N-terminal parathyrin and calcium in serum. Clin. Chem. 26, 1672-1675 (1980). Charles D. Hawker. SmithKline. Clinical Laboratories,. Inc. St. Louis, MO 63141.
term drift is determined by plotting results for the quality-control pool over time. Our technique of updating the pool limits neither masks nor detracts from the quality-control procedure. If there is a drift, it shows up on the time plot. To the contrary, as additional data are accumulated, the confidence intervals for each pool get narrower and narrower, permitting the laboratory to control the assays to a much “finer tune” than is possible by doing a relatively small number of assays and then locking the confidence limits as many laboratories do. With regard to assay specificity and our assertion that C-terminal assays remain the assay of choice for evaluation of hypercalcemic states, we stand firmly behind our statements, both in the proposed Selected Method and in our subsequent Letter (1). It should be noted that the four distinguished Evaluators agreed with this conclusion as well. As long as there are investigators, there will be controversy, but one should examine the nature of the arguments to place them in their proper perspective. The references the authors cite include two from laboratories that developed “mid-region” antisera for PTH (13, 14) and thus had a bias toward this type of assay, but neither have shown clinical data in primary hyperparathyroidism that were as good as data obtained with the best C-terminal assays (15). Another reference (16) described a 1976 splitspecimen evaluation of the assays of four different reference laboratories, but most of the assays were no longer in use when the paper was published. Moreover, the chief investigator of that project misinterpreted and misrepresented the data despite admonitions from the reference laboratories. The fmal reference is to the correspondents’ own published data (17), but their Cterminal assay data fails to come even close to what we would consider to be the standard of performance for C-terminal assays, because fewer than half of patients with primary hyperparathyroidism had increased values-compared with more than 90% with most C-terminal assays (15). Therefore, their comparison of N-terminal vs C-terminal based on data generated with their inhouse assays is of no value in a debate over the type of assay to use. Finally, with regard to the assertion that our comments about using both assays in renal-failure patients represented an opinion that had no supporting definitive data, the authors should examine the seven literature citations (only two of which were ours) in support of this argument, which were listed consecutively (refs. 28-34) in the proposed Selected Method. We believe that our proposed Selected Method has accomplished its primary
objectives. It described methods that have stood the test of time, although they are not etched in stone and are subject to improvements or replacement. It gave details that readers can use to develop and evaluate their own procedures, including criteria on which critical judgements should be based. And, it communicated information and elicited discussion. We thank those who wrote letters for participating in this process. References 1. Di Bella, F. P., and Hawker, C. D., Comment on the proposed Selected Method for parathyrin. Clin. Chem. 28, 1817-1818 (1982).
2. Lindall, A. W., and Cohn, D. V., Ibid.,
p
al., Clinical utility of radioimmunoassays for parathyroidhormone. Miner. Electrolyte Metab. 3,283-290 (1980). 16. Raisz, L. G., Yajnik, C. H., Beckman, R. S., and Bower, B. F.,Comparison of com-
mercially available parathyroid
Charles SmithKline Inc. St. Louis,
Clinical
Ibid.,
Laboratories,
Francis
p 1817.
4. Berson, S. A., Yalow, R S., Aurbach, G. D., and Potts, J. T., Jr.,Immunoassay ofbovine and human parathyroid hormone. Proc. Nati. Acad. Sci. USA 49,613-617 (1963). 5. Fleisher, M., Oettgen, H. F., Besenfelder, E., and Schwartz, M. K., Measurement of carcinoembryonic antigen. Clin. Chem. 19, 1214-1220 (1973). 6. Auletta, F. J., Zusman, R. M., and Caldwell, B. V.,Developmentand standardization of radioimmunoassay for prostaglandins E, F, and A. Clin. Chem. 20, 1580-1587 (1974). 7. Hawker, C. D., Glass, J. D., and Rasmussen, H., Further studies on the isolation and characterization of parathyroid polypeptides. Biochemistry 5, 344-352 (1966). 8. Arnaud, C. D., Tsao, H. S.,and Littledike, T.,Radioimmunoassay of human parathyroid hormone in serum. J. Clin. Invest. 50, 21-34 (1971). 9. Berson, S. A., and Yalow, R. S., Immuno-
D. Hawker
MO 63141
1817.
3. Schmidt-Gayk, H.,
hormone
immunoassays in the differential diagnosis of hypercalcemia due to primary hyperparathyroidism or malignancy. Ann. Intern. Med. 91,739-740 (1979). 17.Simon, M., and Cuan, J., Diagnostic utility of C-terminal parathyrin measurement as compared with measurements of N-terminal parathyrin and calcium in serum. Clin. Chem. 26, 1672-1675 (1980).
