SEARCH

SEARCH BY CITATION

Keywords:

  • Glomerular filtration rate;
  • graft function;
  • graft survival;
  • kidney;
  • kidney function

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

These analyses assessed whether creatinine based estimates of glomerular filtration rate (eGFR) accurately represent (1) graft function at different times post-transplant and (2) changes in function over time. These analyses compared iothalamate GFR to eGFR in 684 kidney allograft recipients. Changes in graft function over time (GFR slope) were measured in 360 of 459 recipients (78%) who were followed for at least 3 years. Ninety-five percent of the patients were Caucasians and 72% received kidneys from living donors. All eGFR calculations correlated significantly with GFR at all time points. However, eGFR were less precise and less accurate during the first-year post-transplant than thereafter. The average rate of GFR change (slope) was −2.93 ± 11.3%/year (−1.06 ± 5.3 mL/min/1.73m2/year). Fifty-four percent of patients had stable or positive GFR slopes. The GFR and eGFR slopes were highly correlated. However, eGFR slope, particularly when calculated by MDRD, significantly underestimated the number of patients with declining graft function. For example, 165 out of 360 patients (46%) lost GFR faster than −1 mL/min/1.73m2/year. eMDRD identified only 83 of these patients (50%) while the eMayo formula identified 134 (81%). In conclusion, eGFR correlate with GFR but they have relatively low precision and accuracy particularly early post-transplant. eGFR slopes underestimate graft functional loss although some formulas are significantly better than others for this calculation.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Accurate measurements of kidney function are critical in the follow-up of kidney transplant recipients. Changes in renal function allow the clinician to detect acute or chronic allograft injury. However, studies utilizing surveillance biopsies suggest that changes in serum creatinine may be insensitive to both of these events. For example, in a previous study up to 30% of acute rejection episodes were considered to be subclinical that is not associated with changes in serum creatinine (1). Similarly, development of allograft fibrosis and atrophy may not be accurately represented by changes in serum creatinine (2). Theoretically, more precise measures of graft function may provide more sensitive means to detect early graft injury.

The methods available to measure kidney function were evaluated extensively by a panel of experts sponsored by the National Kidney Foundation (3). Direct measures of glomerular filtration rate (GFR) are still considered as the ‘gold standard’ for assessing kidney function. However, these techniques are cumbersome and are rarely applied in routine clinical practice. An alternative approach has been to develop formulas that allow an estimate of the GFR (eGFR) based on the serum creatinine concentration. These formulas may be superior to measuring creatinine clearance in CKD populations (3) but they do not circumvent issues related to the variability in serum creatinine that is independent of changes in GFR. For example, in previous studies we showed that eGFR calculated by the MDRD formula (4) underestimates GFR in healthy individuals at least in part because the relationship between creatinine and GFR is different in health than in patients with kidney disease (5).

Previous studies assessed the accuracy of eGFR formulations in a variety of clinical circumstances including recipients of kidney transplants (6–11). Those studies assessed the relationship between GFR and eGFR at single points in time post-transplant. In contrast, in this study we used and compared GFR with eGFR to (1) measure kidney graft function at different time points after transplantation and (2) to assess changes in graft function over time. These results, based on the analysis of a large cohort of kidney transplant recipients, confirmed and expanded previous results on the relationships between GFR and eGFR: first, this relationship is weaker in transplant recipients than in patients with chronic kidney disease (CKD); secondly, there is a significant variability in the bias, precision and accuracy of eGFR at different time points after transplantation; and lastly, changes in eGFR over time significantly underestimate the number of patients losing graft function and/or the extent in the decline in graft function over time.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Patient selection

These analyses included 684 adult recipients of kidney transplants at the Mayo Clinic from January 1998 to September 2004. All of these patients had GFR measured by iothalamate clearance at 3 weeks and 1 year following transplantation. In addition, 536 of the 684 patients (78%) had their GFR measured at 2 years and 365 (53%) at 3 years. Iothalamate clearance testing at Mayo Clinic routinely includes bladder examination by ultrasound to confirm complete bladder emptying and thus ensuring complete collection of urine excretion during the 2-h test. Recipients of pancreas transplants were excluded from these analyses. Data for the analysis were extracted from electronic databases and from paper records. The extraction and the reporting of these data were approved by institutional review board. GFR was estimated (eGFR) from serum creatinine using several formulations, including the four variables MDRD formula (eMDRD) (4), the Cockroft–Gault formula (eCG)(12) and an additional formula recently described by our group (5) (GFR = Exp[1.911 + (5.249/creatinine) − (2.114/creatinine2) − 0.00686 × age − 0.205 (if female)]; if serum creatinine is less than 0.8 mg/dL, this formula uses the 0.8 value. All of the serum creatinine values used in these calculations were measured within 1 week of iothalamate GFR determination and in the same laboratory using the modified kinetic rate Jaffé reaction on an autoanalyzer (Roche-Hitachi 747; Roche Diagnostics Corp., Indianapolis, IN) that was calibrated daily.

