Site-specific predictors of bone mineral density response to zoledronic acid in postmenopausal osteoporosis: differential roles of bone turnover markers and body mass index across skeletal sites

Abstract

Background

Postmenopausal osteoporosis (PMOP) elevates fracture risk. Zoledronic acid effectively reduces fractures, but individual bone mineral density (BMD) responses vary. Systematic evidence on independent site-specific predictors of BMD changes across femoral neck, total hip, and lumbar spine remains scarce. This study aimed to identify independent factors associated with 12-month BMD changes at the femoral neck (ΔFNBMD), total hip (ΔTHBMD), and lumbar spine (ΔLSBMD) in PMOP patients (T-score ≤ − 2.5) following initial zoledronic acid treatment.

Methods

A single-center retrospective cohort of 145 PMOP patients (BMD T-score ≤–2.5) was analyzed. Baseline BMD, bone turnover markers (β-cross-linked C-terminal telopeptide of type I collagen [β-CTX-I], N-terminal propeptide of type I procollagen [PINP]), blood biochemistry, and body mass index (BMI) were assessed. BMD was reassessed at the 12-month follow-up. Multiple regression and logistic models identified predictors of ΔFNBMD, ΔTHBMD, and ΔLSBMD. Receiver operating characteristic (ROC) analysis evaluated predictive performance.

Results

ΔFNBMD correlated independently with baseline β-CTX-I (β = 0.031, P = 0.017). ΔTHBMD was associated with β-CTX-I (β = 0.036, P < 0.001) and BMI (β = 0.001, P = 0.027). ΔLSBMD linked to PINP (β = 0.000, P = 0.009), serum iron (β = 0.003, P = 0.023), and chloride (β=-0.005, P = 0.023). Logistic regression showed higher baseline PINP increased the odds of effective FNBMD improvement (odds ratio [OR] = 1.042 per 1 ng/mL increase, 95% confidence interval [CI]: 1.018–1.067, P < 0.001), while β-CTX-I (OR = 1.003, P = 0.008) and BMI (OR = 1.156, P = 0.023) influenced THBMD response. ROC analysis confirmed PINP as a robust predictor for FNBMD efficacy (area under the curve [AUC] = 0.724, P < 0.001), and β-CTX-I (AUC = 0.669, P = 0.004) and BMI (AUC = 0.637, P = 0.021) as clinically actionable predictors for THBMD efficacy.

Conclusion

Zoledronic acid-induced BMD changes in PMOP exhibit site-specific associations with distinct biomarkers and clinical parameters. These findings provide preliminary evidence for tailoring treatment based on baseline metabolic and anthropometric profiles, though larger prospective studies are required to validate these markers for clinical application.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12905-026-04671-0.

Introduction

Osteoporosis (OP) is a systemic skeletal disease characterized by reduced bone mass and destruction of the bone microarchitecture. It predominantly affects postmenopausal women and is likely to induce fragility fractures, severely impacting the quality of life and life expectancy of patients [1]. Compared to secondary osteoporosis and male osteoporosis, postmenopausal osteoporosis (PMOP) is the more common type of osteoporosis, and its pathogenesis is mainly related to increased bone resorption and decreased bone formation caused by the decline in estrogen levels [2, 3]. With the acceleration of the global aging process, the incidence of PMOP has been rising year by year and has become an important public health issue. According to statistics, approximately 200 million women worldwide suffer from osteoporosis, with the majority being postmenopausal women [4, 5]. Osteoporotic fractures not only bring great pain and economic burden to patients but also significantly increase their risk of death [6]. Therefore, how to effectively prevent and treat PMOP and reduce the risk of fractures has become an important topic in both clinical and scientific research fields.

Zoledronic acid, a third-generation bisphosphonate drug, inhibits the activity of osteoclasts and reduces bone resorption and can significantly increase bone mineral density (BMD) and reduce the risk of fractures [7, 8]. According to a study of patients with osteoporosis, the use of zoledronic acid was associated with lower all-cause mortality, particularly in patients with hip and vertebral fractures [9]. In recent years, due to its long-acting nature and convenient administration route (intravenous infusion once a year), zoledronic acid has been widely used in the treatment of PMOP. Multiple clinical studies have confirmed that zoledronic acid can significantly increase the BMD of PMOP patients and reduce the risk of vertebral and non-vertebral fractures [10, 11]. However, although zoledronic acid has shown significant clinical effects in increasing BMD and reducing the risk of fractures, there is large variations in the treatment responses among different patients, and the improvement in BMD is not obvious in some patients [12]. This individual difference may be related to various factors such as the patient’s bone metabolism status, body mass index, and dietary habits [13]. Therefore, identifying the independent factors influencing BMD changes after zoledronic acid treatment is crucial for optimizing the treatment regimen and improving the treatment effect.

