Dietary Inflammatory Index is intensity-dependently associated with the Bone Mineral Density at the Lumbar Spine: A Cross- Sectional NHANES Study from 2007 to 2020

doi:10.21203/rs.3.rs-4964246/v1

Background

Bone mineral density (BMD) is a tool to assess bone health. Dietary inflammation index (DII) is an index to reflect the inflammatory effect of a diet. Even though the association between DII and BMD have been extensively studied, the outcome remains controversial. This study aims to reveal the relationship between DII and lumbar spine BMD in both the whole population and in postmenopausal women.

Methods

We utilized data from the National Health and Nutrition Examination Survey (NHANES) 2007–2020 and included 27027 participants after exclusion procedure. Smooth curve fitting and multivariate regression analysis were used both in the entire population and later in the subgroup analysis.

Results

In the entire population, the smooth curve fitting revealed a non-linear association with both declining and increasing trends between DII and lumbar spine BMD. Multivariate regression analysis indicated that, after adjustment, the median DII and BMD was negatively associated and the high DII and BMD was positively associated. In postmenopausal women, smooth curve fitting suggested an inverted U-shaped relationship between DII and BMD. We also observed a positive correlation between low DII and lumbar spine BMD.

Conclusions

DII has both positive and negative relationship with lumbar spine BMD in both the entire population and menopausal women, depending on the intensity of DII. The different trends of the smooth curve fitting may be explained by the low-inflammation state in postmenopausal women mediated by estrogen but more evidence is needed to explain the reason.

NHANES

dietary inflammatory index

lumbar spine

bone mineral density

postmenopausal women

Bone mineral density (BMD) at the femoral neck or spine assessed via dual-energy X-ray absorptiometry (DXA) serves as a quantitative measure for calculating the T-score which is critical for diagnosing osteoporosis (1). Bone loss is a prevalent senile condition in the United States and data from the 2005–2010 National Health and Nutrition Examination Survey (NHANES) indicate that approximately 8.2 million females and 2.0 million males aged 50 and older suffer from osteoporosis, with an additional 27.3 million women and 16.1 million men exhibiting low bone mass (2).

The dietary inflammatory index (DII) is an innovative tool designed to evaluate the impact of a diet on inflammation, utilizing 45 food parameters. Each component to calculate DII can positively or negatively influence specific inflammatory markers, including interlukin-1𝛽 (IL-1𝛽), interlukin-4 (IL-4), interlukin-6 (IL-6), interlukin-10 (IL-10), tumor necrosis factor-α (TNF-α), and C-reactive protein (CRP). Diets are classified as pro-inflammatory with a positive DII score, which correlates with elevated levels of these biomarkers, or as anti-inflammatory with a negative DII score, associated with reduced levels of inflammatory cytokines (3).

Even though the impact of inflammatory biomarkers on bone resorption and osteogenesis have been extensively studied (4–6), the association between DII and lumbar spine BMD remains controversial. Despite a research has established a negative relationship between a pro-inflammatory diet and the lumbar spine BMD (7), some studies have reported there was no significant association between them (8, 9). The different results in these studies may be attributed to the methodological limitations of the research. Some studies did not exclude postmenopausal women who are younger than the age limitation that was set at the beginning of the research, and some studies failed to include sufficient cofounders in their multivariate regression models.

We made several methodological improvements in this study. First of all, we treated DII as a continuous variable, and analyzed the relationship between it and BMD across different DII tertiles. Second, our analysis incorporated a broad population by utilizing NHANES data from 2007 to 2020. Third, we excluded participants in conditions that were likely to cause bias the outcomes. Finally, we included a comprehensive range of confounding factors and employed smooth curve fitting techniques to visualize the relationship between DII and BMD.

2.1 Study design and participants

In this cross-sectional study, we utilized data from NHANES, a nationwide survey conducted by the National Center for Health Statistics at the U.S. Centers for Disease Control and Prevention. NHANES aims to assess the nutritional and health status of the U.S. population and is conducted in two-year cycles with representative sample weights. The study protocol of NHANES has been approved by the institutional review board, and all participants provided written informed consent in accordance with the principles of the Declaration of Helsinki. The data used in the study are available on the CDC (Center for Disease Control and Prevention) website: https://www.cdc.gov/nchs/nhanes/. Clinical trial number: not applicable.