Table 1 displays the characteristics of this patient population. The protocols used for post-transplant immunosuppression have been described in detail in previous publications (13). In brief, induction immunosuppression included thymoglobulin in 67% of patients and anti-CD25 antibodies in 14%. The remaining 19% of the patients received no induction. Seventy-six percent of patients received maintenance immunosuppression with prednisone, tacrolimus and mycophenolate mofetil. Instead of tacrolimus, sirolimus was used in 15% of patients and cyclosporine in 9%. The dose of tacrolimus was adjusted to achieve trough levels of 10–12 ng/mL during the first 4 months post-transplant and 6–8 ng/mL thereafter. Target sirolimus levels were from 15–20 ng/mL for the first 4 months post-transplant and 10–15 ng/mL thereafter. Target cyclosporine levels were 200–250 ng/mL in the first 4 months and 100–150 thereafter. All of these patients received tripmetroprim/sulfamethoxazol during the first 3 months following the transplant.

Table 1.  Patient characteristics (N = 684)
ParameterValue
Recipient age (range)50.8 ± 13.6 (18–81)
Recipient gender (% males)55.1%
Recipient race (% Caucasians)94.5%
Recipient weight (kg)79.0 ±18.1
Recipient BMI27.6 ±6.1
Dialysis pre-transplant (%)56%
Diabetes (%)18.1%
Donor age (range)41.3 ± 13.1 (6–75)
Donor gender (% males)46.2%
Donor weight (kg)82.5 ± 19.0
Donor BMI28.1 ± 5.6
Type of donor
 Living related72.3%
 Living unrelated21.4%
 Deceased6.3%
 Follow-up years (range)3.4 ± 1.6 (1.0–7.8)

Data analysis

eGFR were calculated for each GFR measurement. At each time following transplantation, eGFR was compared to GFR by assessing (1) bias of eGFR (the difference between eGFR and GFR); (2) percent bias (bias as a percentage of GFR); (3) precision between GFR and eGFR (R2) and (4) accuracy expressed as the percent of the eGFR that fell within 30% or 50% of the GFR. To assess the rate of change in graft function over time we analyzed data from 360 patients who had at least four determinations of iothalamate GFR over a 3-year period of time. Linear and loglinear regression were used to estimate the per-patient slope of the GFR and eGFR over time. These slopes were expressed as percent change per year. Data were expressed as mean and standard deviation throughout the manuscript. Correlations between GFR and eGFR measures were assessed using Pearson's correlation coefficient. Linear regression was used to identify factors associated with eGFR bias and bias in eGFR slope.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

Using the accepted scoring of chronic kidney disease (CKD) (3), at 1-year post-transplant 38% of kidney allograft recipients had a GFR >60 mL/min/1.73 m2 (CKD stage 1–2); 53% had a GFR between 30 and 59 mL/min/1.73 m2 (CKD stage 3); and 9% had a GFR <30 mL/min/1.73 m2 (CKD stage 4–5). Of interest, 3% of these patients had GFR >90 mL/min/1.73 m2.

Performance of eGFR formulas at different times post-transplant

Table 2 displays GFR and eGFR average values, calculated with different formulas at different times after transplantation. The eMDRD underestimated GFR at both 3 weeks and 1-year post-transplant (p < 0.001, Wilcoxon). In contrast, after the first-year eMDRD overestimated the GFR (p = 0.004). The eCG and the eMayo significantly overestimated the GFR at all time points (p < 0.001, Wilcoxon) (see Table 2). The relationship between GFR and eMDRD is displayed graphically in Figure 1. It should be noted that the eMDRD primarily underestimated GFR in recipients with higher GFR levels. For example, among transplant recipients with a 1-year GFR ≥60 mL/min/1.73 m2 (N = 268) the mean GFR was 73 ± 13 and the eMDRD 61 ± 17 (p < 0.0001, paired Wilcoxon). In contrast, among patients with GFR <60 mL/min/1.73 m2 (N = 416) the mean GFR was 43 ± 11 and the eMDRD 45 ± 13 (p = 0.001, paired Wilcoxon). In contrast, the eMayo significantly overestimated GFR but equally at all levels of GFR (data not shown).

