Abstract
Introduction: Kidney transplant recipients (KTRs) have increased risk of cardiovascular disease (CVD) mortality. We investigated vascular biomarkers, angiopoietin-1, and angiopoietin-2 (angpt-1, -2), in CVD development in KTRs. Methods: This ancillary study from the FAVORIT evaluates the associations of baseline plasma angpt-1, -2 levels in CVD development (primary outcome) and graft failure (GF) and death (secondary outcomes) in 2000 deceased donor KTRs. We used Cox regression to analyze the association of biomarker quartiles with outcomes. We adjusted for demographic; CVD- and transplant-related variables; medications; urine albumin-to-creatinine ratio; and randomization status. We calculated areas under the curves (AUCs) to predict CVD or death, and GF or death by incorporating biomarkers alongside clinical variables. Results: Participants’ median age was 52 IQR [45, 59] years: with 37% women and 73% identifying as white. Median time from transplantation was 3.99 IQR [1.58, 7.93] years and to CVD development was 2.54 IQR [1.11–3.80] years. Quartiles of angpt-1 were not associated with outcomes. Whereas higher levels of angpt-2 (quartile 4) were associated with about 2 times the risk of CVD, GF, and death (aHR 1.85 [1.25–2.73], p < 0.01; 2.24 [1.36–3.70)], p < 0.01; 2.30 [1.48–3.58], p < 0.01, respectively) as compared to quartile 1. Adding angiopoietins to preexisting clinical variables improved prediction of CVD or death (AUC improved from 0.70 to 0.72, p = 0.005) and GF or death (AUC improved from 0.68 to 0.70, p = 0.005). Angpt-2 may partially explain the increased risk of future CVD in KTRs. Further research is needed to assess the utility of using angiopoietins in the clinical care of KTRs. Conclusion: Angpt-2 may be a useful prognostic tool for future CVD in KTRs. Combining angiopoietins with clinical markers may tailor follow-up to mitigate CVD risk.
Plain Language Summary
We have identified compelling insights into the role of two proteins called angiopoietin 1 and 2, which are needed, in balance, for healthy blood vessels to form. In the case of kidney transplants from deceased donors, we found that high levels of angiopoietin-2, suggest a higher risk of future heart problems, kidney transplant failure, and even death. As such, our study reaffirms that high levels of angiopoietin-2 in the blood is a marker of blood vessel injury and inflammation. Notably, when we included information about these proteins along with the regular medical details, we were better at predicting important clinical outcomes. Understanding these proteins could help us figure out which KTRs are at a higher risk, so we can take better care of them and potentially improve their outcomes.
Introduction
While cardiovascular disease (CVD) remains the leading cause of death worldwide, it targets kidney transplant recipients (KTRs) 30 times more frequently than the general population [1]. Given its biased targeting of KTRs, there is a need for heightened awareness and treatment in this population. However, recent studies have highlighted the under-treatment of CVD in KTRs [2]. Stratification of KTRs for CVD risk may aid in increasing awareness and developing more targeted follow-up and treatment approaches for high-risk patients. To better phenotype KTRs for risk of CVD, we measured two plasma vascular biomarkers, angiopoietin-1 (angpt-1) and angiopoietin-2 (angpt-2), to assess their role in CVD outcomes.
The procedure of transplanting a deceased donor kidney into a recipient is an extremely vascular process involving the clamping and reattachment of vessels in a timely and intricate manner [3]. Therefore, there is inevitable ischemic injury to the kidneys [4], which subsequently activates pathways of vascular injury and repair, specifically the angiopoietin pathway [5]. A deceased donor graft experiences ischemia prior to reperfusion in the recipient body [6]. This ischemic insult activates certain vascular pathways that dictate the recovery of the graft and collateral organs. Ischemic kidney injury has been shown to be associated with local release of angpt-2 in the renal arteries and veins of both living and deceased donor re-perfused kidneys [5]. This release of angpt-2 can predict poor outcomes several years after the kidney insult [7]. Additionally, vascular injury is prominent in patients with CVD [8‒10], as it activates an inflammatory cascade leading to a pro-atherogenic state, which is involved in the development of coronary artery disease and diastolic dysfunction [11, 12]. We hypothesize that the activation of the angiopoietin pathway after kidney transplantation will influence future cardiovascular health in the recipients.
Angpt-1 functions as a protective protein for blood vessels by inhibiting inflammation, limiting vessel leakage, and preventing endothelial cell death [13]. Angpt-2 competitively inhibits angpt-1, thus acting as an antagonist, leading to increased inflammation and vessel instability [14]. In patients with CKD, angpt-2 is associated with higher rates of heart failure, atherosclerotic disease, acute myocardial infarctions, and left ventricular hypertrophy [15]; on the other hand, angpt-1 is associated with improved cardiovascular health and decreased fasting blood glucose levels and inflammation [7, 16]. We have previously shown that in hospitalized patients with acute kidney injury, having higher concentrations of angpt-1 relative to angpt-2 was substantially protective against future heart failure hospitalizations, and significantly improved survival [7].
