The reference range and within-person variability of thyroid stimulating hormone during the first and second trimesters of pregnancy

Transcription

1 03 Haddow cpp 5//04 2:02 PM Page ORIGINAL PAPER The reference range and within-person variability of thyroid stimulating hormone during the first and second trimesters of pregnancy JE Haddow, GJ Knight, GE Palomaki, MR McClain, AJ Pulkkinen J Med Screen 2004;:70 74 See end of article for authors affiliations Correspondence to: James E Haddow, Foundation for Blood Research, PO Box 90, Scarborough, ME , USA; jhaddow@fbr.org Accepted for publication 22 July Objective: To further explore first and second trimester reference ranges for thyroid stimulating hormone (TSH) and examine within-person variability of TSH and thyroid peroxidase (TPO) antibody. Setting: Women coming for routine prenatal care in early pregnancy agreed to participate in a trial of integrated serum screening for Down s syndrome. Two serum samples were obtained from each woman, one each in the first and second trimesters. These samples were also available for TSH and TPO measurements in the present study. Methods: TSH and TPO antibody measurements were performed in 26 women with ultrasounddated pregnancies who provided serum samples in both trimesters. TSH reference ranges were established for the entire cohort and for the antibody-negative subgroup. Within-person variability of TSH measurements between trimesters was examined. Results: Median TSH values are lower in the first trimester than in the second (.00 versus.29 miu/l), but 98th centile values are higher (5.20 versus 4.8 miu/l). High correlation exists between individual women s first and second trimester TSH measurements (r=0.75, r 2 =0.56, p<0.00). Among 23 women with TSH values above the 98th centile in the second trimester, 7 (74%) were over the 95th centile in the first trimester. TPO antibody measurements are also highly correlated between trimesters (r=0.97, r 2 =0.94). Conclusion: Proper interpretation of TSH measurements during pregnancy requires that laboratories establish and monitor appropriate reference ranges. TSH levels show high within-person consistency between trimesters. INTRODUCTION In 999, our research group reported an association between undiagnosed thyroid deficiency in the mother during early pregnancy (usually mild) and neurodevelopmental problems in the offspring. In that study, the primary determinant of thyroid deficiency was a thyroid stimulating hormone (TSH) level at or above the 98th centile for the study population of 25,26 viable pregnancies. Although the potential for improving fetal neurodevelopment by early diagnosis and treatment has not yet been defined, our study and several others dealing with various thyroid-related pregnancy complications 2 7 have led to increased awareness on the part of both obstetricians and pregnant women as to the need for taking this condition into account during routine evaluations. Professional groups have issued conflicting guidelines about population screening, 8 however, indicating that consensus on an optimal policy has not yet been reached. Regardless of which recommendations are followed, there continues to be a need for better-defined reference ranges for commonly used tests of thyroid function during pregnancy. This is most important for TSH measurements. In this study, we provide such data for TSH using a cohort of Caucasian women receiving routine prenatal care at clinics and practices throughout Maine as the reference population. Blood samples were originally obtained from each of these women, in both the first and second trimesters, in a trial of integrated serum screening for Down s syndrome. This makes it possible to examine withinperson variability of thyroid measurements for the first time. PATIENTS AND METHODS The serum specimens for the current study are a subset from a larger prospective intervention study that evaluated a new method of prenatal screening for fetal Down s syndrome (US Maternal and Child Health Bureau [MC0095]). This method of screening, called integrated serum screening, requires that two serum samples be drawn during each woman s pregnancy: one at the time of her first prenatal visit and the other early in the second trimester. First trimester samples were obtained from,387 women and 928 of these women also supplied second trimester samples. Approximately 98% of the study population was Caucasian. The study was approved by the Institutional Review