Serum T50 predicts cardiovascular mortality in individuals with type 2 diabetes: A prospective cohort study

Individuals with type 2 diabetes (T2D) have a higher risk of cardiovascular disease, compared with those without T2D. The serum T50 test captures the transformation time of calciprotein particles in serum. We aimed to assess whether serum T50 predicts cardiovascular mortality in T2D patients, independent of traditional risk factors.

323 ± 63 min.Higher age, alcohol use, highsensitive C-reactive protein, and plasma phosphate were associated with lower serum T 50 levels.Higher plasma triglycerides, venous bicarbonate, sodium, magnesium, and alanine aminotransferase were associated with higher serum T50 levels.After a follow-up of 7.5 [5.4-10.7]years, each 60 min decrease in serum T50 was associated with an increased risk of cardiovascular (fully adjusted HR 1.32, 95% CI 1.08-1.50,and p = 0.01) and all-cause mortality (HR 1.15, 95%CI 1.00-1.38,and p = 0.04).Results were consistent in sensitivity analyses after exclusion of individuals with estimated glomerular filtration rate <45 or <60 mL/min/1.73m 2 and higher plasma phosphate levels.
Conclusions.Serum T50 improves prediction of cardiovascular and all-cause mortality risk in individuals with T2D.Serum T50 may be useful for risk stratification and to guide therapeutic strategies aiming to reduce cardiovascular mortality in T2D.

Background and aims
Despite adequate control of blood glucose levels and traditional risk factors for cardiovascular disease, individuals with type 2 diabetes (T2D) still face a three times higher age-adjusted relative risk for cardiovascular disease, as compared to the gen-eral population [1,2].At least part of the excess risk in persons with T2D is driven by a susceptibility to cardiovascular disease [3].The novel serum T50 test presents an innovative approach to evaluating the transformation of calciprotein particles (CPP), providing valuable insights into the future risk of cardiovascular disease that methods detecting prevalent calcifications such as computed tomography (CT) scans do not capture.
CPPs physiologically occur in serum, where they play a role in the precipitation of supersaturated calcium and phosphate [4].Calciprotein monomers (CPM) undergo spontaneous aggregation into primary CPPs, which can then transform into secondary CPPs.These secondary CPPs are associated with promoting vascular media calcification and contribute to pre-atherosclerotic changes in the vessel wall, including intimal hyperplasia [5].The serum T50 test has been developed and validated to quantify the capacity of human serum to prevent CPP2 formation [6].The test measures the rate by which primary CPPs are transformed into secondary CPPs in a blood sample, with lower serum T50 values (i.e., shorter conversion time) indicating a higher propensity for pathophysiological changes, including calcification [6,7].
In the evolving landscape of cardiovascular risk assessment, the serum T50-test emerges as a minimally invasive, cost-effective tool that offers the potential for early intervention, positioning it as a promising prognostic marker.Previous studies demonstrated that a lower serum T50 is independently associated with cardiovascular mortality in individuals suffering from chronic kidney disease (CKD) [8][9][10] and in the general population [11].In the latter study, this association seemed particularly strong in a subgroup of individuals with T2D at baseline.Furthermore, glycated hemoglobin was identified as a relevant independent determinant of serum T50 [12].At present, the relationship between serum T50 and (cardiovascular) mortality risk has not been assessed in a dedicated cohort study in individuals with T2D.
Therefore, in this study, we aimed to assess whether serum T50 predicts cardiovascular (CV) and all-cause mortality inindividuals with T2D.Second, we aimed to assess determinants of T 50 in this population.

Study design
The DIAbetes and lifestyle cohort twente study (DIALECT) is a prospective observational cohort study conducted in the Netherlands [13].The aim of the study is to examine the impact of lifestyle and dietary habits, as well as pharmacological manage-ment, on outcomes in persons with T2D treated in a secondary healthcare center.A total of 668 patients were enrolled between 2009 and 2019, of which 621 had available follow-up data in April 2023.This is a post hoc analysis using samples that have been added to a biobank that was built when the baseline visit of the DIALECT cohort took place.The study adhered to good clinical practice guidelines and the Declaration of Helsinki, with all participants providing written informed consent prior to participation.The study received approval from local institutional review boards (METC-registration numbers NL57219.044.16 and 1009.68020).