The accompanying pathologies in both the kidneys and the heart may stem from not only shared risk factors and co-dependence of the two organs, but also the activation of certain pathways, such as the angiopoietin pathway, that influence both organs [17]. Further understanding of the link between kidney transplantation and CVD is imperative to guide the care of recipients posttransplantation and improve outcomes. Here, we performed an ancillary study using the Folic Acid for Vascular Outcome Reduction in Transplantation Trial (FAVORIT) to understand the role of angpt-1 and angpt-2 in the risk of future CVD and the secondary outcomes of graft failure (GF) and death [18].
Methods
Study Population
Details of the FAVORIT design have been published previously [18]. Briefly, this was a multicenter double-blinded randomized controlled trial conducted at 30 clinical sites across the USA, Canada, and Brazil. Participants were randomized to a multivitamin that includes either high-dose or low-dose folic acid (5 or 0 mg), vitamin B6 (50 or 1.4 mg), and vitamin B12 (1,000 or 2 mg). Participants included KTRs between the ages of 35–75 years with clinically stable kidney function for at least 6 months posttransplantation and elevated homocysteine levels. Participants with chronic illness limiting life expectancy to less than 2 years and those with a recent CVD event or procedure were excluded. Participants were followed for a minimum of 4 years and up to 8 years to assess the development of incident or recurrent CVD.
Outcomes
We assessed potential associations between angpt-1 and angpt-2 with the composite primary outcome of any CVD event, which was defined as fatal or nonfatal myocardial infarction, fatal or nonfatal stroke, resuscitated sudden death, CVD-related death, and any procedure.
We also assessed the secondary outcomes of GF and death. GF was defined as re-transplantation or return to dialysis. Death was defined as all-cause mortality.
Plasma Samples and Biomarker Measurement
We requested 2000 baseline plasma samples (at time of enrollment) of deceased donor KTRs from the FAVORIT trial and measured angpt-1 and angpt-2 concentrations. Samples were centrifuged at 2,000 g for 10 min at 4°C, separated into 1-mL aliquots, and immediately stored at −80°C. Biomarkers were measured using the Meso Scale Discovery platform (Meso Scale Diagnostics, Gaithersburg, MD, USA), which uses electrochemiluminescence detection combined with patterned arrays. The mean inter-assay coefficient of variationand the reportable range of detection for each biomarker are shown in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000538878). All laboratory personnel were blinded to the participants’ disease status and outcomes.
Statistical Analysis
We used a Cox proportional hazards regression model to examine associations between angiopoietins and the primary outcome of CVD, as well as secondary outcomes of GF and death. Spline plots between the continuous biomarkers and the outcomes are shown in online supplementary Figure 1. We assessed the relationship between the continuous log-transformed biomarkers and biomarker quartiles with outcomes. When interpreting the log-transformed analyses, the associations were for each doubling in the biomarker concentrations; when interpreting the categorical analyses, quartile 1 was used as the reference group. We then assessed the proportional hazards assumption for outcomes by testing weighted Schoenfeld residuals. We subsequently evaluated two different multivariable clinical models and compared the predictive power of the biomarkers to the clinical models. The first model included all variables of clinical interest, while the second was a parsimonious model, including only statistically significant variables.
Variables in the full clinical model included demographics (age, sex, race, and ethnicity); CVD-related factors (hypertension, diabetes mellitus, body mass index, history of CVD, baseline estimated glomerular filtration rate); transplant-related variables (having a pancreas transplant and graft vintage at randomization); medications (lipid-lowering drugs, cyclosporine, tacrolimus, sirolimus, mycophenolate, azathioprine, and prednisone); urine albumin-to-creatinine ratio; and randomization status. The parsimonious model included the following variables: demographics (age, sex, race, and ethnicity); CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate); and graft vintage at randomization.
In a sensitivity analysis, we accounted for the competing risk of death with the outcomes of CVD and GF by creating two composite outcomes of CVD or death, and GF or death. We used the Fine and Gray model to compare biomarker quartiles on outcome occurrence [19]. In supplementary analysis, we also evaluated the associations between outcomes and angpt-1-to-angpt-2 ratio both as log transformed and quartiles to remain consistent with our prior publications on the ratio and outcomes in a non-transplant setting [7].
To assess the predictive power of angiopoietins, we calculated the area under receiver operating characteristic (ROC) curve for the clinical model (included all variables from the parsimonious model), angiopoietins (angpt-1 and -2), and the combination of the biomarkers with the clinical model with the combined outcome of CVD or death and GF or death at 7.5 years (75th percentile of the follow-up time). Significance in the change in AUC was calculated using the DeLong test.
Statistical analyses were conducted using R version 4.2.2 and SAS 9.4. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the “Declaration of Istanbul on Organ Trafficking and Transplant Tourism.” This study adhered to the Declaration of Helsinki and the Institutional Review Boards for the participating investigators approved this study.
