Insights from the 4C-T Study suggest increased cardiovascular burden in girls with end stage kidney disease before and after kidney transplantation

Rizky I. Sugianto, PhD, Nima Memaran, MD, Bernhard M.W. Schmidt, MD MSc, Anke Doyon, MD, Daniela Thurn-Valsassina, MD, Harika Alpay, MD, Ali Anarat, MD, Klaus Arbeiter, MD, Karolis Azukaitis, MD, Aysun K. Bayazit, MD, Ipek K. Bulut, MD, Salim Caliskan, MD, Nur Canpolat, MD, Ali Duzova, MD, Jutta Gellerman, MD, Jerome Harambat, MD PhD, Denise Homeyer, MA, Mieczyslaw Litwin, MD, Francesca Mencarelli, MD PhD, Lukasz Obrycki, MD, Dusan Paripovic, MD, Bruno Ranchin, MD, Rukshana Shroff, MD, PhD, Uwe Tegtbur, MD, Jeannine von der Born, MD, Ebru Yilmaz, MD, Uwe Querfeld, MD, Elke Wühl, MD, Franz Schaefer, MD, Anette Melk, MD PhD


RESULTS
• Girls are more susceptible towards the development of arterial stiffening.
• Susceptibility in girls is associated with magnitude and duration of impaired kidney function.
• This finding might contribute to the higher mortality risk shown for girls with CKD.

4C-T cohort: 235 children with CKD and transplantation
Prospective multicenter study 704 children with CKD • Annual CV assessments • Total visits n=1368 n=80 n=155 Endpoint: PWV z-score AIMS Arterial stiffness differs between sexes in children with chronic kidney disease (CKD) and transplantation (Tx).
Longitudinal assessment of PWV, a measure of vascular stiffness and predictor cardiovascular (CV) mortality, to determine sex differences and potential contributing factors.

Pulse wave velocity (PWV)
general population CKD ± Tx population

Introduction
Overall childhood mortality rates are declining 1 . In the general population, boys show higher mortality in most regions of the world 2, 3 largely due to more accidents 1 , prematurity, respiratory distress during infancy 2,4 and sepsis occurring post-puberty 3 .
Inferior survival in girls is associated with poverty, marginalization and a sociocultural preference for male offspring 2 . Mortality in children with end stage kidney disease (ESKD) is more than 30 times higher than the general population 5 . Data from USRDS on 14,024 children on kidney replacement therapy (KRT) suggest a higher mortality risk in girls (HR: 1.36, 95% CI: 1.25-1.50) due to their greater risk for cardiovascular death 6 . Despite declining overall mortality rates in children with functioning grafts, the proportion of cardiovascular mortality remains unchanged and is about 20% higher in girls 7 .
Cardiovascular events as the most common causes of death in children with ESKD account for about one third of deaths in children on dialysis and a quarter of those undergoing transplantation 8 . Data from the Australian and New Zealand Dialysis and Transplant Registry suggested an even higher mortality in pediatric kidney transplant recipients due to cardiovascular causes 9 . The post-transplant mortality due to cardiovascular causes is greater than that related to non-functioning grafts 10 .
Early measures of arterial stiffness such as increased aortic pulse wave velocity (PWV) are highly predictive for cardiovascular events and mortality 11 and associated with a faster decline in estimated glomerular filtration rate (eGFR) in adults with CKD 12 . Aortic PWV can be measured non-invasively and reproducibly in children 13,14 .

J o u r n a l P r e -p r o o f
Findings in adults indicate that the global survival advantage of females is lost in ESKD 19 , a phenomenon that is not sufficiently explained by disparities of access to transplantation due to higher levels of panel reactive antibodies in women 20 and pregnancy-induced-incompatibility 21 . In the pediatric population, girls are less likely to undergo pre-emptive transplantation 6,22 and show poorer graft survival than boys, the latter being partly explained by receiving male donor organs 23,24 . Our own data indicated a greater susceptibility of girls for cyclosporin A-associated hypertension 25 , which could contribute to poorer graft survival and increased cardiovascular mortality.
Here we aimed to study the course of arterial stiffness in children with ESKD who underwent transplantation either pre-emptively or after prior dialysis, to uncover potential sex differences.

