Deciphering the Threshold and Segmented Nonlinear Association Between Systemic Inflammation Response Index and Spinal Bone Density: Insights From a Large-Scale Population Study

Abstract

Study Design

Cross-sectional study.

Objective

To examine the association between SIRI and spinal BMD and assess the influence of age, hypertension, and diabetes.

Methods

We analyzed data from 13,950 participants aged ≥20 years. SIRI was calculated using neutrophil, monocyte, and lymphocyte counts, and spinal BMD was measured by DXA. Linear regression, generalized additive models, and segmented regression were used, with subgroup analyses based on age, hypertension, and diabetes.

Results

A threshold effect was observed at SIRI = 0.68. Below this threshold, SIRI negatively correlated with spinal BMD (β = −0.0412, P = 0.0494), while above it, a positive correlation was found (β = 0.0079, P < 0.0001). Subgroup analyses showed stronger positive associations in older adults (≥65 years, β = 0.0136, P < 0.0001), and those with hypertension (β = 0.0089, P = 0.0004) and diabetes (β = 0.0187, P < 0.001).

Conclusion

A segmented nonlinear relationship exists between SIRI and spinal BMD, with age, hypertension, and diabetes as significant modifiers. SIRI may serve as a biomarker for osteoporosis risk.

Keywords

systemic inflammation response index spinal bone mineral density osteoporosis inflammation biomarkers population-based study bone health

Introduction

Osteoporosis is a prevalent global public health issue, particularly among older adults,¹ leading to increased risks of fractures, restricted mobility, and a substantial decline in quality of life, thereby placing a significant burden on patients, families, and health care systems.^2-6 As a crucial indicator of bone health, spinal bone mineral density (BMD) is not only a primary criterion for diagnosing osteoporosis but is also directly linked to fracture risk. Understanding the factors that influence spinal BMD and the underlying biological mechanisms is therefore essential for the early prevention and intervention of osteoporosis.^7,8 In recent years, inflammation has gained considerable attention as an influential factor in bone health. Numerous studies indicate that chronic low-grade inflammation is closely associated with various metabolic diseases and may play an essential role in bone metabolism, particularly by accelerating bone loss and consequently reducing bone density.^9,10 However, the specific mechanisms linking inflammation to bone health remain unclear, warranting further investigation.

The Systemic Immune-Inflammation Response Index (SIRI), calculated from neutrophil, monocyte, and lymphocyte counts, is an integrative marker of inflammation.^11,12 In recent years, SIRI has shown high sensitivity and reliability as a marker of systemic inflammation burden in assessing and predicting risks for cardiovascular disease, metabolic syndrome, diabetes, and cancer.^13-16 Despite its promising applications in these fields, the relationship between SIRI and spinal BMD, especially in terms of osteoporosis risk, has yet to be thoroughly explored and widely discussed. The purpose of this study is to investigate the association between SIRI and spinal BMD in a large population sample, with a specific focus on potential segmented nonlinear relationships and threshold effects. This study aims to provide new insights and perspectives on the role of inflammation in bone health.

This study draws on data from the 2005-2018 cycles of the National Health and Nutrition Examination Survey (NHANES), a nationally representative program employing a multistage stratified sampling design.¹⁷ By analyzing the relationship between SIRI and spinal BMD in a large general population, we aim to assess the independent impact of systemic inflammation on bone health after adjusting for key demographic and clinical variables. Using multivariable regression and nonlinear modeling approaches, we further explore whether this association follows a threshold or segmented pattern. These findings may offer new insights into the inflammatory mechanisms of osteoporosis and support the potential utility of SIRI as a biomarker for bone health monitoring.

Methods

Data Source

We used demographic data from the NHANES database in the United States. Conducted every two years since 1999, this survey employs a multi-stage, stratified sampling method to ensure representation across all age groups in the U.S. population. During the survey, participants provide informed consent and complete questionnaires with assistance from trained staff, followed by physical examinations conducted in mobile labs. Ethical approval is provided by the National Center for Health Statistics (NCHS). For our analysis, we selected data from the 2005 to 2018 cycles, which included demographic information, questionnaire responses, physical exam findings, lab results, and imaging data. This publicly accessible data is available at https://www.cdc.gov/nchs/nhanes/about_nhanes.htm.

Study Design

This study utilized data from five NHANES cycles: 2005-2006, 2007-2008, 2009-2010, 2013-2014, and 2017-2018, with a total of 50,463 participants. Initially, individuals without spinal BMD data from Dual-Energy X-ray Absorptiometry (DXA) were excluded, as well as those lacking the required data to compute SIRI (neutrophil, monocyte, and lymphocyte counts). Participants under 20 years of age were also excluded. The final analysis included 13,950 participants (Figure 1).

Figure 1.

Flowchart of participant selection and analytical steps.

Exposure and Outcome Variables

SIRI is calculated as (neutrophil count × monocyte count)/lymphocyte count, with these cell counts obtained from laboratory blood tests. Total spinal BMD is measured by DXA. Based on SIRI quartiles, the 13,950 participants were divided into four groups: Q1, Q2, Q3, and Q4. This study examined variations in different variables across these quartiles. Although converting continuous variables into quartiles may reduce statistical precision, this approach helps to limit bias, particularly in cases of skewed distributions. Additionally, it facilitates the identification of potential relationships between independent and dependent variables at different levels.

Assessment of Other Covariates

Other covariates included age, sex, race, marital status, education level, smoking status, drinking status, history of hypertension, history of diabetes, waist circumference, weight, height, and body mass index (BMI, kg/m²). Some variables were collected using standardized questionnaires: smoking was defined as having smoked at least 100 cigarettes in one’s lifetime; drinking was defined as consuming at least 12 alcoholic beverages of any type within a single year. Participants who answered “Yes” to the question “Has a doctor ever told you that you have diabetes/hypertension?” were classified as having diabetes or hypertension. Blood samples were analyzed in the laboratory to determine total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), and low-density lipoprotein cholesterol (LDL-C).