Results
Baseline Characteristics of KTRs
This study included 2,000 KTRs from the FAVORIT who had stable kidney function for a minimum of 6 months after kidney transplantation (of 2,371 patients from the parent FAVORIT cohort). The median age of recipients at time of enrollment was 52 years as shown in Table 1. About 37% of recipients were women and 20% were Black. The majority of the recipients were American (75.7%), and 12.9% and 11.4% were from Canada and Brazil, respectively. At time of enrollment, recipients had their grafts for a median of 3.99 years. 13.5% of participants were dual kidney and pancreas recipients. About 92% were on prednisone as part of their immunosuppression regimen, 65.2% were on Mycophenolate, 50.4% were on cyclosporine, and 41.1% were on Tacrolimus. The vast majority (92.5%) of recipients carried a diagnosis of hypertension, while 43.8% had diabetes and 21.1% had CVD at baseline. The median eGFR was 47 mL/min/1.73 m2. The characteristics of all deceased donor KTRs from the parent cohort (n = 2,371) are shown in online supplementary Table S2, and characteristics of the ancillary cohort stratified according to quartiles of angpt-1 and angpt-2 levels are shown in online supplementary Tables S3 and S4.
Baseline characteristics
Variable . | Median (IQR) or N (%) total (N = 2,000) . |
---|---|
Age, years | 52 (45, 59) |
Women | 746 (37.3) |
Race | |
White | 1,461 (73.1) |
Black | 413 (20.7) |
Other | 122 (6.1) |
Treatment group | |
High-dose vitamin | 971 (48.5) |
Low-dose vitamin | 1,029 (51.5) |
Location | |
USA | 1,513 (75.7) |
Canada | 258 (12.9) |
Brazil | 229 (11.4) |
Graft vintage, years | 3.99 (1.58, 7.93) |
Kidney and pancreas transplant | 270 (13.5) |
Medications | |
Cyclosporine | 1,008 (50.4) |
Tacrolimus | 821 (41.1) |
Azathioprine | 330 (16.5) |
Sirolimus | 165 (8.2) |
Mycophenolate | 1,303 (65.2) |
Prednisone | 1,841 (92.1) |
Aspirin | 860 (43.0) |
Statin | 1,081 (54.1) |
ACEi/ARB | 879 (43.6) |
Medical History | |
CVD | 421 (21.1) |
Diabetes | 876 (43.8) |
Hypertension text | 1,849 (92.5) |
Smoker | |
Never | 978 (48.9) |
Former | 770 (38.5) |
Current | 233 (11.7) |
Exam findings | |
SBP, mm Hg | 135 (122.5, 148) |
DBP, mm Hg | 77.5 (70, 84.5) |
BMI, kg/m2 | 28.04 (24.58, 31.66) |
Lab results | |
Total cholesterol, mg/dL | 179.5 (154, 207) |
LDL, mg/dL | 97 (78, 118) |
HDL mg/dL | 44 (36, 530) |
Triglycerides, mg/dL | 166 (112.2, 236) |
Creatinine, mg/dL | 1.6 (1.3, 2.0) |
eGFR mL/min/1.73 m2 | 47.31 (36.27, 61.23) |
ACR, mg/g | 25.06 (9.01, 38.32) |
Variable . | Median (IQR) or N (%) total (N = 2,000) . |
---|---|
Age, years | 52 (45, 59) |
Women | 746 (37.3) |
Race | |
White | 1,461 (73.1) |
Black | 413 (20.7) |
Other | 122 (6.1) |
Treatment group | |
High-dose vitamin | 971 (48.5) |
Low-dose vitamin | 1,029 (51.5) |
Location | |
USA | 1,513 (75.7) |
Canada | 258 (12.9) |
Brazil | 229 (11.4) |
Graft vintage, years | 3.99 (1.58, 7.93) |
Kidney and pancreas transplant | 270 (13.5) |
Medications | |
Cyclosporine | 1,008 (50.4) |
Tacrolimus | 821 (41.1) |
Azathioprine | 330 (16.5) |
Sirolimus | 165 (8.2) |
Mycophenolate | 1,303 (65.2) |
Prednisone | 1,841 (92.1) |
Aspirin | 860 (43.0) |
Statin | 1,081 (54.1) |
ACEi/ARB | 879 (43.6) |
Medical History | |
CVD | 421 (21.1) |
Diabetes | 876 (43.8) |
Hypertension text | 1,849 (92.5) |
Smoker | |
Never | 978 (48.9) |
Former | 770 (38.5) |
Current | 233 (11.7) |
Exam findings | |
SBP, mm Hg | 135 (122.5, 148) |
DBP, mm Hg | 77.5 (70, 84.5) |
BMI, kg/m2 | 28.04 (24.58, 31.66) |
Lab results | |
Total cholesterol, mg/dL | 179.5 (154, 207) |
LDL, mg/dL | 97 (78, 118) |
HDL mg/dL | 44 (36, 530) |
Triglycerides, mg/dL | 166 (112.2, 236) |
Creatinine, mg/dL | 1.6 (1.3, 2.0) |
eGFR mL/min/1.73 m2 | 47.31 (36.27, 61.23) |
ACR, mg/g | 25.06 (9.01, 38.32) |
Results are presented as median (IQR) or n (%). At baseline, there were 4 participants with missing Race status, 3 participants with missing history of kidney or pancreas transplant, 2 participants with a missing history of diabetes, and 19 participants with missing smoking status.
