CLIN. CHEM. 31/2, 296-298 (1985) The Upper Limit of Normal for Thyrotropin Is 3 or 4 milli-int. units/L A Paul Durham Recently Musto et al. (C/in Chem 30: 329-330, 1984) noted importance of accurate measurement of thyrotropin at low concentrations, the upper limit of normal is well below the stated upper limit of normal of most commercial tests. Here I amplify their statement. that, despite the clinical Additional Keyphrases: ods pituitary hormones reference interval ‘kit” meth- In a recent discussion of thyrotropin (thyroid-stimulating hormone) measurement, Musto et al. (1) draw attention to the clinical importance of good performance at low concentrations. A thyrotropin assay that has adequate reproducibility throughout the “normal” range and is essentially free of nonspecific matrix effects can be expected to confer two benefits. First, it will minimize the number of borderline results that will require the thyroliberin (thyrotropin-releasing factor, TRF) test as followup. Second, it will provide more meaningful baseline values for comparison with values obtained after TRF stimulation. The high upper limit of normal that is characteristic of most commercially available thyrotropin assays is symptomatic of their inability to provide precise and accurate measurements of thyrotropin at low concentrations. Musto et al. (1) state without documentation that, according to the literature, the true upper limit of normal for thyrotropin is less than 5.0 milli-int. units/L. Clinicians who regularly encounter only results for thyrotropin generated by commercially available kits may well be inclined to doubt this assertion. But the point made by Musto et al. is an important one, and this Note is intended to document its truth. Studies aimed at determining the reference interval for normal individuals can provide important insights into the acceptability of an assay system. This is especially true of assays for thyrotropin, where an upper limit of normal significantly higher than those reported for fully optimized radioimmunoassays may well be symptomatic of shortcomings that can seriously compromise clinical efficacy (1). There is, in fact, a striking contrast between most but not all thyrotropin kits now on the market and the several “research” radioimmunoassays (2-6) tabulated in Table 1. The latter characteristically show an upper limit of normal of approximately 3 or 4 milli-int. units/L. The stated normal range for most kits, on the other hand, has until recently extended up to 10 milli-int. units/L. While there has been an encouraging trend towards more accurate thyrotropin kits, an upper limit of 7 milli-int. unitsfL represents, even today, the norm for nearly all commercially available assays. This shows that the great majority of thyrotropin kits still err, on borderline samples, by a factor of two relative to more definitive assay systems. Indeed, it is rare to encounter any thyrotropin procedure with a reasonable assay time (same-day or overnight) that Diagnostic Products Corp., 5700 West 96th St., Los Angeles, CA 90045. Received October 18, 1984; accepted November 27, 1984. 296 CLINICAL CHEMISTRY, Vol. 31, No. 2, 1985 Table 1. Upper Limits of Normal in Research Assays for Thyrotropin Thyrotropin, mull-mt. unlts/L Assay (and ret.) Radioimmunoassays Melbourne, Australia (2) Newcastle, U.K. (3) Munich, F.R.G. (4) Boston, MA (5) Los Angeles, CA (6) Other methods Wick chromatography/RIA (16) Adsorption to concanavalin A/RIA (17) Immunochemiluminescence assay (18) Cytochemical bioassay (19, 20) measures accuracy Research 3.1 2.8 3.8 3.0 3.5 3.0 4.0 4.5 2.4, 3.7 up to standards set by the research assays and precision at low concentrations (7, 8). Assays for vs Kit Methods The systems here designated as “research” assays have strong claims to being regarded as reference methods for thyrotropin. First, they were established more than a decade ago, and since then have contributed significantly to our understanding of the physiology and clinical relevance of this pituitary hormone. This is reflected not only in countless journal articles but also in standard medical textbooks and reviews, where values ranging from 3.0 to 5.0 milli-int. units/L are usually quoted as the accepted upper limit of normal (1, 5, 9-11). Second, these research assays resemble the typical thyrotropin kit of today in being nonequilibrium, competitive, double-antibody radioirnmunoassays. They differ prinulpally in the degree to which they have been optimized for sensitivity and specificity. Comparisons between the research assays and the kit methods are therefore in order. Characteristically, the