1. Introduction
Parathyroid carcinoma (PC) is an extremely rare endocrine malignancy, and the prevalence is only 0.005% of all malignancies [
1,
2]. The annual incidence of parathyroid carcinoma is ~3.5–5.7 cases per 10,000,000 population. PC is associated with <1% of all primary hyperparathyroidism cases. The peak incidence of PC occurs in the fifth decade of life, with no sex predilection [
3]. The etiology of PC is mainly unknown; however, there is an increased risk of parathyroid carcinoma in patients with the hyperparathyroidism-jaw tumor syndrome (HPT-JT) [
4]. The risk of PC may be increased in other endocrine diseases, such as familial isolated hyperparathyroidism and multiple endocrine neoplasia type 1 (MEN1) [
5,
6]. Several genetic mutations have been proposed in PC, including, retinoblastoma (RB), P53, cyclin D1/parathyroid adenomatosis gene 1 (PRAD1), and BRCA2; however, a strong body of evidence implicates the gene
CDC73 (formerly called
HRPT2) first identified in the context of HPT-JT [
7,
8,
9]. Clinical features that raise concern for parathyroid malignancy include serum calcium >14 mg/dL, PTH >5–10 times the upper limit of normal or PTH >500 pg/mL, palpable neck mass, concomitant skeletal, renal disease, and features of parathyroid crisis, e.g., altered mental status [
10].
The diagnosis of PC is difficult because it overlaps clinically, radiologically, and histologically with parathyroid adenoma/hyperplasia and relies on evidence of unequivocal local tissue invasion and/or metastasis [
10]. However, such evidence is often absent at the time of presentation [
11]. PC is usually indolent and slowly progressive, with low rates of lymph node (<5%) and systemic metastasis (<2%) [
12], and higher rates of local recurrence (25–80%). Mortality is usually from complications of hypercalcemia and not due to tumor burden [
10]. Previously reported 5-year and 10-year survival rates are 76–85% and 49–77%, respectively [
11]. PC is an uncommon condition with very little literature outlining its clinical aspects, diagnostic modalities, optimum treatment options, and prognostic markers [
13,
14,
15].
We present one of the largest and most up-to-date database studies aimed at investigating the demographic, clinical, and pathological factors affecting the prognosis and survival of patients with PC.
2. Materials and Methods
The Surveillance, Epidemiology and End Results (SEER) database initiated by the National Cancer Institute in 1972 covers approximately 28% of the U.S. population. The SEER*Stat software (Version 8.3.5) was used to collect data using the International Classification of Diseases version 3 (ICD-O-3) and anatomical code (C75.0). There were 18 registries in all: Alaska Native Tumor Registry, Arizona Indians Tumor Registry, Cherokee Nation Tumor Registry, Connecticut tumor registry, Detroit tumor registry, Georgia Center for Cancer registry, Greater Bay Area Cancer tumor Registry, Greater California registry, Hawaii Tumor Registry, Iowa Tumor Registry, Kentucky Tumor Registry, Louisiana Tumor Registry, New Jersey Tumor Registry, Seattle-Puget Sound Tumor Registry, and Utah Tumor Registry from SEER software (
https://seer.cancer.gov/seerstat/, accessed 5 March 2022). The data were exported to Statistical Product and Service Solutions (SPSS©) version 20.2 (IBM Corporation, Armonk, NY, USA).
Demographic and clinical data included age, race, sex, histologic variant, tumor differentiation, tumor size, tumor stage, lymph node status, surgical treatment, radiotherapy, chemotherapy, overall survival, survival with surgery, and survival with radiation therapy. Excluded from the final study cohort were patients with in situ malignancies, those with nonspecific tumor origins, and those without histological confirmation of cancer. This study used the “backward Wald” method to calculate odds ratios (OR) and identify the independent factors that affect survival. Data that were either unidentified or missing were removed from multivariate analysis. The chi-square test, paired t-test, and multivariate analysis were used to analyze the data. Statistical significance was defined as p < 0.05.
4. Discussion
This study presents one of the largest cohorts of patients with PC, with mean age of diagnosis of 62 ± 10 years, slightly higher incidence in men, 25% with nodal metastasis, and tumor size 4 cm associated with poor outcome.
