Research. Although thyrotropin (thyroidstimulated

Transcription

1 Research OBSTETRICS Free T4 immunoassays are flawed during pregnancy Richard H. Lee, MD; Carole A. Spencer, PhD; Jorge H. Mestman, MD; Erin A. Miller, BS; Ivana Petrovic, MS; Lewis E. Braverman, MD; T. Murphy Goodwin, MD OBJECTIVE: The purpose of this study was to evaluate the diagnostic accuracies of 2 free thyroxine immunoassays during pregnancy. STUDY DESIGN: Serum was collected from healthy, thyroid peroxidase antibody-negative women during each trimester and nonpregnant controls. Thyrotropin, total T4 (TT4), free T4 index (FT4I), and 2 different FT4 immunoassays were studied. RESULTS: As expected, TT4 was elevated in all 3 trimesters compared to controls (P.001). FT4I was elevated in the 1st trimester as compared with controls (P.05) and returned to the nonpregnant range in the 2nd and 3rd trimesters. In contrast, 1st trimester FT4 immunoassay values were either comparable or lower than controls and by the 2nd and 3rd trimesters had decreased to approximately 65% of controls. CONCLUSION: Neither FT4 immunoassay accurately reflects established free T4 changes during pregnancy. TT4 and the FT4I retained an appropriate inverse relationship with TSH throughout pregnancy and appear to provide a more reliable free T4 estimate. Key words: free, immunoassay, pregnancy, thyroid, thyroxine Cite this article as: Lee RH, Spencer CA, Mestman JH, et al. Free T4 immunoassays are flawed during pregnancy. Am J Obstet Gynecol 2009;200:260.e1-260.e6. From the Departments of Obstetrics and Gynecology (Drs Lee, Mestman, and Goodwin and Ms Miller) and Medicine (Drs Spencer and Mestman and Ms Petrovic), Women s and Children s Hospital, Los Angeles County University of Southern California Medical Center, Los Angeles, CA, and the Department of Medicine, Boston University Medical Center, Boston, MA (Dr Braverman). Received May 2, 2008; revised Aug. 9, 2008; accepted Oct. 7, 2008 Reprints not available from the authors /free 2009 Mosby, Inc. All rights reserved. doi: /j.ajog For Editors Commentary, see Table of Contents Although thyrotropin (thyroidstimulated hormone; TSH) is generally considered the primary test for evaluating thyroid status during pregnancy, in some situations it is imperative for clinicians taking care of pregnant patients to have access to an accurate and reliable way to estimate free thyroxine (FT4) concentrations. This is especially true in light of reports suggesting that isolated hypothyroxinemia may be associated with impaired fetal psychomotor development and decreased intelligent quotient. 1,2 Because of the pregnancyrelated changes in the thyroid axis due to increases in thyroxine-binding globulin (TBG) and the rise in human chorionic gonadotropin (hcg), a stimulus for increased thyroid synthesis and secretion of T4 and triiodothyronine (T3), the free thyroxine index (FT4I) was developed (total T4 mathematically corrected for TBG). Over the years, it has shown changes in pregnancy consistent with the expected effects of TBG and hcg. Pregnancy poses unique challenges to FT4 methodologies. Current FT4 immunoassays are actually FT4 estimate tests that do not measure FT4 directly and are known to be sensitive to alterations in binding proteins. 3 It is thus not surprising that pregnancy-induced increases in thyroxine-binding globulin, as well as decreases in albumin, can influence FT4 immunoassays in a method-specific manner. 4,5 Whereas the law of mass action dictates that some lowering of FT4 should be expected in pregnancy because of the high TBG state, the unexpected high prevalence of low third-trimester FT4 immunoassay values versus the nonpregnant lower limit questions its reliability. 5,6 It should be noted that when FT4 is measured by the current reference methods (equilibrium dialysis, gas chromotography/mass spectrometry), there is less than a 10% decrease in thirdtrimester mean versus nonpregnant controls. 