We obtained data from NHANES from 2007 to 2020, which included five complete 2-year cycles and one combined cycle (2017-2020). Our study focused on participants who completed 24-hour dietary recall and underwent DXA testing. The detailed selection and exclusion procedure was as follows: (1) Exclude participants less than 20 years. (2) Exclude pregnant participants. (3) Exclude participants with incomplete dietary data. (4) Exclude participants with unreliable calories intake (men < 500 kcal or > 8000 kcal, women <500 kcal or > 5000 kcal) (5) Exclude participants with self-reported congestive heart failure, coronary heart disease, pulmonary emphysema, chronic bronchitis, chronic obstructive pulmonary disease, cancer or malignancy. (6) Excluded participants with unreliable eGFR (<1% or >99%). A total of 27027 participants (including 6062 postmenopausal women) were finally included in this study and the detailed selecting procedure was presented in Figure 1.

Figure 1 The participants enrollment. A total of 27027 people were included from NHANES (from 2007 to 2020)

2.2 Dietary Data and Dietary inflammatory Index (DII)

We calculated the DII following the methods described by Shivappa et al (3) and the dietary data for calculation were obtained from NAHANES. Our study included 28 out of 45 food parameters including energy, carbohydrate, protein, total fat, dietary fiber, cholesterol, total saturated fatty acids, total monounsaturated fatty acids, total polyunsaturated fatty acids, ω-3 polyunsaturated fatty acids, ω-6 polyunsaturated fatty acids, vitamin A, vitamin B1, vitamin B2, vitamin B3, vitamin B6, vitamin B12, vitamin C, vitamin D, vitamin E, folic acid, alcohol, β-carotene, caffeine, iron, magnesium, zinc and selenium. We treated DII as a continuous variable and stratified it into three tertiles for further statistical analysis.

2.3 Lumbar Bone Mineral Density

BMD was assessed using DXA. For the periods 2007-2010, 2013-2014, and 2017-2020, we calculated the average BMD from lumbar vertebrae L1 to L4 to represent lumbar spine BMD. However, for the years 2011-2012 and 2015-2016, we used total spine BMD measurements as proxies for lumbar spine BMD. Further details on the DXA measurement protocols can be found at www.cdc.gov/nchs/nhanes/.

2.4 Covariates

Covariates in our study included age (in years), systolic blood pressure (SBD, in mmHg), diastolic blood pressure (DSB, in mmHg), estimated glomerular filtration rate (eGFR, in ml/min/1.73 m²), glycohemoglobin (%), total cholesterol (TC, in mg/dL), low density lipoprotein (LDL, in mg/dL), calcium intake (in mg), caffein intake (in mg), alcohol intake (in mg), gender (categorized as male and female), race/ethnicity (categorized as Mexican American, Other Hispanic, non-Hispanic white, non-Hispanic black, or other race – including multi-racial), education level (categorized as more than high school and no more than high school), marital status (categorized as married/living with a partner, widowed/divorced/separated and never married), income, smoking status, body mass index (BMI), arthritis.

We deleted patients whose DSB was 0 and took the averages of remained measured SBP and DBP as the statistical data. By using serum creatinine (Scr, in mg/dL) and age, eGFR is calculated through the following formula: eGFR = 186*Scr-1.154*age-0.203*(0.742, female) (ml/min/1.73 m²). Calcium intake, caffein intake and alcohol intake data were obtained in 24-hour food recall.

Income was classified according to ratio of family income to poverty (PIR) as poor (<1), near poor (1-3), not poor (≥3) (10). Smoking status was classified according to the NCHS classifications, where individuals who had smoked fewer than 100 cigarettes in their lifetime were considered never smokers, those who had smoked more than 100 cigarettes but were not currently smoking at the time of the survey were classified as ever smokers, and those who had smoked more than 100 cigarettes in their lifetime and were currently smoking at the time of survey were categorized as current smokers. BMI was categorized as underweight or healthy weight (<25 kg/m²), overweight (25-29.9 kg/m²), obese (30-34.9 kg/m²) and severely or extremely obese (≥35 kg/m²). Arthritis was defined as “a doctor told you had arthritis”. Postmenopausal women were selected by “do you still have regular periods”.

2.5 Statistical Analysis

We performed statistical analyses using R (version 3.5.3) and EmpowerStats (www.empowerstats.com; X&Y Solution Inc.). We investigated the association between DII and lumbar BMD in the overall population and specifically in menopausal women. Our analysis included baseline characteristic analysis, smooth curve fitting and multivariate logistic regression.

In the baseline characteristics section, categorical variables were presented as percentages, and continuous variables were expressed as means ± standard deviation (SD). Differences among DII tertiles were assessed using weighted linear regression for continuous variables and weighted chi-square tests for categorical variables

To visually depict the relationship between DII and lumbar BMD, we employed smooth curve fitting with covariates in the entire population, specially in menopausal women. Furthermore, to accurately assess the impact of DII on BMD across different DII tertiles, we utilized weighted multivariate regression analysis in the entire population and among menopausal women, employing three different adjusting models.