PC is common in the fifth decade of life, with no sex predilection [
16,
17]. Lee et al. [
1] reported 3–19% nodal metastases and 3–4% distant metastasis at initial operation, and previous studies showed rates of less than 5% and less than 2%, respectively [
10]. Reports about the impact of nodal involvement on prognosis have been contradictory: Asare et al. and other studies performed on large databases reported that survival and recurrence rates are not affected by positive nodal status [
1,
2,
14,
18]. However, single institutional studies, such as those from Harari et al. [
19] and Sandelin et al. [
20], reported that nodal involvement is associated with decreased survival. Lower rates of lymph node examination were reported compared to our study results [
19,
21]. In some multicenter studies, tumor size did not affect survival [
20] and, in Asare et al., in a National Cancer Data Base (NCDB) study, tumors > 4 cm were associated with decreased survival [
1,
2,
14,
22].
Missing information from large databases and referral bias from single institutional studies may explain somewhat this discrepancy. An additional problem is that tumors with histologic features suggestive of malignancy but without evidence of invasion or metastases are classified as “atypical parathyroid adenoma/neoplasm” but may be reclassified as PC after either local recurrence or metastases have been documented, which may happen several years later [
2]. A staging system for PC was recently included in the 8
th edition of The American Joint Committee on Cancer Staging manual; however, prognostic stage groups are not currently available due to lack of high-quality information. In this system, the extent of local invasion, but not size, is considered. For the nodal stage, only the site: central (pN1a) vs. lateral (pN1b) neck is considered [
10,
23].
Treatment strategies for PC mainly rely on surgical removal of the tumor [
24]. In a study using data from the NCDB, Asare et al. [
14] reported that complete or partial tumor resection was associated with improved survival compared with those who did not undergo surgery. A high preoperative clinical suspicion of PC should prompt the surgeon to perform a more aggressive or en bloc resection to optimize disease outcomes, since the prognosis is largely dependent on the completeness of resection and avoidance of negative complications, such as intraoperative tumor rupture [
21,
23,
25,
26].
Adjuvant therapy options for unresectable tumors include radiotherapy, chemotherapy, immunotherapy, and ablation. Management of hypercalcemia includes variable combinations of bisphosphonates, calcimimetic agents, and the osteoclast inhibitor denosumab [
10,
27]. Deaths due to PC are typically associated with intractable hypercalcemia rather than the effect of the tumor burden itself. Treatment of symptomatic hypercalcemia associated with PC may involve intravenous hydration, furosemide diuretics, calcitonin, mithramycin, and dialysis [
22,
28]. Calcimimetic agents, such as cinacalcet, which act on the calcium-sensing receptor (CaSR) responsible for regulation and hemostasis of calcium through actions on the parathyroid gland and kidney [
29], are effective in reducing serum calcium levels by at least 1 mg/dl in 62% of patients with PC [
9]. Intravenous bisphosphonates, including zolendronate and pamidronate, are effective in treating hypercalcemia by inhibiting the mobilization of calcium from the bone into the circulation. Zoledronate is more effective than other IV bisphosphonates in hypercalcemia associated with PC [
10]. Oral bisphosphonates have not been reported to be effective for hypercalcemia associated with PC [
30]. Denosumab, a monoclonal antibody targeting the receptor activator of nuclear factor kappa-b ligand, has potent antiresorptive actions in bone by blocking osteoclast formation, function, and survival [
31] and has been used to treat refractory hypercalcemia associated with PC [
32].
Current data on adjuvant and neoadjuvant treatments are limited. Asare et al. reported no survival benefit in 51 (7.0%) patients who received radiation treatment [
14]. Some centers have reported that PC are radioresistant [
23,
33]; however, a retrospective review of 16 patients with parathyroid carcinoma reported 5- and 10-year survival rates of 100 vs. 80% and 69 vs. 43% for patients who received combined surgery and radiation vs. surgery alone, respectively [
33]. Nevertheless, large-scale studies examining the effectiveness of combination therapy are not available.
