Ex Vivo Intact Tissue Analysis Reveals Alternative Calcium-sensing Behaviors in Parathyroid Adenomas

Abstract

Context

The biochemical basis for clinical variability in primary hyperparathyroidism (PHPT) is poorly understood.

Objective

This study aimed to define parathyroid tumor biochemical properties associated with calcium-sensing failure in PHPT patients, and to relate differences in these profiles to variations in clinical presentation.

Methods

Preoperative clinical data from a sequential series of 39 patients undergoing surgery for PHPT at an endocrine surgery referral center in a large, public university hospital were evaluated for correlation to parathyroid tumor biochemical behavior. An intact tissue, ex vivo interrogative assay was employed to evaluate the calcium-sensing capacity of parathyroid adenomas relative to normal donor glands. Tumors were functionally classified based on calcium dose-response curve profiles, and clinical parameters were compared among the respective classes. Changes in the relative expression of 3 key components in the calcium/parathyroid hormone (PTH) signaling axis—CASR, RGS5, and RCAN1—were evaluated as potential mechanisms for calcium-sensing failure.

Results

Parathyroid adenomas grouped into 3 distinct functional classes. Tumors with diminished calcium sensitivity were the most common (18 of 39) and were strongly associated with reduced bone mineral density (P = 0.0009). Tumors with no calcium-sensing deficit (11 of 39) were associated with higher preoperative PTH (P = 0.036). A third group (6/39) displayed a nonsigmoid calcium/PTH response curve; 4 of these 6 tumors expressed elevated RCAN1.

Conclusion

Calcium-sensing capacity varies among parathyroid tumors but downregulation of the calcium-sensing receptor (CASR) is not an obligate underlying mechanism. Differences in tumor calcium responsiveness may contribute to variations in PHPT clinical presentation.

Primary hyperparathyroidism (PHPT) is a common endocrine neoplastic disorder biochemically defined by constitutively elevated hypersecretion of the parathyroid hormone (PTH), leading to chronic hypercalcemia and a spectrum of clinical sequelae, including bone mass attrition and increased risk of pathological fracture (1, 2). Osteoporosis and osteopenia secondary to hyperparathyroidism represent a significant clinical challenge, with as many as 44% of hyperparathyroidism patients exhibiting loss of bone mineral density (3-5). While the central physiological deficit in PHPT—the failure to maintain calcium homeostasis—is well recognized, the underlying molecular mechanisms that drive this systemic dysfunction and that account for the divergent patterns of clinical presentation among PHPT patients have not been fully characterized.

In the normal parathyroid gland, PTH secretion is tightly coupled to serum calcium concentration. Reduced serum calcium levels promote PTH secretion to stimulate hypercalcemic activities in the kidney (increased calcium retention), the intestine (increased 1,25[OH]2D-dependent calcium absorption), and bone (enhanced scavenging of mineralized calcium from skeletal stores). Conversely, increased serum calcium levels inhibit PTH secretion via activation of the calcium-sensing receptor (CASR), a C-type G-protein coupled receptor. Genetic evidence supports the pivotal role of CASR in calcium homeostasis: constitutively activating gain-of-function mutations in CASR induce a hypocalcemic phenotype (autosomal dominant hypocalcemia), while inactivating heterozygous loss-of-function germline mutant alleles cause familial hypocalciuric hypercalcemia, a mild form of hyperparathyroidism (6). However, complete loss of CASR activity is not likely to be a common underlying mechanism in PHPT. Somatic mutations in CASR are rarely found in human parathyroid tumors and do not account for the vast majority of PHPT incidence (7). While reduced expression of CASR has been associated with primary and secondary hyperparathyroidism (7, 8), neoplastic parathyroid tissues retain the capacity to initiate CASR-dependent intracellular signaling (9-11), and CASR activity can be mobilized by calcimimetics for the management of secondary hyperparathyroidism (12) and biochemical control in primary hyperparathyroidism (13). These observations suggest that qualitative differences in the relative responsiveness of CASR-mediated calcium sensing may be more likely to explain phenotypic differences in PHPT presentation than complete abrogation of the CASR biochemical signaling pathway. Possible mechanisms could include increased opposition to CASR downstream signaling by RGS5, a regulator of G-protein signaling (14, 15), or an uncoupling of PTH secretion from CASR by enhanced activation of the regulator of calcineurin RCAN1 and the FGF23 signaling pathway (16).

Consistent with the idea that loss of CASR expression is not the sole determinant of calcium-sensing deficiency in PHPT, previous work from our laboratory and others has revealed heterogeneity in the kinetics and amplitude of CASR-mediated G-protein signaling among human parathyroid tumors (8, 11, 17). The net effect of this heterogeneity produced measurable differences between tumors in the calcium concentrations required to initiate CASR-dependent G-protein coupled downstream biochemical signaling cascades (10). Importantly, tumors manifesting a greater degree of impairment in CASR signaling were selectively associated with patients who had bone mineralization deficits (18).

While compelling, these prior studies focused on proximal signaling events by cells in culture rather than on the physiologically relevant endpoint of PTH secretion by intact parathyroid gland tissue. The absence of normal donor parathyroid tissue as a reference standard was another important limitation. A major obstacle to studying calcium sensing in normal and neoplastic parathyroid tissue has been the lack of experimentally tractable model systems that can faithfully reproduce the dynamic calcium response behaviors of the intact organ. To address this, our group developed an ex vivo intact tissue assay system for interrogative, dynamic assessment of PTH secretion by intact normal and neoplastic parathyroid glands in response to changes in ambient calcium concentration. We used this approach to test the hypothesis that discrete patterns of tumor calcium-sensing behavior, as revealed by variations in the index features of their respective calcium/PTH biochemical response curves, may be linked to differences in preoperative presentation and disease course in patients with PHPT. Here, we describe the existence of distinct biochemical response profiles associated with different PHPT clinical presentations and demonstrate a specific tumor biochemical signature associated with increased risk of bone mass attrition.

Methods

Bone Density Measurements

Bone mineral density (BMD) was determined by dual-energy x-ray absorptiometry (DXA) and reported as a T-score, a metric indicating the number of standard deviations a subject’s BMD differs from the average BMD of a healthy 30-year-old adult. A diagnosis of osteoporosis is based upon a T-score of more than 2.5 SD below the mean (T ≤ −2.5). A T-score between −1.0 and −2.5 is diagnostic for osteopenia. A T-score of −1 or greater indicates no BMD deficit. The T-scores reported here represent the lowest value among all skeletal sites examined for each patient who underwent this testing.

