Antithyroid antibodies do not affect pregnancy outcome in women undergoing assisted reproduction *

William H. Kutteh1,4, William B. Schoolcraft2 and Richard T. Scott, Jr3

1 Room D324, Division of Reproductive Endocrinology, Department of Obstetrics and Gynecology, 956 Court Avenue, University of Tennessee, Memphis, Tennessee 38163–2116, 2 Colorado Center for Reproductive Medicine, Englewood, CO and 3 Saint Barnabus Medical Center, Livingston, NJ, USA


    Abstract
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
The purpose of this study was to determine the association of antibodies to thyroglobulin and thyroid peroxidase and pregnancy outcome in women undergoing assisted reproductive techniques. The study included three centres and retrospectively evaluated patient sera for antithyroid antibodies, then related the results to pregnancy outcome. Enzyme-linked immunosorbant assays for thyroglobulin and thyroid peroxidase antibodies were performed using two different commercially available kits. Controls included 200 healthy women of reproductive age. Women (n = 873) who were undergoing assisted reproductive techniques for pelvic adhesions, endometriosis, ovarian dysfunction, or unexplained/other were included. All women utilized a standard regimen of gonadotrophin releasing hormone agonist down-regulation followed by ovulation induction. Embryos were cultured for 3 days, at which time embryo transfer occurred. Statistical analysis utilized the two-tailed Fisher's exact test. Antithyroid antibodies were positive in 143 of 873 (16.4%) women undergoing assisted reproductive techniques while 29 of 200 (14.5%) normal controls were positive (not significant). Deliveries were achieved in 396 of 730 (54.2%) of women who had no thyroid antibodies versus 78 of 143 (54.5%) of women who had thyroid antibodies (not significant). No difference in biochemical pregnancies (not significant) or clinical pregnancy losses (not significant) were detected. Antithyroid antibodies were found no more frequently in women undergoing assisted reproductive techniques than in normal controls. There were no differences in pregnancy outcome based on antithyroid antibody positivity in women undergoing assisted reproductive techniques. These data do not support the testing or treatment for antithyroid antibodies of women undergoing assisted reproductive techniques.

Key words: antithyroid antibodies/autoimmunity/infertility/in-vitro fertilization


    Introduction
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Autoimmune thyroid disorders are characterized by the presence of antithyroid antibodies, specifically antithyroglobulin and antithyroid peroxidase. Thyroglobulin (TG) antibodies are present in the majority of autoimmune thyroid disease patients. TG is a molecule produced by the thyroid cells and stored in thyroid colloid. The primary function of TG is the storage and synthesis of thyroid hormones. Thyroid hormones are synthesized on TG which subsequently serves in the synthesis of T3 and T4 (O'Connor and Davies, 1990Go). Elevations in the serum concentration of this autoantibody may be identified in any condition which results in an abnormal or damaged thyroid follicular structure (Bayraktar et al., 1990Go; Parkes et al., 1994Go). Antibodies to thyroid peroxidase (TPO), previously referred to as microsomal antibody, are less prevalent in some studies. TPO is an enzyme responsible for iodination of tyrosine residues along with coupling of iodinated residues to form thyroid hormones. The role of thyroid stimulating hormone (TSH), the primary trophic hormone of the thyroid, is thyroid growth and development; therefore, thyroid cell activation stimulates thyroid hormone synthesis (O'Connor and Davies, 1990Go).

Antithyroid antibodies have been reported in apparently healthy populations and are observed more frequently in women during their reproductive years (Geva et al., 1997Go). It has been suggested that immunological factors may play an important role in the reproductive processes of fertilization, implantation and placental development (Billingham and Head, 1981Go). Women have a high degree of immunological responsiveness which is reflected by their increased susceptibility to non-organ specific and organ specific autoimmune (Chiovato et al., 1993Go). Such increased susceptibility is supported by the fact that thyroid autoantibodies have been associated with an increased risk for pregnancy loss (Stagnaro-Green et al., 1990Go). It has also been reported that 5–10% of postpartum women demonstrate evidence of thyroid dysfunction (Amino et al., 1982Go; Hayslip et al., 1988Go; Jansson et al., 1988Go). Recent studies have suggested an association between autoimmune factors and reproductive wastage (Gleicher et al., 1989Go).

