ORIGINAL ARTICLE
A Prospective Comparative Study of Parathyroid Dual-Phase Scintigraphy, Dual-Isotope Subtraction Scintigraphy, 4D-CT, and Ultrasonography in Primary Hyperparathyroidism Martin Krakauer, PhD,* Bente Wieslander, PhD,* Peter S. Myschetzky, MD,† Anke Lundstrøm, DMSc,† Theis Bacher, MSc,* Christian H. Sørensen, DMSc,‡ Waldemar Trolle, MD,‡ Birte Nygaard, DMSc,§ and Finn N. Bennedbæk, PhD§ Purpose: Preoperative localization of the diseased parathyroid gland(s) in primary hyperparathyroidism allows for minimally invasive surgery. This study was designed to establish the optimal first-line preoperative imaging modality. Patients and Methods: Ninety-one patients were studied consecutively in a prospective head-to-head comparison of dual isotope (99mTc-MIBI vs 123I) subtraction parathyroid scintigraphy (PS), dual-phase PS, 4-dimensional (4D) CT, and ultrasonography (US). Surgery, histological confirmation, and postoperative normalization of Ca++ and parathyroid hormone were the reference standard. Results: Ninety-seven hyperfunctioning parathyroid glands (HPGs) were identified by the reference standard. Sensitivity and specificity for subtraction PS, dual-phase PS, 4D-CT, and US were 93%, 65%, 58%, and 57% as well as 99%, 99.6%, 86%, and 95%, respectively. Interrater agreement was excellent for subtraction PS (κ = 0.96) while only fair for 4D-CT (κ = 0.34). Pinhole imaging and subtraction of delayed images (the latter especially in case of a nodular thyroid gland) increased the sensitivity of subtraction PS. SPECT/low-dose CT did not increase sensitivity but aided in the exact localization of the HPGs. Of 7 negative subtraction PS studies, 4D-CT and US were able to locate 3 and 1 additional HPGs, respectively. Conclusions: Dual isotope pinhole subtraction PS has higher diagnostic accuracy compared with dual-phase PS, 4D-CT, and US as a first-line imaging study in primary hyperparathyroidism. In case of a negative scintigraphy or suspicion of multiglandular disease, 4D-CT and/or US is recommended as a second-line modality. However, diagnostic algorithms should be adapted in accordance with local availability and expertise. Key Words: hyperparathyroidism, scintigraphy, subtraction, 4D-CT, ultrasonography (Clin Nucl Med 2016;41: 93–100)
P
rimary hyperparathyroidism (PHP) is characterized by elevated levels of blood calcium due to the hypersecretion of parathyroid hormone (PTH) from 1 or more hyperfunctioning parathyroid glands (HPGs). Definitive treatment is parathyroidectomy (PTx) preferably with a minimally invasive technique.1,2 Preoperative imaging allows for minimally invasive PTx. Imaging modalities include parathyroid scintigraphy (PS),3 ultrasonography (US),4 4-dimensional (4D) CT,5,6 and, recently, 11C-methionine PET/CT7 and 18F-choline PET/CT.8 Although promising, the PETbased modalities are still not widely available. Received for publication May 26, 2015; revision accepted July 26, 2015. From the Departments of *Nuclear Medicine and †Radiology, Gentofte University Hospital, Hellerup; ‡Department of Otolaryngology, Head and Neck Surgery, Copenhagen University Hospital, Rigshospitalet, Copenhagen; and §Department of Endocrinology, Herlev University Hospital, Herlev, Denmark. Conflicts of interest and sources of funding: none declared. Correspondence to: Martin Krakauer, MD, PHD, Department of Nuclear Medicine, Gentofte Hospital, Kildegaardsvej 28, DK-2900 Hellerup, Denmark. E-mail: [email protected]. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0363-9762/16/4102–0093 DOI: 10.1097/RLU.0000000000000988
Many studies have evaluated the available imaging modalities, but few have assessed them prospectively in a head-to-head comparison. Moreover, imaging protocols vary considerably between studies. Parathyroid scintigraphy can be performed with subtraction scintigraphy or, more commonly, as a dual-phase study. In addition, there are differences between studies of PS in terms of tracers, collimator types, energy windows, software, and the use of SPECT or SPECT/CT. Ultrasonography can be performed with or without color Doppler. Accuracy is dependent in the experience of the ultrasonographer and the presence of thyroid pathology.9 Four-dimensional CT exploits the differential washout of an iodinated contrast agent from HPGs using multiple acquisitions in various contrast phases. Four-dimensional CT protocols also differ between studies.10 We wanted to establish the optimal imaging modality in a prospective, blinded, head-to-head comparison of subtraction PS, dual-phase PS, 4D-CT, and US in consecutive patients with PHP eligible for PTx.
PATIENTS AND METHODS Patients Patients were recruited consecutively from the endocrinology department. We included adult patients with PHP who were candidates for PTx according to national guidelines. Exclusion criteria were pregnancy, allergy to iodine-containing contrast agents, and impaired renal function. If treated with thyroid hormone or cinacalcet, patients were instructed to pause this for 3 weeks and 5 days, respectively, before imaging (except for US). Informed written consent was obtained from all participants, and the protocol was approved by the local ethics committee. All patients underwent preoperative PS and 4D-CT on the same day. Six patients did not undergo US due to logistical reasons. The operating surgeons were experienced head and neck surgeons and had the results of all modalities available when deciding the surgical approach. A drop of more than 50% in per-operative plasma PTH levels and a minimum of 3 months' follow-up were used to verify successful PTx.
Imaging Protocols and Interpretation Scintigraphy Planar and pinhole images were acquired on a Philips Skylight γ-camera (Philips Healthcare, the Netherlands). SPECT/lowdose CT and 4D-CT were performed on a Philips Precedence SPECT/16-slice CT scanner. Patients were injected with 12 MBq (0.32 mCi) 123I IV and, after 2 hours, 600 MBq (16.2 mCi) 99mTcMIBI. Dual-isotope planar images (10 minutes, LEHR collimator) and pinhole images (15 minutes; aperture size, 3 mm) were obtained (matrix, 256 256; 99mTc-window, 140 keV −7/7; 123I asymmetrical window, 159 keV −4/+10). Delayed planar and pinhole images were obtained 2 hours later. SPECT/low-dose CT/4D-CT were initiated after the early planar images. First, a 40-cm FOV low-dose (120 kV, 50 mA)
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
93
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
FIGURE 1. Four-dimensional CT. Nonenhanced low-dose CT (A), late arterial phase diagnostic CT (B), and hepatic phase diagnostic CT (C). Arrow indicates an HPG posterior to the right inferior thyroid lobe. Note the rapid contrast wash-in and wash-out.
non–contrast-enhanced CT of the neck and thorax was performed. Then, 100 mL nonionic iodinated IV contrast was infused (Omnipaque 300 mg I/mL) at 3.5 mL/s. A late arterial phase diagnostic CT (from the lower edge of the mandible to the top of the aortic arch) was performed 22 seconds after bolus tracking in the left carotid artery at the level of the thyroid (120 kV; 150 mA; pitch, 0.813; rotation, 0.75 second; slice thickness, 2 mm; increment, 1 mm; collimation, 16 1.5; FOV, 250 mm). Thirty seconds later, a hepatic phase CT scan was done. SPECT (360-degree orbit; 128 angles; 14 s/angle; matrix, 128 128) was reconstructed using Ordered Subsets Expectation Maximization, 16 subsets and 3 iterations. Dual-phase planar images were evaluated side-by-side. Subtraction images were analyzed using a custom-made program (Matlab; Mathworks, Massachusetts) where variable subtraction and prefiltering and postfiltering were available. 123I images were subtracted from 99mTc-MIBI images with increasing multiplication until counts in the thyroid bed were equal to 99mTc background counts on the neck. SPECT/low-dose CTwere used as an adjunct to planar imaging for accurate localization of the findings. In 4D-CT images, HPGs were identified as well-defined structures with a characteristic contrast-uptake pattern.6 Briefly, the HPGs appeared hypodense compared with thyroid tissue in precontrast images, with rapid
wash-in and wash-out (Fig. 1). Images were examined on a Vitrea workstation (Toshiba Medical Systems Corporation, Tokyo, Japan). Gray scale and color Doppler ultrasound was performed on a Logiq 5 US scanner (GE Medical Systems, Milwaukee, WI) with a 12-MHz linear probe. A suspected HPG presented as an ovoid hypoechoic lesion at the posterior edge of the thyroid gland, occasionally with an echogenic curvilinear capsule seen anteriorly, and an arch of prominent vessels surrounding and leading into the nodule. The readers of PS, 4D-CT, and US were all blinded to the other modalities. The location of a putative HPG was given in relation to the thyroid. Certainty of the findings was graded on a 3-point scale. At follow-up, the reports from each of the imaging modalities were scored in a consensus meeting between all readers and the surgeon against the reference standard (surgical findings and histology). To assess interobserver agreement, a second nuclear medicine specialist and a radiologist evaluated planar subtraction PS and 4DCT, respectively.