Human Parathyroid Tissue Collection

Normal tissue controls

Normal human parathyroid tissue was obtained from eucalcemic donors through our institution’s solid organ transplant service, using a fully authorized tissue procurement protocol for the recovery of viable, intact parathyroid glands. Dissected glands were immediately placed in University of Wisconsin cold storage solution (UW) (19) on ice and sectioned into a series of 5 to 10 mg pieces on an ice-cold tissue culture plate in a laminar flow tissue culture hood. Tissue pieces were then allocated for fixation in 4% paraformaldehyde (PFA) for histology, UW for viability assessment, and secretion medium (MEM-EBSS-CMF medium supplemented with 0.5mM calcium, 0.5mM magnesium, 0.2% bovine serum albumin, and 20mM HEPES [pH 7.4]) for ex vivo calcium-response PTH secretion assays.

Parathyroid adenoma collection

Parathyroid adenoma specimens were obtained under an institutional review board (IRB)-approved protocol (IRB protocol number 19-27072) from patients undergoing surgery for primary hyperparathyroidism at our high-volume endocrine surgery center. Clinical, demographic, and pathological patient data were collected from the medical record and anonymized by the study clinical research coordinator in compliance with IRB requirements. The tumor samples were placed in UW solution (19) on ice upon extirpation. The live tissue was then subdivided as described above for the parathyroid donor samples. Sectioned tissue pieces were allocated into 4% PFA for histology, UW solution for viability assessment, or secretion medium (MEM-EBSS-CMF medium supplemented with 0.5mM Ca, 0.5mM Mg, 0.2% bovine serum albumin, and 20mM HEPES [pH 7.4]) for ex vivo calcium-response PTH secretion assays.

Ex Vivo Calcium-Response PTH Secretion Assay

Dynamic changes in PTH secretion from normal and parathyroid adenoma tissue in response to changes in ambient calcium concentrations were evaluated using a real-time ex vivo interrogative biochemical assay (20, 21). Parathyroid tissue specimens were placed in individual transwell cell culture inserts (polycarbonate membrane, 0.4 μm pore; Corning, catalog number 3413) with 100 μL of 0.5mM calcium chloride secretion medium in the interior chamber. At least 3 different tissue samples from each tumor specimen were assayed in parallel to verify intraspecimen reproducibility. The Transwell insert was placed in 24-well dishes containing 600 μL of 0.5mM calcium chloride secretion medium per well outside of the Transwell chamber. The 24-well dish was then transferred to a humidified tissue culture incubator at 37 °C and 5% CO₂ and allowed to equilibrate for 60 minutes, replacing the secretion medium on both sides of the Transwell at 30 minutes. After the equilibration period, the Transwell chambers containing the tissues were sequentially transferred to new wells containing 600 μL of medium with a range of calcium concentrations (0.5, 0.75, 1.0, 1.25, 1.5, 2.0, and 3.0mM), changing the inner Transwell media to the new calcium concentration during each transfer. The tissues were incubated for 15 minutes at each concentration point, and then the media from the exterior, cell-free compartment of each well was collected in individual microfuge tubes and were briefly spun at 16000g for 15 seconds to remove particulate material or inadvertently transferred cells. A final, low calcium (0.5mM) incubation was performed at the end of the series as a repeat maximal secretion stimulus in confirm that intracellular PTH stores had not been depleted. The concentration of PTH secreted into the media was determined in triplicate by an enzyme-linked immunosorbent assay (ELISA) assay specific for intact human PTH (aa 1-84), with a detection range of 11 to 1323 pg/mL (Quidel, catalog number 60-3100). PTH concentrations were plotted as a function of log ambient calcium concentration using GraphPad Prism 9.0. For EC50 curves, PTH secretion was expressed relative to the maximal amount of PTH produced (at 0.5mM calcium, the lowest calcium concentration) and plotted as a function of log[Ca2+]. Each parathyroid tissue piece was stained with Hoechst 33342 and propidium iodide for viability verification at the end of the PTH assay.

Tissue Histology

Parathyroid tissue was fixed in 4% paraformaldehyde (PFA) in 0.1M phosphate-buffered saline (pH 7.6) overnight at room temperature. After fixation, the tissue was rinsed with ddH2O and the PFA was replaced with 70% ethanol for storage. The tissue was embedded in paraffin and 5-micron sections were stained by hematoxylin and eosin (H&E). Tissue sections were examined on an Olympus CX43 upright microscope and images were captured with a SPOT Idea 5.0 Mp CMOS color digital camera (SPOT Imaging).

Immunofluorescence Staining

Sections of normal parathyroid tissue and parathyroid adenomas were stained with primary antibodies against CASR, RGS5, RCAN1, HRPT2, GCM2, MEN1, RET, and CDKN1B. Sections underwent deparaffinization and antigen retrieval using standard conditions, followed by a 1-hour incubation at room temperature with 3% bovine serum albumin (Fisher Scientific, catalog number BP1600-100) in Tris-buffered saline for blocking. Immunofluorescence detection of CASR, RGS5, RCAN1, HRPT2, GCM2, MEN1, RET, and CDKN1B in human parathyroid gland sections were performed with a custom anti-CASR polyclonal antibody (VA610, raised against CASR epitope ADDDYGRPGIEKFREEAEERDI, was generated and kindly provided to us by Dr. Wenhan Chang of UCSF) (1 μg/mL in blocking solution), anti-RGS5 monoclonal antibody (Origene, catalog number TA503075S; 1:1000 in blocking solution), anti-RCAN1 polyclonal antibody (Novus, catalog number NBP1-46853; 1:1000 in blocking solution), anti-HRPT2 monoclonal antibody (Santa Cruz Biotechnology, catalog number sc-33638; 1:200 in blocking solution), anti-GCM2 polyclonal antibody (Abcam, catalog number ab201170; 1:500 in blocking solution), anti-MEN1 polyclonal antibody (Sigma, catalog number HPA030342; 1:100 in blocking solution), anti-RET polyclonal antibody (Sigma, catalog number HPA008356; 1:300 in blocking solution), or anti-CDKN1B polyclonal antibody (Sigma, catalog number HPA059086; 1:100 in blocking solution). Negative controls were stained with their isotype matched IgG controls, respectively. Slides were washed 3 times in phosphate-buffered saline, and secondary antibodies (Alexa 488 goat anti-mouse IgG or Alexa 488 goat anti-rabbit IgG (Invitrogen, catalog numbers A11001 and A11034) were used at a 1:500 dilution in blocking solution for staining. Plasma membrane was stained with wheat germ agglutinin 594 (Invitrogen, catalog number W11262) at 5 μg/mL for 10 minutes at room temperature. Stained slides were cover slipped using ProLong Gold antifade mounting media with NucBlue (Invitrogen, catalog number P36981). Excitation wavelengths of 385 nm, 475 nm, and 567 nm were used for nuclear, CASR, RGS5, RCAN1, HRPT2, GCM2, MEN1, RET, CDKN1B, and plasma membrane visualization. All photos were taken at the same exposure, gain, and offset using a Zeiss AxioVert A1 microscope (Carl Zeiss) equipped with a Colibri 7 LED illuminator (Carl Zeiss), Axiocam 506M camera (Carl Zeiss), and ZEN Pro software. The staining patterns for each antibody were verified by confirmation experiments utilizing different visualization methods, including alternate fluorophore or HRP-conjugated secondary antibodies.

