This prospective, single-center study was carried out in Erasmus Medical Center, the Netherlands, between 10 January 2019 and 13 August 2020. Power calculation for the study sample size can be found in the Supplementary Material. Patients who met the eligibility criteria were asked to participate upon admission for transplant. All patients 50 years or older were eligible for inclusion. Patients from 30 years of age or older were eligible if they had at least one of the following risk factors for vascular disease: diabetes mellitus, 1 year or longer dialysis duration, smoking history of at least 10 pack years or a history with peripheral arterial disease, ischemic heart disease or a cerebrovascular accident (19). In addition, South-East Asian ethnicity was considered a risk factor as previous studies found that these patients were at increased risk of cardiovascular disease after adjustment for confounders (20). Combined liver-kidney transplant recipients, HLA incompatible recipients and dual transplant recipients were excluded from the study. The study was performed according to the Declaration of Helsinki and received approval from the local Medical Ethical Committee (MEC 2018-1401). The study was prospectively registered in the Netherlands Trial Register (NTR7641).

Patients who gave written informed consent underwent a non-contrast-enhanced abdominal CT-scan upon admission for transplant. All scanners used were modern ≥128-multislice CT systems (Siemens Healthineers). A scanning protocol was developed specifically for the study to ensure that there would be no differences in scan parameters that could affect measurement of the calcification score. A low-dose CT-scan (at fixed 120 kVp) was performed and reconstructed to 3.0 mm slice thickness and 1.5 mm increment using a dedicated quantitative calcium scoring kernel (B35f or Qr36) without iterative reconstructions. Consequently, the scan was analyzed in Intellispace Portal (Philips) with the HeartBeat-CS application. This application is designed to calculate the Agatston score, which is a continuous quantification score for coronary calcification with a high specificity for the absence of coronary artery disease (21, 22). The calculation is based on the weighted density score given to the highest attenuation value multiplied by the area of the calcification speck. A CT attenuation threshold of 130 Hounsfield units (HU) is used for the detection of calcification, with only contiguous voxels totaling ≥1 mm² in area counted as lesions to reduce the influence of image noise(21). For calculation of the AIC score as an adapted version of the Agatston score, the aorto-iliac trajectory was divided into anatomical segments, as explained in Figure 1. These anatomical segments included the infrarenal aorta until the iliac bifurcation (segment I), the right and left common iliac artery until the internal iliac artery branch (segment II), and the right and left external iliac artery (starting from the internal iliac artery branch until Poupart’s ligament) (segment III). The total AIC score was calculated as the sum of these separate calcification scores. The ipsilateral AIC score consisted of the sum of the aorta and ipsilateral common iliac artery, depending on the implantation side. The external iliac artery calcification score was not included in this score for 2 reasons. Firstly, the external iliac artery is not entirely in the inflow trajectory of the donor kidney, depending on whether the anastomosis is made with the proximal or distal part. Secondly, the transition from external iliac artery to common femoral artery was often unclear because of the use of a low-dose CT-scan which could result in misclassification.

The primary outcome of the study was eGFR in the first year post-transplant. All measurements of eGFR from day 1 until 400 days were used, as not all patients had their 1-year eGFR measurement exactly at 365 days post-transplant. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) formula, which is standard in our hospital as recommended by the kidney disease improving global outcomes (KDIGO) guidelines (23, 24). Secondary outcomes were the incidence of delayed graft function (DGF), primary non-function (PNF), the presence and number of biopsy-proven acute rejection episodes in the first year post-transplant, uncensored and death-censored graft survival, and the need for a peri-operative or pre-operative vascular intervention.