research assays involve use of meticulously purified tracer, incubation times of six days or more, and standard curves prepared with human serum. These features can be expected to minimize nonspecific “matrix effects,” resulting in lower and more accurate values throughout the normal reference interval (12). Some of these features appear to be essential for meaningful results, particularly the steps taken to purify the tracer and to eliminate matrix differences between standards and unknowns. It has been shown repeatedly that neither buffered human serum albumin nor an animal serum matrix is an adequate substitute for a human serum matrix (8, 1215). Further corroboration for an upper limit of approximately 3 or 4 milli-int. unitsfL comes from studies performed with various “nonstandard” methodologies (16-20), including immunoassays based on chemiluminescence, separation by wick chromatography, or sample concentration, as well as the modern cytochemical In addition, experiments bioassay (Table 1). in which samples were collected from normal adults and assayed by radioimniunoassay, both before and after suppression by treatment with thyroid hormones, provide evidence that the true median value for circulating thyrotropin is barely 1 milhi-int. unitJL (21). Clinical Significance From studies performed with the research assays, a great deal has been learned as a result of optimizing for sensitivity and specificity, thanks to the superior delineation of the normal reference interval that this entails. Thus, we now know that thyrotropin exhibits a circadian rhythm, with lower values in the morning and higher values near midnight (15, 22). We know more exactly the pattern exhibited by thyrotropin during pregnancy (16,23). We know also that there are significant age- and sex-related variations, with an increased prevalence of moderately increased thyrotropin concentrations in ostensibly normal elderly women (24-26). The very concept of “subclinical hypothyroidism”-a condition characterized by “normal” concentrations of thyroid hormones in the presence of an increased thyrotropin concentration-had to await the development of fully optimized radioimmunoassays for thyrotropin (27). Of even greater clinical significance is the fact that improvements in thyrotropin methodology have reduced the need for TRF tests to clarify borderline results. Optimizing a thyrotropin assay not only tends to lower values throughout the reference interval for normal, it also tends to improve the precision and discrimination that can be achieved in this region, rendering more meaningful the comparison of basal values with those measured after TRF stimulation (14, 28, 29). Indeed, Bigos et a!. (30) showed excellent correlation between such values, as measured with one of the research assays. Wide and Dahlberg (29) also showed improved correlation when steps were taken to optimize their thyrotropin assay for low-end sensitivity. Furthermore, although perfect discrimination of hyperthyroid samples is not yet possible, in well-optimized thyrotropin assays a basal value exceeding the median for norma! persons renders a diagnosis of hyperthyroidism highly unlikely (31). Practical Considerations major practical considerations bear on the interpreannotation of thyrotropin results at the stage of them to physicians. First, reference intervals for thyrotropin that are derived from blood donors or laboratory volunteers are clearly inappropriate for newborns and may be inappropriate for certain other groups as well. In particular, they should not be blindly adopted for the interpretation of results on samples from a maternity ward (16,23) or a geriatric unit (25, 26). Thus, depending on the reference group, 3 or 4 miuli-int. units/L may not represent the true upper limit of normal. Accordingly, each laboratory should establish by experiment its own expected values (32). Moreover, clinical decision limits should respect the distribution of values and disease in the local patient population. Decision limits designed to optimize the predictive value of a test may not correspond to conventional reference interval limits based on 95% coverage for normal individuals. Use of the thyrotropin assay as a primary screen for congenital hypothyroidism provides a classic illustration of this point (33). Finally, one must remember that in some contexts a thyrotropin result that is only slightly above the reference interval for normal individuals may not by itself warrant taking any corrective action (27). Three tation al reporting In deciding whether to accept or reject a thyrotropin kit, an important criterion at the stage of method evaluation ought to be its ability to yield values for a normal adult population that are similar to values found by well-established, fully optimized referencemethods forthyrotropin (1).. The true upper limit of normal for young men and nonpregnant women in good health is close to 3 or 4 milli-int. unitsfL, according to the literature summarized in this report. References 1. Musto JD, Pizzolante thyrotropin measurement (1984). JM, Chesarone VP. A comment on the and evaluation. Clin Chem 30, 329-330 2. Patel YC, Burger HG, Hudson B. Radioimmunoassay of serum thyrotropin: Sensitivity and specificity. J Clin Endocrinol Metab 33, 768-774 (1971). 