Statistical Analysis

In this study, continuous variables are presented as means (standard deviation [SD]) or medians with interquartile ranges (25th and 75th percentiles), while categorical variables are expressed as frequencies and percentages (n, %). Differences in normally distributed continuous variables were evaluated using independent-sample T-tests or one-way ANOVA. For non-normally distributed continuous variables, the Wilcoxon-Mann-Whitney rank-sum test was used. Chi-square tests were applied to assess differences in categorical variables. Differences across SIRI quartiles were examined. To investigate the association between SIRI and total spinal BMD, multivariable stepwise regression models were used to estimate β coefficients and 95% confidence intervals (CIs). Additionally, GAM and curve fitting, implemented with the mgcv package in R, were used to assess the potential impact of SIRI on total spinal BMD. The segmented package in R was applied to explore potential saturation effects and to identify breakpoint values via threshold analysis. Subgroup analyses were performed to evaluate whether the relationship between SIRI and spinal BMD varied across specific population characteristics, including age, gender, smoking, alcohol consumption, hypertension, and diabetes. Interaction terms were included to assess statistical significance for effect modification. The log-likelihood ratio test was conducted to compare the standard linear model with the segmented generalized additive model. All statistical analyses were performed in R (version 4.2.3), with statistical significance set at a two-sided P < 0.05. Data management and supplementary statistical analyses were performed using IBM SPSS Statistics 25.

Results

Baseline Characteristics

A total of 13,950 participants were included in this study, with 48.7% male and 51.3% female. Participants were grouped by SIRI quartiles (Q1-Q4), with the lowest quartile serving as the reference group. Significant differences were observed across SIRI groups in various baseline characteristics: sex, age, race, marital status, smoking and drinking status, diabetes and hypertension diagnoses, weight, BMI, waist circumference, LDL-C, HDL-C, and total cholesterol (all P < 0.001); education level (P = 0.007); and triglyceride levels, which showed no statistically significant difference (P = 0.112). As SIRI increased, the proportion of males rose, as did the percentage of non-Hispanic White participants, while education levels showed a slight decrease. Higher SIRI groups had elevated rates of smoking, drinking, diabetes, and hypertension. Spinal bone mineral density also showed significant differences among the groups, with slightly higher bone density observed in the high SIRI group, suggesting a potential link between SIRI and bone health (Table 1).

Table 1.

Baseline Characteristics of Participants Grouped by SIRI Levels: A Cross-Sectional Analysis.

	Q1 (SIRI ≤ 0.68)	Q2 (0.68 < SIRI ≤ 1.00)	Q3 (1.00 < SIRI ≤ 1.43)	Q4 (1.43 < SIRI)	P Value
	N = 3486	N = 3638	N = 3336	N = 3490	P Value
Gender, n (%)					<0.0001
Male	1416 (40.6%)	1649 (45.3%)	1728 (51.8%)	1998 (57.2%)
Female	2070 (59.4%)	1989 (54.7%)	1608 (48.2%)	1492 (42.8%)
Race/ethnicity, n (%)					<0.0001
Mexican American	614 (17.6%)	728 (20.0%)	668 (20.0%)	601 (17.2%)
Other hispanic	318 (9.1%)	393 (10.8%)	308 (9.2%)	280 (8.0%)
Non-hispanic white	983 (28.2%)	1588 (43.7%)	1667 (50.0%)	1998 (57.2%)
Non-hispanic black	1221 (35.0%)	626 (17.2%)	473 (14.2%)	409 (11.7%)
Other race	350 (10.0%)	303 (8.3%)	220 (6.6%)	202 (5.8%)
Education level, n (%)					0.0070
Less than 9th grade	409 (11.7%)	2052 (56.4%)	386 (11.6%)	395 (11.3%)
9-11th grade	511 (14.7%)	215 (5.9%)	486 (14.6%)	544 (15.6%)
High school grad/GED or equivalent	775 (22.2%)	425 (11.7%)	803 (24.1%)	865 (24.8%)
Some college or AA degree	972 (27.9%)	131 (3.6%)	931 (27.9%)	1013 (29.0%)
College graduate or above	818 (23.5%)	541 (14.9%)	725 (21.7%)	671 (19.2%)
Refused	0 (0.0%)	3 (0.1%)	1 (0.0%)	1 (0.0%)
Don’t know	1 (0.0%)	1 (0.0%)	4 (0.1%)	1 (0.0%)
Marital status, n (%)					<0.0001
Married	1902 (54.6%)	416 (11.4%)	1860 (55.8%)	1861 (53.3%)
Widowed	196 (5.6%)	552 (15.2%)	240 (7.2%)	324 (9.3%)
Divorced	369 (10.6%)	795 (21.9%)	387 (11.6%)	447 (12.8%)
Separated	118 (3.4%)	1023 (28.1%)	114 (3.4%)	119 (3.4%)
Never married	610 (17.5%)	849 (23.3%)	516 (15.5%)	498 (14.3%)
Living with partner	291 (8.3%)	1 (0.0%)	219 (6.6%)	240 (6.9%)
Refused	0 (0.0%)	2 (0.1%)	1 (0.0%)	1 (0.0%)
Smoking, n (%)					<0.0001
Yes	1312 (37.6%)	1570 (43.2%)	1519 (45.5%)	1917 (54.9%)
No	2174 (62.4%)	2068 (56.8%)	1817 (54.5%)	1573 (45.1%)
Drinking, n (%)					<0.0001
Yes	2357 (67.6%)	2655 (73.0%)	2433 (72.9%)	2631 (75.4%)
No	1129 (32.4%)	983 (27.0%)	903 (27.1%)	859 (24.6%)
Diabetes mellitus, n (%)					<0.0001
Yes	346 (9.9%)	377 (10.4%)	403 (12.1%)	492 (14.1%)
No	3140 (90.1%)	3261 (89.6%)	2933 (87.9%)	2998 (85.9%)
Hypertension, n (%)					<0.0001
Yes	1060 (30.4%)	1121 (30.8%)	1080 (32.4%)	1393 (39.9%)
No	2426 (69.6%)	2517 (69.2%)	2256 (67.6%)	2097 (60.1%)
Age (yrs), mean (SD)	48.10 (15.384)	48.70 (15.808)	49.33 (16.456)	50.650 (17.195)	<0.0001
Weight (kg)	77.4018 (17.92302)	78.3187 (17.88563)	81.0814 (18.73564)	81.626 (19.642)	<0.0001
Standing height (cm), mean (SD)	166.2919 (10.06240)	166.6514 (10.23051)	167.6664 (10.00837)	167.155 (9.887)	<0.0001
Body mass index (kg/m²), mean (SD)	27.9779 (5.84027)	28.1402 (5.73753)	28.7964 (6.02517)	29.188 (6.543)	<0.0001
Waist circumference (cm), mean (SD)	95.0080 (14.23132)	96.4773 (14.10846)	98.8397 (14.75708)	100.041 (15.448)	<0.0001
LDL-cholesterol (mmol/L), mean (SD)	3.0615 (0.92791)	3.0351 (0.89231)	3.0345 (0.92415)	2.959 (0.913)	<0.0001
Direct HDL-cholesterol (mmol/L), mean (SD)	1.4389 (0.42496)	1.3779 (0.41000)	1.3407 (0.40814)	1.336 (0.404)	<0.0001
Triglyceride (mmol/L), mean (SD)	1.4354 (1.4851)	1.5029 (1.14422)	1.5204 (1.14942)	1.706 (1.020)	0.1120
Total cholesterol (mmol/L), mean (SD)	5.1493 (1.07495)	5.1545 (1.08352)	5.1039 (1.04399)	5.057 (1.085)	<0.0001
Spinal BMD (g/cm²), mean (SD)	1.0248 (1.15928)	1.0141 (0.15196)	1.0277 (0.1511)	1.0301 (0.15782)	<0.0001

