Correlation and mediation analysis between plasmapheresis donation behavior and bone mineral density and bone metabolism biomarkers: a cross-sectional study based on plasmapheresis donors at high risk of osteoporosis in China

Study design and participants

This is a multicenter, cross-sectional study. From July 1, 2022 to September 30, 2022, plasmapheresis stations in Linwu (Hunan Province) and Jiange (Sichuan Province) recruited plasma donors as research subjects (the specific process of including and excluding subjects can be found in the Fig. S1). In order to minimize the interference of both primary and secondary osteoporosis risk factors as confounding factors, we used the International Osteoporosis Foundation (IOF) Osteoporosis Risk 1-Minute Test Questions for donor screening to ensure that no donor already exists the risk of osteoporosis. This questionnaire is recommended as a commonly used tool for osteoporosis risk assessment in the Guidelines for the Diagnosis and Management of Primary Osteoporosis 2017 (Xia et al., 2019).

Inclusion criteria: (1) Registered qualified plasmapheresis donors (repeat donors,we named this group donors) who have donated more than 50 times and have continued to donate in the past six months, as well as new donors (we named this group controls) who have no history of plasmapheresis donation; (2) Male donors are over 50 years old, and female donors are postmenopausal; (3) Complete “the International Osteoporosis Foundation (IOF) Osteoporosis Risk 1-Minute Test Questions”, all answers are no; (4) Obtain the consent of the plasmapheresis donor and sign the “Informed Consent Form of the Plasmapheresis Donor”. Exclusion criteria: (1) Donors who may be affected by X-rays are excluded; (2) Taking steroid hormones or drugs that affect BMD and bone metabolism in the past 3 months should be excluded.

This study passed the review of the Ethics Committee of the Institute of Blood Transfusion, Chinese Academy of Medical Sciences (approval number: NO: 2022029), and had each subject sign an informed consent form.

Sample size estimation

This article uses a cross-sectional study sample size calculation formula for estimation (Wang & Cheng, 2020), and the formula is as follows where p represents the expected prevalence of osteoporosis in the donor population. Since there is no survey on the prevalence of osteoporosis in the donor population, and the average age of the population surveyed in this survey is over 50 years old. Therefore, using the results of the 2018 China Osteoporosis Epidemiological Survey in people over 50 years old  (National Health Commission of the People’s Republic of China, 2018), it can be seen that p is 19.2%. q = 1 − p, d is the allowable error, and z is the significance test statistic. Set α = 0.05, d = 0.2 × p, z = 1.96. According to the formula, the sample size needs to be at least 421 people. ${\displaystyle n=\frac{pq}{{\left(d/z_{\alpha }\right)}^2}=\frac{z_{\alpha }^2\times pq}{d^2}.}$

Blood sampling and information collection

A 3 ml venous blood was collected from each donor in a vacuum blood collection tube with a separating gel before donating plasma. Centrifugation (3,000r/min, 10min) was used to separate the serum, and the serum was stored at low temperature in a −20 °C freezer, and finally detected in the same lab. Blood test items include: serum total calcium, albumin (ALB), 25OHD, PTH, P1NP, β-CTX, SF levels.

Indicators of plasmapheresis donation behaviors including the total number of plasmapheresis donations, the recent plasma donation frequency (i.e., the number of plasmapheresis donations in the last 12 months) and the interval from the last plasma donation were retrieved from the information system of the plasmapheresis station. Self-made questionnaires were used to collect background information such as age, BMI, annual household income, meat intake, calcium supplementation, and vitamin D supplementation. The International Physical Activity Questionnaire (IPAQ) was used to collect the physical activity of plasmapheresis donors.

Laboratory analysis and assays

P1NP was detected by total type I collagen amino-terminal elongated peptide detection kit (batch number: 653011, Roche, Basel, Switzerland), and the detection instrument was Roche E602. PTH was detected using a parathyroid hormone assay kit (batch number: 220901, Mindray, Shenzhen) and the testing instrument was Mindray CL-6000i. β-CTX was detected with β-collagen special sequence detection kit (batch number: 635899, Roche, Basel, Switzerland), and the detection instrument was Roche E602. Albumin was detected using an albumin detection kit (batch number: AUZ1125, Beckman, Brea, CA, USA) and the detection instrument was Beckman 5800. Serum calcium was detected by calcium assay kit (batch number: AUZ1199, Beckman, Brea, CA, USA), and the detection instrument was Beckman 5800. 25OHD was detected by total 25-hydroxyvitamin D assay kit (batch number: 230501, Mindray, Shenzhen), and the detection instrument was Mindray CL-6000i. SF was detected using chemiluminescence method using a ferritin assay kit (batch number: 47177UDOO, Abbott, Chicago, IL, USA) and the detection instrument was Abbott Alinity.

