Comparison of Models for Predicting Renal Disease Following I-131 Therapy for Feline Hyperthyroidism

Morrison J., Jergens A., Deitz K., et al.

Conference Proceedings, (2010). American College of Veterinary Internal Medicine, Anaheim: p.744-745


Hyperthyroidism is a common endocrine disorder in middle aged to older cats. Therapy with I-131 is considered to be curative for > 90% of affected cats. However, the hyperthyroid state may mask underlying renal insufficiency and rates of subsequent renal disease following I-131 therapy approach 30%. Initial prospective analysis of a population of hyperthyroid cats identified pre-treatment serum creatinine concentration and glomerular filtration rate (GFR, as determined by nuclear scintigraphy) to be significantly associated with the development of renal disease following I-131 therapy. A receiver operating characteristic (ROC) curve was developed for each of the following scores: pre-treatment creatinine concentration, pre-treatment GFR, and a linear combination score using weighted values of creatinine and GFR, i.e. (2.6953)(creatinine) –(0.9617)(GFR). The coefficients in the linear combination were de- termined by a logistic regression model with the two explanatory variables. Area under the curve (AUC) and sensitivity and specificity values were calculated for each model. Forty-six hyperthyroid cats were treated with I-131 and 13/46 (28%) were determined to have renal disease following therapy. For cats that developed renal disease following I-131 therapy, mean pretreatment creatinine concentration was 1.29 mg/dl (range: 0.6– 2.2 mg/dl) and mean GFR was 2.01 ml/min/kg (range: 1.03–2.8 ml/ min/kg). In cats that maintained normal renal function following therapy, mean pre-treatment creatinine concentration was 0.83 mg/ dl (range: 0.2–1.7 mg/dl) and mean GFR was 2.91 ml/min/kg (range: 1.09–7.41 ml/min/kg). Estimated AUC was highest for the combination model (0.853), followed by the creatinine concentration AUC (0.798) and the GFR AUC (0.769). The Youden index, defined as max (Se1Sp-1), was used to determine the optimal cutoff value for each of the 3 models. In the combination model, a combination score 4 0.7818 (the cut- off value was chosen based on a cat with a pre-treatment creatinine concentration of 1.1 mg/dl and pre-treatment GFR value of 2.27 ml/ min/kg) predicted the development of renal disease with 85% sen- sitivity and 85% specificity. For the creatinine only model, a pre- treatment creatinine concentration > 1.4 mg/dl predicted the development of renal disease with 77% sensitivity and 76% specificity. In the GFR only model, a pre-treatment GFR value < 2.18 mL/min/ kg predicted post treatment renal disease with 85% sensitivity and 73% specificity. The results of this analysis demonstrated the model based on both pre-treatment serum creatinine concentration and GFR afforded the best predictive power for the development of renal disease following I-131 treatment. This model was superior to models that evaluated the variables independently.