3.1 Participant Characteristics

A total of 27,072 participants were included in the study, comprising 13,629 males and 13,443 females. Baseline characteristics of the entire population, stratified by DII tertiles (weighted), are presented in Table 1. Participants in the highest DII tertile (DII tertile 3) tended to be younger (44.74 years vs. 45.92 years) and more likely to be female (61.59% vs. 38.22%). They also exhibited lower levels of lumbar BMD (1.029 g/cm² vs. 1.040 g/cm²), eGFR (105.824 ml/min/1.73 m² vs. 111.070 ml/min/1.73 m²), calcium intake (652.63 mg vs. 1310.65 mg), caffeine intake (156.61 mg vs. 184.54 mg), and alcohol intake (7.32 mg vs. 15.56 mg), along with higher levels of total cholesterol (TC) (192.19 mg/dL vs. 190.77 mg/dL), low-density lipoprotein (LDL) (114.19 mg/dL vs. 111.57 mg/dL), and glycohemoglobin (5.63 % vs. 5.56 %) compared to those in the lowest DII tertile (DII tertile 1). Additionally, higher proportions of non-Hispanic Black individuals (15.49 % vs. 8.33 %), other Hispanic individuals (6.95 % vs. 6.13 %), individuals with a maximum education level of high school (46.21% vs. 30.24%), individuals who were widowed, divorced, separated, or never married (19.40% vs. 13.25%, 22.90% vs. 19.28%), economically disadvantaged individuals (18.83% vs. 11.50%, 39.18% vs. 30.47%), current smokers (24.56% vs. 14.31%), individuals with a BMI <30 (18.13% vs. 12.98%, 22.17% vs. 19.26%), and those with arthritis (22.20% vs. 19.83%) were associated with adherence to a high DII diet.

Table 1. Baseline characteristics of the entire population by tertiles of dietary inflammatory index (weighted).
	Dietary inflammatory index
	Tertile 1 (-5.731-0.300)	Tertile 2 (0.300-2.229)	Tertile 3 (2.229-5.405)	P value
DII ^a	-1.23 ± 1.11	1.30 ± 0.55	3.19 ± 0.64	<0.0001
Age (years) ^a	45.29 ± 15.45	45.25 ± 16.06	44.74 ± 16.90	0.0445
Systolic blood pressure (mmHg) ^a	120.65 ± 15.89	121.37 ± 16.05	121.33 ± 17.50	0.0044
Diastolic blood pressure (mmHg) ^a	71.97 ± 10.62	72.21 ± 10.77	71.55 ± 11.32	0.0004
Lumbar BMD (g/cm²) ^a	1.04 ± 0.16	1.04 ± 0.15	1.03 ± 0.15	<0.0001
eGFR (ml/min/1.73 m²) ^a	111.07 ± 37.75	108.09 ± 37.91	105.82 ± 38.19	<0.0001
Glycohemoglobin (%) ^a	5.56 ± 0.85	5.60 ± 0.90	5.63 ± 0.95	<0.0001
TC (mg/dL) ^a	190.77 ± 39.72	193.54 ± 40.61	192.19 ± 40.41	<0.0001
LDL (mg/dL) ^a	111.57 ± 34.36	114.34 ± 35.31	114.19 ± 34.61	0.0002
Calcium intake (mg) ^a	1310.65 ± 680.23	927.23 ± 461.28	652.63 ± 362.31	<0.0001
Caffein intake (mg)^a	184.54 ± 223.90	169.78 ± 207.44	156.61 ± 197.53	<0.0001
Alcohol intake (mg)^a	15.56 ± 32.07	13.11 ± 31.79	7.32 ± 22.73	<0.0001
Gender (%)				<0.0001
Male	61.78	49.17	38.41
Female	38.22	50.83	61.59
Race/ethnicity (%)				<0.0001
Mexican American	10.20	9.51	8.28
Other Hispanic	6.13	6.45	6.95
Non-Hispanic White	66.43	63.77	61.99
Non-Hispanic Black	8.33	11.89	15.49
Other races including multi-racial	8.92	8.39	7.29
Education level (%)				<0.0001
More than high school	69.76	61.57	53.79
No more than high school	30.24	38.43	46.21
Marital status (%)				<0.0001
Married/living with a partner	67.47	64.08	57.70
Widowed/divorced/separated	13.25	16.05	19.40
Never married	19.28	19.87	22.90
Income (%)				<0.0001
Poor	11.50	12.42	18.83
Near poor	30.47	36.05	39.18
Not poor	58.03	51.53	41.98
Smoking status (%)				<0.0001
Current	14.31	17.45	24.56
Ever	25.36	23.34	19.26
Never	60.33	59.22	56.19
BMI group (%)				<0.0001
Underweight or healthy weight	12.98	17.31	18.13
Overweight	19.26	21.59	22.17
obese	34.83	32.28	30.85
Severely or extremely obese	32.93	28.81	28.85
Arthritis (%)				0.0005
Yes	19.83	21.07	22.20
No	80.17	78.93	77.80

3.2 Association of DII with Lumbar Spine BMD

The smoothed curve fitting result was presented in Figure 2. In the entire population, there was a nonlinear relationship between DII and lumbar spine BMD, exhibiting varying trends depending on the DII values, but generally showing a declining overall trend. The analysis adjusted for variables including age, gender, race/ethnicity, smoking status, BMI categories, arthritis, SBP, DBP, eGFR, TC, LDL, calcium intake, caffein intake, and alcohol intake.

