Increased Subclinical Coronary Artery Pathology in Type 2 Diabetes With Albuminuria

Diabetes affects the kidneys, and presence of albumin in the urine (albuminuria) reflects widespread vascular damage (1) and is a risk factor for cardiovascular disease (CVD) and mortality (2–4). Still, the pathophysiological association between albuminuria and CVD remains incompletely understood.

Recent advances in noninvasive imaging enable functional assessment of coronary artery pathology and present an opportunity to explore the association between albuminuria and CVD.

The radiopharmaceutical ¹⁸F-sodium fluoride (18NaF) is a positron emission tomography (PET) imaging tracer enabling assessment of active microcalcification processes, detecting the earliest stages of atherosclerosis (5). The PET imaging tracer, ⁶⁴Cu-DOTATATE, is used to quantify plaque inflammation, and quantification of cardiac adipose tissue (CAT) can provide an estimate of epicardial inflammation. The myocardial flow reserve (MFR) quantified from ⁸²Rb (⁸²Rb) cardiac PET/computed tomography (CT) provides risk information for CVD and mortality (6).

In this cross-sectional study, we evaluated the presence of subclinical coronary artery pathology in people with type 2 diabetes, stratified by albuminuria. Using multimodal imaging, we assessed the presence of microcalcification, plaque and epicardial inflammation, myocardial microvascular dysfunction, and calcifications in the epicardial arteries (coronary artery calcium score [CACS]). We hypothesize that albuminuria is related to a more severe phenotype of myocardial microvascular and macrovascular pathology.

This was a prespecified explorative substudy from the DiaHEART study, a cross-sectional study with long-term follow-up, aiming to identify subpopulations at high risk of developing CVD, with the MFR as the primary end point. We included 90 people with type 2 diabetes between 2021 and 2023, free of overt CVD, and aged between 50 and 85 years.

Exclusions criteria were history of CVD, second- or third-degree atrioventricular block, sick sinus syndrome, severe kidney impairment (estimated glomerular filtration rate [eGFR] < 30 mL/min/1.73 m²), chronic obstructive pulmonary disease, severe reaction to iodinated contrast, or pregnancy and lactating.

Participants were stratified into two groups based on previous urine albumin-to-creatine ratio (UACR) in two consecutive measurements (i.e., normoalbuminuria [UACR] <30 mg/g), and any albuminuria (UACR ≥30 mg/g). The albuminuria groups were prespecified to include three groups (UACR <30 mg/g [n = 30], UACR >30 >300 mg/g [n = 30], and UACR >300 mg/g [n = 30]), but, because of recruitment issues, we combined microalbuminuria and macroalbuminuria into an “any albuminuria” group (UACR ≥30 mg/g [n = 60]).

All participants were examined with a 1) hybrid coronary CT angiography (cCTA) in combination with an 18NaF PET/CT scan, 2) ⁶⁴Cu-DOTATATE PET/CT scan, and 3) cardiac ⁸²Rb PET/CT myocardial perfusion scan—on separate days. All scans were performed at Rigshospitalet, Copenhagen, Denmark. Figure 1 shows the timeline of the visits.

The study was approved by the Danish National Committee on Health Research Ethics (H-19063311), Copenhagen, Denmark, and performed in compliance with the Declaration of Helsinki. All participants gave informed written consent.

Clinical measurements were collected prior to the PET/CT scans. The UACR was measured in three consecutive morning urine samples by an enzyme immunoassay and calculated as the geometric mean. The 24-h blood pressure was recorded using a cuff device (TM2430; Takeda, Tokyo, Japan).

The 18NaF PET was performed using a Siemens Biograph mCT 128 (Siemens, Munich, Germany), using an injection-to-scan delay of 60 min. Subsequently, an electrocardiogram (ECG)-gated contrast-enhanced cCTA was obtained using the same scanner. The cCTA images were used to extract anatomic data of the coronary arteries using the software Autoplaque (version 2.5; Cedars-Sinai Medical Center), to ensure accurate coregistration of the PET and the cCTA (7). The coronary microcalcification activity (CMA) represents the overall microcalcification in the coronary arteries depending on the intensity and the volume of the 18NaF PET imaging tracer activity (8). The PET images were corrected for coronary motion, using FusionQuant (9), and injection-to-scan delay in minutes (10). Total CMA for each person was calculated as the sum of CMA in the three main coronary arteries.

The ⁶⁴Cu-DOTATATE PET/CT was aimed at injection-to-scan delays of 60 min, with Siemens Biograph mCT or Biograph Vision systems (Siemens Healthineers). The PET images were corrected for ECG motion using FusionQuant (8). Quantification of the coronary inflammation activity (CIA) above threshold was performed in FusionQuant (Cedars-Sinai Medical Center) in a procedure similar to the 18NaF PET analysis. The total CIA was calculated as the sum of inflammation activity above threshold in the three main coronary arteries.

