Circulating miR-122-5p and miR-375 as Potential Biomarkers for Bone Mass Recovery after Parathyroidectomy in Patients with Primary Hyperparathyroidism: A Proof-of-Concept Study

Primary hyperparathyroidism (PHPT) is the leading cause of secondary osteoporosis. Although bone mineral density (BMD) tends to recover after parathyroidectomy in PHPT patients, the degree of recovery varies. Circulating microRNAs (miRNAs) profiles are known to be correlated with osteoporosis and fracture. We aimed to investigate whether osteoporotic fracture-related miRNAs are associated with postoperative BMD recovery in PHPT. Here, 16 previously identified osteoporotic fracture-related miRNAs were selected. We analyzed the association between the preoperative level of each miRNA and total hip (TH) BMD change. All 12 patients (among the 18 patients enrolled) were cured of PHPT after parathyroidectomy as parathyroid hormone (PTH) and calcium levels were restored to the normal range. Preoperative miR-19b-3p, miR-122-5p, and miR-375 showed a negative association with the percent changes in TH BMD from baseline. The association remained robust for miR-122-5p and miR-375 even after adjusting for sex, age, PTH, and procollagen type 1 N-terminal propeptide levels in a multivariable model. In conclusion, preoperative circulating miR-122-5p and miR-375 levels were negatively associated with TH BMD changes after parathyroidectomy in PHPT patients. miRNAs have the potential to serve as predictive biomarkers of treatment response in PHPT patients, which merits further investigation.


Introduction
Primary hyperparathyroidism (PHPT), a leading cause of secondary osteoporosis, is characterized by hypercalcemia and bone loss owing to the overproduction of parathyroid hormone (PTH) [1,2]. In patients with PHPT, the only curative treatment is surgery [3]. Previous studies reported that the risk of fracture reduced and bone density improved in patients who underwent parathyroidectomy as opposed to patients who did not [3,4]. However, there is a large variance in bone mineral density (BMD) changes after parathyroidectomy among PHPT patients [5,6]. Sharma and colleagues demonstrated that BMD reduced after surgery in 31% of patients with PHPT [5]. PTH alone does not explain all changes in BMD after surgery [5,7]. Although alkaline phosphatase (ALP), type 1 cross-linked C-terminal telopeptide collagen (CTx), and procollagen type 1 N-terminal propeptide (P1NP) have been suggested as predictors of BMD changes after surgery in PHPT patients, these observations remain controversial [6,[8][9][10]].

Study Design
Among the 164 patients who underwent parathyroidectomy for PHPT at Severance Hospital between July 2017 and April 2019, 18 were selected and enrolled in the study. The following groups were also excluded from the study: (i) patients who were administered a drug known to affect bone metabolism, such as bisphosphonate, denosumab, calcimimetics, glucocorticoid, thiazolidinedione, thiazide, antidepressant, lithium, aromatase inhibitor, and estrogen at the time of blood collection; (ii) patients who did not undergo dualenergy X-ray absorptiometry (DXA) for BMD measurement 1 year after parathyroidectomy; (iii) patients who had samples with hemolysis ratio higher than 7 or visible hemolysis; (iv) patients who had a history of liver failure, chronic kidney disease stage 4 and above, metabolic bone disease, and other uncontrolled endocrine diseases affecting bone metabolism; (v) patients who had a family history of PHPT or multiple endocrine neoplasia 1; (vi) patients who had experienced any fractures in the last 6 months; and (vii) patients with atypical parathyroid adenoma or parathyroid carcinoma confirmed in the histopathologic report. BMD was measured using (DXA) (Discovery W, Hologic Inc., Bedford, MA, USA) at the lumbar spine (LS), femoral neck (FN), and total hip (TH) at baseline and post 1 year after parathyroidectomy. The coefficients of variation (CVs) in our institution for areal BMD at LS, FN, and TH were 1.2%, 2.1%, and 1.7%, respectively. The primary outcome of the study was the association of each osteoporotic fracture-related miRNA with the percentage change in TH BMD (%) at 1 year after surgery for the following reasons: (i) there is a greater reduction of cortical bone than cancellous bone in patients with PHPT [26][27][28][29]; (ii) TH BMD is one of the regions that improves 1 year after parathyroidectomy [30]; and (iii) CVs for TH BMD are lower than that for FN in our hospital, which is consistent with the reported literature (CV for TH BMD, 0.8-1.7%; CV for FN BMD, 1.1-2.2%) [31]. The association between each miRNA and percentage changes in LS BMD and FN BMD from baseline, 1 year after surgery, were analyzed as secondary outcomes. The serum CTx level was measured using an electrochemiluminescence immunoassay (Elecsys β-CrossLaps; Roche Diagnostics, Rotkreuz, Switzerland; intra-assay CV < 3.5%, inter-assay CV < 8.4%). PTH and P1NP levels were also measured using an electrochemiluminescence immunoassay (ECLIA; Roche Diagnostics, Rotkreuz, Switzerland; PTH intraassay CV < 2.5%, PTH inter-assay CV < 5.7%; P1NP intra-assay CV < 3.6%, inter-assay CV < 3.9%). Osteoporosis, osteopenia, and normal BMD were classified according to the International Society of Clinical Oncology guidelines [32]. The T-score was applied for postmenopausal women and men aged 50 years and older, and Z-score was applied for premenopausal women and men aged < 50 years.