All patients with autoimmune thyroid disease have T cells in their blood and within the thyroid gland which recognize the specific thyroid molecules TG, TPO and the TSH receptor. Some of these T cells are able to kill `self' thyroid cells and activate B-cells to secrete antibodies which bind to these same thyroid molecules (O'Connor and Davies, 1990Go). Detection of thyroid reactive T cells, however, is difficult and currently unrealistic as a clinical tool.

Alternatively, thyroid autoantibody measurement can be used as a marker of autoimmune thyroid disease related to activated B-cells. Such measurements have become increasingly more advanced, and have proven to be useful clinical tools. It has been suggested that antithyroid antibodies may serve as peripheral markers for abnormal T cell function that may be responsible for pregnancy loss (Stagnaro-Green et al., 1992Go). This was further supported by Pratt et al. (1993) who demonstrated that detection of thyroid antibodies before conception carried an increased risk of pregnancy loss (Pratt et al., 1993Go).

Both TG and TPO are important factors in hormone synthesis and are major autoantigens in thyroid autoimmune disease. We previously reported no differences in the prevalence of antithyroid antibodies in 688 women undergoing assisted reproductive techniques compared to 200 normal, healthy female controls (Kutteh et al., 1999Go). The purpose of this study was to determine if the presence of antithyroid antibodies altered the pregnancy outcome in women undergoing assisted reproductive technologies.


    Materials and methods
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Patient population
Serum samples were collected routinely on all patients participating in the in-vitro fertilization (IVF) programmes at University Fertility Associates of the University of Tennessee, the Institute for Reproductive Medicine and Services at Saint Barnabas Medical Center, and at the Colorado Center for Reproductive Medicine. Between April 1996 and April 1997 serum from all basal samples was separated and then frozen at –70°C for later use. To prevent any potential observation or treatment bias, all samples were maintained frozen until after the last patient's pregnancy test.

A total of 873 women undergoing assisted reproductive techniques was included. Women were included if they had an evaluation including history and physical examination, hysterosalpingogram or hysteroscopy, and mid-luteal progesterone or late-luteal phase endometrial biopsy. These women had a prior history of infertility which included tubal factors (19%), ovarian dysfunction (26%), endometriosis (17%), and those who suffered from male factor/unidentified forms of infertility (38%). Women with a history of more than one prior pregnancy loss, known thyroid disease, or previously diagnosed autoimmune disease were excluded. If repeat cycles were performed on the same patient, only the first cycle was included in this study. Serum samples from all patients who underwent embryo transfer were available for analysis. Only couples who had embryos available for transfer were included for analysis so that pregnancy rates (PR) and outcome could be evaluated. None of the women had any additional treatment for antithyroid antibodies as this was a retrospective study and antibody results were not known at the time of IVF.

The control population consisted of 200 healthy, non-pregnant, reproductive-aged, parous women without a history of reproductive problems. The use of frozen blood samples from patients and controls conformed to human use guidelines established by the Institutional Review Board at the University of Tennessee.

Ovulation induction
Patients underwent a standard regimen of gonadotrophin releasing hormone agonist (GnRHa)-induced pituitary down-regulation followed by ovulation induction. In some cycles, GnRHa was used in a flare ovulation induction protocol. Ovarian stimulation was carried out using i.m. gonadotrophin injections followed by human chorionic gonadotrophin (HCG) administration upon ultrasound documentation of follicular maturity. Transvaginal ultrasound-guided oocyte retrieval was performed ~35 h after HCG administration. Embryos were maintained in culture for 3 days, at which time embryo transfer was performed. When clinically indicated, intracytoplasmic sperm injection (ICSI) and/or assisted hatching were performed. All serum samples were stored for analysis until after completion of the final cycle.

