Pregnancy Reduces the Accuracy of the Estimated Glomerular Filtration Rate Based on Cockroft-Gault and MDRD Formulas

Abstract

Objective: This study aims to determine the effect of pregnancy on the accuracy of 3 commonly used methods to estimate glomerular filtration rate ([GFR] creatinine clearance, the Cockroft-Gault, and modification of diet in renal disease [MDRD] formulas) using the inulin clearance as a reference. Design: Longitudinal study design. Setting: University hospital. Population: A total of 44 parous nonsmoking Caucasian women. They had a history of uneventful pregnancy (n = 9), preeclampsia (n = 27), and intrauterine fetal demise (n = 8). Methods: Measurements were performed both in pre-pregnancy and early pregnancy (8 weeks of gestation) and included inulin infusion, blood pressure, and 24-hour urinary and serum creatinine. Agreement between methods to estimate GFR was assessed by the Bland and Altman method. Main outcome measures: GFR estimated by inulin and creatinine clearances and the Cockroft-Gault and MDRD formulas. Results: During early pregnancy, the GFR measured by inulin increased 32% compared with the pre-pregnant value (from 115 ± 18 to 150 ± 23 mL/min·1.73 m⁻²), whilst the GFR measured by the indirect methods only increased 20%. The observed bias and limits of agreements are larger in early pregnancy relative to the pre-pregnant state for all 3 methods. Conclusion: The renal hyperfiltration during pregnancy decreases further the accuracy of the creatinine clearance and the Cockroft-Gault and MDRD formulas to estimate GFR.

Keywords

glomerular filtration rate pregnancy Cockroft-Gault formula MDRD formula

Introduction

Pregnancy induces marked changes in the cardiovascular and volume regulatory systems. Already in the fifth week of pregnancy, the kidneys display marked hyperfiltration. Eventually, the glomerular filtration rate (GFR) will rise approximately 60% above the pre-pregnant value. These renal changes are accompanied by systemic vasodilatation and plasma volume expansion, and these adaptations will be fully established by midpregnancy.^1,2 Several techniques are available to measure the GFR. Nonetheless, there is currently no consensus about the most suitable method to estimate GFR in pregnancy.

The gold standard for measuring GFR is the inulin clearance. However, the routine use of this method in clinical practice is costly and cumbersome. Therefore, the creatinine clearance is frequently used instead to approximate GFR. It requires a 24-hour urine collection, which is often incomplete, particularly in the outpatient setting, thus reducing the reliability of this method. Prediction equations based on single-point serum parameters are also commonly used as noninvasive estimates for GFR. The most widely used prediction equations are the Cockroft-Gault and the modification of diet in renal disease (MDRD) formulas. The Cockroft-Gault formula is based on serum creatinine, weight, age, and gender, whereas MDRD uses serum creatinine, age, gender, and race in the prediction equation.^3,4 Both equations have been validated in cohort studies using nonpregnant patients with mild-to-moderate renal insufficiency.

Two previous studies reported on the use of creatinine-based formulas to estimate GFR in pregnancy. One cross-sectional study compared the MDRD formula with inulin clearance during pregnancy and postpartum.⁵ Another study compared the Cockroft-Gault and MDRD formulas using the creatinine clearance as a reference in a group of pregnant women with preeclampsia.⁶ These studies indicated that both single-point formulas are unsuitable for estimating GFR in pregnancy.

Renal hyperfiltration, as seen in type 2 diabetes, both exaggerates the underestimation and reduces the accuracy of these methods.^7,8 We postulate that pregnancy, which is also characterized by marked renal hyperfiltration,⁹ has a similar effect on the accuracy of these methods. The present study aims to determine the impact of pregnancy on the accuracy of 3 commonly used methods to estimate GFR using inulin clearance as a reference. To this end, in a longitudinal study design, we compared the creatinine clearance and the Cockroft-Gault and MDRD formulas using the inulin clearance as a reference, both before and in early pregnancy.

Participants and Methods

Participants

We performed this longitudinal study in 44 parous women, of which 12 were multiparous. Each participant underwent 2 measurements, 1 before and 1 in early pregnancy. All women enrolled were Caucasian and nonsmokers and had participated in a previous study on hemodynamic changes in response to pregnancy in the period from 1996 till 1999.¹⁰ We used the data found in that previous study. In this study 69 women where included. We included those women with complete data on both measurements. Women with a history of preeclampsia and/or intrauterine fetal demise were recruited at the outpatient clinic at the time of their 6 weeks postpartum checkup, whereas the control participants were recruited by advertisement. In all, 9 women had a previous uneventful pregnancy, 27 participants had a history of preeclampsia, and 8 had a history of intrauterine fetal demise most likely caused by placental insufficiency. Participants were at least 5 months postpartum at the time of the first measurement. All participants gave written informed consent. The study was approved by the hospital’s medical ethical committee.