Board at the Foundation for Blood Research and the informed consent process included permission to use the serum samples in secondary studies, such as this one. The following criteria were used to select a reference population for the thyroid measurements among women with viable pregnancies: Gestational dates were confirmed by ultrasound The first trimester sample was obtained between 8 and 3 weeks of pregnancy The second trimester sample was obtained between 5 and 2 weeks of pregnancy (7 8 weeks after the first sample) Sufficient sample volume was available for testing Pregnancies that satisfied these criteria were taken consecutively. Journal of Medical Screening 2004 Volume Number 4

2 03 Haddow cpp 5//04 2:02 PM Page 7 Thyroid stimulating hormone in early pregnancy 7 Due to limited resources, only the first 26 of the 928 paired samples available for testing were processed. The TSH measurements were obtained on the Immulite 2000 (Diagnostic Products Corporation, Los Angeles, CA, USA) using the Rapid TSH 30-minute procedure. The lower limit of sensitivity is 0.0 miu/l. Coefficients of variation over the four weeks of testing were 6.8 and 5.2% for inhouse pools of.22 and 3.98 miu/l, respectively. Matched samples from individual women were always tested in the same run. The thyroid peroxidase (TPO) antibody measurements were also obtained on the Immulite Coefficients of variation over the same time period were 9.0 and 8.6% for manufacturer s pools of 45 and 57 IU/ml, respectively. Correlation coefficients and regression analyses were calculated after a logarithmic transformation. Statistical significance was two-tailed at the p=0.05 level. Selected data from the probability plot were fitted to a linear model to estimate the standard deviation. RESULTS Table displays selected centiles of TSH measurements on a week-by-week basis for the late first trimester (8 3 weeks) and the early second trimester (5 2 weeks). Collective measurements are also shown for each of the trimesters. In this study, all of the second trimester TSH measurements were obtained from the same women as those whose measurements are shown in the first trimester. No samples were collected during week 4. The collective median value is higher in the second trimester than in the first trimester (.29 versus.00 miu/l). Week-by-week median values within each of the trimesters are relatively stable. If the 98th centile were to be selected as the cutoff level for initiating further action, a single cutoff value could be justified for each trimester (5.2 and 4.2 miu/l in the first and second trimesters, respectively). Figure is a graphic representation of the full dataset shown in Table. It underscores the substantially greater variability of TSH levels within the population of pregnant women during the earlier weeks of pregnancy. The most important contributor to this variability is an excess of very low TSH levels. This phenomenon is well understood and can be attributed to a weak TSH-like effect of high human chorionic gonadotropin (hcg) levels. 2 As pregnancy proceeds, the variance of TSH levels diminishes. This means that TSH (miu/l) Gestational age (weeks) 98th th 75th 50th 25th Figure Selected centiles of TSH by gestational age in the first and second trimester of pregnancy for 26 women. The centiles of TSH are plotted on the logarithmic vertical axis versus the completed week of gestation on the horizontal axis. TSH values rise slightly during pregnancy. The distribution of TSH measurements is broader in the first trimester, especially for lower measurements. All women are included in the figure, regardless of TPO antibody measurements. commonly used methods for establishing cutoffs, such as standard deviations, will not effectively define a consistent group of high-risk women over the range of gestational weeks being considered in this study. Centiles are being used in our analyses to help avoid this problem. Table also contains selected centiles by trimester for women with TPO antibody levels 35 IU/ml. Overall, 2 women (.7% of the total) had a positive TPO antibody measurement (>35 IU/ml) in at least one of their samples. When these women are removed from the reference population, the TSH values defining the selected upper centiles are lower. For example, the 98th centile for all 26 women in the second trimester is 4.8 miu/l; when the 2 women with positive TPO antibodies are removed the 98th centile is 3.7 miu/l. TPO measurements in both trimesters are < IU/ml (undetectable) in 834 of the 26 women (74%). Figure 2 shows the antibody measurements among the remaining 292 women. The results are highly correlated (r=0.97, r 2 =0.94), but the first trimester measurements are consistently higher by about 60%. 