Figure 2 Smooth curve fitting results after adjustment revealed a non-linear relationship between DII and lumbar BMD in the entire population. The solid red line represents the smooth curve fit between variables. Blue bands represent the 95% confidence interval from the fit.

We conducted further analysis to explore the associations between Dietary Inflammatory Index (DII) and lumbar spine Bone Mineral Density (BMD) across different DII tertiles within the entire population (Table 2). The results indicated a significant inverse correlation between median DII (Tertile 2) and lumbar spine BMD, which persisted after adjustment (model 1: β = -0.0076; 95% CI: -0.0146, -0.0006, p=0.0325; model 2: β = -0.0078; 95%CI: -0.0147, -0.0010, p=0.0251; model 3: β=-0.0126; 95% CI: -0.0227, -0.0025, p=0.0147). Conversely, a positive correlation was observed between high DII (Tertile 3) and lumbar spine BMD, reaching significance after adjustment in model 3 (model 3: β = 0.0102; 95% CI: 0.0013, 0.0190, p = 0.0241). However, no significant correlation was found between low DII (Tertile 1) and lumbar spine BMD in either unadjusted or adjusted models.

Table 2. The association between DII and the lumbar BMD in the entire population (weighted).
	Tertile 1 (-5.731-0.300)		Tertile 2 (0.300-2.229)		Tertile 3 (2.229-5.405)
Model	β (95%CI)	P value	β (95%CI)	P value	β (95%CI)	P value
Model 1	-0.0014 (-0.0049, 0.0021)	0.4361	-0.0076 (-0.0146, -0.0006)	0.0325*	0.0008 (-0.0052, 0.0067)	0.8024
Model 2	-0.0004 (-0.0039, 0.0030)	0.8045	-0.0078 (-0.0147, -0.0010)	0.0251*	0.0012 (-0.0046, 0.0069)	0.6895
Model 3	-0.0010 (-0.0061, 0.0042)	0.7118	-0.0126 (-0.0227, -0.0025)	0.0147*	0.0102 (0.0013, 0.0190)	0.0241*
Model 1: unadjusted. Model 2: adjusted for age, gender, race/ethnicity. Model 3: adjusted for age, gender, race/ethnicity, smoker, SBP, DBP, eGFR, BMI group, glycohemoglobin, calcium intake, caffein intake, alcohol intake, TC, LDL, arthritis. *P<0.05

3.3 Subgroup Analysis

In the subgroup analysis, a total of 6062 postmenopausal women were included, and baseline characteristics stratified by DII tertiles (weighted) were presented in Table 3. Postmenopausal women adhering to a high DII diet (DII tertile 3) exhibited predominantly higher levels of eGFR (83.69 ml/min/1.73 m² vs. 77.97 ml/min/1.73 m²), glycohemoglobin (5.87 % vs. 5.73 %), along with lower levels of calcium intake (585.42 mg vs. 1065.30 mg), caffeine intake (151.67 mg vs. 165.61 mg), and alcohol intake (3.93 mg vs. 7.75 mg) compared to those adhering to a low DII diet (DII tertile 1). Additionally, postmenopausal women who were Mexican American (6.21% vs. 5.02%), other Hispanic (6.14 % vs. 5.33 %), non-Hispanic black (15.34% vs. 9.33%), widowed/divorced/separated (37.47 % vs. 28.41 %) or never married (8.98 % vs. 6.95 %), economically disadvantaged (17.19 % vs. 7.73 %, 41.33% vs. 30.47%), currently smoking (19.54 % vs. 9.12%), and with an education level no higher than high school (49.90 % vs. 31.34 %), a BMI≥30 (24.78 % vs. 17.59 %), and arthritis (46.29 % vs. 41.32 %) were more likely to adhere to a high DII diet.

