A retrospective analysis of the diagnostic performance of 11C-choline PET/CT for detection of hyperfunctioning parathyroid glands after prior negative or discordant imaging in primary hyperparathyroidism

Background Identifying the correct location of a parathyroid adenoma in patients with primary hyperparathyroidism (pHPT) is crucial as it can guide surgical treatment. This study aimed to determine the diagnostic performance of 11C-choline PET/CT in patients with pHPT as a next in-line scan after primary negative or discordant first-line imaging. Methods This was a retrospective single-center cohort study. All patients with pHPT that were scanned utilizing 11C-choline PET/CT, after prior negative or discordant imaging, between 2015 and 2019 and who subsequently underwent parathyroid surgery were included. The results of the 11C-choline PET/CT were evaluated lesion-based, with surgical exploration and histopathological examination as the gold standard. Results In total, 36 patients were included of which three patients were known to have Multiple Endocrine Neoplasia (MEN) syndrome. In these 36 patients, 40 lesions were identified on 11C-choline PET/CT and 37 parathyroid lesions were surgically removed. In 34/36 (94%) patients a focused parathyroidectomy was performed, in one patient a cervical exploration due to an ectopically identified adenoma, and in one patient a bilateral exploration was performed because of a double adenoma. Overall, per-lesion sensitivity of 11C-choline PET/CT was 97%, the positive predictive value was 95% and the accuracy was 94% for all parathyroid lesions. Conclusions In patients with pHPT and prior negative or discordant first-line imaging results, pathological parathyroid glands can be localized by 11C-choline PET/CT with high sensitivity and accuracy.

The treatment of pHPT is generally surgical removal of the hyperfunctioning parathyroid gland(s). Since a solitary adenoma is the predominant cause, parathyroid surgery is preferably performed through a minimally invasive parathyroidectomy (MIP) in which only the suspected adenoma is resected in a focused manner. The other classic surgical approach is a bilateral neck exploration (BNE), in which all four parathyroid glands are exposed and inspected, and subsequently the enlarged glands are resected.
Therefore, this study aimed to analyze the diagnostic performance of 11 C-choline PET/CT after prior negative or discordant first-line imaging in patients with pHPT undergoing parathyroid surgery with an optimized imaging protocol.

Study design and patient selection
This is a retrospective single-center cohort study of patients ≥ 18 years with biochemically proven pHPT who underwent parathyroid surgery after localization utilizing 11 C-choline PET/CT and negative or discordant first-line imaging in a teaching and tertiary referral hospital in the Netherlands between 2015 and 2019. Patients with a germline mutation predisposing for familial hypocalciuric hypercalcemia were excluded. First-line imaging results were classified as discordant if a potential adenoma was identified on only one imaging modality or both imaging modalities showed an adenoma but in different locations.
The medical charts were reviewed to determine the outcome of all preoperative imaging tests. The location was recorded from imaging reports. Localization of suspected parathyroid adenoma(s) was described in relation to the position of the thyroid midline/trachea, i.e. right cranial; right caudal; left cranial; left caudal; ectopic (e.g. paratracheoesophageal or mediastinal) or no parathyroid adenoma identified. Corrected serum calcium levels were calculated [20].
A no objection notice of subjects was obtained after the protocol had been approved by the Medical Ethics Committee UMC Groningen (2018/658).

Cervical ultrasonography
cUS was performed in the University Medical Center Groningen (UMCG) or referring hospitals on various ultrasound systems by radiologists, as described previously [6]. In short; patients were examined in a supine position with a hyperextended neck using a high-frequency linear transducer. The neck was scanned from the level above the thyroid to the clavicle caudally. In case the cUS was performed outside the UMCG, images were retrieved, however re-interpretation was not performed, due to the highly operator dependent nature of this imaging modality.