Confocal Imaging

Confocal images were obtained using a Nikon ECLIPSE Ti spinning disk confocal microscope (Nikon) equipped with 488 and 561 nm lasers, a CSU-X confocal spinning disk unit (Andor), a Zyla 5.5-megapixel sCMOS camera (Andor), and a 60× oil-immersion objective (NA 1.4; Nikon). Images were acquired using Metamorph software Version 7.10.2.240 (Molecular Devices). Images were 1800 × 1800 pixels in size and were analyzed using ImageJ (NIH). Histograms of the merge channel images were held constant with those of the individual wavelength channels. Z-stack images of each field of view were collected with a Z-dimension step size of 1 µm and a total of 7 steps. Colocalization analysis of CASR and the plasma membrane marker wheat germ agglutinin (WGA) were performed using the Coloc 2 module in ImageJ. Manders’ colocalization coefficients tM1 (the proportion of CASR signal that co-localizes with the plasma membrane marker) and tM2 (the proportion of the plasma membrane marker signal that co-localizes with CASR) were calculated to evaluate CASR and plasma membrane colocalization for the different biochemical groups (22). Images were processed using an auto-threshold setting to remove background and nonspecific image noise using a standard ImageJ algorithm. Colocalization analysis was performed using the entire field of view in each image to eliminate potential sampling bias. Approximately 200 to 300 cells were analyzed in each image.

Fluorescence Intensity Calculation and Quantification

The sum intensity (I_sum) of the epifluorescence output channel of interest in each immunofluorescent image was acquired in ZEN Pro software (Carl Zeiss). The background intensity (I_b) of each image was calculated based on the pixel number per image multiplied by the average pixel intensity of 4 defined regions of interest (ROIs) without tissue presence. Background-subtracted intensity per cell (I_bs) for CASR, RGS5, RCAN1, and their matched IgG controls were calculated in each image according to the following equation:

$${\displaystyle \begin{array}{l}{\displaystyle I_{bs}=\frac{I_{sum}-I_b}{Totalcellnumber}}\end{array}}$$

Absolute intensity values per cell (I_abs) for CASR, RGS5, and RCAN1 were calculated by:

$${\displaystyle \begin{array}{l}{\displaystyle I_{abs(CASR)}=I_{bs(anti-CASR)}-I_{bs(anti-CASRmatchedIgG)}}\end{array}}$$

$${\displaystyle \begin{array}{l}{\displaystyle I_{abs(RGS5)}=I_{bs(anti-RGS5)}-I_{bs(anti-RGS5matchedIgG)}}\end{array}}$$

$${\displaystyle \begin{array}{l}{\displaystyle I_{abs(RCAN1)}=I_{bs(anti-RCAN1)}-I_{bs(anti-RCAN1matchedIgG)}}\end{array}}$$

The baseline reference mean and SD for each primary antibody signal was calculated from imaging sections from the normal parathyroid glands used in the signaling studies, obtained from 5 independent donors. The relative fluorescence intensity scoring in the adenoma samples was calculated by subtracting the mean absolute intensity values of normal parathyroid tissue (N = 5) from the individual parathyroid adenoma absolute intensity values. The fluorescence intensity scoring categories of the parathyroid adenoma specimens relative to the mean normal tissue signal are defined as follows:

-	>1 SD below mean normal value
+	Within ± 1 SD of mean normal value
++	>1-2 SD above mean normal value
+++	>2 SD above mean normal value

Statistical Analysis

GraphPad Prism 9.0 (GraphPad Software) was used to create figures and perform statistical analyses, including standard curve plotting and interpolation, nonlinear regression curve fit, EC50 determination, Chi-squared analysis, Fisher exact test, analysis of variance (ANOVA), Pearson correlation coefficient calculations, and 2-tailed Student t tests.

Results

Study Cohort

Parathyroid adenomas were obtained under an IRB-approved protocol with fully informed consent from a sequential series of 39 patients undergoing surgery for primary hyperparathyroidism at our institution. All study participants were nonfamilial, sporadic PHPT patients. The characteristics of the study population are shown in Table 1. Preoperative PTH ranged from 45 to 300 pg/mL and correlated inversely with vitamin D levels, as has been previously reported (23) (Fig. S1 (24)). The patient cohort comprised predominantly older (median age = 65 years) Caucasian women, consistent with the current demographic distribution of PHPT patients in the United States (2, 25). A deficit in BMD was present in 25 of the 39 patients, 7 of whom presented with a DXA finding of osteoporosis (lowest T-score of < −2.5). For comparison, a total of 7 viable parathyroid glands from 5 independent, eucalcemic donors (4 female, 1 male; mean age = 47 years) were obtained and analyzed as normal tissue reference controls.

Table 1.

Study population characteristics

Patient characteristics	(n = 39)
Demographics
Age at surgery (years) – median (IQR)	65 (55, 72)
Sex
Female	35 (90%)
Male	4 (10%)
Race
White	33 (85%)
Black or African American	0
Other/Unknown	6 (15%)
Ethnicity
Hispanic or Latino	3 (8%)
Non-Hispanic or Latino	36 (92%)
Clinical presentation
Nephrolithiasis	8 (21%)
Osteoporosis	7 (18%)
Osteopenia	18 (46%)
Number of glands removed
1	33 (85%)
2	5 (13%)
3	1 (2%)
4	0
Preoperative biochemistry	Median (IQR)
Intact PTH (pg/ml)	104 (85, 133)
Serum calcium (mg/dL)	10.8 (10.4, 11.0)
Serum creatinine (mg/dL)	0.76 (0.68, 0.99)
Vitamin 25 D (ng/mL)	33.65 (28.25, 42.75)
Preoperative bone density	Median (IQR)
Lowest T-score	−1.95 (−2.78, −1.03)

Parathyroid adenomas were obtained under an IRB-approved protocol with fully informed consent from a sequential series of 39 patients undergoing surgery for primary hyperparathyroidism at our institution. All study participants were nonfamilial, sporadic PHPT patients. Race and ethnicity were self-reported by the study participants. The presence of nephrolithiasis, osteoporosis, or osteopenia was noted from the patients’ preoperative medical record and referral history. Preoperative biochemical values were determined from blood drawn at the patient’s clinic appointment prior to scheduling surgery. T-scores were determined from the patient’s referral medical records.

Abbreviations: IQR, interquartile range; PTH, parathyroid hormone.

