Impact of Localization Studies and Clinical Scenario in Patients With Hyperparathyroidism Being Evaluated for Reoperative Neck Surgery FREE

Preoperative localization studies (LSs) for parathyroid reexploration, unlike for initial parathyroid surgery, are crucial for operative success. Numerous studies^1- 10 in the literature have reported favorable outcomes of reoperative parathyroid surgery, as high as 97%, and have focused on the accuracy of LSs in directing surgeons where to explore. However, the success rates of curing patients with persistent or recurrent hyperparathyroidism (HPT) and the accuracy of LSs have been derived only from patients who do, in fact, undergo reoperative parathyroid surgery after LSs have successfully localized the site(s) of the abnormal parathyroid gland(s). On the other hand, those who do not meet the criteria for reexploration owing to nonlocalizing studies have been excluded from the analysis, skewing the actual success of reoperative parathyroid surgery and LSs in previous series.

Although there is a general consensus that LSs are a prerequisite for reexploration, conscientious surgeons also expend a great amount of time and effort reviewing the patient's clinical history, such as the initial diagnosis, type of surgical approach used in previous explorations, operative findings, and surgical pathology results, before performing reoperative parathyroid surgery.^10- 11 Clinical history can help provide information about the disease status, stratifying the patient as having a single diseased gland or multiple diseased glands remaining in the neck and/or mediastinum. Consequently, patients who are predicted to have single-site disease (SSD) by clinical history are candidates for single-site exploration (SSE), and those who are predicted to have multiple-site disease (MSD) are candidates for multiple-site exploration (MSE). Clinical scenario (CS) can assure surgeons that the LSs correspond to their clinical expectations and is a helpful strategy in decision making for the reoperative approach. Although this clinical stratification process may be obvious and already practiced by many surgeons, it is an important concept that is often not systematically scrutinized or reported in the literature.

This study was designed to analyze our clinical strategies in the management of patients with HPT requiring reoperative neck surgery. Information provided by LSs and clinical history was applied in the stratification of patients by disease status and reoperative approach. To our knowledge, this is the first study to consider a very critical view of the success rates of LSs and reoperative parathyroid surgery by including all patients with HPT being evaluated for reoperative neck surgery in an intent-to-treat analysis, even those who were not reexplored owing to nonlocalizing studies.

Between May 1, 2000, and November 30, 2009, 203 patients with HPT being evaluated for reoperative neck surgery at the Cleveland Clinic (Cleveland, Ohio) by 4 endocrine surgeons (M.M., J.M., E.B., and A.S.) were studied from a prospective database. These patients had symptoms and/or associated conditions of HPT that warranted reoperation. They also had a history of neck operations involving the same surgical territory as in parathyroid exploration (previous parathyroid, thyroid, and esophageal surgical procedures). Diagnoses of primary, secondary, persistent, and recurrent HPT were confirmed. Persistent HPT was defined as hypercalcemia (serum calcium level >10.5 mg/dL [to convert to millimoles per liter, multiply by 0.25]) occurring within 6 months of parathyroidectomy, and recurrent HPT was defined as hypercalcemia recurring after at least 6 months of postoperative normocalcemia. Clinical scenario (determined by the type of HPT and findings at previous exploration[s] obtained from referring notes, operative notes, and surgical pathology reports) stratified patients into the following disease categories: (1) strongly suspected of having a single remaining abnormal gland, (2) strongly suspected of having multiple remaining abnormal glands, and (3) unknown (inadequate clinical information to determine clinical status). The various clinical conditions used to stratify these patients are listed in Table 1.

All 203 patients underwent surgeon-performed ultrasonography and sestamibi-iodine subtraction scan (MIBI) with or without computed tomography (CT) (co-registration of single-photon emission CT with CT images performed with MIBI starting in July 2005). When the results of these studies were negative or inconclusive, parathyroid fine-needle aspiration biopsy under ultrasonographic guidance, magnetic resonance imaging, CT (before the use of single-photon emission CT/CT with MIBI), and/or selective venous sampling for intact parathyroid hormone (PTH) was performed. The success rates of each LS were evaluated on a per-patient basis, as previously described by Shen et al.⁴ A study was defined as completely correct if it identified all abnormal parathyroid glands in a patient with no false-positive or false-negative findings. The term correct with false-positive results was used if the LS identified all abnormal glands in a patient but also had false-positive results (imaged suspected tumors not identified at surgery). The term correct with false-negative results was used if the localization study identified all abnormal glands in a patient but also had false-negative results (additional nonimaged tumors found at exploration). Finally, a study was defined as completely incorrect if it produced the following results: (1) a completely negative study result in a patient with an abnormal gland(s), (2) a completely false-positive study result that localized all abnormal glands in incorrect positions, or (3) the localization of the abnormal gland(s) remaining unknown after reexploration. The sensitivity of LSs was defined as the proportion of studies that correctly located all abnormal parathyroid glands (completely correct and correct with false-positive results) in patients with abnormal glands.