Enzyme-linked immunosorbent assays (ELISA)
All serum samples were evaluated for the presence of TG and TPO antibodies using a standard, highly sensitive ELISA. Samples were assayed in duplicate and mean values were reported. These commercial ELISA test kits (ImmunoWELL®; GenBio, San Diego, CA, USA) consisted of TG and TPO antibody coated plates to which both a set of known negative and positive sera and the patient sample sera were applied and incubated at room temperature. Sample plates were washed several times with wash buffer (0.01 mol/l phosphate buffered saline and 0.05% Tween), prior to the addition of a conjugate containing peroxidase-conjugated goat antihuman immunoglobulin (IgG) in phosphate buffered saline and carrier protein. The sample plates were then washed again prior to the addition of a colour-developing substrate followed by addition of the stop solution (0.25 mol/l oxalic acid). The optical density of the wells was determined on a spectrophotometer at a wavelength of 405 nm. Test results from 100 control women were correlated to a second commercially available ELISA kit (Kronus, San Clemente, CA, USA). Correlation coefficients were excellent between both manufacturers' ELISA kits (r = 0.95 for TG and r = 0.96 for TPO).

In order to standardize the assay, this optical density value was used to determine the normalized absorbance of the specimen (An), calculated through the following equation: An = As/Ac x EV. (As = absorbance of specimen; Ac = mean absorbance of the positive control; EV = expected value for the positive control). An values were then used to determine the respective values in IU/ml. Based on normalized data obtained from the manufacturer, results were reported as: negative, <=0.17 An (<=67 IU/ml) for TG and <=0.26 An (<=40 IU/ml) for TPO; borderline, 0.18–0.23 An (68–119 IU/ml) for TG and 0.27–0.37 An (41–64 IU/ml) for TPO; and positive >=0.24 An (>=120 IU/ml) for TG and >=0.38 An (>=65 IU/ml) for TPO. For data analysis, borderline values were considered to be negative. Interassay and intra-assay variation for antibodies to TG and TPO were 5.8% and 6.2% respectively.

Concentrations of TSH were also determined using a commercial ELISA test kit (Genzyme, San Carlos, CA, USA). Sample sera and control standards were added to microwells coated with murine monoclonal anti-TSH. Goat-anti TSH conjugated to horseradish peroxidase was then added to the samples prior to a 3 h incubation. The sample plates were rinsed with tap water, and tetramethylbenzidine substrate reagent was then added followed by a stop solution (2 mol/l hydrochloric acid). The absorbance was read at 450 nm and results were then converted to µIU/ml. The normal range for TSH was 0.45–4.5 µIU/ml. Interassay and intrassay variation was <8.0%.

Statistical analysis
The null hypothesis was that there were no differences in the pregnancy outcome of women undergoing assisted reproductive techniques based on the presence or absence of antithyroid antibodies. Statistical analysis between groups utilized the two-tailed Fisher's exact test. The odds ratio (OR) was used as an approximation of the relative risk in this retrospective study to evaluate the prevalence of antithyroid antibodies in controls versus women undergoing assisted reproductive techniques. The OR was also used to evaluate differences in pregnancy outcome after assisted reproductive techniques in women with and without antithyroid antibodies.


    Results
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Prevalence of antithyroid antibodies in control women versus women undergoing assisted reproductive techniques
Blood samples for 200 controls (healthy, reproductive-aged women) were assayed using commercial ELISA kits for the presence of TG and TPO antibodies (Table IGo). Overall, 29 out of 200 (14.5%) control women had antithyroid antibodies. Out of these 29, eight (4.0%) had antibody to TG alone; seven (3.5%) had antibody to TPO alone, and 14 (7.0%) control females revealed the presence of both TG and TPO antibodies. Out of the 873 assisted reproductive techniques patients, 143 (16.4%) tested positive for the presence of thyroid autoantibodies (Table IGo). There was no statistical difference between the prevalence of antithyroid antibodies in women undergoing assisted reproductive techniques compared to the control population. Out of the total group, 43 (4.9%) of the women had TG antibodies alone, while 48 (5.5%) tested positive for TPO antibody alone. Fifty-two (6.0%) of these women had both TG and TPO antibodies. None of these three groups, TG alone, TPO alone, or TG and TPO together, were statistically different when compared with the control population based on P value or OR. There were no differences in the prevalence of TG or TPO antibodies in women with different infertility diagnosis compared to controls.