3. Hall R, Amos J, Ormston BJ. Radioimmunoassay of human serum thyrotropin. Br Med J 1, 582-585 (1971). 4. Marschner I, Erhardt FW, Scriba PC. Ringversuch zur radioimmunologischen Thyrotropin-bestimmung (hTSH) im Serum. J Gun Chem Clin Biochem 14, 345-351 (1976). 5. Ridgway EC, Kourides LA, Maloof F. Thyrotropin. In Nuclear Medicine In Vitro, B Rothfeld, Ed., J.B. Lippincott, Philadelphia, PA, 1974, pp 205-219. 6. Pekary AE, Hershman JM, Parlow AF. A sensitive and precise radioimmunoassay for human thyroid-stimulating hormone. J Clin Endocrinol Met.ab 41, 676-679 (1975). 7. Bigos ST, Pekary AE, MacLean J, et al. A thyrotropin radioimmunoassay kit evaluated vs two reference assays. Gun Chem 30, 437-439 (1984). 8. Wood WG. A “same day” TSH radioimmunoassay kit with acceptableprecision and accuracy. NucCompact 11, 60-63 (1980). 9. Hershman JM (Ed.). Endocrine Pathophysiology: A PatientOrien,ted Approach. Lea and Febiger, Philadelphia, PA, 1977, p 39. 10. Williams RH (Ed.). Tcrtbook of Endocrinology, 6th ed., W.B. Saunders, Philadelphia, PA, 1981, p 137. 11. Wintrobe MM (Ed.). Harrison’s Principles of Internal Medicine, 7th ed., McGraw-Hill, New York, NY, 1974, p 470. 12. Erhardt FW, Marschner I, Wood WG. Quality control and some sources of error in radioimmunoassays. Pediatr Adolesc Endocrinol 12, 170-180 (1983). 13. Hall R. Separation of bound from free labelled antigen. In Radioimmunoassay Methods, KE Kirkham, WM Hunter, Eds., Churchill Livingstone, Edinburgh, U.K., 1971, pp 313-314. 14. Kubasik NP, Ricotta M, Hunter T, Sine HE. Clinical evaluation of two thyrotropin radioimmunoassay kits: Human serum matrix calibrators and bovine serum matrix calibrators. Clin Chem 27, 504 (1981). Letter. 15. Parker DC, Pekary AE, Hershman JM. Effect of normal and reversed sleep-wake cycles upon nyctohemeral rhythmicityof plasma thyrotropin: Evidencesuggestiveofan inhibitory influence in sleep. J Clin Endocrinol Metab 43, 318-329 (1976). 16. Weeke J, Dybk,jr L, Granlie K, et al. A longitudinal study of serum TSH, and total and free iodothyronines during normal pregnancy. Acta Endocrinol 101, 531-537 (1982). 17. Nisula BC, Louvet J-P. Radioimmunoassay of thyrotropin concentrated from serum. J Clin Endocrinol Metab 46, 729-733 (1978). 18. Weeks I, Sturgess M, Siddle K, et al. A high sensitivity immunochemiluminometric assay for human thyrotrophin. Clin Endocrinol 20, 489-495 (1984). 19. Petersen V, Smith BR, Hall R. A study of thyroid stimulating activity in human serum with the highly sensitive cytochemical bioassay. J Clin Endocrinol Metab 41, 199-202 (1975). 20. Faglia G, Bitensky L, Pinchera A, et a!. Thyrotropin secretion in patients with central hypothyroidism: Evidence for reduced biological activity of immunoreactive thyrotropin. J Clin Endocrinol Metab 48, 989-998 (1979). 21. Adams DD, Kennedy TH, Utiger RD. Comparison of bioassay and immunoassay measurements of serum thyrotropin (TSH) and study of TSH levels by immunoassay of serum concentrates. J Clin Endocrinol Metab 34, 1074-1079 (1972). CLINICAL CHEMISTRY, Vol. 31, No. 2, 1985 297 22. Weeke J. The influence of the circadian thyrotropin rhythm on the thyrotropin response to thyrotropin-releasing hormone in normal subjects. Scand J Glin Lab Invest 33, 17-20 (1974). 23. Braunstein GD, Hershman JM. Comparison of serum pituitary thyrotropin and chorionic gonadotropin concentrations throughout pregnancy. J Clin Endocrinol Metab 42, 1123-1126 (1976). 24. Nystrom E, Bengtsson C, Lindquist 0, et a!. Thyroid disease and high concentration of serum thyrotrophin in a population sample of women: A 4-year follow-up. Acts Med Scand 210, 39-46 (1981). 25. Sawin CT, Chopra D, Azizi F, et a!. The aging thyroid: Increased prevalence of elevated serum thyrotropin levels in the elderly. JAm Med Assoc 242, 247-250 (1979). 26. Tunbridge WMG, Evered DC, Hall R, et a!. The spectrum of thyroid disease in a community: The Whickham survey.Glin Endocrinol 7, 481-493 (1977). 