In addition, the accuracy of these LSs in guiding the surgeon to perform an appropriate and curative exploration was assessed based on the aggregate interpretation of LSs for each patient. For this study, LSs were considered to be accurate if a focal signal led to a curative exploration and excision of only the single localized gland (completely correct) or if multifocal signals guided the surgeon to perform a curative exploration and excision of at least 2 localized glands (completely correct or correct with false-negative results). Surgical cure was determined by the assessment of serial serum calcium and PTH values at least 6 months postoperatively (range, 6 months to 9 years), in which the patient did not meet the criteria for persistent or recurrent primary HPT. After reexploration and from surgically confirmed pathology laboratory findings, the positive predictive value (PPV) of disease status (SSD or MSD) derived only from the CS was analyzed. Last, the PPV of disease status based on the interpretation of LSs in the context of the patient's CS was also evaluated.

Statistical analysis was performed using a commercially available statistical software package (JMP 8.0; SAS Institute, Inc). We determined statistical significance between the PPVs using a z score for proportions. The level of statistical significance for this study was defined as P < .05.

Of the 203 patients, 147 were women and 56 were men (female to male ratio of 2.6 to 1). Their mean age was 58 years (age range, 21-87 years). Indications for previous cervical operations are listed in Table 2. Previous surgical procedures included parathyroid surgery in 126 patients (62.1%), parathyroid and thyroid surgery in 24 (11.8%), and nonparathyroid surgery (thyroid or esophageal) in 53 (26.1%). Indications for reoperative parathyroid surgery included persistent HPT (primary, secondary, and tertiary HPT and multiple endocrine neoplasia type 1) in 110 patients (54.2%), recurrent HPT (primary, secondary, and tertiary HPT and multiple endocrine neoplasia types 1 and 2A) in 40 (19.7%), primary HPT in 52 (25.6%), and secondary HPT in 1 (0.5%). Of the 203 patients, 176 (86.7%) were candidates for reexploration due to concordance on 2 LSs (n = 165, 93.8%), localization by only selective venous sampling when all other study results were negative (n = 6, 3.4%), or localization by only 1 study when also undergoing thyroid surgery (n = 5, 2.8%). On the other hand, 27 patients (13.3%) did not meet the criteria for reexploration due to nonlocalizing studies. Of the 176 reoperative candidates, localization by only ultrasonogram and MIBI (with or without co-registration of CT images) was achieved in 63 patients (35.8%). Thirty-seven patients (21.0%) required fine-needle aspiration biopsy as the only additional preoperative study, and 46 (26.1%) needed other noninvasive imaging modalities (CT and/or magnetic resonance imaging) with or without fine-needle aspiration biopsy. Magnetic resonance imaging was used more frequently in the earlier years of this study and is now rarely used owing to its low sensitivity, and CT alone without MIBI was performed before July 2005. Thirty patients (17.0%) underwent selective venous sampling for PTH. A total of 517 LSs were performed in these reoperative candidates. The success rates of these LSs analyzed individually are demonstrated in Table 3.

For the 176 reoperative candidates, their CSs as previously described were reviewed, and they were stratified as having SSD (n = 145, 82.4%), MSD (n = 16, 9.1%), or unknown disease status (SSD or MSD) (n = 15, 8.5%). For each group, the aggregate interpretation of LSs (focal or multifocal signal), surgical approach (SSE or MSE), operative success (cure, fail, or negative exploration), and surgically confirmed outcome (SSD, MSD, or no disease) were evaluated. For details, refer to the Appendix and eFigures 1, 2, and 3.