View this table:
[in this window]
[in a new window]
 
Table I. Prevalence of antibodies to thyroglobulin (TG) and thyroid peroxidase (TPO) in controls and women undergoing assisted reproductive techniques
 
Thyroid stimulating hormone (TSH) concentrations
TSH concentrations were also assessed in the assisted reproductive techniques and control populations (Table IIGo). From the group of 200 control patients, 11 (5.5%) had abnormally high TSH concentrations. Out of these, seven (3.5%) were abnormal coupled with positive antibodies to TG and/or TPO. Only four (2.0%) from the total control population yielded abnormal TSH values while testing negative for both antibodies to TG and TPO. Of the 873 women undergoing assisted reproductive techniques, 70 (8.0%) had abnormally high TSH concentrations, which was not significantly different from the controls. The majority of these patients (43 or 4.9%) had abnormal TSH concentrations but tested negative for both TG and TPO antibodies, while 27 (3.1%) of these patients had abnormal TSH concentrations along with positive antibodies to TG and/or TPO values. Neither of these values were statistically different from the control population.


View this table:
[in this window]
[in a new window]
 
Table II. Elevated thyroid stimulating hormone (TSH) concentrations in control women and women undergoing assisted reproductive techniques
 
Pregnancy outcome after assisted reproductive techniques
A total of 873 women who underwent embryo transfer had samples available for analysis and met all inclusion and exclusion criteria. Overall 570 women (65.3%) became pregnant and 474 women (54.3%) delivered (Table IIIGo). An average of 3.5 ± 0.32 embryos was transferred in each case. ICSI was used in 38% of the cases and selective assisted hatching was used in one or more embryos being transferred in 68% of cases. There were no differences in pregnancy outcomes when evaluating data from cases where ICSI or selective assisted hatching was performed (data not shown).


View this table:
[in this window]
[in a new window]
 
Table III. Pregnancy outcome in women undergoing assisted reproductive techniques related to the presence or absence of thyroid antibodies
 
More importantly, there were no differences in pregnancy outcomes between the 143 women who tested positive for one or both antithyroid antibodies and the 730 women who tested negative for both antibodies (Table IIIGo). Deliveries occurred in 78 of 143 women (54.5%) who tested positive for one or both antibodies compared to 396 of 730 women (54.2%) of women who tested negative for both antibodies (not significant). Similarly, there were no statistical differences in biochemical pregnancy losses, clinical losses or those who did not achieve pregnancy, based on the presence or absence of antithyroid antibodies.

To determine if those women with the highest antithyroid antibody titres were at greater risk of poor pregnancy outcome, we further stratified the data on pregnancy outcome based on arbitrary threshold values of positive antithyroid antibodies as follows: level 1 positive, 120–420 IU/ml for TG and 65–370 IU/ml for TPO; and level 2 positive, 421–1021 IU/ml for TG and 371-1070 IU/ml for TPO. When the 143 patients with positive antithyroid antibodies were stratified in this arbitrary fashion, 89 (62.2%) were categorized in level 1 and 54 (37.8%) were categorized in level 2 (Table IVGo). There were no significant differences in assisted reproductive techniques outcome when patients with positive antithyroid antibodies were analysed in this fashion.


View this table:
[in this window]
[in a new window]
 
Table IV. Stratification of assisted reproductive treatment pregnancy outcome in 143 women with positive antithyroid antibodies related to concentrations of positivity
 

    Discussion
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
These data suggest that elevated antithyroid autoantibody titres do not occur more frequently in women undergoing assisted reproductive techniques when compared to controls. A multicentred study, which evaluated 544 pregnant women in the first trimester, directed attention to a correlation between spontaneous abortion and elevated titres of autoantibodies to TG and TPO (Stagnaro-Green et al., 1990Go). This increased risk of miscarriage could not be explained by TSH concentrations, anticardiolipin antibodies, maternal age, or relevant history. Unfortunately the authors failed to evaluate these women for other known causes of miscarriages. Similarly, Lejeune et al. (Lejeune et al., 1993Go) reported on 730 consecutive pregnant women attending an antenatal clinic. Elevated antibodies to TPO and TG were associated with an increased miscarriage rate. In a subsequent study of 487 assisted reproductive techniques pregnancies, antithyroid antibodies were found in 106 patients (22%) (Singh et al., 1995Go). There was a clinical miscarriage rate of 32% in the antibody positive group compared with a 16% miscarriage rate in the antibody negative group; however, no significant difference was noted in the incidence of biochemical pregnancies. Our larger study failed to detect an increased prevalence of antibodies to TG/TPO in non-pregnant women undergoing assisted reproductive techniques compared to non-pregnant control women.

