Clinical Chemistry 43: 957-962, 1997;
(Clinical Chemistry. 1997;43:957-962.)
© 1997 American Association for Clinical Chemistry, Inc.
Misleading results from immunoassays of serum free thyroxine in the presence of rheumatoid factor
Anthony G. W. Norden1,a,
Rodwin A. Jackson2,
Lorraine E. Norden1,
A. Jane Griffin3,
Margaret A. Barnes4 and
John A. Little5
1
Departments of Chemical Pathology and
3
Rheumatology, Chase Farm Hospitals Trust, The Ridgeway, Enfield, Middlesex EN2 8JL, UK.
2
Department of the Care of the Elderly, Enfield
Community Care Trust, The Ridgeway, Enfield, Middlesex EN2 8JL, UK.
4
Eagle House Surgery, 291 High St., Ponders End,
Enfield, Middlesex EN3 4DN, UK.
5
North-East Thames Regional Immunoassay (NETRIA)
Laboratory, St. Bartholemew's Hospital & Medical College, 51/53
Bartholemew Close, London EC1A 7BE, UK.
a Author for correspondence. Fax 0181 342 0558; e-mail 100127.1142{at}compuserve.com
 |
Abstract
|
|---|
A novel interference with measurements of serum free thyroxine
(FT4) caused by rheumatoid factor (RhF) is described. We
found misleading, sometimes gross, increases of FT4 results
in 5 clinically euthyroid elderly female patients with high RhF
concentrations. All 5 patients had high FT4 on Abbott
AxSYM® or IMx® analyzers. "NETRIA"
immunoassays gave misleading results in 4 of the 5 patients;
Amerlex-MAB® in 2 of 4 patients; AutoDELFIA®in 2 of the 5; and Corning ACS-180® and Bayer Diagnostics
Immuno 1® in 1 of the 5. BM-ES700® system
results for FT4 in these women remained within the
reference range. Results for serum T4, thyroid-stimulating
hormone, free triiodothyronine, thyroid-hormone-binding globulin, and
FT4 measured by equilibrium dialysis were normal in all 5
patients. Drugs, albumin-binding variants, and anti-thyroid-hormone
antibodies were excluded as interferences. Addition to normal serum of
the RhF isolated from each of the 5 patients increased the apparent
FT4 (Abbott AxSYM). Screening of 83 unselected patients
demonstrated a highly significant positive correlation between
FT4 (Abbott AxSYM) and RhF concentrations. Discrepant,
apparently increased FT4 with a normal result for
thyroid-stimulating hormone should lead to measurement of the
patient's RhF concentration.
Key Words: indexing terms: hyperthyroidism • rheumatoid arthritis • fluoroimmunoassay • polyethylene glycol • affinity chromatography • immunoabsorption • equilibrium dialysis • thyroid-stimulating hormone
|
Introduction
|
|---|
Diagnosis and management of patients with suspected thyroid
disease depend on reliable measurements of serum concentrations of
thyroid hormones and thyroid-stimulating hormone
(TSH).1
These measurements have become readily available through the
widespread use of automated immunoassays. Measurements of "free"
thyroid hormones, particularly free thyroxine (FT4),
are central to the TSH/FT4 testing strategy
(1)(2). Analytical interferences with
measurements of serum FT4 assays are well recognized and
include the presence of anti-thyroid hormone antibodies,
thyroid-hormone-binding albumin variants (found in familial
dysalbuminemic hyperthyroxinemia), heterophile antibodies, and various
drugs (including salicylate, for some methods) (3)(4)(5)(6). The
frequency with which the different assay methodologies in use are
affected is not precisely known (2)(4).