Statistical Analysis Although approximately 13% of the population have supernumerary glands,11 a “per-parathyroid gland” analysis was done under the assumption that each patient had 4 parathyroid glands
FIGURE 2. Patient enrollment, flowchart. Ninety-one patients participated in the final analysis. 94
www.nuclearmed.com
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
85% (80%–89%) 86% (81%–90%)
97% (95%–99%) 89% (85%–92%)
NPV (95% CI)
Significance
of which at least one was diseased. This was done to calculate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the modalities and tested with the nonparametric Friedman test. The terms specificity and negative predictive value on a per-patient basis are not meaningful because all patients were assumed to have at least 1 HPG (being the very definition of PHP). Per-patient analyses were assessed with the nonparametric McNemar test. Binary logistic regression analysis was done in patients with a single HPG to assess correlations between true-positive findings and the weight of the HPGs, P-PTH, and PCa++ levels. Statistical significance was defined at a P value of 0.05 or less. Interrater agreement was calculated using Cohen κ.
P < 0.001 vs all others P = 0.206 vs 4D-CT P = 0.182 vs US P = 0.602 vs US —
Imaging in Primary Hyperparathyroidism
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
86% (81%–90%) 95% (91%–97%) 60% (50%–70%) 80% (68%–89%) 58% (47%–68%) 57% (46%–67%) 41 39 37 13 230 237 56 51
99% (97%–100%) 99.6% (98%–100%) 98% (92%–100%) 98% (92%–100%) 93% (86%–97%) 65% (55%–74%) 7 34 2 1 265 266 90 63
PPV (95% CI) Sensitivity (95% CI) False Neg False Pos True Neg
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pos, positive; Neg, negative.
24/67 (26%/74%) 66 (28–85) 665 (145–6000) 1.46 (1.30–1.85) 125 (42–463) 32/91 (35)
4D-CT US (n = 340)
Sex, male/female Age, median (range), y HPG weight, median (range), mg P-Ca++, median (range), mmol/L P-PTH, median (range), ng/L Nodular thyroid, n/n (%)
Subtraction PS Dual-phase PS
TABLE 1. Baseline Patient Characteristics
True Pos
The results of subtraction PS, dual-phase PS, 4D-CT, and US are listed on a per-parathyroid gland basis in Table 2 and Table 3. Table 4 shows the per-patient analysis in cases with a single HPG. Subtraction PS was significantly more sensitive than all other modalities studied both on a per-patient (95%, P < 0.001) and on a per-parathyroid gland analysis (93%, P < 0.001; Figs. 3, 4; Tables 2–4). There were no statistically significant differences in sensitivity between dual-phase PS, 4D-CT, and US (65%, 58%, and 57%, respectively, on a per-parathyroid gland analysis). Fourdimensional CT specificity (86%) was significantly lower than all the other modalities (P = 0.001). The specificity of subtraction PS, dual-phase PS, and US was 99%, 99.6%, and 95%, respectively. The number of positive scans with the rating “certain,” “probable,” and “uncertain” for subtraction PS was (76%, 18%, 6%),
TABLE 2. All Patients, Per-Parathyroid Gland (91 Patients, 364 Parathyroids)
General Analyses
Specificity (95% CI)
RESULTS Patient recruitment is summarized in Figure 2. Of 108 enrolled patients, 12 patients never had surgery due to patient reluctance and/or borderline indication for PTx and were excluded. Five patients were excluded as no HPG was found during surgery despite image-guided extensive bilateral neck exploration. In one of these, right-sided hemithyroidectomy resulted in normalization of PTH and Ca++, but histology never confirmed an HPG. In this patient, 4D-CT indicated an HPG in the right upper quadrant while the other modalities were negative. The remaining 4 patients still had PHP postoperatively. In these patients, neither dual-phase PS nor subtraction PS indicated any HPGs, whereas 4D-CT and US indicated an HPG in 4 and 2 cases, respectively. Ninety-one patients were thus available for final analysis, none of whom had had previous PTx. All 91 patients had normal blood calcium and PTH at follow-up (minimum, 3 months). Table 1 lists baseline patient characteristics, median weight of the removed HPG, and thyroid status. Eighty-five patients (93%) had a single HPG (80 adenomas, 4 hyperplastic carcinomas, and 1 parathyroid carcinoma). One of these had an ectopic HPG in the anterior mediastinum. Six patients (7%) had multiglandular disease (2 patients with 2 adenomas and 4 patients with 2 hyperplastic glands). The total number of HPGs was 97.
www.nuclearmed.com
95
96
www.nuclearmed.com
4
7 5
Dual-phase PS
4D-CT US (n = 5)
11 10
12
11
True Neg
1 0
0
1
False Pos
5 5
8
3
False Neg
58% (29%–84%) 50% (20%–80%)
33% (11%–65%)
75% (43%–93%)
Sensitivity (95% CI)
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
49 (58%; 47%–67%) 46 (58%; 46%–68%)
4D-CT US (n = 80)
58% (46%–68%) 78% (54%–87%)
99% (92%–100%) 98% (90%–100%)
PPV % (95% CI)
*Because 4D-CT always reported a possible location of an HPG, even when uncertain, this number is 0. Pos, positive; Neg, negative.
81 (95%; 88%–98%) 59 (69%; 59%–78%)
Subtraction PS Dual-phase PS
True Pos (Sensitivity %; 95% CI)
TABLE 4. Patients With a Single HPG, Per-Patient (85 Patients)
Pos, positive; Neg, negative.
9
Subtraction PS
True Pos
0* 21 (26%; 18%–37%)
3 (4%; 1%–10%) 25 (29%; 21%–40%)
69% (42%–88%) 67% (39%–87%)
60% (36%–80%)
79% (49%–94%)
NPV (95% CI)
36 (42%; 32%–53%) 13 (16%; 10%–26%)
1 (1%; 0%–6%) 1 (1%; 0%–6%)
False Positive (%; 95% CI)
92% (60%–100%) 100% (66%–100%)
100% (70%–100%)
92% (60%–100%)
Specificity (95% CI)
False Neg (%; 95% CI)
88% (47%–99%) 100% (46%–100%)
100% (40%–100%)
90% (54%–100%)
PPV (95% CI)
TABLE 3. Patients With >1 HPG, Per-Parathyroid Gland (6 Patients, 24 Parathyroids)
P < 0.001 vs all others P = 0.144 vs 4D-CT P = 0.123 vs US P = 0.86 vs US —
Significance
P = 0.046 vs dual phase P = 0.317 vs 4D-CT P = 0.18 vs US P = 0.157 vs 4D-CT P = 0.157 vs US P = 0.317 —
Significance
Krakauer et al Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
Imaging in Primary Hyperparathyroidism
Interobserver agreement (κ) for planar subtraction PS and 4D-CT was 0.96 (almost perfect) and 0.34 (fair), respectively.12
Subgroup Analyses
FIGURE 3. Sensitivity, all patients (per-parathyroid gland analysis). Error bars indicate 95% confidence interval.
followed by 4D-CT (71%, 20%, 9%), dual-phase PS (63%, 27%, 10%), and US (56%, 36%, 8%), respectively. Thirty-five percent of the patients had a nodular thyroid gland based on the findings on the 123I thyroid scintigram and/or US. There were no statistically significant differences in sensitivity of any of the imaging modalities with respect to thyroid morphology (data not shown). Dual-phase PS sensitivity correlated positively with the size of the HPG, whereas this was not the case for any of the other modalities. None of the modalities correlated with blood PTH or Ca++ levels (data not shown).