Antithyroid testing has been advocated by some as an early marker for implantation failure in the prediction of pregnancy after assisted reproductive techniques. Some have even suggested that those patients be treated with intravenous gammaglobulin (Sher et al., 1998Go). Antithyroid antibodies may be associated with an increased risk of subsequent miscarriage in pregnancies conceived naturally or through assisted reproductive techniques (Stagnaro-Green et al., 1990Go; Singh et al., 1995Go), however current data do not support an association of antithyroid antibodies and implantation failure. Antithyroid antibodies are known to occur in normal, healthy populations, and these autoantibodies are five times more common in women than in men (Chiovato et al., 1993Go). Because of the relatively common occurrence of antithyroid antibodies in normal women, interpreting the significance of these antibodies in women with reproductive problems remains difficult. Therefore, aggressive and expensive treatment of women with antithyroid antibodies who are undergoing assisted reproductive techniques may not be necessary.

Differences in age between the control and patient test groups could not account for the failure to find different prevalences of antithyroid antibodies observed in our study. Women undergoing assisted reproductive techniques were older (35.6 ± 4.1 years) than the controls (30.8 ± 6.2 years) and might have been expected to have an elevated prevalence of autoantibodies; however, this was not observed in our study. Some, but not all studies have revealed an increase in the prevalence of autoantibodies with age. Kontiainen et al. (Kontiainen et al., 1994Go) found an increase in the amount of TPO antibodies with age; however, this association was not statistically significant. Similarly, antithyroid antibody assay methodology cannot be used as an explanation to account for the failure to find any differences in our study. We utilized two different commercially available test kits to evaluate control samples and found that the correlation coefficients between tests were >=0.95. Interassay and intra-assay reproducibilities were excellent.

It has been suggested that antithyroid antibodies may be associated with an increased risk of pregnancy loss. Two working hypotheses concerning the possible pathophysiological role of TG and TPO antibodies and pregnancy loss exist. The first is the proposal that the biochemical interaction between hormones and elevated thyroid autoantibodies may in some way directly result in pregnancy loss. Since thyroid function is normal in many patients when thyroid antibodies are detected, many investigators have questioned the possible role of thyroid antibodies in women with pregnancy loss. Alternatively, others consider TG and TPO autoantibodies as secondary markers of autoimmune disease rather than the actual cause of pregnancy loss (Singh et al., 1995Go). These antithyroid antibodies may reflect an abnormal immunological response which results in pregnancy loss. However, the absence of an effect on pregnancy outcome in women with antithyroid antibodies undergoing IVF most probably discounts both theories as clinical concerns.

In summary, there were no differences detected in the prevalence of antithyroid antibodies in non-pregnant women undergoing assisted reproductive techniques and healthy, non-pregnant control women. Moreover, the pregnancy outcome of women who underwent embryo transfer after ovulation induction was not different with respect to biochemical pregnancy, clinical loss, delivery or failure to become pregnant, based on the presence or absence of antibodies to thyroglobulin or thyroid peroxidase. The clear lack of association between these antibodies and pregnancy outcome suggests that women undergoing assisted reproductive techniques (with no other medical indications) do not need to be tested and do not require treatment for antithyroid antibodies. The expense of testing coupled with the costs and potential risks of treatment cannot be justified by a laboratory test result that does not appear to affect the outcome. We suggest that any treatments utilized for antithyroid antibodies in otherwise healthy women undergoing assisted reproductive techniques be reserved for patients participating in prospective, randomized, placebo-controlled trials.


    Acknowledgments
 
The authors appreciate the generous donation of antithyroid antibody ELISA kits from the research committees at GenBio, San Diego, CA, USA and Kronus, San Clemente, CA, USA.