Table 2.  Serum creatinine, GFR and different estimates of GFR at different times post-transplant
VariableTime post-transplant
3 weeks1 year2 years3 years
  1. The number of patients included at each time point is indicated in parentheses.

  2. 1Values represent means ± standard deviation of the mean (number of determinations).

  3. 2p < 0.0001 compared to GFR (paired Wilcoxon).

Serum creatinine1.51 + 0.4 (684)11.52 + 0.4 (684)1.56 + 0.5 (593)1.57 + 0.5 (459)
GFR55.8 ± 18 (684) 54.7 ± 18 (684)53.6 ± 18 (536)52.3 ± 20 (365)
MDRD52.2 ± 16 (683)251.5 ± 15 (684)55.4 ± 18 (506)55.6 ± 19(346) 
CG64.9 ± 20 (679)265.8 ± 20 (684)265.4 ± 23 (506)266.6 ± 25 (346)2
Mayo63 ± 22 (683)262 ± 21 (684)2 59.3 ± 21 (506)259.5 ± 22 (346)2
image

Figure 1. Relationship between GFR estimated by the MDRD formula (X-axis) and by iothalamate GFR (Y-axis). Figure displays data collected at 3 weeks (A), 1 year (B) and 2 years (C) after the transplant.

Download figure to PowerPoint

Table 3 displays characteristics of the relationship between GFR and eGFR at different time points after transplantation. In this table, only those patients who had GFR measured at all time points are included to allow paired comparisons (N = 360). The precision and accuracy of eGFR, calculated by all formulas, were statistically weaker at 3 weeks post-transplant than at later time points (data not shown). It should be noted also that these parameters were significantly better for eMDRD and eMayo than for eCG. The bias of eMDRD and the direction of the bias varied at different times post-transplant. During the first year there was a small negative bias (that is the eMDRD underestimated GFR) while the bias was positive after the first year (that is the eMDRD overestimated GFR). In contrast, the bias for the eMayo and the eCG was relatively constant at all time points indicating that these two formulas overestimated GFR (Table 3).

Table 3.  Comparison between measured GFR and GFR estimated by different formulations
GFR vs.TimePrecision (R2)Bias% BiasAccuracy (within 30%)Accuracy (within 50%)
  1. Only those patients with four determinations of GFR over three or more years (3 weeks, 1, 2, 3+ years, N = 360) are included in this analysis.

MDRD3 weeks0.51−4.33−1.7661.784.7
1 year0.69−5.04−4.8572.290.6
2 years0.620.806.7164.983.4
3 years0.712.4912.9665.282.7
C-G3 weeks0.428.6622.955.676.7
1 year0.5310.1924.655.072.5
2 years0.4911.6427.8652.170.9
3 years0.5313.5835.7048.665.8
Mayo3 weeks0.547.0218.0455.877.2
1 year0.696.3914.6368.986.7
2 years0.674.6611.9865.882.7
3 years0.746.9718.4562.081.8

Measuring changes in graft function over time

Out of the 459 patients who had a functioning allograft for at least 3 years, 360 (78%) had at least 4 GFR measurements and these patients were selected for these analyses. Compared to patients not selected for these analyses, these 360 patients did not differ significantly in recipient or donor demographic characteristics (age, sex, race and BMI), percent of living kidney donors or GFR at 3 weeks and 1 year (data not shown). The average change in GFR was −2.93 ± 11.3%/year (or −1.06 ± 5.29 mL/min/1.73 m2/year). One hundred and ninety-four patients (54%) had a GFR slope greater (more positive) than −1 mL/min/1.73 m2, the expected rate of loss in kidney function due only to aging (14). That is, in these patients the GFR was considered to be stable or improved over the period of observation. In contrast, 166 patients (46%) lost GFR at a rate faster than −1 mL/min/1.73 m2/year.