Exploring Linear and Nonlinear Relationships of SIRI with Spinal BMD

We employed multivariable regression models to assess the linear association between SIRI and spinal BMD. The crude model (Model 1, Table 2) indicated a significant positive linear association between SIRI and spinal BMD, where each unit increase in SIRI corresponded to an increase of 0.0057 in spinal BMD (95% CI: 0.0027 to 0.0087, P = 0.0002). After adjusting for sex, race, marital status, education level, smoking status, drinking status, history of diabetes, and history of hypertension, the minimally adjusted model (Model 2, Table 2) still demonstrated a significant linear effect of SIRI on spinal BMD (β = 0.0052, 95% CI: 0.0023 to 0.0082, P = 0.0005). In the fully adjusted model (Model 3, Table 2), which included additional continuous variables such as age, weight, height, BMI, waist circumference (WC), TC, HDL-C, LDL-C, and TG, the linear effect of SIRI on spinal BMD remained significant and even increased (β = 0.0082, 95% CI: 0.0040 to 0.0124, P = 0.0001). The regression coefficients for all independent variables are listed in Supplemental 1.

Table 2.

Multivariable Regression Analysis Results of the Relationship Between SIRI and Spine BMD.

	Crude Model (Model 1)		Minimally Adjusted Model (Model 2)		Fully Adjusted Model (Model 3)
	β*(95% CI)	P Value	β*(95% CI)	P Value	β*(95% CI)	P Value
SIRI	0.0057 (0.0027 ∼ 0.0087)	0.0002	0.0052 (0.0023 ∼ 0.0082)	0.0005	0.0082 (0.0040 ∼ 0.0124)	0.0001
SIRI (quartile)
Q1
Q2	−0.0107 (−0.0179 ∼ −0.0035)	0.0035	−0.0023 (−0.0092 ∼ 0.0047)	0.5258	−0.0005 (−0.0096 ∼ 0.0086)	0.9137
Q3	0.0030 (−0.0044 ∼ 0.0104)	0.4266	0.0094 (0.0023 ∼ 0.0166)	0.0101	0.0103 (0.0006 ∼ 0.0199)	0.0379
Q4	0.0054 (−0.0018 ∼ 0.0127)	0.1428	0.0094 (0.0021 ∼ 0.0167)	0.0111	0.0124 (0.0025 ∼ 0.0223)	0.0145
P for trend	0.0110		0.7734		0.1498

β*: The β coefficient is used to explain the effect of the independent variable on the dependent variable. When the independent variable (e.g., SIRI value) increases by one unit, the expected change in the dependent variable (spinal BMD), while controlling for other variables, is represented by the value of the β coefficient.

Model 1: An initial model without any covariate adjustment, containing only the original effect of SIRI on spinal BMD.

Model 2: Based on Model 1, all categorical variables were added, including sex, race, marital status, education level, smoking status, drinking status, diabetes history, and hypertension history, to control for the confounding effects of basic demographic and health status variables on spinal BMD.

Model 3: Building on Model 2, all continuous variables were further included: age, weight, height, BMI, WC, TC, HDL-C, LDL-C, and TG. This model aims to fully adjust for all potential confounding factors, thereby evaluating the independent effect of SIRI under comprehensive control.

Although the multivariable regression models indicated a significant relationship between SIRI and spinal BMD, the linear trend test (Table 2) suggested that this association might not be strictly linear. In the crude model, the linear trend P-value was 0.011, reaching statistical significance. However, in the minimally adjusted model (P = 0.7734) and the fully adjusted model (P = 0.1498), this trend did not reach significance. To further explore the potential nonlinear relationship between SIRI and spinal BMD, we employed a GAM (Figure 2).