Molecular Insights, Ongoing Investigations, and Future Perspectives
The pathogenesis of PC likely involves the interplay of genetic and environmental factors, similar to many other malignancies. Childhood exposure to radiation, as well as concurrent thyroid or parathyroid disease increase the risk of PC [
3,
4]. The most consistent genetic abnormality associated with PC is mutation of
CDC73 (formerly called
HRPT2), first identified in the context of HPT-JT [
7,
8,
9], which encodes the protein parafibromin, a tumor suppressor gene [
33,
34]. The genetic syndrome most closely associated with PC is HPT-JT (at least 15% of cases). Germline
CDC73 mutations are present in up to 25% of seemingly sporadic PC cases [
9]. Somatic
CDC73/HRPT2-inactivating mutations can be demonstrated in 67% of sporadic cases [
9,
35,
36,
37]. PC has been linked to syndromes such as MEN types 1 and 2A, familial hypocalciuric hypercalcemia, and germline mutations of
CASR, RET, and
p53 [
6,
10,
11,
14]. To date, no parathyroid cell line system has been standardized or validated, hampering preclinical research. Patient-derived tumor organoids are another preclinical model system under investigation that could greatly facilitate both basic and translational research of PC. Genetically engineered mouse models with inactivating mutations of CDC73 and/or loss of parafibromin protein expression are being investigated. Walls et al. demonstrated mice with deleted CDC73 do not develop PC [
38]. However, they developed parathyroid adenoma (25%) and atypical parathyroid adenoma (75%) [
38]. Amplification of
CCND1 and overexpression of cyclin D1 were observed in 41% and 82% of PC, respectively [
34]. Other common somatic mutations in PC are
CDC73 (38%), TP53 (38%), MEN1 (31%), TERT (31%), PTEN (25%),
NF1, TSC2, KDR, and
CDK2A/B (12% each), along with amplification of 4q12 (
PDGFRA, KIT, and KDR) and 20q12 (
AURKA, SRC, TOP1, ZNF217, ARFRP1) (6% each). The tumor mutation burden was low (1.7 m/Mb) except in three cases with
CDC73 mutations > 20 m/Mb [
39]. The most frequent mutation was
CDC73 and TP53, which were mutually exclusive [
39].
Small next-generation sequencing (NGS) studies with < 20 patients have reported actionable genetic alterations in approximately one half to two thirds of the cases [
40]. Notably, NTRK1 was present in 5% and a high tumor mutational burden in approximately 20% [
24,
40]. The inclusion of standardized reporting systems, such as the AJCC 8th edition of the International Collaboration on Cancer Reporting (ICCR), is expected to yield useful data for establishing prognostic stages. Similarly, consistent and reliable imaging modalities for diagnosis and tumor response assessment are much needed. Fluorescent guidance has been used to identify parathyroid in thyroid surgeries to avoid hypoparathyroidism [
41]. Trials testing different modes of this technology in PC to prevent incomplete resection are currently underway (
Table 9). There is still an unmet need for effective therapeutic options for the management of complications, such as hypercalcemia and tumor burden. Due to the absence of consistent targetable driver mutations, use of next-generation sequencing technology will have immense value. Since MAPK, PI3K, and VEGF are overexpressed in PCs, regimens with multi-target tyrosine kinase inhibitors, such as cabozantinib, sorafenib, vandetinib + everolimus, and lenvatinib + everolimus, based on the affected pathway, have been explored with good palliative effect in a few cases [
34]. A brief response to cabozantinib has been observed in patients with the
KDR T888 K mutation, with a drop in PTH level and radiologic response [
39].
Temozolomide-based regimens for high O6-methylguanine DNA methyltransferase (MGMT) promoter methylation status led to prolonged remission in a case of metastatic PC, highlighting the success of precision oncology [
42]. Tumor-agnostic agents, such as pembrolizumab or nivolumab, for high tumor mutation burden or microsatellite instability-high tumors and NTRK inhibitors for PCs with NTRK mutations are applicable in selected cases. No effective adjuvant chemotherapies are currently available. Enrollment in clinical trials of novel agents should be considered for all patients when available. Genomic basket trials targeting actionable mutation(s) or immune signatures are likely the most viable options [
34]. Multikinase tyrosine kinase inhibitors may be attempted on a clinical trial, with the drug either matched to the altered pathway or used empirically to block commonly altered pathways. Off-label use of the same agent as a standard of care option may be considered on a case-by-case basis. Immunotherapy against parathyroid hormone in advanced and refractory parathyroid carcinomas has shown tumor regression with hormonal and biochemical normalization [
43]. Ongoing surveillance is warranted because of the risk of recurrence and distant metastasis, even after complete resection; however, surveillance intervals and modalities are not well defined.
Table 9 enlists select ongoing trials enrolling patients with PCs that are registered at clinicaltrials.gov.
Limitations of our study include the limited dataset of the retrospective registry, such as incomplete data on tumor size, invasion of lymph nodes or local tissue, presence of metastasis, and specific treatment and survival outcomes. Information on the levels of calcium and PTH, as well as the effect of hypercalcemia treatment on tumor burden, survival rates, renal disease, and bone mineral density associated with PCs are not available. Furthermore, the dataset lacks information on genetic mutations, as well as association of familial cancer with PCs. An organized, national, and/or international effort is needed to gather and sequence newly identified PCs based on the criteria and protocols established collectively.