A nomogram for predicting osteoarthritis based on serum biomarkers of bone turnover in middle age: A cross-sectional study of PTH and β-CTx

Abstract

The objective of this study was to investigate the diagnostic model of osteoarthritis by bone turnover markers in Chinese middle-aged subjects. The study was designed as a cross-sectional investigation with 305 participants aged 45 to 64. Radiographs of tibiofemoral knee joints were used for diagnose osteoarthritis. Radiographic grading, evaluated using the Kellgren and Lawrence grading scale (K-L), was scored by 2 experienced observers who were blinded to the source of subjects. An optimal model was developed by logistic regression. And the prognostic performance of the selected model was assessed by the area under the receiver operating characteristic curve. The prevalence of osteoarthritis was 52.29% (n = 137/262) in middle age. β-CTx levels tended to increase according to the K-L grades, whereas PTH levels significantly decrease. levels of 25(OH)D, β-CTx, and PTH were each significantly associated with osteoarthritis risk (P < .05). Based on the estimated parameters of the optimal model, a nomogram was constructed for predicting osteoarthritis. These data suggest that the combination of PTH and β-CTx could significantly improve the prognosis of osteoarthritis in middle age, and that the nomogram can assist primary physicians in the identification of high-risk men.

1. Introduction

Osteoarthritis (OA) is the most common form of arthritis, and a major cause of disability and impaired quality of life.^[1] Its societal burden is only expected to increase further due to aging of the population, higher life expectancy, and the ongoing obesity epidemic.^[2] It has been estimated to affect 1 in 10 people at age 50 and >1 in 2 people at age 75.^[3] Thus, identified and treated early for middle-aged patients is crucial.

A hallmark of osteoarthritis is the progressive degeneration of articular cartilage and subsequent joint space narrowing. However, when a radiological diagnosis is established, significant joint damage has often already occurred. An increasing line of evidence suggests that bone metabolism markers might be useful for the assessment and prediction because of their faster response compared with × rays, but so far no definite conclusions have been made.^[4]

Most previous studies^[1,5,6] investigating the value of bone turnover markers (BTM) have several limitations, which include the measurement of a few markers, the observation of a broad age spectrum, the focus on the association of markers reflecting metabolism of different tissues, whereas guidance of clinical operation is relatively ignored.

Serologic biochemical markers (biomarkers) are receiving increased attention for guiding evidence-based patient management, and have been widely used in clinical practice.^[7,8] Therefore, in the current study, serum levels of the turnover markers were assessed in Chinese middle-aged subjects (45–64 years) with either no or radiographic OA (Kellgren and Lawrence, K&L). Then, associations of the BTM levels with osteoarthritis were modeled and designed to predict diagnosis and progression.

2. Methods

2.1. Study design and participants

This cross-sectional study is a sub-cohort analysis of a study on BTM levels conducted from the outpatients and inpatients in the Department of Orthopedic and Traumatology and Health Examination Center of The 72nd Group Military Hospital of the People’s Liberation Army between December 2018 and January 2020, using a total of 305 participants aged 45 to 64^[9,10] (data on file). All participants completed a standardized questionnaire before their scan (including age, medical history, drugs, smoking, etc.). Height, weight and body mass index (BMI, kg/m²) were measured. Study exclusion criteria included metabolic and endocrine diseases; bone tumors; renal insufficiency; secondary causes of osteoporosis, such as Cushing’s disease, hyperthyroidism, Crohn’s disease, or rheumatoid arthritis; use of a bone active agent or hormone therapy within half a year. The presented study was approved by the Ethical Committee of The 72nd Group Military Hospital of the People’s Liberation Army.

2.2. Radiographic imaging

Radiographs of tibiofemoral knee joints were made in a posteroanterior view, weight-bearing, semiflexed.^[11] Radiographs were evaluated using the Kellgren and Lawrence grading scale (K-L)^[12] where grade 0 = normal; grade 1 = possible osteophytes only; grade 2 = definite osteophytes and possible joint space narrowing; grade 3 = moderate osteophytes and/or definite joint space narrowing; and grade 4 = large osteophytes, severe joint space narrowing, and/or bony sclerosis. Radiographic grading was scored by 2 experienced observers who was blinded to the source of subjects.