As can be seen in Table 4, compared to the GFR slope, the slopes of eGFR were statistically different and the direction and/or the magnitude of the change in graft function differed among the different formulas. Thus, in contrast to the GFR slope, the eMDRD slope was positive, suggesting that there is an average gain in graft function in the population. The eCG and eMayo slopes were concordant with the GFR slope indicating an overall loss in graft function. However, the eCG and the eMayo slopes indicated a significantly smaller rate of change in graft function than the GFR slope. Because the relationships between eGFR and GFR were particularly weak at 3 weeks post-transplant (see Table 3) we re-calculated GFR and eGFR slopes starting at 1-year post-transplant (Table 4). Compared to the slopes calculated using all time points, the GFR slope after 1 year was significantly more negative (p = 0.002, Wilcoxon) but there were no significant changes in the eMDRD, eCG or eMayo slopes. Thus, the relationships between GFR and eGFR slopes did not change significantly by this second calculation.

Table 4.  Change in graft function calculated as slope of GFR or slope of eGFR calculated by different formulations (N = 360)
VariableGFR slope (%/year)GFR slope (mL/min/year)
  1. In the first part of this table the slopes were calculated using all data points, starting at 3 weeks post-transplant. In the second part the slopes were calculated starting at 1-year post-transplant.

  2. 1p < 0.0001 vs. the GFR slope by paired Wilcoxon.

  3. 2p = 0.002 vs. the GFR slope by paired Wilcoxon.

  4. 3p = 0.018 vs. the GFR slope by paired Wilcoxon.

 Mean ± SDMedianMean ± SDMedian
 Slopes calculated from 3 weeks to the end of follow-up
GFR−2.93 ± 11.3−1.22−1.06 ± 5.29−0.61
eMDRD2.29 ± 9.0612.121.43 ± 4.5211.06
eCG−0.38 ± 8.7410.070.31 ± 5.5510.07
eMayo−1.56 ± 10.42−4.19−0.66 ± 5.353−0.04
 Slopes calculated from 1 year to the end of follow-up
GFR−3.92 ± 13.1−1.90−1.31± 5.88−0.95
eMDRD2.17 ± 10.3212.201.53 ± 5.0611.05
eCG−1.07 ± 10.721−0.82−0.07 ± 6.431−0.45
eMayo−1.53 ± 11.811−0.55−0.43 ± 5.851−0.31

Figure 2 (top) displays graphically the relationship between GFR and eMDRD slopes. There was a statistically significant relationship between these two parameters (r = 0.613; p < 0.0001). However, as can be seen in this figure there is significant scattering of the data between the GFR and eMDRD slopes. Among patients losing GFR at a rate faster than −1 mL/min/1.73m2/year (N = 165, 46%) only 83 (50%) were identified correctly by eMDRD slope as losing graft function. To study further this relationship in Figure 2 (bottom) we classified patients in three groups according to their GFR slope: Group 1 included recipients losing GFR at a rate faster than −10%/year (N = 69, 19%); Group 2 including recipients with GFR slopes between −10 to +10%/year (N = 268, 74%) and Group 3 including recipients with GFR slopes greater than 10%/year (N = 23, 6.4%). In Group 1 patients (Figure 2, bottom) the GFR slope was −20.3 ± 10.5%/year (range −10.1% to −56.9%) and the eMDRD slope was significantly higher, −5.8 ± 10.6%/year (p < 0.0001). In 27% of these patients, the eMDRD and the GFR slopes were comparable that is less than −10%/year. This subgroup included those patients with the most rapid loss of GFR (mean −32 + 12%/year, range −14 to −57). In contrast, in 73% of Group 1 patients the eMDRD slope underestimated the rate of loss in graft function. In Group 2 patients, that is those with GFR slopes between −10 and +10%/year, the mean GFR slope was −0.09 ± 4.84 and the mean eMDRD slope was significantly higher at 3.62 ± 6.6%/year (p < 0.0001). Finally, in Group 3 patients the GFR slope (16.3 ± 4.8%/year) and the eMDRD slope (11.3 ± 12.2) differed only marginally (p = 0.06).

imageimage

Figure 2. Top: Relationship between GFR slope (Y-axis) and eMDRD slope (X-axis) in 360 kidney transplant recipients. Slopes data are expressed as %/year. Males recipients (+) and female recipients (•) are identified. Bottom: eMDRD slopes in patients classified into three groups according to their GFR slopes indicated in the X-axis. Boxed areas indicate that range of GFR slopes in each of the groups.