Tumor Calcium Responsiveness

We employed a modified ex vivo interrogative assay (21) to quantitate the dynamic modulation of PTH secretion from these tissues in response to changes in ambient calcium concentration. In this system, intact tumor or normal parathyroid gland tissue is challenged with a range of different extracellular calcium concentrations, and PTH secretion into the media over a 15-minute period is then quantitated. The relationship between calcium concentration and PTH production in normal parathyroid tissue can be fitted to a sigmoid dose-response curve (Fig. 1A, blue curve) reflective of the cooperative allosteric inhibition kinetics of calcium ions acting upon multiple binding sites in the extracellular domain of the calcium-sensing receptor (CASR) (26-29). The mean calcium concentration required to suppress PTH secretion to half of maximum (EC50) in normal parathyroid tissue (n = 5) is 1.022mM (95% CI, 0.8718-1.102). In contrast, 18 of the adenomas demonstrated a rightward shift of the calcium response curve, with an elevated mean EC50 of 1.278mM (95% CI, 1.209-1.348) indicating diminished sensitivity to calcium challenge (Fig. 1A, red curve). Eleven tumors were found to have EC50 values similar (P = 0.569) to normal tissue in this assay (Fig. 1B, purple curve), with a sigmoid dose-response curve closely aligned to the normal tissue profile and a mean EC50 of 0.9359mM (95% CI, 0.8738-0.9980). A third group of 6 tumors generated a nonsigmoid dose-response curve (Fig. 1C). At least 3 different tissue samples from each adenoma or normal parathyroid tissue specimen were assayed in parallel. We observed no significant differences in calcium response behaviors among replicate samples in our study cohort (Fig. 1D). The EC50 values derived from multiple sampling sites within each tumor were highly consistent. The SD to mean EC50 ratios among intraspecimen sampling sites were less than 10% in each calcium response category, confirming that the observed differences in calcium sensitivity between the EC50 classes were highly reproducible throughout each tumor and were not due to sampling artifact (Table S1 (24)). Variability within each calcium response category was smaller than the aggregate variability metric for the entire cohort, as expected given the differences in overall EC50 values between the groups.

Figure 1.

Calcium/PTH dose-response curves. Live parathyroid tissue was interrogatively challenged with a series of ambient calcium concentrations, and PTH secretion over a 15-minute period at each calcium concentration was determined by ELISA. PTH secretion expressed as a fraction of the maximal value for each specimen is plotted as a function of log calcium concentration. Data point symbols represent the mean ± SD from 5 independent normal donor response curves, each consisting of at least 3 independent biological replicates per sample, or the mean and SD from the indicated number of adenoma samples, each with at least 3 independent biological replicates per sample. The response curve profiles were plotted using a variable slope, 4 parameter log(inhibitor) vs response model (Prism 9.0). (A) Comparison of normal parathyroid tissue (n = 5, blue line) to parathyroid adenomas with attenuated calcium sensitivity (n = 18, red line). The calcium concentration required for 50% inhibition of maximal PTH secretion (EC50) is indicated by the thin black line. The mean EC50 for normal tissue (N) is 1.012mM calcium. The mean EC50 for the adenomas in this group is 1.242mM calcium. (B) Comparison of normal parathyroid tissue (blue line) with parathyroid tumors with no loss of calcium responsiveness (purple line). The mean EC50 for the adenomas in this group (n = 12) is 0.9934mM, compared with 1.012mM for normal tissue. (C) The calcium/PTH relationship for nonsigmoid response curve tumors (n = 6) is shown relative to the normal tissue response curve (n = 5). (D) Tumor calcium response behavior is highly consistent across different sampling regions. Calcium/PTH response curves were generated from 3 different sections of an individual tumor. Representative curves from a high EC50 tumor, a nonsigmoid response curve tumor, and a low EC50 tumor are shown. Symbols represent the mean ± SD at each calcium concentration point. Lines represent replicate 1 (red), replicate 2 (green), and replicate 3 (blue). (E) Calcium/PTH dose-response curves. PTH secretion normalized relative to maximal production at 0.5mM extracellular calcium is plotted as a function of calcium concentration. Symbols represent the mean ± SD at each concentration for high EC50 tumors (red line/symbols; n = 18), low EC50 tumors (green line/symbols; n = 11), nonsigmoid response curve tumors (black line/symbols; n = 6), or normal parathyroid tissue (blue line/symbols; n = 5). (F) Calcium EC50 distribution among PHPT adenomas is bimodal. The range of calcium EC50 values was binned into 30 equivalent segments representing 0.03mM increments on the x-axis. The number of tumors in each bin is plotted on the y-axis. Calcium EC50 values obtained from normal donor parathyroid tissue are shown in blue. Black bars represent parathyroid tumors. (G) Absolute PTH production comparison between high EC50 (red line), low EC50 (green line), nonsigmoid response curve tumors (blue line), and normal parathyroid tissue (black line) across the same extracellular calcium concentrations used to generate the setpoint curves. PTH values represent the total concentration of PTH in pg/mL secreted into the media after a 15-minute incubation period, per mg wet weight of tissue. Symbols represent the mean and SD for biological triplicates (3 different sampling regions of each specimen) for tumors or tissues in each category. (H) Suppressibility of PTH secretion at 3mM extracellular calcium. PTH suppressibility represents the concentration of secreted PTH at 3mM calcium relative to the concentration of secreted PTH at 0.5mM calcium, expressed as a ratio. Each dot represents a different specimen. Horizontal lines represent the mean value for each group.

Four specimens did not secrete PTH sufficiently above background to plot a dose-response curve; these specimens were found to contain minimal parathyroid tissue upon subsequent histological examination. None of the specimens in the series were found to be oxyphil-dominant adenomas (defined as >75% oxyphil content (30)). The difference between the mean EC50s of the high and low EC50 groups was highly significant (P ≤ 0.0001 by 2-tailed Student t test), as was the difference between the normal tissue and high EC50 tumor means (P = 0.0080 by 2-tailed Student t test) (Fig. 1E). A histogram plot of EC50 values among the parathyroid adenomas was suggestive of a bimodal distribution (Fig. 1F). Absolute PTH production was higher in low EC50 tumors than in normal tissue or high EC50 tumors (Fig. 1G).

All 3 classes of tumors demonstrated a failure to suppress PTH secretion at high (3mM) calcium concentrations compared with normal tissue (Fig. 1H) (high EC50 vs normal, P = 0.0010; low EC50 vs normal, P = 0.0409; nonsigmoid vs normal, P = 0.0338; all by 2-tailed Student t test). Maximal suppressibility did not differ between the 3 tumor classes (P = 0.9779 by ANOVA), but as a group the tumors were significantly different from normal tissue (P = 0.0424 by ANOVA). In addition, there was reduced maximal suppression, reduced cooperativity (smaller Hill coefficient), and a relatively impaired PTH secretory response at low Ca2+ concentration. Mean tumor weight and preoperative serum calcium were not different between the 3 calcium response groups (by ANOVA, P = 0.9864 for tumor weight and P = 0.1591 for preoperative calcium).