2.3. Biochemical markers of bone turnover assessment

Blood samples were collected between 9:30 am and 2:00 pm in a non-fasting state, centrifuged and stored at −80°C until analyzed. The following assays were performed: OC (osteocalcin), t-P1NP (N-terminal procollagen type 1 extension pro-peptide), β-CTx (beta-isomerized type I collagen C-telopeptide breakdown products), 25(OH)D (25-hydroxy-cholecalciferol), and PTH (parathormone).^[13–15] The concentrations of BTMs were measured enzymatically on an autoanalyzer (COBAS e601; Diagnostics, Germany) following the manufacturer’s recommendations. The intra-assay coefficient variation (CV) of the method was <10%, and the inter-assay CV was <15%. The repeatability and intermediate precision were as follows: t-P1NP (Elecsys® total P1NP): <3.3% and < 3.8%; β-CTx (Elecsys® β-CrossLaps/serum): <4.8% and < 5.8%; PTH (Elecsys® PTH): <2.1% and < 3.5%; 25(OH)D (Elecsys® Vitamin D total): <6.9% and < 13.2%; OC (Elecsys® N-MID Osteocalcin): <1.4% and < 2.4%. The lower limit of quantification (LLOQ) and upper limit of quantification (ULOQ) were described: t-P1NP, 5 to 1200 ng/mL; β-CTx, 0.010 to 6.00 ng/mL; PTH, 1.20 to 5000 pg/mL; 25(OH)D, 3.00 to 70.0 ng/mL; OC, 0.500 to 300 ng/mL.

2.4. Statistical analysis

R software (version 4.0.1, R Foundation for Statistical Computing, Vienna, Austria) and SPSS (version 25.0, SPSS Science, Chicago, IL) for Windows were used for statistical analyses. Normal distribution of the data was evaluated by QQ-plot and histogram. Different variables were described by the mean ± SD/SE. When comparing the means between groups, the Student t test or One-way ANOVA was used. Logistics regression analysis was used to consider whether the probability of having OA was related to biochemical markers of bone turnover. Then, a nomogram plot was drawn on the basis of regression analysis. Finally, receiver operating characteristic (ROC) curves were plotted for validating the logistic model by the maximum sensitivity and specificity. A P value < .05 was taken to indicate statistical significance.

3. Results

3.1. Demographics of participants

A total of 262 of the 305 participants ranged in age from 45 to 64 years were included in this study, and 43 subjects were excluded for reasons. A flow diagram of the inclusion process is demonstrated in Figure 1.

Figure 1.: Flowchart of the study. HC = healthy control, OA = osteoarthritis.

Baseline characteristics are shown in Table 1. 137 participants were diagnosed with osteoarthritis at baseline comprised 73.72% females with an overall age of 59.37 ± 7.71 years and BMI of 22.32 ± 3.06 kg/m². Patients with knee OA did not differ from controls for age, height, weight, BMI, and menopause (P > .05). However, and as expected, patients with knee OA had a higher proportion of women (101 [38.55%] vs 36 [13.74%], P < .01) than controls in gender distribution.

Table 1 - Clinical characteristics of participants.

Parameters	HC group1	OA group	P value	HC group 2	Grade Ⅰ/Ⅱ	Grade Ⅲ/Ⅳ	P value
N	125	137		23	18	22
Gender
Male	60	36	<.01	7	5	6	.969
Female	65	101		16	13	16
Age (yr)	57.92 ± 7.17	59.37 ± 7.71	.117	67.26 ± 11.90	66.11 ± 9.89	66.73 ± 7.56	.936
Height (cm)	162.28 ± 7.42	161.45 ± 5.94	.328	160.82 ± 2.97	160.94 ± 6.68	160.23 ± 3.53	.858
Weight (kg)	58.79 ± 7.91	58.16 ± 6.14	.480	56.43 ± 6.47	56.88 ± 7.12	57.54 ± 6.31	.853
BMI (kg/m2)	22.30 ± 2.45	22.32 ± 2.12	.964	21.82 ± 2.53	21.97 ± 2.53	22.42 ± 2.44	.712
Menopause	44.09 ± 3.44	44.18 ± 3.06	.852

BMI = body mass index, HC = healthy control, OA = osteoarthritis.