Study design, setting, participants
The 4C-T (Cardiovascular Comorbidity in Children with Chronic Kidney Disease -Transplantation) sub-study is part of the 4C study 26 , a prospective observational study. Seven-hundreds-four pediatric CKD patients (age 6-17 years) with an eGFR below 60ml/min/1.73m 2 not yet receiving KRT were enrolled between 2009 and 2011.
Ethical aspects and details of the data acquisition were described previously 26 . The median follow-up time was 6 years with a maximum of 9 years.

Data sources/measurements
PWV was assessed annually using the oscillometric Vicorder device (SMT medical, Würzburg, Germany) as previously described 13,14 . Every 6 months blood and urine samples, anthropometrics, casual blood pressure (BP), medical history updates were J o u r n a l P r e -p r o o f obtained per standardized protocol. Laboratory measurements were performed centrally. eGFR was calculated using the Schwartz formula 27 .

Variables
Sex-and height-adjusted standardized scores (z-scores) were calculated for PWV 13 as the primary endpoint.
The following parameters were considered as covariates: eGFR decline, body mass index (BMI), BP, lipids, hemoglobin, sodium, potassium, calcium, phosphorus, bicarbonate, parathyroid hormone (PTH), uric acid, and urea. Kidney diseases were categorized as congenital anomalies of the kidney and urinary tract (CAKUT) or non-CAKUT. Supplemental Table S1 provides a more granular classification of primary renal diseases. Antihypertensive and immunosuppressive medications (including trough levels) were recorded. Systolic and diastolic BP (sex-,age-,height-adjusted) 28 z-scores, height and BMI (sex-,age-adjusted) 29 z-scores were calculated.
As ambulatory blood pressure measurements (ABPM) were only available in a subgroup of patients, we provide data for the correlation between the BP and ABPM in the Supplemental Table S2.

Time Variable
Time (years) was assessed by the following variables: time since inclusion, time pre-KRT (time since inclusion but before KRT-start), time post-transplantation (time since transplantation), time from eGFR≤30 to transplantation and time on dialysis (see Supplemental Figure S1 for more details).

Healthy Control Cohort
Longitudinal PWV measurements in 307 (girls n=145) healthy children from Rebirth Active School Study were used to assess possible sex differences in the J o u r n a l P r e -p r o o f physiological development of PWV . The study investigated cardiovascular   parameters in healthy children during a school based physical activity program 30 with   two repetitive PWV measurements with an interval of 12.7±3.3 months between 2017 and 2018.

Analyses Steps
The analyses for PWV z-score (PWVz) were performed in three analysis steps: (1) All data comprising the whole observation time. Then divided into two separate analyses according to transplantation: (2) "pre-KRT" and (3) "post-transplantation" (Figure 1).

Step 1: All Data
We included patients with at least one visit during the observation period representing the complete observation time (n=235, Figure 2). This includes data before KRT, on dialysis and after transplantation.

Step 2: Pre-KRT Data
All data before the KRT start was included to study the development of PWVz during CKD progression. Patients with at least one visit pre-KRT was included (n=230, J o u r n a l P r e -p r o o f Covariates that highly correlated to each other (BP values, lipid levels) were grouped.
If two or more covariates from the same group were eligible, the one with the better model fit (lower Akaike Information Criterion, AIC) was selected.
To assess CKD progression eGFR decline was calculated. Delta eGFR (∆eGFR) for each patient i at visit v, was calculated as the difference between eGFR at visit (v) and the previous visit (v-1) divided by the time (T) interval (years) between both visits ) (Supplemental Figure S2). In case of a missing eGFR value between two visits, ∆eGFR was interpolated.

Step 3: Post-transplantation Data
Patients with at least one visit post-transplantation were included (n=199, Figure 2). For further investigation patients with visits at pre-KRT and post-transplantation were included to assess eGFR slope pre-KRT (n=195, Figure 2). Covariates were then screened using the above pre-defined basic model to identify contributing factors.
Covariates showing significant associations with PWVz (p<0.05) and/or eliminating the association (p>0.05) between PWVz and the variable "sex" were included in the backward selection. Similar to the analysis for pre-KRT, if two or more covariates were eligible but highly correlated to each other, the one with the lower AIC was included.
We calculated individual eGFR slopes using the eGFR measurements pre-KRT to ≤30mL/min/1.73m 2 (pre-emptive) or since dialysis start (after prior dialysis).