Tc-MIBI-SPECT(/CT)
At the UMCG patients were scanned on a Symbia T16 gamma camera with CT (Siemens), resulting in SPECT/ CT images. 99m Tc-MIBI was used for preoperative localization as dual phase technique, which was combined with a dual tracer subtraction technique for thyroid only visualization with 99m Tc-pertechnetate. Some 99m Tc-MIBI scans were performed in referring hospitals with slightly different imaging protocols (only dual phase technique and/or only SPECT). However, all protocols adhered to the international guidelines on parathyroid imaging [21]. The 99m Tc-MIBI-SPECT(/CT) were re-interpreted by UMCG nuclear medicine physicians in case it had been performed outside the UMCG. 11 C-choline PET/CT was performed as described previously by Noltes et al. [19]. In short; patients had to fast for six hours while drinking one liter of water, tea or coffee without milk and sugar prior to the PET/CT procedure. First, a low dose CT (ldCT) was performed for attenuation correction of the PET images, with 120 kV, Quel ref mAs of 30 and a pitch of 1.5 on a 40 or 64-slice CT (Biograph mCT, Siemens). Scan area involved one bed position, was from the lower jaw to the heart and was recorded in listmode. In the first five patients, PET/ CT images were taken directly after injection of a median dose of 556 MBq (range 510-635 MBq) 11 C-choline until 40 to 60 min postinjection. After optimizing our imaging protocol [19], all subsequent patients were scanned dynamically 20 min after injection of a median dose of 410 MBq (range 360-440 MBq) for 10 min. All images were iteratively reconstructed using three iterations, 21 subsets with 5 mm FWHM Gaussian filter, including time of flight and resolution modelling. A lesion was called positive and suspected to be a parathyroid adenoma in case there was focal uptake above background, and the location could match with a parathyroid location.

Surgery
Parathyroidectomy was performed by experienced surgeons either via a MIP or a BNE. Intraoperative parathyroid hormone (ioPTH) was measured at T0 (before incision), T1 (excision of adenoma), T2 (T1 + 5 min), T3 (T1 + 10 min) and T4 (T1 + 15 min). A decrease of ioPTH of ≥ 50% [22] was classified as sufficient. Final localization of the adenoma during surgery was retrieved from the surgical report. The pathology report was reviewed for the final diagnosis. Cure was defined as normal serum calcium levels (< 2.57 mmol/L) 6 months after surgery [23].
Lesion based localization was defined as true positive, true negative, false positive, and false negative, depending on the final outcome (i.e. surgical and pathology reports). A true positive scan was defined as suspected adenoma(s) localized to the correct quadrant. A scan was considered true negative when no pathological parathyroid gland(s) was found at surgery and the scan suggested no presence of hyperactive parathyroid tissue. A false positive scan was defined as suspected adenoma(s) localized to the incorrect quadrant or a positive scan in a quadrant while during surgery no pathological parathyroid gland was found in this quadrant. A false negative scan was defined as removal of a pathological parathyroid gland during surgery, although the scan had not demonstrated the presence of hyperactive parathyroid tissue in this quadrant.
Per-lesion sensitivity, PPV and accuracy (of intraoperatively inspected parathyroid glands) of 11 C-choline PET/ CT was calculated.

Statistics
Data were analyzed using descriptive statistics. Categorical variables are displayed as count (n) and percentage (%). Continuous variables are displayed by median with range. Statistical analyses were performed using IBM SPSS Statistics version 25.0 (IBM Corp., Armonk, NY, USA).

Patients
Between 2015 and 2019, 54 patients underwent a 11 C-choline PET/CT. After exclusion of 18 patients, 36 patients were included in this study (Fig. 1). Data from 13 patients were previously published [24].

Preoperative imaging
A cUS was performed in 30/36 patients (24 at the UMCG and six at other hospitals), of which in nine patients a parathyroid adenoma could not be identified (Table 2). In the remaining 21 patients, an adenoma was suspected on cUS, but this lesion could not be confirmed on 99m Tc-MIBI-SPECT(/CT) (discordant findings n = 16) or a  Tc-MIBI-SPECT(/CT) was negative in 21 patients and positive but discordant with cUS in eight patients ( Table 2). 11 C-choline PET/CT was positive in all 36 patients, and 40 lesions suspicious for a parathyroid adenoma were identified on 11 C-choline PET/CT. Figure 2 shows representative images of a patient with a negative 99m Tc-MIBI-SPECT(/CT) and a positive 11 C-choline PET/CT. Of the excluded 16 patients who did not undergo surgery, in four patients (25%) a parathyroid lesion could not be identified on 11 C-choline PET/CT.