5th Table Gestational week-specific TSH values in a cohort of 26 pregnant women with viable pregnancies, sampled in both the first and second trimester Observed Centile of TSH (miu/l) Completed week Number of of gestation observations 5th 25th 50th 75th 95th 98th NC All First Trimester TPO Antibody Negative NC NC All Second Trimester TPO Antibody Negative NC = not calculated: <50 observations. Those with TPO antibody measurements >35 IU/ml in at least one sample are not included Journal of Medical Screening 2004 Volume Number 4

3 03 Haddow cpp 5//04 2:02 PM Page Haddow, Knight, Palomaki, McClain, Pulkkinen In Figure 3, first trimester TSH measurements for each of the 26 women are plotted against the same woman s measurement in the second trimester. These results are highly correlated after the 5% of women with at least one very low value (<0. miu/l) are excluded (r=0.75, r 2 =0.56, p<0.00). The corresponding Deming lines are: TSH_first = Second trimester TPO (IU/mL) First trimester TPO (IU/mL) Figure 2 Detectable TPO antibody measurements in the first and second trimester of pregnancy for 292 women. Both TPO measurements are below IU/ml (undetectable) in 834 of the 26 women (74%). Among the remaining 292 women (26%), the first and second trimester TPO measurements are shown on the logarithmic horizontal and vertical axes, respectively. The solid line indicates the line of identity (Y=X). Vertical and horizontal dashed lines indicate a cutoff level of 35 IU/ml that is recommended by the manufacturer for defining antibody-positive women. Open circles are used to indicate the 2 women (.7% of all women) who have a positive TPO antibody identified in either the first or second trimester *TSH_second and TSH_second =.5034*TSH_first When the woman s TSH value is elevated in the second trimester, it is useful to know what her value would have been in the first trimester. The present dataset offers an opportunity to gain that insight. The analysis looks backward to examine first trimester values of 23 women whose TSH values were above the 98th centile in the second trimester. Seventeen of these 23 women (74%) had TSH values at or above at least the 95th centile in the first trimester, and all were above the median. It is worth noting that eight of women with values above the 99th centile were in that category in both trimesters, and six of the eight were TPO antibody positive. Similarly, for the 23 women in our study whose first trimester TSH values were above the 98th centile, sixteen (70%) remained at or above at least the 95th centile in the second trimester, and all but two remained above the median. Figure 4 shows a probability plot for both the first and second trimester TSH measurements in all 26 women. Neither distribution is log Gaussian over the entire range (indicated by the data fitting a straight line). The first trimester measurements fit a line well only between the 30th and 95th centiles, with a standard deviation of A cluster of extremely low TSH values ( 0.0 miu/l) in the first trimester means that it is virtually impossible, in the absence of clinical findings consistent with hyperthyroidism, to make a meaningful interpretation of such measurements at this period in gestation, especially when testing is being carried out in an unselected population. However, the second trimester TSH measurements fit a line well over a much broader range (between the th and 99th centiles, with a standard deviation of 0.242). In the second trimester, it is possible to reliably predict extreme upper centiles (e.g. 98th or 99th) by extrapolation from the mean value and logarithmic standard deviation. For example, the predicted 98th centile in the second trimester is 4.06 miu/l ( (log (.29) * 0.242) ), close to the observed value of 4.8 miu/l. 0 Second trimester TSH (miu/l) 0. TSH (miu/l) First trimester TSH (miu/l) Figure 3 A comparison of TSH measurements in the first and second trimester in 26 pregnant women. The TSH measurements are shown on logarithmic scales, with an individual pregnant woman s first trimester measurement on the horizontal axis and the same woman s second trimester measurement on the vertical axis. The correlation is high (r=0.75, r 2 =0.56) among the 68 