Abstract
Purpose
Patients with primary hyperparathyroidism routinely undergo preoperative imaging to localize the abnormal gland to facilitate a guided parathyroidectomy. These techniques include neck ultrasound (US), dual phase planar technetium-99 m ( 99m TC) sestamibi (MIBI) scans, single photon emission computed tomography (SPECT), combined SPECT/CT, and four dimensional CT scans (4D CT). Despite appropriate preoperative imaging, non-localization of abnormal glands does occur. This study aims to determine whether non-localization is the result of radiologic interpretive error or a representation of a subset of truly non-localizing parathyroid adenomas.
Materials and methods
A retrospective study was performed; two senior radiologists reinterpreted the preoperative imaging (US and MIBI scans) of 30 patients with initially non-localizing studies. All patients underwent parathyroidectomy for primary hyperparathyroidism at a tertiary referral center. Both radiologists were blinded to the scores of his colleague. The results were compared for inter-reader reliability using Cohen’s kappa test.
Results
Twenty-nine of thirty nuclear studies were found to be negative on reinterpretation. The readers agreed in 86.67% of their observations, with a kappa (κ) value of 0.706 (SE = ± 0.131, 95% confidence interval for κ = 0.449–0.962). One of eighteen ultrasounds had positive localizations on reexamination, however, the inter-observer agreement was only 55.6%, with a kappa value of 0.351 (SE = ± 0.139, and 95% confidence interval for κ = 0.080–0.623). Overall, no statistically significant difference in preoperative and retrospective interpretation was found.
Conclusion
This study identifies a subset of parathyroid adenomas that do not localize on preoperative imaging despite sound radiographic evaluation.
1
Introduction
Primary hyperparathyroidism is an overproduction of parathyroid hormone from a single adenomatous gland (85%) or less commonly by multiple adenomatous or hyperplastic glands (15%) . Overproduction of the hormone may have a significant impact on the health of the affected individuals. Surgical excision of the hyperactive gland or glands is curative . Localization with pre-operative imaging allows for the possibility of performing a guided parathyroidectomy procedure rather than the traditional bilateral neck exploration . Guided procedures allow for smaller incisions, decreased operative time, and less associated morbidity .
To facilitate a guided parathyroidectomy, patients with primary hyperparathyroidism routinely undergo preoperative imaging to localize the abnormal gland . These techniques include neck ultrasound (US), dual phase planar technetium-99 m ( 99m TC) sestamibi (MIBI) scans, single photon emission computed tomography (SPECT), combined SPECT/CT, and four dimensional CT scans (4D CT) . A combination of imaging modalities is often used as no single imaging modality is 100% sensitive or specific. MIBI and US have reported sensitivities of 54% to 88% and 59% to 89% respectively .
While prior localization aids in surgical planning, a percentage of individuals fail to localize despite preoperative imaging. This study aims to investigate possible radiographic associations with non-localizing parathyroid adenomas. We aim to determine whether non-localization is the result of radiologic interpretive error or if it is truly a representation of pathologically non-localizing parathyroid adenomas.
2
Materials and methods
A retrospective chart review of 364 patients who underwent parathyroidectomy by two parathyroid surgeons over a 5-year period at the Penn State Milton S. Hershey Medical Center was performed in order to identify patients with non-localizing adenomas. Inclusion criterion was determined as patients undergoing their first surgery for treatment for primary hyperparathyroidism with negative preoperative imaging. Variables used for statistical analysis included age, gender and, preoperative imaging results. Also included were surgical and pathologic findings including gland location, single or multiple gland disease, preoperative parathyroid hormone (PTH) and calcium levels, intra-operative PTH values, and postoperative outcomes. After review, patients with multiple endocrine neoplasia (MEN) as well as those with secondary, tertiary, recurrent, persistent, or non-MEN-familial hyperparathyroidism were excluded in order to determine the sensitivity of imaging for hyperactive glands. Patients who were included in the study were those who had available studies with reports within the electronic medical record and imaging system prior to surgery. Of the 364 patients reviewed, 30 patients met inclusion criteria for this study.
Technetium-99 m sestamibi imaging was performed at our center using the conventional dual phase technique, with immediate and delayed planar images. After administration of the standard dose of radioisotope (methoxyisobutylisonitrile “MIBI,” labeled with Technetium-99 m), anterior planar images of the neck and upper chest were obtained immediately and again two hours after radiopharmaceutical injection. The initial images corresponded to the thyroid phase. The delayed images corresponded to the parathyroid phases, where the radioisotope had washed out of the thyroid gland but was retained within the parathyroid adenoma. SPECT or SPECT/CT images were acquired of the neck and chest in most cases. For those who underwent SPECT/CT, a low dose computed tomographic (CT) study was obtained concurrent with the SPECT acquisition; the SPECT and CT images were reconstructed in the standard axial, coronal, and sagittal projections and were fused ( Fig. 1 ).
Two senior radiologists, proficient in both nuclear and ultrasound radiology from our tertiary center, reinterpreted the images of the 30 patients in this study independently of each other. These radiologists were not the original readers of the images. A simple, semi-quantitative scoring system was created for this study. Both radiologists reviewed and scored the studies separately, blinded to the scores of his colleague. The two readers’ results were subsequently compared for inter-reader reliability using Cohen’s kappa test. The scoring system was as follows:
- MIBI Score
Absent uptake = 0; Equivocal = 1 and Positive uptake = 2. This score was applied to all MIBI cases, with or without SPECT/CT.