The Upjohn
Company
Kalamazoo,
MI 49001
P. Di Bella
Treatmentwith Heparinand Resultsfor Free Thyroxin:An In VIvo or an in Vitro Effect? To the Editor:
The recent Letters of Boss et al. (1) and Lundberg et al. (2) serve as timely reminders of the difficulties of measuring factors relating to thyroid function in the plasma of patients who are undergoing therapy with heparin. Artefactual increases of as much as 50% in total thyroxin (T4) estimated by competitive-binding protein assay (CPBA) and as much as 30% in triiodothyronine
(T3) resin
uptake
have pre-
chemical heterogeneity of parathyroid hormone in plasma. J. Clin. Endocrinol. Metab.
viously been reported (3) for patients receiving heparin. It is well known (4) 28, 1037-1047 (1968). that nonesterified fatty acids (NEFA) 10. Hawker, C. D., and Di Bella, F. P., are released by lipoprotein lipases Parathyroid hormone inchronicrenalfailure: stimulated in vivo by heparin, but it has Studies with two different parathyroidhornot been fully realized until recently (5) mone radioimmunoassays. Contrib. Nephrol. that most of the NEFA found in the 20,21-37 (1980). plasma of heparinized patients arises 11. Flueck,J.A., Di Bella, F. P., Edis, A. J., from very rapid and continuing lipolytic et al., Immunoheterogeneity of parathyroid hydrolysis of triglycerides after blood hormone in venous effluent serum from byhas been drawn. Variations in the hanperfunctioning parathyroid glands. J. Clin. dling of blood in vitro may explain why Invest. 60, 1367-1375 (1977). frequent observations of significantly 12. Yalow, R. S., Significance of the heteroincreased concentrations of free thygeneity of parathyroid hormone. In Endocrinology of Calcium Metabolism. D. H. roxin (VI’4) associated with comparable Copp and R. V. Talmage, Eds.,Excerpts Medica, Amsterdam, 1977, pp 308-312. 13.Lindall, A. W., Ross,B.,and Cecchettin, M., Potential clinical usefulness of new glandular and circulating parathyroid peptides illuminated by region specific radioimmunoassay. In 3rd mt. Symp. on Calciotropic Hormones, Methods and Clin. Applications, Lorenzini Found. Symp. 121, 1981, p 26. 14. Wood, W. G., Marschner, I., and Scriba, P. C., Tests on three antisera and subsequent development of radioimmunoassay for different regions of human parathyrin. Horm. Metab. Res. 11,309-317 (1979). 15. Martin,K. J.,Hruska,K.,Freitag, J.,et
increases
of NEFA
these patients
in the
plasma
of
are not always confirmed
(3).
NEFAs are potent
competitors
with
T4 (and T3) for binding siteson serum albumin (6). It is to be expected that the
large amounts of these substances generated during the preparation of plasma from heparinized subjects may perturb the equilibria between bound and free T4. This can lead to changes in FF4 and T3-uptake estimations. Calculations (7) indicate that complete blockage of albumin-binding sites by NEFA in the plasma of an euthyroid patient with
CLINICAL CHEMISTRY, Vol. 28, No. 12, 1982
2441
normal concentrations of the three plasma proteins and a T4 concentration of 100 mmolfL would lead to an increase in FF4 concentration of only about 20%. However, considerably greater percentage
changes
are
seen
in various of FT4 (1,3,
methods for the estimation 7) when results are compared before and immediately
after
treatment
with hep-
arm. This suggests the presence ondary
interference
of sec-
mechanisms,
re-
lated to the methodology of FT4 measurement. Equilibrium dialysis FT4 assays appear to be especially sensitive to interfering substances in the plasma of heparinized patients. It is not uncommon for FT4 values in post-heparin plasmas to be as much as sixfold higher (3, 7) than the pre-heparin values for euthyroid patients. It has been known for at least 27 years (8) that the NEFAs that bind most avidly to the albumin
T4-binding sites are effectively alyzed through a semipermeable brane
of the type
used
not dimem-
in equilibrium
dialysis, whereas T4 is fully permeable. Mass-action equilibrium considerations show (7) that the negative concentration gradient of NEFA that is set up across the membrane between the sample and buffer compartments must be compensated by an equivalent increase in FF4, thermodynamically engineered in the buffer compartment. FT4 estimations by equilibrium dialysis, for serum or plasma samples grossly contaminated by NEFA, may therefore be considerably greater than the actual concentrations existing in undiluted samples before dialysis. Similar effects may also occur in other techniques that depend on FT4 transport across a membrane
serum
NEFA
mmol/L,
concentration
the decrease
of 1.0
in Amerlex
RIA values was still clinically insignificant at 4%. At very high concentrations of NEFA (5-10 mmolfL) all thyroid function tests were affected. For example, T3 uptake test results were increased by 30%, equilibrium dialysis FT4 by 661%, total T4 by CPBA by 203%, and total T4 RIA by 14%. Under these conditions
a maximum
decrease
roid-function
tests
in the plasma
volunteers.