All donor kidneys were transplanted into the right or left iliac fossa by using the standard Gibson incision, depending on surgeon preference. Firstly, the anastomosis of the renal vein was performed end-to-side with the external iliac vein. Consequently, the arterial anastomosis was performed end-to-side with the external iliac artery using prolene 5.0 or 6.0. In the case of severe aorto-iliac calcification without a soft spot to implant the kidney, an endarterectomy could be performed with patch angioplasty prior to the arterial anastomosis. For the ureter, an extra-vesical anastomosis was performed as described by Lich-Gregoir. The ureter anastomosis was protected with an external splint or double J stent, depending on randomization arm of an ongoing randomized controlled trial investigating urologic complications. Standard immunosuppression regime of transplant recipients consisted of induction with basiliximab followed by triple-therapy with tacrolimus, mycofenolate mofetil and prednisone. Prednisone was gradually tapered and stopped after 4 months.

Continuous variables were presented as median and interquartile range (IQR) for non-normally distributed variables and mean with standard deviation for normally distributed variables. Categorical variables were summarized as number and percentage. Baseline characteristics were compared between patients with a median or lower ipsilateral AIC score and patients with an above median AIC score. Categorical baseline characteristics were compared with chi-square tests or Fisher’s exact test. Continuous baseline characteristics were compared using Mann-Whitney U test. Correlations between arterial segments were calculated using Spearman’s correlation coefficient. The association between the ipsilateral AIC score as a continuous variable and eGFR trajectory was analyzed using a linear mixed model with random intercepts and random slopes. An unstructured covariance matrix was used because of unbalanced outcome data. We used a predefined model to correct for donor, recipient and transplant-related confounders. We included the following variables in our fixed effects: follow-up time, ipsilateral AIC score, recipient age, recipient sex, recipient diabetes, recipient smoking, a previous kidney transplant, total dialysis duration (including hemodialysis and peritoneal dialysis), coronary artery disease, peripheral arterial disease, donor type, donor age, donor diabetes, donor last creatinine, pre-emptive transplantation, cold preservation technique (static cold storage or hypothermic machine perfusion), total human leukocyte antigen (HLA) mismatch, virtual panel reactive antibodies (vPRA), cold ischemic time, postoperative dialysis and 1 or more rejection episodes in the first year after transplant. The univariable analysis can be found in the Supplementary Table S1. In case of non-linearity for continuous variables, a natural cubic spline with 3 degrees of freedom was used (25). Knots for splines were selected based on the observed non-linear trajectories. Multicollinearity was investigated by calculating the generalized variance inflation factor (GVIF), where a value below 5 indicates no multicollinearity. The following clinically plausible interactions were tested with likelihood ratio tests: time and ipsilateral calcification score, time and donor type and time and delayed graft function. The main analysis included all eGFR measurements of the recipients, independent of graft failure. We performed one sensitivity analysis where we imputed an eGFR of 10 mL/min/1.73 m² after graft failure occurred if the measured eGFR was above 10 mL/min/1.73 m². Secondary outcomes were analyzed using unadjusted analyses due to a lack of statistical power for these outcomes. Differences in survival were calculated with the log-rank test. Median follow-up time was calculated with the reversed Kaplan-Meier method. R statistical software version 4.0.4. was used for data analysis (packages “nlme,” “lme4,” “splines2,” “survival”).

A total of 140 kidney transplant recipients were included in the study and received a non-contrast-enhanced CT-scan. The inclusion flowchart is shown in Figure 2. 272 patients received a kidney transplant during the inclusion period of which 69 did not meet the inclusion criteria. Two-hundred-and-three patients were eligible for inclusion with a consent rate of 69.0%. The consent rate was lower than expected due to logistical reasons and a mandatory study stop due to the COVID-19 pandemic. Median follow-up time in the cohort was 622 days (IQR 416-757).

Median AIC scores per segment are illustrated in Table 1. The right and left external iliac artery had the lowest calcification scores with a median of 48 (IQR 0–412) and 52 (0–466), respectively. Scatter plots to identify correlations between calcification scores of separate arterial segments are presented in Figure 3. All AIC scores of arterial segments were highly correlated, but the highest correlation was found between the right and left common iliac artery (Spearman’s ρ = 0.90, p < 0.001). The lowest correlation coefficients were found between the right external and common iliac artery (Spearman’s ρ = 0.70, p < 0.001) and left external and common iliac artery (Spearman’s ρ = 0.65, p < 0.001).

