Role of Runx2 polymorphisms in risk and prognosis of ossification of posterior longitudinal ligament

Abstract

Background

Our study was aimed at finding out if Runx2 SNPs (single‐nucleotide polymorphisms) are related to susceptibility to and prognosis of ossification of posterior longitudinal ligament (OPLL).

Methods

We selected 80 OPLL patients and another 80 independent patients without OPLL from September 2013 to November 2014. Serum was collected to detect the genotypes of rs1321075, rs12333172, and rs1406846 on Runx2 with direct sequencing analysis.

Results

Differences in clinical characteristics, including age, weight, height, sex ratio, as well as smoking and drinking history, between OPLL and control groups appeared to be insignificant (all P‐value >.05). The allele of rs1406846 (A) emerged as a key element in raising OPLL risk with the biggest statistical significance (P<.001). Conversely, alleles of rs967588 (T) and rs16873379 (C) were associated with reduced predisposition to OPLL less remarkably (both P=.033). Regarding rs16873379, the case group exhibited a smaller frequency of homozygote CC in comparison with TT genotype than the control group (P=.016). Furthermore, the improvement rate based on calculation of JOA score suggested that genotype AA of rs6908650 was beneficial for OPLL patients' recovery from posterior laminoplasty surgery (P<.05), while genotypes of rs16873379 (CC), rs1406846 (AA), and rs2677108 (CC) significantly restrained this process (P<.05). Besides, rs16873379, rs1406846, and rs2677108 were significantly associated with number of ossification segments (P<.05).

Conclusions

Runx2 SNPs (e.g., rs16873379, rs1406846, and rs2677108) were strongly correlated with onset and treatment efficacy of OPLL, and they might regulate severity of OPLL.

1. INTRODUCTION

Ossification of posterior longitudinal ligament (OPLL) is classified as a subset of bone‐forming disease that could result in radiculopathy and myelopathy particularly for the elderly.1 It is a commonly observed disorder in Asia with the morbidity achieving 2%–4% in Japan, where males are twice as likely as females to suffer from OPLL.2 The prevalence of OPLL in Europe and the United States also ranges between 0.01% and 1.7%, though a bit lower than that in Asia.3 Since OPLL patients' spinal cords and nerve roots are compressed by ossified ligaments, severe neurological dysfunctions (e.g., myelopathy) may appear.4 Nonetheless, conventional treatments make it hard to alleviate the functional disorder for that myelopathy progresses in a chronic but hidden manner.3 Therefore, seeking for novel and efficacious treatments enabled us to continuously unveil key risk factors of OPLL, yet, little evidence has been released. As suggested by twin and pedigree studies, hereditary factors or genetic polymorphisms (e.g., COL6A1 promoter‐572T) seemed to impose significant impacts on OPLL, suggesting that identification of susceptible genes may lead us to discover early interventions to effectively impede OPLL progression.5, 6

Previous researches reported that the expression of Runx2 (Runt‐related transcription factor 2) was enhanced in spinal ligament cells of OPLL patients.7 Besides, it was speculated that Runx2 may affect progression of heterotopic ossification as silencing of Runx2 could inhibit the expression level of angiopoietin‐1 in cultured OPLL cells.8 Two years later, Runx2 was confirmed to inhibit the formation of heterotopic ossification in an independent manner.9 Consequently, we suspected that Runx2 might be closely associated with the morbidity of OPLL.

As a transcription factor, Runx2 appears to manipulate an intricate gene‐regulator network during the process of osteoblastogenesis.10 To be specific, numerous osteoblast lineage‐specific genes, such as Ocn (osteocalcin), Osx (osterix), and Bsp (bone sialoprotein), could be up‐regulated by Runx2, while the expression of non‐osteoblast genes, including MyoD (myogenic differentiation) and PPAR‐γ (peroxisome proliferator‐activated receptor gamma), may be down‐regulated by Runx2.10 Moreover, Runx2 has been hypothesized to control gene expressions through its unique interaction with various co‐regulatory factors,11 nevertheless, how this interaction goes and what the functional polymorphisms of Runx2 are still remain challenging.10

Hence, this retrospective case–control study enabled us to find out the functional polymorphisms of Runx2 that were crucial to the occurrence of OPLL, which could provide evidence on discovering early interventions for OPLL.

