All parathyroid lesions that were missed on preoperative 4-dimensional computed tomographic scans are shown. Most missed glands were found in the expected anatomic location at the time of surgery. Missed glands were more common in the inferior position.
The arrowheads point to the location of a candidate intrathyroidal lesion. The top panel shows that an ectopic right inferior gland in the mediastinum was missed on initial review. A, Axial noncontrast computed tomography through the superior mediastinum shows a small nodule anteromedial to the right brachiocephalic vein, just deep to the manubrium of the sternum. B, Axial arterial phase computed tomography shows some enhancement in this lesion, the degree of which is not well determined due to streak artifact from the adjacent inflowing contrast bolus (arrowhead). C, Axial venous phase computed tomography shows similar or slightly greater attenuation in this lesion compared with the arterial phase. This finding was misinterpreted as a reactive lymph node. Two other candidate lesions were identified on this study reflecting hyperplastic orthotopic right superior and left superior glands (not shown), both of which were also abnormal at surgery. The bottom panel shows that a right inferior intrathyroidal parathyroid lesion was mistaken on initial review as a thyroid nodule. D, Axial noncontrast computed tomography is shown. E, This lesion is hyperattenuating to the thyroid on the arterial phase. F, This lesion is isoattenuating to the thyroid on the venous phase. Although this enhancement pattern is also typical of parathyroid adenoma, the lesion was initially dismissed as a likely benign thyroid nodule due to the presence of 2 other high-suspicion candidates, both of which were also abnormal at surgery.
eFigure. Comparison of Preoperative US and 4D-CT Findings
Customize your JAMA Network experience by selecting one or more topics from the list below.
Sho S, Yuen AD, Yeh MW, Livhits MJ, Sepahdari AR. Factors Associated With Discordance Between Preoperative Parathyroid 4-Dimensional Computed Tomographic Scans and Intraoperative Findings During Parathyroidectomy. JAMA Surg. 2017;152(12):1141–1147. doi:10.1001/jamasurg.2017.2649
What factors contribute to discordance between preoperative parathyroid 4-dimensional computed tomographic scans and intraoperative findings?
In an analysis of a database that included 411 patients with primary hyperparathyroidism who underwent preoperative 4-dimensional computed tomographic scans followed by parathyroidectomy, we found that the presence of multigland disease, multinodular goiter or thyroid nodule, parathyroid lesion size of 10 mm or less, and parathyroid lesion in the inferior position were associated with discordance between preoperative parathyroid 4-dimensional computed tomographic scans and intraoperative findings.
Understanding potential limitations and factors affecting the accuracy of preoperative 4-dimensional computed tomographic scans may enable surgeons to fully leverage this new localization technique in preoperative planning and intraoperative troubleshooting.
Parathyroid 4-dimensional computed tomographic scans (4D-CTs) have emerged as an accurate and cost-effective initial localization study for patients with primary hyperparathyroidism. However, potential limitations and factors affecting the accuracy of preoperative 4D-CTs remain poorly defined.
To characterize factors associated with missed parathyroid lesions on preoperative 4D-CTs and to investigate patterns of commonly observed errors.
Design, Setting, and Participants
A prospectively accrued patient database was analyzed from September 1, 2011, through October 31, 2016. The study was performed in a tertiary referral center. Consecutive patients with primary hyperparathyroidism undergoing preoperative 4D-CTs and subsequent parathyroidectomy were included in the study.
Main Outcomes and Measures
Discordance between preoperative 4D-CTs and intraoperative findings in the number and location of abnormal parathyroid lesions.