Stimulated by the finding of a spuriously increased
FT4 result in one elderly female patient, we surveyed
prospectively the thyroid-screening workload of our general hospital
laboratory for similar problems. Over a 6-month period we found another
four patients with a similar pattern of results. All five patients were
elderly women with high concentrations of serum rheumatoid factor
(RhF). A comparison of our FT4 results in these patients
with results from other laboratories using similar methods demonstrated
widespread interference with FT4 measurements.
|
Materials and Methods
|
|---|
Serum FT4 in the five patients described were
measured with the Abbott Diagnostics systems, all within 4 h of
blood collection; FT4 in serum stored at 4 °C was stable
for at least 48 h. Sera from these patients were also stored at
-80 °C in 1-mL aliquots for the further studies described. The
aliquots were thawed once only and kept for no longer than 24 h at
ambient temperature before analysis for FT4 in other
analytical systems or for other thyroid-related hormones and proteins;
RhF was isolated from parallel aliquots within ~2 h after thawing.
The survey of patients' sera with positive RhF was performed on
specimens frozen at -20 °C for as long as 3 months before assay.
Procedures followed for obtaining patients' samples were in accordance
with the Helsinki Declaration of 1975, as revised in 1983.
TSH was measured by the following techniques: AxSYM®
(Abbott Diagnostics Div., Berks., UK); TSH MAIAclone® IRMA
(cat. no. 370023; Bio-Stat, Stockport, UK); "NETRIA" 2-step
immunoenzymometric assay [North-East Thames Regional Immunoassay
(NETRIA) Laboratory, London, UK]; AutoDELFIA® (Wallac UK,
Milton Keynes, UK); ACS-180® (Ciba Corning Diagnostics,
Halstead, Essex, UK).
RhF was isolated from patients' sera as follows, all steps done at
room temperature: We dialyzed 1-mL serum aliquots for 5 h against
2 L of 10 mmol/L sodium phosphate buffer, pH 7.4, containing KCl, 2.7
mmol/L, and NaCl, 500 mmol/L ("0.5M PBS"). We then
applied 0.5 mL of the dialyzed serum to 0.8 x 5 cm columns of
human IgG immobilized on agarose (Sigma–Aldrich, Poole, UK; cat. no.
A-6284, lot 07H8824; 5–10 mg of IgG bound per milliliter of settled
gel). The eluent was "0.5 M PBS" containing bovine
albumin, 10 g/L, and the flow rate was 0.1 mL/min. We applied the
dialysate by washing it onto the column with 0.25 mL of eluent and
stopping the flow for 15 min; the first 5 mL of eluate following the
sample application was collected. We then applied to the column freshly
prepared potassium thiocyanate, 3 mol/L in "0.5M PBS"
containing 10 g/L bovine albumin ("3M KSCN"), and
collected the next 5 mL of eluate, which contained most of the RhF
activity. We promptly dialyzed both 5-mL eluate fractions exhaustively
against 10 mmol/L sodium phosphate buffer, pH 7.4, containing KCl, 2.7
mmol/L, and NaCl, 138 mmol/L (phosphate-buffered saline; PBS), and
concentrated the dialysates to 0.5 mL in an Amicon (Stonehouse, UK) B15
concentrator. Inclusion of bovine albumin in the eluents was essential
for good recovery of RhF. Serum protein electrophoresis and
immunofixation was performed on the Paragon system [Beckman
Instruments (UK), High Wycombe, UK] as recommended by the
manufacturer.
To measure RhF in eluates from IgG-agarose columns, we used the Beckman
Array 360®, having diluted 0.05 mL of each eluate 1:5
in PBS containing 10 g/L bovine albumin. To measure the apparent
FT4-increasing activity of the RhF in eluates from
IgG-agarose columns, we mixed 0.125 mL of "normal" serum (serum
without high concentrations of RhF or FT4) with 0.125 mL of
the concentrated eluates from IgG-agarose, eluted with either
"0.5M PBS" or "3M KSCN," from each of the
five patients and from a "normal" serum and then measured the
FT4. Endogenous FT4 concentrations in the
eluates, measured by mixing 0.125 mL of each dialyzed concentrated
eluate with an equal volume of PBS, were negligible. Apparent
FT4-increasing activity was calculated as: (apparent
FT4 in the mixture of patient's eluate and "normal"
serum) - (apparent FT4 found in the corresponding eluate
from the "normal" serum).
|
Patients
|
|---|
Patient 1 was an 84-year-old woman with rheumatoid arthritis that
was controlled with enteric-coated prednisolone, 2.5–5 mg per day. Her
case was reviewed in the outpatient department because of weight loss
of ~5 kg in the previous 6 months. Marked rheumatoid deformities of
both hands was evident but no signs of thyrotoxicosis (2).
