Radioiodine therapy for Graves' disease: multivariate analysis of pretreatment parameters and early outcome.

Twenty-one clinical, biochemical, scan and tracer-kinetic parameters were documented in 76 patients with Graves' disease who had received a standard 5-mCi therapy dose of 131I. Linear discriminant analysis was then undertaken to determine what combination of variables best predicted outcome. One year after therapy, 40 patients were euthyroid, 11 were hypothyroid, and 25 were still thyrotoxic. Linear discriminate functions combining 24-h 131I uptake, the presence or absence of thyroid eye signs and a computer-derived measurement of thyroid cell mass best discriminated the three outcome groups. The proportion of patients correctly reclassified according to outcome using these functions was, however, only just over 50%. It is concluded that no single or combination of pretreatment variables predicts early outcome with sufficient confidence to justify a rigorously 'scientific' approach to the administration of 131I therapy for Graves' disease.