7 Thus, FT4 immunoassays may not reliably identify hypothyroxinemic pregnant patients, especially in the first trimester when maternal FT4 concentrations are important for early fetal brain maturation before the fetal thyroid functions. The performance of 2 commonly used FT4 immunoassays in a select population of clinically euthyroid subjects in the nonpregnant state and through each trimester of pregnancy was evaluated. The TT4 and FT4I have been used for clinical management of thyroid disease in pregnancy over the past 3 decades with consistent results. In recent years, FT4 immunoassays have largely replaced these 2 tests; nevertheless, there are limited methodspecific and trimester-specific data for these direct immunoassays. These limitations have caused considerable confusion about the interpretation of thyroid tests during pregnancy. The primary purpose of this study was to compare the diagnostic performance of 2 different FT4 immunoassays to traditional approaches for estimating free thyroxine (total T4 and FT4I) relative to the physiological TSH changes that are known to occur throughout pregnancy. 260.e1 American Journal of Obstetrics & Gynecology MARCH 2009

2 Obstetrics Research TABLE 1 Nonpregnant vs pregnancy group characteristics of TPO Ab-negative subjects Test Nonpregnant (n 93) First trimester (n 111) Second trimester (n 47) Third trimester (n 63) P Gestational age (wk) Urinary iodine ( g/dl) TSH (miu/l) (method A) 1.60 ( ) 1.20 ( ) a 1.60 ( ) 1.60 ( ).001 TSH (miu/l) (method B) d 2.10 ( ) 1.32 ( ) a 1.79 ( ) 1.90 ( ).001 n 26 n 77 n 38 n 43 TT4 ( g/dl) (method A) 8.70 ( ) b ( ) ( ) ( ).001 FT4 immunoassay (ng/ dl) (method A) 1.10 ( ) c 1.20 ( ) 0.85 ( ) 0.89 ( ).001 FT4 immunoassay (ng/ 1.30 ( ) b 0.97 ( ) a 0.69 ( ) 0.70 ( ).001 dl) (method B) d n 30 n 76 n 36 n 40 FT4I ( g/dl) (method A) 7.70 ( ) 8.70 ( ) a 7.50 ( ) 8.00 ( ).001 THBR (TBG estimate) 1.10 ( ) b 1.19 ( ) 1.41 ( ) 1.43 ( ).001 CI, confidence interval; FT4, free thyroxine; FT4I, free thyroxine index; SD, standard deviation; TBG, thyroxine-binding globulin; THBR, thyroid hormone-binding ratio; TPOAb, thyroid peroxidase antibodies; TSH, thyroid-stimulating hormone; TT4, total thyroxine. Data are presented as mean SD or median (95% CI). Post hoc analysis vs nonpregnant. a P.002 vs first trimester; b P.001 vs all 3 trimesters of pregnancy; c P.001 vs second and third trimesters; d Samples for method B were analyzed if there was remaining serum, leading to the decreased overall sample size for this methodology; otherwise, the sample size for each test is indicated by the top row. MATERIALS AND METHODS Subjects Pregnant women who were obtaining prenatal care at the Los Angeles County and University of Southern California Women s and Children s Hospital were recruited in the first trimester of pregnancy. Participants had sequential samples of venous blood drawn in the first trimester ( 14 weeks), second trimester ( /7 weeks), and in the third trimester ( 28 weeks). A spot urine was obtained to measure iodine at each visit. All participants were over the age of 18 and had singleton, viable pregnancies confirmed by ultrasound performed in the first trimester. Exclusion criteria included: known thyroid disease, autoimmune disease, multiple gestation, diabetes, hyperemesis gravidarum, or pregnancy within the last 12 months. Premenopausal nonpregnant subjects aged years were recruited from the gynecology clinic and matched for ethnicity. Exclusion criteria included known thyroid disease, autoimmune disease, use of hormone-containing contraceptives, or pregnancy within the last 12 months. Both blood and urine were collected once. These subjects were different from the pregnant subjects. Laboratory methods TSH, TT4, a thyroid hormone-binding ratio (THBR) estimate of TBG, and an FT4 estimate made by immunoassay were all measured using the Elecsys (Roche, IN) platform (method A). Free hormone index corrections for TBG effects (FT4I) were calculated by dividing total hormone values by the THBR result. When sufficient specimen was available, TSH and FT4 immunoassay values were also obtained using the Tosoh A1A-600 analyzer (Ramsey, MN) (method B). Thyroid peroxidase (TPO) antibodies were measured using the Nichols Advantage (San Juan Capistrano, CA) platform as well as the Kronus (Boise, ID) radioassay method. TPO antibody-positive subjects were excluded from the analysis in accordance with recommendations for establishing normal ranges of thyroid function. 