Before the
oleic acid was added, the serum total T4 (Amerlex T4 RIA) was 108 nmol/L and the FT4 (Amerlex FT4 RIA) was 16.7 pmol/L. Oleic acid was added to aliquots of the serum pool to give final concentrations ranging from 0.125 to 50 mmol/L. FT4 values were determined by the Amerlex FT4 RIA method and by a direct equilibrium dialysis assay (9); values for total T4 were determined with the Amerlex T4 RIA and an “in-house” T4 CPBA method; and T3 uptake values were measured by the “T3 (MAA) Uptake” method. Compared with the equilibrium dialysis FT4 assay, the Amerlex FF4 RIA was virtually unaf-
fected until extremely
high concentra-
tions of NEFA were reached. For example, 0.5 mmol of oleic acid per liter
produced no detectable erlex FT4 RIA results untreated 2442
control
change in Am(relative to the
samples).
CLINICAL CHEMISTRY,
With
a
of pa-
tients with chronic renal failure undergoing hemodialysis and injected with heparmn. Lipoprotein lipase is stimulated in the blood of these patients by heparin. It could be shown that, unless precautions were taken to minimize triglyceride hydrolysis in vitro after blood sampling (for example, by rapid cooling),
the production
of NEFA
led to
changes in the values of thyroid-function tests mimicking those found when oleic acid was added to normal serum. An example of this effect is given in Figure 1 for FT4 measured by equilibrium dialysisand by the Amerlex FT4 kit. Whereas FT4 values by equilibrium
We have studied the effects of NEFA on thyroid-function tests, particularly FT4 assays. As a representative NEFA, oleic acid was added in vitro to a pool of freshly prepared serum from a group of healthy
of
about 30-50% in Amerlex FT4 values was observed, similar to previous reports (1, 2). The equilibrium dialysis FT4 assay was especially sensitive at low to intermediate concentrations (0.5-1.0 mmol/L) of oleic acid, giving artefactual increases in FT4 of 140-220% relative to the control samples. The Amerlex FT4 RIA is clearly more robust to increased NEFA than the equilibrium dialysis method at intermediate NEFA values (up to 1 to 2 mmol/L), but at very high (10 mmol/L) cOncentrations of NEFA, neither method is reliable. Similar effects were seen for all thy-
(1).
euthyroid
FT4
70
60
50
#{149} Ia
140
30
20
Nofl’iPI
dialysis increased considerably when the post-heparinblood was allowed to stand
at room temperature before separation of the plasma and freezing, the Amerlex values
were much
less affected,
though
there was evidence of a small downward trend in values. We can now identify four areas of influence by NEFA on FF4 assay: (a)) the time interval between heparmn injection and taking of blood (1, 2); (b) further increases in NEFA from enzyme hydrolysis
of triglycerides
in vitro
(as in
heparin therapy)-this effect is dependent on time and temperature during plasma sample preparation; (c) possible physiological increases in blood NEFA arising out of the patient’s condition (for example, caloric deprivation or liver disease);and (d) artefacts in measurements of FF4 by some methods (especially equilibrium dialysis). The reasons for the lowering of Amerlex FT4 values by high NEFA concentrations in serum are complex and may in part be related to some weak residual binding of [1251]T4 analog tracer to serum albumin. The affinity constant of the tracer for albumin is considerably weaker (10) than that for T4, which in turn is about 100-fold weaker than for NEFA (6). Our own studies suggest that the displacement of the residual [‘2511T4 analog bound to plasma albumin by NEFA onto the antiserum may not completely account for the 30-50% less FT4 values seen in the more extreme cases (1, 2). At these extreme concentrations of NEFA, general interference begins to appear even in total T4 RIAs. It is therefore possible at extremely high concentrations of NEFA that nonspecific changes in antiserum binding of the tracer may occur in the Amerlex FF4 assay, given the relatively low concentrations of antiserum and tracer and the differences in size and charge distribution of the T4 analog, as compared with T4. In summary, the interferences in FF4 assays reported by Boss et al. (1) and Lundberg et al. (2) are largely the result of uncontrolled conversion of plasma triglycerides in vitro by lipolytic activity stimulated in vivo by heparmn. The question posed by Boss et al. (1) as to which FF4 method gives the “true” result invites the reply that no assay can give accurate
