Numerous studies suggested that oxidative stress (OS) played a central role in the onset and development of postmenopausal osteoporosis (PO); however, conflicting results were obtained as to the association of OS-related biomarkers and PO. This meta-analysis aimed to identify the association between these markers and PO, and explore factors that may explain the inconsistencies in these results. A systematic literature search was conducted in relevant database. Search terms and selection criteria were priorly determined to identify and include all studies that detected markers of OS in PO patients. We pooled data with a random effects meta-analysis with standardized mean differences and 95% confidence interval. Total 17 studies including 12 OS markers were adopted. The results showed that superoxide dismutase (SOD) in erythrocytes, catalase (CAT), total antioxidant status (TAS), hydroperoxides (HY), advanced oxidation protein products (AOPP), malondialdehyde (MDA), and vitamin B12 (VB12) in plasma/serum were not statistically different between the PO and control group, whereas significantly increased level of homocysteine (Hcy) and nitric oxide (NO), along with decreased SOD, glutathione peroxidase (GPx), folate, and total antioxidant power (TAP) in plasma/serum were obtained in the PO group. In summary, OS might serve as potential biomarkers in the etiopathophysiology and clinical course of PO.
Postmenopausal osteoporosis (PO) is one of the most common bone diseases, characterized by low bone mineral density (BMD) and pathological fracture, which leads to significant morbidity [
In spite of the remarkable progress achieved in the understanding of how estrogen deficiency induces PO, the underlying pathogenic mechanisms have been found to be complex and multifaceted [
OS is generated as a result of insufficient activity of the endogenous antioxidant defense system against reactive oxygen species (ROS). On the one hand, excessive ROS are able to exert oxidative damage to lipids, protein, and DNA [
Experimental studies demonstrated that OS is an important factor in bone remodeling [
In addition, there is a wide range of OS biomarkers and laboratory techniques available, each of which has its own strength and limitation [
Identification of the studies was carried out through an extensive literature search using the PubMed database, ISI Web of Science, EMBASE, and Google Scholar mainly based on the search terms, with English restriction, and updated to February 2016. The search strategy included the terms “oxidative stress”, “bone mineral density”, and “postmenopausal osteoporosis” and they were used in text word searches, and the “related articles” function was used to broaden the search. Publications cited in references found using these search terms were also reviewed for any relevant studies, which were not already identified; in addition, all searches were conducted prior to February 2016 with no time span specified.
We searched all abstracts for potentially relevant publications. Studies meeting the following criteria were included: (1) having measured levels of one or more of the following OS markers in both PO patients and healthy controls: SOD, CAT, Hcy, GPx, protein carbonyl, 3-nitrotyrosine, NO, vitamins, folate, lipid peroxidation, TAP, and TAS; (2) being reported in an original research paper in a peer-reviewed journal; and (3) adequately describing their samples (e.g., diagnostic criteria, source of samples, and storage) and methods such that the experiments could be replicated (or included appropriate references). For all included studies, the study design, sample type, age, and BMI of each group and biomarkers of interest were recorded.
Papers were excluded (1) if the study only enrolled subjects with postmenopausal osteoporosis; (2) if the outcomes were not reported as the mean ± standard deviation (SD); (3) if the BMD was not evaluated by dual-energy X-ray absorptiometry (DXA); and (4) if postmenopausal women took estrogen replacement therapy (ERT) before the clinical trial. Finally, studies were checked carefully to ensure that the diagnosis criteria of PO were similar among the studies. When several reports from the same study were published, only the most recent or informative one was included in our meta-analysis. Only biomarkers that were the object of at least 2 independent studies were included.
To reduce the heterogeneity, the studies included in the meta-analysis were only carried out on the same biological sample, except for plasma and serum. All the studies had a cross-sectional design, with cases mostly diagnosed according to the BMD T-score (the number of standard deviations below the average for a young adult at peak bone density) lower than 2.5 standard deviations from BMD peak at either femoral neck or lumbar spine, according to WHO guidelines.
The meta-analysis was conducted using Stata statistical software version 12.1 (Stata, College Station, TX, USA). Standardized mean differences were used to construct forest plots of continuous data.