Multiple parathyroid glands were removed from 4 patients in our series. We performed calcium EC50 analysis on each of the tissue specimens resected from these patients. One case where 3 glands were removed was subsequently found to be due to a single adenoma. The first 2 glands removed proved to be histologically normocellular, and intraoperative PTH did not decline until removal of the third, hypercellular gland. Calcium EC50 was elevated in the hypercellular gland but normal in the first 2 glands. In a second multi-gland case, the tissue samples received in the laboratory contained insufficient parathyroid tissue for biochemical analysis. In a third case, both right and left superior glands were found to be enlarged upon bilateral neck exploration. Intraoperative PTH did not decline until both enlarged glands were removed. Both of the resected glands demonstrated elevated an EC50 setpoint (1.263mM/right superior and 1.167mM/left superior). In the fourth and final multi-gland case, the superior and inferior right parathyroid glands were removed. Intraoperative PTH did not decline until both suspect glands were removed. Both glands were found to be hypercellular on histopathological analysis. The larger, right superior gland manifested no EC50 shift (0.867mM), while the right inferior gland had an elevated EC50 (1.350mM).

Clinical Correlates to Tumor Biochemical Behavior

To identify clinical correlates to the respective calcium response profiles, we examined preoperative laboratory and bone density data for the 39 patients in the study cohort and probed for potential associations between tumor biochemical behavior and phenotypic presentation. Bone mineral density was determined by DXA and is reported as the lowest T-score among all skeletal sites examined for each patient who underwent this testing. Twenty of the 25 T-scores reported in our study cohort were determined from cortical bone sites (14 at the femoral neck, 4 at the distal radius, and 2 at the hip). Five scores were from trabecular bone at the lumbar spine. The lumbar spine T-scores were not associated with the patient’s EC50 group; of the 5 lumbar spine scores, 2 were in the high EC50 group, 1 was in the low EC50 group, and 2 were in the nonsigmoid curve group. Patients whose tumors displayed an elevated EC50, defined as >1.10mM (the upper 95% CI limit for normal tissue), were more likely to present with osteoporosis (Fig. 2A), with a relative risk of 3.556 (P = 0.0152 by 2-tailed Fisher exact test). The mean EC50 among the 7 patients with osteoporosis was 1.249mM, significantly higher than that of the 10 patients with T-scores indicating no BMD deficit (0.9921mM) (Fig. 2B) (P = 0.0009). Patients presenting with osteopenia (n = 7) had a mean EC50 of 1.100mM, intermediate between osteoporotic patients and patients with no BMD deficit, although the difference fell short of statistical significance (P = 0.0621). Tumor EC50 values correlated significantly with the patient’s lowest T-score (Fig. 2C) (P = 0.0147). Mean patient age at time of surgery did not vary significantly between the EC50 groups (P = 0.2753 by ANOVA), and tumor EC50 as a continuous variable did not correlate with patient age (P = 0.470).

Figure 2.

PHPT patients with high EC50 tumors are more likely to present with osteoporosis. Adenomas with a calcium EC50 greater than 1.10mM (the upper limit of the 95% CI for this metric in normal tissue) are classified as high EC50 tumors. Adenomas with an EC50 less than 1.10mM are classified as low EC50 tumors. (A) Distribution of patients with osteoporosis (T-score < −2.5) or no bone mineral density (BMD) deficit (T-score > −1.0), grouped by EC50 status. (B) Scatter plot of individual EC50 scores among patients with osteoporosis, osteopenia (T-score between −1 and −2.5), or no BMD deficit; the mean value for each group is indicated by the horizontal black line. The pink region demarcates the high EC50 range. The blue region marks the low EC50 range. The mean EC50 in the osteoporosis group is significantly different than the mean EC50 in the group with no BMD deficit (P = 0.0009 by 2-tailed Student t test). (C) Correlation between tumor EC50 and lowest T-score. EC50 values are plotted as a function of the patient’s lowest T-score. Pearson’s correlation coefficient testing demonstrated a highly significant linear relationship between these variables (P = 0.0147).

While tumors with EC50 values in the normal range appeared to retain responsiveness to extracellular calcium challenge, the overall amount of PTH that these tumors produced ex vivo per mg of tissue was significantly higher than that of high EC50 tumors (Fig. 3). High EC50 tumors produced a mean of 83.84 ± 52.89 pg PTH per mg wet weight of tissue over 15 minutes at 0.5mM calcium, compared with 315.7 ± 208.6 pg PTH per mg wet weight of tissue secreted by low EC50 tumors under the same conditions (P = 0.0002). Nonsigmoid response curve tumors produced a mean of 151.3 ± 143.5 pg PTH per mg tissue, not significantly different from the high EC50 group (P = 0.1051) or from the low EC50 group (P = 0.1078). Consistent with the ex vivo data, patients with low EC50 tumors had a higher mean preoperative PTH than patients with high EC50 tumors (141.3 ± 67.70 pg/ml vs 94.56 ± 32.74 pg/mL; P value for difference = 0.0171) (Fig. 4). Further supporting the existence of a clinical difference between the high and low EC50 subgroups, the ratio of preoperative PTH to preoperative serum Ca2+ was significantly higher in patients with low EC50 tumors (14.54 ± 6.106, n = 10) than in patients with high EC50 tumors (8.578 ± 3.066, n = 18) (P = 0.0019 for difference by 2-tailed t test). Mean preoperative PTH in patients with nonsigmoid response curve tumors (101.7 ± 30.65 pg/mL) was not statistically different from patients with high EC50 tumors (P = 0.1953) or low EC50 tumors (P = 0.6448).

Figure 3.

Maximal PTH production in PHPT adenomas. The total amount of PTH secreted into the media over 15 minutes at 0.5mM ambient calcium was determined by ELISA and then normalized to the wet weight in mg of tissue used in the assay. Bars indicate the mean and SD for each group. High EC50 in black (n = 18); low EC50 in white (n = 11); nonsigmoid response curve profile in gray (n = 6). Maximum PTH production by low EC50 tumors is significantly higher than for high EC50 tumors (P = 0.0002 by 2-tailed Student t test).

Figure 4.

Preoperative PTH is higher in patients with low EC50 tumors. Scatter plots of individual preoperative PTH levels among patients with high EC50 tumors (n = 18) and low EC50 tumors (n = 11). Horizontal black lines represent the mean preoperative PTH levels for each group. The mean preoperative PTH level in low EC50 tumors is significantly higher than in high EC50 tumors (P = 0.0171 by 2-tailed Student t test).