Currently, there are few studies on the influencing factors of BMD changes in PMOP patients after receiving zoledronic acid treatment, especially a systematic study on the independent related factors of BMD changes in different sites (such as the femoral neck, total hip, and lumbar spine) [14]. Existing studies have shown that bone turnover markers (BTMs) such as β-isomerized carboxy-terminal telopeptide of type I collagen (β-CTX-I) and amino-terminal propeptide of type I procollagen (PINP) are of great value in evaluating bone metabolism status and predicting treatment effects [15, 16]. Research evidence suggests that incorporating these markers of bone metabolism into assessment algorithms can help improve the predictive accuracy of fracture risk [15]. They also summarized the collection methods of bone metabolism markers to increase sample stability during testing [16]. In addition, factors such as body mass index (BMI) and dietary habits are also considered to potentially affect the changes in BMD [17, 18]. For example, dairy products in the diet provide minerals and nutrients for bone growth. However, intake of carbonated beverages showed a significant inverse association with BMD as well as fracture risk [18]. However, the roles of these factors in the changes in BMD at different sites have not been fully verified.

Despite zoledronic acid’s established efficacy in reducing fractures and increasing BMD in PMOP, substantial inter-patient variability in treatment response persists. Current clinical guidelines and prior studies largely assess BMD changes as a composite outcome or focus on isolated skeletal sites, leaving a critical gap in the systematic identification of independent predictors driving site-specific BMD responses across the femoral neck, total hip, and lumbar spine. This study aimed to identify independent site-specific predictors of 12-month BMD changes at the femoral neck, total hip, and lumbar spine in postmenopausal osteoporosis patients receiving initial zoledronic acid treatment. Distinct from prior studies focusing on overall or single-site BMD changes, this work provides a comprehensive characterization of site-specific predictors across the femoral neck, total hip, and lumbar spine in PMOP patients receiving zoledronic acid, addressing a key gap in the current understanding of heterogeneous treatment responses. By concurrently assessing bone turnover markers, BMI, and biochemical indices, our findings reveal heterogeneous association patterns across anatomical sites, providing preliminary evidence to inform differentiated individualized treatment decisions beyond one-size-fits-all approaches.

Materials and methods

Study participants

This was a single-center retrospective analysis. A total of 145 PMOP patients who were treated at Zhejiang Provincial People’s Hospital from January 2020 to December 2020 were included. Despite the COVID-19 pandemic and associated movement restrictions, our hospital sustained specialized outpatient clinics for chronic disease management, utilizing staggered appointments and community coordination to facilitate patient visits. Importantly, patients with any confirmed history of SARS-CoV-2 infection prior to baseline assessment or during the 12-month follow-up period were strictly excluded from this study to eliminate acute infection-related confounding. The patients’ ages ranged from 49 to 91 years, with a mean age of (79.00 ± 3.61) years. Among them, 72 patients had hypertension, 19 had type 2 diabetes mellitus, and 45 had chronic gastritis.

Inclusion criteria: ① Diagnosed with osteoporosis (the bone mineral density T-score of the femoral neck, total hip, or lumbar spine 1–4 ≤ -2.5) as determined by dual-energy X-ray absorptiometry (DXA) [1]; ② Able to have regular reexaminations of BMD.

Exclusion criteria: ① Patients with diseases such as chronic renal insufficiency (creatinine clearance rate ≤ 30 mL/min), hypocalcemia, and advanced malignant tumors; ② Patients who had received bisphosphonate treatment within the past 3 years; ③ Patients who had received zoledronic acid treatment before; ④ Patients who had previously been treated with osteoporosis medications; ⑤ Patients with a confirmed history of SARS-CoV-2 infection (COVID-19) prior to baseline assessment or during the 12-month follow-up period.

Research methods

All patients underwent a general condition survey, including age, menopause duration, height, weight, and concomitant diseases (including hypertension, type 2 diabetes mellitus, chronic gastritis), etc. The duration of menopause (age - age at menopause) and body mass index were calculated, and they underwent BMD measurement, bone metabolism marker and blood biochemistry tests at baseline. The BMD was rechecked 12 months after receiving zoledronic acid treatment. All blood sampling and testing followed standardized operating procedures (International Federation of Clinical Chemistry, IFCC) to ensure data accuracy and reproducibility.