women with both measurements >0. miu/l. All women are included in the figure, regardless of TPO antibody measurements Centile Figure 4 A probability plot comparing measurements in the first and second trimester in 26 pregnant women. The TSH measurements in the first (small closed circles) and second (larger open circles) trimesters are shown on the logarithmic vertical scale. The horizontal scale shows the Gaussian centile, based on the rank of the observation within its category. If the points follow a straight line closely, the distribution is considered log Gaussian. The thin line shows the results of a regression analysis of the data between the th and 99th centiles (TSH= log(.29) + z score * ) Journal of Medical Screening 2004 Volume Number 4

4 03 Haddow cpp 5//04 2:02 PM Page 73 Thyroid stimulating hormone in early pregnancy 73 DISCUSSION The availability of both first and second trimester serum samples from this entire cohort of pregnancies allows within-person variability of TSH and TPO antibody levels to be assessed for the first time. The high degree of correlation between TSH measurements in the two trimesters adds support to the validity of interpretation for clinical purposes, especially for measurements above the 98th centile where deficient thyroid function is suspected. By contrast, TSH measurements at the low end of the distribution are not highly correlated with hyperthyroidism due to suppression by the high levels of hcg in early pregnancy. TPO antibody measurements are also highly correlated, even though they are systematically lower in the second trimester than in the first. In 990, Glinoer and his associates examined distributions of maternal TSH measurements throughout pregnancy. 3 They documented a downward shift in TSH levels during the first trimester attributable to a weak TSH-like effect of hcg followed by a gradual rise as pregnancy proceeded. The pattern of TSH measurements described in the present study for the first and second trimesters is consistent with this earlier work. Glinoer subsequently documented somewhat higher TSH levels among the sub-population of women with elevated antibody levels and also demonstrated that TPO levels became lower by 60% as pregnancy progressed. 2 These findings are also confirmed in the present study. Among antibody-negative women, the median TSH measurement during the first trimester is 0.94 miu/l in the present study, as opposed to 0.75 miu/l in Glinoer s study. 3 This difference is probably due to standards differences between assays and serves as a reminder that reference ranges are most reliable when they are established by the laboratory providing the testing service. A lesser degree of variability can occur between kit lots of reagents made by the same manufacturer, indicating that quality assurance protocols need to be in place within the laboratory to document shifts and make the necessary adjustments to ensure that interpretations will be consistent. In establishing reference data, it is also important to consider possible differences in TSH and TPO antibody levels between various racial and ethnic groups. 4 When a cutoff is being chosen for any given screening or diagnostic test, it is important to have knowledge relating to the correlation between that test and the clinical disorder for which the test is being done. 5,6 Obtaining such knowledge can prove difficult in situations such as the one in this study, where mildly elevated TSH levels may not be associated with clinical manifestations in the pregnant woman at the time when testing is being done. Ten years after delivery in our previous study, it was possible to retrospectively document the natural clinical history of a group of women whose second trimester TSH values were at or above the 98th centile. TSH measurements were not performed on the stored serum samples until the time when the follow-up study was begun. The 98th centile was defined in terms of the second trimester viable pregnancy population as a whole (including an estimated % of women with positive antibody studies). In that study, 48 of the 62 women with TSH values at or above the 98th centile (77%) were not known to be thyroid deficient during pregnancy. It was possible to obtain follow-up on 45 of those 48 women years later. Among those 45 women, 29 (64%) were now documented to be clinically hypothyroid. Approximately 80% of those women had tested positive for thyroid antibodies during pregnancy. It was also possible to