This patient, like the other patients reported here, had never received
medication with thyroid preparations. Thyroid-function tests in March
1995 showed an increased serum FT4 inconsistent with the
normal TSH concentration (Table 1
). Review of her thyroid-function tests performed over the
previous 3 years (1992–1995) showed markedly high FT4
concentrations with normal values for TSH (Fig. 1
). Concern about the inconsistent thyroid-function results led
to reinvestigation of this patient (Tables 1
and 2
). Results of liver-function tests, serum protein
electrophoresis, immunoglobulins (G, A and M), thyroid-hormone-binding
globulin, anti-thyroid-microsomal and anti-thyroglobulin antibodies,
screen for familial dysalbuminemic hyperthyroxinemia, and
anti-T4 and anti-T3 antibodies were normal.
View larger version (17K):
[in this window]
[in a new window]
|
Figure 1. (Top curve) Apparent serum
FT4 (reference range, 9.0–24 pmol/L) and
(bottom curve) TSH concentration (reference range
0.3–5.0 mU/L) in Patient 1 from September 1992 to March 1995.
Measurements were made with the Abbott IMx in 1992–4 and the Abbott
AxSYM in 1995.
|
|
Four additional patients with persistent isolated unexplained increases
of FT4 were found during a prospective 6-month survey
(Table 1
) of ~9100 combinations of FT4/TSH screening
results, of which 1365 were from women older than 70 years. Four of the
five patients received steroid medication, three of whom were taking
prednisolone, <10 mg/day. All patients had normal concentrations of
thyroid-binding globulin, a negative screen for familial dysalbuminemic
hyperthyroxinemia, and no anti-T4 and
anti-T3 antibodies.
|
Results
|
|---|
In addition to the assays stated above, serum from each of the
five patients was also analyzed for FT4 with the
Abbott IMx® analyzer and was sent to colleagues in other
laboratories for analysis of FT4 by other analytical
systems (Table 2
). The results demonstrate marked analyzer-dependent
variation. Depending on the analytical system in use by the laboratory,
a patient's FT4 result might vary from normal (consistent
with the measured TSH) to increased by >400%.
In contrast, measurements of TSH in each of the patients shown in Table 1
and by the four quite different assay systems described in
Materials and Methods did not show marked analyzer-dependent
variation. All 5 patients had TSH concentrations within the assay
reference range for all 5 different TSH methods examined, except for
one measurement of TSH in Patient 3, which demonstrated a borderline
increase (5.19 mU/L) compared with an upper reference limit of 5.0
(Table 1
)
Because the five patients all had high concentrations of RhF, we
assayed additional subjects' samples to examine whether there was any
general relation between the FT4 measured by
immunoassay and the RhF concentrations in serum (Fig. 2
). The criteria for inclusion of patients were a positive RhF by
immunonephelometry (>20 kU/L), a normal TSH concentration, and
clinical absence of thyroid disease. The results shown in Fig. 2
(which
exclude the five original patients) demonstrate a highly significant
statistical association between apparent FT4 and
concentrations of RhF (Spearman rank correlation coefficient =
0.53, P <0.001).
View larger version (16K):
[in this window]
[in a new window]
|
Figure 2. Plot of serum FT4 against serum RhF
concentration measured in 83 individuals.