Surgically confirmed outcomes of HPT in reoperative neck surgery in 176 patients who were candidates for reexploration were as follows: SSD removed in 144 cured patients (81.8%), MSD excised in 25 cured patients (14.2%), SSD removed in 4 failed patients (2.3%), and no disease found in 3 patients (1.7%). The SSE group achieved cure by removing 90% SSD and 6% MSD (with the last 4% representing 3 failures and 3 negative explorations), the MSE group achieved cure by excising 19% SSD and 81% MSD, and the unknown disease status group achieved cure by removing 67% SSD and 27% MSD (with the last 6% representing 1 failure). The LSs accurately guided reexploration (either SSE or MSE) in 84.7% of patients (149 of 176), consisting of 88% in the SSE group, 75% in the MSE group, and 67% in the unknown disease status group. However, when including the 27 patients who were not reexplored owing to nonlocalizing studies, the success of LSs to direct reexploration was decreased to 73%. The cure rate in reoperated patients was 96% (3 negative explorations and 4 persistent disease). When including the 27 nonexplored patients, the overall cure rate of the entire cohort was reduced to 83%.

Of the 173 reexplored patients in whom pathologic abnormalities were found (excluding 3 negative explorations), 83.2% (n = 144) were found to have SSD and 16.8% (n = 29) MSD (Table 4). The PPV of the disease status derived only from the CS was then evaluated. When patients were strongly suspected of having a single remaining abnormal gland and, therefore, were candidates for SSE based on CS alone, 92.3% (131 of 142) had surgically confirmed SSD and 7.7% (11 of 142) had MSD. Conversely, when patients were strongly suspected of having multiple remaining abnormal glands and, therefore, were candidates for MSE based on the CS alone, 81.3% (13 of 16) had surgically confirmed MSD and 18.8% (3 of 16) had SSD. Therefore, when patients were stratified by CS, 92% actually had SSD (increased from 83%, P = .008) and 81% actually had MSD (increased from 17%, P < .001). Although the stratification process could not be applied in the group with unknown or unclear history, 66.7% (10 of 15) had surgically confirmed SSD, and 33.3% (5 of 15) had MSD.

The PPV of the disease status from the interpretation of LSs in the context of the patient's CS was then analyzed (Table 5). From the entire cohort with a focal signal on LSs, 91.8% of patients (135 of 147) were found to have SSD based on completely correct LSs alone. However, when an accurate focal signal was seen in the SSE group, 95.3% of patients (123 of 129) had surgically confirmed SSD. When a focal signal was seen in the MSE group, 60.0% of patients (3 of 5) had SSD. Lastly, when a focal signal was seen in the group with unknown disease status, 69.2% of patients (9 of 13) actually had SSD. On the other hand, from the entire cohort with multifocal signals on LSs, 73.1% (19 of 26) were found to have MSD based on LSs alone (completely correct or correct with false-negative results). However, when accurate multifocal signals were seen in the SSE group, 53.8% of patients (7 of 13) had surgically confirmed SSD. When multifocal signals were seen in the MSE group, 100% of patients (11 of 11) had MSD. Lastly, when multifocal signals were seen in the group with unknown disease status, 50.0% of patients (1 of 2) actually had MSD. From this analysis, we can conclude that a focal signal on LSs has a PPV for SSD of 92%. However, if the patient can be stratified by CS to have SSD, the PPV of the focal signal on LSs is increased to 95% (P = .12). Conversely, multifocal signals on LSs have a PPV for MSD of 73%. However, if the patient can be stratified by CS to have MSD, the PPV of the LSs is increased to 100% (P = .03). Although the PPV of disease status improved for both groups, it was statistically significant only for the MSE group, supporting previous studies^12- 14 that have shown LSs to be superior at detecting single adenomas vs multigland disease.

Thirty-four of 176 patients (19.3%) had more extensive exploration than indicated by LSs due to concomitant thyroid surgery (29 in the SSE group and 5 in the unknown disease status group). However, for this study, patients were considered to have SSE if a completely correct focal signal on LSs guided the surgeon to explore only the index site, and, therefore, the concomitant thyroid surgery would not have changed the surgical approach. Only 3 of 34 patients were considered to have MSE due to completely or partially incorrect LSs that led the surgeon to explore other nonlocalized sites, even if no thyroidectomy was being performed.