Circulating miRNA Analysis
Overnight fasting blood samples were collected 0-2 days before surgery and stored at −80 • C in the Yonsei University College of Medicine. Selected osteoporotic fracture-related miRNAs were extracted and amplified from plasma according to the manufacturer's instructions and methods, as described in the literature [33,34].
The miRNeasy Serum/Plasma Kit (Qiagen) was used for purification of cell-free total RNA. Briefly, plasma samples were thawed on ice and 1 mL lysis reagent was added to 200 µL sample. Isolated RNA was used for reverse transcriptase polymerase chain reaction (RT-PCR) using the miRCURY LNA RT Kit (Qiagen). The isolated RNA was thawed and mixed with 5× miRCURY RT reaction buffer followed by 10× miRCURY RT Enzyme. To adjust each template RNA to 5 ng/µL, we used 2 µL of the eluate for 200 µL of the plasma samples. After mixing the reverse transcription master mix on ice, the samples were incubated for 60 min at 42 • C and then incubated for 5 min at 95 • C to inactivate the reverse transcriptase enzyme. After cooling to 4 • C, RT-PCR was performed. PCR conditions were: 95 • C for 2 min, 40 cycles of 95 • C for 10 s, and 56 • C for 1 min.
The 5s ribosomal RNA (5s rRNA) and miR-16 were used as housekeeping genes, and the stability of the reference gene transcripts was calculated using Delta Ct, BestKeeper, Normfinder, and Genorm algorithms [35][36][37][38]. The 5s rRNA gene was identified as a stable housekeeping gene. All analyses were based on the baseline value (2 −∆Ct ). Ct values were normalized to the internal control, and miRNA expression was presented as ∆Ct = (Ct gene of interest-Ct internal control). If the miRNA Ct was ≥40, the value was considered undetermined and was set to ∆Ct = 40. Samples with a hemolysis ratio (calculated by Ct hsa-miR-23a-3p minus Ct hsa-miR-451a ) of higher than 7 were excluded from the final analysis [33].

Statistical Analysis
Logarithmic transformation (log 2) was used for miRNA analysis. The paired-sample t-test and Wilcoxon signed-rank test were used to compare continuous variables, as appropriate. Univariate and multivariate regression analyses were performed to evaluate the association of each miRNA with BMD gain at 1 year after adjustment for age, sex, and baseline PTH level. All statistical analyses were performed using STATA 16.0 (Stata Corp LP., College Station, TX, USA). A two-sided p-value of < 0.05 was considered statistically significant.

Target Gene Predictions
Potential target mRNAs for the selected miRNAs were identified via bioinformatics using four different algorithms within the miRWalk3.0 database, namely miRWalk, TargetScan, miRDB, and miRTarBase [39].