Treatment method

Intravenous infusion of 5 mg of zoledronic acid (produced by Novartis Pharma Stein AG, Switzerland, batch number SYC68), and the infusion duration was not less than 30 min. After that, the patients took calcitriol soft capsules (0.5 µg, once a day, produced by Catalent Germany Eberbach GmbH, Switzerland, batch number B4414) and calcium carbonate D3 tablets (1 tablet (containing 600 mg calcium, 125 IU vitamin D), once a day, produced by Wyeth Pharmaceuticals Inc., USA, batch number EH6T) for a total of 12 months. All patients showed good treatment compliance.

BMD measurement

The Lunar Prodigy dual-energy X-ray bone densitometer (GE Company, USA) was selected, and a dedicated person was responsible for measuring and recording the 12-month BMD at the femoral neck (ΔFNBMD), total hip (ΔTHBMD), and lumbar spine (ΔLSBMD) of each PMOP patient. For FNBMD, we measured bilateral proximal femurs and selected the lowest BMD in both sides as the measurement. The densitometer was calibrated daily prior to measurements. For obese patients (BMI > 30), the High Penetration Mode was enabled. At the same time, standardize the patient position and reduce the generation of artifacts. △BMD is the difference between the BMD 12 months after treatment and that before treatment. All bone density measurements were completed by the same trained operator after the equipment was calibrated, and the measurement repeatability error was controlled within 1%.

Blood biochemistry test and bone metabolism marker test

In the early morning (6:30 − 7:30), 3 mL of venous blood was drawn from the elbow on an empty stomach. Part of the blood was directly used to detect albumin, blood glucose, creatinine, uric acid, serum potassium, serum sodium, serum chloride, serum calcium, serum phosphorus, and serum iron using the 7600 fully automatic biochemical analyzer of Hitachi Company, Japan. For β-CTX-I measurement, 3 mL of venous blood was collected in EDTA tubes, centrifuged at 4 °C within 30 min, and plasma aliquots were stored at -70 °C until analysis. This protocol adheres to IFCC recommendations to ensure marker stability [12]. The Roche Cobas e 601 immunochemiluminescence analyzer (Japan) was selected to measure plasma β-isomerized carboxy-terminal telopeptide of type I collagen (β-CTX-I), amino-terminal propeptide of type I procollagen (PINP), osteocalcin, parathyroid hormone, and 25-hydroxyvitamin D (25(OH)D).

Statistical processing

All data were analyzed using SPSS statistical software (version 17.0). Measurement data were expressed as mean ± standard deviation (Mean ± SD). The independent samples t-test was used to analyze the differences between groups. Spearman correlation analysis was used to analyze the correlation between continuous variables. Finally, the multiple regression analysis method was selected to screen the independent related factors of △FNBMD, △THBMD, and △LSBMD. A P value less than 0.05 was considered statistically significant.

Results

Correlation analysis of △FNBMD, △THBMD, and △LSBMD in PMOP patients

Baseline demographic, clinical, and biochemical characteristics of the 145 enrolled postmenopausal osteoporosis (PMOP) patients are summarized in Table 1. The cohort had a mean age of 79.0 ± 3.6 years and a mean body mass index (BMI) of 22.82 ± 3.61 kg/m². Baseline bone mineral density (BMD) T-scores confirmed established osteoporosis across all skeletal sites (femoral neck: −2.9 ± 0.5; total hip: −2.7 ± 0.4; lumbar spine: −3.1 ± 0.6). Levels of bone turnover markers, including β-CTX-I (0.52 ± 0.27 ng/mL) and PINP (53.71 ± 10.03 ng/mL), reflected the high bone turnover state typical of PMOP. The Spearman correlation analysis was used to analyze the correlations. The variables included age, years of menopause, body mass index, β-CTX-I, PINP, osteocalcin, parathyroid hormone, 25 (OH) D, albumin, blood glucose, creatinine, uric acid, serum potassium, serum sodium, serum chloride, serum calcium, serum phosphorus, serum iron, as well as △FNBMD, △THBMD, and △LSBMD. The results showed that △FNBMD was correlated with β-CTX-I and PINP (both P < 0.05), △THBMD was correlated with years of menopause, body mass index, β-CTX-I, and PINP (all P < 0.05), and △LSBMD was correlated with PINP (P < 0.05), while weak correlations with serum chloride and serum iron were observed (both P < 0.05, but correlation coefficients r < 0.2) (Fig. 1A-J). These weak associations should be interpreted with caution due to their low magnitude. In addition, there were no significant correlations between other variables and △FNBMD, △THBMD, and △LSBMD (all P > 0.05) (Table 1).