The cardiac software, Syngo.via Frontier—Cardiac risk assessment (Siemens, Healthcare Sector) was used to calculate the CAT. The software determines the amount of adipose tissue surrounding the heart by drawing the boundaries of the myocardium and the adipose tissue based on the density of fat (−150 to −50 Hounsfield units).

After administering of 1,100 MBq ⁸²Rb, the cardiac ⁸²Rb PET/CT myocardial perfusion scans were performed at rest and during pharmacologically induced stress (infusion of 140 μg/kg/min adenosine for 6 min) using the same PET/CT scanner as the 18NaF examinations. The Syngo.via software (syngo.mbf vb20a) automatically measured the myocardial blood flow (11). The MFR was calculated as the ratio between myocardial blood flow at rest and under stress. The left ventricular ejection fraction (LVEF) reserve was calculated as the difference between the LVEF under stress and at rest using ECG-gated reconstructions of the ⁸²Rb PET/CT.

A semiautomated software (Syngovia 4.0) was used to calculate total CACS as the total coronary artery calcium content in the three main coronary arteries.

This prespecified substudy is hypothesis generating and explorative, and limited evidence for sample size calculation for these novel measures is not yet available. Continuous nonnormal distributed variables are presented as median and interquartile range (IQR) and log2 transformed in all analyses. A value of 1 was added to the CACS and the CMA before the transformation, since the unequal distribution of these variables included values of zero. Continuous normal distributed variables are presented as mean and SD. Categorical variables are summarized as numbers and percentages.

Unpaired sample t test and the χ² test or Fisher exact test were applied, as appropriate, to assess whether clinical characteristics differed between participants with normoalbuminuria and albuminuria. Analysis of covariance was applied to compare differences in CMA, CIA, CAT, high-sensitivity C reactive protein (hs-CRP), MFR, and CACS among the participants with and without albuminuria after adjustment for sex, age, HbA_1c, 24-h systolic blood pressure, LDL cholesterol, eGFR, and smoking.

Unadjusted regression models were applied to determine the association between current UACR and eGFR levels as continuous variables. CMA, CIA, CAT, MFR, and CACS in the total population, and standardized regression coefficients (β), were reported. Next, we applied adjusted linear regression models, including sex, age, HbA_1c, 24-h systolic blood pressure, LDL cholesterol, eGFR, and smoking. Analyses, including MFR, were further adjusted for resting heart rate during the ⁸²Rb PET/CT scan.

Sensitivity analysis was as follows: 1) A multivariate stepwise analysis was performed to understand which potential confounders could explain the univariable associations. 2) To account for the impact of current albuminuria unpaired samples, t test was applied to assess whether CMA, CIA, CAT, hs-CRP, MFR, and CACS differed between participants with current UACR <30 mg/g and ≥30 mg/g. 3) To explore the impact of increased inflammation, focus was placed on the participants with hs-CRP >2 mg/L. Unadjusted regression models were applied to investigate the association between current hs-CRP and UACR, eGFR, CMA, CIA, CAT, MFR, and CACS.

Two-sided P values <0.05 were considered statistically significant. Statistical analyses were performed using R.Studio software (version 4.2.0).

The data sets for the current study are available upon reasonable request.

Figure 2 shows the development of the cohort. Clinical characteristics of the participants with or without albuminuria were similar between groups and are presented in Table 1.

The group with albuminuria had a higher median CMA of 0.75 [IQR 0.30–1.85] compared with the group with normoalbuminuria 0.43 [0.24–0.87] (P = 0.04). However, the difference did not remain after adjustment (P = 0.09). The multivariate step analysis indicated that blood pressure was a potential confounder explaining the univariable association.

Mean MFR (SD) was lower in the group with albuminuria (2.5 [SD 0.7]) compared with the group with normoalbuminuria (2.9 [SD 0.7]), with a significant group difference (P = 0.02), but not after adjustment (P = 0.07), while age, eGFR, and blood pressure were identified as potential confounders explaining the univariable association. The LVEF reserve was lower in the group with albuminuria (5% [SD 5]) compared with the group with normoalbuminuria (7% [SD 3]) (P = 0.02), and also after adjustment (P = 0.01). The CIA, CAT, and CACS were similar between groups (P ≥ 0.50). Cardiac measurements in the two groups are presented in Fig. 3.

In this study, we evaluated subclinical myocardial microvascular and macrovascular pathology by multimodal imaging in people with type 2 diabetes without CVD, stratified by albuminuria. The main findings were 1) people with any albuminuria had elevated CMA and lower myocardial microvascular function, and 2) coronary inflammation was similar in people with and without albuminuria.