2. MATERIALS AND METHODS

2.1. Subjects

We selected 80 patients (case group) who were diagnosed as OPLL between September 2013 and November 2014 in Affiliated Shanxi Provincial People's Hospital, Shanxi Medical University. All patients in the case group were diagnosed using both clinical symptoms and imaging. Computed tomography (CT) and magnetic resonance imaging (MRI) were introduced to judge compression of spinal cord, morphology of spinal canalis, as well as size and range of ossifications. On the other hand, the control group included 80 unrelated patients who were confirmed as OPLL negative and they were admitted in our hospital during the same period. The baseline clinical characteristics are enlisted in Table 1. Individuals in the two groups were all of the Han nationality. Patients in the case group would be accepted if they: (1) had no history of spine‐relevant surgery; (2) owned a ossification scope that was confined to C_3~7 scope, and was involved with two segments; (3) have a >50% occupational ratio of spinal canal. Furthermore, patients with ankylosing spondylitis, diffuse idiopathic bone hypertrophy, spinal infection, and tumor were excluded. Patients who had taken drugs were also excluded since these drugs may have impact on bone metabolism. This study was approved by the Ethical Committee of Affiliated Shanxi Provincial People's Hospital, Shanxi Medical University. Each participant has signed the informed consent and they agreed to participate in this study.

Table 1.

The PCR primer sequences for seven SNPs on RUNX2

SNP	Region	Location	Gene sequence	Primer sequence
rs967588 C>T	Intron	4.5E+07	TATAAAACAT[C/T]CAAGTATGCA	Sense: TTTGGGTTCAGCTATACTATTTCGT
				Anti‐sense: AATCTCATGCATACTTGGATGTTTT
rs16873379 T>C	Intron	4.5E+07	TATCTGACTA[C/T]GCCAAACCAC	Sense: TTAATATCTGACTATGCCAAACCAC
				Anti‐sense: AGCCTATAATTTAACTCCAGAGCC
rs3749863 A>C	Intron	4.5E+07	CACCCGCCAA[A/C]AGGGAGGAGT	Sense: GCACAGCCCAGCCCC
				Anti‐sense: TTGCCTCTCGACTCCTCCC
rs6908650 G>A	Intron	4.5E+07	CGTCTTCCTT[A/G]GCAGTACCCC	Sense: CGAACTCCACTCTAAGAGGTTC
				Anti‐sense: AGAAGAAAAACAGGTTTCCAGGTTG
rs1321075 A>C	Intron	4.5E+07	AGATCCTAAA[A/C]CCTCAACCTG	Sense: TCCGCAGGTTGAGGTTTTAG
				Anti‐sense: AATCTGGATCGGGTTATGGAGAG
rs1406846 T>A	Intron	4.5E+07	TGTGAGCCTC[A/T]TTTTATTTCA	Sense: ATTCATGCTAATGAGATACTGTCAC
				Anti‐sense: ATTCTACCGGAGGCTGTGAG
rs2677108 C>T	Intron	4.5E+07	AGTCTTAACG[C/T]TTATAATAAA	Sense: ACTAGTTTAGACTCCTGGCTGG
				Anti‐sense: AATTGGTTAGCAAAGAGAGGCT

PCR, polymerase chain reaction; SNP, single‐nucleotide polymorphism; RUNX2, Runt‐related transcription factor 2.

2.2. Operation of laminoplasty and post‐operative treatment

After general anesthesia, patients accepting laminoplasty routinely exposed their C_2~7 vertebral plate and lateral mass. The side assuming serious physical symptoms was chosen as the opened side. Muscles that attached to C₂ spinous process were reserved, and the opening range remained C_3~7 vertebral plate with the angle ranging from 45° to 60°. The contracture tract outside the dura mater was loosen to the degree that dural sac achieved distention and pulsation. The vertebral plate was turned on, and it was fixed on joint capsule for routine indwelling drainage.

