GFR ≤25 Years Postdonation in Donors with (vs without) a 1^st Degree Relative with ESRD

Abstract

An increased risk of ESRD has been reported for living kidney donors, and appears to be higher for those donating to a relative. The reasons for this are not clear. One possibility is that ESRD may be due the nephrectomy-related reduction in GFR, followed by an age-related decline that may be more rapid in related donors. Between 1/1/1990 – 12/31/2014, we did 2002 living donor nephrectomies. We compared long-term postdonation eGFR trajectory for donors with (n=1245) versus without (n=757) a first-degree relative with ESRD. Linear mixed-effects models were used to model the longitudinal trajectory of eGFR. With all other variables held constant, there was a steady average increase in eGFR until donors reached age 70; 1.12 (95% CI: 0.92–1.32) ml/min/year between 6 weeks–5 years postdonation; 0.24 (0.00–0.49) ml/min/year between 5–10 years; and 0.07 (−0.10–+0.25) ml/min/year between 10–20 years for donors with attained age less than 70. After age 70, eGFR declined. After adjustment of predonation factors, the difference in eGFR slopes between related and unrelated donors was 0.20mL/min/1.75m²/year (0.07–0.33) Our data suggests that: postdonation, kidney donor eGFR increases each year for a number of years and that eGFR trajectory does not explain an increase in ESRD after donation.

Introduction

A living kidney donor loses about 50 percent of kidney function with unilateral nephrectomy. The remaining kidney undergoes compensatory hypertrophy, and within 6 weeks of donor nephrectomy, glomerular filtration rate (GFR) returns to approximately 70% of preoperative values. It has been thought that, in the short-term, there may be an additional small GFR improvement, followed by stabilization. However, during middle age, some individuals experience a slow, but steady, decline in GFR. If kidney function is normal at the beginning of this decline, it rarely leads to development of end-stage renal disease (ESRD). It was long believed, based on long-term clinical observations, comparing donors to the general population, that living kidney donors had sufficient renal function to withstand this decline in GFR, and to live a normal life without increased risk of ESRD (1–4). However, 2 recent studies, comparing donors with matched healthy controls, have suggested that donors have a slightly increased lifetime risk of ESRD (5, 6). In these studies, the majority of ESRD was seen in those donating to a relative. Confounding this observation is that, in the absence of donation, relatives of those with ESRD are at increased risk of developing ESRD (7–13). Additionally, in the first few decades of clinical transplantation, most donors were related to their recipient, so that, in general, related donors have longer follow-up than unrelated donors (14).

An understanding of the pathogenesis of, and risk factors for, ESRD in donors is important for donor selection and counseling. For some, nephrectomy might not leave sufficient reserve for the “normal” age-related decline in GFR. It is also possible that donors might experience a more rapid decline in kidney function than observed in nondonors. A third possibility is that donors experience the same steady decline in GFR as seen in nondonors, but that some develop new-onset kidney or systemic disease that accelerates the decline in GFR (15–17).

Determining whether (postdonation) normal age-related GFR decline, alone, explains the increased risk of ESRD is important for counseling potential donors; if so, a higher GFR threshold at evaluation could be considered, especially in younger donors. It would also be important to determine whether donors having a family member with ESRD had a faster decline in eGFR compared to those without. If the increased risk is limited to those having a relative with ESRD, then those without a relative with ESRD could be counseled that they may not be at increased risk.

In this study, we analyzed estimated GFR (eGFR) over time in 2002 white kidney donors to determine: 1) whether the typical decline in eGFR over time can explain the increased rate of ESRD in donors, and 2) whether slope differs between those donating to a first-degree relative with ESRD (or having a first degree relative with ESRD) compared to donors without a first-degree relative with ESRD.

Materials & Methods

We have a long-standing, IRB-approved study of long-term kidney donor outcomes (IRB HSC# 0301M39762). Our protocols have been described in detail (2). Briefly, donors are contacted every 3 years (1/3 of the cohort each year) and asked to provide an updated medical and psychosocial history, including development of new conditions, and to send us any intervening laboratory results. Donors not having recent laboratory results are asked to undergo a routine health checkup, including laboratory tests. All information is entered into our IRB-approved database.

