Improvements in imaging localization of parathyroid adenomas in patients with primary hyperparathyroidism (PHPT) before surgical excision have allowed the transition from traditional bilateral neck exploration to selective excision of an adenoma localized preoperatively. The newest techniques of minimally invasive parathyroidectomy require precise preoperative localization to be successful [1]. The current techniques used for routine preoperative imaging localization of a parathyroid adenoma primarily consist of nuclear medicine sestamibi scanning with or without cervical ultrasound. The two techniques are complementary: The nuclear medicine scan offers functional information, whereas ultrasound shows more detailed anatomic information.

Despite improvements in ultrasound and nuclear medicine technology, neither technique will detect all parathyroid adenomas. A recent meta-analysis showed ^99mTc-sestamibi scanning outperformed ultrasound for the detection of single adenomas with respective mean sensitivities mean of 88% versus 78%. Mean sensitivity for the detection of multiple adenomas was substantially less—30% for sestamibi scanning and 16% for ultrasound [2]. Some institutions are now routinely using SPECT/CT for localization as a single test that can offer both functional and anatomic information. A study by Lavely et al. [3] showed the addition of early phase SPECT/CT improved accuracy and interobserver agreement for the localization of adenomas. However, in a retrospective study published in 2005, Gayed et al. [4] found that SPECT/CT did not add significant clinical value compared with conventional planar and SPECT images.

Adenoma localization is especially important in the setting of persistent or recurrent PHPT after surgical exploration. Before proceeding with reoperation, the current consensus is that two concordant preoperative imaging studies should localize hyperfunctional parathyroid tissue to the same anatomic region of the neck [5]. The goal of accurate localization is to minimize morbidity and maximize the likelihood of success [6]. Recent reports in the surgical literature describe the use of single-phase contrast-enhanced CT to localize parathyroid adenomas, particularly in patients who have already undergone surgical exploration of the neck [7, 8]. In a 2009 article, Randall et al. [9] described the imaging features of parathyroid adenomas on single-phase contrast-enhanced CT. Retrospectively, they found a clearly visible correlation of CT findings with surgical findings in 69.6% of adenomas and an additional 17.7% of adenomas were considered probably seen but the quality of the CT images were limited by technical factors.

Four-dimensional CT has been described as a potentially more robust method of localization because it exploits the differential early enhancement of parathyroid adenomas [1, 6]. The sensitivity for precise localization to a quadrant of the neck has been found to be higher for 4D CT (range, 70–88%) than for sestamibi scanning (range, 33–65%) and ultrasound (range, 29–57%) [1, 6].

The purpose of this study, which was performed at our institution, was to evaluate the accuracy of 4D CT in the preoperative localization of suspected parathyroid adenomas in patients with previously unsuccessful surgery or negative nuclear medicine and ultrasound examinations through correlation of 4D CT findings with surgical, pathologic, and clinical outcomes.

Patient demographics and surgical history and the locations of the adenomas are summarized in Table 1. The mean sensitivity and specificity for precise CT localization of surgically proven adenomas was 82% (range, 79–88%) and 92% (75–100%), respectively. Sensitivity and specificity, respectively, for reader 1 was 87.5% (95% CI, 67.6–97.3%) and 100% (39.8–100%); reader 2, 79.2% (57.9–92.9%) and 100% (39.8–100%); and reader 3, 79.2% (57.9–92.9%) and 75% (19.4–99.4%). Interobserver reliability ranged from fair to excellent across all reader pairs and adenoma locations (κ = 0.18–1.0), with all but three kappa values ≥ 0.60. There were two low kappas at location B, where readers 1 and 3 disagreed with reader 2 (κ = 0.17 and 0.26, respectively). These cases amounted to five and six disagreements, respectively. There was also a kappa of 0.46 between readers 1 and 3 at location E, left of midline, with two disagreements. Despite these lower kappas, there was no evidence of significant differences among these kappas and the overall kappa of 0.70 (0.60–0.79) was excellent. Retrospective review of the 4D CT studies after unblinding of the surgical results allowed identification of the surgically proven parathyroid adenomas in all cases.

All of the patients with adenomas resected at surgery showed a biochemical response, with a mean preoperative parathyroid hormone assay of 128 pg/mL (range, 63–237 pg/mL) and 20-minute postparathyroidectomy hormone assay of 27 pg/mL (range, 7–65 pg/mL). The mean weight of the resected parathyroid adenomas was 627 mg (range, 118–2,583 mg). The mean adenoma diameter at pathology was 10 mm (range, 5–18.7 mm).

The maximum mean densities of the confirmed adenomas were 41, 128, 138, and 109 HU at 0, 30, 60, and 90 seconds, respectively. In 10 patients, normal lateral compartment lymph nodes were identified with mean HU densities of 39, 58, 92, and 100 HU at 0, 30, 60, and 90 seconds, respectively. Compared with enhancement of the lymph nodes, enhancement of the adenomas was significantly greater in the same patients at 30 (p = 0.0001) and 60 (p = 0.006) seconds, but not at 90 seconds (p = 0.5953) (Fig. 2).