Table 1.

Univariate analysis of △BMD in 145 PMOP patients

Variables		△FNBMD	△THBMD	△LSBMD
Age (years)	79.00 ± 3.61	-0.058	-0.123	-0.056
Menopause years (years)	27.26 ± 10.32	-0.083	-0.226*	-0.007
Body Mass Index (kg/m2)	22.82 ± 3.61	0.111	0.182*	-0.048
β-CTX-I (ng/mL)	0.52 ± 0.27	0.199*	0.263**	0.157
PINP (ng/mL)	53.71 ± 10.03	0.222*	0.201*	0.222*
Osteocalcin (ng/mL)	22.04 ± 5.59	0.146	0.142	0.148
Parathyroid hormone (pg/mL)	46.58 ± 20.60	-0.065	-0.014	0.032
25 (OH) D (ng/mL)	18.01 ± 8.62	0.057	-0.109	-0.029
Albumin (g/L)	43.82 ± 3.62	-0.146	-0.038	-0.012
Blood glucose (mmol/L)	5.59 ± 1.31	0.017	-0.026	-0.134
Creatinine (µmol/L)	82.23 ± 14.47	0.146	0.052	0.032
Uric acid (µmol/L)	306.60 ± 20.95	0.058	-0.011	-0.001
Blood potassium (mmol/L)	4.07 ± 0.32	0.025	0.010	0.009
Blood sodium (mmol/L)	142.17 ± 2.66	-0.033	-0.047	-0.125
Blood chlorine (mmol/L)	103.85 ± 2.99	-0.112	-0.129	-0.181*
Blood calcium (mmol/L)	2.36 ± 0.13	0.088	0.057	0.009
Blood phosphorus (mmol/L)	1.10 ± 0.16	0.168	0.084	-0.119
Blood iron (mmol/L)	14.60 ± 5.04	0.111	0.154	0.190*

FNBMD Femoral Neck Bone Mineral Density, THBMD Total Hip Bone Mineral Density, LSBMD Bone Mineral Density of Lumbar Spine 1–4, β-CTX-I Carboxy-terminal Telopeptide of Type I Collagen, PINP Amino-terminal Propeptide of Type I Procollagen; 25(OH)D 25-hydroxyvitamin D

*P < 0.05; **P < 0.01

Fig. 1.

Analysis of influencing factors of △BMD in PMOP patients

Univariate analysis of △FNBMD, △THBMD, and △LSBMD in PMOP patients

The changes in △BMD among different subgroups of concomitant diseases in PMOP patients were analyzed by univariate analysis. As shown in Table S1, there were no statistically significant differences in △FNBMD, △THBMD, and △LSBMD among the subgroups of hypertension, chronic gastritis, and type 2 diabetes mellitus in PMOP patients (P > 0.05), which suggested that △BMD in PMOP patients was not associated with concomitant diseases.

Multiple regression analysis of △FNBMD, △THBMD, and △LSBMD in PMOP patients

Taking the indicators correlated with △FNBMD, △THBMD, and △LSBMD in the univariate analysis as independent variables, and △FNBMD, △THBMD, and △LSBMD as dependent variables, a Stepwise regression analysis was performed. The inclusion probability criterion was ≤ 0.050, and the exclusion probability criterion was ≥ 0.100. As shown in Table S2, PINP was an independent related factor of △FNBMD in PMOP patients. β-CTX-I did not meet the entry standard, indicating that β-CTX-I was not an independent related factor of △FNBMD. In addition, both β-CTX-I and BMI were independent related factors of △THBMD in PMOP patients (Table S3). The duration of menopause and PINP did not meet the entry standard, indicating that neither of them was an independent related factor of △THBMD. Moreover, PINP was identified as an independent related factor of △LSBMD in PMOP patients. Although serum iron and serum chloride showed statistical associations in multiple regression (Table S4), their weak correlation coefficients and lack of clear biological plausibility limit their interpretability. In addition, the linear distribution results of the multiple regression analysis of △FNBMD, △THBMD, and △LSBMD are shown in Fig. 2A-C. The predicted cumulative probability and the measured cumulative probability of △FNBMD, △THBMD, and △LSBMD showed a linear correlation, indicating that the residuals conformed to a normal distribution and the model was stable.

Fig. 2.