Of 411 patients studied (mean [SD] age, 59  years; 325 [79.1%] female), 123 (29.9%) had discordance between preoperative 4D-CTs and intraoperative findings. Among the 411 patients, 75 (18.2%) had major discordance, including incorrectly localized adenoma on the contralateral side of the neck, missed double adenoma, and absence of any abnormal lesion detected on 4D-CTs. Compared with concordant cases, discordant cases had higher frequencies of multigland disease (66.7% [82 of 123] vs 24.3% [70 of 288], P < .001) and multinodular goiter or thyroid nodule (40.7% [50 of 123] vs 29.2% [84 of 288], P = .02). Missed parathyroid lesions were smaller (mean [SD], 0.86 [0.29] vs 1.24 [0.50] cm; P < .001) and were more likely to be in the inferior position (65.4% [87 of 133] vs 38.1% [177 of 465], P < .001). Parathyroid lesion size of 10 mm or less (odds ratio [OR], 4.37; 95% CI, 2.24-8.54), multigland disease (OR, 7.63; 95% CI, 3.49-16.69), multinodular goiter or thyroid nodule (OR, 1.82; 95% CI, 1.01-3.28), and parathyroid lesion in the inferior position (OR, 6.82; 95% CI, 3.10-14.99) were independently associated with discordant 4D-CT results.
Conclusions and Relevance
Multigland disease was most strongly associated with discordance between preoperative 4D-CTs and intraoperative findings, followed by parathyroid lesion in the inferior position and parathyroid lesion size of 10 mm or less. Awareness of these potential pitfalls may allow surgeons to better leverage this new localization technique in preoperative planning and intraoperative troubleshooting.
Preoperative imaging in patients with primary hyperparathyroidism guides surgical planning and improves operative outcomes.1-3 In addition to conventional localization techniques of ultrasound (US) and sestamibi single-photon emission computed tomography (SPECT), 4-dimensional computed tomographic scans (4D-CTs) have emerged in major hospitals as a highly accurate and cost-effective initial localizing study for patients with parathyroid disease.4-8
Although 4D-CTs have been shown to have superior sensitivity compared with US and sestamibi SPECT, missed lesions can still occur with this technique, which is especially true in cases of multigland disease (MGD), for which 4D-CT sensitivity remains as low as 32% to 53%.9-11 In clinical practice, failure of preoperative imaging to identify all abnormal parathyroid glands may increase the risk of failed parathyroidectomy. Unlike well-established localization techniques of US and sestamibi SPECT, 4D-CTs remain a new technique to many surgeons. As such, it is critical that surgeons become aware of potential limitations and factors that may influence the accuracy of 4D-CTs.
The aim of this study was to characterize factors associated with discordant results between preoperative 4D-CTs and intraoperative findings during parathyroidectomy. Furthermore, we sought to identify commonly observed error patterns, including locations of abnormal parathyroid glands that are more likely to be missed on 4D-CTs.
Patients with biochemical evidence of primary hyperparathyroidism who underwent 4D-CTs were included in a prospective database between September 1, 2011, and October 31, 2016. The study was performed in a tertiary referral center. Relevant clinical information, CT-derived data, and operative findings were recorded. Consecutive patients with primary hyperparathyroidism undergoing preoperative 4D-CTs and subsequent parathyroidectomy were analyzed in the present study, including primary and reoperative patients with parathyroidectomy. Patients who subsequently underwent a successful parathyroidectomy, defined as an intraoperative parathyroid hormone (PTH) level drop of 50% or more from the baseline value, were included in this study. For patients with available follow-up information, calcium levels were checked to assess for eucalcemia at or beyond 6 months after surgery. This study was performed with approval from the University of California, Los Angeles (UCLA) Institutional Review Board and with a waiver of informed consent and a waiver of Health Insurance Portability and Accountability Act of 1996 authorization.
The 4D-CT protocol included noncontrast, arterial phase, and delayed phase images from the hard palate to the carina as previously described.12 Parathyroid lesions were characterized using a system based on anteroposterior location relative to the expected course of the recurrent laryngeal nerve in the tracheoesophageal groove. Lesions posterior to the expected course of the nerve were defined as superior parathyroid glands, and those anterior to the recurrent laryngeal nerve were defined as inferior parathyroid glands. Location along a superoinferior axis was also described using the thyroid isthmus and the lower edge of the thyroid gland as landmarks. Radiology reports were generated by one of us (A.R.S.), who is a subspecialty-certified neuroradiologist with more than 10 years of experience in CT interpretation, including 6 years of experience in interpretation of 4D-CTs.