The selection of literature for included studies was shown in Figure
The characteristics of the included studies.
Location | Study design | Sample (patients/controls) | Age (patients/controls) | BMI (patients/controls) | Biomarker |
---|---|---|---|---|---|
Cervellati et al., Italy [ |
Cross-sectional study | 56 versus 38 | 58.40 ± 4.30 versus 53.70 ± 4.60 | 24.20 ± 3.20 versus 26.40 ± 4.10 | HY, AOPP, |
|
|||||
Cervellati et al., Italy [ |
Cross-sectional study | 30 versus 98 | 57.70 ± 4.9 versus 53.90 ± 5.00 | 24.40 ± 3.50 versus 25.40 ± 3.50 | AOPP |
|
|||||
Altindag et al. [ |
Cross-sectional study | 39 versus 26 | 56.70 ± 9.4 versus 54.15 ± 58 | 27.40 ± 5.20 versus 26.50 ± 4.20 | TOS |
|
|||||
Yilmaz and Eren [ |
Cross-sectional study | 34 versus 15 | 55.90 ± 6.5 versus 54.10 ± 4.7 | 21.40 ± 3.10 versus 26.10 ± 2.60 | Hcy |
|
|||||
Yousefzadeh et al. [ |
Cross-sectional study | 22 versus 22 | 59.27 ± 4.26 versus 56.91 ± 6.23 | 25.39 ± 5.03 versus 27.47 ± 3.88 | TBARS |
|
|||||
Akpolat et al. [ |
Cross-sectional study | 66 versus 60 | 62.88 ± 6.59 versus 55.40 ± 7.88 | 27.29 ± 4.06 versus 32.02 ± 8.05 | Hcy |
|
|||||
Ozgocmen et al. [ |
Cross-sectional study | 59 versus 22 | 56.75 ± 5.38 versus 55.86 ± 6.01 | SOD | |
|
|||||
Sendur et al. [ |
Cross-sectional study | 45 versus 42 | 55.60 ± 2.90 versus 56.60 ± 2.40 | 29.40 ± 4.40 versus 27.90 ± 4.30 | CAT |
|
|||||
Zinnuroglu et al. [ |
Cross-sectional study | 23 versus 23 | 67.60 ± 8.50 versus 62.24 ± 7.60 | 29.21 ± 4.13 versus 28.45 ± 4.42 | MDA |
|
|||||
Ouzzif et al. [ |
Cross-sectional study | 58 versus 64 | 61.90 ± 9.90 versus 53.50 ± 5.30 | 28.50 ± 4.40 versus 32.30 ± 6.20 | Hcy |
|
|||||
Bozkurt et al. [ |
Cross-sectional study | 38 versus 48 | 57.30 ± 7.90 versus 51.40 ± 8.90 | 25.70 ± 3.80 versus 28.20 ± 3.70 | Hcy |
|
|||||
Haliloglu et al. [ |
Cross-sectional study | 25 versus 53 | 55.70 ± 0.50 versus 53.50 ± 0.60 | 26.60 ± 5.83 versus 28.20 ± 5.19 | Hcy |
|
|||||
Baines et al. [ |
Cross-sectional study | 110 versus 110 | 68.90 (41–86) versus 67.60 (45–84) | 24.52 ± 4.09 versus 27.84 ± 4.96 | Hcy |
|
|||||
Cagnacci et al. [ |
Cross-sectional study | 28 versus 72 | 54.70 ± 0.90 versus 52.50 ± 0.60 | 25.60 ± 0.80 versus 27.50 ± 0.60 | Hcy |
|
|||||
Maggio et al. [ |
Cross-sectional study | 75 versus 75 | 70.40 ± 8.50 versus 68.80 ± 3.50 | 25.30 ± 2.90 versus 28.10 ± 3.40 | VB A |
|
|||||
Sharma et al. [ |
Cross-sectional study | 35 versus 30 | 58.00 ± 6.00 versus 53.00 ± 5.00 | 28.29 ± 7.50 versus 29.79 ± 6.20 | SOD |
|
|||||
Wu et al. [ |
Cross-sectional study | 60 versus 60 | 63.46 ± 7.45 versus 61.65 ± 6.30 | 23.29 ± 3.29 versus 25.39 ± 3.60 | AOPP |
Values are mean ± SD.