Figure 2.

Adjusted associations between SIRI and spinal BMD using GAM (A) unadjusted model - The smooth function displays the baseline association between SIRI and spinal BMD without adjustment for any covariates. (B) Minimally adjusted model - Adjusted for basic demographic characteristics, including sex, race, marital status, education level, smoking and drinking status, history of diabetes, and hypertension. Shows the independent effect of SIRI on spinal BMD. (C) Fully adjusted model - Adjusted for all key covariates, including sex, race, marital status, education level, lifestyle factors (smoking and drinking), diabetes and hypertension history, as well as age, weight, height, BMI, WC, TC, HDL-C, LDL-C, and TG, showing the effect of SIRI on spinal BMD with comprehensive variable control.

The GAM results revealed a significant nonlinear association between SIRI and spinal BMD (P < 0.001). As shown in panels (A), (B), and (C) of Figure 2, different levels of adjustment produced subtle variations in the trend. In the unadjusted model (Figure 2A), spinal BMD showed a slight increase at lower SIRI levels, followed by a more consistent upward trend at higher SIRI values. The minimally adjusted model (Figure 2B) indicated a similar pattern, with spinal BMD showing an initial dip at moderate SIRI levels before trending upward. The fully adjusted model (Figure 2C) confirmed a more complex nonlinear relationship, with spinal BMD initially decreasing and then sharply increasing as SIRI levels rose. This trend suggests that higher SIRI levels may have a more pronounced positive effect on spinal BMD, particularly when other covariates are comprehensively controlled for.

Segmented Regression and Nonlinear Analysis Results

To further investigate the potential segmented effect, we constructed a two-segment regression model (Model 2, Table 3). The results showed that the β coefficients for SIRI’s effect on total spinal BMD before and after the breakpoint of 0.68 were −0.0412 (P = 0.0494) and 0.0079 (P < 0.0001), respectively. After the breakpoint, the positive impact of SIRI on spinal BMD significantly increased, reaching statistical significance. This indicates a significant difference in the relationship between SIRI and spinal BMD across the breakpoint. Additionally, the log-likelihood ratio test indicated that the segmented model was statistically superior to the standard linear model (P = 0.0008, Table 3).

Table 3.

Results of Standard Linear Regression and Segmented Regression Analysis of SIRI and Total Spine BMD.

Model	Total Spine BMD Adjusted β (95% CI)	P Value
Model 1
The standard linear model	0.0057 (0.0027 ∼ 0.0087)	0.0002
Model 2
Turning point (K)	0.68
SIRI <0.68	−0.0412 (−0.0822 ∼ −0.0001)	0.0494
SIRI ≥0.68	0.0079 (0.0045 ∼ 0.0113)	<0.0001
Log likelihood ratio test	0.0008

Model 1 was a standard linear model, and model 2 was a piecewise model. Log likelihood ratio test was used to compare two models.

Overall, the standard linear model, GAM, and segmented regression models consistently indicated a positive association between SIRI and spinal BMD, with this effect becoming more apparent at higher SIRI levels (Figure 2). The segmented regression analysis showed that at lower SIRI levels (<0.68), the effect of SIRI on spinal BMD was negative, but beyond this threshold (≥0.68), the effect turned positive and significantly increased. This finding supports the hypothesis of a nonlinear or segmented effect of SIRI on spinal BMD, suggesting that the impact becomes more pronounced as SIRI levels rise.

Subgroup Analysis Results

Subgroup analysis revealed notable differences in the association between SIRI and spinal BMD across various population characteristics (Table 4). Overall, a significant positive relationship was observed in the entire cohort (β = 0.0082, 95% CI: 0.0040 ∼ 0.0124, P = 0.0001). When stratified by age, a stronger association was observed in participants aged ≥65 years (β = 0.0136, 95% CI: 0.0070 ∼ 0.0203, P < 0.0001) compared to those aged <65 years (β = 0.0064, 95% CI: 0.0030 ∼ 0.0098, P < 0.0001), with a significant interaction effect (P for interaction = 0.0241).

Table 4.

Subgroup Analysis of the Association Between SIRI and Spinal BMD.

Variables	n (%)	β*(95% CI)	P	P for Interaction
All patients	13950 (100.00%)	0.0082 (0.0040 ∼ 0.0124)	0.0001
Age				0.0241
Age≥ 65	2835 (20.32%)	0.0136 (0.0070 ∼ 0.0203)	<0.0001
Age< 65	11115 (79.68%)	0.0064 (0.0030 ∼ 0.0098)	<0.0001
BMI				0.3141
BMI <27	6354 (45.55%)	0.0031 (−0.0013 ∼ 0.0074)	0.1710
BMI ≥27	7596 (54.45%)	0.0061 (0.0021 ∼ 0.0101)	0.0031
Gender				0.8512
Male	6791 (48.68%)	0.0010 (−0.0026 ∼ 0.0047)	0.5521
Female	7159 (51.32%)	0.0031 (−0.0020 ∼ 0.0082)	0.2282
Smoking				0.9481
Yes	6318 (45.29%)	0.0039 (−0.0001 ∼ 0.0080)	0.0559
No	7632 (54.71%)	0.0063 (0.0017 ∼ 0.0108)	0.0069
Alcohol consumption				0.4721
Yes	10076 (72.23%)	0.0056 (0.0020 ∼ 0.0091)	0.0020
No	3874 (27.77%)	0.0030 (−0.0026 ∼ 0.0086)	0.2923
Hypertension				0.0312
Yes	4654 (33.36%)	0.0089 (0.0040 ∼ 0.0139)	0.0004
No	9296 (66.64%)	0.0021 (−0.0017 ∼ 0.0060)	0.2748
Diabetes mellitus				<0.0001
Yes	1618 (11.60%)	0.0187 (0.0102 ∼ 0.0272)	<0.0001
No	12332 (88.40%)	0.0029 (−0.0004 ∼ 0.0061)	0.0818

β*: Regression coefficient indicating the change in spinal BMD associated with each unit increase in SIRI within the respective subgroup.