Quantitation of CASR, RGS5, and RCAN1 Protein Abundance in Parathyroid Tumors

To determine whether loss or downregulation of CASR abundance could be a common factor in PHPT tumors, immunofluorescence was used to visualize CASR protein levels in the adenomas relative to normal parathyroid tissue. All image fields were acquired under identical conditions with respect to background thresholding, illumination intensity, exposure time, and gain, and the pixel counts in the fluorochrome channel of interest were generated in a linear relationship to signal strength over the entire signal intensity range. Threshold-corrected, IgG-background-subtracted pixel counts for the anti-CASR immunofluorescence signal output for each image field were normalized to cell number to generate a CASR signal intensity score. The mean signal intensity in normal parathyroid tissue was classified as 1+, and a SD value was calculated based on the variation observed among 5 independent normal donor parathyroid glands. Sections with an intensity score more than 1 SD below the normal tissue mean were classified as negative (“−”). Sections within ± 1 SD of mean normal were classified as 1+. Sections with intensity scores greater than 2 or 3 SD above the normal tissue mean were classified as 2+ and 3+, respectively (Fig. 5). Among the 39 tumors, silencing of CASR expression was not uniformly observed. CASR abundance was greater than in normal tissue in 6 tumors (3 scored as 3+, 3 scored as 2+), equivalent to normal tissue (1+) in 18 tumors, and less than in normal tissue (−) in 15 tumors (Table 2). In all cases where the protein was detectable, CASR appeared to be appropriately localized to the plasma membrane. Image analysis for colocalization of CASR signal and a plasma membrane marker (wheat germ agglutinin, WGA) in planar 63× oil-immersion objective images (Fig. 6A) and in spinning disk confocal imaging (Fig. 6B) demonstrated no differences in CASR plasma membrane localization in CASR-expressing tumors in our study cohort and no correlation with tumor EC50 category (Table 3). Using the Coloc2 module of Image J, Manders Colocalization Coefficients were calculated to quantitate and compare the positional overlap of CASR with a plasma membrane marker (WGA) in confocal image planes from each tumor type. The proportion of CASR signal that colocalized with the WGA marker was designated tM1, and the proportion of the WGA signal that colocalized with CASR was designated tM2. The tM1 and tM2 mean values did not differ significantly between the different biochemical groups of tumors (by ANOVA, the P value for difference among tM1 means = 0.1135; P value for difference among tM2 means = 0.5691). CASR expression or relative abundance did not correlate with calcium response category, as all 3 profile types shared similar proportions of CASR+ and CASR− tumors (P = 0.4227 by Chi-square analysis).

Figure 5.

Scoring example expressing immunofluorescence signal intensity of CASR in tumors relative to normal parathyroid tissue. Magnification is 630× for all images; scale bar = 50 microns. Representative examples of CASR expression levels in adenomas (left panels) compared to normal donor tissue (right panel). Adenomas expressing CASR at a level more than 1 SD below normal tissue are scored as negative (“−”). Levels equivalent to or within 1 SD of normal are scored as 1+ (“+”), more than 1 SD above normal are scored as 2+ (“++”), or more than 2 SD above normal are scored as 3+ (“+++”). Blue = DAPI; green = anti-CASR; red = wheat germ agglutinin AlexaFluor 594.

Table 2.

Immunofluorescent staining intensity of CASR, RGS5, and RCAN1 in parathyroid tumors

	Intensity score	High EC50 (n = 18)	Low EC50 (n = 11)	Nonsigmoid (n = 6)
CASR	+++	6% (1)	0% (0)	33% (2)
	++	17% (3)	0% (0)	0
	+	44% (8)	64% (7)	13% (1)
	-	33% (6)	36% (4)	50% (3)
RGS5	+++	17% (3)	18% (2)	33% (2)
	++	0	0	0
	+	6% (1)	18% (2)	0
	-	78% (14)	64% (7)	67% (4)
RCAN1	+++	0	(18%) 2	67% (4)
	++	17% (3)	9% (1)	0
	+	39% (7)	55% (6)	33% (2)
	-	44% (8)	18% (2)	0

Threshold-corrected, IgG-background-subtracted pixel counts for each immunofluorescent signal were normalized to cell number to generate a signal intensity score. The mean signal intensity in normal parathyroid tissue is defined as 1+, and a SD value was calculated based on the variation observed among 5 independent normal donor parathyroid glands. Intensity scores more than 1 SD below the normal tissue mean were classified as negative (“-”). Sections within ± 1 SD of mean normal were classified as +1. Sections with intensity scores greater than 2 or 3 SD above the normal tissue mean were classified as 2+ and 3+, respectively.

Figure 6.

CASR localization is not altered in tumors with different calcium response profiles. (A) Representative immunofluorescent images from normal parathyroid tissue and from tumors of each of the 3 response curve groups are shown. Images were captured with a 63× oil-immersion objective lens. Green = CASR; red = wheat germ agglutinin, a plasma membrane marker; blue = DAPI. Scale bar = 50 microns. (B) Confocal microscopy images demonstrate no differences in CASR subcellular localization in the plasma membrane in the 3 calcium-sensing classes of parathyroid tumors. Spinning disk confocal microscopy was employed to generate optical sections from normal parathyroid tissue or from high EC50, low EC50, or nonsigmoid response curve tumors. Each image was extracted from a Z-stack series of 7 frames with a step size of 1 micron. CASR signal is in green. WGA = wheat germ agglutinin/AlexaFluor 594, a plasma membrane marker, shown in red. The 2 different color output channels were merged to create overlays, which demonstrate colocalization of the CASR and WGA signal (yellow). Magnification = 600×. Scale bars = 50 microns.

Table 3.

Mander’s colocalization coefficients for CASR and WGA in parathyroid adenomas

	Normal (n = 5)	High EC50 (n = 13)	Low EC50 (n = 6)	Nonsigmoid (n = 5)
tM1	0.4998 ± 0.059	0.4130 ± 0.091	0.4114 ± 0.047	0.4948 ± 0.060
tM2	0.6475 ± 0.079	0.6504 ± 0.102	0.6770 ± 0.066	0.5833 ± 0.139

The tM1 coefficient represents the proportion of the CASR signal that colocalizes with the wheat germ agglutinin (WGA) plasma membrane marker. The tM2 coefficient indicates the proportion of the WGA marker that colocalizes with the CASR signal. Values represent the mean ± SD for each tumor type.

Because CASR abundance appeared to be as high or higher than normal tissue levels in 24 of 39 tumors, the expression levels of RGS5 and RCAN1, 2 key components known to affect PTH production in the parathyroid gland, were examined. The relative expression of these 2 proteins was measured via immunofluorescence using the same scoring system described above for CASR.

RGS5 was overexpressed relative to normal tissue in 9 of 39 tumors (8 at 3+, 1 at 2+) (Fig. S7 (24) and Table 2). Three tumors expressed RGS5 at normal tissue levels, and 27 tumors had levels below that of normal tissue. The level of RGS5 expression did not correlate with CASR abundance, tumor EC50, response curve profile group, or clinical parameters, including preoperative PTH, preoperative serum calcium, age at surgery, presence or absence of nephrolithiasis, or BMD.