BMI: body mass index (kg/m2).
The relationship between enzymatic antioxidant and risk of PO.
First author | Biomarker | Biologic sample | Sample (patients/controls) | SMD | Heterogeneity | |
---|---|---|---|---|---|---|
(95% CI) |
|
|
||||
Ozgocmen [ |
CAT | Erythrocytes (fasting) | 59 versus 22 | −1.82 (−2.39, −1.26) | ||
Sendu [ |
CAT | Erythrocytes | 45 versus 42 | −0.30 (−0.72, 0.12) | ||
|
|
|
|
|
|
|
Maggio [ |
SOD | Erythrocyte | 75 versus 75 | −2.70 (−3.14, −2.26) | ||
Ozgocmen [ |
SOD | Erythrocytes (fasting) | 59 versus 22 | 0.16 (−0.33, 0.65) | ||
|
|
|
|
|
|
|
Sharma [ |
SOD | Serum | 35 versus 30 | −4.03 (−4.89, −3.17) | ||
Maggio [ |
SOD | Plasma | 75 versus 75 | −2.05 (−2.45, −1.66) | ||
|
|
|
|
|
|
|
|
||||||
Maggio [ |
GPx | Plasma | 75 versus 75 | −2.00 (−2.39, −1.61) | ||
Sharma [ |
GPx | Serum | 35 versus 30 | −5.51 (−6.59, −4.43) | ||
|
|
|
|
|
|
|
The relationship of free radicals products/antioxidants and risk of PO.
First author | Biomarker | Biologic sample | Sample (patients/controls) | SMD | Heterogeneity | |
---|---|---|---|---|---|---|
(95% CI) |
|
|
||||
Sendur [ |
MDA | Plasma | 45 versus 42 | 0.75 (0.31, 1.19) | ||
Akpolat [ |
MDA | Serum | 66 versus 60 | 1.15 (0.78, 1.53) | ||
Maggio [ |
MDA | Plasma | 75 versus 75 | −0.16 (−0.48, 0.16) | ||
Wu [ |
MDA | Plasma | 60 versus 60 | 0.29 (−0.07, 0.65) | ||
|
|
|
|
|
|
|
Wu [ |
AOPP | Plasma | 60 versus 60 | 1.07 (0.68, 1.45) | ||
Cervellati [ |
AOPP | Serum | 30 versus 63 | 0.09 (−0.32, 0.51) | ||
Cervellati [ |
AOPP | Serum | 56 versus 38 | 0.14 (−0.29, 0.58) | ||
|
|
|
|
|
|
|
Cervellati [ |
HY | Serum | 56 versus 38 | 0.27 (−0.14, 0.68) | ||
Cervellati [ |
HY | Serum | 30 versus 63 | 0.05 (−0.38, 0.49) | ||
|
|
|
|
|
|
|
Akpolat [ |
NO | Serum | 66 versus 60 | 0.72 (0.36, 1.09) | ||
Sendu [ |
NO | Plasma | 45 versus 42 | 0.60 (0.17, 1.03) | ||
|
|
|
|
|
|
|
Altindag [ |
TAS | Plasma | 39 versus 26 | −3.00 (−3.72, −2.28) | ||
Yilmaz [ |
TAS | Plasma | 34 versus 15 | −84.54 (−101.64, −67.44) | ||
|
|
|
|
|
|
|
Cervellati [ |
TAP | Serum | 56 versus 38 | −0.22 ( −0.63, 0.19) | ||
Yousefzadeh [ |
TAP | Plasma | 22 versus 22 | −0.62 (−1.22, −0.01) | ||
Cervellati [ |
TAP | Serum | 30 versus 63 | 0.04 (−0.40, 0.48) | ||
Sharma [ |
TAP | Serum | 35 versus 30 | −13.18 (−15.53, −10.83) | ||
|
|
|
|
|
|
|
Ouzzif [ |
VB12 | Plasma | 58 versus 64 | −0.13 (−0.48, 0.23) | ||
Bozkurt [ |
VB12 | Serum | 38 versus 48 | 0.05 (−0.37, 0.48) | ||
Haliloglu [ |
VB12 | Serum | 25 versus 53 | −0.19 (−0.66, 0.29) | ||
Baines [ |
VB12 | Serum | 110 versus 110 | −0.08 (−0.34, 0.19) | ||
Cagnacci [ |
VB12 | Serum | 28 versus 72 | 0.46 (0.02, 0.90) | ||
|
|
|
|
|
|
|
Haliloglu [ |
Folate | Serum | 25 versus 53 | −0.46 (−0.94, 0.02) | ||
Baines [ |
Folate | Serum | 110 versus 110 | −0.39 (−0.66, −0.12) | ||