ACEi/ARB, angiotensin converting enzyme inhibitors/angiotensin receptor blockers; ACR, albumin creatinine ratio; BMI, basic metabolic index; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; HDL, high density lipoprotein; LDL, low density lipoprotein; SBP, systolic blood pressure.
Angiopoietins and CVD
A total of 305 (15.25%) recipients developed a CVD event, with a median time to event of 2.54 IQR (1.11–3.80) years from enrollment. Angpt-1 levels did not show significant differences between individuals who developed CVD and those who did not, as detailed in online supplementary Table S5. However, angpt-2 levels were notably higher in individuals with CVD compared to those without (2,221.20 [1,673.1–3,079.7] pg/mL vs. 1,836.90 [1,404.8–2,501.6] pg/mL, p < 0.001) (online suppl. Table S5). No significant associations were observed between angpt-1 and CVD outcomes (Table 2).
Associations between angiopoietin-1 and outcomes
Outcome . | Biomarker . | Unadjusted model . | Full model . | Parsimonious model . |
---|---|---|---|---|
Any primary CVD | Log-transformed | 0.96 (0.91–1.02) p = 0.161 | 0.94 (0.88–1.00) p = 0.05 | 0.96 (0.91–1.02) p = 0.21 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.07 (0.79–1.46) p = 0.67 | 0.95 (0.69–1.32) p = 0.76 | 0.98 (0.72–1.34) p = 0.91 | |
Quartile 3 vs. 1 | 0.90 (0.65–1.25) p = 0.55 | 0.79 (0.56–1.12) p = 0.18 | 0.89 (0.64–1.23) p = 0.47 | |
Quartile 4 vs. 1 | 0.93 (0.68–1.29) p = 0.68 | 0.87 (0.61–1.23) p = 0.43 | 0.98 (0.70–1.36) p = 0.89 | |
Death | Log-transformed | 0.95 (0.89–1.01) p = 0.077 | 0.94 (0.88–1.00) p = 0.06 | 0.96 (0.90–1.02) p = 0.22 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 0.94 (0.67–1.33) p = 0.73 | 0.80 (0.60–1.24) p = 0.41 | 0.89 (0.63–1.27) p = 0.52 | |
Quartile 3 vs. 1 | 0.92 (0.65–1.30) p = 0.63 | 0.77 (0.52–1.14) p = 0.19 | 0.95 (0.66–1.36) p = 0.77 | |
Quartile 4 vs. 1 | 0.82 (0.58–1.17) p = 0.28 | 0.81 (0.55–1.21) p = 0.30 | 0.90 (0.62–1.30) p = 0.57 | |
GF | Log-transformed | 0.95 (0.88–1.02) p = 0.13 | 0.93 (0.85–1.00) p = 0.06 | 0.94 (0.87–1.02) p = 0.13 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.22 (0.82–1.82) p = 0.34 | 1.06 (0.69–1.63) p = 0.79 | 1.01 (0.67–1.52) p = 0.96 | |
Quartile 3 vs. 1 | 1.04 (0.69–1.59) p = 0.84 | 0.88 (0.56–1.39) p = 0.58 | 0.94 (0.61–1.44) p = 0.76 | |
Quartile 4 vs. 1 | 0.82 (0.52–1.28) p = 0.38 | 0.75 (0.46–1.23) p = 0.25 | 0.81 (0.51–1.28) p = 0.37 |
Outcome . | Biomarker . | Unadjusted model . | Full model . | Parsimonious model . |
---|---|---|---|---|
Any primary CVD | Log-transformed | 0.96 (0.91–1.02) p = 0.161 | 0.94 (0.88–1.00) p = 0.05 | 0.96 (0.91–1.02) p = 0.21 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.07 (0.79–1.46) p = 0.67 | 0.95 (0.69–1.32) p = 0.76 | 0.98 (0.72–1.34) p = 0.91 | |
Quartile 3 vs. 1 | 0.90 (0.65–1.25) p = 0.55 | 0.79 (0.56–1.12) p = 0.18 | 0.89 (0.64–1.23) p = 0.47 | |
Quartile 4 vs. 1 | 0.93 (0.68–1.29) p = 0.68 | 0.87 (0.61–1.23) p = 0.43 | 0.98 (0.70–1.36) p = 0.89 | |
Death | Log-transformed | 0.95 (0.89–1.01) p = 0.077 | 0.94 (0.88–1.00) p = 0.06 | 0.96 (0.90–1.02) p = 0.22 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 0.94 (0.67–1.33) p = 0.73 | 0.80 (0.60–1.24) p = 0.41 | 0.89 (0.63–1.27) p = 0.52 | |
Quartile 3 vs. 1 | 0.92 (0.65–1.30) p = 0.63 | 0.77 (0.52–1.14) p = 0.19 | 0.95 (0.66–1.36) p = 0.77 | |
Quartile 4 vs. 1 | 0.82 (0.58–1.17) p = 0.28 | 0.81 (0.55–1.21) p = 0.30 | 0.90 (0.62–1.30) p = 0.57 | |
GF | Log-transformed | 0.95 (0.88–1.02) p = 0.13 | 0.93 (0.85–1.00) p = 0.06 | 0.94 (0.87–1.02) p = 0.13 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.22 (0.82–1.82) p = 0.34 | 1.06 (0.69–1.63) p = 0.79 | 1.01 (0.67–1.52) p = 0.96 | |
Quartile 3 vs. 1 | 1.04 (0.69–1.59) p = 0.84 | 0.88 (0.56–1.39) p = 0.58 | 0.94 (0.61–1.44) p = 0.76 | |
Quartile 4 vs. 1 | 0.82 (0.52–1.28) p = 0.38 | 0.75 (0.46–1.23) p = 0.25 | 0.81 (0.51–1.28) p = 0.37 |
Variables in the full model included: demographics (age, sex, race, and ethnicity); CVD-related factors (hypertension, diabetes mellitus, body mass index, history of CVD, baseline estimated glomerular filtration rate); transplant-related variables (having a pancreas transplant, and graft vintage at randomization); medications (lipid-lowering drugs, cyclosporine, tacrolimus, sirolimus, mycophenolate, azathioprine, and prednisone); urine albumin-to-creatinine ratio; and randomization status. The Parsimonious model included the following variables: demographics (age, sex, race, and ethnicity); CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate); and graft vintage at randomization.