In 81 cases, both pinhole and parallel-hole collimated images were available for subtraction PS. Pinhole subtraction images correctly identified the HPGs in 77 (95%) of these, whereas this was the case in only 62 (77%) for parallel-hole subtraction images (P < 0.001; Fig. 5). In 2 cases, subtraction of the delayed pinhole images as opposed to the early images correctly identified the HPG. In 14 cases, delayed subtraction images provided some additional certainty in the diagnosis. Nine (64%) of these had a nodular thyroid. Among patients with a single HPG (n = 85), there were 4 false-negative subtraction PS studies. In 2 (50%) of these, 4D-CT was able to correctly identify the HPG. Ultrasonography did not locate any of the single HPGs missed by subtraction PS. In patients with 2 HPGs (n = 6), subtraction PS correctly identified 9 of 12 HPGs (75%). Both 4D-CT and US were able to locate 1 of the 3 HPGs (33%) that were missed by subtraction PS.
DISCUSSION This study compared the diagnostic performance of subtraction PS, dual-phase PS, 4D-CT, and US in the preoperative workup in patients with PHP. The strength of the study is the prospective, consecutive design. We found significantly better stand-alone performance of subtraction PS regarding sensitivity, PPV, specificity, and NPV compared with all the other modalities investigated. In addition, subtraction PS showed excellent interrater reproducibility.
FIGURE 4. Dual-phase PS versus subtraction PS. A and B, Dual-phase PS. Early (A) and delayed (B) planar scintigrams. C–E, Subtraction PS. 123I (C), 99mTc-MIBI (D), and subtraction (E) pinhole scintigrams. Dual-phase PS is negative, whereas subtraction PS shows an inferior right-sided HPG. © 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
97
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
FIGURE 5. Parallel-hole versus pinhole subtraction PS. A–C, Parallel-hole subtraction PS showing an inferior right-sided HPG. D–F, Pinhole subtraction PS showing the same lesion and an additional inferior left-sided HPG.
Subtraction PS is arguably slightly more complex to perform, mostly concerning the postprocessing. The duration of the image acquisition is, however, comparable to that of dual-phase PS as delayed imaging is rarely needed. In Europe, 123I is readily availably on a daily basis and costs approximately US $150 plus delivery for 3 patient doses. A wide range of diagnostic accuracies of PS has previously been reported. Accordingly, a systematic review found an overall sensitivity for detecting a single HPG by PS of 88% and 79% for US.13 A recent meta-analysis found a pooled sensitivity for planar dual-phase PS, SPECT, and SPECT/CT of 70%, 74%, and 86%, respectively,14 with significant heterogeneity between studies. It is likely that the efficacy of an imaging modality is dependent on imaging protocols, appropriate hardware and software, and on local expertise. The finding that subtraction PS was superior to dualphase PS and US is, however, not surprising as this is in accordance with previous studies favoring subtraction PS.15–20 However, to achieve high sensitivity, it is important to avoid oversubtraction and undersubtraction, aiming to subtract just as many 123I counts as necessary to achieve background 99mTc-MIBI activity in the thyroid bed. Pinhole imaging is highly recommended as it increases sensitivity markedly as also previously reported.20–24 Planar parallel-hole images or SPECT should, however, not be omitted because ectopic HPGs may otherwise be missed. When early imaging in subtraction PS is negative, the addition of delayed subtraction images increases sensitivity slightly, especially in the setting of a nodular thyroid. SPECT/low-dose CT after planar PS often offered an accurate anatomical localization of the HPG(s). The study was, however, not designed to evaluate the sensitivity of SPECT/low-dose CT as a stand-alone modality because SPECT/low-dose CT was always used as an adjunct to planar imaging. The advantage of the higher spatial resolution of pinhole collimation is not exploited in 98
www.nuclearmed.com
SPECT/CT because it is performed with parallel-hole collimators. Although there are previous studies using pinhole SPECT, there are, to our knowledge, no published reports on pinhole SPECT/ CT.25,26 Dual isotope subtraction SPECT/CT holds promise but needs further evaluation.27,28 The relatively low accuracy of 4D-CT compared with PS is in contrast to many previous findings.5,29–32 The reason for this is not clear but may, in part, be influenced by differences in study designs. In many studies, 4D-CT is used as an adjunct to PS, the latter often performed with the less sensitive dual-phase protocol. It is also possible that some selection bias exists in retrospective studies without consecutive patient recruitment. In addition, 4D-CT has been shown to have a higher sensitivity than PS in reoperative patients, an indication that was not evaluated in our study.33–35 Ultrasonography has been increasingly used to localize HPG not least due to accessibility, low cost, and its nonionizing properties. Reported accuracy varies considerably, and in a few studies, US was even found to be superior to dual-phase PS.36 The relatively low accuracy of US in our study is probably due to several factors such as patient material (relatively small HPGs) and the lack of state-of-the-art equipment, but results are still comparable with those reported by others36 and seen noninferior to 4D-CT. A nodular thyroid gland did not reduce the performance of any of the studied modalities. For subtraction PS, this is in concordance with some previous studies20,37 while in contrast to others.38,39 To obtain low numbers of false positives, careful examination of the 123 I thyroid image is crucial to identify hypofunctioning thyroid lesions that mimic HPGs. There was a significant positive correlation between a truepositive dual-phase PS and the size of the HPGs but not with PTH or Ca++ levels. This is in accordance with previous findings and consistent with the notion that it is the amount of oxyphilic cells that determines the uptake of 99mTc-MIBI and not the © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
hormone-producing chief cells.40 The same correlation with HPG size was not found for subtraction PS. This could be due to the low number of false negatives in subtraction PS. Surprisingly, a true-positive 4D-CT or US did not correlate with HPG size or any of the biochemical markers. When reporting a positive finding, the confidence of the reader was higher in subtraction PS and 4D-CT than in US and dual-phase PS. However, even when only considering a positive US or 4D-CT finding that was rated certain, these modalities still showed a substantial number of false-positive findings. By contrast, subtraction PS and dual-phase PS showed only very few false-positive findings, making these techniques highly specific. In addition, the interrater reproducibility was exceptionally high for subtraction PS (κ = 0.96), illustrating the robustness of the technique. In the few cases with a negative subtraction PS, both 4D-CT and, to a lesser extent, US were in some cases able to identify the HPG.
CONCLUSIONS An increasing body of evidence including the present study supports the use of pinhole subtraction PS for preoperative imaging in PHP. SPECT/CT can be added for accurate anatomical localization, further aiding in the planning of surgery. Four-dimensional CT and US could serve as an add-on in case of a negative PS. However, the success of all imaging modalities is dependent on local expertise, imaging equipment, and software. This must be taken into account when devising local diagnostic algorithms. In future studies, emerging modalities such as PET/CT and perfusion CT should be evaluated in a prospective head-to-head comparison with pinhole subtraction PS. ACKNOWLEDGMENTS The authors thank Jane Helene Andersen, Dorte Klarskov Bakke, and Inge Mortensen for their skilled technical assistance. REFERENCES 1. Trolle W, Møller H, Bennedbaek FN, et al. Minimally invasive surgery for hyperparathyroidism. Ugeskr Laeger. 2010;172:33–38. 2. Kunstman JW, Udelsman R. Superiority of minimally invasive parathyroidectomy. Adv Surg. 2012;46:171–189. 3. Hindié E, Ugur O, Fuster D, et al. 2009 EANM parathyroid guidelines. Eur J Nucl Med Mol Imaging. 2009;36:1201–1216. 4. American Institute of Ultrasound in Medicine, American College of Radiology, Society for Pediatric Radiology, et al. AIUM practice guideline for the performance of a thyroid and parathyroid ultrasound examination. J Ultrasound Med. 2013;32:1319–1329. 5. Rodgers SE, Hunter GJ, Hamberg LM, et al. Improved preoperative planning for directed parathyroidectomy with 4-dimensional computed tomography. Surgery. 2006;140:932–940; discussion 940–941. 6. Hoang JK, Sung W, Bahl M, et al. How to perform parathyroid 4D CT: tips and traps for technique and interpretation. Radiology. 2014;270:15–24. 7. Weber T, Maier-Funk C, Ohlhauser D, et al. Accurate preoperative localization of parathyroid adenomas with C-11 methionine PET/CT. Ann Surg. 2013;257:1124–1128. 8. Orevi M, Freedman N, Mishani E, et al. Localization of parathyroid adenoma by 11C-choline PET/CT: preliminary results. Clin Nucl Med. 2014;39: 1033–1038. 9. Chandramohan A, Sathyakumar K, Irodi A, et al. Causes of discordant or negative ultrasound of parathyroid glands in treatment naïve patients with primary hyperparathyroidism. Eur J Radiol. 2012;81:3956–3964. 10. Noureldine SI, Aygun N, Walden MJ, et al. Multiphase computed tomography for localization of parathyroid disease in patients with primary hyperparathyroidism: how many phases do we really need? Surgery. 2014;156: 1300–1307.