Study Cohort and outcomes

Because donors at our institution are primarily white, and because ESRD rates differ by ethnicity, we limited our analysis to whites (18). From January 1, 1990 through December 31, 2014, we did 2002 living donor nephrectomies in whites for whom we have serum creatinine information more than 6 weeks postdonation. During this interval, we made no changes to our medical candidacy criteria except that we now consider white donors, ≥ 55 years of age, with hypertension that is well-controlled with a single drug (19). Follow-up on all patients continued until June 30, 2016.

To estimate the effect of family history of ESRD on eGFR, we categorized donors into those with a first-degree relative - parent, sibling, or child - with ESRD (i.e., either donating to a relative with ESRD or, by medical history, having a first-degree relative with ESRD, n = 1245) and those without a first-degree relative with ESRD (n = 757) (640 unrelated; 117 distant relatives).

The primary outcome of our study was eGFR, estimated using the CKD-EPI equation (20). Measurements of eGFR within 6 weeks of donation were excluded from the analysis because the goal was to model long-term trends in kidney function and not transient changes after donation.

Statistical analyses

Continuous covariates were summarized as median (25^th, 75^th percentile) and binary characteristics were summarized as n (%) by first-degree family history of ESRD. To test univariable differences between donors with and without a first-degree family history of ESRD, we used the Wilcoxon rank sum tests (continuous covariates) and Pearson-chi square tests (categorical covariates). To model the effect of family history on the longitudinal trajectory eGFR, we used linear mixed-effects models.

Specifically, we fit a model with fixed effects for donation year, gender, age at donation, smoking status at donation, predonation eGFR, predonation body mass index, family history of ESRD, time since donation, and the interaction between family history and time since donation. Considering the post-donation compensatory increase in GFR, we allowed the eGFR slope to change at 5, 10, and 20 years postdonation. This linear spline basis allows for sufficient flexibility in the post-donation trajectory while remaining more interpretable than other basis expansions. The knot points at 5, 10, and 20 years were chosen based on modeling the eGFR trajectory using a natural (cubic) splines (see Appendix for sensitivity analysis). To account for correlation among repeated measurements for the same donor, we included a random subject-specific intercept and random slopes. Because age-related decline in kidney function begins in middle age and may be more pronounced for older patients, we allowed eGFR slope to change at 70 years of age. Donation year was included to account for secular trends in creatinine determination, postdonation care, donor selection, and donor characteristics not accounted for in the model. Of key interest in this analysis is the interaction between family history and time since donation as that would indicate whether or not donors with a first-degree relative with ESRD experience a different eGFR slope.

The model was created using information from living white donors between January 1, 1990–December 31, 2014 because during that interval we had increased our routine donor surveillance. In addition, we repeated the analysis using donors of all races donating since 1990 (n = 2125) and our entire white donor population donating since 1963 (n = 3151). Additional details of the model and model building process are provided in the Supplementary Appendix.

All analyses were performed using SAS Version 9.4 (SAS System, Cary, NC) or R Version 3.3.1 (R Foundation for Statistical Computing, Vienna, Austria). All statistical tests were two-sided tests with p < 0.05 indicating statistical significance. Code to reproduce the analyses is available as a Github repository at http://github.com/docvock.

Results

Donor characteristics

Donors with a first-degree relative with ESRD were younger (p <0.001), donated during an earlier era (p <0.001), were more likely to be male (p =0.057), and be smokers at the time of donation (p < 0.001) (Table 1). Also, donors with a first-degree relative with ESRD had longer follow-up (median 10.0 years versus 6.1 years, p < 0.001) but similar number of eGFR measurements postdonation (median, 4 versus 3 measurements). Other characteristics were similar between groups.

Table 1.

Characteristics of donors at donation and number of eGFR measurements. Continuous covariates are summarized as median (25^th, 75^th percentile) and categorical covariates are summarized as N(%).