Clinical Characteristics of Study Subjects
A total of 12 patients with PHPT (10 women, mean age 55.2 years) were included in the final analysis ( Figure 1). Of the 10 women, 7 were postmenopausal and 3 pre-menopausal. Among the two men, one was above 50 years of age and the other below 50 years of age.
According to baseline BMD, three patients had osteoporosis, four patients had osteopenia, and five patients had normal BMD (Table 1). Each patient underwent surgical treatment for the following reasons. Four patients were <50 years old, seven patients had serum calcium levels > 1 mg/dL compared to the normal range, two patients had estimated glomerular filtration rate (eGFR) < 60, one patient had nephrolithiasis, and five patients had a T-score of −2.5 or less. Among the five patients with a T-score of −2.5 or less, two had a forearm T-score of −2.5 or less. One patient was provided surgical treatment because she wanted surgical treatment, and bone loss was progressed. All patients were successfully cured of PHPT as PTH and corrected Ca levels improved to normal ranges within 1-3 days after surgery (median PTH, 109.7 to 40.6 pg/mL, p < 0.001; median corrected calcium, 11.0 to 9.1 mg/dL, p = 0.001). P1NP, CTx, and 25-OH-vitamin D were also significantly recovered 2 weeks after surgery (median P1NP, 72.5 to 91.8 ng/mL, p = 0.003; median CTx, 0.665 to 0.390 ng/mL, p = 0.002; mean 25-OH-vitamin D, 19 to 27 ng/mL, p = 0.001). Serum calcium levels were maintained in the normal range for more than 6 months after surgery, and we decided that PHPT patients were cured.

Clinical Characteristics of Study Subjects
A total of 12 patients with PHPT (10 women, mean age 55.2 years) were included in the final analysis ( Figure 1). Of the 10 women, 7 were postmenopausal and 3 premenopausal. Among the two men, one was above 50 years of age and the other below 50 years of age. According to baseline BMD, three patients had osteoporosis, four patients had osteopenia, and five patients had normal BMD (Table 1). Each patient underwent surgical treatment for the following reasons. Four patients were < 50 years old, seven patients had serum calcium levels > 1 mg/dL compared to the normal range, two patients had estimated glomerular filtration rate (eGFR) < 60, one patient had nephrolithiasis, and five patients had a T-score of −2.5 or less. Among the five patients with a T-score of −2.5 or less, two had a forearm T-score of −2.5 or less. One patient was provided surgical treatment because she wanted surgical treatment, and bone loss was progressed. All patients were successfully cured of PHPT as PTH and corrected Ca levels improved to normal ranges within 1-3 days after surgery (median PTH, 109.7 to 40.6 pg/mL, p < 0.001; median corrected calcium, 11.0 to 9.1 mg/dL, p = 0.001). P1NP, CTx, and 25-OH-vitamin D were also significantly recovered 2 weeks after surgery (median P1NP, 72.5 to 91.8 ng/mL, p = 0.003; median CTx, 0.665 to 0.390 ng/mL, p = 0.002; mean 25-OH-vitamin D, 19 to 27 ng/mL, p = 0.001). Serum calcium levels were maintained in the normal range for more than 6 months after surgery, and we decided that PHPT patients were cured.  Data are presented as frequency (%), mean ± standard deviation, and median (interquartile ranges). Abbreviations: P1NP, procollagen type 1 N-terminal propeptide; CTX, C-terminal

Correlation between miRNAs and Study Covariates
The relationships between age, PTH, CTx, P1NP, and each miRNA are shown as a heat map (Supplementary Figure S1). Although none of the miRNAs were significantly associated with age, baseline PTH, CTx, and P1NP (p > 0.05), there was a high correlation between most of the miRNAs, indicating that the selected miRNAs in this study were indeed involved in bone metabolism. All miRNAs, except miR-122-5p and miR-124-3p, showed statistically significant associations with each other (p < 0.05).