Among the 360 patients included in these analyses 21 (6%) lost their allograft, not due to patient death, after a period of follow-up of 67 + 18 months. Of those 21 patients, 15 (71%) had a GFR slope faster than −10%/year during the first 3 years post-transplant. In contrast, only 8 of the 21 (38%) had an eMDRD slope of similar magnitude.

There is also a significant relationship between GFR and eMayo slopes (r= 0.730, p < 0.0001) (Figure 3, top). Among patients losing GFR at a rate faster than −1 mL/min/1.73m2/year (N = 165, 46%), 134 (81%) were

imageimage

Figure 3. Top: Relationship between GFR slope (Y-axis) and eMayo slope (X-axis) in 360 kidney transplant recipients. Slopes are expressed as %/year. Males recipients (+) and female recipients (•) are identified. Bottom: Slope of eMayo in patients classified into three groups according to their GFR slopes indicated in the X-axis. Boxed areas indicate that range of GFR slopes in each of the three groups.

identified correctly by eMayo slope as losing graft function. Figure 3 (bottom) displays the eMayo slopes of patients classified in the three groups described previously. In Group 1 patients, that is losing GFR at a rate faster than −10%/year, the mean GFR slope −20.3 ± 10.5%/year and the eMayo slope was −13.5 ± 12%/year (p < 0.0001, paired Wilcoxon). In 65% of these patients the eMayo and the GFR slopes were comparable. However, 35% of the patients were misclassified by the eMayo slope as having lost less function than indicated by the GFR slope. In Group 2 patients the mean GFR slope was −0.09 ± 4.84%/year and the mean eMayo slope was 0.5 ± 6.6 (p = 0.06). Finally, in Group 3 patients the GFR slope was 16.3 ± 4.8%/year and the eMayo slope 9.9 ± 12.1 (p = 0.05).

We next assessed whether the bias of the eGFR slopes (eGFR slope – GFR slope) was related to particular recipient characteristics. In the case of eMDRD, the bias of the slope related only to the sex of the recipient such that the bias was greater in females than in male recipients (p < 0.0001). Figure 2 displays graphically the differences in the relationship between GFR and eMDRD slopes in males and females recipients. In female recipients the eMDRD significantly overestimated the GFR slope (GFR slope −4.1 ± 12; eMDRD slope 5.4 ± 10.2%/year, p < 0.0001). In male recipients these differences were smaller but statistically significant (GFR slope −1.95 ± 10.6; eMDRD slope −0.22 ± 7.1, p = 0.001). The relationship between recipient sex and slope in the eMDRD bias was noted when those slopes were calculated using all time points or only time points after 1 year. Other donor or recipient demographic characteristics did not relate significantly with the eMDRD slope bias. In contrast to these findings there were no significant correlates of the relationship between GFR and eMayo slopes. In particular, as can be appreciated in Figure 3 (top) the sex of the recipient did not correlate with the bias in the slope of the eMayo.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

These results, based on a large cohort of kidney recipients, confirm and expand previous studies on the relationships between GFR and eGFR: first, this relationship is weaker in transplant recipients than in patients with chronic kidney disease (CKD); secondly, there is a significant variability in the bias, precision and accuracy of eGFR at different time points after transplantation and lastly, changes in eGFR over time significantly underestimate the number of patients losing graft function and/or the extent in the decline in graft function over time.

These analyses allowed us to quantitate kidney allograft function and changes on that function over time, using direct GFR measurements, in a large cohort of kidney allograft recipients. One year following transplantation 38% of these patients achieved GFR levels greater than 60 mL/min/1.73m2 and 3% greater 90 mL/min/1.73m2. In addition, in more than half of these recipients (54%), the GFR level remained stable during a period of observation of at least 3 years. These data, not previously described, are indeed encouraging for the future of these recipients and of kidney transplantation in general although admittedly the period of observation is relatively short.