Imaging in Primary Hyperparathyroidism
11. Akerström G, Malmaeus J, Bergström R. Surgical anatomy of human parathyroid glands. Surgery. 1984;95:14–21. 12. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. 13. Ruda JM, Hollenbeak CS, Stack BC Jr. A systematic review of the diagnosis and treatment of primary hyperparathyroidism from 1995 to 2003. Otolaryngol Head Neck Surg. 2005;132:359–372. 14. Wong KK, Fig LM, Gross MD, et al. Parathyroid adenoma localization with 99m Tc-sestamibi SPECT/CT: a meta-analysis. Nucl Med Commun. 2015;36: 363–375. 15. Tunninen V, Varjo P, Schildt J, et al. Comparison of five parathyroid scintigraphic protocols. Int J Mol Imaging. 2013;2013:921260. [Internet]. 2013. Available at: http://www.hindawi.com/journals/ijmi/2013/921260/abs/. Accessed October 25, 2013. 16. Taieb D, Hindie E, Grassetto G, et al. Parathyroid scintigraphy: when, how, and why? A concise systematic review. Clin Nucl Med. 2012;37:568–574. 17. O'Doherty MJ, Kettle AG. Parathyroid imaging: preoperative localization. Nucl Med Commun. 2003;24:125–131. 18. Hindié E, Mellière D, Jeanguillaume C, et al. Unilateral surgery for primary hyperparathyroidism on the basis of technetium Tc 99m sestamibi and iodine 123 subtraction scanning. Arch Surg. 2000;135:1461–1468. 19. Caveny SA, Klingensmith WC 3rd, Martin WE, et al. Parathyroid imaging: the importance of dual-radiopharmaceutical simultaneous acquisition with 99m Tc-sestamibi and 123I. J Nucl Med Technol. 2012;40:104–110. 20. Heiba SI, Jiang M, Rivera J, et al. Direct comparison of neck pinhole dualtracer and dual-phase MIBI accuracies with and without SPECT/CT for parathyroid adenoma detection and localization. Clin Nucl Med. 2015;40: 476–482. 21. Klingensmith WC 3rd, Koo PJ, Summerlin A, et al. Parathyroid imaging: the importance of pinhole collimation with both single- and dual-tracer acquisition. J Nucl Med Technol. 2013;41:99–104. 22. Arveschoug AK, Bertelsen H, Vammen B. Presurgical localization of abnormal parathyroid glands using a single injection of Tc-99m sestamibi: comparison of high-resolution parallel-hole and pinhole collimators, and interobserver and intraobserver variation. Clin Nucl Med. 2002;27:249–254. 23. Tomas MB, Pugliese PV, Tronco GG, et al. Pinhole versus parallel-hole collimators for parathyroid imaging: an intraindividual comparison. J Nucl Med Technol. 2008;36:189–194. 24. Ho Shon IA, Yan W, Roach PJ, et al. Comparison of pinhole and SPECT 99m Tc-MIBI imaging in primary hyperparathyroidism. Nucl Med Commun. 2008;29:949–955. 25. Carlier T, Oudoux A, Mirallié E, et al. 99mTc-MIBI pinhole SPECT in primary hyperparathyroidism: comparison with conventional SPECT, planar scintigraphy and ultrasonography. Eur J Nucl Med Mol Imaging. 2008;35: 637–643. 26. Spanu A, Falchi A, Manca A, et al. The usefulness of neck pinhole SPECT as a complementary tool to planar scintigraphy in primary and secondary hyperparathyroidism. J Nucl Med. 2004;45:40–48. 27. Dontu VS, Kettle AG, O'Doherty MJ, et al. Optimization of parathyroid imaging by simultaneous dual energy planar and single photon emission tomography. Nucl Med Commun. 2004;25:1089–1093. 28. Hassler S, Ben-Sellem D, Hubele F, et al. Dual-isotope 99mTc-MIBI/123I parathyroid scintigraphy in primary hyperparathyroidism: comparison of subtraction SPECT/CT and pinhole planar scan. Clin Nucl Med. 2014;39: 32–36. 29. Cheung K, Wang TS, Farrokhyar F, et al. A meta-analysis of preoperative localization techniques for patients with primary hyperparathyroidism. Ann Surg Oncol. 2012;19:577–583. 30. Chazen JL, Gupta A, Dunning A, et al. Diagnostic accuracy of 4D-CT for parathyroid adenomas and hyperplasia. AJNR Am J Neuroradiol. 2012;33: 429–433. 31. Starker LF, Mahajan A, Björklund P, et al. 4D parathyroid CT as the initial localization study for patients with de novo primary hyperparathyroidism. Ann Surg Oncol. 2011;18:1723–1728. 32. Suh YJ, Choi JY, Kim SJ, et al. Comparison of 4D CT, ultrasonography, and 99mTc Sestamibi SPECT/CT in Localizing Single-Gland Primary Hyperparathyroidism. Otolaryngol Head Neck Surg. 2014;152: 438–443. 33. Mortenson MM, Evans DB, Lee JE, et al. Parathyroid exploration in the reoperative neck: improved preoperative localization with 4D-computed tomography. J Am Coll Surg. 2008;206:888–895; discussion 895–896.
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
99
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
34. Lubitz CC, Hunter GJ, Hamberg LM, et al. Accuracy of 4-dimensional computed tomography in poorly localized patients with primary hyperparathyroidism. Surgery. 2010;148:1129–1137; discussion 1137–1138. 35. Brown SJ, Lee JC, Christie J, et al. Four-dimensional computed tomography for parathyroid localization: a new imaging modality. ANZ J Surg. 2015;85:483–487. 36. Adkisson CD, Koonce SL, Heckman MG, et al. Predictors of accuracy in preoperative parathyroid adenoma localization using ultrasound and Tc99m-sestamibi: a 4-quadrant analysis. Am J Otolaryngol. 2013;34:508–516. 37. Rink T, Schroth HJ, Holle LH, et al. Limited sensitivity of parathyroid imaging with (99m)Tc-sestamibi/(123)I subtraction in an endemic goiter area. J Nucl Med. 2002;43:1175–1180.
100
www.nuclearmed.com
38. Erbil Y, Barbaros U, Yanik BT, et al. Impact of gland morphology and concomitant thyroid nodules on preoperative localization of parathyroid adenomas. Laryngoscope. 2006;116:580–585. 39. Barczynski M, Golkowski F, Konturek A, et al. Technetium-99m-sestamibi subtraction scintigraphy vs. ultrasonography combined with a rapid parathyroid hormone assay in parathyroid aspirates in preoperative localization of parathyroid adenomas and in directing surgical approach. Clin Endocrinol (Oxf). 2006;65:106–113. 40. Mehta NY, Ruda JM, Kapadia S, et al. Relationship of technetium Tc 99m sestamibi scans to histopathological features of hyperfunctioning parathyroid tissue. Arch Otolaryngol Head Neck Surg. 2005;131:493–498.
