In this study, we evaluated, for the first time, the prognostic potential of major cardiovascular markers, high-sensitivity cTnI (hs-cTnI), NT-proBNP, and adiponectin, for ACM and CVM in stable HD patients over a long follow-up period of 7 years. We found that hs-cTnI and NT-proBNP, but not adiponectin, were significant risk factors for ACM and CVM in univariate analysis using the Kaplan–Meier and Cox proportional hazards methods. The C-index of ACM for hs-cTnI and NT-proBNP decreased similarly over seven years, whereas the C-index of CVM at to 4–7 years was likely to be higher for hs-cTnI than for NT-proBNP. In multivariate Cox proportional regression analysis using clinical risk factors, including both hs-cTnI and NT-proBNP, the independent risk factors for ACM were hs-cTnI, age, and CRP, but not NT-proBNP, whereas the independent risk factors for CVM were hs-cTnI and DM, but not NT-proBNP. Furthermore, when hs-cTnI was added to the main model, including NT-proBNP in addition to age, CRP, and DM as independent prognostic risk factors, the NRI and IDI for ACM and CVM increased significantly throughout the entire period, leading to long-term improvement of the prognostic ability. Though the discriminatory ability of NT-proBNP was similarly evaluated, the NRI for the CVM increased significantly only in the first year. These results suggest that hs-cTnI is superior to NT-proBNP and adiponectin in predicting ACM and CVM over seven years in stable HD patients.
The correlation between biomarkers may have influenced the results of the analysis of independent risk factors. In this study, more than 99% of the measurements of hs-cTnI, NT-proBNP, and adiponectin were above the LOQ, and this confirmed the accuracy of the measurements. Based on this analysis, the main positive independent determinant of hs-cTnI was NT-proBNP, whereas the main positive determinants of NT-proBNP were DM, CRP, and hs-cTnI. In earlier reports on dialysis patients, the NT-proBNP level was positively associated with inflammation or CRP9,25,26. The association between NT-proBNP and several factors may partially explain why it was not selected as an independent risk factor for mortality after adjusting for age, DM and hs-cTnI levels. Adiponectin was not an independent risk factor for mortality in this study, probably because it was also independently and positively correlated with hs-cTnI levels, higher HDL-C levels, and female sex. Previous studies have also reported significant positive associations between adiponectin and cTnI27 and NT-proBNP13,28.
In the present study, the middle tertile (14.6–30.9 ng/L) and the high tertile (> 31.1 ng/L) were independently correlated with ACM and CVM, which demonstrated that even a mild increase in cTnI within the reference range was associated with poor prognosis. A large-scale prospective study of the general population revealed a cTnI concentration-dependent prediction of CVM at 13 years even in the lower half of the reference range, that is, ≤ 6 ng/L. This indicated that cTnI measurements within the reference range reflected the status of subclinical myocardial stress29. Myocardial troponin release is caused not only by necrosis and apoptosis of acute myocardial injury but also by chronic diseases, such as myocardial overload, myocardial wall strain, and myocardial ischemia7. Chronic myocardial stress leads to myocyte turnover, intracellular cTnI degradation, increased plasma membrane permeability, and plasma membrane bleb formation30,31. cTnI may be released from stressed viable cardiomyocytes via plasma membrane shedding of vesicular blebs containing cytoplasmic TnI30. In previous studies on dialysis patients, a cTnI cut-off value of 14–30 ng/L within the reference range had significant predictive value for ACM at 1–4 years2,9,32, with which our findings are aligned. The elevated cTnI levels in patients with renal insufficiency can be attributed to myocardial damage/stress and decreased clearance3,7,9. However, cTnI measurements were only slightly affected by reduced clearance, as clearance occurs in the liver and kidneys33. Approximately one-third of our participants had cTnI levels above the reference range. In contrast, NT-proBNP levels exceeded the reference range in almost all participants in our study because NT-proBNP mainly undergoes renal clearance3.
One of the most important findings of this study was that the baseline levels of hs-cTnI, but not NT-proBNP, were significant independent predictors of ACM and CVM at 7 years and had better prognostic discriminatory power for ACM and CVM than NT-proBNP throughout the 7-year follow-up period. cTnI and NT-proBNP had almost identical prognostic significance with 4-year follow-up9, and that cTnI had a slightly stronger predictive power for ACM than NT-proBNP, although both were significant predictors2. In these studies, most NT-proBNP measurements were above the LOQ, whereas more than 50% of the cTnI measurements were below the LOQ (17 and 30 ng/L)2,9. This partial imprecision of cTnI values may have resulted in failure to confirm the superiority of cTnI. In a study of HD patients using a highly sensitive assay with an LOQ of 3.8 ng/L, cTnI had a greater discriminatory power for mortality at 5 years compared to NT-proBNP34. In contrast, in a study of HD and peritoneal dialysis patients using a low-sensitivity cTnI assay with an LOQ of approximately 100 ng/L, NT-proBNP had a significantly greater predictive power for mortality at several years compared to cTnI26,35. Another reason may be differences in the intra-individual variability of the measurements. The reference change values of high-sensitivity cTnI were approximately half of those of NT-proBNP36,37,38. Accordingly, high-sensitivity cTnI may be more likely than NT-proBNP to reflect fundamental and persistent cardiac conditions in a single measurement and may be more strongly associated with long-term prognosis than NT-proBNP. Additional explanation may be the differences in dialysis modalities, considering that in studies on patients on peritoneal dialysis, NT-proBNP was reported to be a stronger predictor of mortality than high-sensitivity troponin25,36.