Surgery
In 34/36 patients (94%) a unilateral parathyroidectomy was performed. In one patient with an ectopically identified adenoma on 11 C-choline PET/CT, the adenoma was extirpated from the mediastinum through a cervical incision. In the second patient, a BNE was performed because of a right and left-sided identified parathyroid lesion. In 33 patients, ioPTH was used and showed a sufficient decrease in 32 patients. The patient without a sufficient decrease was diagnosed with postoperative mildly persisting HPT.
In one patient with MEN-I syndrome, 11 C-choline PET/ CT identified two parathyroid adenomas, but only one adenoma was resected (true positive). Since this MEN-I patient already had two parathyroid glands resected, the decision was made to perform a MIP and leave one parathyroid gland in-situ to avoid hypoparathyroidism. In one patient with sporadic pHPT, 11 C-choline PET/CT identified three possible parathyroid adenomas on both sides of the neck, but during surgery only the right side was explored. Since a sufficient decrease of ioPTH was found after removal of one parathyroid adenoma, the decision was made not to explore the left side, resulting in one true positive and one false positive lesion. Thus, in these two patients, two lesions on 11 C-choline PET/ CT were not confirmed as parathyroid adenomas and were excluded, leaving 38 11 C-choline PET/CT lesions for analysis.  (Table 3). In eight patients, the parathyroid gland on the ipsilateral site was explored as well, showing normal glands (true negative). The two false positive and the one false negative cases were analyzed (Table 4).  There was one patient with MEN-IIA syndrome where the 11 C-choline PET/CT showed a lesion in the right lower quadrant, but the adenoma was localized elsewhere during surgery (right upper). Hence, a false positive and false negative localization was registered (Table 4). This false negative lesion could in retrospect be identified on 11 C-choline PET/CT. The other false positive lesion includes the aforementioned patient with adenomas on both sides of the neck. In retrospect, the false positive lesions may have been due to reactive lymph nodes and thyroiditis, as was found in the pathological examination (Table 4).
Overall, per-lesion sensitivity of 11 C-choline PET/CT for all parathyroid lesions was 97% (100% for sporadic lesions and 67% for MEN related lesions, respectively), the PPV 95% (97% sporadic and 67% MEN) and the accuracy 94% (98% sporadic and 50% MEN).  Tc-pertechnetate (a) and early 99m Tc-MIBI (b) showing physiological uptake in the right thyroid lobe. Planar subtraction image (early 99m Tc-MIBI minus 99m Tc-pertechnetate image c) also does not show the parathyroid adenoma. The uptake in the right thyroid lobe was also observed on the 99m Tc-MIBI-SPECT image d. The low dose-CT of the 99m Tc-MIBI-SPECT/CT e retrospectively did show a focus suspect for a parathyroid adenoma located paraoesophageally on the right side (red arrow). The 11 C-choline PET/CT (f fused PET/CT image and g, h PET only image) showed this same lesion (1.80 cm and 1.00 g at pathology) located paraoesophageally on the right side, suspicious for a parathyroid adenoma (red arrow). The 11 C-choline PET/CT also showed the same physiological uptake in the right thyroid lobe (f-h) respectively (Table 5). In total, 34/36 patients were cured two months post-surgery and 10/12 patients (24 loss to follow up) 6 months post-surgery. One patient that was not cured was diagnosed with MEN-I and during surgery, the decision was made to leave one parathyroid gland in-situ to avoid hypoparathyroidism. The other patient presented with mildly persisting HPT. Additional imaging did not show a definitive localization and due to the mildly elevated albumin-corrected calcium levels (2.61 mmol/L) no further surgical or medical intervention was deemed necessary.