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
A Prospective Comparative Study of Parathyroid Dual-Phase Scintigraphy, Dual-Isotope Subtraction Scintigraphy, 4D-CT, and Ultrasonography in Primary Hyperparathyroidism Martin Krakauer, PhD,* Bente Wieslander, PhD,* Peter S. Myschetzky, MD,† Anke Lundstrøm, DMSc,† Theis Bacher, MSc,* Christian H. Sørensen, DMSc,‡ Waldemar Trolle, MD,‡ Birte Nygaard, DMSc,§ and Finn N. Bennedbæk, PhD§ Purpose: Preoperative localization of the diseased parathyroid gland(s) in primary hyperparathyroidism allows for minimally invasive surgery. This study was designed to establish the optimal first-line preoperative imaging modality. Patients and Methods: Ninety-one patients were studied consecutively in a prospective head-to-head comparison of dual isotope (99mTc-MIBI vs 123I) subtraction parathyroid scintigraphy (PS), dual-phase PS, 4-dimensional (4D) CT, and ultrasonography (US). Surgery, histological confirmation, and postoperative normalization of Ca++ and parathyroid hormone were the reference standard. Results: Ninety-seven hyperfunctioning parathyroid glands (HPGs) were identified by the reference standard. Sensitivity and specificity for subtraction PS, dual-phase PS, 4D-CT, and US were 93%, 65%, 58%, and 57% as well as 99%, 99.6%, 86%, and 95%, respectively. Interrater agreement was excellent for subtraction PS (κ = 0.96) while only fair for 4D-CT (κ = 0.34). Pinhole imaging and subtraction of delayed images (the latter especially in case of a nodular thyroid gland) increased the sensitivity of subtraction PS. SPECT/low-dose CT did not increase sensitivity but aided in the exact localization of the HPGs. Of 7 negative subtraction PS studies, 4D-CT and US were able to locate 3 and 1 additional HPGs, respectively. Conclusions: Dual isotope pinhole subtraction PS has higher diagnostic accuracy compared with dual-phase PS, 4D-CT, and US as a first-line imaging study in primary hyperparathyroidism. In case of a negative scintigraphy or suspicion of multiglandular disease, 4D-CT and/or US is recommended as a second-line modality. However, diagnostic algorithms should be adapted in accordance with local availability and expertise. Key Words: hyperparathyroidism, scintigraphy, subtraction, 4D-CT, ultrasonography (Clin Nucl Med 2016;41: 93–100)
P
rimary hyperparathyroidism (PHP) is characterized by elevated levels of blood calcium due to the hypersecretion of parathyroid hormone (PTH) from 1 or more hyperfunctioning parathyroid glands (HPGs). Definitive treatment is parathyroidectomy (PTx) preferably with a minimally invasive technique.1,2 Preoperative imaging allows for minimally invasive PTx. Imaging modalities include parathyroid scintigraphy (PS),3 ultrasonography (US),4 4-dimensional (4D) CT,5,6 and, recently, 11C-methionine PET/CT7 and 18F-choline PET/CT.8 Although promising, the PETbased modalities are still not widely available. Received for publication May 26, 2015; revision accepted July 26, 2015. From the Departments of *Nuclear Medicine and †Radiology, Gentofte University Hospital, Hellerup; ‡Department of Otolaryngology, Head and Neck Surgery, Copenhagen University Hospital, Rigshospitalet, Copenhagen; and §Department of Endocrinology, Herlev University Hospital, Herlev, Denmark. Conflicts of interest and sources of funding: none declared. Correspondence to: Martin Krakauer, MD, PHD, Department of Nuclear Medicine, Gentofte Hospital, Kildegaardsvej 28, DK-2900 Hellerup, Denmark. E-mail: [email protected]. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved. ISSN: 0363-9762/16/4102–0093 DOI: 10.1097/RLU.0000000000000988
Many studies have evaluated the available imaging modalities, but few have assessed them prospectively in a head-to-head comparison. Moreover, imaging protocols vary considerably between studies. Parathyroid scintigraphy can be performed with subtraction scintigraphy or, more commonly, as a dual-phase study. In addition, there are differences between studies of PS in terms of tracers, collimator types, energy windows, software, and the use of SPECT or SPECT/CT. Ultrasonography can be performed with or without color Doppler. Accuracy is dependent in the experience of the ultrasonographer and the presence of thyroid pathology.9 Four-dimensional CT exploits the differential washout of an iodinated contrast agent from HPGs using multiple acquisitions in various contrast phases. Four-dimensional CT protocols also differ between studies.10 We wanted to establish the optimal imaging modality in a prospective, blinded, head-to-head comparison of subtraction PS, dual-phase PS, 4D-CT, and US in consecutive patients with PHP eligible for PTx.
PATIENTS AND METHODS Patients Patients were recruited consecutively from the endocrinology department. We included adult patients with PHP who were candidates for PTx according to national guidelines. Exclusion criteria were pregnancy, allergy to iodine-containing contrast agents, and impaired renal function. If treated with thyroid hormone or cinacalcet, patients were instructed to pause this for 3 weeks and 5 days, respectively, before imaging (except for US). Informed written consent was obtained from all participants, and the protocol was approved by the local ethics committee. All patients underwent preoperative PS and 4D-CT on the same day. Six patients did not undergo US due to logistical reasons. The operating surgeons were experienced head and neck surgeons and had the results of all modalities available when deciding the surgical approach. A drop of more than 50% in per-operative plasma PTH levels and a minimum of 3 months' follow-up were used to verify successful PTx.
Imaging Protocols and Interpretation Scintigraphy Planar and pinhole images were acquired on a Philips Skylight γ-camera (Philips Healthcare, the Netherlands). SPECT/lowdose CT and 4D-CT were performed on a Philips Precedence SPECT/16-slice CT scanner. Patients were injected with 12 MBq (0.32 mCi) 123I IV and, after 2 hours, 600 MBq (16.2 mCi) 99mTcMIBI. Dual-isotope planar images (10 minutes, LEHR collimator) and pinhole images (15 minutes; aperture size, 3 mm) were obtained (matrix, 256 256; 99mTc-window, 140 keV −7/7; 123I asymmetrical window, 159 keV −4/+10). Delayed planar and pinhole images were obtained 2 hours later. SPECT/low-dose CT/4D-CT were initiated after the early planar images. First, a 40-cm FOV low-dose (120 kV, 50 mA)
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016 Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
93
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
FIGURE 1. Four-dimensional CT. Nonenhanced low-dose CT (A), late arterial phase diagnostic CT (B), and hepatic phase diagnostic CT (C). Arrow indicates an HPG posterior to the right inferior thyroid lobe. Note the rapid contrast wash-in and wash-out.
non–contrast-enhanced CT of the neck and thorax was performed. Then, 100 mL nonionic iodinated IV contrast was infused (Omnipaque 300 mg I/mL) at 3.5 mL/s. A late arterial phase diagnostic CT (from the lower edge of the mandible to the top of the aortic arch) was performed 22 seconds after bolus tracking in the left carotid artery at the level of the thyroid (120 kV; 150 mA; pitch, 0.813; rotation, 0.75 second; slice thickness, 2 mm; increment, 1 mm; collimation, 16 1.5; FOV, 250 mm). Thirty seconds later, a hepatic phase CT scan was done. SPECT (360-degree orbit; 128 angles; 14 s/angle; matrix, 128 128) was reconstructed using Ordered Subsets Expectation Maximization, 16 subsets and 3 iterations. Dual-phase planar images were evaluated side-by-side. Subtraction images were analyzed using a custom-made program (Matlab; Mathworks, Massachusetts) where variable subtraction and prefiltering and postfiltering were available. 123I images were subtracted from 99mTc-MIBI images with increasing multiplication until counts in the thyroid bed were equal to 99mTc background counts on the neck. SPECT/low-dose CTwere used as an adjunct to planar imaging for accurate localization of the findings. In 4D-CT images, HPGs were identified as well-defined structures with a characteristic contrast-uptake pattern.6 Briefly, the HPGs appeared hypodense compared with thyroid tissue in precontrast images, with rapid
wash-in and wash-out (Fig. 1). Images were examined on a Vitrea workstation (Toshiba Medical Systems Corporation, Tokyo, Japan). Gray scale and color Doppler ultrasound was performed on a Logiq 5 US scanner (GE Medical Systems, Milwaukee, WI) with a 12-MHz linear probe. A suspected HPG presented as an ovoid hypoechoic lesion at the posterior edge of the thyroid gland, occasionally with an echogenic curvilinear capsule seen anteriorly, and an arch of prominent vessels surrounding and leading into the nodule. The readers of PS, 4D-CT, and US were all blinded to the other modalities. The location of a putative HPG was given in relation to the thyroid. Certainty of the findings was graded on a 3-point scale. At follow-up, the reports from each of the imaging modalities were scored in a consensus meeting between all readers and the surgeon against the reference standard (surgical findings and histology). To assess interobserver agreement, a second nuclear medicine specialist and a radiologist evaluated planar subtraction PS and 4DCT, respectively.