Participants
The study population included individuals with age ≥18 years and T2D who were receiving routine secondary care treatment at the outpatient clinic.Patients who were undergoing renal replacement therapy or lacked adequate proficiency in the Dutch language were excluded from participation.All 621 individuals within this cohort were included based on their possession of follow-up data.All individuals had complete serum T 50 measurements.

Baseline data
Patients underwent eligibility screening via electronic patient records before being asked to attend a research appointment.At the outpatient department, a thorough medical assessment was carried out.Measurements of height, weight, waist, and hip circumference were taken to compute the body mass index (BMI) as weight divided by the square of height (kg/m 2 ).Blood pressure was assessed while the patient was lying down, using an automated apparatus (Dinamap; GE Medical Systems) for a duration of 15 min at 1-min intervals [14,15].The mean systolic and diastolic pressures from the final three readings were utilized for subsequent analyses.
The presence of microvascular disease was determined by evaluating the occurrence of nephropathy, neuropathy, or retinopathy.Nephropathy was identified as an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73m 2 with or without albuminuria.Neuropathy was evaluated using monofilament and VibraTip.Retinopathy was assessed by analyzing fundus images taken with a retinal camera, which were then examined by an ophthalmologist at 1-2-year intervals.
The existence of macrovascular disease at the start of the study was established by a record of prior coronary heart disease, cerebrovascular disease, or peripheral arterial disease, as noted in the patient's medical history.The diagnosis of coronary heart disease was confirmed by a clinician's diagnosis of conditions, such as unstable angina, myocardial infarction, interventions like percutaneous coronary angioplasty, or surgical procedures like coronary artery bypass grafting.A background of cerebrovascular disease was denoted by episodes of transient ischemic attacks or stroke, whereas peripheral arterial disease was ascertained through confirmed arterial disease via angiography or magnetic resonance angiography, interventions like percutaneous transluminal angioplasty, or surgical procedures such as peripheral arterial bypass grafting.
Non-fasting venous blood was collected to perform routine laboratory tests.This includes blood count, estimated renal function, liver function, HbA1c, and cholesterol profile.NT-proBNP levels were measured by electrochemiluminescence immunoassay in plasma EDTA.The eGFR was calculated using the 2009 CKDEpidemiology Collaboration formula.All blood samples, 24-h urine collections, and morning urine specimens were preserved in a biobank at a temperature of −80°C for subsequent examination.Serum T50 measurements were conducted following established methods, with minor modifications to the original method outlined by Pasch et al. [6].Serum samples (40 µL) were transferred to transparent 96-well plates, followed by sequential addition of calcium solution (35 µL) and phosphate solution (25 µL) at pH 7.4 and 37°C.Pipetting and mixing were performed using a Tecan Freedom Evo 100 liquid handling robot.The plate was sealed and placed in a temperature-controlled nephelometer (NEPHELOstar PLUS) at 37.0°C for kinetic measurements every 3 min over 10 h.T50 analysis was done using Calciscon's proprietary software.The measurable T50 range was 40-500 min, with intraassay CV of 2.2% and inter-assay CV of 3.4%.The reference range for healthy individuals was determined as 270-470 min based on sera from 253 Swiss blood donors.

Endpoints
The primary endpoint was cardiovascular mortality, whereas the secondary endpoint focused on mortality from all causes.The hospital's outpatient department employs a perpetual monitoring system linked with the municipal death registration to maintain current data on patient status (alive or deceased), with follow-up extending until April 2023.Causes of death were classified in accordance with the 10th revision of the International Classification of Diseases.Specific causes of death analyzed in this study included ischemic heart diseases (codes I20-I25), cardiomyopathy (I42), sudden cardiac death (I46), heart failure (I50), cerebrovascular diseases (I60-I69), and atherosclerosis (I70).