Cagnacci [ |
Folate | Serum | 28 versus 72 | −5.52 (−6.41, −4.64) | ||
Bozkurt [ |
Folic Acid | Serum | 25 versus 53 | 0.04 (−0.39, 0.46) | ||
Ouzzif [ |
Folate | Plasma | 58 versus 64 | −0.24 (−0.60, 0.11) | ||
Akpolat [ |
Folate | Serum | 66 versus 60 | −1.03 (−1.40, −0.66) | ||
|
|
|
|
|
|
|
Akpolat [ |
Hcy | Plasma | 66 versus 60 | 0.09 (−0.26, 0.44) | ||
Yilmaz [ |
Hcy | Plasma | 34 versus 15 | 0.25 (−0.36, 0.86) | ||
Ouzzif [ |
Hcy | Plasma | 58 versus 64 | 0.51 (0.14, 0.87) | ||
Bozkurt [ |
Hcy | Serum | 38 versus 48 | 0.43 (−0.00, 0.88) | ||
Haliloglu [ |
Hcy | Serum | 25 versus 53 | 1.15 (0.64, 1.65) | ||
Baines [ |
Hcy | Plasma | 110 versus 110 | 0.24 (−0.03, 0.50) | ||
Cagnacci [ |
Hcy | Serum | 28 versus 72 | 1.22 (0.75, 1.68) | ||
|
|
|
|
|
|
|
Search strategy flow diagram.
Forest plot of meta-analysis of the relationship between enzymatic antioxidant and risk of PO.
Forest plot of meta-analysis of the relationship between TAP/TAS and risk of PO.
Forest plot of meta-analysis of the relationship of free radicals products and risk of PO.
Forest plot of meta-analysis of the relationship of nutrient status and risk of PO.
All the identified papers were published between 2003 and 2016. Eight studies [
The levels of antioxidant enzymes in PO cases and controls were reported in 12 articles.
CAT activity in erythrocytes was measured in 2 papers [
SOD activity was measured in 3 papers [
GPx activity was reported in 3 papers, while 2 studies reported on GPx activity in the plasma/serum samples. After meta-analysis, a significantly lower GPx activity was found in PO subjects than that in controls (−3.72, 95% CI −7.16 to −0.28), (
The meta-analysis including 4 trials with 408 subjects revealed that TAP level was significantly decreased in the PO group compared to the control group under a random-effects model (−2.74, 95% CI −4.60–1.08).
In the present study, the meta-analysis including 2 trials with 124 subjects revealed that, with regard to TAS level, there was no statistical difference between PO group and control group under a random-effects model (−43.30, 95% CI −123.21–36.60).
A forest plot that provided suitable data for statistical pooling revealed that there was no significant difference obtained for MDA levels between PO group and control group (0.50, 95% CI −0.08–1.08).
From the results, we could find that only 3 papers measured the AOPP and 2 papers measured the HY activity in PO. The final results of meta-analyses showed that no significantly higher AOPP (0.44, 95% CI −0.20–1.08) and HY (0.17, 95% CI −0.13–0.47) appeared in the PO subjects.
Meta-analysis of 2 trials with 168 subjects revealed that the NO level was statistically higher in the PO group than in the control group under a random-effects model (0.67, 95% CI 0.40–0.95).