All patients were post‐operatively given antibiotics and hormones routinely through intravenous injection. The drainage tubes would be removed when the drainage liquid for patients without leakage of cerebrospinal was <50 mL/day. Besides, the patients were kept in the dorsal elevated position, and their wounds were appropriately treated with pressure. If necessary, subdural puncture was applied for catheter drainage. Finally, patients receiving laminoplasty should wear neck circumstances for 6 weeks after operation.

2.3. Efficacy evaluations

All OPLL patients were followed up for 12 months, and the clinical efficacy of laminoplasty on them was evaluated according to Japanese Orthopaedic Association (JOA) score. The scores of vertebrate JOA peaked at 17 and bottomed at 0, and lower scores were representative of more evident functional disorders. The recovery rate after operation was calculated in light of the following formula:

$$\frac{\text{Post‐operative JOA score}-\text{Pre‐operative JOA score}}{17-\text{Pre‐operative JOA score}}\times 100.$$

2.4. Collection of genomic DNA from peripheral blood

We collected 3 mL of peripheral venous blood from all participants before they had breakfast in the morning. Then, all blood samples were kept in heparin anticoagulant tubes in which serum was separated once samples were properly mixed. After that, the blood samples were stored in an ultra‐low‐temperature (−70°C) freezer. Then, we added 10 mL of EDTA anticoagulant into each peripheral blood sample obtained from patients in the two groups and all samples were maintained in a sterile condition. The extraction and purification of genomic DNA from peripheral blood was implemented by DNA purification kit (Qiagen Company in Germany).

2.5. Genotyping of three SNPs in Runx2

Gene sequences of RUNX2 were searched via NCBI Gene, and were also located with NCBI SNP. With assistance of NCBI Primer‐BLAST, primer sequences for proliferation of SNPs were designed, and the primers featured by fine specificity and hard to dimerization. After primers were designed (Table 1), the 50 μL PCR reaction system adopted to proliferate DNA fragment was specifically described as follows: 10×LA PCR buffer (5 μL), 25 mmol/L Mg²⁺ (5 μL), 25 mmol/L dNTPs (8 μL), 10 mol/L upstream primer (1 μL), 10 mol/L forward primer (1 μL), 10 mol/L reverse primer (1 μL), 5U/μL Taq enzyme (0.5 μL), and DNA template (2.5 μL). The particularized amplification condition seemed to be the combination of denaturation (95°C for 5 min), 35 cycles of annealing (98°C for 10 s, 55°C for 5 s, and 72°C for 30 s), and extension (72°C for 5 min).

Direct sequencing analysis was performed for the above purified PCR products on an ABI3130xl genetic analyzer (Applied Biosystems, Waltham, MA, USA). The sequence diagram is shown in Figure 1A–G, while analysis of linkage disequilibrium (LD) was conducted with assistance of Hapmap (http://www.hapmap.org), and the results are indicated in Figure 1H. Besides, the four SNPs (rs967588, rs16873379, rs3749863, and rs6908650) in Block 1 were in high LD, and another three SNPs belonged to Block 2 (rs13212075, rs1406846, and rs2677108) were in high LD.

Figure 1.

Direct sequencing results of A, rs967588; B, rs16873379; C, rs3749863; D, rs6908650; E, rs1321075; F, rs1406846; G, rs2677108, as well as LD H, results of the above SNPs. LD, linkage disequilibrium; SNP, single‐nucleotide polymorphism.

2.6. Statistical analysis

Data analysis was performed by SPSS19.0 statistical software (New York, USA). Measurement data were expressed in the form of mean ± standard deviation. Genotype and allelic frequency were obtained using frequency count and Hardy‐Weinberg equilibrium analysis. Comparison of genotype distribution and allele frequency were achieved using the chi‐square test. The LD analysis was implemented with application of Haploview software (Boston, MA, USA). The associations between various factors and ossification of cervical vertebra longitudinal ligament were assessed by logistic regression.