Prior to each experiment, the participant used a standardized sodium intake (100 mmol/d sodium) for 1 week. All participants consulted a dietitian prior to study. We checked the women’s compliance with the dietary regimen by quantifying the 24-hour sodium output on the day before the experiment. None of the women used vitamins, oral contraceptives, nonsteroidal anti-inflammatory drugs, or glucocorticosteroids in the 2 weeks prior to the first study session. Participants did not drink caffeine- or alcohol-containing beverages and did not eat for at least 10 hours prior to each study session. Eight women were chronic hypertensive, if applicable, we advised them to discontinue the antihypertensive drugs at least 1 week prior to the first measurement session. All measurements were performed at day 5 ± 2 and again by 8 weeks of gestation under standardized environmental conditions in a temperature-controlled room (25°C) with as little external disturbance as possible. The measurement session started at 8.00 AM, with the insertion of a 20-gauge catheter into a vein of the right forearm to assess GFR, and a similar second catheter into a vein of the contralateral forearm for the collection of blood samples. During the measurement session, patients were in supine position on a comfortable bed. The same person, trained as a biochemical analyst, executed and read the tests results.

Measurement of (GFR)

We measured GFR by the continuous infusion of inulin (Inutest; Laevosan Gessellschaft, Linz, Austria). At least 2 hours after the start of the inulin infusion, we sampled blood from the contralateral arm for the later assay of the circulating levels of inulin. Inulin was measured spectrophotometrically. The GFR was corrected for body surface area and expressed in mL/min·1.73 m⁻². On the day prior to measurement, the participants had collected a 24-hour urine sample. The box below specifies the creatinine clearance, Cockroft-Gault, and MDRD formulas. Creatinine was measured in blood samples using the Jaffe method.

\begin{array}{l} 24 - h creatinine clearance: \\ \frac{(urine creatinine / serum creatinine (μ mol/l)) \times 1 .000 .000}{1440 \times (1 .73 / body surface area)} \\ Cockroft-Gault: \frac{(140 - age) \times weight (kg) \times 1,04}{serum creatinine μ mol/l} \\ MDRD: 175 (serum creatinine (μ mol/l) \times 0 {.011312)}^{- 1 .154} \\ \times {age}^{- 0 .205} \times 0 .742 \end{array}

Statistical Analysis

Data were analyzed using SPSS version 15.0. We evaluated the data distribution using histograms. Differences between the measurements on both occasions were tested using a paired t test. Agreement between methods was assessed by the Bland and Altman method.¹¹ For each GFR estimate, we constructed a graph plotting the difference between the estimated and measured GFR against their mean. Finally, we calculated the limits of agreement by taking the mean difference ± 1.96 SD. By multiple linear regression analysis, we tested whether several independent variables are a predictor of the GFR inulin. A P value below .05 was considered significant.

Results

Table 1 lists the demographics of the 44 women enrolled in this study. In early pregnancy, serum creatinine along with the systolic and diastolic blood pressures had decreased relative to the pre-pregnant state. Meanwhile, weight and body mass index (BMI) had not changed in early pregnancy compared with the pre-pregnant state. The mean time between the 2 measurements was 7 ± 5 (range 2-18) months.

Table 1.

Clinical Characteristics in the Pre-Pregnant State and by 8 Weeks Amenorrhea (n = 44) ^a

	Pre-Pregnancy	Early Pregnancy
Age (years)	30 ± 4	30 ± 4
Weight (kg)	68.4 ± 15.1	69.1 ± 15.4
BMI (kg/m²)	24.4 ± 5.3	24.7 ± 5.4
Creatinin (μmol)	69 ± 11	58 ± 8^b
Natriuresis (mmol/24 h)	104 ± 37	116 ± 56
Systolic blood pressure (mm Hg)	117 ± 14	112 ± 11^b
Diastolic blood pressure (mm Hg)	72 ± 12	64 ± 9^b

Abbreviation: BMI, body mass index.

^a Data are listed as mean ± SD.

^b P < .05 compared with early pregnancy.

Table 2 lists GFR in pre-pregnancy and early pregnancy measured by the gold standard (inulin clearance) and the 3 indirect methods. Before pregnancy, only the Cockroft-Gault method gave an estimate for GFR closely corresponding with the inulin clearance, as opposed to an underestimation by 10% and 23% of the pre-pregnant GFR estimated by the creatinine clearance and the MDRD method, respectively. In early pregnancy, the inulin clearance had increased by 32% relative to the pre-pregnant value (from 115 ± 18 to 150 ± 23 mL/min 1.73 m⁻²) as opposed to the estimated increase of only about 20% by the other 3 methods.

Table 2.