Covariate	No 1st Degree Family History of ESRD N = 757	1st Degree Family History of ESRD N = 1245	p-valuea
Donation Age (Years)	45.3 (37.7, 51.9)	41.2 (33.8, 49.3)	<0.001
Female	479 (63.3)	733 (58.9)	0.057
Predonation eGFRb (mL/min/1.73 m2)	90.4 (80.2, 102)	91.6 (80.3, 102.6)	0.283
Predonation creatinine (mg/dL)	0.86 (0.77, 1.0)	0.9 (0.8, 1.0)	0.012
Predonation BMI (kg/m2)	26 (23, 29)	26 (24, 29)	0.877
Predonation glucose (mg/dL)	91 (86, 97)	92 (86, 99)	0.053
Predonation hemoglobin (g/L)	14.0 (13.2, 14.9)	14.0 (13.1, 14.9)	0.549
Predonation systolic blood pressure (mmHg)	122 (114, 131)	120 (112, 131)	0.010
Predonation smoker	120 (15.9)	294 (23.7)	<0.001
Year of donation	2005 (2001, 2009)	2000 (1995, 2006)	<0.001
Follow-upc length (years)	6.1 (2.1, 11)	10.0 (2.8, 15.9)	<0.001
Follow-upc > 1 year	648 (85.6)	1079 (86.7)	0.545
Number of eGFR measurements	4 (2, 6)	3 (2, 6)	0.023

^aComparison between donors with and without first degree family history of ESRD using Pearson chi-square test for categorical covariates and Wilcoxon rank sum test for continuous covariates.

^bGFR, estimated using the CKD-EPI formula using serum creatinine

^cFollow-up is defined as the time from donation until the last serum creatinine measure

Longitudinal model of postdonation eGFR

For several years post-donation, the typical donor had increasing or stable eGFR (Figure 1) (The coefficient estimates from the longitudinal model of postdonation eGFR are shown in Table 2). For donors with a first-degree relative with ESRD who had not yet reached 70 years of age, the average yearly change in eGFR between 6 weeks and 5 years postdonation was 1.12 mL/min/1.73 m² (95% CI: 0.92–1.32); between 5–10 years postdonation, 0.24mL/min/1.73 m² (0.00–0.49); between 10 –20 years post-donation, 0.07 mL/min/1.73 m² (−0.10 − +0.25); and for after 20 years postdonation, −0.19 mL/min/1.73 m² (−0.75 – +0.37).

Figure 1.

Estimated average adjusted eGFR (mL/min/1.73 m²) trajectory after 1 year post-donation by family history of ESRD (p=0.003 for comparison of curves by family history).

Table 2a.

Coefficient estimates from the linear mixed model of longitudinal measures of eGFR^a post-donation.

Effect	Estimate	SE	p-value
Age at donation (per decade before 35 years)	−4.20	0.67	<0.001
Age at donation (per decade after 35 years)	−2.08	0.29	<0.001
Female (ref: male) gender	1.44	0.41	<0.001
Donation eGFRa (per 10 mL/min/1.73 m2)	4.72	0.15	<0.001
Donation BMI (per 5 kg/m2)	−0.67	0.25	0.008
Smoker at donation (ref: non or former smoker)	1.43	0.52	0.006
No family Hx of ESRD (ref: family Hx of ESRD)	−0.49	0.51	0.34
Donation year (per 1 year)	0.15	0.04	<0.001
Slope (per year, 6 wks −5 yrs post-donation, Age < 70 years, Family Hx of ESRD)	1.12	0.10	<0.001
Slope (per year, 5–10 years post-donation, Age < 70 years, Family Hx of ESRD)	0.24	0.12	0.049
Slope (per year, 10–20 years post-donation, Age < 70 years, Family Hx of ESRD)	0.07	0.09	0.42
Slope (per year, >20 years post-donation, Age <70 years, Family Hx of ESRD)	−0.19	0.28	0.50
ΔSlope: Age > 70 years versus Age < 70 years	−0.38	0.15	0.010
ΔSlope: No Family Hx of ESRD versus Family Hx of ESRD	0.20	0.07	0.003