DXA was used to measure the BMD of anterior and posterior lumbar spine (L1-L4), bilateral femoral neck and non-dominant radial BMD of plasmapheresis donors with GE LunarDPX Prodigy. In both Linwu and Jiange plasmapheresis stations, Haier Blood Technology XJ-II blood component apheresis machine was used for plasmapheresis donation.

Study outcome parameters

P1NP was measured by electrochemiluminescence with normal reference values ranging from 16.3 to 78.2 ng/mL, and data were cited from the Guidelines for Clinical Application of Bone Turnover Biomarkers (Li, Xia & Zhang, 2021). β-CTX was measured by electrochemiluminescence, and the normal reference value ranged from 0.114–0.628 ng/mL, and the data were quoted from the 2021 Guidelines for the Clinical Use of Biochemical Markers for Osteotransition. Albumin was determined by BCG, with normal reference values ranging from 40 to 55 g/L, and data were quoted from the Hong, Yusan & Ziyu (2014). Serum calcium was determined by colorimetric method, and the normal reference value ranged from 2.2–2.7 mmol/L, and the data were quoted from the “National Clinical Laboratory Operating Regulations”. PTH was determined by chemiluminescence method, and the normal reference value ranged from 12-65 pg/mL, and the data were quoted from the “National Clinical Laboratory Operating Regulations”. 25OHD was determined by electrochemiluminescence, and vitamin D inadequacy was indicated at <20 ng/ml, as shown in the Guidelines for the Use of Vitamin D and Bone Health in Adults (Liao et al., 2014). According to the WHO criteria for iron accumulation (World Health Organization, (2020)), male SF > 200 ug/L and female SF > 150 ug/L are classified as iron accumulation. BMD is presented as g/cm² and as T-score. Based on the DXA classification criteria for BMD, a T-value of ≥-1 is normal, a T-value of −2.5 <T <-1 is low bone mass, and a T-value of ≤−2.5 is osteoporosis, as shown in the Guidelines for the Diagnosis and Management of Primary Osteoporosis 2017 (Xia et al., 2019). According to the guideline, a T-value ≤−2.5 at either site is diagnosed as osteoporosis.

Statistical analysis

Graphpad Prism 8.0 and SPSS 24.0 (Armonk, NY, USA) were used for data collation and statistical analysis. Firstly, the overall population was descriptively statistically analyzed, and the basic differences between the two groups of new plasmapheresis donors and repeat plasmapheresis donors were compared. For the continuous data, the data conforming to the normal distribution were expressed as mean and standard deviation, and one-way ANOVA was used. Data that do not conform to the normal distribution are expressed by median and interquartile ranges, and Wilcoxon rank-sum test is used. The frequency (composition ratio) was used to express the counting data, and the chi-square test was used. The overall population was divided into groups according to the type of donor, and whether there were statistically significant differences between groups in bone biomarkers. The statistically significant indexes were analyzed according to three different indicators of donation behaviors (i.e., the total number of plasmapheresis donations, recent frequency, and the interval from the last plasma donation), and multiple linear regression analysis was used to further assess the association between donation behaviors and bone metabolism biomarkers. We use Restricted Cubic Splines (RCS) to better demonstrate the nonlinear trend of change between bone metabolism biomakers and plasmapheresis donation behaviors. The Bootstrap method was used to test the mediating effect of SF on bone metabolism markers in relation to plasma donation behaviors. Finally, analyze whether there are statistically significant differences in BMD at different locations among different types of plasmapheresis donors (new and repeated donors). When conducting multiple comparisons, the Bonferroni method is used to adjust the P-value.

The test level was α = 0.05, and p < 0.05 indicated that the difference was statistically significant.

Quality control

In the information collection stage, a unified and standardized project operation manual and questionnaire were produced and the staffs involved in the project were trained through online meetings to ensure the standardization of data collection. In the data entry stage, two independent members entered separately and then cross-checked to ensure the accuracy of the data. In the statistical analysis stage, the detection value of serum total calcium was corrected by the correction formula [corrected calcium = measured calcium + (40-measured albumin) × 0.02] and included in the statistical analysis.

Baseline characteristics

A total of 553 donors were included in this study, including 276 new donors (i.e., the control group) and 277 repeat donors (i.e., the donor group), respectively. There were statistically significant differences in age, annual household income, exercise intensity, protein intake and calcium supplementation between the two groups (P < 0.05), as shown in Table 1. There were significant differences in the levels of 25OHD, P1NP, β-CTX and PTH between the two groups (P < 0.05). The inadequacy rate of vitamin D in the repeat donor group (82.31%) was significantly higher than that in the new donor group (68.48%) (P < 0.001).