Statistical Analysis Although approximately 13% of the population have supernumerary glands,11 a “per-parathyroid gland” analysis was done under the assumption that each patient had 4 parathyroid glands
FIGURE 2. Patient enrollment, flowchart. Ninety-one patients participated in the final analysis. 94
www.nuclearmed.com
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
85% (80%–89%) 86% (81%–90%)
97% (95%–99%) 89% (85%–92%)
NPV (95% CI)
Significance
of which at least one was diseased. This was done to calculate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of the modalities and tested with the nonparametric Friedman test. The terms specificity and negative predictive value on a per-patient basis are not meaningful because all patients were assumed to have at least 1 HPG (being the very definition of PHP). Per-patient analyses were assessed with the nonparametric McNemar test. Binary logistic regression analysis was done in patients with a single HPG to assess correlations between true-positive findings and the weight of the HPGs, P-PTH, and PCa++ levels. Statistical significance was defined at a P value of 0.05 or less. Interrater agreement was calculated using Cohen κ.
P < 0.001 vs all others P = 0.206 vs 4D-CT P = 0.182 vs US P = 0.602 vs US —
Imaging in Primary Hyperparathyroidism
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
86% (81%–90%) 95% (91%–97%) 60% (50%–70%) 80% (68%–89%) 58% (47%–68%) 57% (46%–67%) 41 39 37 13 230 237 56 51
99% (97%–100%) 99.6% (98%–100%) 98% (92%–100%) 98% (92%–100%) 93% (86%–97%) 65% (55%–74%) 7 34 2 1 265 266 90 63
PPV (95% CI) Sensitivity (95% CI) False Neg False Pos True Neg
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pos, positive; Neg, negative.
24/67 (26%/74%) 66 (28–85) 665 (145–6000) 1.46 (1.30–1.85) 125 (42–463) 32/91 (35)
4D-CT US (n = 340)
Sex, male/female Age, median (range), y HPG weight, median (range), mg P-Ca++, median (range), mmol/L P-PTH, median (range), ng/L Nodular thyroid, n/n (%)
Subtraction PS Dual-phase PS
TABLE 1. Baseline Patient Characteristics
True Pos
The results of subtraction PS, dual-phase PS, 4D-CT, and US are listed on a per-parathyroid gland basis in Table 2 and Table 3. Table 4 shows the per-patient analysis in cases with a single HPG. Subtraction PS was significantly more sensitive than all other modalities studied both on a per-patient (95%, P < 0.001) and on a per-parathyroid gland analysis (93%, P < 0.001; Figs. 3, 4; Tables 2–4). There were no statistically significant differences in sensitivity between dual-phase PS, 4D-CT, and US (65%, 58%, and 57%, respectively, on a per-parathyroid gland analysis). Fourdimensional CT specificity (86%) was significantly lower than all the other modalities (P = 0.001). The specificity of subtraction PS, dual-phase PS, and US was 99%, 99.6%, and 95%, respectively. The number of positive scans with the rating “certain,” “probable,” and “uncertain” for subtraction PS was (76%, 18%, 6%),
TABLE 2. All Patients, Per-Parathyroid Gland (91 Patients, 364 Parathyroids)
General Analyses
Specificity (95% CI)
RESULTS Patient recruitment is summarized in Figure 2. Of 108 enrolled patients, 12 patients never had surgery due to patient reluctance and/or borderline indication for PTx and were excluded. Five patients were excluded as no HPG was found during surgery despite image-guided extensive bilateral neck exploration. In one of these, right-sided hemithyroidectomy resulted in normalization of PTH and Ca++, but histology never confirmed an HPG. In this patient, 4D-CT indicated an HPG in the right upper quadrant while the other modalities were negative. The remaining 4 patients still had PHP postoperatively. In these patients, neither dual-phase PS nor subtraction PS indicated any HPGs, whereas 4D-CT and US indicated an HPG in 4 and 2 cases, respectively. Ninety-one patients were thus available for final analysis, none of whom had had previous PTx. All 91 patients had normal blood calcium and PTH at follow-up (minimum, 3 months). Table 1 lists baseline patient characteristics, median weight of the removed HPG, and thyroid status. Eighty-five patients (93%) had a single HPG (80 adenomas, 4 hyperplastic carcinomas, and 1 parathyroid carcinoma). One of these had an ectopic HPG in the anterior mediastinum. Six patients (7%) had multiglandular disease (2 patients with 2 adenomas and 4 patients with 2 hyperplastic glands). The total number of HPGs was 97.
www.nuclearmed.com
95
96
www.nuclearmed.com
4
7 5
Dual-phase PS
4D-CT US (n = 5)
11 10
12
11
True Neg
1 0
0
1
False Pos
5 5
8
3
False Neg
58% (29%–84%) 50% (20%–80%)
33% (11%–65%)
75% (43%–93%)
Sensitivity (95% CI)
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
49 (58%; 47%–67%) 46 (58%; 46%–68%)
4D-CT US (n = 80)
58% (46%–68%) 78% (54%–87%)
99% (92%–100%) 98% (90%–100%)
PPV % (95% CI)
*Because 4D-CT always reported a possible location of an HPG, even when uncertain, this number is 0. Pos, positive; Neg, negative.
81 (95%; 88%–98%) 59 (69%; 59%–78%)
Subtraction PS Dual-phase PS
True Pos (Sensitivity %; 95% CI)
TABLE 4. Patients With a Single HPG, Per-Patient (85 Patients)
Pos, positive; Neg, negative.
9
Subtraction PS
True Pos
0* 21 (26%; 18%–37%)
3 (4%; 1%–10%) 25 (29%; 21%–40%)
69% (42%–88%) 67% (39%–87%)
60% (36%–80%)
79% (49%–94%)
NPV (95% CI)
36 (42%; 32%–53%) 13 (16%; 10%–26%)
1 (1%; 0%–6%) 1 (1%; 0%–6%)
False Positive (%; 95% CI)
92% (60%–100%) 100% (66%–100%)
100% (70%–100%)
92% (60%–100%)
Specificity (95% CI)
False Neg (%; 95% CI)
88% (47%–99%) 100% (46%–100%)
100% (40%–100%)
90% (54%–100%)
PPV (95% CI)
TABLE 3. Patients With >1 HPG, Per-Parathyroid Gland (6 Patients, 24 Parathyroids)
P < 0.001 vs all others P = 0.144 vs 4D-CT P = 0.123 vs US P = 0.86 vs US —
Significance
P = 0.046 vs dual phase P = 0.317 vs 4D-CT P = 0.18 vs US P = 0.157 vs 4D-CT P = 0.157 vs US P = 0.317 —
Significance
Krakauer et al Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
Imaging in Primary Hyperparathyroidism
Interobserver agreement (κ) for planar subtraction PS and 4D-CT was 0.96 (almost perfect) and 0.34 (fair), respectively.12
Subgroup Analyses
FIGURE 3. Sensitivity, all patients (per-parathyroid gland analysis). Error bars indicate 95% confidence interval.
followed by 4D-CT (71%, 20%, 9%), dual-phase PS (63%, 27%, 10%), and US (56%, 36%, 8%), respectively. Thirty-five percent of the patients had a nodular thyroid gland based on the findings on the 123I thyroid scintigram and/or US. There were no statistically significant differences in sensitivity of any of the imaging modalities with respect to thyroid morphology (data not shown). Dual-phase PS sensitivity correlated positively with the size of the HPG, whereas this was not the case for any of the other modalities. None of the modalities correlated with blood PTH or Ca++ levels (data not shown).