All parathyroid lesions were classified as correctly or incorrectly localized on 4D-CTs by correlating the operative notes with the original radiology reports and using anatomic landmarks reported in the operative and radiology reports. Gland sizes were obtained from the operative notes and were based on the largest dimension measured. Determination of single-gland disease (SGD) (ie, 1 abnormal gland) or MGD (ie, 2-3 abnormal glands or 4-gland hyperplasia) was made based on operative findings (size, color, and overall appearance) in conjunction with sufficient drop in PTH level (intraoperative PTH level drop ≥50%). Based on surgeon preference, patients included in the study underwent either routine 4-gland exploration with direct visual inspection of all parathyroid glands or focused image-guided parathyroidectomy. To avoid the surgical approach from confounding the evaluation of 4D-CT accuracy, only those glands that were removed for high clinical suspicion with pathological confirmation of hypercellular parathyroid tissue were considered for determination of 4D-CT accuracy. Correct localization of all abnormal glands on 4D-CTs was defined as concordant, while those with 1 or more missed or incorrectly localized lesions were defined as discordant. Surgical findings were considered to be the criterion standard for identifying abnormal parathyroid glands.
Discordant 4D-CTs were further classified as having major or minor discordance based on potential implications for the surgical approach and outcomes. Major discordance included the following types of errors: (1) abnormal gland found on the opposite side of the neck (right vs left), (2) SGD indicated by 4D-CTs but MGD found at surgery, (3) MGD indicated by 4D-CTs but SGD found at surgery, and (4) no abnormal lesions found on 4D-CTs. These types of discordance can either lead to failed unilateral neck exploration (1 and 2) or influence surgeons to choose 4-gland exploration when focused parathyroidectomy was in fact possible (3 and 4). Minor discordance included abnormal gland found on the same side of the neck but in the wrong location (superior vs inferior) and MGD found on both 4D-CTs and surgery but with the wrong number or location of abnormal lesions noted on 4D-CTs. These errors were deemed minor because they have minimal influence on surgical planning and outcomes.
Concordant and discordant cases were compared to identify factors contributing to inaccurate lesion localization by 4D-CTs. Clinical factors included age, sex, body mass index (BMI), SGD vs MGD, lesion size and location, multinodular goiter or thyroid nodule (>5 mm), prior thyroid or parathyroid surgery, and history of hypothyroidism. These factors were further evaluated using univariate and multivariable logistic regression analysis. The χ2 test was used to evaluate categorical data, and the t test was used for continuous variables. Two-sided P < .05 was the threshold used for statistical significance for all tests. Statistical analyses were performed using a software program (SAS Enterprise, version 4.2; SAS Institute Inc).
During the study period, there were 640 patients with primary hyperparathyroidism who underwent a total of 660 4D-CTs. Patients were excluded if they did not have a parathyroidectomy performed at UCLA (n = 227) or if their parathyroidectomy was not successful (intraoperative PTH level drop <50%) (n = 2). The final study cohort consisted of the remaining 411 patients (mean [SD] age, 59  years; 325 [79.1%] female), 123 (29.9%). Follow-up information of at least 6 months was available for 165 of 411 patients (40.1%). For patients with 6-month follow-up data available, all had remained eucalcemic at or beyond the 6-month benchmark.
Discordance between preoperative 4D-CTs and intraoperative findings was noted in 123 of 411 patients (29.9%). Among these patients, 107 cases were associated with 1 or more missed parathyroid adenomas. Preoperative, surgeon-performed US was done in all patients. The US and 4D-CT results were similar in 184 of 411 patients (44.8%). When US and 4D-CT findings did not match, most intraoperative findings were concordant with the 4D-CT findings as opposed to the US findings (178 of 227 [78.4%] vs 20 of 227 [8.8%]); in the remaining cases (12.7%), the intraoperative finding matched with neither US nor 4D-CTs (eFigure in the Supplement). In 20 cases, preoperative US had identified the parathyroid lesion missed on 4D-CTs. The sensitivity and specificity for quadrant localization in SGD were 90.7% and 94.7%, respectively. For MGD detection, the sensitivity and specificity were 45.3% and 87.2%, respectively. When defining an accurate 4D-CT localization study as lacking major discordance, the overall accuracy was 81.7%. Using a stricter definition for accurate localization (ie, lack of major and minor discordances), the overall accuracy was 70.0%.