CI: Confidence interval, representing the range of values within which the true regression coefficient is expected to fall with 95% confidence.

P: P-value indicating the statistical significance of the association within each subgroup.

P for interaction: P-value assessing the statistical significance of interaction between SIRI and subgroup variables on spinal BMD.

The relationship between SIRI and spinal BMD also varied significantly by hypertension status (P for interaction = 0.0312). In participants with hypertension, a significant positive association was identified (β = 0.0089, 95% CI: 0.0040 ∼ 0.0139, P = 0.0004), while no significant association was observed in those without hypertension (P = 0.2748). Similarly, diabetes mellitus emerged as a key modifier (P for interaction <0.0001). Among participants with diabetes, the association was particularly pronounced (β = 0.0187, 95% CI: 0.0102 ∼ 0.0272, P < 0.001), whereas no significant relationship was observed in non-diabetic individuals (P = 0.0818).

For other subgroup variables, including gender, smoking status, alcohol consumption, and BMI categories, no significant interaction effects were observed (P for interaction >0.05). However, the general trends remained consistent across these groups. These findings underscore the potential role of age, hypertension, and diabetes as modifiers of the relationship between SIRI and spinal BMD, while also suggesting that this association remains robust across different levels of body weight. Together, they highlight the complex and multifactorial influence of systemic inflammation on bone health.To further investigate age-specific and sex-specific variations, we stratified participants into four age groups (<60, 60-69, 70-79, and ≥80 years) and conducted subgroup analyses by sex within each age group. As presented in Supplemental 2, a significant positive association between SIRI and spinal BMD was observed among females aged <60 years (β = 0.0093, 95% CI: 0.0035 to 0.0150, P = 0.0016), while no significant associations were found in older female subgroups. Among males, the associations across age groups were generally weaker and nonsignificant. Notably, in the 70-79 years subgroup, the direction of the association differed between males (β = −0.0030, 95% CI: −0.0160 to 0.0101, P = 0.6524) and females (β = 0.0010, 95% CI: −0.0148 to 0.0167, P = 0.9045), although no significant interaction effect by sex was detected (P for interaction = 0.7052). These results suggest that the SIRI–BMD relationship may vary subtly across age and sex strata, particularly among younger female participants.

Discussion

BMD is a crucial indicator of bone health, widely used for diagnosing osteoporosis and assessing fracture risk.^18-21 In recent years, attention has increasingly focused on the impact of inflammation on bone metabolism, especially with the SIRI as a comprehensive marker reflecting an individual’s inflammatory burden.^22,23 SIRI has demonstrated high sensitivity in predicting the risk of several diseases, including cardiovascular disease, metabolic syndrome, diabetes, and cancer.^24-26 However, the relationship between SIRI and spinal BMD, particularly any potential nonlinear associations, remains underexplored. This study is the first to uncover a segmented nonlinear relationship between SIRI and spinal BMD, identifying a significant threshold effect, thus providing new insights into the mechanisms by which inflammation influences bone metabolism.

This study utilized a large population sample from the NHANES database to systematically analyze the relationship between SIRI and spinal BMD, revealing a complex association in a stepwise manner. Initially, we employed linear regression analysis, progressively adjusting for variables to examine the overall impact of SIRI on spinal BMD. Results indicated that the initial unadjusted model showed a significant positive correlation between SIRI and spinal BMD. After adjusting for categorical variables such as sex, race, and marital status, the positive trend remained significant. Further adjustments for continuous variables like age, weight, BMI, and waist circumference strengthened the positive impact of SIRI on spinal BMD. Subsequently, we used a GAM to explore the nonlinear relationship between SIRI and spinal BMD. Results demonstrated that SIRI’s effect on spinal BMD exhibited a nonlinear pattern: within the lower SIRI range, spinal BMD showed minimal change with SIRI variations, whereas at higher SIRI levels, spinal BMD increased significantly. To further confirm this segmented characteristic, we conducted a segmented regression analysis, dividing SIRI into two intervals for linear fitting. The analysis revealed that when SIRI was below 0.68, it was negatively correlated with spinal BMD, whereas above 0.68, the correlation was significantly positive.

In addition to these findings, subgroup analyses further illuminated how specific population characteristics modified the association between SIRI and spinal BMD. The positive relationship was stronger among older adults (≥65 years), individuals with hypertension, and those with diabetes, suggesting that these populations may experience heightened or more complex inflammatory responses influencing bone metabolism. For example, the association in participants aged ≥65 years (β = 0.0136) was more pronounced than in younger individuals (β = 0.0064). Similarly, participants with hypertension or diabetes showed stronger positive correlations compared to those without these conditions, highlighting potential population-specific differences in how inflammation impacts bone density. These subgroup findings underscore the multifaceted nature of inflammation’s effects on bone metabolism, which may be influenced by underlying health conditions or age-related changes in inflammatory responses.