CVD, cardiovascular disease.
For each doubling of baseline angpt-2 levels, there was a 29% increased risk of CVD (aHR 1.29, 95% CI: [1.08–1.54], p = 0.006) as shown in Table 3. Recipients with angpt-2 levels in quartile 4 had an 85% higher risk of CVD compared to those in quartile 1 (aHR 1.85, 95% CI: [1.25–2.73], p = 0.002). The Kaplan-Meier curves illustrate reduced CVD over time with decreasing quartiles of angpt-2 (Fig. 1a). Additional details regarding biomarker-event associations comprising the composite CVD outcome are presented in online supplementary Tables S6 and S7.
Associations between angiopoietin-2 and outcomes
Outcome . | Biomarker . | Unadjusted model . | Full model . | Parsimonious model . |
---|---|---|---|---|
Any primary CVD | Log-transformed | 1.67 (1.44–1.93) p<0.001 | 1.29 (1.08–1.54) p = 0.006 | 1.39 (1.18–1.65) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.35 (0.91–2.00) p = 0.13 | 1.19 (0.79–1.80) p = 0.40 | 1.20 (0.81–1.79) p = 0.37 | |
Quartile 3 vs. 1 | 2.22 (1.54–3.14) p<0.001 | 1.89 (1.29–2.77) p<0.001 | 1.83 (1.26–2.64) p<0.001 | |
Quartile 4 vs. 1 | 3.09 (2.18–4.37) p<0.001 | 1.85 (1.25–2.73) p = 0.002 | 2.14 (1.48–3.09) p<0.001 | |
Death | Log-transformed | 2.03 (1.73–2.37) p<0.001 | 1.55 (1.29–1.87) p<0.001 | 1.70 (1.42–2.03) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.35 (0.85–2.13) p = 0.21 | 1.04 (0.63–1.70) p = 0.89 | 1.12 (0.70–1.78) p = 0.64 | |
Quartile 3 vs. 1 | 2.29 (1.50–3.48) p<0.001 | 1.90 (1.22–2.96) p = 0.004 | 1.78 (1.16–2.73) p = 0.008 | |
Quartile 4 vs 1 | 3.64 (1.45–5.41) p<0.001 | 2.30 (1.48–3.58) p<0.001 | 2.40 (1.58–3.65) p<0.001) | |
GF | Log-transformed | 1.86 (1.53–2.26) p<0.001 | 1.53 (1.24–1.88) p<0.001 | 1.70 (1.39–2.08) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.22 (0.74–2.01) p = 0.44 | 1.38 (0.80–2.40) p = 0.25 | 1.18 (0.70–1.99) p = 0.54 | |
Quartile 3 vs. 1 | 2.06 (1.30–3.25) p = 0.002 | 2.34 (1.41–3.87) p<0.001 | 2.21 (1.37–3.55) p<0.001 | |
Quartile 4 vs. 1 | 2.60 (1.66–4.06) p<0.001 | 2.24 (1.36–3.70) p = 0.002 | 2.55 (1.59–4.09) p<0.001 |
Outcome . | Biomarker . | Unadjusted model . | Full model . | Parsimonious model . |
---|---|---|---|---|
Any primary CVD | Log-transformed | 1.67 (1.44–1.93) p<0.001 | 1.29 (1.08–1.54) p = 0.006 | 1.39 (1.18–1.65) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.35 (0.91–2.00) p = 0.13 | 1.19 (0.79–1.80) p = 0.40 | 1.20 (0.81–1.79) p = 0.37 | |
Quartile 3 vs. 1 | 2.22 (1.54–3.14) p<0.001 | 1.89 (1.29–2.77) p<0.001 | 1.83 (1.26–2.64) p<0.001 | |
Quartile 4 vs. 1 | 3.09 (2.18–4.37) p<0.001 | 1.85 (1.25–2.73) p = 0.002 | 2.14 (1.48–3.09) p<0.001 | |
Death | Log-transformed | 2.03 (1.73–2.37) p<0.001 | 1.55 (1.29–1.87) p<0.001 | 1.70 (1.42–2.03) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.35 (0.85–2.13) p = 0.21 | 1.04 (0.63–1.70) p = 0.89 | 1.12 (0.70–1.78) p = 0.64 | |
Quartile 3 vs. 1 | 2.29 (1.50–3.48) p<0.001 | 1.90 (1.22–2.96) p = 0.004 | 1.78 (1.16–2.73) p = 0.008 | |
Quartile 4 vs 1 | 3.64 (1.45–5.41) p<0.001 | 2.30 (1.48–3.58) p<0.001 | 2.40 (1.58–3.65) p<0.001) | |
GF | Log-transformed | 1.86 (1.53–2.26) p<0.001 | 1.53 (1.24–1.88) p<0.001 | 1.70 (1.39–2.08) p<0.001 |
Quartile 1 | (reference) | |||
Quartile 2 vs. 1 | 1.22 (0.74–2.01) p = 0.44 | 1.38 (0.80–2.40) p = 0.25 | 1.18 (0.70–1.99) p = 0.54 | |
Quartile 3 vs. 1 | 2.06 (1.30–3.25) p = 0.002 | 2.34 (1.41–3.87) p<0.001 | 2.21 (1.37–3.55) p<0.001 | |