In contrast, elevated RCAN1 expression was preferentially found in tumors exhibiting nonsigmoid calcium/PTH response curves (P = 0.0428 by Fisher exact test) (Fig. S8 (24) and Table 2). Four of the 6 tumors with nonsigmoid response curve profiles expressed RCAN1 above normal tissue levels (3 at 3+, 1 at 2+). Among the 18 high EC50 tumors, 3 expressed increased levels of RCAN1 (all at 2+). Increased RCAN1 was found in 3 of the 11 low EC50 tumors (2 at 3+, 1 at 2+). RCAN1 expression did not correlate with RGS5 or CASR abundance and was not associated with preoperative PTH, gland weight, preoperative serum calcium, age at surgery, presence or absence of nephrolithiasis, or BMD.

In parallel, expression of the parathyroid tumor markers MEN1 (Fig. S9 (24)), Gcmb (Fig. S10 (24)), RET (Fig. S11 (24)), CDKN1B (Fig. S12 (24)), and CDC73 (Fig. S13 (24)) were evaluated in normal and adenoma tissue by immunofluorescence. None of the expression patterns of these markers correlated either individually or collectively with tumor EC50 status or with any of the clinical parameters collected in our study.

Discussion

To address the question of whether differences in tumor-intrinsic properties among PHPT patients could be an underlying driver of phenotypic variation, this study utilized a novel ex vivo, intact tissue interrogative assay to investigate the calcium-sensing properties of a series of parathyroid adenomas and related these behaviors to patient preoperative clinical parameters. The results presented here reveal that while a failure to achieve normal tissue levels of maximal PTH suppression under high calcium conditions appears to be a common characteristic shared by all of the parathyroid tumors in our study, significant diversity exists in the relative sensitivity and calcium signaling behavior of these tumors. The emergence of 3 distinct biochemical response curve profiles in the study cohort suggests that multiple underlying mechanisms may contribute to the disruption of calcium-sensing capacity in parathyroid tumors and the development of PHPT. The fact that low EC50 tumors appear to produce higher absolute amounts of PTH suggests that the underlying biochemical abnormality in these tumors could affect tonic secretion levels as opposed to calcium-sensing behavior. In contrast, the relatively low absolute levels of PTH secreted by high EC50 tumors could imply that the physiological phenotype produced by these tumors is driven primarily by a sensing deficit. The behavior of high EC50 tumors in our system is consistent with prior work using collagenase-digested dispersed parathyroid cell suspensions, where the tumor cells were found to produce less overall PTH than normal tissue while demonstrating diminished calcium responsiveness (31). Tumor-intrinsic calcium-sensing properties appear to correlate with certain features of PHPT patient clinical presentation. Tumors with attenuated calcium responsiveness (calcium EC50 > 1.10mM) were strongly associated with elevated risk of osteoporosis. Consistent with the known catabolic effects of PTH in cortical bone (2, 32), BMD deficits in the study cohort were primarily observed at cortical bone sites. Six of the 7 patients with osteoporosis and 6 of the 7 patients with osteopenia had BMD deficits detected in cortical bone. Patients whose tumors manifest an intrinsic deficit in calcium-sensing capacity could be more likely to develop chronic BMD loss due to a moderately increased but inappropriately sustained level of PTH secretion irrespective of endogenous fluctuations in serum calcium concentrations. In contrast, tumors with normal calcium responsiveness (calcium EC50 < 1.10mM) appear to secrete more PTH per mg of tissue and may have a higher tonic threshold of secretion in vivo, as patients with these tumors have a higher mean preoperative PTH. It is possible that PHPT arises in patients with this class of tumors due to an increase in basal, tonic PTH production rather than a failure in relative reactivity to changes in ambient calcium concentration. Tumor biochemical data from 2 patients with proven multi-gland disease is currently too limited to draw firm conclusions as to etiology or mechanism; in 1 case both tumors demonstrated elevated EC50 values while in the second case, only 1 of the 2 tumors were found to have an elevated EC50. Further examination of additional PHPT patients presenting with multi-gland disease will be required to determine if tumor biochemical characteristics can be correlated with clinical presentation in this subset of patients.

In their seminal work analyzing the hormonal secretory properties of bovine, murine, and human dispersed parathyroid cells, Brown et al were the first to demonstrate the existence of a calcium-sensing mechanism in these cells (33, 34) and to show that certain parathyroid tumors were deficient in this critical function (35). While primary ablation of CASR activity via experimental manipulation or in rare germline mutational kindreds has been shown to induce calcium-sensing failure and many of the hallmarks of PHPT, the great majority of human PHPT patients do not manifest germline or tumor somatic mutations in CASR (8, 36, 37). This fact along with the wide variation in clinical presentation among PHPT patients is suggestive of diverse etiological mechanisms beyond direct mutational inactivation of CASR itself.

Our data indicate that diminished CASR expression is not the sole mechanism for attenuated calcium responsiveness in parathyroid tumors causative of PHPT. CASR expression level was not predictive of the calcium EC50 biochemical phenotype, suggesting that additional factors may contribute to the uncoupling of calcium sensing from PTH secretion in PHPT adenomas.

While we observed downregulation of CASR abundance in a substantial proportion (15 of 39) of tumors, more than half of the adenomas in our cohort (24 of 39) retained CASR expression at or above the levels observed in normal tissue. These findings might appear at odds with earlier studies (38), where Kifor and coworkers reported that parathyroid adenomas and hyperplastic parathyroid tissue manifested a nearly 60% reduction in CASR immunostaining intensity when compared with tumor-adjacent normal parathyroid tissue. However, Kifor et al noted significant variability in CASR staining intensity among pathological glands, and the reference standard used for comparison (nontumor parathyroid tissue from patients with parathyroid adenomas) may not be representative of a true normo-calcemic baseline of CASR expression. Considering these factors, we feel that our observation of variability in CASR expression among PHPT adenomas is consistent with the Kifor study, and the more modest degree of CASR downregulation we observed may be attributable at least in part to the fact that our reference standard was normal parathyroid tissue from eucalcemic donors rather than tumor-adjacent parathyroid tissue where compensatory upregulation of CASR might have occurred in response to the patient’s hypercalcemic milieu, creating a greater apparent difference in CASR expression between tumor and nontumor parathyroid tissue.

In contrast to another earlier study (39), our data show that reduced CASR abundance in parathyroid tumors does not correlate with attenuated calcium sensitivity. This difference may be due to experimental design, as the PTH secretory dynamics reported here are measured across a broad range of concentrations encompassing both hypo- and hypercalcemic conditions while the earlier paper focused solely on inhibition from baseline PTH upon calcium infusion. The results reported by Cetani et al did not demonstrate a correlation between the magnitude of PTH inhibition (% suppression from pre-infusion baseline) and relative CASR abundance. Since PHPT patients by definition have elevated baseline PTH, it is possible that the Cetani study may not have been able to detect normal CASR-dependent reactivity in the context of increased tonic PTH levels. Collectively, our data suggest that changes in the activity or abundance of factors beyond CASR itself contribute to the failure of calcium sensing in PHPT.