8 Batches of specimens were separated and the sera stored at 4 C for no longer than 3 days before being analyzed. Spare specimen was archived at -20 C for any further testing. The reference ranges established by the respective laboratories for the methods were: TSH (methods A and B), miu/l: TT4, g/dl; THBR, ; FT4I, g/dl; and FT4, immunoassay (method A, ng/dl; and method B, ng/dl). Iodine was measured as previously described, and iodine sufficiency for the study population was defined as a median urinary concentration greater than 5 g/dl. 9 Statistical analysis Comparison of demographic data between pregnant and nonpregnant subjects was performed using independent t test and 2. Comparison of thyroid function tests was performed using nonparametric analysis of variance (ANOVA). If the ANOVA comparison was considered significant, post hoc analysis using the Tamhane T2 adjustment was performed to determine significant pairwise comparisons. A P value.05 was considered significant. Statistical analysis was per- MARCH 2009 American Journal of Obstetrics & Gynecology 260.e2

3 Research Obstetrics FIGURE 1 Individual TSH values and group medians for nonpregnant vs pregnant patients as measured by 2 different methods (A and B) Log TSH miu/l method A B A B A B A B group nonpregnant 1st trimester 2nd trimester 3rd trimester a P <.001 vs nonpregnant formed using SPSS v.13.0 (SPSS, Inc, Chicago, IL). This study was approved by our institutional review board at the University of Southern California. RESULTS Between January 2004 and April 2007 a total of 134 pregnant subjects and 107 nonpregnant subjects were enrolled. Nonpregnant subjects were older ( vs years; P.001) and more parous ( vs ; P.02). There was no difference in gravidity and percentage with Hispanic ethnicity between the 2 groups (results not shown). A high percentage of the population was Hispanic: 98.2% and 94.0% in the nonpregnant and pregnant groups, respectively. TPO antibodies were detected in 23/134 (17.2%) pregnant subjects and 14/107 (13.1%) nonpregnant subjects. After excluding TPO antibodypositive subjects, 111 pregnant patients a a a Significance relative to the nonpregnant group. The shaded areas represent the nonpregnant reference range; sample sizes: nonpregnant (A 93, B 26), 1st trimester (A 111, B 77), 2nd trimester (A 47, B 38), and 3rd trimester (A 63, B 43). TSH, thyroid-stimulating hormone and 107 nonpregnant controls remained for analysis. Of the 111 pregnant subjects recruited in the first trimester, follow-up samples were obtained in 47 in the second trimester and 63 in the third trimester. Both the nonpregnant and pregnant subjects were iodine sufficient among all trimesters. The mean gestational age in the first trimester was 9.0 weeks, second trimester weeks, and third trimester weeks (Table 1). Figure 1 shows individual log TSH values, together with group medians, for nonpregnant versus pregnant patients versus the nonpregnant laboratory reference range. TSH values were lower in the first trimester compared with nonpregnant controls and returned to nonpregnant values in the second and third trimesters using both methods. Ten patients had first-trimester TSH values below 0.3 miu/l using method A; these patients also had low values measured by method B. Only 3 patients (2.7%) had values above 2.5 miu/l (2.6, 2.6, and 3.1 miu/l) using method A. Method B values available for 2 of these patients were 3.3 and 3.1 miu/l, respectively. Figure 2 shows the different approaches for estimating free T4 status (total T4, FT4I, and FT4 immunoassays). As expected, the estrogen-mediated rise in TBG resulted in higher TT4 values relative to nonpregnant in all 3 trimesters (P.001). The TBG effect was present very early in gestation, because first-trimester median TT4 values were not statistically