Effects of different donation behavior indicators on bone metabolism biomarkers

In this part of the study, the statistically different indicators related to bone metabolism biomarkers in Table 1 was grouped according to three donation behavior indicators (i.e., total number of donations, the recent frequency of donation, and interval between the last donation). The results showed that the level of 25OHD and β-CTX was lower while PTH and PINP were higher in the repeat donor group. High total number of donations, high recent frequency and short interval have significant influence on these biomarkers. After 100 times of plasmapheresis donation in Fig. 1, it has an impact on the bone formation indicators (P1NP). When the frequency of plasmapheresis donation increases to more than 17 times recently, both bone resorption (β-CTX) and bone formation (P1NP) indicators undergo significant changes. When the interval between plasmapheresis donation exceeds 14 days, it has no effect on the levels of bone resorption (β-CTX) and bone formation (P1NP) indicators. When the interval between plasmapheresis donation exceeds 28 days, the levels of 25OHD and PTH return to before plasmapheresis donation. In summary, the main influencing factors of bone metabolism related indicators are recent plasmapheresis donation frequency. When it increases to a certain level (i.e., more than 17 times/nearly 12 months, note: the current upper limit in China is 24 times/nearly 12 months), both bone resorption and bone formation indicators undergo significant changes. When the interval between plasmapheresis donation exceeds 14 days (the current minimum interval requirement in China), the levels of bone resorption and bone formation indicators return to before plasmapheresis donation. When the interval between plasmapheresis donation exceeds 28 days, the levels of 25OHD and PTH return to before plasmapheresis donation. As the total number of plasmapheresis donation increases, it first affects bone resorption indicators, and after 100 plasmapheresis donations, it will further affect bone formation indicators.

Multiple linear regression analysis on bone metabolism biomarkers

Multiple linear regression analysis was performed for bone metabolism biomarkers (25OHD, P1NP, β-CTX, PTH) with significant differences in Table 1, and adjusted for a large number of possible confounding factors (age, body mass index (BMI), annual household income, physical activity, protein intake, calcium supplementation and vitamin D supplementation). Only indicators of plasmapheresis donation behaviors that were clearly risk factors were shown (Table 2). The results showed that the level of 25OHD was positively correlated with the total number of plasmapheresis donations, and negatively correlated with the frequency of plasmapheresis donation. The level of P1NP was positively correlated with the total number of donations, the higher the number of donations, the higher the level of P1NP. PTH level was positively correlated with the frequency of donation, while β-CTX level was negatively correlated with the frequency of donation. The interval from the last plasma donation has no correlation with changes in the indicators after adjustment.

Nonlinear trends between indicators of plasmapheresis donation behaviors and bone metabolism biomarkers

From Fig. 2, it can be seen that non-linear relationships existed between biomarkers and behavior indicators in Table 2. The 25OHD level decreased rapidly at first with the increase of the total numbers, then began to rise towards 100 times, and then remained stable after 200 times. However, it decreased rapidly with the increase of recent frequency. It may be noteworthy that even new donors have vitamin D inadequacies, which may be exacerbated by plasma donation. Both PTH and P1NP levels showed an upward trend as the total number and recent frequency increased. β-CTX decreased slowly as the recent frequency decreased, but values remained within the normal range.

Baseline comparison and mediation analysis of SF

The results indicate that there is a statistical difference between the two groups at the SF level. Compared with the control group, the donor group had lower SF levels. The iron accumulation rate in the group of women who repeatedly donate plasma is significantly lower than that in the group of new donors (P < 0.001). There is no statistically significant difference in iron accumulation rate between male repeat donors and new donors (Table S1).

Total number of donations, the recent frequency of donation, interval between the last donation, SF, β-CTX and P1NP were included in the Pearsoncorrelation analysis to analyze the correlation between the two pairs. There is a correlation between total numbers and the levels of SF and P1NP, with correlation coefficients of −0.306 and 0.148, respectively. The level of SF is correlated with total numbers and P1NP, with correlation coefficients of −0.306 and −0.187, respectively. P1NP is correlated with total numbers and SF (Table S2). The above results indicate that there is a correlation between total numbers, SF, and P1NP pairwise. Therefore, total numbers, SF, and P1NP were selected for the next analysis. Use Bootstrap method for mediating effect analysis. According to the research purpose of this article, the variable details are as follows: total numbers is the independent variable, SF level is the mediator variable, and P1NP level is the dependent variable, as shown in Fig. 3. According to the results in Table 3, it can be concluded that SF partially mediates the relationship between the total number of plasma donations and the level of P1NP. The total effect is c = 0.156, the direct effect is c′ = 0.122, and the indirect effect is a * b = 0.034. The mediating effect percentage is a∗b/c = 21.8%.

Analysis related to BMD

There was no significant difference in BMD between the two groups (P > 0.05). There was no difference in the prevalence of osteoporosis and low bone mass rate between the two groups, indicating that there was no significant difference in BMD between the two groups (Table 4). Meanwhile, we conducted multiple linear regression analysis on BMD in different parts and found that plasmapheresis donation related behaviors were not independent factors affecting BMD (Table S3), which is consistent with the single factor results. Therefore, this indicates that plasmapheresis donation does not affect the donor’s BMD.