In 81 cases, both pinhole and parallel-hole collimated images were available for subtraction PS. Pinhole subtraction images correctly identified the HPGs in 77 (95%) of these, whereas this was the case in only 62 (77%) for parallel-hole subtraction images (P < 0.001; Fig. 5). In 2 cases, subtraction of the delayed pinhole images as opposed to the early images correctly identified the HPG. In 14 cases, delayed subtraction images provided some additional certainty in the diagnosis. Nine (64%) of these had a nodular thyroid. Among patients with a single HPG (n = 85), there were 4 false-negative subtraction PS studies. In 2 (50%) of these, 4D-CT was able to correctly identify the HPG. Ultrasonography did not locate any of the single HPGs missed by subtraction PS. In patients with 2 HPGs (n = 6), subtraction PS correctly identified 9 of 12 HPGs (75%). Both 4D-CT and US were able to locate 1 of the 3 HPGs (33%) that were missed by subtraction PS.
DISCUSSION This study compared the diagnostic performance of subtraction PS, dual-phase PS, 4D-CT, and US in the preoperative workup in patients with PHP. The strength of the study is the prospective, consecutive design. We found significantly better stand-alone performance of subtraction PS regarding sensitivity, PPV, specificity, and NPV compared with all the other modalities investigated. In addition, subtraction PS showed excellent interrater reproducibility.
FIGURE 4. Dual-phase PS versus subtraction PS. A and B, Dual-phase PS. Early (A) and delayed (B) planar scintigrams. C–E, Subtraction PS. 123I (C), 99mTc-MIBI (D), and subtraction (E) pinhole scintigrams. Dual-phase PS is negative, whereas subtraction PS shows an inferior right-sided HPG. © 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
97
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
FIGURE 5. Parallel-hole versus pinhole subtraction PS. A–C, Parallel-hole subtraction PS showing an inferior right-sided HPG. D–F, Pinhole subtraction PS showing the same lesion and an additional inferior left-sided HPG.
Subtraction PS is arguably slightly more complex to perform, mostly concerning the postprocessing. The duration of the image acquisition is, however, comparable to that of dual-phase PS as delayed imaging is rarely needed. In Europe, 123I is readily availably on a daily basis and costs approximately US $150 plus delivery for 3 patient doses. A wide range of diagnostic accuracies of PS has previously been reported. Accordingly, a systematic review found an overall sensitivity for detecting a single HPG by PS of 88% and 79% for US.13 A recent meta-analysis found a pooled sensitivity for planar dual-phase PS, SPECT, and SPECT/CT of 70%, 74%, and 86%, respectively,14 with significant heterogeneity between studies. It is likely that the efficacy of an imaging modality is dependent on imaging protocols, appropriate hardware and software, and on local expertise. The finding that subtraction PS was superior to dualphase PS and US is, however, not surprising as this is in accordance with previous studies favoring subtraction PS.15–20 However, to achieve high sensitivity, it is important to avoid oversubtraction and undersubtraction, aiming to subtract just as many 123I counts as necessary to achieve background 99mTc-MIBI activity in the thyroid bed. Pinhole imaging is highly recommended as it increases sensitivity markedly as also previously reported.20–24 Planar parallel-hole images or SPECT should, however, not be omitted because ectopic HPGs may otherwise be missed. When early imaging in subtraction PS is negative, the addition of delayed subtraction images increases sensitivity slightly, especially in the setting of a nodular thyroid. SPECT/low-dose CT after planar PS often offered an accurate anatomical localization of the HPG(s). The study was, however, not designed to evaluate the sensitivity of SPECT/low-dose CT as a stand-alone modality because SPECT/low-dose CT was always used as an adjunct to planar imaging. The advantage of the higher spatial resolution of pinhole collimation is not exploited in 98
www.nuclearmed.com
SPECT/CT because it is performed with parallel-hole collimators. Although there are previous studies using pinhole SPECT, there are, to our knowledge, no published reports on pinhole SPECT/ CT.25,26 Dual isotope subtraction SPECT/CT holds promise but needs further evaluation.27,28 The relatively low accuracy of 4D-CT compared with PS is in contrast to many previous findings.5,29–32 The reason for this is not clear but may, in part, be influenced by differences in study designs. In many studies, 4D-CT is used as an adjunct to PS, the latter often performed with the less sensitive dual-phase protocol. It is also possible that some selection bias exists in retrospective studies without consecutive patient recruitment. In addition, 4D-CT has been shown to have a higher sensitivity than PS in reoperative patients, an indication that was not evaluated in our study.33–35 Ultrasonography has been increasingly used to localize HPG not least due to accessibility, low cost, and its nonionizing properties. Reported accuracy varies considerably, and in a few studies, US was even found to be superior to dual-phase PS.36 The relatively low accuracy of US in our study is probably due to several factors such as patient material (relatively small HPGs) and the lack of state-of-the-art equipment, but results are still comparable with those reported by others36 and seen noninferior to 4D-CT. A nodular thyroid gland did not reduce the performance of any of the studied modalities. For subtraction PS, this is in concordance with some previous studies20,37 while in contrast to others.38,39 To obtain low numbers of false positives, careful examination of the 123 I thyroid image is crucial to identify hypofunctioning thyroid lesions that mimic HPGs. There was a significant positive correlation between a truepositive dual-phase PS and the size of the HPGs but not with PTH or Ca++ levels. This is in accordance with previous findings and consistent with the notion that it is the amount of oxyphilic cells that determines the uptake of 99mTc-MIBI and not the © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
hormone-producing chief cells.40 The same correlation with HPG size was not found for subtraction PS. This could be due to the low number of false negatives in subtraction PS. Surprisingly, a true-positive 4D-CT or US did not correlate with HPG size or any of the biochemical markers. When reporting a positive finding, the confidence of the reader was higher in subtraction PS and 4D-CT than in US and dual-phase PS. However, even when only considering a positive US or 4D-CT finding that was rated certain, these modalities still showed a substantial number of false-positive findings. By contrast, subtraction PS and dual-phase PS showed only very few false-positive findings, making these techniques highly specific. In addition, the interrater reproducibility was exceptionally high for subtraction PS (κ = 0.96), illustrating the robustness of the technique. In the few cases with a negative subtraction PS, both 4D-CT and, to a lesser extent, US were in some cases able to identify the HPG.
CONCLUSIONS An increasing body of evidence including the present study supports the use of pinhole subtraction PS for preoperative imaging in PHP. SPECT/CT can be added for accurate anatomical localization, further aiding in the planning of surgery. Four-dimensional CT and US could serve as an add-on in case of a negative PS. However, the success of all imaging modalities is dependent on local expertise, imaging equipment, and software. This must be taken into account when devising local diagnostic algorithms. In future studies, emerging modalities such as PET/CT and perfusion CT should be evaluated in a prospective head-to-head comparison with pinhole subtraction PS. ACKNOWLEDGMENTS The authors thank Jane Helene Andersen, Dorte Klarskov Bakke, and Inge Mortensen for their skilled technical assistance. REFERENCES 1. Trolle W, Møller H, Bennedbaek FN, et al. Minimally invasive surgery for hyperparathyroidism. Ugeskr Laeger. 2010;172:33–38. 2. Kunstman JW, Udelsman R. Superiority of minimally invasive parathyroidectomy. Adv Surg. 2012;46:171–189. 3. Hindié E, Ugur O, Fuster D, et al. 2009 EANM parathyroid guidelines. Eur J Nucl Med Mol Imaging. 2009;36:1201–1216. 4. American Institute of Ultrasound in Medicine, American College of Radiology, Society for Pediatric Radiology, et al. AIUM practice guideline for the performance of a thyroid and parathyroid ultrasound examination. J Ultrasound Med. 2013;32:1319–1329. 5. Rodgers SE, Hunter GJ, Hamberg LM, et al. Improved preoperative planning for directed parathyroidectomy with 4-dimensional computed tomography. Surgery. 2006;140:932–940; discussion 940–941. 6. Hoang JK, Sung W, Bahl M, et al. How to perform parathyroid 4D CT: tips and traps for technique and interpretation. Radiology. 2014;270:15–24. 7. Weber T, Maier-Funk C, Ohlhauser D, et al. Accurate preoperative localization of parathyroid adenomas with C-11 methionine PET/CT. Ann Surg. 2013;257:1124–1128. 8. Orevi M, Freedman N, Mishani E, et al. Localization of parathyroid adenoma by 11C-choline PET/CT: preliminary results. Clin Nucl Med. 2014;39: 1033–1038. 9. Chandramohan A, Sathyakumar K, Irodi A, et al. Causes of discordant or negative ultrasound of parathyroid glands in treatment naïve patients with primary hyperparathyroidism. Eur J Radiol. 2012;81:3956–3964. 10. Noureldine SI, Aygun N, Walden MJ, et al. Multiphase computed tomography for localization of parathyroid disease in patients with primary hyperparathyroidism: how many phases do we really need? Surgery. 2014;156: 1300–1307.
