Arterial Stiffness Predicts Mortality in Individuals With Type 1 Diabetes

Cardiovascular disease (CVD) is the leading cause of the excess morbidity and mortality observed in individuals with type 1 diabetes, and the standardized mortality ratio is known to increase by each stage of diabetic nephropathy (1,2). This predisposition is only partly attributable to traditional risk factors, and, in fact, cardiovascular risk scores developed for the general population and people with type 2 diabetes are poorly applicable in those with type 1 diabetes (3). Thus, a unique risk factor profile is likely to prevail in these individuals and merits further characterization.

Arterial stiffness is a well-known predictor of mortality in the general population and in selected groups, including those with type 2 diabetes, yet no longitudinal studies have been carried out in individuals with type 1 diabetes (4,5). Interestingly, arterial stiffening seems to occur early in individuals with type 1 diabetes, as their pulse pressure (PP), a crude estimate of arterial stiffness, increases up to 15–20 years earlier than in healthy control subjects (6). This phenomenon of early arterial aging made us hypothesize that arterial stiffness might be an important mediating factor leading to premature death in type 1 diabetes. Because microangiopathy is a major manifestation of complicated type 1 diabetes, we further hypothesized that early signs of arterial stiffening could be detected by the augmentation index (AIx), which, as a measure of arterial pulse wave reflections, is particularly affected by stiffness in the small resistance arteries (7). We previously showed that AIx correlates with microvascular and macrovascular complications in type 1 diabetes in a cross-sectional setting (8). The aim of this study was therefore to explore whether AIx predicts all-cause as well as cardiovascular and/or diabetes-related mortality in type 1 diabetes.

This prospective observational follow-up study is part of the ongoing nationwide Finnish Diabetic Nephropathy (FinnDiane) Study, in which >5,400 individuals with type 1 diabetes have been characterized since 1997. The study protocol has been approved by the local ethics committees, and written informed consent was obtained from each participant. The FinnDiane protocol has been previously described in detail (9). Briefly, baseline data on cardiovascular risk factors and diabetic complications are collected from standardized questionnaires and medical files, as well as through clinical examination and biochemical measurements. Since 2001, noninvasive assessment of arterial stiffness and central hemodynamics through pulse wave analysis has been included in the baseline examination of those individuals studied at the FinnDiane center in Helsinki, Finland.

In this substudy, individuals with available baseline data on arterial stiffness by the year 2015 were included. Further inclusion criteria were age >18 years, type 1 diabetes diagnosed by 40 years of age, and insulin treatment initiated within 1 year of the diagnosis. The baseline population comprised 906 individuals (416 men) with a mean age of 43.2 years (SD 12.2), including 134 individuals with chronic kidney disease (CKD), defined as an estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m², ongoing hemodialysis, or having received a renal transplant, as well as 98 individuals with a previous CVD event, defined as myocardial infarction, coronary revascularization, stroke, lower-extremity revascularization, or nontraumatic amputation.

Applanation tonometry (SphygmoCor; AtCor Medical, Sydney, New South Wales, Australia) is a noninvasive reproducible method to estimate central (aortic) blood pressure variables and arterial stiffness through pulse wave analysis (10,11). A high-fidelity micromanometer (SPC-301; Millar Instruments, Houston, TX) is used to record peripheral pressure waveforms from the radial artery of the right arm, and three readings with a pattern of at least 20 valid waveforms are selected for the analysis. The software generates a central pressure waveform using a validated transfer function. This enables determination of central systolic blood pressure (CSBP) and diastolic blood pressure (CDBP), central PP (CPP) = CSBP − CDBP, central mean arterial pressure (CMAP) = 1/3 × CSBP + 2/3 × CDBP, central end-systolic pressure (CESP), ejection duration, and subendocardial viability ratio (SEVR), which indirectly estimates myocardial perfusion. To assess stiffness in the small resistance arteries, AIx is calculated as a quotient of two measures: the difference of the second and the first systolic peak of the pressure waveform (corrected for heart rate) and CPP.

Mortality data were obtained from the cause-of-death statistics and the archive of death certificates maintained by Statistics Finland. Cardiovascular deaths of individuals with diabetes are not uncommonly classified as diabetes-related deaths in Finland, especially in cases in which no autopsy has been performed. Therefore, we combined cardiovascular and/or diabetes-related causes of death into one secondary end point in the survival analysis, alongside all-cause and cardiovascular mortality.

Univariable analyses of established cardiovascular risk factors and hemodynamic variables from pulse wave analysis were run to detect differences between those who died during follow-up and those who survived. The χ² tests were used for dichotomous variables and t tests or Mann–Whitney U tests for continuous variables. Data are presented as mean ± SD (normally distributed) or median with interquartile range (nonnormally distributed) for continuous variables and as percentages for dichotomous variables.