In agreement with previous studies (6–11) these analyses indicate that the relationship between eGFR, measured by the MDRD formula, and GFR in kidney transplant recipients is significant but statistically weaker than in patients with native chronic kidney disease (4). Perhaps this is not unexpected because most eGFR formulas were developed in patients with native CKD and not in kidney transplant recipients. In contrast, the Nankivell eGFR formula (15) was developed in transplant recipients. However, this formula performed poorly in these studies and those results are not reported here. Among the available formulas the MDRD was derived in patients with advanced CKD (4) while the Mayo formula was derived from a population that included normal as well as CKD individuals (5). Likely for this reason, the MDRD underestimates particularly higher levels of allograft GFR which are more likely to be present early after transplantation. The relationship between GFR level and MDRD bias has been noted in previous studies (5,16–18). Other factors may contribute to the poor performance of eGFR formulas in kidney transplant recipients. For example, these formulas include correction factors for age, body size and sex that may not be appropriate for transplant recipients where these parameters may differ from the recipient to the donor. Consistent with the relevance of these demographic variables estimating GFR with Cystatin C requires a different formulation for patients with native CKD and transplant recipients (19). Another possible explanation for the poor performance of eGFR is that the relationship between serum creatinine and GFR is different in transplant patients than in patients with native CKD. For example, recent studies showed that muscle mass is reduced in patients on dialysis and that this parameter changes little after transplantation (20). Low production of creatinine may explain the overestimation of GFR by creatinine-based eGFR formulas in transplant recipients. The profound changes that occur in the recipient (e.g. weight gain) and the allograft during the first-year post-transplant may also explain why the accuracy and precision of all eGFR are weaker during the first year than thereafter.

It should be noted that the average differences between eGFR and GFR (bias) at different time points post-transplant are generally not large (see Table 2). However, these biases are variable giving some uncertainty about the validity of assessing graft function in the clinic by eGFR alone in individual patients. In addition, eGFR bias, over time, have a rather large effects on eGFR slope. The relationship between bias and slopes is represented graphically in Figure 4. This figure represents an idealized recipient with a starting GFR of 55 mL/min/1.73m2 and losing GFR at −10%/year. In the figure, at each time point, the GFR value was modified according to the average bias of the eMDRD and the eMayo formulas that is shown in Table 2. As can be seen, compared to the GFR slope the eMDRD slope is significantly flatter due to the underestimation of GFR early post-transplant and the overestimation later on. This indeed gives the false impression that this patient is losing graft function at a slower rate than indicated by the GFR. Because the eMayo formula overestimates GFR at all time points, the line describing the changes in eMayo is shifted upwards but the slope of that line is similar to the GFR slope. The clinical implications of these biases in eGFR slopes are large. The eMDRD slope fails to identify half of the patients who are losing graft function at a faster than expected rate and it also underestimates the rate of functional loss. Furthermore, this underestimation of graft function loss have implications for the identification of patients at risk for subsequent graft failure. For example, in 71% of the patients graft failure, after the first 3 years, was preceded by a GFR slope faster than −10%/year. In contrast, only 38% of the patients who subsequently lost their allograft had a negative eMDRD slope of similar magnitude. The eMayo slope identifies most patients losing GFR (81%) but also underestimates, albeit to a lesser degree than the eMDRD, the rate of GFR decline. The CG formula was not included in the slope analysis because eCG was clearly inferior at all time points to the other formulas used here.

image

Figure 4. Consequences of eGFR bias on slopes. The GFR line (•––•) is formed starting with the mean GFR at 3 weeks and then decreasing the GFR at −10%/year. The eMDRD (□––□) and the eMayo lines (▴… . ▴) were plotted modifying the value of the GFR at each time by the eGFR bias calculated in Table 2.

Download figure to PowerPoint

The bias in eGFR slopes imply that at least in some kidney transplant recipients the serum creatinine increases less than expected when the GFR declines. Physiologically, this can be explained by a reduced production of creatinine after transplantation and/or by increased secretion of creatinine by renal tubules and/or bowel. These two possibilities could be addressed at least in part by repeated measurements of urinary creatinine excretion. Unfortunately, that information is not available to us since 24-h creatinine clearance is not used routinely in our program.