Compared with concordant cases, discordant cases were associated with higher frequencies of MGD (66.7% [82 of 123] vs 24.3% [70 of 288], P < .001) and multinodular goiter or thyroid nodule (40.7% [50 of 123] vs 29.2% [84 of 288], P = .02). Abnormal glands missed on 4D-CTs were smaller compared with those that were identified before surgery (mean [SD], 0.86 [0.29] vs 1.24 [0.50] cm; P < .001) and were more likely to be in the inferior position (65.4% [87 of 133] vs 38.1% [177 of 465], P < .001). Patients with prior thyroid or parathyroid surgery were observed less frequently in the discordant group. Intended surgical approach was not associated with concordance between preoperative 4D-CTs and intraoperative findings (Table 1).
In univariate analysis, factors associated with discordant 4D-CT findings included parathyroid lesion size of 10 mm or less (odds ratio [OR], 8.27; 95% CI, 4.62-14.82), MGD (OR, 5.68; 95% CI, 3.59-8.98), multinodular goiter or thyroid nodule (OR, 1.66; 95% CI, 1.07-2.58), and parathyroid lesion in the inferior position (OR, 4.85; 95% CI, 2.57-9.15). Prior thyroid or parathyroid surgery was protective against discordant 4D-CT findings (OR, 0.32; 95% CI, 0.18-0.60), while BMI and intended surgical approach were not statistically significant predictors (Table 2).
In multivariable analysis, parathyroid lesion size of 10 mm or less (OR, 4.37; 95% CI, 2.24-8.54), MGD (OR, 7.63; 95% CI, 3.49-16.69), multinodular goiter or thyroid nodule (OR, 1.82; 95% CI, 1.01-3.28), and parathyroid lesion in the inferior position (OR, 6.82; 95% CI, 3.10-14.99) remained associated with discordance between preoperative 4D-CTs and intraoperative findings. Prior thyroid or parathyroid surgery remained protective against discordant 4D-CT results (Table 2).
We analyzed the 4D-CT results and corresponding surgical findings of the 123 discordant cases. We found major discordances in 75 cases, of which 41.3% (n = 31) occurred in patients with SGD found at the time of surgery (Table 3). The most common errors included SGD incorrectly identified as MGD on 4D-CTs (n = 16) and no abnormal glands detected on 4D-CTs (n = 12). Finding a single abnormal gland on the side opposite from 4D-CT localization was rare, occurring in only 3 patients. Errors leading to missed MGD were also common, occurring in 58.7% (44 of 75). These cases were frequently misdiagnosed as SGD on preoperative imaging when in fact patients had MGD identified during surgery.
Minor discordance was identified in 48 cases (Table 3). Most minor discordances were in patients with MGD, occurring in 81.3% (n = 39). These discordances were due to correct identification of MGD on 4D-CTs but incorrect numbers or locations of abnormal glands.
Locations of missed parathyroid lesions were determined by correlating the operative notes with the corresponding original radiology reports. Most missed glands were found in the expected anatomic location at the time of surgery (Figure 1). Missed glands were more common in the inferior position (34.5% [46 of 133] superior vs 65.4% [87 of 133] inferior). Glands that were either intrathyroidal or flattened against the thyroid comprised 7.5% (2 of 133) of all missed glands. Figure 2 (top panel) shows a missed inferior mediastinal lesion in a patient with MGD. Figure 2 (bottom panel) shows a missed right inferior intrathyroidal lesion that was mistaken for a thyroid nodule in a patient with MGD.
