Baseline characteristics
Demographic data for the cohort as related to chronological age are presented in Table 1. The difference in the red series is observed, presented in the lower level of hemoglobin and RBC in the older group (p<0.01). Also significant differences are observed in the number of the platelets and lymphocytes, the older group has fewer platelets and lymphocytes (p<0.05). And the levels of AST, TC, TG, TC/HDL-c, and glucose increased with age. On the contrary, levels of Alb decreased with age. Furthermore, the older group has a higher levels of GDF-15 and lower expression of hTERT mRNA, and PBMC telomerase activity. No significant differences in the distribution of the number of WBC, Neu, Mono, neutrophil to lymphocyte ratio, and levels of ALT, BUN, creatinine and LDL-c are observed.
Table 2 shows the baseline characteristics as stratified by GDF-15 tertiles. The medians of GDF-15 levels varied from 338.52 pg/mL in the bottom tertile up to 640.71 pg/mL in the top tertile. Patients with high GDF-15 levels are older, had higher number of blood monocytes, when compared to patients with low GDF-15 levels (p<0.05). Kim et al, showed that the frailty and frailty criterion were significantly associated with lower erythrocyte levels of long-chain n-3 PUFA[16]. Consistent with the previous study, in our study, the older group has less RBC. And the levels of AST and TC/HDL-c increased with GDF-15 levels. With regard to the numbers of WBC, Neu, LYM, neutrophil to lymphocyte ratio, platelets and levels of Hb, ALT, BUN, creatinine, TC, TG, HDL-c, LDL-c and glucose, no significant statistical difference is observed among different GDF-15 tertiles.
Telomere-related parameters across different GDF-15 tertiles.
Telomere length in PBMCs, relative levels of hTERT mRNA in leukocytes, and the activity of PBMC telomerase stratified by GDF-15 tertiles are presented in Table 3. Significant difference in the telomere-related parameters were observed, presented in the lower level of hTERT mRNA and T/S ratio in individuals with high GDF-15 levels (p<0.01). Moreover, the measurements of PBMC telomerase activity showed a consistent direction of effect over GDF-15 groups. Patients with high GDF-15 levels had a lower PBMC telomerase activity.
Correlations among GDF-15 and various biomarkers.
Fig. 1 shows the correlation among GDF-15 levels with various biomarkers. A statistically significant association between GDF-15 levels and hematological parameters is observed, presented in the negative correlation between GDF-15 and RBC (r=-0.303, p=0.003), and positive correlation between GDF-15 and Mono (r=0.198, p=0.031). A statistically significant positive association was found with lipids, such as LDL-c and TC/HDL-c (r=0.207, p=0.034; r=0.196, p=0.044, respectively). Also, GDF-15 levels positively correlated with AST (r=0.336, p<0.001), and negatively correlated with Alb (r=-0.336, p=0.001).
Correlations among age, T/S ratio, leukocyte hTERT mRNA levels, PBMC
telomerase activity and GDF-15.
To further assess the correlation among age, T/S ratio, leukocyte hTERT mRNA levels, PBMC telomerase activity and GDF-15, the Spearman correlation method was performed. As shown in Fig. 2, GDF-15 levels are positively correlated with age (r=0.664, p<0.001) (Fig. 2A), and negatively correlated with PBMC telomerase activity (r=-0.231, p=0.012), telomere length (r=-0.434, p<0.001), and WBC hTERT mRNA levels (r=-0.206, p=0.024) (Fig. 2B, C and D).
Univariate and multivariate linear regression
To further explore the association of GDF-15 with T/S ratio, leukocyte hTERT mRNA levels, and PBMC telomerase activity, we performed multivariable linear regression, respectively.
Table 4 displays the results of the univariate and multivariate regression analysis of PBMC telomerase activity and GDF-15. In the univariate linear regression analysis, PBMC telomerase activity positively associated with numbers of LYM, RBC, PLT, and levels of Alb, glucose, TC. In addition, PBMC telomerase activity negatively associated with GDF-15 levels. In the multivariate linear regression analysis, PBMC telomerase activity showed a statistically significant linear and negative relationship with GDF-15 levels (β-coefficient=-0.583, 95% CI -1.044 to -0.122, p=0.014) after adjusting for Neu, RBC, Alb and TC.
Table 5 shows that telomere length is negatively associated with GDF-15 levels (β-coefficient=-0.200, 95% CI -0.305 to -0.094, p<0.001) after adjusting for glucose, TG, and HDL-c.
The association between hTERT mRNA and GDF-15 is also further evaluated by multivariate regression analysis. As summarized in Table 6, the expression of hTERT mRNA showed a statistically significant linear and negative relationship with GDF-15 levels (β-coefficient=-0.207, 95% CI -0.312 to -0.102, p<0.001) after adjusting for glucose.