collect -year follow-up information on 20 control women whose TSH values during pregnancy were below the 98th centile. In the interval, five of these women (4%) developed clinical hypothyroidism. Based on these data, it appears reasonable to use the 98th centile of TSH as the cutoff for recommending treatment or close follow-up, at least until further data accumulate to link the next lower tier of TSH levels with outcome. To our knowledge, no other comparable follow-up data exist for pregnant women that might serve as a guide for defining an appropriate TSH cutoff. Based on this knowledge, our initial recommendation for setting a TSH cutoff would be to use the 98th centile as the decision point for considering further action. If we were to use a reference range that included only antibody-negative women, the 98th centile cutoff would result in 4.3% of first trimester women and 3.3% of second trimester women being identified as high risk. To achieve our intended target of 2% (based on clinical correlation), we would need to set the cutoff higher. ACKNOWLEDGMENTS We thank Diagnostic Products Corporation, Los Angeles, California, USA for their donation of reagents for this study Authors affiliations James E. Haddow, Foundation for Blood Research, Scarborough, George J. Knight, Foundation for Blood Research, Scarborough, Maine 04070, USA Glenn E. Palomaki, Foundation for Blood Research, Scarborough, Monica R. McClain, Foundation for Blood Research, Scarborough, Andrea J. Pulkkinen, Foundation for Blood Research, Scarborough, REFERENCES Haddow JE, Palomaki GE, Allan WC, et al. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. N Engl J Med 999;34: Allan WC, Haddow JE, Palomaki GE, et al. Maternal thyroid deficiency and pregnancy complications: implications for population screening. J Med Screen 2000;7: Pop VJ, Brouwers EP, Vader HL, et al. Maternal hypothyroxinaemia during early pregnancy and subsequent child development: a 3-year follow-up study. Clin Endocrinol 2003;59: Davis LE, Leveno KJ, Cunningham FG. Hypothyroidism complicating pregnancy. Obstet Gynecol 988;72: Leung AS, Millar LK, Koonings PP, et al. Perinatal outcome in hypothyroid pregnancies. Obstet Gynecol 993;8: Jones WS, Man EB. Thyroid function in human pregnancy. VI. Premature delivers and reproductive failures of pregnant women with low serum butanol-extractable iodines. Maternal serum TBG and TBPA capacities. Am J Obstet Gynecol 969;4: Stagnaro-Green A, Roman SH, Cobin RH, et al. Detection of at-risk pregnancy by means of highly sensitive assays for thyroid autoantibodies. JAMA 990;264: Surks MI, Ortiz E, Daniels GH, et al. Subclinical thyroid disease: scientific review and guidelines for diagnosis and management. JAMA 2004;29: American College of Obstetricians and Gynecologists, Committee on Practice Bulletins. ACOG Practice Bulletin. Clinical management guidelines for obstetricians-gynecologists. Number 32, November 200 (replaces Technical Bulletin Number 8, June 993, and Committee Opinion Number 24, September 2000). Thyroid disease in pregnancy. Obstet Gynecol 200;98: Gharib H, Cobin RH, Dickey RA. Subclinical hypothyroidism during pregnancy: position statement from the American Association of Clinical Endocrinologists. Endocrine Practice 999;5: Wald NJ, Watt HC, Hackshaw AK. Integrated screening for Down s syndrome on the basis of tests performed during the first and second trimesters. N Engl J Med 999;34: Glinoer D, de Nayer P, Bourdoux P, et al. Regulation of maternal thyroid during pregnancy. J Clin Endocrinol Metab 990;7: Journal of Medical Screening 2004 Volume 4

5 03 Haddow cpp 5//04 2:02 PM Page Haddow, Knight, Palomaki, McClain, Pulkkinen 3 Glinoer D, Riahi M, Grun JP, et al. Risk of subclinical hypothyroidism in pregnant women with asymptomatic autoimmune thyroid disorders. J Clin Endocrinol Metab 994;79: Hollowell JG, Staehling NW, Flanders WD, et al. Serum TSH, T4, and thyroid antibodies in the United States population (988 to 994): National Health and Nutrition Examination Survey (NHANES III). J Clin Endocrinol Metab 2002;87: Wald NJ, Cuckle H. Reporting the assessment of screening and diagnostic tests. BJOG 989;96: Haddow JE, Palomaki GE. ACCE: A model process for evaluating data on emerging genetic tests. In: Khoury M, Little J, Burke W, eds. Human Genome Epidemiology: A Scientific Foundation for Using Genetic Information to Improve Health and Prevent Disease. Oxford University Press; 2003, Journal of Medical Screening 2004 Volume Number 4