A selective, Bayesian approach to intraoperative PTH monitoring

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

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).

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.

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).

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.