Longitudinal analysis was performed using Kaplan-Meier survival curves with log-rank tests. For multivariable analyses, continuous covariates were standardized by dividing the difference of each value and the covariate mean by the SD of that covariate. The best-fit regression model for each end point was selected using the Akaike information criterion and further adjusted for sex and the variable of interest from pulse wave analysis. Independent associations with mortality were determined by Cox regression analysis and are presented as standardized hazard ratios (sHRs) with 95% CI. P values <0.05 were considered statistically significant.

After a median follow-up of 8.2 years (interquartile range 5.7–9.7), 67 (7.4%) individuals had died (Table 1). These individuals were older, had a longer diabetes duration, and had higher office SBP, PP, HbA_1c, and triglycerides, as well as lower BMI and eGFR at baseline, compared with those who survived. Similarly, those who died had more often been prescribed antihypertensive and lipid-lowering medication and had more often had a previous CVD event at baseline. In the pulse wave analysis, AIx (28% [21–33] vs. 19% [9–27]), CSBP (138 [121–150] vs. 119 [109–131] mmHg), CMAP (96 [91–105] vs. 91 [85–98] mmHg), CPP (61 [44–80] vs. 41 [34–52] mmHg), and CESP (115 [106–125] vs. 105 [96–116] mmHg) were higher, whereas SEVR (116% [102–138] vs. 142% [123–164]) was lower at baseline in those who died during follow-up.

With division into tertiles based on AIx values, those in the highest tertile showed the highest rate of all-cause mortality (Fig. 1). Furthermore, AIx was associated with all-cause mortality in an unadjusted Cox regression model (Table 2). After adjustments for sex, age, BMI, and HbA_1c, AIx remained in the model with an sHR of 2.14 (95% CI 1.42–3.23; P < 0.001). In the final model, after correction for two additional strong predictors of mortality, eGFR and previous CVD event, AIx was still associated with all-cause mortality (sHR 1.71 [1.10–2.65]; P = 0.017). Similarly, CSBP (sHR 1.29 [1.03–1.62]; P = 0.028), CMAP (sHR 1.30 [1.05–1.62]; P = 0.019), and SEVR (sHR 0.67 [0.47–0.94]; P = 0.022) were independently associated with all-cause mortality in the final model. In a subanalysis including only individuals with CKD, AIx showed an even stronger association with all-cause mortality (sHR of 3.39 [1.66–6.91]; P = 0.001), when adjusted for sex, BMI, and previous CVD event.

No adjustments for peripheral blood pressure variables were made in the regression model to avoid multicollinearity. AIx correlated with SBP, DBP, and PP with Pearson correlation coefficients of 0.432, 0.284, and 0.344, respectively. For comparison with AIx, these variables were added separately to the final Cox regression model (sex, age, BMI, HbA_1c, eGFR, and previous CVD event). Independent, yet weaker association with all-cause mortality was seen for SBP (sHR 1.34 [1.07–1.67]; P = 0.011), DBP (sHR 1.28 [1.01–1.63]; P = 0.041), and PP (sHR 1.28 [1.00–1.64]; P = 0.046).

Of the deaths that occurred during the follow-up, 53 were classified as cardiovascular and/or diabetes-related. In an adjusted Cox regression model (Table 3), AIx was independently associated with the composite end point of cardiovascular and/or diabetes-related mortality (sHR 2.30 [1.38–3.83]; P = 0.001). When cardiovascular mortality (N = 36) was analyzed separately, AIx was associated with both cardiovascular mortality (sHR 2.36 [1.22–4.53]; P = 0.010) and noncardiovascular mortality (sHR of 2.11 [1.25–3.56]; P = 0.005) in adjusted Cox regression analyses. For the other than cardiovascular and/or diabetes-related causes of death (N = 14), however, AIx was not a significant risk factor in a Cox regression model adjusted for sex (sHR 1.56 [0.82–2.99]; P = 0.179).

In this study population of 906 individuals with type 1 diabetes followed up for a median of 8.2 years, AIx was an independent risk factor for all-cause mortality even after adjustments for well-known risk factors, including renal function. The same observation was made regarding cardiovascular and/or diabetes-related mortality as a composite secondary end point, as well as in a subanalysis of only individuals with CKD. Other measures of central hemodynamics that showed an independent association with all-cause mortality included CSBP, CMAP, and SEVR. When comparing sHRs, AIx outperformed both the central and office blood pressure variables in predicting all-cause mortality, when separately included in the final multivariable model.

While arterial stiffness indices have been increasingly studied in type 2 diabetes, this is the first study to investigate the association between AIx, a surrogate measure of stiffness in the small resistance arteries, and mortality in individuals with type 1 diabetes. Earlier studies in type 1 diabetes have evaluated how the PP, a crude estimate of stiffness in the large arteries, predicts CVD and mortality (12,13). Prospective studies using the gold-standard measure of arterial stiffness, pulse wave velocity (PWV), are so far limited to type 2 diabetes.