Few studies have examined the long-term prognostic value of biomarkers over 5 years in HD patients. In an earlier study of Japanese HD patients, the AUCs of NT-proBNP for both ACM and CVM gradually decreased at 1, 3, and 5 years39, which was similar to our findings for NT-proBNP throughout the 7-year study period. In a report using non-high-sensitivity cTnI, the discrimination ability for sudden cardiac death was significantly greater with NT-proBNP than with cTnI at both 3 and 5 years26. In contrast, in the present study, hs-cTnI was superior to NT-proBNP in the discrimination ability for ACM and CVM from 1 to 7 years of age, with CVM being much greater. The difference between the earlier report and ours may be attributed to the difference in the sensitivity of the cTnI assay.
As shown in Table 2, ERI showed a univariate predictor of ACM but not CVM, while preexisting CVD was a univariate weak predictor of clinical outcomes in contrast to hs-cTnI as the powerful predictor. These unique associations of ERI and preexisting CVD with the outcomes can be explained as follows: ERI showed a univariate positive association with age (β = 0.236, P < 0.001) and negative ones with BMI (β = − 0.333, P < 0.001), GNRI (β = − 0.331, P < 0.001) and TG levels (β = − 0.278, P < 0.001). ERI was associated with 3 univariate predictors of ACM, age, GNRI, and TG, and with one univariate predictor of CVM: only age. This difference may cause a stronger univariate association of ERI with ACM than CVM (Table 2). Next, preexisting CVD showed weak univariate positive associations with risk factors of the outcomes: DM (β = 0.265, P < 0.001), CRP (β = 0.162, P < 0.05), hs-cTnI (β = 0.230, P < 0.001) and NT-proBNP (β = 0.160, P < 0.05). CVD patients were considered to take preventive treatments for recurrence before the participation, which may lead to reduced levels of CRP, hs-cTnI, and NT-proBNP. Consequently, associations between these risk factors and preexisting CVD may have weakened. Furthermore, no significant difference in age was found between CVD and non-CVD patients (68.7 ± 10.9 vs 65.9 ± 13.2 years, P = 0.12). These can explain why associations of preexisting CVD with ACM and CVM were much weaker than those of hs-cTnI.
Finally, this study found weak positive associations of eGFR with cTnI, NT-proBNP, and age, contrary to general expectations. In this study, eGFR values were calculated from the Japanese equation using Cr levels16. Consequently, they depended mainly on Cr levels, where the two factors had a strong inverse relationship (β = − 0.823, P < 0.001). Since Cr levels were associated negatively with age (β = − 0.513, P < 0.001) and adiponectin (β = − 0.287, P < 0.001) and positively with GNRI (β = 0.413, P < 0.001), eGFR was resultantly associated positively with age and adiponectin and negatively with GNRI as shown in the Result section. Furthermore, considering an earlier report that 69% of HD patients had residual renal function (RRF) loss, defined as a urine volume below 200 mL/day, 20 months after HD initiation40, most of our participants may have presented with RRF loss, because of having the median HD vintage of 86 months. Taking these into account, eGFR only marginally reflected renal clearance and showed no negative association with cTnI or NT-proBNP. Furthermore, cTnI and NT-proBNP were associated positively with age and adiponectin and negatively with GNRI. These correlations were similar to those found for eGFR, which may have led to the weak positive associations of eGFR with cTnI and NT-proBNP.
This study has several limitations. First, the results of this single-centre small-scale clinical study need to be validated in a large-scale, long-term prospective multicentre study that quantifies the hs-cTnI. Second, the methods for measuring hs-cTnI, NT-proBNP, and adiponectin levels have not been standardised or harmonised, and the cut-off values cannot yet be strictly applied to uraemic patients. Third, this study evaluated the prognostic performance of a single measurement but not the average value of multiple measurements. Some studies reported no difference in the prognostic value between single and multiple measurements41, whereas others reported that multiple measurements were better36,37. Fourth, though cTnI may be adsorbed onto dialyser membranes42, this study did not compare the pre- and post-dialysis levels; therefore, further investigation on this aspect is needed.
In conclusion, after adjusting for clinical risk factors, we demonstrated that hs-cTnI was superior to NT-proBNP and adiponectin for predicting and discriminating ACM and CVM over 7 years in patients who were undergoing HD, suggesting the significance of baseline hs-cTnI measurements in long-term management of patients who require HD. To validate these findings, a large-scale, long-term, prospective, multicentre study using hs-cTnI assay is required.
- The Renal Warrior Project. Join Now
- Source: https://www.nature.com/articles/s41598-024-62491-4