Collectively, these findings support the hypothesis of a segmented association between SIRI and spinal BMD, offering new perspectives on the complex and bidirectional role of inflammation in bone metabolism. They also underscore the need to consider individual characteristics, such as age and comorbidities, when evaluating the skeletal impact of systemic inflammation. From a clinical standpoint, inflammation-driven ossification plays a key role in conditions such as spondylodiscitis and cervical arthroplasty, where excessive bone formation may complicate healing and spinal fusion outcomes. Anti-inflammatory medications are commonly employed in the acute phase to minimize such complications. These observations reinforce the potential utility of inflammatory biomarkers like SIRI, not only for risk stratification in osteoporosis, but also for guiding individualized treatment decisions in inflammation-related spinal disorders. To more intuitively demonstrate the nonlinear relationship between SIRI and spinal BMD, we created a scatter plot and fitted curve in Figure 3 to further validate the segmented association observed in our analysis. Figure 3 illustrates the overall trend between SIRI and spinal BMD, as well as the segmented relationships at different SIRI levels. In Figure 3A, a slight upward trend is observed across the overall range of SIRI and spinal BMD. To examine this segmented effect more closely, we further divided the data, as presented in Figure 3B and C. Figure 3B shows the situation when SIRI is below 0.68, where the relationship between SIRI and spinal BMD is slightly negative, aligning with our threshold effect hypothesis. Figure 3C displays the relationship when SIRI is above 0.68, showing a significant positive correlation between SIRI and spinal BMD, suggesting that higher SIRI levels may promote bone density. These visualizations clearly support the segmented nonlinear association between SIRI and spinal BMD, providing key visual evidence to enhance our understanding of their relationship.

Figure 3.

Scatter plot of the segmented relationship between SIRI and Spinal BMD (A) Displays the overall association between SIRI and spinal BMD, illustrating the general trend across the full range of SIRI values. (B) Shows the relationship between SIRI and spinal BMD for SIRI values below the identified threshold of 0.68. This segment indicates a slight negative association, aligning with the threshold effect observed in the analysis. (C) Illustrates the association for SIRI values equal to or above 0.68, where a positive relationship between SIRI and spinal BMD becomes evident, supporting the hypothesis that higher SIRI levels are associated with higher spinal BMD values.

This study provides significant insights into the relationship between inflammation and bone metabolism, marking the first to reveal a segmented nonlinear relationship between SIRI and spinal BMD. This finding not only suggests a potential biomarker for osteoporosis screening and individualized risk assessment but also highlights the complex, multi-level impact of inflammatory burden on bone density.

We propose that low-level inflammation may contribute to bone resorption by activating osteoclast activity, thereby accelerating bone loss and reducing spinal BMD. Chronic low-grade inflammation has been shown to promote osteoclast differentiation, enhancing bone resorption without adequate compensatory bone formation, which can negatively impact bone density.²⁷ Inflammatory cytokines, such as TNF-α and IL-6, are often induced by low levels of inflammation and have been observed to activate osteoclasts, contributing to bone loss and supporting the view that low-level inflammation detrimentally affects spinal BMD.^28,29 Additionally, pro-inflammatory cytokines like IL-1β and IL-6 are known to enhance osteoclast differentiation and activity under low inflammatory conditions, reinforcing the idea that low-grade inflammation increases bone resorption by influencing osteoclast activity.^30-34

However, when inflammation reaches higher levels, compensatory mechanisms may be triggered within the bone microenvironment. For instance, increased osteoblast activity or the upregulation of anti-inflammatory factors may regulate bone metabolism and mitigate some of the negative effects of inflammation.^34,35 At this stage, the adverse impact of inflammation on bone density may be reduced or even reversed, potentially resulting in an increase in spinal BMD. High levels of pro-inflammatory cytokines can, in some cases, trigger a compensatory osteoblastic response, promoting bone formation and potentially enhancing spinal BMD.^36,37 Furthermore, elevated inflammatory markers have been shown to stimulate adaptive responses within the skeletal system, enhancing osteoblast activity and raising spinal BMD. At higher inflammation levels, specific cytokines, such as IL-1β and RANKL, may simultaneously activate both osteoblasts and osteoclasts, with adaptive responses making osteoblast activity more prominent under these conditions.^38,39

This segmented relationship suggests that the impact of inflammation on bone metabolism exhibits dual characteristics depending on its intensity. Low-level inflammation may accelerate bone resorption, whereas high-level inflammation may under certain conditions promote bone formation. Given SIRI’s sensitivity to inflammatory burden, it may serve as an effective biomarker for dynamically monitoring these varied effects on bone density. This discovery provides new insights into the complex role of inflammation in bone health and lays a scientific foundation for the application of inflammatory biomarkers in osteoporosis prevention and management.

In future research, validating this mechanism will further deepen our understanding of the complex role of inflammation in bone metabolism. Specifically, future studies could utilize longitudinal designs to observe the relationship between changes in SIRI and dynamic spinal BMD variations, thereby determining a causal relationship between the two. Additionally, experimental studies could focus on the regulatory mechanisms of osteoclast and osteoblast activity under varying levels of inflammation, especially examining the specific roles of pro-inflammatory and anti-inflammatory factors within the bone marrow microenvironment. Further exploration of the synergistic effects between SIRI and other inflammatory markers (such as NLR and PLR) would also help clarify the multifaceted impact of inflammation on bone metabolism. Through animal models and in vitro experiments, specific molecular pathways (such as the RANKL/OPG and Wnt signaling pathways) can be observed in inflammatory conditions to delineate the precise influence of inflammation on bone resorption and formation. Additionally, with the advancement of multi-omics technologies, future research could integrate genomics, proteomics, and metabolomics to construct a comprehensive map of the effects of different inflammation levels on bone health. This approach could help identify potential molecular biomarkers and advance novel strategies for osteoporosis prevention and treatment, moving toward more personalized management of bone health.

Despite revealing a segmented nonlinear association between SIRI and spinal BMD, this study has certain limitations. First, as a cross-sectional study, it cannot establish causality, making it unclear whether changes in SIRI directly lead to changes in spinal BMD. Second, although several confounding variables (such as age, sex, ethnicity, lifestyle, and history of chronic disease) were controlled, there may still be unmeasured confounders, such as diet, nutritional intake, and physical activity, that could influence spinal BMD. Third, as this study is based on a U.S. population sample from the NHANES database, its generalizability may be limited, especially in applicability to other ethnicities and regions. Finally, the study relies on SIRI as a single indicator; future research could incorporate multiple inflammation-related markers to better understand the complex impact of inflammation on bone density.