3. RESULTS

3.1. Baseline characteristics of patients in the case and control group

Clinical and demographic characteristics of patients were revealed and compared in Table 2. The case group included a total of 80 OPLL patients (45 males, 35 females). The average age, height, and weight of patients in the case group were 49.0±9.4 years old, 164.7±9.6 cm, and 59.6±8.6 kg, respectively. The control group had an equivalent sample size of OPLL‐negative individuals (49 males, 31 females) with an average age, height, and weight of 46.3±7.7 years old, 162.7±5.1 cm, and 58.4±6.9 kg, respectively. Moreover, the case group included a total of 39 smokers and another 24 patients who usually consumed alcohols. Another 33 smokers and 32 individuals with a history of alcohol consumption were included in the control group. The Mann‐Whitney test indicated that no significant difference in age, gender, and height existed between case and control groups (P>.05). It was also drawn from chi‐square test that case and control groups have similar sex ratio, as well as frequencies of smoking and drinking history (P>.05). Moreover, OPLL patients' JOA recovery rate was averaged (59.7±8.5)%, and their mean ossification segment number was 1.9±1.2.

Table 2.

The main characteristics of subjects in OPLL and control groups

Group	OPLL (n=80)	Control (n=80)	P‐value
Mean age (years old)	49.0±9.4	46.3±7.7	.286a
Mean weight (kg)	59.6±8.6	58.4±6.9	.101a
Mean height (cm)	164.7±9.6	162.7±5.1	.173a
Gender (%)b
Male	45 (56.25)	49 (61.25)	.521
Female	35 (43.75)	31 (38.75)
Smoking history (%)b
Yes	39 (48.75)	33 (41.25)	.340
No	41 (51.25)	47 (58.75)
Drinking history (%)b
Yes	24 (30.00)	32 (40.00)	.185
No	56 (70.00)	48 (60.00)
JOA scorec
Before surgery	8.6±1.9	‐
After surgery	13.7±2.2	‐
Follow‐up time (months)	12	‐
JOA improvement rate (%)	59.7±8.5
Ossification segment number	1.9±1.2	‐

OPLL, ossification of posterior longitudinal; JOA, Japanese Orthopaedic Association.

^aMann‐Whitney test.

^bChi‐square test.

^cAll patients received posterior laminoplasty surgery.

3.2. Genotype distribution and allele frequency of Runx2 gene loci in the case and control group

The genotyping and allelic frequencies of SNPs that were situated in Runx2 are listed in Table S1. In the allelic model (Table 3), rs1406846 (T>A) was most prominently associated with elevated OPLL risk (OR=5.67, 95% CI: 2.71‐11.85, P<.001), while rs967588 (C>T) and rs16873379 (T>C) were correlated with declined susceptibility to OPLL, respectively (OR=0.47, 95% CI: 0.23‐0.94, P=.033; OR=0.48, 95% CI: 0.25‐0.94, P=.033). Besides, the mutated homozygotes of rs16873379, rs1321075, and rs1406846 appeared to confer subject‐regulated possibility of OPLL risk than their wide homozygotes (CC vs TT, OR=0.06, 95% CI: 0.01‐0.58, P=.016; AA vs CC, OR=0.06, 95% CI: 0.01‐0.58, P=.016; AA vs TT, OR=3.38, 95% CI: 1.34‐8.54, P=.010). Notably, the crucial role of rs1406846 in predisposing to OPLL was also enhanced by three additional models, namely dominant model (AA+AT vs TT, OR=3.64, 95% CI: 1.64‐8.05, P=.001), heterozygote model (AT vs TT, OR=3.20, 95% CI: 1.25‐8.18, P=.015), and recessive model (AA vs TT+AT, OR=1.07, 95% CI: 1.01‐1.13, P=.024).

Table 3.