different from secondtrimester values (10.60 [ ] vs [ ] g/dl, P.74). The FT4I appeared to successfully correct for the TBG effects in that first-trimester values were higher than nonpregnant subjects due to the peak hcg stimulatory effect, returning to nonpregnant levels in the second and third trimesters. The inverse relationship between free thyroxine and TSH appeared appropriate for all methods (Figure 2). In contrast, neither of the 2 FT4 immunoassays demonstrated the expected first-trimester increase in FT4 relative to the nonpregnant controls. In fact, the median FT4 measured by method B was significantly lower than nonpregnant in the first trimester (1.30 [ ] vs 0.97 [ ] ng/dl, P.001). In the second and third trimesters, a high percentage of women had FT4 immunoassay results below the manufacturer s lower limit (Table 2). By method A, 7.5% of nonpregnant subjects and 5.4% of first-trimester subjects had FT4 values below the manufacturer s lower limit, whereas 66.0% had low FT4 in the second trimester and 57.1% in the third trimester. For method B, 0% of nonpregnant and 6.6% of first-trimester subjects were below the manufacturer s lower limit, while 63.9% and 67.5% were below this threshold in the second and third trimesters, respectively. In contrast, no subjects in all 4 groups had an FT4 index or TT4 below the manufacturer s lower limit. COMMENT We have demonstrated that in an iodinesufficient, TPO antibody-negative popu- 260.e3 American Journal of Obstetrics & Gynecology MARCH 2009

4 Obstetrics Research FIGURE 2 TT4, FT4I, and FT4 immunoassay A B results TT4/FT4I µg/dl Total T4 (TT4) Low TSH < 0.3 miu/l Free T4 Index lation, the measurement of free thyroxine by 2 different immunoassays did not reflect the expected physiological hcgmediated rise in the first trimester. In addition, the expected return to nonpregnant concentrations in the second and third trimesters was not seen. Instead, a continued decline in FT4 was identified, resulting in 57-68% of women falling into a range that would be classified as hypothyroxinemic by the manufacturer s recommended ranges. In contrast, FT4 Immunoassay (A) FT4 Immunoassay (B) a a a a a a a a a NP 1st 2nd 3rd NP 1st 2nd 3rd NP 1st 2nd 3rd NP 1st 2nd 3rd a P <.05 vs NP the FT4I performed as expected, with a physiologic increase in the first trimester with normalization to nonpregnant levels in the second and third trimesters. 10 This pattern of change during pregnancy corresponds to that described using the gold standard methods of equilibrium dialysis and tandem mass spectrometry. 7,11,12 Because these latter methods are expensive and technically demanding, they are not suitable for widespread clinical use, but to the extent that they FT4 Immunoassay ng/dl 1.2 Individual and median values for nonpregnant (NP) vs 1st, 2nd, and 3rd trimesters for total thyroxine (TT4) vs 3 free thyroxine estimate tests: free T4 index (FT4I) and 2 different immunoassays (A Roche Elecsys; B Tosoh A1A). a Significance relative to the nonpregnant group. Patients with TSH below 0.3 miu/l are indicated by solid symbols. Sample sizes are as listed in Table 1. FT4, free thyroxine; T4, thyroxine; TSH, thyroid-stimulating hormone validate the trends seen with the FT4I, they support its use. One of the impediments to the use of the FT4I in current practice is that it requires 2 distinct tests, adding to expense. Alternatively, TT4 may serve as an appropriate approximation of FT4. As shown in Table 3, the TT4, whether measured directly or indirectly, has shown remarkably consistent ranges throughout pregnancy over many years approximately % of nonpregnant values. Adjusting the TT4 in pregnancy by a factor of 1.5 compared with nonpregnant reference ranges yields a workable estimate of FT4. In practice, the nonpregnant reference range of g/dl would become g/dl. Applying this adjustment to our study population (Table 2), there was no statistical difference in the percentage of subjects below the reference range for the FT4 index and the pregnancyadjusted TT4. It should be noted that in the seminal study of Haddow et al, showing that maternal thyroid deficiency during pregnancy negatively impacted neuropsychological development of the child, mean TT4 was 7.5 g/dl. 13 Since TT4 is an inexpensive assay well suited to automation, the simple adjustment described above would provide clinicians with an inexpensive and readily available method for estimating free thyroxine. 