Table 3. Baseline characteristics of the postmenopausal women by tertiles of dietary inflammatory index (weighted).
	Dietary inflammatory index
	Tertile 1 (-5.512-0.894)	Tertile 2 (0.896-2.622)	Tertile 3 (2.623-4.982)	P value
DII ^a	-0.72 ± 1.22	1.79 ± 0.49	3.49 ± 0.55	<0.0001*
Age (years) ^a	59.96 ± 11.73	59.52 ± 12.80	59.64 ± 13.04	0.4926
Systolic blood pressure (mmHg) ^a	125.37 ± 18.77	126.76 ± 18.65	127.92 ± 20.64	0.0002*
Diastolic blood pressure (mmHg) ^a	71.35 ± 10.45	71.59 ± 11.02	71.15 ± 12.04	0.4885
Lumbar BMD (g/cm²) ^a	0.97 ± 0.16	0.98 ± 0.17	0.97 ± 0.16	0.1508
eGFR (ml/min/1.73 m²) ^a	77.96 ± 30.65	81.11 ± 33.29	83.69 ± 37.32	<0.0001*
Glycohemoglobin (%) ^a	5.73 ± 0.78	5.80 ± 0.87	5.87 ± 1.00	<0.0001*
TC (mg/dL) ^a	206.75 ± 39.46	204.81 ± 40.59	204.45 ± 42.79	0.1666
LDL (mg/dL) ^a	120.10 ± 36.40	119.40 ± 35.48	120.07 ± 38.17	0.8984
Calcium intake (mg) ^a	1065.30 ± 511.66	797.89 ± 407.28	585.42 ± 314.01	<0.0001*
Caffein intake (mg)^a	165.61 ± 167.68	169.02 ± 217.09	151.67 ± 154.67	0.0091*
Alcohol intake (mg)^a	7.75 ± 17.11	6.17 ± 15.62	3.93 ± 14.14	<0.0001*
Race/ethnicity (%)				<0.0001*
Mexican American	5.02	5.53	6.21
Other Hispanic	5.33	4.68	6.13
Non-Hispanic White	73.06	73.23	66.55
Non-Hispanic Black	9.33	10.88	15.34
Other races including multi-racial	7.26	5.67	5.76
Education level (%)				<0.0001*
More than high school	68.66	57.00	50.10
No more than high school	31.34	43.00	49.90
Marital status (%)				<0.0001*
Married/living with a partner	64.63	61.73	53.54
Widowed/divorced/separated	28.41	31.45	37.47
Never married	6.95	6.83	8.98
Income (%)				<0.0001*
Poor	7.74	10.13	17.19
Near poor	30.47	37.99	41.33
Not poor	61.79	51.88	41.47
Smoking status (%)				<0.0001*
Current	9.12	12.97	19.54
Ever	27.49	26.01	20.70
Never	63.39	61.02	59.76
BMI group (%)				<0.0001*
Underweight or healthy weight	33.49	25.24	23.19
Overweight	31.68	30.78	31.09
obese	17.59	22.36	24.78
Severely or extremely obese	17.24	21.62	20.94
Arthritis (%)				0.0064*
Yes	41.31	44.31	46.29
No	58.69	55.69	53.71

The smoothed curve fitting result in postmenopausal women was presented in Figure 3. DII and lumbar BMD exhibited an inverse U-shaped association (Figure 3). The analysis adjusted for variables including age, race/ethnicity, smoking status, BMI groups, arthritis, SBP, DBP, eGFR, LDL, calcium intake, caffein intake and alcohol intake.

Figure 3 Smooth curve fitting results after adjustment revealed an inverse U-shaped relationship between DII and lumbar BMD in postmenopausal women.The solid red line represents the smooth curve fit between variables. Blue bands represent the 95% confidence interval from the fit.

We conducted further analysis to examine the associations between DII and lumbar spine BMD across different DII tertiles among postmenopausal women (Table 4). When DII was in the median tertile (Tertile 2), we initially found a significant negative correlation, but this significance was not maintained in the fully adjusted model (model 1: β=-0.0289; 95%CI: -0.0475, -0.0103, p=0.0023; model 2: β=-0.0253; 95%CI: -0.0435, -0.0071, p=0.0065; model 3: β = -0.0043; 95% CI: -0.0303, 0.0217, p=0.7458). Conversely, when DII was in the low tertile (Tertile 1), we observed a positive correlation between DII and lumbar spine BMD in unadjusted and adjusted models (model 1: β=0.0111; 95%CI: 0.0042, 0.0179, p=0.0015; model 2: β=0.0114; 95%CI: 0.0048, 0.0181, p=0.0008; model 3: β=0.0098; 95%CI: 0.0001, 0.0196, p=0.0.492). However, we did not find a significant correlation between DII and lumbar spine BMD in either unadjusted or adjusted models when DII was high (Tertile 3).