A total of 5 studies reported results on VB12. All the separated papers found no statistical difference between cases and controls after combining all the raw data. The meta-analysis also showed no statistically decreased VB12 level in the PO group than in the control group under a random-effects model (0.00, 95% CI −0.20–0.21).
Folate activity was evaluated in 6 studies. The heterogeneity was significant (
Seven papers on Hcy were adopted in the current meta-analysis. The heterogeneity was significant (
Given the small number of studies, we performed a one-study removed sensitivity analysis by excluding each study individually. The effect size of MDA, AOPP, TAP, VB12, folate, and Hcy remained essentially unchanged in direction and magnitude after the removal of each study individually. We intended to assess publication bias, but the studies for each outcome of interest were too few to derive meaning from funnel plots.
To our knowledge, this is the first meta-analysis to clarify and quantify the relationship between OS-related biomarkers and PO patients. Our research further supported the presence of oxidative damage in PO patients. The results showed increased Hcy and NO in the PO subjects, while it showed decreased levels of folate and TAP, along with lower activity of SOD and GPx in these subjects.
ROS are usually highly reactive and unstable and have a very short half-life, thus making them difficult to measure directly. Oxidized biomolecule products generated by ROS are much more stable and commonly used as ROS markers. In addition, ROS could also be assessed indirectly by measuring antioxidant levels or antioxidant enzymes activity [
Antioxidant system would stop the radical chain reaction and direct the resultant ROS to target where it would cause less injury [
Numerous researches suggested that NO acted as an important regulator on bone metabolism [
Information about nutrient status of PO was also taken into consideration in this meta-analysis. Studies suggested that Hcy played an important role in bone metabolism and had been involved in osteoporotic facture incidence [
VB12 is essential for folate cycling and is known to be a determinant of total Hcy concentration. Our results revealed no significant change of VB12 in PO group compared with healthy control, which was not in agreement with a previous meta-analysis on the relationship between VB12 and PO [
Folate status is another determinant of total Hcy concentration. In line with the results of Hcy in PO, a significantly decreased folate level appeared in PO groups in our meta-analysis. Although a similar trend was shown in Zhang H’s study, no statistical difference was obtained in their meta-analysis [
In the present meta-analysis, we investigated possible causes of heterogeneity among studies. Host factors can be ruled out because most studies were matched by age and gender and the studies were only carried out on PO subjects. Meanwhile, we pooled only measurements carried out on the same (or similar) biological sample and with reasonably comparable methods; thus, in a few cases, the heterogeneity persisted. In addition, all the OS markers and antioxidant system were taken into consideration in our meta-analysis, which made a solid foundation for comprehensive evaluation of the relationship between OS and PO. However, there are several limitations of the present study. The positive results achieved with OS markers are, however, flawed by the small sample size and lack of evidence that these molecules actually exerted an antioxidant effect in vivo. Secondly, although we attempted to consider as many confounding factors as possible, we cannot exclude the possibility that the observed associations could be attributed to uncontrolled factors that affect the condition of OS, such as 25-hydroxyvitamin D levels or years since menopause. Thirdly, the study population was mostly Turks and Italians, hence we could not be certain that our results will be applicable in other populations. Clinical trials in the future should be carried out to further test whether these biomarkers could be the “gold standard” for diagnostics and prevention.
In summary, our meta-analysis suggests a significant association between OS and PO. The imbalance of ROS and antioxidant system may contribute to functional and structural remodeling that favors the occurrence of PO. Despite many efforts made to effectively diagnose and therapeutically prevent PO occurrence, the results with several antiosteoporotic agents are not well satisfactory. Scavenging ROS overproduction or regulation of antioxidants activity could be investigated to see whether this may represent a novel therapeutic approach to prevent PO occurrence.
The authors declare that they have no competing interests.
Qiaozhen Zhou and Li Zhu contributed equally to this work.
The authors thank Dr. Chunjie Li for his great help with meta-analysis techniques and language polishing. This work was supported by the Natural Science Foundation of China (nos. 81271186, 81500817), Zhejiang Provincial Natural Science Foundation of China (no. LY15H140008), Health Science and Technology Project of Zhejiang Province (2016KYB184), and Wenzhou Public Technical Research Medical Program (Y20140708).