    Notes
 
4 To whom correspondence should be addressed Back

* This paper was presented at the 54th Annual Meeting of the American Society for Reproductive Medicine, San Francisco, CA, USA, October 4–9, 1998. Back


    References
 Top
 Abstract
 Introduction
 Materials and methods
 Results
 Discussion
 References
 
Amino, N., Mori, H., Iwatani, Y. et al. (1982) High prevalence of transient postpartum thyrotoxicosis and hypothyroidism. N. Engl. J. Med., 306, 849–852.[ISI][Medline]

Bayraktar, M., Ergin, M., Boyacioglu, A. et al. (1990) A preliminary report of thyroglobulin release after fine needle aspiration biopsy of thyroid nodules. J. Int. Med. Res., 18, 253–255.[ISI][Medline]

Billingham, R. and Head, J. (1981) Current trends in reproductive immunology: an overview. J. Reprod. Immunol., 3, 253–265.[ISI][Medline]

Chiovato, L., Lapi, P., Fiore, E. et al. (1993) Thyroid autoimmunity and female gender. J. Endocrinol. Invest., 16, 384–389.[ISI][Medline]

Geva, E., Amit, A., Lerner-Geva, L. et al. (1997) Autoimmunity and reproduction. Fertil. Steril., 67, 599–611.[ISI][Medline]

Gleicher, N., El-Roeiy, A., Confino, E. et al. (1989) Reproductive failure because of autoantibodies: unexplained infertility and pregnancy wastage. Am. J. Obstet. Gynecol., 160, 1378–1380.

Hayslip, C., Fein, H., O'Donnell, V. et al. (1988) The value of serum antimicrosomal antibody testing in screening for symptomatic postpartum thyroid dysfunction. Am. J. Obstet. Gynecol., 159, 203–209.[ISI][Medline]

Jansson, R., Dahlberg, P. and Karlsson, F. (1988) Postpartum thyroiditis. Bailliere's Clin. Endocrinol. Metab., 2, 619–635.[ISI][Medline]

Kontiainen, S., Melamies, L., Miettinen, A. et al. (1994) Thyroid autoantibodies in serum samples with abnormal TSH concentrations. APMIS, 102, 716–720.[ISI][Medline]

Kutteh, W.H., Yetman, D.L., Carr, A.C. et al. (1999) Increased prevalence of antithyroid antibodies identified in women with recurrent pregnancy loss but not in women undergoing assisted reproduction. Fertil Steril., 71, 843–848.[ISI][Medline]

Lejeune, B., Grun, J.P. and de Nayer, P. (1993) Antithyroid antibodies underlying thyroid abnormalities and miscarriage or pregnancy induced hypertension. Br. J. Obstet. Gynaecol., 100, 669–672.[ISI][Medline]

O'Connor, G. and Davies, T.F. (1990) Human autoimmune thyroid disease: a mechanistic report. Trends Endocrinol. Metab., 1, 266–272.[ISI]

Parkes, A., Black, E., Adams, H. et al. (1994) Serum thyroglobulin: an early indicator of autoimmune post-partum thyroiditis. Clin. Endocrinol., 41, 9–14.[ISI][Medline]

Pratt, D., Kaberlein, G., Dudkiewicz, A. et al. (1993) The association of antithyroid antibodies in euthyroid non-pregnant women with recurrent first trimester abortions in the next pregnancy. Fertil. Steril., 60, 1001–1004.[ISI][Medline]

Sher, G., Maassarani, G., Zouves, C. et al. (1998) The use of combined heparin/aspirin and immunoglobulin G therapy in the treatment of in vitro fertilization patients with antithyroid antibodies. Am. J. Reprod. Immunol., 39, 223–225.[ISI][Medline]

Singh, A., Dantas, Z.N., Stone, S.C. et al. (1995) Presence of thyroid antibodies in early reproductive failure: biochemical versus clinical pregnancies. Fertil. Steril., 63, 277–281.[ISI][Medline]

Stagnaro-Green, A., Roman, S., Cobin, R. et al. (1990) Detection of at-risk pregnancy by means of highly sensitive assays for thyroid autoantibodies. J. Am. Med. Assoc., 264, 1422–1425.[Abstract]

Stagnaro-Green, A., Roman, S.H., Cohen, R.H. et al. (1992) A prospective study of lymphocyte-initiated immunosuppression in normal pregnancy: evidence of a T cell etiology for post partum thyroid dysfunction. J. Clin. Endocrinol. Metab., 74, 645–653.[Abstract]

Submitted on May 4, 1999; accepted on August 12, 1999.