Discussion
In this study, 11 C-choline PET/CT showed overall a high lesion-based sensitivity, PPV and accuracy of 97%, 95% and 94%, respectively in 36 patients operated for pHPT with prior non-conclusive first-line imaging. The surgeon performed a successful resection in all 36 patients of which in 34 patients a MIP was sufficient and an extensive neck exploration was avoided. The overall sensitivity of 97% in our study is comparable to three other studies, showing a sensitivity of 87.0%, 92.3% and 98.8% [16][17][18]. Our sensitivity would have been even higher if we had excluded MEN-IIA patients from our analysis (100% for sporadic pHPT and 100% for MEN-I patients). Liu et al. conducted a study of 87 patients undergoing 11 C-choline PET/CT after negative or discordant first-line imaging and subsequently parathyroid surgery [18]. They defined lesions with both positive and inconclusive uptake as positive 11 C-choline PET lesions, which might have overestimated their diagnostic accuracy. Two other studies both included patients with pHPT undergoing 11 C-choline PET/CT and parathyroid surgery in whom the adenoma was also detected on MIBI imaging [16,17]. The main strength of our study is that it describes and reflects the accuracy of 11 C-choline PET/ CT after prior negative or discordant first-line imaging in daily practice.
We included MEN-I and MEN-IIA patients since preoperative imaging is pivotal in both MEN-I patients with recurrent disease and MEN-IIA patients [23]. This patient group was, however, limited in our cohort and more research is warranted into additional preoperative imaging in MEN patients.
The use of a highly sensitive preoperative imaging technique, such as 11 C-choline PET/CT might lead to more Table 4 Outcomes of the false positive (n = 2) and false negative (n = 1) 11    minimally invasive procedures. Correspondingly, the surgeon performed a successful MIP in 34/36 patients. A MIP has several advantages compared to a BNE. There is a lower risk of postoperative hypocalcemia and laryngeal nerve injury, it results in less unnecessary scarring with better opportunity for later surgery and has a shorter duration of surgery [25,26]. It is, however, not recommended to employ 11 C-choline PET/CT as a first-in-line imaging technique since the AAES guideline advises initial imaging with cUS and 99m Tc-MIBI-SPECT/CT (or four dimensional-CT), because this is the most costeffective strategy [23]. If these first-line imaging results are negative, imaging upscaling can be performed, e.g. with 11 C-choline PET/CT. More research into the costeffectiveness of 11 C-choline PET/CT is warranted to study its feasibility as a first-line imaging technique.
The studied imaging technique, 11 C-choline PET/CT, has potential drawbacks, such as that the tracer production of 11 C-choline is complicated and expensive, includes the need for a cyclotron on site because of its short half-life (20 min) and is therefore restricted in its availability. Therefore, other PET imaging techniques for the detection of parathyroid lesions have been evaluated. 18 F-fluorocholine, a tracer comparable to 11 C-choline, is a more widely available tracer because of its stable biodistribution and longer half-life (110 min). It, therefore, does not require a cyclotron on site. However, biodistributions of these tracers may not be completely identical. This will remain an open question since a direct comparison between the two choline tracers will not likely be performed. Thus far, the reported sensitivity of 11 C-choline for the detection of parathyroid adenomas is in the range of 92-99%, and studies utilizing 18 F-fluorocholine report 90-96%, indicating a near-similar performance [8-10, 16, 18]. We estimate a doubling of radiation exposure for 18 F-fluorocholine compared to 11 C-choline based on differences in half-life time, positron-energy and biological processes such as metabolism and excretion [27]. Therefore, for our clinical setting, 11 C-choline seems to be the most feasible. Another feasible option for the detection of parathyroid lesions is 11 C-methionine PET/CT or four dimensional-CT, with a sensitivity of 72-86% and 80-90%, respectively [6,7,11,12,[28][29][30]. In our center, we found a sensitivity of 72% for 11 C-methionine PET/ CT in the same clinical setting as the current study [6]. A head-to-head comparison, including a cost-effectiveness analysis, is needed to assess which of these imaging techniques is superior.
Our study has known limitations for retrospective analyses. First, this study was subjected to a referral bias since we only included patients who underwent 11 C-choline PET/CT and subsequent parathyroid surgery. However, we feel that referral bias was not substantial since in 75% of patients not undergoing surgery, a possible location for a parathyroid lesion was identified on 11 C-choline PET/ CT. Assuming that in the non-operated patients 11 C-choline PET/CT was correctly positive in 12 patients (75%) and false negative in four patients (25%), this would still have resulted in a high sensitivity of 90% (47/52) for this technique patient-based. Second, due to the retrospective nature, there are missing values in the follow-up. Most patients were lost to follow-up two months postoperatively. Thus, we cannot make firm conclusions about the final cure rate of this patient cohort. However, this was not the aim of our retrospective analysis since our primary objective was to correlate preoperative 11 C-choline PET with surgical localization of parathyroid glands and pathology outcome. Another potential drawback is that not all parathyroid glands were routinely inspected during surgery. This was only the case in 8/36 patients. Some positive lesions were therefore not confirmed as true or false positive, which might also have overestimated the sensitivity of 11 C-choline PET/CT.

Conclusion
In patients with pHPT and prior negative or discordant first-line imaging results, pathological parathyroid glands can be localized preoperatively with high accuracy by 11 C-choline PET/CT. In this challenging clinical setting, 11 C-choline PET/CT showed a high sensitivity, PPV and accuracy of 97%, 95% and 94%, respectively. The use of a highly sensitive preoperative imaging technique such as 11 C-choline PET/CT leads to more focused surgical procedures instead of bilateral neck explorations. Further studies are needed, including a head-to-head comparison and cost-effectiveness analysis on the various anatomical and nuclear imaging techniques to determine the order in which to use them in the setting of pHPT.