Statistical analyses
Statistical analyses were performed with R version 3.4.2(Vienna, Austria), designating a twotailed p-value under 0.05 as statistically significant.The normality of distributions was verified through histograms and probability plots.Variables adhering to normal distribution were expressed as mean ± standard deviation, whereas those with skewed distributions were denoted as median [interquartile range].Counts and percentages were used to describe categorical variables.Comparative analysis of baseline characteristics across serum T 50 tertiles involved the use of oneway ANOVA, Kruskal-Wallis test, or chi-square tests, contingent on the nature of the data.Levene's test was employed to assess the homogeneity of variances.For Cox regression analyses involving non-normally distributed data, a logtransformation was utilized.We addressed missing data (less than 10%) with multiple imputation techniques based on predictive mean matching with regression models.
The determinants of serum T50 were assessed using multivariable linear regression analyses.Univariable regression analyses were performed for all baseline variables, and those with a p-value less than 0.10 were included in the multivariable linear regression analysis.
To examine the relationship between serum T 50 levels and both cardiovascular and all-cause mortality, multivariable Cox regression models were utilized.Covariates were selected for the multivariable models if they were significantly associated with serum T 50 in multivariable linear regression analysis and with all-cause mortality in the univariable Cox regression analysis (p-value less than 0.20) or considered clinically relevant according to previous studies, such as smoking, alcohol use, BMI, and systolic blood pressure.A visual representation of the association between T 50 and both cardiovascular and all-cause mortality, adjusted for all covariates used in the Cox regression model, was depicted in a figure .The predictive value of T50 was assessed using the Harrell C index, reflecting the model's discriminative value, analogous to the area under the curve.Higher values of the Harrell C index (scale 0-1) indicate better discrimination.Although Harrell's C index measures the concordance between predicted risks and actual outcomes, providing an overall estimate of model performance, it does not inherently test for the statistical significance of individual predictors.To evaluate the statistical contribution of T 50 to the model, we conducted a likelihood ratio (LR) test by comparing nested Cox proportional hazards models with and without the T 50 variable.The LR test offered a formal hypothesis test for the significance of the inclusion of T 50 , presenting a chi-square statistic and corresponding p-value to assess the improvement in model fit attributed to T 50 .
To explore potential nonlinear relationships, we employed restricted cubic spline transformations with three knots in a Cox model (25th, 50th, and 75th percentile) and compared them with linear splines to indicate whether incorporating nonlinear relationships contributes to a better model fit.
We also investigated effect modification by age, gender, and smoking for the association between T50 and cardiovascular and all-cause mortality.Effect modification was addressed by incorporating multiplicative interaction terms in instances involving continuous data, subsequently conducting stratified Cox regression analysis predicated on median values.
We performed sensitivity analyses by repeating identical Cox regression models in individuals with eGFR ≥45 mL/min/1.73m 2 and eGFR ≥60 mL/min/1.73m 2 at baseline and individuals with phosphate levels below ≤1.00 mmol/L.We conducted these sensitivity analyses to determine if impaired renal function, as a marker of disrupted mineral metabolism, or high phosphate levels, indicating increased phosphate availability, strengthen the relationship between T 50 and mortality.Also, to ensure that the association was not driven by extreme outliers in T 50 , we performed a sensitivity analysis with exclusion of the upper and lower 5% serum T 50 measurements.

Baseline characteristics
Baseline characteristics are presented in Table 1.The mean age was 64.2 ± 9.8 years, 61% was male, average HbA1c levels were 58.5 ± 12.1 mmol/mol, and 65% of patients used insulin, of whom 16% as monotherapy.The average serum T50 time was 323 ± 63 min.Participants in the lowest tertile of serum T50 were older, had lower systolic blood pressure, hemoglobin, magnesium, and venous bicarbonate levels, and had higher plasma phosphate levels at baseline.

Determinants of serum T50
Upon univariable and multivariable linear regression analysis, age, alcohol use, high-sensitive C-reactive protein (hsCRP), triglycerides, plasma phosphate, plasma sodium, venous bicarbonate, plasma magnesium, and alanine aminotransferase (ALAT) were significant determinants of serum T50 (Table 2).Plasma phosphate was the strongest determinant of serum T50.The total variance explained by the multivariable model was 40%.