FF4 estimates
for patients
undergoing heparmn therapy until more veffective ways are found to prevent 0L blood samples from being contaminated Nn,,le. by NEFA during plasma preparation. In Fig. 1. Amerlex FF4 RIA and equilibrium our experience, even free thyroxin index dialysis estimations of FT4 in plasma of measurements may be unreliable, besix patients with chronic renal failure cause of distortions of the T3 uptake #{149}, Pre-hemodialysis (before heparin therapy) dialysis, a method plasma samples allowed to stand for >8 hat 20#{176}Cresults. Equilibrium hitherto held to be thermodynamically before separation and freezing; V, post-hemodialysis (with heparin therapy) plasma samples immeimpeccable in all cases, appears to be the diately cooled to 0 #{176}C, separated, and frozen; X, most susceptible of all FT4 assay techPost-hemodialysis (with heparin therapy) plasma niques to interference by NEFA. Our samples allowed to stand for >8 hat 20#{176}C before results suggest that dialysis FT4 methseparation and freezing
Vol. 28, No. 12, 1982
i
-
-
-
FT4 Pin
-
-
ods may give falsely increased
FT4 values, even at the intermediate increases of NEFA that are known (7) to occur in caloric-deprivation states, nonthyroidal illness, and late pregnancy. In contrast, the Amerlex FT4 RIA appears to maintain accuracy in FT4 estimations
radioimmunoassay of free thyroxine in serum dialysates. Clin. Endocrinol. 16, 101-105 (1982). 10. Beyer, H. K., Amerlex FT4 RIA-a new
investigation principle: Description, evaluation and comparison with the Dow-Lepetit FT4 RIA test. Nuel. Med. 21, 72-78 (1982).
with moderate increases in NEFA concentrations. It is perhaps also worth
commenting that the data of Lundberg et a!. (2) suggest that for untreated patients the Amerlex FT4 RIA may be superior to traditional methods in the diagnosis of mild hyperthyroidism. From our results the most important issue cannot be concerned with criteria for selection of a “best” FF4 assay through extrapolation from a situation where all tests are invalid. The essential question is how the clinician or clinical chemist can best deal with patients from intensive-care or renal-dialysis units who are undergoing heparin treatment. We believe that (a) all those who use thyroid-function tests should be aware of the pitfalls of carrying out T3 uptake, T4 CPBA, and FT4 assays during heparm therapy; (b) where possible, thyroid-function tests should be carried out on blood samples taken before heparinization or when the patients, after completion of anticoagulant treatment, have been taken off heparin therapy for a sufficient time (some hours) to allow lipolytic activity to subside; or that (c) research should be stimulated to find a safe additive to be incorporated into blood-collection tubes to prevent the development of interfering substances in vitro. References 1. Boss, M., Kingstone, D., Chan, M. K., and Varghese, Z., Contradictory findings in the measurement of free thyroxin after administration of heparin. Clin. Chem. 28, 1238-1239 (1982). Letter. 2. Lundberg, P. A., Jagenburg, R., Lindstedt, G., and Nystrom, E., Heparin in vivo effect on free thyroxin. Clin. Chem. 28, 1241-1242 (1982). Letter.
T. A. Wilkins J. E. M. Midgley A. F. Giles Clinical Reagents Development Amersham International plc White Lion Rd. Amersham, Bucks, HP7 911, U.K. Ed. note: ci. also a Letter in N. Engl. J. Med. 307: 126 (1982),inwhich results from analog methods “appear to support the misleadingly high free T4 index in familial T4 excess.”