Quartile 4 vs. 1 | 2.60 (1.66–4.06) p<0.001 | 2.24 (1.36–3.70) p = 0.002 | 2.55 (1.59–4.09) p<0.001 |
Variables in the full model included: demographics (age, sex, race, and ethnicity); CVD-related factors (hypertension, diabetes mellitus, body mass index, history of CVD, baseline estimated glomerular filtration rate); transplant-related variables (having a pancreas transplant, and graft vintage at randomization); medications (lipid-lowering drugs, cyclosporine, tacrolimus, sirolimus, mycophenolate, azathioprine, and prednisone); urine albumin-to-creatinine ratio; and randomization status. The Parsimonious model included the following variables: demographics (age, sex, race, and ethnicity); CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate); and graft vintage at randomization.
CVD, cardiovascular disease.
a Kaplan-Meier curve of the probability CVD-free survival stratified by angpt-2 quartiles. b Kaplan-Meier curve of the probability of survival stratified by angpt-2 quartiles. c Kaplan-Meier curve of the probability of GF-free survival stratified by angpt-2 quartiles.
a Kaplan-Meier curve of the probability CVD-free survival stratified by angpt-2 quartiles. b Kaplan-Meier curve of the probability of survival stratified by angpt-2 quartiles. c Kaplan-Meier curve of the probability of GF-free survival stratified by angpt-2 quartiles.
Competing risk analysis to account for death showed similar results (online suppl. Table S8). However, in the competing risk analysis, for each doubling of angpt-1 levels there was an 11% significant decrease in CVD (aHR 0.89, 95% CI: [0.82–0.96]).
In supplementary analysis, we have explored the relationship between angpt-1and angpt-2 ratio to remain consistent with our prior publication [7] and have reported our findings in online supplementary Table S9. We identified that for each doubling of angpt-1-to-angpt-2 ratio, there was a significant 7% lower risk in CVD (aHR 0.93, 95% CI: [0.88–0.99], p = 0.018), 10% lower risk of death (aHR 0.90, 95% CI: [0.84–0.96], p < 0.001), and 12% lower risk of GF (aHR 0.88, 95% CI: [0.82–0.95], p < 0.001). When analyzing the data using quartiles of the ratio, we identified significant independent associations between the highest quartile of angpt-1-to-angpt-2 (quartile 4) as compared to the lowest quartile (quartile 1) for the outcomes of death and GF, where there was a 42% lower risk (aHR 0.58, 95% CI: [0.39–0.87], p = 0.008) of death and a 48% lower risk (aHR 0.52, 95% CI: [0.32–0.85], p = 0.009) of GF.
Angiopoietins and Death
A total of 247 (12.4%) recipients died during study follow-up, with a median time to death of 2.72 IQR (1.38, 4.03) years. While angpt-1 levels did not significantly differ between survivors and non-survivors (online suppl. Table S5), angpt-2 levels were significantly higher among those who died (2,381.1 [1,713.4–3,359.3] pg/mL vs. 1,840.1 [1,408.1–2,505.5] pg/mL, p < 0.001) (online suppl. Table S5).
There were no significant associations found between angpt-1 and mortality, as indicated in Table 2. However, for each doubling of angpt-2 levels, there was a 55% increased risk of CVD (aHR 1.55, 95% CI: [1.29–1.87], p < 0.001), as demonstrated in Table 3. Similarly, recipients with angpt-2 levels in quartile 4 exhibited a 2.3-fold higher risk of death compared to those in quartile 1 (aHR 2.30, 95% CI: [1.48–3.58], p < 0.001). The Kaplan-Meier curves illustrate higher survival rates over time with decreasing angpt-2 quartiles (Fig. 1b).
