
I Window
9,886
8,645
8,349
9,894
± 11,659
±6,061
± 4,533
± 12,292
Other
Calculated
Contaminating
(cpm)*
12 ± 2 4
2 ±3.7
1 ± 1.6
0
4 ±7.5
4 ±7.8
0
0
X (n) = number of RIA tests of different types in each isotope group.
* Figures shown are mean values ± SD for each group of tests,
t (n) = number of patients receiving each radioisotope.
Results
During the two-month period encompassing this study,
10,650 patient serum samples were counted for extraneous
gamma radiation before RIA testing. Forty-three patient
samples, requiring 105 RIA tests, were identified. The
contaminated whole serum patient samples ranged from
1 to 5 mL (0.001-0.005 L), and the amount of contaminating gamma radiation is noted in Table 3. These spurious counts ranged from 109 to 111,356. The average
initial contaminating counts ranged from 498 for gallium
67 to just more than 31,000 for iodine 131. The average
spurious counts in the test aliquots, however, were much
less. These test aliquots ranged in size from 10 to 200 nL,
and the average contaminating count ranged from 30 to
just more than 1,400 counts per minute (cpm) on the
iodine 125 test window. The counts before assay in the
contaminating isotope windows ranged from 60 to 2,867.
After completion of the RIA tests, these counts were significantly different. As expected, the counts on the iodine
125 test label window ranged between 8,000 and 10,000
cpm because this is the counted phase of the radioassay.
The mean counts in the contaminating radioisotope windows were from 0 to 12, with an individual count range
from 0 to 113. On calculating the amount of these spurious
counts that would have occurred in the iodine 125 test
window, this range was from 0 to 35 counts, with mean
counts of only 0 to 4.
The Student's Mest of statistical significance for paired
data on the differences of the assay counts with and without the contaminating spurious isotope counts was performed. It was shown that there was no significant statistical difference between the assay counts before and after
the subtraction of the spurious, contaminating radioisotope counts. In the case of iodine 131, this was significant
to a P value of less than 0.01. For thallium 201, the P
value was less than 0.002. For technetium 99 the P value
was less than 0.05, and for gallium 67 the P value was
less than 0.0001.
To support the assumption that the spurious counts
occurring in the contaminating isotope windows were
from parenteral radiopharmaceuticals, and no other
source, statistical ^-testing was also performed on the negative control group as compared to background "noise"
counts obtained from empty wells. Statistical analysis here
showed that there was no statistical difference between
the negative controls and the background counts in the
empty wells, with a P value of 0.05 for iodine 131, 0.025
for thallium 201, 0.025 for technetium, and 0.0002 for
gallium.
One problem encountered was that of the simultaneous
vitamin B12 and folate RIA tests. These tests use a cobalt
Table 4. Types of RIA Tests Performed in Presence of Potentially Interferring Isotope
RIA Test
Iod ine '.131
Thallium 201
Technetium 99
Gallium 67
Total
T4
T 3 resin uptake
TSH
FSH
LH
•Testosterone
Free T 3
B-HCG
B12/folate
PTH
Prolactin
Total
8
5
10
10
8
13
2
3
4
1
2
2
2
2
1
1
22
18
29
1
1
3
10
3
5
12
1
105
• Testosterone = total and free.
1
4
5
5
3
3
3
2
4
1
622
RICCIO ET AL.
57 radioisotope assay label in addition to an iodine 125
label. The range of counts on the iodine 125 window in
these tests was from 3,300 to 6,181, and on the interfering
isotope window a range of 2,548 to 2,944 was obtained.
However, the cobalt 57 radiolabel used in the cyanocobalamin (vitamin B12) test overlaps portions of all of the
interfering radioisotope energy windows.
Discussion
Different radionuclides have unique gamma emission
spectra characterized by finite, narrow energy peaks.
However, gamma radiation detection instruments commonly in use in clinical laboratories inherently detect and
represent these distinct energy peaks as continuous energy
curves (Fig. 1). This distortion of a spectral line into a
hump or peak results from the inherent nature of resolution of these detection systems. Many different processes
cause photon interaction and scattering represented by a
scintillation counter as back scatter into lower energy
windows. This distortion is a function of numerous factors, including a small loss of scintillation light before
entering the phototube, nonuniform sensitivity of the
photocathode, statistical variation in the electron path
from dynode to dynode, bremsstrahlung effect, and, more
importantly, a mathematically predictable Compton
scattering with concomitant energy escape from the crystal." For these reasons, the presence of extraneous radionuclides could theoretically cause back scatter of gamma
counts into a lower energy window used to measure an
iodine 125 radionuclide tagged analyte. Laboratorians and
clinicians alike have been concerned for a long time about
the potential effect of this interference on RIA test results.
For many years The College of American Pathologists
(CAP) has included questions concerning this in their inspection checklist.
In the early days of RIA, absorption systems, using particulate matter such as charcoal, talc, or dextran, or salting
out techniques had relatively high levels of nonspecific
binding, especially when the absorbed phase was counted.5
It is in the nonspecific binding phase of the assay in which
significant amounts of extraneous radioactivity could be
incorporated, altering the assay counts and, therefore, results. Modern RIA techniques using coated tubes or beads
offer excellent separation efficiency. Furthermore, in some
of these tests, the solid phase can be washed extensively
because of chemical linking of the antibody to the solid
phase. This dramatically reduces any contamination in
the nonspecific binding phase. Also, the higher specificity
offered by monoclonal antibodies and the double precipitated antibody systems generally yield excellent separations and increased specificity.