27. Ridgway EC, Cooper DS, Walker H, et al. Peripheral responses to thyroid hormone before and after L-thyroxine therapy in patients with subclinical hypothyroidism. J Glin Endocrinol Metab 53, 1238-1242 (1981). 28. Pekary AE, Hershman JM, Sawin CT. Linear modulation of serum thyrotropin by thyroid hormone treatment in hypothyroidism. Acts Endocrinol 95, 472-478 (1980). 29. Wide L, Dahlberg PA. Quality requirements of basal S-TSH assays in predicting an S-TSH response to TRH. Scand J GUn Lab Invest 155, 101-110 (1980). 30. Bigos ST, Ridgway EC, Kourides IA, Maloof F. Spectrum of pituitary alterations with mild and severe thyroid impairment. ,J Clin Endocrinol Metab 46, 317-325 (1978). 31. Kubasik NP, Same DO, Brodows RG, Sine HE. Assay of thyrotropin in hyperthyroidism. Clin Chem 29, 1688(1983). Letter. 32. Tunbridge WMG, Hall R. Thyrohf-stimulating hormone. In Hormone Assays and Their Clinical Application, 2nd ed., JA Loraine, ET Bell, Eds., Churchill Livingstone, London, U.K., 1976, 175-192. 33. Petersen PH, Rosleff F, RasmussenJ, Hobolth N. Studies on the required and analytical quality of TSH measurements in screening for congenital hypothyroidism. Scand J Clin Lab Invest 155, 85-93 (1980). CLIN. CHEM. 31/2, 298-302 (1985) Preparation of Urine Samples for Liquid-Chromatographic Determination of Catecholamines: Bonded-Phase Phenylboronic Acid, Cation-Exchange Resin, and Alumina Adsorbents Compared Alan H. B. Wu and Terrie G. Gornet We compared results for the liquid-chromatographic determination of free norepinephrine and epinephrine in urine after purifying the catecholsby the followingmethods: (a)acidwashed alumina, (b) weak cation-exchange resin (WCX), (C) a combination of weak cation-exchange resin followed by alumina (WCX-alumina), and (c commercially available phenylboronic acid adsorbent. We evaluated analytical specificity, sensitivity, recovery, and turnaround time. The WCX-alumina combination produced the most sensitive and specific chromatograms for urinary catecholamines; the other methods took less processing time. Neither WCX nor alumina alone was suitable for routine work because of chromatographic interferences in a significant proportion of urines. The phenylboronic acid method is adequately sensitive and specific for norepinephrine and epinephrine, and samples can be assayed faster. Thus it provides a compromise between the high analytical performance of the WCXalumina method and the speed of the WCX and alumina methods. liquid chromatography is widely used for total and free urinary catecholamines with ei- “High-pressure” Department of Pathology and Laboratory Medicine, University of Texas Health Science Center at Houston, Houston, TX 77025. Presented at the 36th national meeting of the AACC, July 1984, Washington, DC. Received September 4, 1984; accepted November 6, 1984. 298 ic towards catecholamines, they cannot be connected to a source of low pressure for rapid isolation and elution. We evaluated a simultaneous dual-step purification procedure involving the use of chemically bonded materials (6) (ion-exchange and phenylboronic acid adsorbents: Bond Elut; Analytichem International, Harbor City, CA 90710) for rapidly isolating free catecholamines and compared results by this method with results obtained on using weak cation-exchange resin and alumina. Materials and Methods AddItIonal Keyphrases: norepinephrine . epinephrine . cer pheochromocytoma neuroblastoma . screening measuring ther electrochemical or fluorometric detectors. In most electrochemical methods, the amines are isolated by a two-step procedure before injection into the chromatograph. Materials used for this include alumina, cation-exchange resins, and Sephadex (1-4). With each of these techniques the catecholamines are adequately separated from other urine components, but each is time consuming. Boric acid gels, which adsorb compounds containing ciadiol groups, have been used recently to isolate urinary catecholamines (2,5). Although these gels are higi’y specif- CLINICAL CHEMISTRY, Vol. 31, No. 2, 1985 Standards and reagents. Norepinephrine (NE), epineph- rime (E), dopamine, and the internal standard, 3,4-dihydroxybenzylaimne, were from Sigma Chemical Co., St. Louis, MO 63178, as were all of the compounds in the interference study.’ Acid-washed alumina (Al203), prepared by the method of Anton and Sayre (7), was from Bioana!ytical Systems, West Lafayette, IN 47905. The weak cation1 Nonstandard abbreviations: NE, norepinephrine; E, epinephrine; WCX, weak cation-exchange; PBA, phenylboronic acid; PSA, primary and secondary amine ion-exchange.