In support of this idea, the relative expression of 2 additional key components in the CASR and PTH secretory pathways appear to be altered in a subset of parathyroid tumors. RGS5 acts in opposition to CASR downstream signaling by regulating the interaction of CASR with its obligate downstream effector G_i/o and G_q/11 proteins (6). Our group has previously reported that RGS5 can induce a hyperparathyroid phenotype when overexpressed in murine parathyroid glands (14, 15). RGS5 is elevated in 9 of 39 tumors in the study cohort, although increased RGS5 expression alone does not appear to be a determinant of diminished calcium responsiveness. This result is consistent with our earlier reports of elevated RGS5 transcripts and protein in an unselected series of parathyroid tumors (14, 15), but also suggests that additional unknown variables must act in concert with RGS5 to alter calcium signaling in parathyroid tissue. In mice, FGF23-mediated suppression of PTH release independent of CASR has been shown to require calcineurin (16), raising the possibility that RCAN1, an endogenous calcineurin inhibitor highly and selectively expressed in parathyroid tissue (40), may contribute to increased PTH secretion in parathyroid tumors. Increased RCAN1 expression has been linked to pathophysiological changes in remodeling cardiac tissue (41), neuron survival (42), degenerative vertebral disks (43) and endocrine tissues (44), and its induction is associated with cellular responses to oxidative stress (45), Alzheimer’s disease (46, 47), aging (48), and altered mineral metabolism (41, 42). Since calcineurin is an essential downstream effector of FGF23-mediated inhibition of PTH release (16), inhibition of calcineurin by RCAN1 would be expected to promote PTH secretion. We find that upregulation of RCAN1 expression is preferentially observed in a subset of tumors (6 of 39) with nonsigmoid calcium/PTH dose-response curves. The calcium response profile of these tumors suggests a subtle uncoupling of PTH secretion from calcium dose-response kinetics as opposed to a quantitative EC50 threshold shift in CASR-dependent signaling. The elevated expression of RCAN1 in tumors with this dose/response phenotype could imply a calcineurin-dependent mechanism that acts directly upon the PTH secretory apparatus downstream and independent of CASR-mediated signaling events. In contrast, the expression patterns for MEN1, Gcmb, RET, CDKN1B, and CDC73 did not correlate with tumor EC50 status or clinical presentation in our study cohort. This result is consistent with previous studies showing that variations in the mutational status, expression levels and subcellular localization of these markers among parathyroid adenomas were not clinically informative (49, 50) and suggest that markers associated with malignant transformation are not indicative of PTH secretory behavior.

While molecular events associated with malignancy do not appear to drive calcium-sensing failure, it is possible that the distinct functional subgroups we describe here may represent different points along a temporal continuum of benign tumor development in the parathyroid. For example, tumors with calcium-sensing deficiency, evidenced by an elevated EC50, could be at a later stage of tumor development compared to low/normal EC50 tumors. A longer period of chronic exposure to elevated PTH could thus contribute to the higher likelihood of BMD loss in the high EC50 cohort. Adenomas with no calcium-sensing defect could arise initially by proliferative expansion, perhaps secondary to a physiological driver such as vitamin D insufficiency (51), with a subsequent somatic mutational event that initiates a preneoplastic lesion. Progressive de-differentiation could then continue until CASR expression and function are eventually diminished, resulting in a high EC50 biochemical phenotype. Alternatively, tumors with different calcium-sensing capacities could arise through distinct, independent mechanisms. Low EC50 tumors could be initiated due to a primary defect in proliferative control, with the PHPT disease phenotype arising from increased tonic PTH secretion due to increased gland cellularity. High EC50 tumors could cause disease by impaired calcium feedback as opposed to elevated tonic PTH secretion, with or without an underlying proliferative component. Consistent with this latter model, low EC50 tumors appear to produce more PTH per mg of tumor tissue and patients with low EC50 tumors present with higher preoperative PTH levels compared with patients with high EC50 tumors. Distinguishing between these various etiological mechanisms will require future studies to resolve the genomic provenance, derivation and lineage, and underlying molecular drivers of parathyroid adenomas with different biochemical profiles and clinical manifestations.

Limitations of the current study include the relatively small study cohort combined with the sequential series experimental design and single accrual center, which may result in under-sampling of certain less common clinical or molecular subsets of PHPT patients. Important differences in the current work from previous studies include the fact that our primary functional assay relies on intact tissue rather than dispersed cells, avoiding potential artifacts of selective cellular extraction or survival bias. Heterogeneity within parathyroid adenomas is a recognized characteristic of these tumors (11), such that the use of intact tissue as opposed to isolated cells for interrogative testing is of significant benefit in reducing experimental variability.

In summary, we have found 3 distinct patterns of calcium-sensing biochemical behavior in a sequential series of parathyroid adenomas from 39 patients undergoing surgery for PHPT. Tumors with diminished calcium sensitivity are associated with increased risk of osteoporosis, while tumors with normal calcium responsiveness are found in patients with higher preoperative PTH levels. The relative abundance of CASR in these tumors does not correlate with tumor-intrinsic calcium-sensing properties or with clinical presentation, although variations in the expression of other components such as RGS5 and RCAN1 are suggestive of currently unknown, multifactorial mechanisms that can uncouple PTH secretion from ambient calcium sensing. Elucidation of these alternative mechanisms will be a major priority for understanding the underlying factors that drive phenotypic variation in patients with PHPT. The identification of functionally distinct classes of parathyroid adenomas provides direct evidence for clinically significant heterogeneity in the etiology and causative mechanisms of this common endocrine disorder.

Glossary

Abbreviations

ANOVA

analysis of variance

BMD

bone mineral density

CASR

calcium-sensing receptor

DXA

dual-energy x-ray absorptiometry

ELISA

enzyme-linked immunosorbent assay

FGF23

fibroblast growth factor 23

IRB

institutional review board

PHPT

primary hyperparathyroidism

PFA

paraformaldehyde

PTH

parathyroid hormone

RCAN1

regulator of calcineurin 1

RGS5

regulator of G-protein signaling 5

UW

University of Wisconsin cold storage solution

WGA

wheat germ agglutinin

Additional Information

Disclosures: J.A.S. is a member of the Data Monitoring Committee of the Medullary Thyroid Cancer Consortium Registry supported by GlaxoSmithKline, Novo Nordisk, Astra Zeneca, and Eli Lilly. She receives institutional research funding from Exelixis and Eli Lilly. All of the authors of this manuscript have no conflicts of interest to disclose.

Data Availability

The data supporting the findings of this study are available on request from the corresponding author. The data are not publicly available due to IRB-mandated privacy restrictions.