Association between RUNX2 SNPs and OPLL risk under five genetic models

LD block	SNP	Group	Allelic model		Dominant model		Homozygote model		Heterozygote model		Recessive model
			OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Block 1	rs967588 C>T	OPLL	0.47 (0.23‐0.94)	.033	0.49 (0.23‐1.05)	.068	1.06 (0.99‐1.12)	.084	0.53 (0.22‐1.27)	.152	1.02 (0.98‐1.07)	.359
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
	rs16873379 T>C	OPLL	0.48 (0.25‐0.94)	.033	0.77 (0.36‐1.65)	.501	0.06 (0.01‐0.58)	.016	1.06 (0.44‐2.56)	.899	1.13 (0.53‐2.41)	.753
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
	rs3749863 A>C	OPLL	0.64 (0.34‐1.21)	.171	0.59 (0.26‐1.35)	.208	0.60 (0.12‐2.89)	.522	0.42 (0.15‐1.14)	.088	0.83 (0.23‐3.05)	.777
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
	rs6908650 G>A	OPLL	0.99 (0.58‐1.68)	.959	1.35 (0.63‐2.92)	.442	1.51 (0.42‐5.42)	.526	1.92 (0.73‐5.07)	.186	0.99 (0.36‐2.66)	.977
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
Block 2	rs1321075 C>A	OPLL	2.22 (1.00‐4.92)	.050	1.83 (0.78‐4.27)	.163	5.93 (1.37‐9.811)	.034	0.94 (0.34‐2.61)	.905	0.87 (0.40‐1.88)	.725
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
	rs1406846 T>A	OPLL	5.67 (2.71‐11.85)	<.001	3.64 (1.64‐8.05)	.001	3.38 (1.34‐8.54)	.010	3.20 (1.25‐8.18)	.015	1.07 (1.01‐1.13)	.024
		Control	Ref.		Ref.		Ref.		Ref.		Ref.
	rs2677108 T>C	OPLL	1.45 (0.73‐2.88)	.295	1.86 (0.85‐4.07)	.122	2.70 (0.29‐5.33)	.384	1.33 (0.53‐3.33)	.541	0.95 (0.11‐8.00)	.959
		Control	Ref.		Ref.		Ref.		Ref.		Ref.

RUNX2: Runt‐related transcription factor 2; SNP, single‐nucleotide polymorphism; LD, linkage disequilibrium; OR, odds ratio; CI, confidence interval.

3.3. Association between Runx2 SNPs and outcomes of OPLL patients after posterior laminoplasty surgery

It was observed in Table 4 that mutated homozygote of rs6908650 (AA) might be beneficial to treatment efficacy of OPLL with a significantly incremental JOA score (15.0±2.5 vs 12.9±2.1) when compared with its wide type (GG) (P<.05). However, post‐operative JOA scores concerning mutated genotypes of rs1321075 (AA), rs1406846 (AA), and rs2677108 (CC) dropped about 16.33%, 35.98%, and 24.84% in comparison with their according wide homozygotes (all P<.05). Besides, the mutation alleles of rs1406846 (A) and rs2677108 (C) were associated with the increased OPLL risk because that their heterozygote carriers had significantly poorer JOA scores than their wide homozygote carriers (TA vs TT: 13.4±1.1 vs 16.4±1.1; TC vs TT: 13.1±1.4 vs 15.3±1.8).

Table 4.

Association between RUNX2 SNPs and outcome of OPLL patients (e.g., JOA score and improvement rate) after posterior laminoplasty surgery