Angiopoietins and GF
A total of 176 (8.8%) recipients experienced GF, with a median time to occurrence of 4.78 IQR (2.85, 5.64) years. Angpt-1 levels did not significantly differ between recipients who developed GF and those who did not (online suppl. Table S5). Angpt-2 levels were significantly higher among those with GF compared to those without (2,138.4 [1,682.7–3,278.2] pg/mL vs. 1,861.0 [1,424.8–2,590.7] pg/mL, p < 0.001) (online suppl. Table S5). Angpt-1 did not show any association with GF as shown in Table 2. However, for every doubling of angpt-2 levels, there was a 53% increased risk of GF, as indicated in Table 3 (aHR 1.53, 95% CI: [1.24–1.88], p < 0.001). Recipients in angpt-2 quartile 4 exhibited a 2.2-fold higher risk of GF compared to those in quartile 1 (aHR 2.24, 95% CI: [1.36–3.70], p = 0.002). The Kaplan-Meier curves displayed a decrease in GF over time with decreasing angpt-2 quartiles (Fig. 1c).
Competing risk analysis, adjusting for death, yielded similar results for angpt-2, as outlined in online supplementary Table S8. In contrast to the primary analysis, angpt-1 exhibited significant and independent associations with GF. Specifically, each doubling of angpt-1 levels corresponded to a 13% decreased risk of GF (aHR 0.87, 95% CI: [0.80–0.94]). Recipients in the 4th quartile of angpt-1 levels experienced a 34% reduction in GF risk compared to recipients in quartile 1 (aHR 0.66, 95% CI: [0.47–0.92]).
Prediction of CVD or Death and GF or Death
For the outcome of CVD or death, the areas under the ROC curves for angiopoietins, the clinical model, and the combined model were 0.63 (95% CI: 0.60–0.66), 0.70 (95% CI: 0.68–0.73), and 0.72 (95% CI: 0.69–0.75), respectively, as seen in Figure 2a. The increase in AUC resulting from the incorporation of the biomarkers into the clinical model demonstrated statistical significance (p = 0.005).
ROC curves for the prediction of CVD, GF and death at 7.5 years. a CVD or death : the addition of angiopoietins (angpt-1 and -2, AUC: 0.63) to the clinical model which included demographics (age, sex, race, and ethnicity); CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate); and graft vintage at randomization (AUC: 0.70) significantly improved prediction of the composite outcome of CVD or death (combined AUC: 0.72, p = 0.005). b GF or death: the addition of angiopoietins (AUC: 0.63) to the clinical model (AUC: 0.68) significantly improved prediction of the composite outcome of GF or death (combined AUC: 0.70, p = 0.005).
ROC curves for the prediction of CVD, GF and death at 7.5 years. a CVD or death : the addition of angiopoietins (angpt-1 and -2, AUC: 0.63) to the clinical model which included demographics (age, sex, race, and ethnicity); CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate); and graft vintage at randomization (AUC: 0.70) significantly improved prediction of the composite outcome of CVD or death (combined AUC: 0.72, p = 0.005). b GF or death: the addition of angiopoietins (AUC: 0.63) to the clinical model (AUC: 0.68) significantly improved prediction of the composite outcome of GF or death (combined AUC: 0.70, p = 0.005).
Regarding the event of GF or death, the areas under the ROC curves for angiopoietins, the clinical model, and the combined model were 0.63 (95% CI: 0.60–0.66), 0.68 (95% CI: 0.65–0.71), and 0.70 (95% CI: 0.67–0.73), respectively, as seen in Figure 2b. The change in AUC with the addition of the biomarkers to the clinical model was statistically significant (p = 0.005).
Discussion
Using an ancillary study of 2,000 deceased donor KTRs from the FAVORIT, we identified that recipients with higher levels of angpt-2, a vascular marker of injury and inflammation, were at a significantly and independently higher risk of future CVD, GF, and death. In our cohort, having higher levels of angpt-1, a marker of vessel stability was not significantly associated with any of the outcomes in deceased donor KTRs. Furthermore, we identified that the addition of angiopoietins to routine clinical variables significantly improved predictions of CVD or death and GF or death at 7.5 years. For predictive analyses, we accounted for participants that could have died prior to having a CVD event or GF by including death in both composite outcomes.
Graft survival has continued to improve throughout the years; however, nonrenal outcomes have not had similar success [20]. CVD and CVD-related deaths are remarkably higher in KTRs compared to the general population [21]. Yet, there are no available clinical tools to phenotype recipients who are at higher risk for CVD to better tailor their posttransplant follow-up. For example, only 36% of KTRs are on statins [22], and approximately 40% of KTRs are obese at the time of transplant, with most gaining around 5–10% of their pretransplant weight within the first year [23, 24]. While transplant clinics are generally not geared to focus on CVD prevention such as lipid management and weight loss counseling, identifying high-risk recipients could allow transplant providers to allocate time and effort to these important CVD preventative measures or communicate with patients the importance of primary care follow-up to discuss these measures further.
