Giant cell arteritis and thyroid dysfunction: multicentre case-control studyBMJ 1999; 318 doi: https://doi.org/10.1136/bmj.318.7181.434 (Published 13 February 1999) Cite this as: BMJ 1999;318:434
- Pierre Duhaut, assistant professor ()a,
- Hubert Bornet, senior physicianb,
- Laurent Pinède, assistant professora,
- Sylvie Demolombe-Ragué, senior physiciana,
- Robert Loire, senior physicianc,
- Dominique Seydoux, medical chiefd,
- Jacques Ninet, senior physiciana,
- Jean Pasquier, senior physician on behalf of the Groupe de Recherche sur l'Artérite à Cellules Géantesa
- aDepartment of Internal Medicine, E Herriot Hospital, Lyons, Cedex 03, France
- bDepartment of Nuclear Medicine, Claude Bernard University, Faculté de Médecine Lyon-Grange-Blanche, France
- cLaboratory of Pathology, Louis Pradel Hospital, Lyons, France
- dControl group, Société de Secours Minière de la Loire, Saint-Etienne, France
- Correspondence to: Dr Duhaut
- Accepted 18 September 1998
The association between giant cell arteritis and thyroid dysfunction remains controversial, but as giant cells are a possible feature of Graves' disease, a common pathway has been suggested. In two series of 101and 98patients, the prevalence of hyperthyroidism was reported to be six times higher in cases of giant cell arteritis than in controls. 1 2 This was not confirmed on smaller series, 3 4 but 15cases of hypothyroidism were reported in 31patients with giant cell arteritis.5
We conducted a multicentre case-control study on cases of giant cell arteritis to investigate this relation.
Subjects, methods, and results
Assuming a prevalence of thyroid dysfunction of 1% in the general population and an odds ratio of 6for hyperthyroidism in the patient group, the sample size requested, with α=0.05 and β=0.2, had been estimated to be 269cases and controls.2
We prospectively studied 285cases of giant cell arteritis (205women, mean age 74.7±8.2years; 80men, 72.7±8.2) newly diagnosed during 1991-96.An experienced pathologist reviewed 262(92%) of the biopsies: temporal arteritis was confirmed in 145—68 were classed as negative (eight did not have a biopsy), and 72were classed as having polymyalgia rheumatica alone (22 did not have a biopsy). Blood samples taken up to 48hours after diagnosis were sent to a reference laboratory.
Controls, randomly selected by computer from residents of Saint-Etienne affiliated to a health insurance company, were matched to cases for age and sex. Of the 222controls participating, 208 (94%) agreed to have a blood sample taken (140women, mean age 74.9±8.7years; 68men, 71.7±8.0) (table). Neither cases nor controls had clinical signs or symptoms of thyroid dysfunction.
We measured concentrations of free thyroxine, thyroid stimulating hormone, and antithyroid peroxidase antibodies by standard radioimmunoassays. Antithyroglobulin antibodies were measured as follows: sera were incubated at room temperature with thyroglobulin labelled with 125-iodine, and the immune complexes were precipitated in fetal veal buffer with polyethylene glycol. A positivity threshhold of 50U/l for a population free of thyroid disease was determined.
We performed multiple logistic regression. Dependent variables were high and low concentrations of thyroid stimulating hormone, high and low concentrations of free thyroxine, and concentrations of positive or negative antiperoxidase antibodies, positive or negative antithyroglobulin antibodies, and positive or negative antithyroid antibodies (antiperoxidase or antithyroglobulin). Independent variables were case or control, geographical origin (north or south), age, sex, and clinical subgroup of patients.
When we took potential confounders into account, we found no difference between cases and controls. Antithyroid antibodies occurred more frequently in women than in men, and prevalence increased with age.
We found no difference between cases and controls when thyroxine or thyroid stimulating hormone titres were outside the normal range, or when antithyroglobulin or antiperoxidase antibody titres were positive (Wilcoxon sum rank test) (table on website).
The prevalence of high concentrations of thyroid stimulating hormone and antithyroid antibody was similar in cases and controls at the onset of the disease. After adjustment for potential confounders, we found a threefold but non-significant increase in the risk of hyperthyroidism in cases when thyroid stimulating hormone concentrations were measured. If the risk was to be significant a sample size of 641patients and 2564controls would be needed; such a sample size with incident cases of giant cell arteritis seems unrealistic. However, a common pathway for Graves' disease or hypothyroidism and giant cell arteritis seems unlikely. Determination of free thyroxine concentrations is probably less reliable in inflammatory syndromes, as thyroxine is bound to sera proteins.
The high prevalence of antithyroid antibodies in the controls should make researchers cautious when describing an association between autoimmune or inflammatory diseases and thyroid dysfunction in elderly patients.
Funding: Hospices Civils de Lyon, Conseil Régional Rhône-Alpes, and Programme Hospitalier de Recherche Clinique 1993,Ministère de la Santé, France.
Competing interests: None declared.
websiteextra: An additional table appears on our websitewww.bmj.com