Table 4. The association between DII and the lumbar BMD in postmenopausal women (weighted).
	Tertile 1 (-5.512-0.894)		Tertile 2 (0.896-2.622)		Tertile 3 (2.623-4.982)
Model	β (95%CI)	P value	β (95%CI)	P value	β (95%CI)	P value
Model 1	0.0111 (0.0042, 0.0179)	0.0015*	-0.0289 (-0.0475, -0.0103)	0.0023*	0.0017 (-0.0136, 0.0170)	0.8247
Model 2	0.0114 (0.0048, 0.0181)	0.0008*	-0.0253 (-0.0435, -0.0071)	0.0065*	-0.0014 (-0.0160, 0.0131)	0.8458
Model 3	0.0098 (0.0001, 0.0196)	0.0492*	-0.0043 (-0.0303, 0.0217)	0.7458	0.0113 (-0.0121, 0.0348)	0.3438
Model 1: unadjusted. Model 2: adjusted for age, race/ethnicity. Model 3: adjusted for age, race/ethnicity, smoker, SBP, DBP, eGFR, BMI group, LDL, glycohemoglobin, calcium intake, caffein intake, alcohol intake, arthritis. *P<0.05

This study represents one of the largest cross-sectional investigations into the association of DII and the lumbar spine BMD in both the general population and postmenopausal women. Utilizing multivariate regression analysis and smooth curve fitting, we explored these relationships across different DII tertiles in these two groups.

In the analysis of the whole population, the smooth curve fitting revealed a non-linear association between DII and lumbar spine BMD, demonstrating both declining and rising trends. Contrary to expectations, the multivariate analysis indicated that diets with the highest pro-inflammatory potential were positively associated with the lumbar spine BMD, whereas diets with the median pro-inflammatory potential were negatively associated with BMD. Similarly surprising, in the subgroup analysis of postmenopausal women, smooth curve fitting suggested an inverted U-shaped relationship between DII and the lumbar spine BMD. The multivariate analysis further indicated that the most anti-inflammatory diets were negatively associated with lumbar spine BMD.

This study reveals both positive and negative associations between dietary inflammation and the lumbar spine BMD in both the general population and menopausal women, a finding previously overlooked. Prior research failed to recognize the non-linear nature of this relationship, likely due to a lack of continuous DII analysis across different tertiles and the absence of smooth curve fitting in diverse populations.

Dietary inflammation may influence BMD via regulating bone remodelling through inflammatory cytokines. Bone remodelling, which involves osteoclasts-mediated bone resorption and osteoblasts-mediated bone synthesis, plays a predominant role in bone structure and function in adults (11). Osteoclasts initiate remodelling by resorbing bone, which is then partially replenished with osteoblast-produced matrix that later mineralizes (5).

Inflammation, which is often accompanied by increased activated cytokines, can lead to excessive activity of osteoclasts and the inhibition of osteoblasts, which causes bone loss and may lead to a decreased BMD. TNF, IL-6 and IL-1 are key cytokines that significantly upregulate osteoclast activity and suppress osteoblast function. TNF signals through nuclear factor-κB (NF-κB) and mitogen-activated protein kinases (MAPKs), while the IL-6 receptor (IL-6R) activates the JAK–STAT pathway to enhance osteoclast activity. In inhibiting osteoblasts, TNF and IL-6 inhibit MAPK, activate NF-κB, and upregulate SMAD ubiquitylation regulatory factor 1 (SMURF1) and SMURF2. The pro-inflammatory cytokines also upregulate Dickkopf-related protein 1 (DKK1) and sclerostin (SOST), which inhibit the WNT–Frizzled pathway (5).

However, these mechanisms could not explain the positive correlation observed between DII and BMD, as well as the distinct phenomena in differed groups. This discrepancy may be explained by the hypothesis that low-level inflammation promotes bone growth through Wnt proteins (12). Additionally, estrogen loss, which happens in menopausal women, leads to chronic low-grade production of the proinflammatory cytokines TNF alpha and IL-17 by converting memory T-cells to effector T-cells (4). Therefore, the smooth curve trend obeserved in menopausal women is similar to the latter two-thirds of the total population.

Research found that it is constitutive low-intensity TNF stimulation rather than short-term and high-intensity one that induces strong and persistent expression of Wnt proteins at bone-forming sites, through the NF-jB (p65) and JNK/AP-1 (c-Jun) pathways, followed by subsequent new bone formation (6) .

Though estrogen and androgen both matter in the bone regulation, the differences of hormone constitution in people from different age and sex groups may explain the differed overall levels of lumbar spine BMD. Androgen stimulated the proliferation of pre-osteoblasts and differentiation of osteoblasts. The converted estrogen suppressed osteoclast formation and resorption activity by blocking the receptor activator of nuclear factor k-B ligand pathway (13). Menopause results from reduced secretion of the ovarian hormones estrogen and progesterone, which takes place as the finite store of ovarian follicles is depleted. In postmenopausal osteoporosis, bone resorption increases by up to 70% due to the increased number of osteoclasts, while the state of bone formation amount cannot make up for the loss. Bone synthesis might increase but to a lesser extent than bone resorption, keep the same as before or decrease by up to 14% (15).