In conclusion, this study identifies a segmented nonlinear relationship between SIRI and spinal BMD, with a significant threshold effect that offers new insights into the role of inflammation in bone health. Although there are certain limitations, this study provides a scientific basis for applying inflammatory biomarkers in osteoporosis risk assessment. Future longitudinal and mechanistic studies are needed to further validate and expand upon these findings.

Conclusion

This study reveals a segmented nonlinear relationship between SIRI and spinal BMD, highlighting a threshold effect that offers new insights into how inflammation may impact bone health. While low levels of SIRI were associated with decreased spinal BMD, likely through increased osteoclast activity, higher SIRI levels appeared linked to compensatory osteoblast activation. Although causality cannot be confirmed due to the cross-sectional design, these findings suggest SIRI’s potential as a biomarker for osteoporosis risk. Future longitudinal studies are encouraged to validate this relationship and explore targeted strategies for bone health management based on individual inflammatory profiles.

Supplemental Material

Supplemental Material - Deciphering the Threshold and Segmented Nonlinear Association Between Systemic Inflammation Response Index and Spinal Bone Density: Insights From a Large-Scale Population Study

Supplemental Material for Deciphering the Threshold and Segmented Nonlinear Association Between Systemic Inflammation Response Index and Spinal Bone Density: Insights From a Large-Scale Population Study by Muradil Mardan, Zeyu Lu, Qingyin Xu, Shaokuan Song, Huoliang Zheng, Hui Deng, Hao Cai, Qizhu Chen, Pengbo Chen, Bo Li, Shengdan Jiang, Leisheng Jiang, and Xin-Feng Zheng in Global Spine Journal.

Footnotes

Acknowledgments

We would like to express our gratitude to all those who contributed to this work. We also acknowledge the helpful discussions and insights provided by colleagues in the field. No specific acknowledgments are required for this study.

Author Contributions

Each author has participated in the drafting and critical revision of the manuscript and has given approval for the final version.

Declaration of Conflicting Interest

The authors declare no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Shanghai Sailing Program (23YF1425900) and the National Natural Science Foundation of China (82272535).

Ethical Statement

ORCID iDs

Muradil Mardan

Ze-yu Lu

Bo Li

Sheng-dan Jiang

Lei-sheng Jiang

Xin-Feng Zheng

Data Availability Statement

The data used in this study are publicly available from the NHANES database. Further details can be accessed at .

Supplemental Material

Supplemental material for this article is available online.

References

Bisignano

James

, et al. Global, regional, and national burden of bone fractures in 204 countries and territories, 1990–2019: a systematic analysis from the global burden of disease study 2019. The Lancet Healthy Longevity. 2021;2(9):e580-e592.

Voltan

Di Giovannantonio

Carretta

, et al. A novel case-finding strategy based on artificial intelligence for the systematic identification and management of individuals with osteoporosis or at varying risk of fragility fracture. Arch Osteoporosis. 2024;19(1):45.

Kumar

Beck

Nery

, et al. Study protocol for the ROLEX-DUO randomised placebo-controlled trial: romosozumab loaded with exercise - dual effects on bone and muscle in postmenopausal osteoporosis and osteopenia. BMJ Open. 2024;14(8):e086708.

Clark

Calo

Torres-Naranjo

, et al. Osteoporosis and fragility fractures in Mexico: a call to action. Arch Med Res. 2024;55(7):103062.

Linton

Schaumberg

Wright

. Dietary inflammatory index is not associated with bone mineral density in functionally able community-dwelling older adults. Eur J Nutr. 2024;63(8):3195-3205.

Stenmanns

Moellmann

Wehling

Frohnhofen

. Evaluation of drug prescriptions on hospital admission in older trauma patients using the Fit fOR the Aged (FORTA) rules. Z Gerontol Geriatr. 2024;63(8):609-615.

Reid

Tobin

McCrosson

, et al. Opportunistic computed tomography: a novel opportunity for osteoporosis screening. Clin Spine Surg. 2024.

Makinen

Solling

McClung

Langdahl

. Romosozumab for the treatment of osteoporosis - a systematic review. J Endocrinol Investig. 2024;48(3):547-572.

Henriques

Berenbaum

Mobasheri

. Obesity-induced fibrosis in osteoarthritis: pathogenesis, consequences and novel therapeutic opportunities. Osteoarthr Cartil Open. 2024;6(4):100511.

10.

Wasserman

Dumenigo

Hornung

Augsburger

Marathe

Lucky

. Prevalence of delayed puberty and low bone density in patients with epidermolysis bullosa: insight from a large single center's experience. Pediatr Dermatol. 2023;40(1):100-106.

11.

Wang

Wen

Jiang

, et al. The clinical value of neutrophil-to-lymphocyte ratio (NLR), systemic immune-inflammation index (SII), platelet-to-lymphocyte ratio (PLR) and systemic inflammation response index (SIRI) for predicting the occurrence and severity of pneumonia in patients with intracerebral hemorrhage. Front Immunol. 2023;14:1115031.

12.

Xia

Zhang

. Systemic immune inflammation index (SII), system inflammation response index (SIRI) and risk of all-cause mortality and cardiovascular mortality: a 20-year follow-up cohort study of 42,875 US adults. J Clin Med. 2023;12(3):1128.

13.

Zhuang

Shen

, et al. A novel systemic inflammation response index (SIRI) for predicting the survival of patients with pancreatic cancer after chemotherapy. Cancer. 2016;122(14):2158-2167.

14.

Zhang

Xing

Zhou

Jiang

. The predictive role of systemic inflammation response index (SIRI) in the prognosis of stroke patients. Clin Interv Aging. 2021;16:1997-2007.