We performed a subgroup analysis among only patients with confirmed eucalcemia at or beyond 6 months after surgery (n = 165). A multivariable analysis within this subgroup identified the same variables as found above (Clinical Factors Associated With Discordant 4D-CT Findings subsection in the Results section) to be independent risk factors of discordant 4D-CT results, including the following: parathyroid lesion size of 10 mm or less (OR, 6.52; 95% CI, 2.59-18.45), MGD (OR, 5.27; 95% CI, 1.85-16.38), multinodular goiter or thyroid nodule (OR, 2.28; 95% CI, 1.05-5.09), and parathyroid lesion in the inferior position (OR, 7.41; 95% CI, 2.73-22.76).
Parathyroid 4D-CTs have emerged as one of the most accurate preoperative imaging modalities to localize abnormal parathyroid glands.10,11 Despite this use, missed lesions and incorrect localization still occur. To our knowledge, the present study represents the largest series to date evaluating factors associated with discordance between preoperative 4D-CTs and intraoperative findings. Parathyroid lesion size of 10 mm or less, MGD, multinodular goiter or thyroid nodule, and parathyroid lesion in the inferior position were found to be risk factors for discordant 4D-CT findings.
The discordance rate between preoperative 4D-CTs and intraoperative findings was 29.9% (123 of 411) in our study. Most were cases of major discordance that could potentially influence the surgical approach and outcome. Multigland disease was most strongly associated with discordance between preoperative 4D-CTs and intraoperative findings, which is consistent with previous reports that suggest suboptimal sensitivities of 4D-CTs for MGD at 32% to 53%.4,10,11 While some studies9,11 have shown superior sensitivity of 4D-CTs in detecting MGDs over US or sestamibi SPECT, our findings indicate that MGD remains a major source of error in 4D-CTs. This result is reflected in the finding that 67.4% (83 of 123) of all errors in 4D-CT localization occurred in patients with MGD. Furthermore, MGD was the most common source of error in 4D-CT localization studies overall, suggesting that the reported relative superiority of 4D-CTs over US and sestamibi SPECT in MGD detection is only modest. Although 4D-CTs may be the least operator dependent compared with US or sestamibi SPECT, detection of MGD continues to remain a significant challenge similar to other existing localization techniques. Our group and others have previously developed a prospectively validated scoring system to predict the risk of MGD using data from 4D-CTs and biochemical factors.13-16 Routine use of the scoring system in conjunction with 4D-CTs may improve preoperative identification of patients with MGD, in whom surgeons must be aware of the important limitations of 4D-CTs in guiding surgical planning.
Multinodular goiter or thyroid nodule was also associated with a higher risk of 4D-CT discordance. A prior study by Galvin et al9 also identified thyroid nodules as a potential risk factor for missed parathyroid lesions on 4D-CTs. Thyroid goiters show nodular projections of thyroid tissue that may be intrinsically low density yet demonstrate a wash-in or washout pattern of enhancement, making it difficult to distinguish between thyroid tissue and a juxtathyroidal parathyroid adenoma. Thyroid nodules can mimic parathyroid adenomas because they can occur in similar locations and appear oval or round, and they can have enhancement characteristics similar to those of parathyroid adenomas. The addition of US may enable correct identification of abnormal parathyroid glands in a patient with thyroid nodules.
Prior results have indicated that parathyroid lesion size may be an important factor in predicting successful preoperative localization, but the size limit of detection with 4D-CTs is largely unknown.17 Our findings suggest that the risk of missed glands increases below 10 mm. Because small lesions (<7 mm) occur more frequently in MGD, detection of a small abnormal lesion or no lesions should alert surgeons to the increased possibility of MGD.13,18 In addition, parathyroid lesions in the inferior position, especially partially or fully intrathyroidal glands or those flattened against the surface of the thyroid gland, were more likely to be missed on 4D-CTs. Inferior parathyroid adenomas are often in close proximity to inferior thyroid veins. As a result, vascular enhancement can obscure these inferior parathyroid lesions. Streak artifact from the inflowing contrast bolus, beam hardening artifact from the bony structures and soft tissues near the thoracic inlet, and residual thymic tissue can also obscure inferior parathyroid lesions. It is important to note that these potential “blind spots” on 4D-CTs are areas that are easily accessible and straightforward to evaluate during surgery. On the other hand, lesions that are missed by US and sestamibi SPECT are often located in areas that are more difficult to explore surgically, such as the posterior or retroesophageal location. The upshot of this finding in our practice has been that 4D-CTs, even when nonlocalizing or discordant, are highly predictive of surgical success because lesions that are difficult to access have been effectively ruled out.