Baseline characteristics of the cohort are displayed in Table 2. To demonstrate differences between patients with a low or high ipsilateral AIC score, baseline characteristics were compared according to the median AIC score. Recipient age upon transplantation was significantly higher in recipients with a high ipsilateral calcification score (median age 66.5 (IQR 61.8–72.5) compared to 63.2 (IQR 54.7–68.7), p = 0.005). Patients with a high ipsilateral AIC score were more often current or former smokers (overall p = 0.007). Furthermore, patients with a high AIC score were more often presenting with a history of coronary artery disease (p < 0.001) or cerebrovascular disease (p = 0.039). Other baseline characteristics were not statistically different.

Patients had a median of 31 (IQR 26–39) eGFR measurements available in the first year after transplant. Figure 4 shows the spaghetti plot for eGFR trajectory, stratified according to donor type (Figure 4B) or AIC score quartile (Figure 4C). All AIC quartiles showed a similar pattern with a steep increase in the first period followed by a small decrease and stabilization phase. eGFR trajectories during the stabilization phase were visually different with a lower eGFR trajectory in the highest calcification quartile.

To assess whether this visual difference in eGFR trajectory during the first year post-transplant could also be demonstrated when corrected for potential confounders, a linear mixed model was performed. The results of the linear mixed model are presented in Table 3. No direct association was observed between AIC score and post-transplant eGFR (β 0.02 (standard deviation (SD) 0.18), p = 0.919). Between day 50 till day 100 after transplant, eGFR increased significantly with 30.90 mL/min/1.73 m² (SD 3.59, p < 0.001). After 100 days follow-up time, eGFR increased significantly with 11.75 mL/min/1.73 m² (SD 1.80, p < 0.001). Patients who received a DCD graft had a lower eGFR (14.41 mL/min/1.73 m² lower (SD 4.918, p = 0.003). An increase in donor age led to a decrease in eGFR with 0.29 mL/min/1.73 m² per year of age (SD 0.08, p < 0.001). Furthermore, the need for postoperative dialysis and acute rejection in the first year were both associated with a lower eGFR with (9.11 mL/min/1.73 m² and 6.69 mL/min/1.73 m², respectively (SD 3.21, p = 0.005 and SD 2.82, p = 0.019, respectively)). The interactions between follow-up time and donor type and follow-up time and delayed graft function showed no significant addition to the model (likelihood ratio test: p = 0.097 and 0.134, respectively). However, a significant interaction was observed between follow-up time and ipsilateral AIC score (overall likelihood ratio test p < 0.001). The predicted values for eGFR during the follow-up time based on the mixed model are plotted in Figure 5, stratified according to ipsilateral AIC score quartile. This shows a sharp increase in eGFR measurements in the first period (day 0–50) for all values of ipsilateral AIC score. Further in the follow-up, it is shown that there is a relation between follow-up time and ipsilaterial AIC score; higher AIC scores show lower values of eGFR over time. The calculated GVIF values showed no important multicollinearity. Our sensitivity analysis, included in the Supplementary Table S2, showed similar results as our main analysis.

The secondary outcomes are noted in Table 4 and compared between patients with a high AIC score and patients with a low AIC score. No significant differences were observed for the incidence of DGF, PNF and acute rejection within the first year after transplant. Seven patients in the high AIC group received a perioperative endarterectomy with patch angioplasty compared to none in the low AIC group, which was a statistical significant difference (p = 0.013). One patient in the high AIC group received a preoperative percutaneous transluminal angioplasty (PTA) with stenting compared to none in the below median AIC group (p = 1.000). Patient survival during the follow-up was inferior in the high AIC score group (log-rank test: p = 0.010). No difference was observed for death-censored graft survival.