Intraoperative PTH assays were used in 171 of 176 patients (97.2%). After excision of a single abnormal parathyroid gland, operative cure was achieved in 120 patients, where the Miami criterion¹⁵ (an intraoperative PTH decline ≥50% from the highest of either the preincision or preexcision level 10 minutes after gland excision) was met in 110 patients (64.3%) but was not satisfied in 10 (5.8%). In patients with multiple abnormal glands removed, operative cure was achieved in 28, where the Miami criterion was met in 26 patients (15.2%) but was not satisfied in 2 (1.2%). In 11 patients, the preincision or preexcision value was within the reference range and decreased at least 50% after gland excision in 8 patients (4.7%) but fell less than 50% after gland excision in 3 (1.8%). Of the remaining 12 patients, intraoperative PTH measurements were not interpretable in 6 (3.5%), 3 (1.8%) failed postoperatively (2 met the Miami criterion and 1 did not), and 3 (1.8%) had negative explorations and did not meet the Miami criterion. Overall, the Miami criterion correctly predicted postoperative cure in 144 of 159 interpretable intraoperative PTH values (90.6%).

From this cohort, 220 parathyroid specimens were excised in 173 patients (excluding 3 negative explorations). As shown in the Figure, 136 parathyroid glands (61.8%) were located in their normal anatomical positions (24 left lower, 43 left upper, 24 right lower, and 45 right upper). Of the remaining 84 ectopic parathyroid specimens (38.2%), 15 were previously autotransplanted parathyroid tissue in the sternocleidomastoid or strap muscles, and the most common ectopic location was intrathymic (Table 6).

Reoperative neck surgery for HPT in patients with previous cervical explorations can pose diagnostic and technical difficulties.^{1- 6,8- 9,16- 21} With the higher risk of morbidity, surgeons need to rely heavily on the patient's clinical history and LSs to perform curative reexploration. When sufficient data can be obtained from both clinical information and imaging studies, reoperative parathyroid surgery can have success rates that are comparable with those of initial parathyroidectomy.^4- 5,8,10 However, most of the recognition has been given to the success of LSs, whereas the wealth of data that can be gained from the clinical history has not been as well acknowledged. In addition, previous studies^1- 10 in the literature have not analyzed patients in whom the LSs are unable to localize the abnormal gland(s), and, therefore, only report the success of LSs in those who actually undergo reoperative parathyroid surgery. In this study, we took an intent-to-treat approach by evaluating all patients presenting for reoperative neck surgery. The ability of LSs to accurately direct reexploration was 85% in those who met the criteria for reexploration; however, it was decreased to 73% when including those who never underwent reexploration. Similar to other series, the present cure rate in reoperated patients was 96%; however, 27 nonexplored patients remain with disease, thereby reducing the overall cure rate to 83%. In conclusion, the failure to cure patients with HPT and previous neck surgery was 4 times more likely to be due to nonlocalizing studies than to a failed reexploration.

One of the aims of this study was to emphasize the importance of clinical history in this patient population. From the entire cohort of patients who were reexplored and in whom pathologic abnormalities were found, 83% had SSD and 17% had MSD. However, by applying CS to the stratification of disease status, the likelihood of actually having either a single or multiple remaining abnormal glands was improved (from 83% to 92% in the SSE group and from 17% to 81% in the MSE group). Furthermore, the PPV of the disease status based on LSs also increased when a patient was stratified as having either a single or multiple remaining abnormal glands (from 92% to 95% in the SSE group and from 73% to 100% in the MSE group). Therefore, the CS allows surgeons to categorize patients according to their disease status, aids in the interpretation of LSs, and helps correctly direct the reoperative approach.

The incidence of concomitant thyroid and parathyroid diseases is growing.^22- 25 Twenty-four percent of this patient population initially underwent thyroid surgery, and 12% had simultaneous parathyroid and thyroid operation as their previous cervical exploration. Furthermore, 19% of patients had concomitant thyroidectomy during reexploration for HPT, which meant requiring more extensive exploration than indicated by LSs. This represents realistic groups of patients who are being evaluated for reoperative neck surgery, and our approach to these patients is extremely relevant to the true clinical setting.

Although the best strategy is to prevent reoperative surgery, failures do occur after initial surgery for HPT. Owing to the potential difficulties of reoperative neck explorations, surgeons must seek data that can help reduce the complexity and morbidity of these operations. These results demonstrate the significance of LSs and clinical history used in conjunction to achieve operative cure. With the use of all available information, the success of reexploration for HPT has the potential to equal that of initial parathyroid surgery in patients with concordant LSs. However, there remain many patients with unsuccessful LSs who continue to live with the disease.