J o u r n a l P r e -p r o o f
As the underlying kidney diseases differ between sexes, additional analyses had to be performed. Patients with at least one visit were included (n=235). Two mixed models for PWVz were performed.
(1) Mixed model adjusted for the interaction term |time since inclusion*kidney disease category| to understand whether the PWVz development differs between CAKUT and non-CAKUT patients. Corrected means and the 95% confidence intervals of PWVz adjusted for the respective models were calculated for CAKUT and non-CAKUT groups.
(2) Mixed model adjusted for the interaction term |time since inclusion*sex and kidney disease category (girls-CAKUT, girls-non-CAKUT, boys-CAKUT, boys-non-CAKUT)| to understand how sex influences the PWVz course in each kidney disease category.

General statistical analysis
Data are given as median and IQR, or absolute and relative frequencies. T-tests were performed to test differences between sexes. Complete data analyses were performed and covariates with missing>10% were not included in the covariate selection. Supplemental Figure S3 provides the number of observations over time.
The pattern of missing data accounting for the variables included in the final models is provided in Supplemental Tables S3a and S3b. Spline regression and mixed models were performed as described above. In the mixed models, patient ID and center were included as random effects to model the between-subject variation and time since inclusion as repeated effect to model within-subject variation 31 . Statistical analysis was performed using SAS 9.4 (SAS Institute, Cary, NC, USA). This manuscript was written according to the STROBE guidelines 32 .

The effect of sex on PWV pre-KRT
We analyzed 230 patients. A higher PWVz increase of 0.15 per year was shown in girls compared to boys (p=0.039; Table 2, a: basic model). One-hundred-fifty-eight patients were included in the covariate screening for final model (Suppl . Table S5).
The final model revealed that ∆eGFR was a strong predictor for PWVz in girls. An eGFR decline of -4ml/min/1.73m 2 per year pre-KRT was associated with a higher PWVz of 0.16 in girls (p=0.017) compared to boys. A Higher diastolic BP z-score and higher LDL were associated with a higher PWVz in both sexes ( Table 2, b: final model). Supplemental Figure S4 illustrates the sex difference on the effect estimate of the influencing factors on PWVz as a result from the respective model stratified by sex (Suppl. Table S6).

The effect of sex on PWV post-transplantation
We analyzed 199 patients. PWVz for girls was 0.44 higher when compared to boys  (Table 3, a: basic model; Suppl. Table S7 shows the basic model separated by the transplantation type i.e. pre-emptive and after prior dialysis).
We further analyzed 195 patients and screened for potential covariates using the basic model (Suppl . Table S8) Importantly, the association of female sex and higher PWVz persisted (p=0.01) (  Figure S5a). We then explored the potential differences between the combinations of the sex and kidney disease category (girls-CAKUT, girls-non-CAKUT, boys-CAKUT, boys-non-CAKUT). The mixed model adjusted for the interaction term |time since inclusion*category for sex and kidney disease| also showed that the development of PWVz did not differ between the four categories (Suppl. Figure S5b). This demonstrated that the higher PWVz in girls was independent of the kidney disease distribution.