SNP (w>m)	Genotype	JOA score		Improvement rate (%)
		Before surgery	After surgery
rs967588 C>T	CC	8.5±1.8	14.5±2.4	60.7±9.1
	CT	8.7±2.0	12.9±1.7	59.6±6.7
	TT	9.0±2.1	11.8±1.5	53.5±6.0
rs16873379 T>C	TT	8.7±2.0	14.5±2.5	61.7±9.0
	TC	8.4±1.8	13.5±1.6	58.5±6.4
	CC	9.0±2.2	12.8±2.1	52.4±6.1a
rs3749863 A>C	AA	8.3±1.9	14.8±2.4	62.1±9.7
	AC	8.6±1.9	13.5±1.8	59.0±6.7
	CC	9.3±1.9	12.5±2.8	57.1±7.9
rs6908650 G>A	GG	8.7±2.1	12.9±2.1	57.4±6.3
	GA	8.5±1.7	14.0±2.2	59.0±8.1
	AA	8.6±2.2	15.0±2.5a	66.8±7.7a
rs1321075 C>A	CC	8.6±1.8	14.7±2.1	61.1±7.2
	CA	8.9±1.9	12.1±1.3b	57.2±10.1
	AA	7.3±2.4	12.3±2.8a	57.9±6.6
rs1406846 T>A	TT	8.5±2.0	16.4±1.1	66.0±7.0
	TA	8.7±1.8	13.4±1.1a	57.7±7.5
	AA	8.3±2.1	10.5±1.0a, c	54.7±4.7a, c
rs2677108 T>C	TT	8.5±2.0	15.3±1.8	62.5±8.3
	TC	8.9±1.8	13.1±1.4b	58.6±7.8
	CC	7.5±1.3	11.5±2.9a, c	54.5±5.1a, c

RUNX2: Runt‐related transcription factor 2; SNP, single‐nucleotide polymorphism; JOA score, Japanese Orthopaedic Association Score; w: wide allele; m: mutant allele.

^amm vs ww, P<.05.

^bwm vs ww, P<.05.

^cmm vs wm, P<.05.

As for the improvement rate that synthetically compared pre‐operative and post‐operative JOA scores, it generally remained high between (52.4±6.1)% and (66.8±7.7)%. Furthermore, it was suggested that genotype AA of rs6908650 might largely aid in the fine treatment efficacy of OPLL (P<.05), while homozygotes CC of rs16873379, AA of rs1406846, and CC of rs2677108 were tightly linked with unfavorable prognosis of OPLL (P<.05).

3.4. Association between Runx2 SNPs and ossification segment number of OPLL patients

The chi‐square test (Table 5) indicated that the ossification segment number of OPLL patients varied notably with changes of genotypes of rs16873379, rs1406846, and rs2677108, implying that the three SNPs might regulate severity of OPLL with statistical significance (P<.05). In contrast, the remaining four SNPs (i.e., rs967588, rs3749863, rs6908650, and rs1321075) showed hardly significant associations with number of ossification segments (P>.05).

Table 5.

Association between RUNX2 SNPs and ossification segment number of OPLL patients

SNP	Genotype	Ossification segment number						ANOVA
		1	2	3	4	5	6	Chi‐square	P‐value
rs967588 C>T	CC	21	16	2	1	1	1	6.363	.784
	CT	14	11	6	0	1	1
	TT	1	3	1	0	0	0
rs16873379 T>C	TT	23	18	0	0	0	1	30.946	.001
	TC	12	10	8	0	2	1
	CC	1	2	1	1	0	0
rs3749863 A>C	AA	15	11	0	0	0	1	15.098	.129
	AC	16	15	8	0	2	1
	CC	5	4	1	1	0	0
rs6908650 G>A	GG	9	10	2	1	0	1	6.028	.813
	GA	20	15	6	0	2	1
	AA	7	5	1	0	0	0
rs1321075 C>A	CC	26	18	6	1	1	1	8.189	.610
	CA	7	11	1	0	1	1
	AA	3	1	2	0	0	0
rs1406846 T>A	TT	18	7	0	0	0	0	21.231	.020
	TA	17	14	6	1	2	1
	AA	1	9	3	0	0	1
rs2677108 T>C	TT	22	10	2	0	0	0	18.736	.044
	TC	12	14	7	1	2	1
	CC	2	6	0	0	0	1

RUNX2: Runt‐related transcription factor 2; SNP: single‐nucleotide polymorphism; OPLL, ossification of posterior longitudinal ligament; ANOVA: analysis of variance.