Among all 262 participants, only 63 of them were measured for all biochemical markers of bone turnover. Of these knees, 18 knees were scored K&L grade Ⅰ/Ⅱ and 22 knees were scored K&L grade Ⅲ/Ⅳ. The baseline demographics were similar between the 3 groups (P > .05).

3.2. Gender and biochemical markers of bone turnover (t-P1NP and β-CTx)

In patients with knee OA, 2 markers of bone turnover were increased compared with controls (Fig. 2). Among them, β-CTx were significantly increased (OA: 610.07 ± 258.97 ng/mL vs Ctrl: 525.53 ± 234.51 ng/mL, P < .05), whereas no significant difference was observed for t-P1NP (OA: 61.12 ± 27.13 ng/mL vs Ctrl: 58.64 ± 25.15 ng/mL, P = .446).

Figure 2.: Compare the differences between OA group and ctrl group in β-CTx as well as t-P1NP. β-CTx = beta-isomerized type I collagen C-telopeptidebreakdown products, OA = osteoarthritis, t-P1NP = N-terminal procollagen type 1 extension pro-peptide.

No significant difference in average t-P1NP levels between male patients (OA: 51.59 ± 22.50 ng/mL vs Ctrl: 55.53 ± 27.65 ng/mL, P = .473) with and without knee OA was seen, and similar to female patients (OA: 64.51 ± 27.91 ng/mL vs Ctrl: 61.51 ± 22.43 ng/mL, P = .468). For β-CTx levels, the difference was significant (OA: 643.24 ± 287.14 ng/mL vs Ctrl: 553.96 ± 251.61 ng/mL, P < .05) in the female group while it was not significant in the male group(OA: 516.99 ± 264.78 ng/mL vs Ctrl: 494.72 ± 212.26 ng/mL, P = .652).

3.3. Biochemical markers of bone turnover of the subgroups divided by radiographic grade

Figure 3 illustrates the difference in biochemical markers of bone metabolism in these subgroups divided by radiographic grade (mean ± SE). Among the various biochemical markers, 25(OH)D levels tended to increase according to the grades, but they were lower in grade Ⅲ than in grade Ⅱ (grade Ⅱ: 28.22 ± 2.94 ng/mL vs grade Ⅰ: 21.36 ± 1.67 ng/mL, P < .05). β-CTx levels tended to increase according to the grades, whereas PTH levels significantly decrease. A significant difference of biochemical bone markers was observed between grades Ⅰ and Ⅲ (383.29 ± 27.61 ng/mL vs 700.59 ± 64.21 ng/mL, P < .01) in β-CTx and grades Ⅰ and Ⅱ (59.26 ± 3.65 pg/mL vs 42.73 ± 4.36 pg/mL, P < .01) and grades Ⅰ and Ⅲ (59.26 ± 3.65 pg/mL vs 38.06 ± 3.14 pg/mL, P < .01) in PTH. (Grade Ⅰ = K&L grade 0; Grade Ⅱ = K&L grade 1/2; Grade Ⅲ = K&L grade 3/4.)

Figure 3.: The changes of BTMs (Bone turnover markers) during different phases. 25(OH)D = 25-hydroxy-cholecalciferol, β-CTx = beta-isomerized type I collagen C-telopeptidebreakdown products, OC = osteocalcin; PTH = parathormone, t-P1NP = N-terminal procollagen type 1 extension pro-peptide.

