Optimal interpretative strategy for preoperative parathyroid scintigraphy

Our objective was to retrospectively compare various scintigraphic methods to determine their relative accuracies and interobserver variabilities in preoperative localization of single-gland disease in patients with primary hyperparathyroidism.

We studied 292 patients who underwent preoperative parathyroid scintigraphy and surgical resection between June 2002 and September 2008. Imaging included early and delayed pinhole Tc-MIBI imaging (including anterior oblique images), similar I imaging, and MIBI SPECT. Seven different imaging combinations were assessed as follows: early MIBI only, delayed MIBI only, comparison of dual-phase early and delayed MIBI, visual subtraction of early MIBI and I images, all planar images, SPECT only, and all planar and SPECT images (ALL). Each data set was reviewed by 2 of 4 experienced nuclear physicians without knowledge of clinical information or final diagnosis. Imaging results were compared with surgical outcomes.

The accuracy of ALL imaging for delineating abnormal/normal parathyroid glands by quadrant was 96%, 93%, 95%, and 95% for the 4 readers. The accuracies of visual subtraction of early MIBI and I images and all planar images were not significantly different from ALL for all 4 readers. All planar and SPECT images were significantly more accurate than early MIBI only, delayed MIBI only, or SPECT only for all 4 readers and better than comparison of dual-phase early and delayed MIBI for 3 of 4 readers.

A rigorous combination of multiple imaging approaches yielded high accuracies across multiple readers for localization of a single offending parathyroid gland, but a more limited technique using MIBI/I subtraction imaging with anterior and anterior oblique pinhole images provided similarly high accuracy and limited interobserver variation of accuracy.