Our study has several strengths that contribute to its scientific value. While previous research has shown a link between angpt-2 and increased inflammation and vessel instability [25, 26], our findings are the first to demonstrate this biology epidemiologically in the kidney transplant setting, thereby providing valuable mechanistic insight. Additionally, our large sample size of 2,000 recipients and recruitment from multiple geographical regions improves the generalizability of our findings. We were able to account for many confounding variables including demographic factors, clinical parameters, medication usage, and transplant-related variables, which improve our ability to isolate the effects of angiopoietins on CVD outcomes, GF and death. Furthermore, our analysis included a competing risk analysis to account for the potential of death to preclude the occurrence of CVD and GF. Notably, our sensitivity analysis confirmed the associations identified between angpt-2 and elevated risks of CVD and GF, while also revealing that angpt-1 was significantly and independently associated with a reduced risk of CVD and GF. By adjusting for the competing risk of death, we were able to better isolate the true effect of angpt-1 on CVD and GF, considering the multiple comorbidities that KTRs develop.
Furthermore, our supplementary analysis evaluating the relationship between the ratio of angpt-1-to-angpt-2 shows a significant and independent reduction in the risk of CVD, death and GF for each doubling of the ratio, which is consistent with our prior findings in the non-transplant setting [7]. However, the most significant lowering of risk was identified between the highest quartile of the ratio and the outcomes of death and GF, but not CVD.
It is important to acknowledge the limitations of our study. We were only able to assess the relationship between angiopoietins and outcomes from a single time point, which was remote from the time of transplantation. Therefore, our study does not address the peri-transplant values of angiopoietins or the role of longitudinal measurements of angiopoietins in the future development of CVD, GF and death. Additionally, our study employed an observational cohort design, which hinders any inferences on the causal relationship between angiopoietins and outcomes. Due to our study design, our identified associations may have also been affected by unmeasured confounding. Although angpt-2 may partially explain why some recipients developed CVD and other unfavorable outcomes posttransplantation, it is important to note that angpt-2 alone did not provide strong predictions of future outcomes as compared to the clinical model alone. As the clinical model included demographics (age, sex, race, and ethnicity), CVD-related factors (diabetes mellitus, history of CVD, and baseline estimated glomerular filtration rate), and graft vintage at randomization, it is unlikely that angpt-1 and angpt-2 alone would better predict outcomes. Future research is needed to assess how a panel of biomarkers capturing the vascular health of the recipient may help improve outcomes as compared to available clinical variables. Lastly, our findings are yet to be externally validated, which limit their translation to the bedside at this time.
In conclusion, our study underscored the importance of angpt-2 in explaining the higher risk of CVD in deceased donor KTRs. Further research is needed to assess if angiopoietins along with existing clinical markers may aid in tailoring follow-up after transplantation to reduce the risk of CVD in those high-risk individuals.
Acknowledgments
The authors would like to thank all of the FAVORIT study participants, research coordinators, and support staff for making this study possible. The opinions expressed in this paper do not necessarily reflect those of the NIDDK, the NIH, the Department of Health and Human Services, or the government of the USA.
The original study was conducted by the FAVORIT Investigators and supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). The data and biospecimens from the FAVORIT reported here were supplied by NIDDK Central Repository (NIDDK-CR) and are available for request at https://repository.niddk.nih.gov. This manuscript was not prepared under the auspices of the FAVORIT study and does not necessarily reflect the opinions or views of the FAVORIT study, NIDDK-CR, or NIDDK.
Statement of Ethics
The parent study (FAVORIT) obtained informed consent from all participants; however, as our study was an ancillary study using pre-existing samples, it was granted an exemption by the Yale Institutional Review Board. It was determined that the investigator is not engaged in research involving human subjects. As such, IRB review and approval are not required.
Conflict of Interest Statement
S.G. Mansour is a Guest Editor for Nutrients-Nutrition in Hemodialysis, Co-director of the American Society of Nephrology AKINOW Basic Science Workgroup, and currently has funding from the Doris Duke Foundation, American Heart Association and the NIH/NIDDK.
Funding Sources
This work was supported by the NIH/NIDDK (Grant 1K23DK127154-01A1).
Author Contributions
S.G. Mansour is the study supervisor and designed and developed the current ancillary study, acquired and interpreted the data, drafted and revised the paper, conducted the statistical analysis, and obtained funding for the study. A. Bostom designed the parent cohort study and acquired the data. L. Brown and N. Gendy analyzed and interpreted the data and drafted the manuscript. M.K. Staunton, K. Garg, Y. Yamamoto, and U. Ugwuowo acquired and analyzed the data as well as conducted the statistical analysis. L. Qian, N.H. Garcilazo, W. Obeid, and L. Al-Qusairi analyzed, acquired or interpreted the data. All authors revised the paper and approved of the final version of the manuscript.
Additional Information
Natalie Gendy and Liam Brown contributed equally to the manuscript.IRB Number: 2000027329.
Data Availability Statement
To optimize the use of participants’ contributions, de-identified participant level data can be made available upon request, from S.G., request for effective use by other members of the scientific community. Data from the Folic Acid for Vascular Outcome Reduction in Transplantation Trial [(V3)/https://doi.org/10.58020/6y8c-jz44] reported here are available for request at the NIDDK Central Repository (NIDDK-CR) website, Resources for Research (R4R), https://repository.niddk.nih.gov/.