Although BMI has been shown to reduce the accuracy of US and sestamibi SPECT for parathyroid localization,17 it did not influence 4D-CT performance in our analysis. Therefore, obese patients may benefit from having 4D-CTs as the initial localization study. Furthermore, patients with prior thyroidectomy or parathyroidectomy were less likely to have discordant 4D-CTs in our analysis. The 4D-CTs provide superior preoperative localization in reoperative cases over sestamibi scintigraphy.19 Although prior surgery can distort normal anatomy, previous thyroidectomy or parathyroidectomy can also simplify radiographic interpretation. Given that multinodular goiter or thyroid nodule was found to be a predisposing factor for discordant 4D-CTs, it follows that their removal can improve localization. Removal of parathyroid glands reduces the number of potential targets, and information gained from a prior neck operation may facilitate subsequent 4D-CT interpretation. Finally, due to the increased risk of reoperative parathyroidectomy, patients without a convincing target on 4D-CTs may not be offered surgery, which may also explain why fewer discordant results were noted in patients with a prior neck operation.
In the present analysis, we have identified several important risk factors associated with potential inaccuracies in 4D-CT localization. These findings afford a more thorough understanding of 4D-CT performance characteristics, which can help surgeons improve their preoperative planning and intraoperative troubleshooting. In the case of a negative 4D-CT, the surgeon may elect to allot additional operative time for what may be a more difficult case. Or, after considering the likelihood of MGD based on biochemical values and the 4D-CT result,14 the surgeon may consider having a more detailed discussion with the patient regarding the potential need for subtotal parathyroidectomy. During surgery, if the surgeon is either not finding the culprit glands or if the PTH level is not dropping, he or she should recall that discordance between intraoperative findings and the 4D-CT results is likely to be explained by MGD, an inferior gland that is flattened against the surface of the thyroid gland, or, less commonly, an intrathyroidal gland.
We acknowledge some limitations to this study. First, given the retrospective study design, follow-up calcium levels were not available in all patients. Consistent with prior studies,8,9 we used the intraoperative PTH level to define a successful parathyroidectomy. We found that all patients who had at least a 50% drop in the intraoperative PTH level and available follow-up data at 6 months or beyond remained eucalcemic. Second, our study was performed within a single institution with one highly experienced neuroradiologist reading all studies. The 4D-CT technique remains novel; therefore, variations in imaging protocol and quality frequently exist among different centers. These factors may limit the generalizability of our results. The evaluation of 4D-CTs in a multi-institutional setting with several interpreting radiologists may further elucidate the potential of this localization technique.
We identified MGD, multinodular goiter or thyroid nodule, small parathyroid lesion size, and parathyroid lesion in the inferior position as factors associated with discordance between parathyroid preoperative 4D-CTs and intraoperative findings. Increased awareness of patterns of potentially missed lesions on 4D-CTs may improve preoperative planning and intraoperative troubleshooting.
Accepted for Publication: April 28, 2017.
Corresponding Author: Shonan Sho, MD, Section of Endocrine Surgery, David Geffen School of Medicine, University of California, Los Angeles, 757 Westwood Plaza, Los Angeles, CA 90095 (firstname.lastname@example.org).
Published Online: August 9, 2017. doi:10.1001/jamasurg.2017.2649
Author Contributions: Dr Sho had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Dr Sho and Mr Yuen contributed equally to this work.
Study concept and design: Sho, Yuen, Yeh, Sepahdari.
Acquisition, analysis, or interpretation of data: Sho, Yuen, Livhits, Sepahdari.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sho, Yuen.
Obtained funding: Yeh.
Administrative, technical, or material support: Yuen, Livhits.
Study supervision: Yeh, Livhits, Sepahdari.
Conflict of Interest Disclosures: None reported.