There are several limitations of these analyses. First, the study population was almost uniformly Caucasian thus this analysis may not be applicable to recipients of other races. This is an important general principle to be considered with any eGFR formulas that generally are derived from patient's cohorts with particular characteristics and strictly speaking should be applied only to patients with similar characteristics. For example, eGFR formulas derived in Caucasians and African Americans, that is, MDRD, do not perform well in South Asians (21). Secondly, at 3 weeks post-transplant patients received tripmetroprim-sulfamethoxazol that may have affected serum creatinine values and eGFR but not GFR. These results are not consistent with a significant effect of this drug because at 3 weeks the magnitude and direction of the eGFR bias was different for all formulas. In addition, the bias for the eMayo and the eCG formulas were similar in magnitude and direction when patients were receiving tripmetroprim-sulfamethoxazol and after it was discontinued. We should also consider that the selection of a subgroup of patients for the slope studies may have introduced a bias in the results by excluding patients transplanted in the most recent past and patients who lost their graft in less than 3 years. However, this is unlikely because the characteristics of the selected patients did not differ significantly from those of the cohort. Finally, these analyses were based on measures of serum creatinine that were not calibrated to a reference laboratory. Measurements in serum creatinine are not well standardized among different laboratories. This lack of measurement ‘calibration’ clearly can influence the accuracy of different eGFR formulations when applied in different laboratories (22). This calibration issue may be in part responsible for the ‘systematic’ overestimation of GFR by serum creatinine noted here. However, calibration cannot explain differences in eGFR slopes because all creatinines were measured in the same laboratory and thus subjected to constant mis-calibration. Furthermore, in previous studies we pointed out that calibration issues cannot explain this phenomenon completely particularly in patients with higher levels of kidney function, including healthy individuals (5).

Periodic accurate assessment of graft function is the cornerstone in the management of kidney allografts. These analyses showed serious pitfalls in this assessment when based on serum creatinine. In most patients differences between eGFR and GFR are not large and likely to be not clinically relevant. However, in some patients that bias is large and could be misleading clinically. For example, at 3 weeks post-transplant 17% of recipients had an eGFR bias greater than +20 mL/min/1.73m2. These data also showed that assessing changes in graft function by serum creatinine and/or eGFR is quite problematic. eGFR underestimate both the number of recipients losing function and the magnitude of the loss in GFR. Potentially, this miscalculation may lead to delayed diagnosis of allograft pathology and inappropriate management of progressive chronic kidney disease (23). These results are in agreement with a previous study showing the limitations of eGFR for assessing changes in kidney allograft GFR in clinical studies (11). Based on these results, it is recommended that direct measurements of GFR should be done in all transplant recipients at some point post-transplant to assess, in each particular patient, the bias in creatinine-based GFR estimates. Indeed, it would be best to repeat the GFR measurement periodically to assess the relationships between GFR and estimated GFR which may vary over time due to changes in muscle mass/creatinine production. However, it is recognized that these tests are difficult to obtain in most clinical practices. The search for alternative methods to measure GFR revealed that serial measures of Cystatin C may provide accurate assessment of changes in kidney function in patients with native kidney disease due to diabetes (24) or polycystic kidney disease (25). Cystatin C relates well to GFR in kidney transplant recipients (19) but the ability of the method to detect changes in function in the renal allograft over time needs to be analyzed.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References