Criteria for inclusion of patients were a positive RhF content (>20
kU/L), normal TSH concentration, and absence of clinical thyroid
disease. A least-squares line of fit is shown.
|
|
To demonstrate directly that RhF was the cause of the misleading
results we had found, we depleted of RhF the sera from each of the five
patients originally studied, using immunoadsorption on columns of
IgG-agarose. The RhF bound to each column was eluted with a chaotropic
solvent, 3 mol/L potassium thiocyanate, and was recovered in the yields
listed in Table 3
. The RhF-depleted and the purified RhF fractions were tested
for their ability to increase the apparent FT4
measured in normal serum (Fig. 3
). Preliminary experiments had demonstrated that mixing normal
serum with the sera from each of the 5 patients would lead to an
apparent FT4 increase.
View larger version (19K):
[in this window]
[in a new window]
|
Figure 3. (Top) RhF concentrations in eluates from serum
applied to IgG-agarose; (bottom)
FT4-stimulating activity (pmol/L) in eluates from serum
applied to IgG-agarose from each of the five patients.
RhF in the corresponding eluates from normal serum was undetectable
(<20 kU/L). Here, we measured the effect on the apparent
FT4 from patient's serum depleted of RhF, and of RhF
purified from each of the five patients studied and mixed with normal
serum.
|
|
Most of the apparent FT4-increasing activity in each
serum was associated with the IgG-agarose-bound fraction, which is
consistent with RhF being the cause of the activity (Fig. 3
).
Electrophoresis of the fractions eluted from IgG-agarose from patient 5
showed that the monoclonal IgA component was not bound to the column
(Fig. 4
), although most of the RhF activity was adsorbed (Fig. 3
, top).
View larger version (100K):
[in this window]
[in a new window]
|
Figure 4. Electrophoretogram on agarose gel of (lane 1)
serum from patient 5 before application to an IgG agarose column and
(lanes 2–4) concentrated eluates from the column.
The gel was stained for protein as described in Materials and
Methods. The anode is uppermost. Lanes: 1, serum from
patient 5 after dialysis against "0.5M PBS" and before
application to IgG agarose column; 2, concentrated
"3M KSCN" eluate with most of the RhF activity;
3, control (PBS alone); and 4, concentrated
"0.5M PBS" eluate with negligible RhF activity. The RhF
activity of each eluate is shown in Table 3
. Immunofixation
demonstrated that the paraprotein in the mid-gamma region was IgA kappa
(Table 1
).
|
|
|
Discussion
|
|---|
All five patients reported had a consistently high
FT4 value in at least two of the eight routine
immunoassays examined; serum from one patient (patient 5) gave abnormal
results in six of the assay systems. FT4 results by
equilibrium dialysis, the "gold-standard" method for
FT4, were normal in all patients (Table 2
) and consistent
with the measurements of TSH (6). The presence of RhF at
high concentrations in all five sera is almost certainly the cause of
the interference. This is supported by the following observations: All
five patients had high concentrations of RhF (Table 1
); precipitation
with polyethylene glycol, a procedure known to precipitate
immunoglobulins, virtually abolished the interference (Table 2
); and
the apparent increase of FT4 in normal serum was almost
entirely confined to the IgG-agarose-bound fraction of each patient's
serum, i.e., the fraction containing purified RhF (Fig. 3
)
Reported causes of interference with FT4 assays
include: thyroxine-binding albumin variants (7); the
presence of heterophile antibodies, e.g., human anti-mouse antibodies
(3); nonspecific immunoglobulins reacting with the assay
system (6); and anti-T4 and
anti-triiodothyronine (T3) antibodies
(1)(3)(5).
Immunoglobulin-mediated interferences in commercial and noncommercial
assays are minimized by adding nonimmune globulin from the same or
related species as the antibody used for the assay ("blockers");
nonetheless, such interferences in immunoassays still occur in
individual patients (9).