We made improvements by excluding certain variables previously used in similar studies. First, in the subgroup analysis, we didn’t adjust for total cholesterol because we found TC had a mediating effect on the lumbar spine BMD in the menopausal women when in causal mediation analysis. Second, we didn’t take vitamin D intake into consideration because a systematic review has reported that vitamin D intake has no significant effect on the lumbar spine BMD (16).

Our study has some limitations. First, we didn’t include physical activity as a covariate. Different types of exercise might have different effects on the lumbar spine BMD despite similar caloric consumption, which cannot be adequately addressed by the available data in the NHANES including metabolic equivalent (MET), weekly frequency, and duration of each activity. A study using NHANES data found that physical activity, as calculated by MET, weekly frequency and duration of each activity, only showed significant association among menopausal women younger than 65 years (17). Second, we didn’t adjust for corticosteroid use because it has dual effects of anti-inflammation and causing bone loss, which is hard to adjust via NHANES data. Thirdly, we didn’t adjust for bisphosphonates and calcitonin because the does and frequency of medicine use matter as well, which are unavailable on a large-scale from NHANES database (18). Fourthly, we did not consider thyroid problems in the logistic regression model because the it is not accurate enough about the actual effect on the BMD by the limited information based on NAHNES. Finally, since this is a cross-sectional study, we are not sure about the causal association between DII and lumbar spine BMD.

In conclusion, the DII is intensity-dependently associated with the lumbar spine BMD in both the overall population and the menopausal women. In the general population, a median DII (0.300 -2.229) is negatively associated with lumbar spine BMD whereas a high DII (2.229–5.405) shows a positive association. In menopausal women, a low DII (-5.512–0.894) is positively associated with the lumbar spine BMD. The differed effects may be influenced by the dual impact of inflammation intensity-dependent Wnt proteins and chronic low inflammation state induced by estrogen loss in postmenopausal women.

BMD

bone mineral density

DII

dietary inflammatory index

NHANES

National Health and Nutrition Examination Survey

DXA

dual-energy X-ray absorptiometry

IL

interleukin

TNF

tumor necrosis factor

CRP

C-reactive protein

SBP

systolic blood pressure

DBP

diastolic blood pressure

eGFR

estimated glomerular filtration rate

TC

total cholesterol

LDL

low density lipoprotein

BMI

body mass index

SD

standard deviation

NF-κB

nuclear factor-κB

MAPKs

mitogen-activated protein kinases

SMURF

SMAD ubiquitylation regulatory factor

DKK

Dickkopf-related protein

SOST

sclerostin

MET

metabolic equivalent

Ethics approval and consent to participate

The studies involving humans were approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Consent for publication (not available)

Availability of data and materials

Publicly available datasets were analyzed in this study. This data can be found at: https://www.cdc.gov/nchs/nhanes/index.htm. Clinical trial number: not applicable.

Competing interest

The author declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The authors declare that no financial support was received for the research, authorship, and/or publication of this article.

Authors’ contributions

The authors’ responsibilities were as follows -- ZS, YJ and XC designed research; YJ, XC, SD drafted the manuscript; YJ, XC, SD, YZ, ZZ, TY collected and analyzed the data; ZS revised the manuscript. YJ, XC and ZS had primary responsibility for the final content. All authors read and approved the final manuscript.

Acknowledgements

We are greatly thankful for all participants in the NAHNES for providing data for this study.