4. DISCUSSION

OPLL, a prevailing disease in Asia, is characterized by abnormal ectopic calcification of ligaments that usually occurred in the cervical spine region.12, 13, 14, 15 Previous studies have suggested that OPLL was classified as a multifactorial disease that may be induced by several genetic and environmental factors, including HLA antigens, hormonal imbalance, infection, diabetes, history of trauma as well as dietary habits.12, 16, 17, 18, 19, 20 It has been reported that OPLL was related to several genes, including FF variant of vitamin D receptor,13 BMP4 (bone morphogenetic protein 4),21 TGF3 (transforming growth factor 3),16 and human retinoic X receptor β.22 Moreover, a number of case–control studies based on different populations provided evidence that OPLL was associated with several SNPs on genes including BMP4 in Chinese population,23 IL15RA (interleukin 15 receptor alpha) in Korean patients,24 TGF‐β1 (transforming growth factor‐β1) in Japanese,25 and COL6A1 (collagen 6A1) in Chinese Han population.6 Nevertheless, key causative genetic mutations that are associated with OPLL have not been identified.16 Therefore, identifying key genetic mutations or SNPs is an informative approach for early interventions.

As a transcription factor, Runx2 is responsible for osteoblast differentiation and it is involved in both bone formation and skeletal development.26, 27, 28, 29 Particularly, a study based on establishment of mice models suggested that the formation of mineralized skeleton could not be completed without assistance of Runx2.30 Phenotypes, including osteoblast deficiency, lamellar bone and marrow cavities, were also observed when Runx2 was knocked out. Moreover, Runx2 deficiency has also been confirmed to make human beings susceptible to cleidocranial dysplasia.31 For in‐depth consideration, Runx2 was deemed to induce conversion of mesenchymal cells into osteoblasts, and simultaneously control their differentiations into either chondrocytes or adipocytes.32

Based on the understanding about the significance of Runx2 in normal formation of osteoblasts, investigations related to osteoblast‐relevant disorders (e.g., OPLL) demonstrated that haploinsufficiency of Runx2 could impede development of OPLL in ENPP1^ttw/ttw mice, implying that normal Runx2 expressions may be essential to trigger OPLL.33 In addition, research findings of Kishiya and his colleagues indicated that Runx2 might play a regulatory role in ectopic calcification through regulation of a downstream effector (i.e., angiopoietin‐1), giving a hint that Runx2 could act as a critical regulator for the development and progression of OPLL.7 However, previous researches were still insufficient to explain how Runx2 is linked with OPLL through its physiological and pathological roles. In this study, we discovered that mutations of certain functional polymorphisms (rs967588, rs16873379, rs1321075, and rs1406846) situated in Runx2 would modify risk of OPLL in Chinese population, which was consistent with a finding reported by Lui et al.20 Besides, it was reasonable to deduce that Runx2 polymorphisms could, to some extent, regulate severity of OPLL as well as the treatment efficacy of posterior laminoplasty, for that the above polymorphisms were significantly correlated with ossification segment number and post‐operative JOA scores of OPLL patients.

However, the above conclusions should be interpreted with caution due to the existence of some limitations. For instance, sample sizes of both the case and control group may not be large enough, and hence some conclusions may appear to be biased. For solving this issue, we recommended a scientific approach to estimate the optimal sample size based on the desired effect size, significance level, and power of the study. Since Runx2 was regarded as a critical mediator for both alkaline phosphatase and osteocalcin gene and it is involved in BMP/Smads pathways which are key regulatory signal transducers in osteoblast differentiation, variations of SNPs that were located in Runx2 are likely to participate in the process of bone formation through BMP/Smad pathways and such mechanism should be verified by further researchers.

5. CONCLUSION

In summary, Runx2 polymorphisms appeared as critical risk factors for OPLL, and they might predict the treatment efficacy of posterior laminoplasty. Taken together, these findings may provide us an informative path to profoundly understand the pathology of OPLL, providing a reference for developing early intervention of OPLL.

CONFLICT OF INTEREST

The authors have declared no conflict of interest.

Supporting information