Our study results showed a mean sensitivity and specificity of 82% and 92%, respectively, for the precise 4D CT localization of occult parathyroid adenomas when prior surgery or localization with ultrasound and sestamibi nuclear medicine scans has been unsuccessful. In addition, we found that adenomas show a characteristic contrast enhancement pattern of early enhancement. This enhancement pattern was significantly different at the 30- and 60-second phases when compared with the progressive enhancement pattern shown in normal lymph nodes (Fig. 3). Along with knowledge of the potential normal and ectopic locations of parathyroid adenomas, this imaging characteristic of adenomas should allow accurate preoperative localization.

Our results are concordant with those of prior studies that have shown 4D CT can localize parathyroid adenomas not seen on ultrasound or nuclear medicine studies [1, 6]. A recent study from 2008 showed significantly improved localization over sestamibi scanning with a sensitivity of 88% for 4D CT versus 54% for sestamibi scanning [6]; all of the patients in that study had undergone prior surgical neck exploration.

As a group, patients with PHPT have a higher risk of surgical failure at repeat surgery, making accurate localization preoperatively even more important. Multiple prior studies have shown recurrent laryngeal nerve injury rates as high as 8% and permanent hypocalcemia in as many as 13% of patients who underwent reoperation for hyperparathyroidism [10–12]. The increased morbidity of repeat surgery also makes precise localization and minimized surgical exploration highly desirable.

Although the sensitivity of 4D CT in our study compares favorably with the sensitivities of ultrasound and nuclear medicine studies [2], 4D CT was performed when nuclear medicine and ultrasound had failed to localize an adenoma. Therefore, although we are unable to directly compare the accuracy of 4D CT in our series with that of ultrasound or nuclear medicine, 4D CT appears to be a valuable addition to the imaging workup in this subset of patients.

Given that most parathyroid adenomas are found using a combined ultrasound and nuclear medicine imaging algorithm, we are not proposing the use of 4D CT in the standard evaluation. However, based on our results, 4D CT will localize most the adenomas successfully in patients in whom conventional imaging is unsuccessful. This knowledge will enable the surgeon to minimize operative times and, if possible, to select a minimally invasive approach. Although 4D CT may offer improved localization in the routine workup of PHPT [1], the increased radiation dose should limit the use of 4D CT for problem-solving cases. In consideration of radiation dose, the results of our study suggest that we may eliminate the 90-second phase because there was no significant difference in enhancement between adenomas and lymph nodes.

MRI may also offer an alternative to CT localization. Although performed in a small number of patients, a recent series showed that the sensitivity for adenoma localization of conventional CT (50%) and that of MRI (47%) were similar in patients undergoing repeat parathyroidectomy [13]. Another recent study showed that MRI may be an effective addition to the imaging algorithm to localize abnormal glands when nuclear medicine and ultrasound findings conflict [14].

In addition to differentiating parathyroid adenomas from lymph nodes, other pitfalls to CT interpretation include prior thyroid surgery, normal vascular structures, and exophytic thyroid nodules [9]. Hypervascular nodal metastases could potentially appear similar to an enhancing parathyroid adenoma and clinical history will be important when interpreting these cases. Besides the importance of identifying normal enhancement of vascular structures on the arterial phase, streak artifact from adjacent contrast material can obscure the lesion on the 30-second phase scan. Therefore, the patient's surgical history and prior imaging examination should be reviewed when selecting a protocol for the 4D CT examination. If the patient has undergone prior cervical exploration with removal of an adenoma, one should consider placing the IV line in the arm on the side of prior surgery given the higher likelihood of an adenoma being in the contralateral neck. This strategy should minimize streak artifact from dense inflowing contrast medium flowing into the subclavian vein and possibly refluxing into the internal jugular vein. In addition, after identifying one adenoma, it is critical to continue searching for additional adenomas because the discovery of a second adenoma may significantly alter the planned surgery and chances for success.

Although this study is limited by its retrospective nature, the design likely minimizes this limitation for several reasons. The 4D CT studies were interpreted in a blinded fashion, whereas in typical clinical practice one would know the clinical history and prior imaging and surgical results in patients, which often would aid in the interpretation. To minimize selection bias, all 4D CT studies performed at our institution were randomly interpreted and included patients who had not yet undergone surgery, patients who had negative four-gland exploration, patients who had multiple adenomas at surgery, and patients who did not have hyperparathyroidism at follow-up. We required commitment from the reader not only to call an examination positive but also to precisely localize the adenoma or adenomas. Because 4D CT is performed relatively infrequently, double or consensus interpretation, which was not done in our study, may increase reader accuracy. In addition, our readers had various levels of experience (1–3 years) interpreting 4D CT studies. This level of experience can be obtained relatively quickly in clinical practice and our overall interreader reliability was excellent.

In summary, parathyroid adenomas in patients with unsuccessful localization on conventional imaging can be accurately shown preoperatively on 4D CT. Parathyroid adenomas show characteristic rapid enhancement that can help differentiate these lesions from lymph nodes. Precise localization is critical to minimizing surgical morbidity, particularly in patients in whom prior surgery has failed.

Address correspondence to M. D. Beland ([email protected]).