P-P plot of standardized residuals of △BMD regression

Influencing factors of the efficacy of zoledronic acid treatment in PMOP patients

The bone mineral densities of the femoral neck, total hip, and lumbar spine (L1-L4) of PMOP patients treated with zoledronic acid were analyzed by dual-energy X-ray bone densitometry. Taking the change in △BMD as the criterion for judging the therapeutic effect of zoledronic acid treatment, the patients were divided into an effective group (△BMD > 0) and an ineffective group (△BMD ≤ 0). The scanning images of the effective group and the ineffective group are shown in Fig. 3. The influencing factors of the efficacy of zoledronic acid treatment in PMOP patients were analyzed by logistic regression. As shown in Table S5, the improvement effects at different skeletal sites were affected by different biomarkers and clinical parameters. The level of the bone formation marker PINP was significantly positively correlated with the efficacy of improving femoral neck bone mineral density (△FNBMD) (β = 0.041, P < 0.001), and its OR value was 1.042 (95%CI: 1.018–1.067), suggesting that for every 1 unit increase in the baseline PINP, the probability of effective treatment increased by 4.2%. The analysis of the influencing factors of total hip bone mineral density (△THBMD) showed that the bone resorption marker β-CTX-I (β = 0.003, P = 0.008) and BMI (β = 0.145, P = 0.023) were both statistically significant. A 1-unit increase in the β-CTX-I level could increase the probability of efficacy by 0.3% (OR = 1.003), while a 1 kg/m² increase in BMI could increase the probability of efficacy by 15.6% (OR = 1.156). It is worth noting that the improvement in lumbar spine bone mineral density (△LSBMD) did not show a significant association with PINP, serum iron, or serum chloride (P > 0.05).

Fig. 3.

Representative images of dual-energy X-ray absorptiometry analysis of PMOP patients treated with zoledronic acid

Predictors of the efficacy of zoledronic acid treatment in PMOP patients

The efficacy of predictors of the efficacy of zoledronic acid treatment in PMOP patients was evaluated by ROC curve analysis. As shown in Table 2; Fig. 4A-C, the improvement in BMD at different skeletal sites was significantly associated with specific biomarkers and clinical parameters, and their predictive values showed anatomical site specificity. In the prediction of △FNBMD, the baseline level of PINP showed the best predictive value, with an AUC of 0.724 (95%CI: 0.627–0.822), and the optimal cut-off value was 42.835, with a sensitivity and specificity of 74.3% and 70.6%, respectively. It is worth noting that the prediction system for △THBMD showed a two-factor characteristic: both the prediction models of the bone resorption marker β-CTX-I (AUC = 0.669, cut-off value 430.20 pg/mL) and BMI (AUC = 0.637, cut-off value 24.064 kg/m²) were clinically operable. The high specificity (87.1%) of BMI indicated its screening value for treatment response. However, the predictive efficacy of △LSBMD was relatively limited, and the AUCs of PINP, serum iron, and serum chloride did not reach statistical significance (P > 0.05). This may be related to the relatively high proportion of trabecular bone in the lumbar spine and the heterogeneity of the bone metabolism regulation mechanism, suggesting that it is necessary to conduct a comprehensive evaluation by combining vertebral morphological indexes or inflammatory factors. It is worth noting that the negative predictive value of serum chloride (sensitivity 80.6% at the cut-off value of 106.050 mmol/L) provides a potential direction for electrolyte balance intervention, but it needs to be verified with a larger sample size.

Table 2.

Efficacy of predictors of the efficacy of zoledronic acid in PMOP patients

Factors	AUC(95% CI)	Cut-off	Sensitivity	Specificity	PPV	NPV	P value
PINP-△FNBMD	0.72(0.63 ~ 0.82)	42.84	74.3%	70.6%	82.1%	60.3%	< 0.001
β-CTX-I-△THBMD	0.67(0.57 ~ 0.77)	430.20	69.9%	64.5%	76.3%	56.8%	0.004
BMI-△THBMD	0.64(0.54 ~ 0.74)	24.06	40.8%	87.1%	64.6%	71.8%	0.021
PINP-△LSBMD	0.62(0.48 ~ 0.76)	32.65	83.9%	42.9%	84.8%	41.2%	0.084
Blood iron -△LSBMD	0.59(0.44 ~ 0.74)	13.15	63.2%	61.9%	70.3%	54.1%	0.191
Blood chlorine -△LSBMD	0.60(0.47 ~ 0.73)	106.05	80.6%	42.3%	37.6%	83.5%	0.105

PPV Positive Predictive Value, NPV Negative Predictive Value

Fig. 4.