Cardiovascular and all-cause mortality
During a median follow-up time of 7.5 [5.4-10.7]years, 164 individuals (26%) died, of whom 48 due to cardiovascular disease.The number of individuals who died due to cardiovascular disease gradually decreased over tertiles of serum T50: 27 individuals (13%) in the first tertile, 12 (6%) in the second tertile, and 9 (4%) in the third tertile of serum T50.After adjustment for potential confounders, each 60 min decrease of serum T50 was associated with a higher cardiovascular mortality risk (fully adjusted model: HR 1.32, 95% CI 1.08-1.50,and p = 0.01) (Fig. 1a, Table 3).Furthermore, each 60 min decrease of serum T50 time was associated with a higher all-cause mortality risk (fully adjusted model: HR 1.15, 95% CI 1.00-1.38,and p = 0.045) (Fig. 1b, Table 3).Results of the univariable Cox regression analyses for all predictors of cardiovascular and all-cause mortality are presented in Table S1.Addition of T50 improved the prediction of cardiovascular (Harrell's C 0.861 vs. 0.863, p < 0.01) and all-cause mortality (Harrell's C 0.784 vs. 0.787 p = 0.04) compared to the final model including traditional cardiovascular risk   factors (Table S2).Serum T50 time was not associated with non-CV mortality (Table S3).We found significant effect modification for the association between serum T50 and cardiovascular mortality (p < 0.001) by baseline NT-proBNP (p-interaction <0.001).Upon stratified analysis, the association was present in individuals with NT-proBNP levels <50 pmol/L, whereas it was absent in those with levels of NT-proBNP ≥50 pmol/L (Table S4).No effect modification by age, sex, and smoking was found for the primary or secondary endpoint (all p-interaction ≥0.10).Employing restricted cubic spline transformations with 3 knots at the 25th, 50th, and 75th percentiles in the Cox models did not enhance the model's predictive performance when compared with linear splines.

Sensitivity analyses
In sensitivity analyses, we investigated whether impaired kidney function, higher plasma phosphate levels, or outliers in serum T50 influenced the associations between serum T50 and cardiovascular mortality.Therefore, we repeated the main analyses in subgroups of individuals with an eGFR ≥45 mL/min/1.73m 2 (N = 523) or eGFR ≥60 mL/min/1.73m 2 (N = 455) at baseline, individuals with phosphate levels ≤1.00 mmol/L at baseline (N = 309) and individuals in the intermediate 90% of serum T50 levels, respectively.These sensitivity analyses yielded results similar to the primary results (Tables S5-S7).

