Preoperative localisation of parathyroid adenoma in primary hyperparathyroidism using 99mTc-sestamibi SPECT/CT : an evolving scanning protocol

Abstract

Primary hyperparathyroidism (pHPT) is caused by one or more hyperfunctional parathyroid gland causing an inappropriately high release of parathyroid hormone (PTH) in relation to the calcium concentration in the blood. PTH acts on the bones to release more calcium and on the kidneys to reabsorb calcium, causing hypercalcemia. Approximately 75% of the patients are women and median age is 62. The only permanent cure is surgical removal of all pathologic parathyroid glands. To minimise the surgical exploration preoperative imaging localisation methods, have for decades been used and refined to pinpoint the culprit gland(s). The performance data for different imaging modalities used for preoperative localisation of hyperfunctional parathyroid glands are difficult to interpret. There are large numbers of studies on different methods with varying protocols and quality, often with insufficient reporting on important influencing factors such as adenoma weight and frequency of multiglandular disease (MGD). In this thesis we have analysed the performance of dual timepoint 99mTc-sestamibi SPECT/CT for preoperative localisation of PTAs with regards to its individual components: 99mTc-sestamibi SPECT alone [S], nonenhanced CT (native phase) [N], contrast-enhanced CT (arterial- and venous phase), [A] and [V] respectively and in combination [AN], [VN], [SN], [ANS], [VNS] and [SNAV]. Additionally, the impact of the adenoma weight and MGD on PTA localisation was also investigated.

In Study I we retrospectively analysed 249 patients examined with nonenhanced 99mTcsestamibi SPECT/CT and found that adding a diagnostic native phase to 99mTc-sestamibi SPECT significantly increased the localisation specificity from 93.5% to 95.9% (p<0.01), but not the sensitivity. In a prospective examination of 192 patients (Study II) we reported that adding an arterial and venous phase to nonenhanced SPECT/CT [SN] significantly increased the localisation sensitivity from 81.1% to 89.9% (p<0.01) without changing the specificity. Using the same cohort, in Study III we showed that adding 99mTc-sestamibi SPECT to different combinations of CT phases increased sensitivity e.g., 80.8% for [AN] as compared to 86.5% for [ANS] (p<0.01). However, the use of both contrast-enhanced phases was found redundant in terms of sensitivity gain, just adding 4 extra mSv.

The specificity was 97.9% for both. Although small parathyroid adenomas are known to be a challenge in preoperative localisation, we showed that it could be overcome using [ANS] or [SNAV]. The performance in patients with MGD remained unsatisfactory for all image sets, with a per-patient sensitivity of merely 30-40%. As a way of mitigating the consequences of this, in Study IV we trained a Machine Learning Classifier to recognise cases were preoperative localisation misclassified patients with MDG as single gland disease (SGD). As predictors, we used a set of pHPT related biochemical variables and the measured adenoma weight on patients cured after parathyroidectomy. On test data, the current classifier reached a 72% true positive prediction rate for MGD-patients and a misclassification rate of 6% for SGD-patients. These results call for further exploration before clinical implementation.