ROC curves of predictors for the efficacy of zoledronic acid in PMOP patients

Discussion

As a third-generation bisphosphonate drug, zoledronic acid effectively inhibits osteoclast activity and reduces bone resorption, demonstrating remarkable effects in increasing bone mineral density and reducing the risk of fractures. Owing to its long-acting nature and convenient annual intravenous dosing, it is widely used in the treatment of PMOP [8]. Nevertheless, in clinical practice, it has been observed that there are significant differences in the responses of different patients to zoledronic acid treatment, and the improvement in bone mineral density in some patients is not satisfactory. Such differences may be closely related to various factors such as the patients’ bone metabolism status, body mass index, and dietary habits [7]. However, up to now, systematic studies on the independent related factors of the changes in bone mineral density at different sites (such as the femoral neck, total hip, and lumbar spine) in PMOP patients after receiving zoledronic acid treatment are relatively scarce. In this study, the method of multiple linear regression analysis was used to screen out the independent related factors of △THBMD, △FNBMD, and △LSBMD from numerous indexes such as age, years of menopause, BMI, concomitant diseases, bone metabolism markers, and blood biochemistry, providing a preliminary insight for the individualized treatment of PMOP patients.

A total of 145 PMOP patients were included in this study, and their general conditions, bone mineral density, bone metabolism markers, and blood biochemistry data were comprehensively collected. The patients received zoledronic acid (5 mg) treatment, and their bone mineral density was rechecked 12 months later. The results of this study showed that the independent related factors of △THBMD, △FNBMD, and △LSBMD in PMOP patients after receiving zoledronic acid treatment were not consistent [19]. The independent related factors of △THBMD were β-CTX-I and BMI, the independent related factor of △FNBMD was PINP, and the independent related factors of △LSBMD were PINP, serum iron, and serum chloride. The independent related factors of △THBMD and △FNBMD were different mainly because the femoral neck is mainly composed of cortical bone, with a relatively low proportion of trabecular bone (accounting for 30–40%), and the bone metabolism is slow. The total hip includes areas such as the femoral neck, greater trochanter, and the upper end of the femoral shaft, which has a higher proportion of trabecular bone and more active bone metabolism. The reason why the independent related factors of △LSBMD are different from those of △THBMD and △FNBMD is not only due to the higher content of trabecular bone in the lumbar spine but also related to the different influences of body weight and limb activities on the lumbar spine and the hip [20]. This site-specific response pattern to zoledronic acid, where predictors vary by skeletal site, aligns with emerging evidence [14, 21]. Our site-specific predictor pattern aligns with emerging evidence that trabecular-rich lumbar spine responds more dynamically to bone turnover suppression than cortical-dominant femoral sites [22].

Although the independent related factors of △THBMD, △FNBMD, and △LSBMD are inconsistent in PMOP patients after receiving zoledronic acid (5 mg) treatment, bone turnover markers (β-CTX-I, PINP) play important roles in the changes in BMD at the three sites. Serum β-CTX-I (a bone resorption marker reflecting osteoclast activity via type I collagen cleavage) and plasma PINP (a bone formation marker indicating osteoblast-derived type I procollagen synthesis) emerged as key factors across different skeletal sites: β-CTX-I independently correlated with △FNBMD and △THBMD, while PINP was linked to △LSBMD [23]. This aligns with prior evidence that bone turnover markers (BTMs) serve as indicators of bone metabolic activity, though their site-specific associations may relate to differences in bone composition (e.g., cortical vs. trabecular bone ratio) [24, 25].

A series of studies in recent years have shown that there is a moderate positive correlation between β-CTX-I and PINP, and both are negatively correlated with BMD [15, 16]. That is, elevated plasma β-CTX-I and PINP levels consistently indicate that anti-bone resorption treatment should be received as early as possible [26].

Due to the rapid decline in estrogen levels in postmenopausal women, the activity of osteoclasts is significantly enhanced, bone loss is accelerated, and the increase in plasma β-CTX-I occurs earlier and faster than that in PINP. Therefore, plasma β-CTX-I is generally preferred as an observation index to determine whether PMOP patients need to receive anti-bone resorption treatment. Eastell R and others integrated the studies on the clinical reference value ranges of serum BTMs in 6 premenopausal (30–44 years old) women in Europe and determined that the reference value range of plasma β-CTX-I is 111–791 pg/mL (with an average value of 297 pg/mL). Therefore, a plasma β-CTX-I level of 300 pg/mL can be used as the cut-off point for high and low turnover [27]. For example, PMOP patients with a plasma β-CTX-I level higher than 300 pg/mL (i.e., high turnover) should start receiving anti-bone resorption treatment, and the higher the plasma β-CTX-I level, the faster the increase in bone mineral density after receiving anti-bone resorption treatment (including zoledronic acid treatment) [28]. However, the current concept of initiating PMOP therapy according to BTM levels has not been reflected in any current guidelines and has not been validated in clinical studies. Therefore, it remains necessary to expand the sample size to more comprehensively explore the reference value of BTMs for the treatment of PMOP.