14 Future studies are needed to determine the financial implications of our results. The FT4 immunoassays may not work well during pregnancy due to their sensitivity to alterations in binding proteins; the pregnancy-induced increase in TBG and decrease in albumin can cause alterations in a method-specific manner. 4,5 TABLE 2 Subjects with thyroid values below the manufacturer s lower reference limit Variable FT4 method A a FT4 method B b FT4 index c TT4 d TT4 1.5 e Nonpregnant 7/93 (7.5) 0/30 (0) 0/93 (0) 0/93 (0) 1st trimester 6/111 (5.4) 5/76 (6.6) 0/111 (0) 0/111 (0) 1/111 (0.9%) 2nd trimester 31/47 (66.0) 23/36 (63.9) 0/47 (0) 0/47 (0) 0/47 (0) 3rd trimester 36/63 (57.1) 27/40 (67.5) 0/63 (0) 0/63 (0) 0/63 (0) Results are expressed as n/n (%). a Free thyroxine (FT4) (0.93 ng/dl); b FT4 (0.75 ng/dl); c FT4 index (4.5 g/dl); d Total thyroxine (TT4) below (4.5 g/dl); e TT4 multiplied by ratio of 1.5 to adjust for pregnancy (6.75 g/dl). MARCH 2009 American Journal of Obstetrics & Gynecology 260.e4

5 Research Obstetrics TABLE 3 Three decades of TT4 measurements Mean nmol/l Mean g/dl Trimester Trimester Mean % nonpregnant Year Author Country Iodine Methodology control 1979 Yamamoto CE 10:459 Japan Sufficient RIA Whitworth CE 17:307 UK Sufficient RIA Glinoer JCEM 71:276 Belgium Deficient RIA Ballabio JCEM 73:824 Thailand Deficient RIA Berghout CE 41:375 The Netherlands Deficient Immunoassay Liberman JCEM 83:3545 USA Sufficient Immunoassay Kung CE 53:725 Hong Kong Deficient IMA Soldin CCA 347:61 USA Sufficient IMA Soldin CCA 347:61 USA Sufficient Mass spectrometry Means TT4, total thyroxine. These findings may also occur in other causes of increased TBG concentrations, such as the estrogen-induced increase in women on the birth control pill or estrogen use in menopausal women. The need for confidence in estimating free thyroxine in pregnancy has been highlighted by studies linking subclinical hypothyroidism or hypothyroxinemia with neurodevelopmental delay. 14 Although most interest has focused on women with an elevated TSH and normal thyroxine, a subset of pregnant women has been described with hypothyroxinemia and normal TSH, which also has been linked to delayed neurodevelopment in offspring. 1,2 It is clear that for the purposes of ongoing research to better understand these putative associations, as well for day-to-day clinical use, a more reliable means of estimating FT4 in pregnancy than the FT4 immunoassay is needed. We acknowledge the limitations of our study: our primarily Hispanic population, the demographic differences between our pregnant and nonpregnant populations, our small sample size, and our lack of pregnancy outcomes. With the vast majority of our population being of Hispanic ethnicity, whether these same findings can be applied to other ethnicities will need to be further studied. Furthermore, although it would be optimal to study the same patient before pregnancy and during pregnancy, the feasibility of approaching such a study would be difficult at our institution. As there was little known about the performance of these tests prior to this study, our initial goal was to recruit over 120 subjects per trimester as recommended by The National Academy of Clinical Biochemistry. 