- US Score
Clearly absent/lesion not seen = 0; Equivocal visualization of a parathyroid adenoma = 1 (equivocal lesions included inconspicuous lesion poorly delineated or with confounding findings and any which were not 100% certain); clearly positive and easily seen on US = 2.
2
Materials and methods
A retrospective chart review of 364 patients who underwent parathyroidectomy by two parathyroid surgeons over a 5-year period at the Penn State Milton S. Hershey Medical Center was performed in order to identify patients with non-localizing adenomas. Inclusion criterion was determined as patients undergoing their first surgery for treatment for primary hyperparathyroidism with negative preoperative imaging. Variables used for statistical analysis included age, gender and, preoperative imaging results. Also included were surgical and pathologic findings including gland location, single or multiple gland disease, preoperative parathyroid hormone (PTH) and calcium levels, intra-operative PTH values, and postoperative outcomes. After review, patients with multiple endocrine neoplasia (MEN) as well as those with secondary, tertiary, recurrent, persistent, or non-MEN-familial hyperparathyroidism were excluded in order to determine the sensitivity of imaging for hyperactive glands. Patients who were included in the study were those who had available studies with reports within the electronic medical record and imaging system prior to surgery. Of the 364 patients reviewed, 30 patients met inclusion criteria for this study.
Technetium-99 m sestamibi imaging was performed at our center using the conventional dual phase technique, with immediate and delayed planar images. After administration of the standard dose of radioisotope (methoxyisobutylisonitrile “MIBI,” labeled with Technetium-99 m), anterior planar images of the neck and upper chest were obtained immediately and again two hours after radiopharmaceutical injection. The initial images corresponded to the thyroid phase. The delayed images corresponded to the parathyroid phases, where the radioisotope had washed out of the thyroid gland but was retained within the parathyroid adenoma. SPECT or SPECT/CT images were acquired of the neck and chest in most cases. For those who underwent SPECT/CT, a low dose computed tomographic (CT) study was obtained concurrent with the SPECT acquisition; the SPECT and CT images were reconstructed in the standard axial, coronal, and sagittal projections and were fused ( Fig. 1 ).
Two senior radiologists, proficient in both nuclear and ultrasound radiology from our tertiary center, reinterpreted the images of the 30 patients in this study independently of each other. These radiologists were not the original readers of the images. A simple, semi-quantitative scoring system was created for this study. Both radiologists reviewed and scored the studies separately, blinded to the scores of his colleague. The two readers’ results were subsequently compared for inter-reader reliability using Cohen’s kappa test. The scoring system was as follows:
- MIBI Score
Absent uptake = 0; Equivocal = 1 and Positive uptake = 2. This score was applied to all MIBI cases, with or without SPECT/CT.
- US Score
Clearly absent/lesion not seen = 0; Equivocal visualization of a parathyroid adenoma = 1 (equivocal lesions included inconspicuous lesion poorly delineated or with confounding findings and any which were not 100% certain); clearly positive and easily seen on US = 2.
3
Results
We identified 364 patients from our institutional parathyroid database who underwent parathyroidectomy between January 2007 and March 2013. The study cohort consisted of 30 (8.2%) patients who had imaging studies that did not localize a hyperfunctioning gland with US or nuclear scan. All patients in the study cohort had MIBI studies performed. Ten percent (n = 3) of patients underwent SPECT imaging and twenty-two (73%) had SPECT/CTs. Eighteen patients (60%) had ultrasounds performed by a radiologist.
The mean age of patients was 56 years (range of 30 to 78). Six were male (20%) and twenty-four were female (80%). Median preoperative level of calcium was 11.05 mg/dL (ranging from 9.4 to 11.9 mg/dL) and diagnostic PTH level was 149 pg/mL (range of 59 pg/mL to 395 pg/mL). Median preoperative PTH level was 191.5 pg/mL (ranging from 57.6 to 395 pg/mL). Immediate postoperative rapid PTH median was 46 pg/mL (range of 10 pg/mL to 116 pg/mL).
All cohorts had an abnormal gland removed at the time of surgery. Exploration began on the left side in 60% (n = 18) of cases, while 40% (n = 12) started on the right. Eighty percent (n = 24) were bilateral parathyroid explorations while the remaining twenty percent (n = 6) were unilateral. The surgeon began the exploration on the correct side of the offending gland 57% of the time. Intraoperatively, 10% (n = 3) of offending glands were found intrathyroidal, 3% (n = 1) in thymus remnants, and 3% (n = 1) in the tracheoesophageal groove. Two patients had multigland adenomas and both received bilateral explorations. The mean weight of glands removed was 0.47 g with a range of 0.03 to 3 g. Gland size ranged from 0.4 × 0.4 × 0.2 cm to 7.0 × 1.5 × 1.0 cm.
The two radiologists scored twenty-nine of thirty nuclear studies in our sample as negative on reinterpretation. One case was felt to have been positive in retrospect, despite having initially been interpreted as negative by another radiologist ( Fig. 2 ). On reinterpretation, both radiologists read this MIBI scan as positive for a left lower parathyroid adenoma. This corresponded with intraoperative findings of a left lower parathyroid adenoma within the tracheoesophageal groove. Overall, the readers agreed in 86.67% of their observations, with a kappa (κ) value of 0.706 (SE = ± 0.131, 95% confidence interval for κ = 0.449–0.962) ( Table 1 ).
