A total of 211 patients on hemodialysis in a single center were enrolled in the restrospective study. Stable patients with no cardiovascular events in the 4 weeks before serum determinations were included. During the follow up [39 (19–56) months], patients with changes in hemodialysis parameters, transferred to another center or transplanted were censored. Investigations were in accordance with the Declaration of Helsinki.

Baseline characteristics were recorded, including age, sex, presence of diabetes mellitus, peripheral vascular disease, previous cardiovascular disease (congestive heart failure determined by echocardiography within the three previous months, myocardial infarction, cerebrovascular disease), dialysis vintage and data regarding the vascular access. Basally, we measured C-reactive protein (CRP), high sensitivity troponin T (hsTnT), CK-MB and N-terminal prohormone of brain natriuretic peptide (Nt-proBNP). All included patients had the same hemodialysis therapy protocol: 4h, three times per week. Routine clinical and biochemical variables were measured by standardized methods on autoanalyzers. CKMB and HsTnT were measured on a Roche/Hitachi Cobas E411 analyzer. Factors related to higher values of CKMB were analyzed.

Echocardiography was recorded in stable patients who had a less than 6-month one before obtaining the sample and within 24h after the last hemodialysis on the day between mid-week dialysis sessions. Diastolic dysfunction was defined as e′ (early mitral annulus velocity) less than 8cm/s, average E (early mitral flow)/e′ over 8, LAVi (left atrium volume index) over 28mL/m2 or Ar−A (time difference between duration of the pulmonary venous atrial reversal wave and duration of the A wave) over 30ms. Systolic dysfunction in turn was defined as a left ventricle ejection fraction of under 45%. The left ventricular mass index (LVMi) was estimated by Devereux's formula.13

New cardiovascular events (ischemic or hemorrhagic cerebrovascular accident, cardiac event [including myocardial infarction and/or congestive heart failure], peripheral vascular events and other ischemic events) were recorded during follow-up. We analyzed the predictor value of CKMB for cardiovascular events.

Values are expressed as the mean (SD) or median (IQR). We established linear regression models for evaluating the distribution of the studied variables according to CKMB levels. To assess the diagnosis value of the different cardiac markers we used the receiving operator characteristics (ROC) curve for each one. Correlations between cardiac markers were performed using Pearson test. Multivariate analysis was performed by Cox regression. Variables were analyzed and only those considered confounders were entered in the final Cox regression model. Different models were used, including CKMB for its different values, in order to determine the best cut-off to assess cardiovascular risk. Cardiovascular events were analyzed using Kaplan–Meier plots, and survival curves were compared using the log-rank test. We used the integrated discrimination improvement (IDI), as described by Pencina et al.14 to interpret the incremental value of CKMB≥2ng/mL added to a risk prediction model including age, sex, previous heart disease, peripheral vascular disease, diabetes mellitus, NtproBNP and HsTnT. IDI is a measure of improvement in model performance and represents the difference between discrimination slopes of two competing models. We calculated the relative IDI, which expresses the relative increase in separation of events and non-events from the separation achieved in the base model (i.e., the difference in discrimination slopes is expressed as a proportion of the discrimination slope of the base model).15 All statistical analyses were performed with the SPSS® 18.0 statistical package (Chicago, IL, USA). A p-value <0.05 was considered statistically significant.

A total of 211 prevalent hemodialysis patients were included in this study to be followed for 39 (19–56) months. One hundred and twenty-three patients (58.3%) were male, with a median age of 73 (60–80). Baseline characteristic are showed in Table 1. The median value of CKMB was 1 (1–2) ng/mL, with the following distribution: 13 patients (6.3%) have basal levels of 0ng/mL, 121 (58.5) have 1ng/mL, 56 (27.1%) have 2ng/mL, 13 (6.3%) have 3ng/dL and only 4 (1.9) have 4ng/dL. In Table 2, baseline characteristics are shown in the different strata according to CKMB values. No patients exceeded the normal range of our laboratory (0–4ng/mL). Univariate analysis revealed that higher levels of CKMB were associated to previous heart disease, diabetes mellitus, peripheral vascular disease and systolic and diastolic dysfunction (data shown in Table 3).

Correlation between cardiac biomarkers showed a positive and significant one between CKMB and NtproBNP (0.163, p=0.020) and between NtproBNP and HsTnT (0.635, p<0.001). No significant correlation was established between CKMB and HsTnT.

A total of 94 cardiovascular events were recorded during the follow up. Cardiac event was the most common event (79.8%), followed by peripheral vascular disease (8.5%), cerebrovascular event (6.4%). Diabetic patients did not show higher incidence of cardiovascular events in our cohort (p=0.3), although they have a trend of higher prevalence of diastolic dysfunction (p=0.056). The area under the ROC curve for cardiovascular events was greater for HsTnT (0.723) and Nt-proBNP (0.688) than CKMB (0.587). CKMB levels were associated to the development of cardiovascular events during follow up [HR 1.46 95% CI (1.15–1.87), p=0.002] as well as the other factors shown in Table 3. In Fig. 1, the survival curve shows the association between cardiovascular events and the different CKMB values confirming that higher levels condition worse prognosis (logRank 8.8, p=0.01). Multivariate Cox regression model adjusted for several cofounders and variables showed that CKMB levels ≥2ng/mL independently increased cardiovascular risk in our cohort of hemodialysis patients. A non-significant trend was observed if the CKMB cut-off was ≥1ng/mL (Table 4).

Fig. 1.

Kaplan–Meier plot illustrating cardiovascular events and CKMB values.

(0.11MB).

IDI analysis was performed to assess the improvement in risk discrimination of adding CKMB (≥2ng/mL) to a cardiovascular event risk prediction model including age, sex, previous heart disease, peripheral vascular disease, diabetes mellitus, NtproBNP and HsTnT. This analysis comprised all individuals and used NtproBNP and HsTnT as dichotomous variables. Adding CKMB resulted in 17% improved risk discrimination for cardiovascular events (IDI 0.026 [0.004–0.053]; relative IDI 9.9%; p=0.04).