^aGFR, estimated using the CKD-EPI formula using serum creatinine

Impact of having a first-degree relative with ESRD

The average eGFR trajectory for donors with and without a first-degree relative with ESRD at different ages at donation (holding all other factors constant) is shown in Figure 2a. Figure 2b shows the standard error for the estimated mean eGFR at various time points. For each group eGFR continued to increase for a number of years (~15 years in the younger donors) before stabilizing. After adjusting for other patient factors, donors with a first-degree relative with ESRD experienced either a smaller increase or larger decline in eGFR -- on average, 0.20 mL/min/1.73m²/year (0.07 – 0.33) more -- than those without a first degree relative with ESRD. For example, a 55 year-old donor at 5-years postdonation, and without a first-degree relative with ESRD, could expect a change in eGFR over the next year of + 0.44 mL/min/1.73 m^2; a similar donor who had a first-degree relative with ESRD, +0.24 mL/min/1.73 m². At 20 years postdonation the same unrelated donor could expect a change of −0.37 mL/min/1.73m²; related, −0.57 mL/min/1.73 m².

Figure 2.

(A) Estimated average eGFR (mL/min/1.73 m²) trajectory after 1 year post-donation by age at donation (p<0.001) and family history of ESRD (p=0.003). All other factors were held constant (pre-donation eGFR 90 mL/min/1.73 m², female gender, non-smoker, pre-donation BMI 25 kg/m², donation year 1995). (B) Standard error estimates for the estimated mean GFR in figure 2a at various follow-up points.

In our additional analyses of the cohort of donors of all races, and of the entire white cohort (since 1963), the findings were similar to those modeled above (Tables S1–S4 and Figure S1–S2 in the Supplementary Appendix).

Impact of age at donation and attained age

Even after controlling for the effects of predonation eGFR, we found a statistically significant association between older donation age and decreased postdonation eGFR (Figure 2a). For example, a 55-year-old donor could expect postdonation eGFR on average to be 8.34mL/min/1.73 m² lower than that of a 25 year-old donor if other characteristics were held constant (95% CI 6.97–9.72; p < 0.001). Once donors reached age 70, they had an average yearly decline in eGFR of 0.38 mL/min/1.73 m² (95% CI 0.09–0.67) more than those <70 years old.

Other factors

Women, more recent donation, and a lower predonation BMI were all associated with increased eGFR postdonation (p <0.01 for all covariates).

Discussion

ESRD is a rare event after living kidney donation. A decline in GFR, which precedes ESRD is much more common; therefore, focusing on eGFR allows for greater power to detect differences by family history of ESRD. Our data suggest that the normal age-related decline in GFR is not responsible for an increased risk of postdonation ESRD. Given the average rate of decline in GFR over time for donors with and without a first-degree relative with ESRD, it would take a prolonged time to develop a GFR <30 mL/min/1.73 m² or ESRD. We did find a small difference in the postdonation eGFR slope between donors with and without a first-degree relative with ESRD. However, for each group, the typical change in eGFR over time was not sufficient to explain an increased rate of postdonation ESRD. Our findings are consistent with those of Kido et al. who demonstrated that donors who developed ESRD (n=8) had a relatively stable postdonation GFR until a new event resulted in an abrupt decline in GFR over time (21). In the donor cohort we describe herein (i.e., our donors since 1990), only 3 donors have developed ESRD. However, in our entire donor cohort since 1963, a total of 39 donors have developed ESRD. Again, similar to the study et al Kido et al, we found that for these 39, serum creatinine levels (eGFR) remained stable until the development of new disease (22).

Steiner et al. argue that nondonors with normal (for age) renal function who develop new-onset disease later in life (e.g., type 2 diabetes) might have a disease-related decline in kidney function; but with sufficient renal reserve that decline would not necessarily result in ESRD (15–17). However, the same decline in donors with less reserve might result in ESRD. Support for this argument comes from a recent report of data from the Scientific Registry of Transplant Recipients (SRTR) showing that there is an increased cumulative incidence of postdonation ESRD due to either diabetes or hypertension (23). Similarly, in our experience (since 1963), development of postdonation diabetes or hypertension doubled the risk of eGFR <30 ml/min/1.73 m² or ESRD as compared to matched donor controls (24).