Body mass index (β = 0.001, P = 0.027) is an independent related factor of △THBMD in PMOP patients after receiving zoledronic acid treatment, which is consistent with the conclusion in relevant studies that a higher body weight has a protective effect on the BMD of PMOP patients [17, 18]. In a 3-year follow-up survey of 400 postmenopausal women, it was found that the average THBMD of women in the higher body mass index group (BMI ≥ 24 kg/m²) was 0.85 g/cm², which was significantly higher than that of women in the lower body mass index group (BMI < 20 kg/m², with an average THBMD of 0.78 g/cm²). Moreover, after receiving zoledronic acid treatment, the increase in THBMD in the higher body mass index group reached 0.06 g/cm², which was significantly higher than that of 0.03 g/cm² in the lower body mass index group. As a mechanical load factor, body weight can promote the formation and mineralization of hip bones, thereby increasing bone mineral density and improving bone strength [29]. Serum iron and serum chloride showed statistical associations with △LSBMD in multiple regression, but their correlation coefficients were low (r < 0.2), and ROC analysis did not support their predictive value for lumbar spine efficacy (AUC < 0.61, P > 0.1) [30]. Due to the lack of strong biological plausibility and reproducibility, these findings are considered preliminary and should not be used to guide clinical practice. Further validation in larger cohorts is required to assess their potential significance [18].

Using the change in BMD as the primary outcome, we analyzed the influencing factors of zoledronic acid efficacy in PMOP patients via logistic regression. The results showed that the efficacy of zoledronic acid in PMOP patients had anatomical site specificity. Previous studies have shown that the efficacy of zoledronic acid shows different specificities at different skeletal sites [31]. For example, at the hip and femoral neck sites, zoledronic acid can significantly increase bone mineral density, while at other sites such as the spine, its effect may not be as obvious as at the hip [31]. This anatomical site specificity may be related to the local differences in bone metabolism. In Allen’s study, zoledronic acid was proven to significantly improve FNBMD, and this improvement was closely related to the change in the level of PINP (a bone formation marker) [32]. Consistent with the results of this study, the improvement in △FNBMD was significantly positively correlated with the baseline level of the bone formation marker PINP. In addition, the effect of zoledronic acid on THBMD was also confirmed, and this effect was significantly positively correlated with β-CTX-I (a bone resorption marker) and body mass index (BMI) [32]. Similarly, the results of our study showed that the efficacy of △THBMD was jointly driven by the bone resorption marker β-CTX-I and BMI, suggesting that active bone resorption and mechanical load stimulation (such as an increase in BMI) may synergistically promote the positive regulation of bone metabolism in the total hip area. Notably, LSBMD did not show a significant association with BTMs or serum electrolytes, which may be related to the high proportion of lumbar cancellous bone and the complex mechanism of metabolic regulation. Comprehensive analysis suggests that bone turnover markers (PINP, β-CTX-I) and body composition parameters (BMI) have predictive potential for zoledronic acid efficacy. In addition, we found that both serum iron and serum chloride were independent factors associated with LSBMD in patients with PMOP. However, there was no significant correlation between these blood electrolyte parameters and the efficacy of zoledronic acid on LSBMD. The effect of blood electrolyte parameters on lumbar efficacy still needs to be verified in larger samples.

While the identified predictors (PINP for FN, β-CTX-I and BMI for TH) were statistically significant, their clinical utility as standalone tools require careful consideration. The odds ratios (e.g., OR = 1.042 for PINP) and beta coefficients, though significant, indicate modest effect sizes. The AUC values for the ROC analyses (0.637–0.724) suggest fair to moderate discriminative ability, highlighting that these biomarkers and BMI are contributory but not definitive predictors [33]. This underscores the complexity of treatment response and the need for integrated clinical models rather than reliance on single parameters.