3.4. Strong association of PTH and β-CTx with knee OA

To assess the relative contributions of the different biochemical markers in determining the osteoarthritis, multivariate logistic model, including imaging diagnosis as the dependent variables (grade 0 = 0; grade 1–4 = 1), and levels of 25(OH)D, β-CTx, and PTH as the independent variables, were used (Table 2). Levels of β-CTx (OR = 1.003, 95% CI: 1.000–1.006, P = .047) and PTH (OR = 0.930, 95% CI: 0.886–0.977, P = .004) were the important predictors of OA.

Table 2: Univariate logistic regression analysis of the OA risk factors based on BTMs.

Furthermore, a discriminating nomogram plot (Fig. 4A and B) was generated to predict osteoarthritis probabilities (AUC = 0.856). The result suggested that the predictions calculated with the nomogram approximated the actual outcomes. For the orthopedic physicians, the plot was available to locate a patient’s levels of β-CTx, and PTH in each axis; to draw a line straight upward to the point axis and sum up the total points; and then to draw a line straight downward to determine their risk of osteoarthritis.

Figure 4.: Nomogram plot of osteoarthritis probabilities and the area under the curve (AUC) of the model are demonstrated. AUC = 0.856. β-CTx = beta-isomerized type I collagen C-telopeptidebreakdown products, PTH = parathormone.

4. Discussion

We investigated the relation between osteoarthritis, even the severity of radiological OA and biochemical markers of bone turnover, and we found significantly increased β-CTx levels compared to the healthy control, especially females, yet PTH levels significantly decrease according to the KL grades. Furthermore, 25(OH)D was significantly higher in grade Ⅱ compared to grade Ⅰ.

Bone turnover has been suggested to be an important factor in OA progression.^[5] Despite years of research, the relationship between bone turnover biomarkers and osteoarthritis is still being discussed owing to its dynamic nature. Previous studies^[16,17] have indicated that increased bone turnover can be a consequence and further contributes to OA. However, no definite conclusions have been made for bone turnover markers, which could predict the incidence and progression of OA. In our study, serum concentrations of BTMs changed in both bone resorption and formation when comparing the healthy control, β-CTx and PTH were strongly associated with radiographic OA progression, suggesting that the high remodeling rates in OA joints. Our results agree with what has been reported by Cloos et al and Karina et al,^[18,19] who found in their studies an increase in resorption markers and a decrease in bone mineral density (BMD). Of note, β-CTx is considered as the first-choice bone resorption markers varied by gender and age. Furthermore, women with knee OA increased their β-CTx levels in middle age compared with men (Fig. 2). Bettica et al^[16] also demonstrated that bone turnover is important for the progression of OA, as increased bone resorption markers, most likely reflecting systemic skeletal changes.

Additionally, PTH is an endogenous hormone with a central role in maintaining calcium homeostasis of the body. We found that PTH levels significantly decreased as OA progressed. That is, endogenous PTH and the stage of knee osteoarthritis showed a negative association. We believe that changes in such indicator often cannot be understood separately. Various studies^[20,21] have indicated that bone-forming therapy, such as PTH analogues, may lead to satisfactory results in the early stage of osteoarthritis. This also confirms the significance of this substance to disease from the side. However, more human data are required.

This study provided the opportunity to explore the utility of bone turnover biomarkers in a cohort evaluated for dynamic bone turnover and radiographic knee OA progression. Moreover, the clinical translation of correlational results has never been mentioned in the same-topic study. But our study has some limitations. The number of patients and serum indicators included in the study were limited. Besides, to maintain the sample size of each group, the cases of different stages have also been integrated, which may affect the universality of statistical modeling. However, this research has great practical significance, and the selection of grouping and modeling methods is fully based on the characteristics of the data, which provides a reference to other similar research.

5. Conclusion

In summary, our study demonstrates that serum β-CTx and PTH are strongly associated with radiographic OA progression in middle age and suggesting their potential utilities as biomarkers for osteoarthritis. BTMs deserve further investigation as biomarkers of bone remodeling to aid in identifying the occurrence and development of diseases.

Author contributions

Data curation: Juntao Xu.

Writing – original draft: Xueqiang Chen.

Writing – review & editing: Xueqiang Chen, Lixin Yu, Houjian Zhang.