The Large “Kit-to-Kit” VariationIn InsulinRadiolmmunoassayIs Mainly Due to Difference in StandardConcentration To the Editor:
In our previous control survey of insulin radioimmunoassay it was disclosed that the “between-kit” variation was the largest of the contributing components (1). The observed “between-kit” variation may have resulted from either the difference in standards, difference in the nature of iodinated antigen, or difference in the characteristics of the antibodies. To detect the most important contributing factor and to minimize it, we made up and distributed samples consisting of the common standards and the quality-control specimen to the participating institutes in which insulin radioimmunoassay was routine. Normal human serum was passed through an anti-insulin antibody-cou-
3. Schatz, D. L., Sheppard, R. H., Steiner, G., et al., Influence of heparin on serum free thyroxin. J. Clin. Endocrinol. Metab. 29, 1015-1022 (1969). 4. Nilsson-Ehle, P.,Garfinkel, A. S.,and Schats, M. C., Lypolytic enzymes and plasma lipoprotein metabolism. Ann. Rev. Biochem. 49, 667-693 (1980). 5. Riemersma, R. A.,Russell,D. C.,and Oliver, M. F., Heparin-induced lipolysis, an exaggerated risk. Lancet ii, 471 (1981). 6. Spector, A. A., Fatty acid binding to albumin. J. Lipid Res. 16, 165-179 (1975). 7. Hunnisett, A. G., Giles, A. F., Midgley, J. E. M., and Wilkins, T. A., False indications
of free thyroxine and other thyroid parameters in non-thyroidal illness. (Submitted for publication in J. Clin. Endocrinol. Metab.) 8. Gordon, R. S., Interaction between oleate and the lipoproteins of human serum. J. Clin. Invest. 36, 477-484 (1955). 9. Giles, A. F., An improved method for the
pled agarose (Sepharose) column (2) to obtain insulin-free human serum. To this normal
human
insulin-free
serum,
we added the standard amount of insulin in terms of the WHO standards to give graded concentrations ranging from 0 to 320 micro-mt. units/mL. The qualitycontrol specimen was prepared to give a concentration of 50 micro-mt. units! mL in a similar manner. The standards and the quality-control specimen were delivered frozen to the participating institutes without disclosing the identity of the samples, and they were asked to measure all the samples (the common standards and the quality-control specimen) with one of the commercially available kits that was routine for them. They were asked to report not only the assay values for all the samples but also their raw counting data for all the samples and those for the various kit standards. All the data were analyzed by use of the data-processing program described by Faden and Rodbard (3) in two different ways: by measuring the quality-control specimen by use of the kit standards (Mode A) and by measuring the specimen by use of the distributed common standards, ignoring the kit standards. The “between-assay” and the “within-assay” variations were calculated from analysis of variance as described by Rodbard (4). Significance of difference was examined by analysis of variance. Table 1 summarizes the assay data of the quality-control specimen measured by use of either the kit standards or the common standards we had distributed. As shown, the values ranged from 41.1 (by institute 3, which used kit f) to 97.8 micro-mt. units/mL (by institute 5, using kit a) when determined by kit standards (Mode A, in Table 1). When the same specimen was measured with use of the common standards, the values ranged from 36.2 (by institute 3, using kit a) to 54.6 micro-mt.
units/mL
(by
Table 1. Measurement of the Quality-Control Sample by Kit Standards (Mode A) and Common Standards (Mode B) Assay values, micro-mt. unlts/mL, found by the I Iv. Instftut.s 1 2 3 4 5
Name of kit
Mode
a) Insulin RIA Dainabot b) IRI-Pharmacia Shionogi c) Insulin Eiken
A B A B A B A
82.6(5%) 46.3(4%) 39.2(16%) 54.1(7%) 49.8(7%) 47.8(12%) 53.9(4%)
Daiichi
B
48.2(6%)
e) lnsulin-RCC
A B A B
49.8(7%) 52.7(17%) 42.8(21%) 48.2(5%)
d) lRl-Pharmacia
f) lnsulin-CIS a
74.0(9%) 40.5(4%) 45.1(12%) 43.4(9%) 53.2(10%) 49.5(12%) 53.5(6%) 45.4(9%) 74.6(8%) 54.6(11%) 52.6(20%) 49.9(10%)
96.5(5%) 65.9(9%) 36.2(10%) 48.2(10%) 55.7(16%) 48.7(9%) 49.5(12%) 53.3(11%) 61.4(6%) 54.2(5%) 50.2(16%) 47.6(9%) 63.5(9%) 46.7(12%) 41.1(25%) 48.8(10%)
97.8(5%) 42.8(8%)
-
71.4(4%)
-
52.0(6%)
-
-
-
-
-
-
In parentheses: the CV In percent, obtained from precision profile.
CLINICAL CHEMISTRY,
Vol. 28, No. 12, 1982
2443