We demonstrated elevated microcalcification activity in the group with any albuminuria and with higher current levels of albuminuria. Significance was lost after adjustment for risk factors, and only a trend remained. However, half of the participants with any albuminuria were so well treated that they currently had normoalbuminuria, and a reduction in albuminuria is associated with improved cardiovascular outcomes (12). We consider that this has weakened the likelihood of finding the “true” difference, supported by our sensitivity analysis investigating people with current UACR above 30 mg/g, demonstrating more microcalcification in the groups with albuminuria, even after adjustment.

Impaired MFR is an independent risk factor for cardiac mortality among people with diabetes (6). In line with previous findings (13), we demonstrated that MFR was lower in the group with any albuminuria compared with normoalbuminuria, suggesting a widespread microvascular dysfunction that affects several microvascular beds.

Previous studies have demonstrated that inflammation measured with different biomarkers is higher with higher levels of albuminuria (14). Only a few studies have evaluated the relationship between CAT and albuminuria (15,16), and findings are conflicting. Measurement of CIA by ⁶⁴Cu-DOTATATE PET/CT is suggested as an early marker of inflammation, but the relationship between albuminuria and CIA has not been investigated previously.

We did not observe any differences in the level of inflammation markers (CAT, CIA, and hs-CRP) between albuminuria groups. The participants had generally well-treated levels of cardiovascular risk factors (Table 1) and had an overall low level of inflammation (median hs-CRP 1.00 [IQR 0.49–1.94] mg/L). We speculate that the low levels hamper the likelihood to detect any association. In both albuminuria groups, there were a high proportion of participants treated with statins, glucagon-like peptide-1 receptor agonist (GLP-1 RA), and aspirin, treatments which also have anti-inflammatory effects (17–19), which might reduce the likelihood of detecting significant differences in the inflammation markers between groups.

Our study has some limitations. First, there is a risk of selection bias, as participants with severe kidney impairment (eGFR <30 mL/min/1.73 m²) were excluded because of the contrast-enhanced CT, which potentially dismisses the people with highest disease burden, and might weaken our ability to show true associations. Second, the participants in general, including those with albuminuria, were well treated, which reduces the likelihood of detecting significant differences between the albuminuria groups, Third, using historical UACR levels for classification could result in misclassification of the participants, as treatment-induced remission of albuminuria may be associated with a lower risk profile compared with sustained albuminuria (82% treated with renin-angiotensin-aldosterone system inhibitors and 87% on antihypertensive treatment), and this might have biased our results. Fourth, the study was explorative; thus we did not adjust for multiple testing, and no primary end point was defined. The limited sample size could have introduced type 2 errors. Fifth, it would have strengthened our study if information on pericoronary fat, which is a powerful determinant of CV events, was available. Sixth, measurement of DOTATATE uptake by PET/CT has shown promising results as a marker for assessing vascular inflammation (20). However, further research and broader validation is essential to establish the role of CIA.

In people with type 2 diabetes, without CVD, the presence of albuminuria was associated with subclinical myocardial microvascular and macrovascular pathology when compared with normoalbuminuria. Higher levels of microcalcifications were related to higher levels of albuminuria.

Acknowledgments. The authors thank our participants and acknowledge the work of the laboratory technicians at Steno Diabetes Center Copenhagen and study nurse J. Maibom.

Funding. The study was funded by grants from the Novo Nordisk Foundation (grant number NNF19OC0054674).

Duality of Interest. P.R. has received speaking fees and/or consultancy to Steno Diabetes Center Copenhagen from Eli Lilly, Novo Nordisk, Sanofi Aventis, Vifor, Boehringer Ingelheim, Astellas, Gilead, Bayer, AstraZeneca, Mundipharma, and MSD. T.W.H. and P.R. have received research grants from Novo Nordisk, and P.R. has received grants from AstraZeneca and has shares in Novo Nordisk. A.K. and R.S.R. have received consultancy fees from Novo Nordisk. R.S.R. has shares in Novo Nordisk. E.H.Z. is a full-time employee at Novo Nordisk. P.S. received grants from Siemens medical systems and consulting honoraria from Synektik, SA. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. I.K.B.R., R.S.R., P.H., E.H.Z., L.H., P.R., A.K., and T.W.H. contributed to the study design and data interpretation. I.K.B.R., R.S.R., K.H.-T., and M.L.L. acquired data. I.K.B.R., A.-C.S.-M., and V.S.W. recruited participants. I.K.B.R. performed statistical analysis. I.K.B.R. drafted the manuscript. The final manuscript was critically read, revised, and approved by all authors. I.K.B.R. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.