Quantitative assessment of thyroid-to-background ratio improves the interobserver reliability of technetium-99m sestamibi thyroid scintigraphy for investigation of amiodarone-induced thyrotoxicosis

Abstract

Amiodarone-induced thyrotoxicosis (AIT) is caused by excessive hormone synthesis and release (AIT I), a destructive thyroiditis (AIT II), or a combination of both (AIT Ind). Although no gold-standard diagnostic test is available, technetium-99m sestamibi thyroid scintigraphy (99mTc-STS) has been previously reported to be an accurate tool for differentiating subtypes with important therapeutic implications. However, the information to guide reporting of 99mTc-STS is qualitative and highly subjective. This study aims to compare the interobserver reliability of 99mTc-STS before and after the use of quantitative thyroid-to-background ratios (TBRs) displayed on a time-activity curve for differentiation of AIT subtypes. A retrospective audit of Nuclear Medicine Departments at Royal Melbourne Hospital (Parkville, Victoria, Australia) and Cabrini Hospital (Malvern, Victoria, Australia) identified 15 consecutive 99mTc-STS studies performed for AIT. Four nuclear medicine physicians reported the studies according to previously established criteria (series 1). Quantitative TBR and estimated 'normal' range TBR were subsequently provided before the studies were reordered and reported again (series 2). Interobserver reliability was calculated using Fleiss' κ statistic for each assessment. The overall percentage of agreement (PoA) and κ statistics for use of conventional 99mTc-STS for diagnosis of AIT improved from 47 to 80% and from 0.30 to 0.67 following the use of quantitative TBR displayed on a time-activity curve with reference to a normal population. Interobserver reliability improved substantially under all diagnostic comparisons, particularly for differentiation of either AIT I (PoA 80% to 94%, κ: 0.48 to 0.84) or AIT Ind (PoA 47% to 82%, κ: -0.05 to 0.51) from other types of AIT. Use of quantitative TBR improves the interobserver reliability of reporting 99mTc-STS for investigation of different types of AIT. There is 'almost perfect' agreement upon differentiation of AIT I from AIT II and AIT Ind, with important implications for rationalizing the use of corticosteroid therapy. Prospective identification of AIT Ind is improved from 'poor' to a 'moderate' level of agreement to facilitate rational use of combination therapy at diagnosis.