However, this study has some limitations. The sample size is only 145, which may limit the statistical power, particularly for the multivariable regression, logistic regression, and ROC analyses conducted. This could affect the stability of the models and the ability to detect weaker associations, implying that our findings should be considered preliminary and require validation in larger cohorts. Future multicenter studies with larger sample sizes are warranted to validate these findings across diverse ethnic populations. In addition, A notable limitation is the suboptimal vitamin D status in our cohort, with a mean baseline 25(OH)D of 18.01 ng/mL and a maintenance dose of only 125 IU daily. This insufficient supplementation may have compromised the maximal anabolic response to zoledronic acid and influenced bone turnover marker levels. Future clinical protocols should prioritize achieving vitamin D sufficiency (> 20 ng/mL) before treatment initiation to ensure robust therapeutic effects. Moreover, we acknowledge that concomitant medications like glucocorticoids, antidiabetics, and proton pump inhibitors, which may affect bone metabolism, were not formally analyzed as confounders, though their low prevalence (≤ 13.1%) and short-term use in our cohort likely had limited impact on 12-month BMD changes, and sensitivity analyses excluding these patients yielded consistent results. While we followed standardized pre-analytical protocols for β-CTX-I (EDTA plasma), future studies could include parallel serum-plasma comparisons to further validate matrix-specific effects. A key limitation is the absence of post-treatment bone turnover marker (BTM) measurements. While baseline β-CTX-I and PINP levels were significantly associated with site-specific BMD responses, we could not evaluate treatment-induced changes in these markers or their correlation with BMD kinetics over time. Future prospective studies should include serial BTM assessments (e.g., at 3–6 months post-treatment) to validate their dynamic role in predicting long-term therapeutic outcomes. Furthermore, the generalizability of our findings is constrained by the single-center retrospective design and modest sample size. Data on key environmental/behavioral confounders, including detailed dietary intake, physical activity, smoking, alcohol use, and the specific effects of some concomitant medications (e.g., low-dose glucocorticoids), were not captured, and the potential influence of some concomitant medications was not analyzed. This limits broader applicability and suggests our results are preliminary, requiring future validation in larger, prospective studies that systematically assess these factors. In addition, although we rigorously excluded patients with confirmed COVID-19 infections, we acknowledge the limitation that undetected subclinical infections or systemic inflammatory states related to the pandemic may have influenced bone turnover markers. While this single-center study minimizes such confounding through strict exclusion, future multicenter research should incorporate systematic serological testing to fully account for this factor. Finally, we acknowledge that menopausal duration, parathyroid hormone (PTH) levels, and antihypertensive medication use may act as important confounders. Although baseline PTH (45.2 ± 12.1 pg/mL) and menopausal duration (median 8 years) were documented, they were not included as covariates in the multivariate models. Additionally, 13.1% of participants used antihypertensive drugs (mostly calcium channel blockers), which could theoretically influence bone metabolism; however, sensitivity analyses excluding these patients showed no alteration in the identified predictors, suggesting their impact was limited in this cohort. It is noted that stepwise selection, while practical for exploratory analysis, has inherent limitations such as potential model instability. Future studies should prospectively collect and adjust for these variables to confirm our findings.

Conclusion

In conclusion, this study demonstrates that zoledronic acid-induced BMD changes in PMOP patients exhibit site-specific associations with distinct biomarkers and clinical parameters. β-CTX-I independently predicted femoral neck and total hip BMD improvement, while PINP and BMI influenced total hip response. Lumbar spine BMD changes were primarily linked to PINP, with weak and preliminary associations with serum iron and chloride that require further validation. ROC analysis highlighted PINP (AUC = 0.724) and β-CTX-I/BMI (AUC = 0.669–0.637) as robust predictors for femoral neck and total hip efficacy, respectively. Although PINP, β-CTX-I, and BMI have certain predictive value for treatment efficacy (AUC 0.637–0.724), the predictive performance of the current model remains at a moderate level. Its accuracy as a standalone clinical decision-making tool is limited, and it needs to be comprehensively judged in combination with other clinical indicators. These hypothesis-generating findings identify potential site-specific correlates of zoledronic acid response, but the small, unicentric design precludes definitive conclusions. Robust clinical tools require validation through multicenter studies with standardized protocols. Further research is required to develop validated predictive tools before translating these insights into practice.

Authors' contributions

CY.X. and PD.B. wrote and revised the main manuscript text. CY.X. performed the investigation. YQ.X. provided the resources. PD.B. supervised and administrated the projected. All authors reviewed the manuscript.

Funding

Authors received no specific funding for this work.

Data availability

Data generated or analyzed during this study are provided in full within the present manuscript.

Declarations

Ethics approval and consent to participate

The research related to human use has been complied with all the relevant national regulations, institutional policies and in accordance with the tenets of the Helsinki Declaration, and has been approved by the ethical committee of Zhejiang Provincial People’s Hospital (Affiliate People’s Hospital, Hangzhou Medical College) (No. 2025 − 194). Written informed consent was obtained from the patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.