Higher Preoperative Circulating miR-122-5p and miR-375 Expression Levels Were Associated with Slow Recovery of TH BMD in Patients with PHPT
The association between each miRNA and postoperative TH BMD percent change from baseline, 1 year after parathyroidectomy, is described in Table 2. In univariable regression, miR-19b-3p (β coefficient = −1.55, p = 0.038), miR-122-5p (β coefficient = −1.61, p = 0.034), and miR-375 (β coefficient = −2.12, p = 0.002) were negatively associated with change in TH BMD. After adjustment for age, sex, PTH, and P1NP (Table 3, model III), the association among miR-122-5p, miR-375, and TH BMD changes was significant, whereas no association was observed between miR-19b-3p and TH BMD changes. According to the pre-specified panel, miR-122-5p and miR-375 commonly targeted runt-related transcription factor 2 (RUNX2) in the predicted target model (Supplementary Materials Table S1). There was no association between each miRNA and LS BMD or FN BMD changes after surgery in patients with PHPT (Supplementary Table S2).

Patients with High miR-122-5p Levels Had Less CTx Change after Parathyroidectomy
In patients with higher preoperative miR-122-5p levels, the postoperative CTx change (%) from baseline, 2 weeks after surgery, was smaller (p = 0.009). There was no difference in the postoperative P1NP change (%) from baseline with respect to miR-122-5p levels (p = 0.310). There was no difference in CTx and P1NP changes after surgery with respect to miR-375 levels (p > 0.05) (Figure 2).

Discussion
In this proof-of-concept study, we investigated whether osteoporotic fracture-related miRNAs were associated with the degree of postoperative BMD changes in patients with PHPT. Previous studies reported potential markers associated with BMD changes after parathyroidectomy in PHPT patients, but the results are inconsistent among studies.

Association between Known Predictors of BMD Response and Postoperative BMD Changes in Patients with PHPT
Predictive markers related to BMD change after parathyroidectomy in patients with PHPT, such as corrected calcium, bone turnover markers (CTx, P1NP), PTH, and age were used in the analysis. Upon conducting the Spearman correlation analysis, corrected calcium was associated with postoperative TH BMD changes (ρ = 0.608, p = 0.036). CTx and P1NP were associated with postoperative FN BMD changes (CTx, ρ = 0.587, p = 0.045; P1NP, ρ = 0.587, p = 0.045). PTH, age, and sex were not significantly correlated with postoperative BMD changes (p > 0.05).