Discussion
The main finding of this prospective cohort study is that individuals with T2D with lower serum T50 levels are at increased risk of cardiovascular mortality.Moreover, serum T50 improved risk prediction compared to a model with traditional risk factors.Furthermore, lower serum T50 was independently associated with an increased risk of allcause mortality.
Mineralization of soft tissue is normally prevented by endogenous inhibitors of calcification, such as the hepatokine fetuin A, predominantly synthesized in the liver, which prevents direct calcium and phosphate deposition as crystalline hydroxyapatite (Ca10(PO4)6(OH)2).Instead, spherical nanoparticles that contain proteins and noncrystallized calcium and phosphate (primary CPPs) are spontaneously formed from CPMs.These primary CPPs then undergo a coordinated maturation to crystalline secondary CPPs, which can exacerbate vascular calcification and contribute to pre-atherosclerotic changes in the vascular wall [5].Thus, the formation of CPPs prevents immediate precipitation of calcium and phosphate in soft tissue, and the transition time from primary to secondary CPPs might be indicative of calcification propensity [6].This is further supported by the observation that amorphous CPP1 has minor pro-inflammatory effects, whereas CPP2 induces oxidative stress and inflammation in macrophages and oxidative stress, inflammation, and calcification in vascular smooth muscle cells (VSMCs) [16][17][18].In contrast to secondary CPPs, CPMs do not trigger cell death, smooth muscle cell calcification, endothelial dysfunction, or inflammation.Instead, CPMs function as inducers of fibroblast growth factor-23 expression in osteoblasts, which causes an increase in phosphaturia and a decrease in calcium uptake, thereby suppressing further CPP formation [5].Assessing the T50 time could be a relevant additional tool for early detection of cardiovascular disease, whereas current methods, primarily based on computed tomography, are unable to quantify the likelihood of future cardiovascular disease development.Moreover, the T50 time could serve as an intermediate endpoint to guide therapeutic strategies aiming to halt the progression of cardiovascular disease [19,20].
The current study extends previous findings in patients with CKD, hemodialysis, kidney transplantation, and the general population [8][9][10] to individuals with T2D.The study by Eelderink et al. in a population-based cohort indicated effect modification by baseline T2D status, setting the stage for the current analysis in a T2D cohort.
Plasma phosphate was the strongest determinant of serum T50 in our study, with higher plasma phosphate levels linked with lower T 50 levels.This finding builds upon previous research on the role of phosphate as a major promoter of cardiovascular disease [21].Additionally, age, use of alcohol, hsCRP, venous bicarbonate, plasma magnesium, sodium, and ALAT were identified as determinants, consistent with previous studies that investigated determinants of serum T50 [10][11][12].Oral magnesium supplementation improved serum T50 individuals with CKD, whereas oral bicarbonate supplementation had no effect on T50 [22][23][24].Dietary interventions such as limiting alcohol and phosphorus intake, as well as interventions involving phosphate-binders or magnesium supplements, may have potential to improve T50 in individuals with T2D.
An increasing number of publications suggest that frequent episodes of hyperglycemia accelerate plasma phosphate-induced osteochondric differentiation of VSMCs [25][26][27].Therefore, individuals with T2D are believed to be particularly susceptible to the onset of phosphate-induced vascular calcifications in comparison to individuals without diabetes, despite having phosphate levels within the normal range.In the current study, we observed no association between glycemic control and T50 time, in contrast with a previous study [12].This may be explained by differences among the study populations, as participants in our study were in secondary diabetes care and exhibited worse glycemic control than the prior study (HbA1c 58.5 ± 12.1 mmol/mol vs. 49 (44-54) mmol/mol, respectively).It could be that the impact of glycemic control on vascular calcification reaches a plateau at a certain level.Beyond this level, other factors may become more significant determinants of vascular calcification or may exceed the impact of glycemic control on vascular calcification.Of note, HbA1c might not be the best parameter reflecting glycemic control, and param-eters such as time in range might be better [28].
In the present study, no data on time in the range were available.
Remarkably, we observed that the association between serum T50 and cardiovascular mortality was primarily evident in individuals with lower, rather than higher, baseline NT-proBNP levels.This suggests that the role of serum T 50 as a potential driver of cardiovascular risk may be overshadowed by other factors in individuals with elevated NT-proBNP levels.It is intriguing to consider that serum T 50 may contribute to endothelial dysfunction, as previously described [29], potentially leading to vascular stiffness and increased NT-proBNP levels.The interaction between CPPs and endothelial cells could play a pivotal role here, impacting endothelial nitric oxide metabolism and promoting oxidative stress, as the authors have suggested.This influence of T 50 might be less pronounced in individuals who already have higher NT-proBNP levels at baseline, yet these individuals may remain susceptible to such elevations in the future.Given that ischemic heart failure is acknowledged as one of the primary causes of heart failure, exploring the relationship between serum T 50 and the risk of developing new heart failure in T2D would be of significant interest.
Strengths of the current study include the complete data on serum T50 levels and follow-up with (CV) mortality endpoints in a well-characterized cohort.Due to a relatively long follow-up time, the event rate for all-cause mortality was high (26%).However, limitations should be mentioned.First, serum T50 was measured at a single timepoint in a single cohort that consisted almost exclusively of Caucasians, limiting extrapolation to other populations.Furthermore, the observational design precludes conclusions on causality, and residual confounding cannot be entirely excluded despite extensive adjustment and sensitivity analyses.Additionally, we did not have access to data on fetuin A levels, a significant inhibitor of CPP2 formation and an identified determinant of serum T 50 .
To conclude, we found that a lower serum T50 is associated with a higher risk of CV and allcause mortality in individuals with T2D.Although prospective intervention studies are necessary to explore the potential of the T50 test to guide therapeutic strategies, it provides additional information for cardiovascular risk stratification beyond traditional risk factors in patients with T2D and underlines the prognostic importance of T 50 in this population.

Fig. 1
Fig. 1 Serum T50 and cardiovascular mortality (a) and all-cause mortality (b).Serum T50 increases cardiovascular (a) and all-cause (b) mortality in individuals with type 2 diabetes.The hazard ratio is shown as a solid line, and the associated pointwise 95% CIs are represented by the shaded area.Figs. are adjusted for age, gender, history of micro and macrovascular disease, smoking, use of alcohol, body mass index (BMI), plasma HDL cholesterol, triglycerides, type 2 diabetes (T2D) years, plasma HbA1c, estimated glomerular filtration rate (eGFR), systolic blood pressure, high sensitive C-reactive protein (hsCRP), and alanine aminotransferase (ALAT).

Table 1 .
Baseline characteristics of the total cohort and per tertile of serum T50.

Table 2 .
Determinants of serum T50 in univariable and multivariable regression model.

Table 2 .
(Continued)Univariable and multivariable linear regression model showing determinants of serum T50.All variables were standardized, except for factors (yes/no).Multivariable models were fitted with all variables that were significant in univariable regression analysis (p < 0.05).Explained variance R 2 = 0.40.

Table 3 .
Serum T50 and cardiovascular and all-cause mortality risk.Hazard ratios and 95% confidence intervals were derived from Cox proportional hazards regression models.N = 621.