Our findings suggest that eGFR trends postdonation might not be able to be extrapolated from population data of nondonors as advocated. First, irrespective of age at donation, eGFR on average steadily increased for the subsequent few years (Figures 1 and 2a). After that (maximum follow-up, 25.8 years), there was a slower continuing increase in eGFR for younger donors; but for older donors (e.g., ≥55 years old), GFR stabilized and then declined once the donor reached 70 years of age. An increasing eGFR in the first few years postdonation has previously been reported (25–30). In a study with short follow-up, Cho et al., using serial 99 mTc-diethylenetriaminepentaacetic acid scans over 2 years postdonation, found increasing GFR (25). In a prospective study of donors and matched controls, Kasiske et al. reported that, over a period of 3 years, the mean (SD) slope of the measured GFR (mGFR) was +1.47 (5.02) mL/min/year in donors, but was −0.36 ( 7.55) mL/min/year in controls, (p = 0.005) (26). Similarly, Lenihan et al., studying postdonation glomerular dynamics, reported increased plasma flow and GFR over 6 years (27). Although also finding an increase in measured GFR for the first 5 years after donation, Courbebaisse et al found that age and body mass index at donation were inversely correlated with the change in GFR (28). We found a similar association between both age at donation and BMI at donation, with postdonation eGFR. (Figure 2a).

Two studies with longer follow-up reported increasing GFR later postdonation. Saran et al. measured GFR in 47 donors at 2 intervals, 10 years apart, postdonation (29). At the time of the first measurement, the donors were 1.4 to 20.7 years postdonation; at the 2nd, 12 to 31 years postdonation. Between the 2 measurements GFR increased at a mean (SD) of 5.97 (17.55) mL/min/ 1.73m² ; median increase, 4 mL/min/ 1.73m². Of the 47 donors, 41 (87%) had a normal (for age) GFR. Fehrman-Ekholm et al. reported on a cross-sectional study of 573 donors who returned to clinic for mGFR determinations (mean time since donation – 15 years; range 2 to 43 years) (30). They found that for a 30-year-old donor, the median GFR increased for the first 17 years, remained stable ~8 years, and then began a slow decline. In contrast, for a 50 year-old donor, the median GFR increased for ~15 years and then slowly fell. Combined with our data, this information suggests that GFR increases for many years postdonation, and that once age-related fall in GFR begins, donors have more reserve than population data might suggest. As a consequence, a higher GFR threshold in younger donors might not be necessary.

Our analyses were limited to white donors from 1990 through 2014. Although our donors before 1990 do have longer follow-up, surveillance was less frequent during that era so only a limited number of serum creatinine levels were documented in the database. As a sensitivity analysis, we reanalyzed our data using all white donors from 1963 through 2014 (see Supplementary Appendix 1). We found no substantive differences in the results between the study cohorts. Importantly, we occasionally performed unrelated donor nephrectomy in the first decades of clinical kidney transplantation; however, it only became common after the mid-1980s (14). In our dataset, related donors were more likely to have donated earlier in the history of our program. Due to this imbalance, we included donation year in our statistical model, thereby accounting for any secular trends in donor care or donor selection. Familial-related development of postdonation hypertension or type 2 diabetes might have accounted for the difference in slope between related and unrelated donors. We found no postdonation differences between groups in incidence of these diagnoses (Tables S5–S6 in the Supplemental Appendix). Of note, since the vast majority of our donor population is white, we limited our analyses to whites. In the general population as well as the donor population, blacks and Hispanics have increased rates of ESRD as compared to whites (18). As a sensitivity analysis, we reanalyzed our data using donors from all racial groups (see Supplementary Appendix 2). We found no significant differences in the post-donation eGFR trajectory between whites and nonwhites, but our observations need to be validated in larger nonwhite donor populations. In our analyses, we compared donors with and without a known first degree relative with ESRD. Of the 757 donors without a first- degree relative with ESRD, the majority 640 (84%) were unrelated. The subgroup of 117 who donated to distant relatives was too small for independent analysis. Removing that subgroup from the analysis does not affect the results. We found a smaller annual increase in GFR for related (vs unrelated) donors (and subsequently a slightly greater decline). Although our data suggests that age-related fall in GFR is unlikely a cause of ESRD in either group, related donors do have less functional reserve should new onset disease develop, In addition, slightly lower GFR may have other clinical impact. For example, there is an inverse relationship between GFT and cardiovascular risk.