This work was supported by grants from the Mayo Clinic Transplant Center and by NIH grant DK062410-02S1. We thank the kidney-pancreas coordinators in our program for their dedication and their help in the collection of data from these patients. We thank Ms. Cynthia Handberg for excellent secretarial assistance.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgments
  8. References
  • 1
    Rush DN, Henry SF, Jeffery JR, Schroeder TJ, Gough J. Histological findings in early routine biopsies of stable renal allograft recipients. Transplantation 1994; 57: 208211.
  • 2
    Solez K, Vincenti F, Filo RS. Histopathologic findings from 2-year protocol biopsies from a U.S. multicenter kidney transplant trial comparing tacrolimus versus cyclosporine. Transplantation 1998; 66: 17361740.
  • 3
    K/DOQI clinical practice guidelines for chronic kidney disease: Evaluation, classification, and stratification. Am J Kidney Dis 2002; 39: S1S266.
  • 4
    Levey AS, Greene T, Schluchter MD et al. Glomerular filtration rate measurements in clinical trials. Modification of Diet in Renal Disease Study Group and the Diabetes Control and Complications Trial Research Group. J Am Soc Nephrol 1993; 4: 11591171.
  • 5
    Rule AD, Larson TS, Bergstralh EJ, Slezak JM, Jacobsen SJ, Cosio FG. Using serum creatinine to estimate glomerular filtration rate: Accuracy in good health and in chronic kidney disease. Ann Intern Med 2004; 141: 929937.
  • 6
    Poge U, Gerhardt T, Palmedo H, Klehr HU, Sauerbruch T, Woitas RP. MDRD equations for estimation of GFR in renal transplant recipients. Am J Transplant 2005; 5: 13061311.
  • 7
    Perico N, Gaspari F, Remuzzi G. Assessing renal function by GFR prediction equations in kidney transplantation. Am J Transplant 2005; 5: 11751176.
  • 8
    White C, Akbari A, Hussain N et al. Estimating glomerular filtration rate in kidney transplantation: A comparison between serum creatinine and cystatin C-based methods. J Am Soc Nephrol 2005; 16: 37633770.
  • 9
    Poggio ED, Wang X, Weinstein DM et al. Assessing glomerular filtration rate by estimation equations in kidney transplant recipients. Am J Transplant 2006; 6: 100108.
  • 10
    Ibrahim HN, Rogers T, Tello A, Matas A. The performance of three serum creatinine-based formulas in estimating GFR in former kidney donors. Am J Transplant 2006; 6: 14791485.
  • 11
    Mariat C, Alamartine E, Barthelemy JC et al. Assessing renal graft function in clinical trials: Can tests predicting glomerular filtration rate substitute for a reference method? Kidney Int 2004; 65: 289297.
  • 12
    Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976; 16: 3141.
  • 13
    Cosio FG, Grande JP, Wadei H, Larson TS, Griffin MD, Stegall MD. Predicting subsequent decline in kidney allograft function from early surveillance biopsies. Am J Transplant 2005; 5: 24642472.
  • 14
    Lindeman RD, Tobin J, Shock NW. Longitudinal studies on the rate of decline in renal function with age. J Am Geriatr Soc 1985; 33: 278285.
  • 15
    Nankivell BJ, Gruenewald SM, Allen RDM, Chapman JR. Predicting glomerular filtration rate after kidney transplantation. Transplantation 1995; 59: 16831689.
  • 16
    Lin J, Knight EL, Hogan ML, Singh AK. A comparison of prediction equations for estimating glomerular filtration rate in adults without kidney disease. J Am Soc Nephrol 2003; 14: 25732580.
  • 17
    Pierrat A, Gravier E, Saunders C et al. Predicting GFR in children and adults: A comparison of the Cockcroft-Gault, Schwartz, and modification of diet in renal disease formulas. Kidney Int 2003; 64: 14251436.
  • 18
    Vervoort G, Willems HL, Wetzels JF. Assessment of glomerular filtration rate in healthy subjects and normoalbuminuric diabetic patients: Validity of a new (MDRD) prediction equation. Nephrol Dial Transplant 2002; 17: 19091913.
  • 19
    Rule AD, Bergstralh EJ, Slezak JM, Bergert J, Larson TS. Glomerular filtration rate estimated by cystatin C among different clinical presentations. Kidney Int 2006; 69: 399405.
  • 20
    Van Den Ham EC, Kooman JP, Schols AM et al. Similarities in skeletal muscle strength and exercise capacity between renal transplant and hemodialysis patients. Am J Transplant 2005; 5: 19571965.
  • 21
    Jafar TH, Schmid CH, Levey AS. Serum creatinine as marker of kidney function in South Asians: A study of reduced GFR in adults in Pakistan. J Am Soc Nephrol 2005; 16: 14131419.
  • 22
    Coresh J, Astor BC, McQuillan G et al. Calibration and random variation of the serum creatinine assay as critical elements of using equations to estimate glomerular filtration rate. Am J Kidney Dis 2002; 39: 920929.
  • 23
    Gill JS, Abichandani R, Kausz AT, Pereira BJ. Mortality after kidney transplant failure: The impact of non-immunologic factors. Kidney Int 2002; 62: 18751883.
  • 24
    Perkins BA, Nelson RG, Ostrander BE et al. Detection of renal function decline in patients with diabetes and normal or elevated GFR by serial measurements of serum cystatin C concentration: Results of a 4-year follow-up study. J Am Soc Nephrol 2005; 16: 14041412.
  • 25
    Rule AD, Torres VE, Chapman AB et al. Comparison of methods for determining renal function decline in early autosomal dominant polycystic kidney disease: The consortium of radiologic imaging studies of polycystic kidney disease cohort. J Am Soc Nephrol 2006; 17: 854862.