8 We acknowledge we were unable to reach this number; however, we believe the presented data demonstrated a dramatic difference in the performance of the 2 FT4 immunoassays versus the FT4I and TT4. Finally, although the prenatal care course was reflected by patients with otherwise uncomplicated courses, the same cannot be concluded about their actual intrapartum/postpartum course without actual delivery outcomes. However, we do believe that because the patients had uncomplicated prenatal care, they do represent a cohort of the population that would otherwise be considered both normal and healthy. In conclusion, FT4 immunoassays in pregnant women yield results that do not correspond to well-established pregnancyrelated changes in the thyroid axis; nor do they correspond to estimates of FT4 by equilibrium dialysis and tandem mass spectrometry. The results diverge so significantly during the second and third trimesters that the vast majority of women would be diagnosed incorrectly as hypothyroxinemic by laboratory criteria alone. Each specific immunoassay needs to have normals and abnormals determined for the pregnant state or immunoassays may underestimate FT4. 15,16 Alternatively, the FT4 index or the TT4 adjusted for pregnancy are reliable methods of estimating free thyroxine status in pregnancy. f REFERENCES 1. Pop VJ, Brouwers EP, Vader HL, Vulsma T, van Baar AL, de Vijlder JJ. Maternal hypothyroxinaemia during early pregnancy and subsequent child development: a 3-year follow-up study. Clin Endocrinol (Oxf) 2003;59: Pop VJ, Kuijpens JL, van Baar AL, et al. Low maternal free thyroxine concentrations during early pregnancy are associated with impaired psychomotor development in infancy. Clin Endocrinol (Oxf) 1999;50: Fritz KS, Wilcox RB, Nelson JC. Quantifying spurious free T4 results attributable to thyroxine-binding proteins in serum dialysates and ultrafiltrates. Clin Chem 2007;53: Roti E, Gardini E, Minelli R, Bianconi L, Flisi M. Thyroid function evaluation by different commercially available free thyroid hormone measurement kits in term pregnant women and their newborns. J Endocrinol Invest 1991;14: Sapin R, d Herbomez M. Free thyroxine measured by equilibrium dialysis and nine immunoassays in sera with various serum thyrox- 260.e5 American Journal of Obstetrics & Gynecology MARCH 2009

6 Obstetrics Research ine-binding capacities. Clin Chem 2003;49: Sapin R, D Herbomez M, Schlienger JL. Free thyroxine measured with equilibrium dialysis and nine immunoassays decreases in late pregnancy. Clin Lab 2004;50: Soldin OP, Tractenberg RE, Hollowell JG, Jonklaas J, Janicic N, Soldin SJ. Trimester-specific changes in maternal thyroid hormone, thyrotropin, and thyroglobulin concentrations during gestation: trends and associations across trimesters in iodine sufficiency. Thyroid 2004; 14: Baloch Z, Carayon P, Conte-Devolx B, et al. Laboratory medicine practice guidelines. Laboratory support for the diagnosis and monitoring of thyroid disease. Thyroid 2003;13: Hollowell JG, Staehling NW, Hannon WH, et al. Iodine nutrition in the United States. Trends and public health implications: iodine excretion data from National Health and Nutrition Examination Surveys I and III ( and ). J Clin Endocrinol Metab 1998;83: Pearce EN, Oken E, Gillman MW, et al. Association of first-trimester thyroid function test values with thyroperoxidase antibody status, smoking, and multivitamin use. Endocr Pract 2008;14: Kahric-Janicic N, Soldin SJ, Soldin OP, West T, Gu J, Jonklaas J. Tandem mass spectrometry improves the accuracy of free thyroxine measurements during pregnancy. Thyroid 2007;17: Mandel SJ, Spencer CA, Hollowell JG. Are detection and treatment of thyroid insufficiency in pregnancy feasible? Thyroid 2005;15: Haddow JE, Palomaki GE, Allan WC, et al. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. N Engl J Med 1999;341: Demers LM, Spencer CA. Laboratory medicine practice guidelines: laboratory support for the diagnosis and monitoring of thyroid disease. Clin Endocrinol (Oxf) 2003;58: Lambert-Messerlian G, McClain M, Haddow JE, et al. First- and second-trimester thyroid hormone reference data in pregnant women: a FaSTER (First- and Second-Trimester Evaluation of Risk for aneuploidy) Research Consortium study. Am J Obstet Gynecol 2008;199:62.e Casey BM, Dashe JS, Spong CY, McIntire DD, Leveno KJ, Cunningham GF. Perinatal significance of isolated maternal hypothyroxinemia identified in the first half of pregnancy. Obstet Gynecol 2007;109: MARCH 2009 American Journal of Obstetrics & Gynecology 260.e6