the Scientific Advisory Board of the European Society for Clinical and Economic Aspects of Osteoporosis and Osteoarthritis (ESCEO) and the Committees of Scientific Advisors and National Societies of the International Osteoporosis Foundation (IOF), Kanis JA, Cooper C, Rizzoli R, Reginster J-Y. Executive summary of the European guidance for the diagnosis and management of osteoporosis in postmenopausal women. Calcif Tissue Int. 2019;104:235–8. 10.1007/s00223-018-00512-x.
Wright NC, Looker AC, Saag KG, Curtis JR, Delzell ES, Randall S, Dawson-Hughes B. The Recent Prevalence of Osteoporosis and Low Bone Mass in the United States Based on Bone Mineral Density at the Femoral Neck or Lumbar Spine. J Bone Min Res. 2014;29:2520–6. 10.1002/jbmr.2269.
Shivappa N, Steck SE, Hurley TG, Hussey JR, Hébert JR. Designing and developing a literature-derived, population-based dietary inflammatory index. Public Health Nutr. 2014;17:1689–96. 10.1017/S1368980013002115.
Wu D, Cline-Smith A, Shashkova E, Perla A, Katyal A, Aurora R. T-Cell Mediated Inflammation in Postmenopausal Osteoporosis. Front Immunol. 2021;12:687551. 10.3389/fimmu.2021.687551.
Redlich K, Smolen JS. Inflammatory bone loss: pathogenesis and therapeutic intervention. Nat Rev Drug Discov. 2012;11:234–50. 10.1038/nrd3669.
Li X, Wang J, Zhan Z, Li S, Zheng Z, Wang T, Zhang K, Pan H, Li Z, Zhang N, et al. Inflammation Intensity–Dependent Expression of Osteoinductive Wnt Proteins Is Critical for Ectopic New Bone Formation in Ankylosing Spondylitis. Arthritis Rheumatol. 2018;70:1056–70. 10.1002/art.40468.
Zhao S, Gao W, Li J, Sun M, Fang J, Tong L, He Y, Wang Y, Zhang Y, Xu Y, et al. Dietary inflammatory index and osteoporosis: the National Health and Nutrition Examination Survey, 2017–2018. Endocrine. 2022;78:587–96. 10.1007/s12020-022-03178-6.
Song D, Kim J, Kang M, Park J, Lee H, Kim D-Y, Park SY, Lim H. Association between the dietary inflammatory index and bone markers in postmenopausal women. PLoS ONE. 2022;17:e0265630. 10.1371/journal.pone.0265630.
Jackson MK, Bilek LD, Waltman NL, Ma J, Hébert JR, Price S, Graeff-Armas L, Poole JA, Mack LR, Hans D et al. Dietary Inflammatory Potential and Bone Outcomes in Midwestern Post-Menopausal Women. Nutrients (2023) 15:4277. 10.3390/nu15194277
Li S, Zeng M. The association between dietary inflammation index and bone mineral density: results from the United States National Health and nutrition examination surveys. Ren Fail. 2023;45:2209200. 10.1080/0886022X.2023.2209200.
Boyle WJ, Simonet WS, Lacey DL. Osteoclast differentiation and activation. Nature. 2003;423:337–42. 10.1038/nature01658.
Collison J. Low-level inflammation promotes bone growth. Nat Rev Rheumatol. 2018;14:249–249. 10.1038/nrrheum.2018.52.
Mohamad NV, Soelaiman I-N, Chin K-Y. A concise review of testosterone and bone health. Clin Interv Aging. 2016;11:1317–24. 10.2147/CIA.S115472.
Takahashi TA, Johnson KM, Menopause. Med Clin North Am. 2015;99:521–34. 10.1016/j.mcna.2015.01.006.
Fischer V, Haffner-Luntzer M. Interaction between bone and immune cells: Implications for postmenopausal osteoporosis. Semin Cell Dev Biol. 2022;123:14–21. 10.1016/j.semcdb.2021.05.014.
Reid IR, Bolland MJ, Grey A. Effects of vitamin D supplements on bone mineral density: a systematic review and meta-analysis. Lancet. 2014;383:146–55. 10.1016/S0140-6736(13)61647-5.
Ji J, Hou Y, Li Z, Zhou Y, Xue H, Wen T, Yang T, Xue L, Tu Y, Ma T. Association between physical activity and bone mineral density in postmenopausal women: a cross-sectional study from the NHANES 2007–2018. J Orthop Surg. 2023;18:501. 10.1186/s13018-023-03976-2.
Khosla S, Hofbauer LC. Osteoporosis treatment: recent developments and ongoing challenges. Lancet Diabetes Endocrinol. 2017;5:898–907. 10.1016/S2213-8587(17)30188-.

No competing interests reported.

supplementmediatingeffectanalysisthepostmenopausalwomen.png
Supplement: mediating effect analysis the postmenopausal women.png Additional file 1: The mediating effect of total cholesterol between DII and BMD in postmenopausal women

Dietary Inflammatory Index is intensity-dependently associated with the Bone Mineral Density at the Lumbar Spine: A Cross- Sectional NHANES Study from 2007 to 2020

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Background

Methods

2.1 Study design and participants

2.2 Dietary Data and Dietary inflammatory Index (DII)

2.3 Lumbar Bone Mineral Density

2.4 Covariates

2.5 Statistical Analysis

Results

3.1 Participant Characteristics

3.2 Association of DII with Lumbar Spine BMD

3.3 Subgroup Analysis

Discussion

Conclusions

Abbreviations

Declarations

Ethics approval and consent to participate

Consent for publication (not available)

Availability of data and materials

Competing interest

Funding

Authors’ contributions

Acknowledgements

References

Additional Declarations

Supplementary Files

Status:

Version 1