The principle of the Abbott AxSYM and IMx assays for
FT4 [Abbott Reference Data List 7A54 (July 1994) and
2222 (Dec. 1990)] is the initial mixing of microparticles coated with
sheep anti-T4 antibody and the patient's serum. A
relatively low concentration of sheep antibody is used so that the
equilibrium between free and bound hormone in the patient's serum is
perturbed as little as possible. After washing, alkaline
phosphatase-labeled T3 is added to the solid phase and
occupies any vacant anti-T4 binding sites. The fluorescence
signal after further washing and adding of a fluorogenic substrate has
an inverse relationship to the FT4 concentration in the
patient's serum. We hypothesize that, in the Abbott AxSYM assay, RhF
reacts with the immobilized sheep anti-T4 and inhibits the
subsequent binding of alkaline phosphatase-labeled T3. The
fluorescence signal will therefore be decreased and the apparent
FT4 concentration will be increased. Immobilization of the
sheep immunoglobulin may make it reactive with the human RhF in a
similar way as immobilization of human IgG on agarose, which we used to
purify RhF from the patients' sera (Fig. 3
). Inhibition of the binding
of reagents by anti-T4 immunoglobulin may also underlie the
interferences seen in the other system in Table 2
. Interspecies binding
of RhF is well recognized (10)(11).
Interestingly, in Patient 5, who had a low concentration of a serum
paraprotein (Table 1
), neither the RhF activity nor the apparent
FT4-increasing activity was associated with the
paraprotein (Fig. 4
and Table 3
). The presence of the paraprotein
appears to be coincidental.
We are surprised that interference from RhF in these assays has not
been previously reported, even in studies of thyroid function in
rheumatoid arthritis (12). Presumably, this is explained
by the variable incidence of the problem among analyzer systems (Table 2
) and the finding that only a minority of RhF-positive patients' sera
have an apparent FT4 above the upper reference limit (Fig. 2
). Furthermore, the laboratories that use only ultrasensitive TSH
assays as a front-line screen for thyroid status (a widely used
strategy in the UK) will, of course, not detect such patients. All five
patients reported here were elderly women, a group with a high
incidence of autoantibodies; one, however, patient 3, did not have
clinical rheumatoid arthritis despite a high RhF concentration.
Furthermore, RhF was measured in two of these patients only after the
misleading FT4 was discovered and then only as part of this
investigation. Because RhF may occur even in healthy individuals as
well as in patients having a wide variety of nonrheumatological
diseases (10)(13), the recognition of RhF or
associated immunoglobulins as a cause of misleading FT4
results may be difficult. The exact incidence of this novel
interference among patients screened for thyroid disease remains to be
determined.
In conclusion, our findings reinforce recent recommendations
(4) for more rigorous assessment of interferences in
commercial diagnostic systems for FT4, which are among the
most widely used of all immunoassays. Indeed, we expect that RhF
interferences with other immunoassays will be found. We recommend that
patients found to have discrepant increases of serum FT4
concentrations and a normal TSH value should be screened for RhF.
Clearly, however, even this strategy will fail to detect misleading
FT4 results if the result is increased only into the
reference range, rather than above the upper limit of the range.
|
Acknowledgments
|
|---|
We thank the many colleagues who performed assays on our behalf and
discussed results with us, including P.M.S. Clark, T. Dunning, J.
Gurney, J. King, V. Lyfar, G.L. Manderino, R.J.Mardell, and R. Passas.
R. Majeed suggested the experiments with IgG-agarose columns. The
interpretations expressed here are our own. We thank D.A. Isenberg for
comment and discussion of the draft manuscript.
|
Footnotes
|
|---|
1 Nonstandard abbreviations: FT4, free thyroxine; RhF, rheumatoid factor; NETRIA, North-East Thames Regional Immunoassay; TSH, thyroid-stimulating hormone; PBS, phosphate-buffered saline; and T3, triiodothyronine. 
|
References
|
|---|
-
Keffer JH. Preanalytical considerations in testing thyroid function. Clin Chem 1996;42:125-134.