Due to different pathophysiological characteristics and the accumulation of CVD at a younger age in type 1 diabetes, extrapolating findings from studies of populations without diabetes or even individuals with type 2 diabetes should be made with caution (14). It is of note that in type 1 diabetes, the macrovascular complications may in part have a microvascular origin. In fact, small vessel disease is the major underlying cause of ischemic stroke in individuals with type 1 diabetes and, interestingly, more common than in individuals with type 2 diabetes (15). In the absence of symptomatic CVD, a reduced coronary vascular reactivity has been shown in young individuals with type 1 diabetes, and another study demonstrated differences in the atherosclerotic morphology of the coronary arteries between the two types of diabetes (16,17). Recently, even an autoimmune component has been proposed to play a role in the pathogenesis of CVD in type 1 diabetes (18). Although the exact pathogenic mechanisms remain to be uncovered, current knowledge implies that type 1 diabetes needs to be considered a separate entity when the risks and prevention of cardiovascular complications are studied (14).

As AIx reflects stiffness in the small resistance arteries, our findings may support the hypothesis of small vessel disease contributing to the pathogenesis of macrovascular complications and premature mortality seen in type 1 diabetes. The Diabetes Control and Complications Trial/Epidemiology of Diabetes Interventions and Complications (DCCT/EDIC) study demonstrated the long-standing effects of hyperglycemia on the risk of diabetic complications and CVD in type 1 diabetes, a phenomenon referred to as “metabolic memory” (19). Whether this is partly mediated by small vessel disease and arterial stiffness is an open question to be addressed in future research.

Although noninvasive and applicable for clinical practice, applanation tonometry is operator-dependent and can be considered time-consuming and costly. However, new operator-independent technologies to capture central hemodynamics by pulse volume plethysmography have been developed in recent years and may improve the feasibility of measuring arterial stiffness (20). Indeed, novel clinical risk markers are needed to be able to predict the increased risk of CVD and mortality in type 1 diabetes. Given our findings, AIx could be a useful tool to detect such high risk of cardiovascular complications, enabling intensive cardiovascular risk control at an early stage for these individuals. Nevertheless, clinical implications require further investigation of the added value of AIx in risk prediction models, especially compared with the traditional blood pressure variables. This study did show a higher risk of mortality per SD increment in AIx as compared with that of SBP, DBP, or PP in separate multivariable models.

The prospective study setting in a large cohort with comprehensive phenotypic data constitutes a major strength in our study, whereas its observational design only allows speculations about causality. With increasing age, there are some limitations to the reliability of AIx. Following a nonlinear pattern, AIx steeply increases in the young while reaching a plateau at older age (21). This could partly be explained by the formula itself: concurrent increases in both augmentation pressure and CPP could result in AIx remaining stable or even declining (22). Central PWV increases later in life, whereas AIx may be preferable in younger populations, which is essential when considering the applicability for early detection and prevention of CVD. It is not clear how AIx changes over time in individuals with type 1 diabetes, or whether there should be a transfer function specifically validated in type 1 diabetes. However, based on the early increase in PP in type 1 diabetes, one could assume earlier plateauing of AIx. Our study population was relatively young, which may have contributed to the predictive value of AIx in this study. As PWV measured by applanation tonometry was introduced to the FinnDiane protocol at a later stage, complete data powered for prospective analysis are still on the way.

To summarize, AIx as an estimate of stiffness in the small resistance arteries is independently associated with all-cause mortality, as well as the composite of cardiovascular and/or diabetes-related mortality in type 1 diabetes. These results together with our earlier findings suggest that detection of early vascular aging in individuals with type 1 diabetes could have complementary value in clinical risk assessment when targeting a more aggressive treatment approach for high-risk individuals.

Acknowledgments. The authors thank the FinnDiane research nurses A. Sandelin, J. Tuomikangas, and M. Korolainen, all affiliated with Folkhälsan Insitute of Genetics (Folkhälsan Research Center, Helsinki, Finland), for technical assistance.

Funding. This study was supported by grants from Academy of Finland (316644 and UAK10121MRI); Biomedicum Helsinki Foundation; Dorothea Olivia, Karl Walter and Jarl Walter Perklén’s Foundation; Finnish Medical Foundation; Finska Läkaresällskapet (Medical Society of Finland); Folkhälsan Research Foundation; Liv och Hälsa Society; Novo Nordisk Foundation (NNF OC0013659); Päivikki and Sakari Sohlberg Foundation; Sigrid Juselius Foundation; Svenska Litteratursällskapet i Finland (the Society of Swedish Literature in Finland); Swedish Cultural Foundation in Finland; Wilhelm and Else Stockmann Foundation; and University of Helsinki (Clinical Researcher stint).

Author Contributions. A.T., C.F., P.-H.G., and D.G. conceived and designed the analysis. A.T., C.F., V.H., and D.G. collected the data. A.T., C.F., and D.G. analyzed and interpreted the data. A.T., C.F., P.-H.G., and D.G. wrote the manuscript. A.T., C.F., V.H., P.-H.G., and D.G. revised and edited the manuscript. P.-H.G. is the guarantor of this study and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 32nd Annual Meeting of the European Diabetic Nephropathy Study Group, Paris, France, 24–25 May 2019, and the 55th Annual Meeting of the European Association for the Study of Diabetes, Barcelona, Spain, 16–20 September 2019.