In large part, the meticulous techniques observed in
RIA testing, as well as the requirement for extremely small
A.J.C.P. • November 1990
sample sizes, are responsible for the insignificant contaminations seen in the final counting phase of the assay as
shown in this study. It is interesting that contaminating
gamma counts as high as 100,000 counts/minute in 4-5
mL (0.004-0.005 L) of sample serum produced an average
of only four potentially contaminating counts back scattering into the test assay energy window. This small number of extraneous gamma counts, when plotted on a typical RIA standard curve, or even an immunoradiometric
assay standard curve, which generally is composed of fewer
counts on the low-dose portion of the curve (making this
test more susceptible to extraneous radioactivity), shows
no significant effect on the clinical value generated from
this standard curve.
In our study, our protocol called for the bound phase
to be counted in all instances. Any laboratory counting
either charcoal phase or free phase counts to determine
RIA results should consider performing a similar evaluation on their assay systems.
The findings of this study support one previously published abstract reporting results of an ongoing study similarly monitoring the effects of extraneous radioactivity
on RIA. It is interesting that this study, using a different
study design, concluded that extraneous isotopes did not
significantly affect the assay result.6 In that study, the design depended on initial assay compared with reassay
when the extraneous radionuclide had significantly deteriorated. The one exception in that study was the ferritin
test, which did show decreased assay results.
In our study, the only exception was a significant contamination present in the counted phase in the simultaneous vitamin B12 and folate assay. Here, the B12 assay
radionuclide marker was cobalt 57 with a major range of
gamma emission between 114 and 136 keV. This energy
range overlaps portions of all of the contaminating radioisotopes that were tested, and, therefore, differential identity of these isotopes cannot be achieved. If significant
levels of a high-energy gamma-emitting radioisotope such
as iodine 131 could effectively be eradicated from the RIA
test system with the use of standard protocol only, it is
likely that the same is occurring in the vitamin B12 assay
as well. Although we did not test our system by omitting
the radiolabeled assay antigen and running the test as a
blank to determine potentially interfering isotope counts,
this may be a method to identify such an isotope. An
additional confounding factor is the use of identical cobalt
57 in the Schilling test for vitamin B12 absorption in vivo.
This, therefore, requires the clinician to be aware of this
problem and draw any necessary RIA tests before radioisotope administration, especially because cobalt 57 has
a half-life of 270 days.
There has been a paucity of studies in the literature
addressing this problem. This controversy has raged behind the scenes for many years, defended by intuitive logic
Vol. 94 • No. 5
DIAGNOSTIC RADIOISOTOPE EFFECT ON RIA TESTING
or small unpublished trials performed individually. This
matter may warrant additional study, especially if any of
the current practices in clinical nuclear medicine change.
Over the last several decades, the dosage of administered
radionuclides for in vivo diagnostic testing has not been
reduced significantly. There has been much advance,
however, in thefieldof radiopharmaceuticals. The degree
of biologic sequestration and compartmentalization of
radiolabeled substances, along with the increased efficiency of excretion and the use of radionuclides with
shorter half-lives (i.e., technetium pyrophosphate), has led
to decreased serum concentrations. However, the benefit
of this decreased serum exposure may be negated if, in
the search for greater image resolution, increased sensitivity of detectors cannot adequately be achieved without
concomitant increase in radionuclide dosage.2 If, in the
future, the field of nuclear medicine requires increased
dosages to increase image resolution, the parameters under
which this study design was undertaken would change. If
this occurs, periodic surveillance studies of this type should
be performed to ensure the continued validity of these
study results.
In summary, this study examined a large volume of
RIA tests in a large community medical center and clinics
and found that less than 1% showed extraneous background radiation before assay. Using differential energy
window gamma counting, we were able to show that significant levels of initial whole serum contamination by
623
radioactivity did not alter the RIA test results significantly.
We conclude that the requirement of screening all RIA
patient serum samples is unnecessary, with the possible
exception of cobalt 57 based assay systems.
References
1. Clark PM, Raggatt PR, Price CP. Antibodies interfering in immunometric assays. Clin Chem 1985;31:1762.
2. Freeman LM, Johnson PM, eds. Clinical scintillation imaging. 2nd
ed. New York: Grune and Stratton/Hartcourt Brace Jovanovich,
1975:755.
3. Lea OA. Characterization of a serum factor interfering with the radioimmunoassay of prostatic acid phosphatase. Clin Chim Acta
1985;149:197-203.
4. Leino A, Kaihola HL, KJeimola V, Kero P. False pathological thyrotropin (TSH) level in mother and infant caused by interfering
antibodies in the TSH radioimmunoassay. Acta Paediatr Scand
1985;74:607-608.
5. Leiva WA. RIA test system quality control. Irvine, California: Beckman Instruments, 1977.
6. Pai SH, Rackliffe DM, Smith C. Effects of extraneous radioactivity
on radioimmunoassay. Am J Clin Pathol !988;80:516.
7. Pearson-Murphy BE, Grigg ERN. Radioassays. Semin Nuc Med
1979;9:169-172.
8. Regester RF, Painter P. False-positive radioimmunoassay pregnancy
test in the nephrotic syndrome. JAMA 1981;246:1337-1338.
9. Sain A, Sham R, Singh A, Silver L. Erroneous thyroid-stimulating
hormoneradioimmunoassayresults due to interfering antibovine
thyroid-stimulating hormone antibodies. Am J Clin Pathol
1979;71:540-542.
10. Vladutiu AO, Sulewski JM, Pudiak KA, Stull CG. Heterophilic antibodies interfering with radioimmunoassay: a false positive pregnancy test. JAMA 1982;248:2489-2490.
11. Wagner HN, ed. Principles of nuclear medicine. Philadelphia: WB
Saunders, 1968.