[Abstract/Free Full Text]
-
Lazarus JH, John R, Hall R. Thyroid disorders. Bouloux P-MG Rees LH eds. Diagnostic tests in endocrinology and diabetes 1994:39-66 Chapman and Hall London. .
-
Fiad TM, Duffy J, McKenna TJ. Multiple spuriously abnormal thyroid function indices due to heterophilic antibodies. Clin Endocrinol 1994;41:391-395.
[Medline]
[Order article via Infotrieve]
-
Nelson JC, Wilcox RB. Analytical performance of free and total thyroxine assays. Clin Chem 1996;42:146-154.
[Abstract/Free Full Text]
-
Liewendahl K, Melamies L, Helenius T, Karonen SL, Liljestrom E, Valimaki M. Automated and manual serum free thyroxine assays evaluated with equilibrium dialysis. Scand J Clin Lab Invest 1994;54:347-351.
[ISI][Medline]
[Order article via Infotrieve]
-
Lai LC, Day JA, Clark F, Peaston RT. Spuriously high free thyroxine with the Amerlite MAB FT4 assay. J Clin Pathol 1994;47:181-182.
[Abstract/Free Full Text]
-
Tokmakjian S, Haines M, Edmonds M. Free thyroxine measured by the Ciba Corning ACS-180 on samples from patients with familial dysalbuminemic hyperthyroxinemia [Letter]. Clin Chem 1993;39:1748-1749.
[ISI][Medline]
[Order article via Infotrieve]
-
Westerhuis LW, Venekamp WJ. Falsely high serum free thyroxine concentration measured with Amerlite-MAB FT4 [Letter]. Clin Chem 1995;41:633-634.
[Free Full Text]
-
Martin-du-Pan RC, Horak M, Bischof P. Clinical significance of invisible or partially visible luteinizing hormone. Hum Reprod 1994;9:1987-1990.
[Abstract/Free Full Text]
-
Hamako J, Ozeki Y, Matsui T, Yamamoto Y, Inoue T, Yukitake J, Titani K. Binding of human IgM from a rheumatoid factor to IgG of 12 animal species. Comp Biochem Physiol 1995;112B:683-688.
-
Koopman WJ, Schrohenloher RE. Utsinger PD Zvaifler NJ Ehrlich GE eds. Rheumatoid arthritis 1985:217-241 JB Lippincott Co. Philadelphia. .
-
Shiroky JB, Cohen M, Ballachey M-L, Neville C. Thyroid dysfunction in rheumatoid arthritis: a controlled prospective survey. Ann Rheum Dis 1993;52:454-456.
[Abstract/Free Full Text]
-
Thomas MJ, Adebajo A, Chapel HM, Webley M. The use of rheumatoid factors in clinical practice. Postgrad Med J 1995;71:674-677.
[Abstract]
The following articles in journals at HighWire Press have cited this article:
|
|
|
|
 
V. Marks
False-Positive Immunoassay Results: A Multicenter Survey of Erroneous Immunoassay Results from Assays of 74 Analytes in 10 Donors from 66 Laboratories in Seven Countries
Clin. Chem.,
November 1, 2002;
48(11):
2008 - 2016.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. A.A. Ismail, P. L. Walker, J. H. Barth, K. C. Lewandowski, R. Jones, and W. A. Burr
Wrong Biochemistry Results: Two Case Reports and Observational Study in 5310 Patients on Potentially Misleading Thyroid-stimulating Hormone and Gonadotropin Immunoassay Results
Clin. Chem.,
November 1, 2002;
48(11):
2023 - 2029.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
A. A. Ismail and J. H Barth
Wrong biochemistry results
BMJ,
September 29, 2001;
323(7315):
705 - 706.
[Full Text]
[PDF]
|
|
|
|
|
|
|
N. Despres and A. M. Grant
Antibody interference in thyroid assays: a potential for clinical misinformation
Clin. Chem.,
March 1, 1998;
44(3):
440 - 454.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