Imaging in Primary Hyperparathyroidism
11. Akerström G, Malmaeus J, Bergström R. Surgical anatomy of human parathyroid glands. Surgery. 1984;95:14–21. 12. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. 13. Ruda JM, Hollenbeak CS, Stack BC Jr. A systematic review of the diagnosis and treatment of primary hyperparathyroidism from 1995 to 2003. Otolaryngol Head Neck Surg. 2005;132:359–372. 14. Wong KK, Fig LM, Gross MD, et al. Parathyroid adenoma localization with 99m Tc-sestamibi SPECT/CT: a meta-analysis. Nucl Med Commun. 2015;36: 363–375. 15. Tunninen V, Varjo P, Schildt J, et al. Comparison of five parathyroid scintigraphic protocols. Int J Mol Imaging. 2013;2013:921260. [Internet]. 2013. Available at: http://www.hindawi.com/journals/ijmi/2013/921260/abs/. Accessed October 25, 2013. 16. Taieb D, Hindie E, Grassetto G, et al. Parathyroid scintigraphy: when, how, and why? A concise systematic review. Clin Nucl Med. 2012;37:568–574. 17. O'Doherty MJ, Kettle AG. Parathyroid imaging: preoperative localization. Nucl Med Commun. 2003;24:125–131. 18. Hindié E, Mellière D, Jeanguillaume C, et al. Unilateral surgery for primary hyperparathyroidism on the basis of technetium Tc 99m sestamibi and iodine 123 subtraction scanning. Arch Surg. 2000;135:1461–1468. 19. Caveny SA, Klingensmith WC 3rd, Martin WE, et al. Parathyroid imaging: the importance of dual-radiopharmaceutical simultaneous acquisition with 99m Tc-sestamibi and 123I. J Nucl Med Technol. 2012;40:104–110. 20. Heiba SI, Jiang M, Rivera J, et al. Direct comparison of neck pinhole dualtracer and dual-phase MIBI accuracies with and without SPECT/CT for parathyroid adenoma detection and localization. Clin Nucl Med. 2015;40: 476–482. 21. Klingensmith WC 3rd, Koo PJ, Summerlin A, et al. Parathyroid imaging: the importance of pinhole collimation with both single- and dual-tracer acquisition. J Nucl Med Technol. 2013;41:99–104. 22. Arveschoug AK, Bertelsen H, Vammen B. Presurgical localization of abnormal parathyroid glands using a single injection of Tc-99m sestamibi: comparison of high-resolution parallel-hole and pinhole collimators, and interobserver and intraobserver variation. Clin Nucl Med. 2002;27:249–254. 23. Tomas MB, Pugliese PV, Tronco GG, et al. Pinhole versus parallel-hole collimators for parathyroid imaging: an intraindividual comparison. J Nucl Med Technol. 2008;36:189–194. 24. Ho Shon IA, Yan W, Roach PJ, et al. Comparison of pinhole and SPECT 99m Tc-MIBI imaging in primary hyperparathyroidism. Nucl Med Commun. 2008;29:949–955. 25. Carlier T, Oudoux A, Mirallié E, et al. 99mTc-MIBI pinhole SPECT in primary hyperparathyroidism: comparison with conventional SPECT, planar scintigraphy and ultrasonography. Eur J Nucl Med Mol Imaging. 2008;35: 637–643. 26. Spanu A, Falchi A, Manca A, et al. The usefulness of neck pinhole SPECT as a complementary tool to planar scintigraphy in primary and secondary hyperparathyroidism. J Nucl Med. 2004;45:40–48. 27. Dontu VS, Kettle AG, O'Doherty MJ, et al. Optimization of parathyroid imaging by simultaneous dual energy planar and single photon emission tomography. Nucl Med Commun. 2004;25:1089–1093. 28. Hassler S, Ben-Sellem D, Hubele F, et al. Dual-isotope 99mTc-MIBI/123I parathyroid scintigraphy in primary hyperparathyroidism: comparison of subtraction SPECT/CT and pinhole planar scan. Clin Nucl Med. 2014;39: 32–36. 29. Cheung K, Wang TS, Farrokhyar F, et al. A meta-analysis of preoperative localization techniques for patients with primary hyperparathyroidism. Ann Surg Oncol. 2012;19:577–583. 30. Chazen JL, Gupta A, Dunning A, et al. Diagnostic accuracy of 4D-CT for parathyroid adenomas and hyperplasia. AJNR Am J Neuroradiol. 2012;33: 429–433. 31. Starker LF, Mahajan A, Björklund P, et al. 4D parathyroid CT as the initial localization study for patients with de novo primary hyperparathyroidism. Ann Surg Oncol. 2011;18:1723–1728. 32. Suh YJ, Choi JY, Kim SJ, et al. Comparison of 4D CT, ultrasonography, and 99mTc Sestamibi SPECT/CT in Localizing Single-Gland Primary Hyperparathyroidism. Otolaryngol Head Neck Surg. 2014;152: 438–443. 33. Mortenson MM, Evans DB, Lee JE, et al. Parathyroid exploration in the reoperative neck: improved preoperative localization with 4D-computed tomography. J Am Coll Surg. 2008;206:888–895; discussion 895–896.
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
www.nuclearmed.com
99
Krakauer et al
Clinical Nuclear Medicine • Volume 41, Number 2, February 2016
34. Lubitz CC, Hunter GJ, Hamberg LM, et al. Accuracy of 4-dimensional computed tomography in poorly localized patients with primary hyperparathyroidism. Surgery. 2010;148:1129–1137; discussion 1137–1138. 35. Brown SJ, Lee JC, Christie J, et al. Four-dimensional computed tomography for parathyroid localization: a new imaging modality. ANZ J Surg. 2015;85:483–487. 36. Adkisson CD, Koonce SL, Heckman MG, et al. Predictors of accuracy in preoperative parathyroid adenoma localization using ultrasound and Tc99m-sestamibi: a 4-quadrant analysis. Am J Otolaryngol. 2013;34:508–516. 37. Rink T, Schroth HJ, Holle LH, et al. Limited sensitivity of parathyroid imaging with (99m)Tc-sestamibi/(123)I subtraction in an endemic goiter area. J Nucl Med. 2002;43:1175–1180.
100
www.nuclearmed.com
38. Erbil Y, Barbaros U, Yanik BT, et al. Impact of gland morphology and concomitant thyroid nodules on preoperative localization of parathyroid adenomas. Laryngoscope. 2006;116:580–585. 39. Barczynski M, Golkowski F, Konturek A, et al. Technetium-99m-sestamibi subtraction scintigraphy vs. ultrasonography combined with a rapid parathyroid hormone assay in parathyroid aspirates in preoperative localization of parathyroid adenomas and in directing surgical approach. Clin Endocrinol (Oxf). 2006;65:106–113. 40. Mehta NY, Ruda JM, Kapadia S, et al. Relationship of technetium Tc 99m sestamibi scans to histopathological features of hyperfunctioning parathyroid tissue. Arch Otolaryngol Head Neck Surg. 2005;131:493–498.
© 2015 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
