A selective, Bayesian approach to intraoperative PTH monitoring

Abstract

Objective: To determine if the performance of intraoperative parathyroid hormone monitoring (IPM) can be optimized by limiting its application to patients with weak preoperative localization.

Background: The value of IPM during minimally invasive parathyroidectomy (MIP) has been questioned, particularly in cases with strong preoperative localization. We describe a novel, Bayesian strategy employing IPM in select patients with a high pretest probability of multiple gland disease (MGD).

Methods: We prospectively examined 361 consecutive patients undergoing surgery for primary hyperparathyroidism. All patients underwent sestamibi (MIBI) scanning and surgeon-performed ultrasound. Intraoperative PTH levels were only used for surgical decision-making in the MIBI-negative, ultrasound-positive patient subset. The following outcomes were analyzed: MGD rate, test performance, success rate, and operative time.

Results: Patients with any positive localization study (91%) were offered MIP. The success rate was 99%. The MGD rate was 3% in MIBI-positive patients and 36% in MIBI-negative patients (10% overall, P < 0.0001). MIBI and surgeon-performed ultrasound were equally sensitive (80% vs. 85%, NS). Among MIBI-negative patients, 71% of whom underwent MIP with IPM, an inadequate fall in the 10-minute postexcision PTH level was highly predictive of MGD, saving 10 failures while causing 1 inappropriate conversion to bilateral exploration (negative likelihood ratio, NLR 28.0). In contrast, among MIBI-positive patients, IPM could have saved 3 failures at the expense of 18 inappropriate conversions (NLR 9.9). IPM increased operative time from 34 to 60 minutes (P < 0.0001).

Conclusion: IPM is more likely to guide the surgeon correctly when used only in MIBI-negative patients, who have a high pretest probability of MGD. This selective strategy maintains high success rates while limiting the frequently adverse impact that IPM carries when used indiscriminately.