Discussion
In this proof-of-concept study, we investigated whether osteoporotic fracture-related miRNAs were associated with the degree of postoperative BMD changes in patients with PHPT. Previous studies reported potential markers associated with BMD changes after parathyroidectomy in PHPT patients, but the results are inconsistent among studies. Lumachi et al. suggested that premenopausal women had a greater LS BMD (L2-4) gain than postmenopausal women [40]. Koumakis et al. reported a correlation between ALP and BMD improvement [8]. Alonso et al. reported that PTH, CTx, and P1NP levels were associated with changes in LS BMD in simple regression analysis in 53 PHPT patients [9]. By contrast, Steinl et al. suggested that preoperative serum calcium levels, total parathyroid weight on computed tomography, sestamibi avidity, and time from pre-to post-operative DXA are associated with LS BMD improvement after surgery, but not PTH levels [41]. Owing to these controversies, we evaluated whether miRNAs can be used as new markers for postoperative BMD recovery in PHPT patients.
Among the selected osteoporotic fracture-related miRNAs, we found that high miR-122-5p and miR-375 levels were associated with lower postoperative TH BMD changes independent of age, sex, PTH, and P1NP. This is consistent with recent clinical studies that have revealed that miR-122-5p and miR-375 are associated with bone deterioration. miR-122-5p was upregulated in 15 postmenopausal women with hip fracture compared to 12 postmenopausal women with osteoarthritis [54], and was also upregulated in 45 osteoporotic patients with fracture compared to 15 non-osteoporotic patients with fracture [55]. In a study of 30 osteoporosis and 30 non-osteoporosis patients, miR-122a had the highest risk of osteoporosis, with the area under the curve of 0.77 among miR-21, miR-23a, miR-24, miR-93, miR-100, miR-122a, miR-124a, miR-125b, and miR-148a [24]. Meanwhile, Mandourah et al. reported that miR-122-5p is downregulated in osteoporosis (with or without fracture) in a study of 139 patients older than 40 years, which is inconsistent with our study [56]. This difference may result from the inclusion of a small number of patients with fracture or steroid exposure 2 years to 1 month prior to the time of study registration in the study of Mandourah et al. [56]. miR-375 was upregulated in 26 patients with low BMD and vertebral fractures compared to 42 healthy postmenopausal women [33]. In a study of 50 osteoporotic and 49 non-osteoporotic women, high levels of miR-375 were associated with a high risk of osteoporosis [57].
Interestingly, in this study, PHPT patients with high miR-122-5p levels had significantly less change in CTx 2 weeks after surgery. Patients with high miR-122-5p levels had relatively high CTx levels within two weeks of surgery, adversely affecting BMD recovery. CTx is cleaved by osteoclasts during bone resorption, which suggests a link between miR-122-5p and osteoclasts [63]. Kelch et al. reported that miR-122-5p is upregulated in osteoporotic osteoclasts during osteoclast differentiation compared to non-osteoporotic cells [64], which is in close agreement with our findings. However, the direct effect of miR-122-5p on osteoclastogenesis has not yet been documented. In addition, it may be an effect of RUNX2 degradation. RUNX2 has been suggested to be a major modulator of osteoblast differentiation, and, recently, Xin et al. reported that this modulation occurs via the AKT/NFATc1/CTSK axis [65]. MiR-122-5p may reduce both osteoblast and osteoclast differentiation through RUNX2 degradation, although there was no difference in P1NP change with respect to the miR-122-5p levels in the present study. In fact, P1NP change 2 weeks after surgery may reflect the increased procollagen type I synthesis as the PTH decreases. In PHPT patients, continuous PTH elevation increases the osteoblastic population and osteoblasts synthesize procollagen type I, while PTH plays a role in preventing the synthesis of procollagen type 1 in osteoblast itself [66,67]. Therefore, immediately after surgical treatment in PHPT patients, the synthesis of procollagen type 1 is temporarily increased with increased osteoblastic population and decreased action of PTH [66,67]. Consistent with this, in our study, P1NP was increased 2 weeks after surgery.
There are several limitations to this study. The miRNAs examined here were related to TH BMD changes but not to LS BMD or FN BMD changes. The number of patients in this study was small owing to the rarity of PHPT. Although PHPT is one of the main causes of secondary osteoporosis, it has an incidence of approximately 80 out of 100,000 (0.08%) in Asians [68]. Despite the small study population, however, the characteristics of the study population were similar to the general characteristics of PHPT, with a ratio of female-to-male of 5:1, and the mean age of 55 years [69]. In addition, the existing markers associated with BMD changes after surgery, such as calcium, CTx, and P1NP [9,41], were also associated with BMD changes in the present study. There was no control group, and miRNAs investigated in the present study were selected by literature review, increasing the chances of false-positive target identification. The mechanisms by which miR-122-5p and miR-375 adversely affect postoperative BMD recovery in PHPT patients are yet to be confirmed by a functional study.
In conclusion, preoperative circulating miR-122-5p and miR-375 levels were negatively associated with TH BMD change after parathyroidectomy in patients with primary hyperparathyroidism. Our findings suggest that miRNAs have a potential to facilitate individualized treatment of PHPT. This study was a preliminary proof-of-concept study that requires additional validation.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/diagnostics11091704/s1, Figure S1: Pearson correlation heatmap of selected osteoporotic fracture-related miRNAs and study covariates. Table S1: The pre-specified panel of selected miRNAs and target genes related to osteoporosis and primary hyperparathyroidism. Table S2: Univariable regression (β coefficient) for the association of log-transformed preoperative miRNAs (−∆Ct) and lumbar spine (LS) and femoral neck (FN) bone mineral density (BMD) percent change from baseline (%), 1 year after parathyroidectomy, in PHPT patients. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.