15.

Jin

Chen

, et al. The associations of two novel inflammation indexes, SII and SIRI with the risks for cardiovascular diseases and all-cause mortality: a ten-year follow-up study in 85,154 individuals. J Inflamm Res. 2021;14:131-140.

16.

Zhang

Cheng

. Prognostic and clinicopathological value of systemic inflammation response index (SIRI) in patients with breast cancer: a meta-analysis. Ann Med. 2024;56(1):2337729.

17.

Johnson

Lamontagne

Bhupathiraju

, et al. Workshop summary: building an NHANES for the future. Am J Clin Nutr. 2024;119(4):1075-1081.

18.

Zhang

Yang

, et al. Correlation study of multiple inflammatory indices and vertebral compression fracture: a cross-sectional study. J Clin Transl Endocrinol. 2024;37:100369.

19.

Greere

Haydar

Grigorescu

, et al. Fine-scale haplotype mapping reveals an association of the FTO gene with osteoporosis and fracture risk in postmenopausal women. Genes. 2024;15(9):1152.

20.

Chandanwale

. Assessing the correlation between anthropometric measurements and bone densitometry as indicators of bone health in adult women in the community. Cureus. 2024;16(8):e68162.

21.

Bowers

Gowland

Hind

. Rickets, resorption and revolution: an investigation into the relationship between vitamin D deficiency in childhood and osteoporosis in adulthood in an 18th-19th century population. Int J Paleopathol. 2024;47:27-42.

22.

Chen

Shi

, et al. Systemic inflammation markers associated with bone mineral density in perimenopausal and postmenopausal women. J Inflamm Res. 2023;16:297-309.

23.

Zhou

Cui

, et al. Frailty phenotype as mediator between systemic inflammation and osteoporosis and fracture risks: a prospective study. J Cachexia Sarcopenia Muscle. 2024;15(3):897-906.

24.

Yang

Fei

. Prognostic value of systemic immune-inflammation index and systemic inflammation response index for oral cancers: a systematic review and meta-analysis. Med Oral Patol Oral Cir Bucal. 2024;29(6):e822-e831.

25.

Zhao

Zhai

Ren

Zhang

Han

. Systemic immune inflammatory index (SII) and systemic inflammatory response index (SIRI) as predictors of postoperative delirium in patients undergoing off-pump coronary artery bypass grafting (OPCABG) with cerebral infarction. BMC Surg. 2024;24(1):338.

26.

Efgan

Acar

Kanter

Kirik

Duman

. Role of systemic immune inflammation index, systemic immune response index, neutrophil lymphocyte ratio and platelet lymphocyte ratio in predicting peritoneal culture positivity and prognosis in cases of spontaneous bacterial peritonitis admitted to the emergency department. Medicina (Kaunas). 2024;60(8):1335.

27.

Cao

Gregoire

Shen

. A high-fat diet decreases bone mass in growing mice with systemic chronic inflammation induced by low-dose, slow-release lipopolysaccharide pellets. J Nutr. 2017;147(10):1909-1916.

28.

Korczowska

Olewicz-Gawlik

Trefler

Hrycaj

Krzysztof

. Does low-dose and short-term glucocorticoids treatment increase the risk of osteoporosis in rheumatoid arthritis female patients? Clin Rheumatol. 2008;27(5):565-572.

29.

Abou-Saleh

Ouhtit

Halade

Rahman

. Bone benefits of fish oil supplementation depend on its EPA and DHA content. Nutrients. 2019;11(11):2701.

30.

Ilesanmi-Oyelere

Schollum

Kuhn-Sherlock

, et al. Inflammatory markers and bone health in postmenopausal women: a cross-sectional overview. Immun Ageing. 2019;16:15.

31.

Langan

Perkins

Vogel

Moudgil

. Microbiota-derived metabolites, indole-3-aldehyde and indole-3-acetic acid, differentially modulate innate cytokines and stromal remodeling processes associated with autoimmune arthritis. Int J Mol Sci. 2021;22(4):2017.

32.

Hua

Liao

. Weakening of M1 macrophage and bone resorption in periodontitis cystathionine gamma-lyase-deficient mice. Oral Dis. 2024;30(2):769-779.

33.

Nakamura

Towheed

. Pathogenesis, assessment, and management of bone loss in axial spondyloarthritis. Semin Arthritis Rheum. 2024;64:152345.

34.

Lin

Jiang

Lou

, et al. Asiatic acid prevents glucocorticoid-induced femoral head osteonecrosis via PI3K/AKT pathway. Int Immunopharmacol. 2024;130:111758.

35.

Tang

Wang

, et al. microRNA-199a counteracts glucocorticoid inhibition of bone marrow mesenchymal stem cell osteogenic differentiation through regulation of Klotho expression in vitro. Cell Biol Int. 2020;44(12):2532-2540.

36.

Raehtz

Bierhalter

Schoenherr

Parameswaran

McCabe

. Estrogen deficiency exacerbates type 1 diabetes-induced bone TNF-alpha expression and osteoporosis in female mice. Endocrinology. 2017;158(7):2086-2101.

37.

Hung

Chang

Hsu

, et al. Therapeutic effect of calcimimetics on osteoclast-osteoblast crosslink in chronic kidney disease and mineral bone disease. Int J Mol Sci. 2020;21(22):8712.

38.

Chen

Gou

, et al. Integrating transcriptomic and proteomic data for a comprehensive molecular perspective on the association between sarcopenia and osteoporosis. Arch Gerontol Geriatr. 2024;125:105486.

39.

Zhang

Sun

Huang

, et al. The chromatin remodeling factor Arid1a cooperates with Jun/Fos to promote osteoclastogenesis by epigenetically upregulating Siglec15 expression. J Bone Miner Res. 2024;39(6):775-790.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.26 MB