In summary, for donors with and without a first-degree relative with ESRD, we found a steady increase in GFR for the first few years postdonation. Although there was a small difference between the slope of GFR in donors with (versus without) a first-degree relative with ESRD, neither group experienced a decline in GFR that would explain an increased incidence of ESRD. Our study supports the possibility that any increased risk of postdonation ESRD is due to new-onset disease. It also emphasizes the importance of donor selection, and of counseling potential donors about risks of new-onset disease that might affect their long-term kidney function, particularly diabetes and hypertension.

Supplementary Material

Supp AppendixS1-5

Figure S1: Estimated average eGFR (mL/min/1.73 m²) trajectory by age at donation and family history of ESRD. All other factors were held constant (pre-donation eGFR 90 mL/min/1.73 m², female gender, non-smoker, pre-donation BMI 25 kg/m², donation year 1995).

Figure S2: Estimated average eGFR (mL/min/1.73 m²) trajectory by age at donation and family history of ESRD. All other factors were held constant (pre-donation eGFR 90 mL/min/1.73 m², female gender, non-smoker, white race, pre-donation BMI 25 kg/m², donation year 1995).

Figure S3. Estimated average eGFR (mL/min/1.73 m²) trajectory after 1 year post-donation by age at donation and family history of ESRD using natural (restricted cubic) splines to model the longitudinal trajectory and using data from white donors donating between 1963 and 2014. All other factors were held constant (pre-donation eGFR 90 mL/min/1.73 m², female gender, non-smoker, pre-donation BMI 25 kg/m², donation year 1995).

Table S1: Characteristics of donors at donation and number of eGFR measurements using our entire white donor population donating since 1963. Continuous covariates are summarized as median (25^th, 75^th percentile) and categorical covariates are summarized as N(%).

Table S2: Coefficient estimates from linear mixed model of longitudinal measures of eGFR^a post-donation using our entire white donor population donating since 1963.

Table S3: Characteristics of donors at donation and number of eGFR measurements using our entire donor population donating since 1990 (all races). Continuous covariates are summarized as median (25^th, 75^th percentile) and categorical covariates are summarized as N(%).

Table S4: Coefficient estimates from linear mixed model of longitudinal measures of eGFR^a post-donation using data from the donor population (all races) donating between 1990 and 2014.

Table S5. Estimates of Cox proportional hazards of model of time to diagnosis of diabetes mellitus post-donation.

Table S6. Estimates of Cox proportional hazards of model of time to diagnosis of hypertension post-donation.

Table S7: Coefficient estimates from linear mixed model of longitudinal measures of eGFR^a post-donation using natural (restricted cubic) splines to model the longitudinal trajectory and using data from the white donor population donating between 1963 and 2014.

Appendix 1: Details of Linear Mixed Effects Model of eGFR Trajectory

Appendix 2: eGFR Trajectory using our entire white donor population donating since 1963

Appendix 3: eGFR Trajectory Using All Races

Appendix 4: Postdonation diabetes and hypertension in those with versus without a family history of ESRD

Appendix 5: Choosing the Functional Form for eGFR Trajectory

Table 2b.

Estimated average yearly change in eGFR by time since donation, current age of donor, and family history of ESRD from the linear mixed model of longitudinal measures of eGFR^a post-donation.

		Family History ESRD		No Family History ESRD
	Time Since Donation	Estimated Slope	SE	Estimated Slope	SE
Current Age < 70 years	6 wks -5 yrs post-donation	1.12	0.10	1.32	0.11
	5–10 years post-donation	0.24	0.12	0.44	0.13
	10–20 years post-donation	0.07	0.09	0.27	0.11
	>20 years post-donation	−0.19	0.28	0.01	0.29
Current Age > 70 years	6 wks -5 yrs post-donation	0.74	0.18	0.94	0.18
	5–10 years post-donation	−0.14	0.19	0.06	0.19
	10–20 years post-donation	−0.31	0.16	−0.11	0.17
	>20 years post-donation	−0.57	0.31	−0.37	0.31

Abbreviations

ESRD

end stage renal disease

eGFR

estimated glomerular filtration rate