Discussion
Our study characterized the evolution of vascular stiffness in girls and boys with progressing CKD and subsequent transplantation. Girls with advanced CKD showed more pronounced arterial stiffening than boys. This is in contrast to the physiological development as demonstrated in a cohort of healthy children. The faster progression of arterial stiffening in girls occurred prior to transplantation reflecting a greater vulnerability of girls' vascular system towards the magnitude as well as the duration of the exposure to an impaired renal function. Our key finding is that time acts differently on the cardiovascular burden between boys and girls during CKD.
Importantly, this was independent of the underlying kidney disease.
A greater susceptibility of females with CKD to develop arterial stiffness in conjunction with renal disease progression has not been described to date. Studies in adults so far have shown more severe arterial stiffness in women compared to men 33 and an association between arterial stiffness and eGFR decline without sex differences 34 . A tendency towards faster decline in renal function in girls compared to boys has been demonstrated especially prior to puberty 22  in mineral metabolism is FGF23 45 , which was measured only at baseline in our cohort. Postmenopausal women without estrogen substitution show higher FGF23 levels than women with estrogen substitution or men 46 . It is conceivable that the FGF23 pathway is more active in pre-pubertal girls or girls with an altered estrogen metabolism due to their uremic state. Similarly, osteoprotegerin (OPG), a cytokine that regulates bone resorption, is associated with cardiovascular events in CKD and hemodialysis patients [47][48][49] . In the CRIC study higher OPG levels were associated with an increased PWV and females had about 10 percent higher OPG levels than male CKD patients 50 , Cholesterol and its subclasses LDL and HDL are known to influence PWV and predict cardiovascular risk 51,52 . In the general population, LDL is associated with an increased risk and HDL with decreased risk. In children with CKD 53, 54 , increased HDL promotes endothelial dysfunction and is associated with vascular damage possibly due to a uremia-associated altered HDL functionality 53 . Our data showed higher HDL levels in girls; another factor that could explain the accelerated vascular damage in girls. This assumption is further supported by the performance of HDL in the model building process. HDL being in the causal pathway between progression of CKD and PWV could explain why the introduction of HDL together with the interaction term "∆eGFR/year*girls" did not reveal a significant result. LDL, however, showed an association with higher PWV in both sexes, but did not explain the sex Notably, we previously showed that girls are more susceptible to cyclosporine Aassociated hypertension than boys 25 .
Patients were allocated to dialysis or preemptive transplantation based on clinical decisions. Bias by indication was overcome by adjusting for all factors that potentially influence the treatment decision (e.g. kidney disease, center, time and kidney function parameter). As not all patients' data was available for the final model due to the timing of examinations, there was a potential selection bias. However, as there were no differences in PWV or sex distribution between patients that were or were not included in the final models (pre-KRT: inclusion n=156, non-inclusion n=74; posttransplantation: inclusion n=187, non-inclusion n=43) this should have not influenced the results of the comparison between sexes. The majority of our study population is Caucasian (88%) and so was the population from which PWV reference values were calculated 13 . This might limit the generalizability of our finding.

Conclusion
The observed higher susceptibility of girls for cardiovascular organ damage in conjunction with kidney disease progression highlights the importance of a closer attention to cardiovascular and kidney function parameters early in the disease course in female patients. Importantly, girls are more vulnerable toward eGFR decline and when exposed to a longer waiting time to transplantation. Early interventions and a faster access of girls to transplantation are crucial to tackle the sex differences in cardiovascular and mortality risk. Strict BP control and management of dyslipidemia are of importance for both sexes.

Disclosure
None. Figure S2. Assessment of changes in eGFR during pre-KRT.

Supplemental
3. Supplemental Figure S3. Number of observations over time since inclusion, differentiated by the modality of kidney replacement therapy at each visit since the inclusion. Figure S4. Effect estimates and 95% confidence interval for factors associated with PWVz during pre-KRT. Figure S5. Additional mixed model for PWVz adjusted for kidney underlying disease category. Table S1. Sub-classifications of primary kidney diseases. Table S2. Correlation between Casual BP and ABPM. Table S3. Matrix of missing data for each final model of pre-KRT (a) and post-transplantation (b). Table S4. Comparison of baseline characteristics between the study population (at first pre-KRT visit pre-KRT) and healthy children cohort (at study inclusion). Table S5. Covariates screening based on basic model for PWVz "pre-KRT". Table S6. Final models for PWVz "pre-KRT" separated by sex. Table S7. Basic mixed models for PWVz "post-transplantation" separated by transplantation type, i.e. pre-emptive transplantation and transplantation after prior dialysis. Table S8. Covariates screening based on basic model for PWVz "post-transplantation".

Supplemental
14. Supplemental Table S9. Sensitivity analysis for the effect of the pre-KRT eGFR slopes calculated from single mixed model on post-transplantation PWVz.
J o u r n a l P r e -p r o o f