Glomerular Fitration Rate (GFR) Measured by the Inulin Clearance (Gold Standard) and 3 Indirect Methods in the Pre-Pregnant State and by 8 Weeks Amenorrhea (n = 44)

GFR Estimates	Pre-Pregnancy	Early Pregnancy	Mean Difference (95% CI)	Relative Change (95% CI)
Inulin clearance (mL/min·1.73 m⁻²)	115 ± 18	150 ± 23^b	35 (31-40)	32% (27-36)
Creatinine clearance (mL/min·1.73 m⁻²)	105 ± 26^a	125 ± 29^a,b	20 (11-30)	24% (14-34)
Cockroft-Gault (mL/min)	116 ± 29	138 ± 32^a,b	23 (18-27)	21% (17-25)
MDRD (mL/min·1.73 m⁻²)	89 ± 14^a	108 ± 17^a,b	19 (15-23)	23% (18-28)

^a P < .05 compared with inulin clearance.

^b P < .001 compared with pre-pregnancy.

Table 3 shows for the 3 indirect methods the observed bias, the limits of agreement, the confidence limits, and the relative number of cases differing by more than 25% from the inulin clearance. In early pregnancy, both the observed bias and the limits of agreement increased for all 3 methods relative to pre-pregnancy state. During pregnancy, the degree of underestimation of the GFR has increased relative to the pre-pregnant state, while the degree of overestimation has decreased significantly.

Table 3.

Observed Bias, Limits of Agreement, and Confidence Limits of the Glomerular Filtration Rate (GFR) Measured by 3 Indirect Methods Using Inulin as Reference and The Relative Number of Cases Below the 25th Percentile or Above the 75th Percentile of the GFR Inulin

	GFR	Bias	95% Limits of Agreement	95% Confidence Limits	% Underestimation	% Overestimation
Pre-pregnancy
Inulin clearance	115 ± 18	–	–	–	–	–
Creatinine clearance	104 ± 30	12 ± 27	−41-64	−55-78	21%	7%
Cockroft-Gault	116 ± 29	−1 ± 30	−60-58	−75-74	11%	18%
MDRD	89 ± 14	26 ±15	−2-55	−10-63	39%	0%
Early pregnancy
Inulin clearance	150 ± 23	–	–	–
Creatinine clearance	125 ± 29	25 ± 25	−25-75	−37-88	34%^a	2%^a
Cockroft-Gault	138 ± 32	12 ± 36	−58-82	−77-101	23%^a	16%^a
MDRD	108 ± 17	42 ± 18	+7-77	−2-87	61%	0%

Abbreviation: MDRD, modification of diet in renal disease; % underestimation, the relative number of cases with a value more than 25% lower than the gold standard; % overestimation, the relative number of cases with a value more than 25% higher than the gold standard.

^a P < .05 compared with pre-pregnancy.

Figures 1 to 6 illustrate the Bland-Altman plots for the 3 indirect methods measured in the pre-pregnant and pregnant state, with the difference between estimated GFR and the inulin clearance on the Y axis and the mean of the estimated GFR and the inulin clearance on the X axis. All indirect methods suffered from wide limits of agreement, which increased in pregnancy relative to the pre-pregnant state. We performed a multiple regression analysis to test whether the independent variables used in both formulas are a predictor of the GFR measured by inulin. Only serum creatinine was related to the GFR inulin in contrast to weight and age (P < .01, P = .90, P = 1.00)

Discussion

This study not only provides convincing evidence for the prediction equations and creatinine clearance to underestimate the pregnancy-induced rise in GFR, these data also illustrate clearly that the inaccuracy of the indirect methods to estimate GFR in the pre-pregnant state is enhanced during pregnancy. This study compared serum creatinine-based formulas (Cockroft-Gault and MDRD formula) and creatinine clearance with the gold standard (inulin clearance) longitudinally in the pre-pregnant and pregnant state. This approach enables us to determine the impact of pregnancy on these tests' performances. Our study is in line with 2 previous cross-sectional studies,^5,6 which reported on underestimation of the GFR in pregnancy by the MDRD method. Moreover, our study confirms a previous finding that in pregnancy the Cockroft-Gault method shows higher values for GFR when compared with creatinine clearance as a reference.⁶ One other longitudinal study described the lack of accuracy of the Cockroft-Gault and MDRD formulas in healthy pregnant women during the second and third stage of pregnancy.¹² Our study approves those findings in the first trimester of pregnancy. Another study also demonstrated a lack of accuracy of the Cockroft-Gault and MDRD formulas when compared to 24 hours creatinine clearance in the third trimester of hypertensive pregnancies.¹³ This observation is in line with our results in the first trimester of both uneventful and eventually complicated pregnancies.

Both prediction equations have been validated in populations with mild-to-moderate renal insufficiency. The Cockroft-Gault formula was validated in 249 participants ranging in age from 18 to 92 years, 96% being male.³ The MDRD formula was validated in 1628 participants,⁴ mostly white, excluding participants with a GFR >70 to 80 mL/min 1.73 m⁻². In general, these formulas are less accurate in the normal to high GFR range. This was confirmed by our study, showing an inaccuracy to estimate GFR by the prediction equations before pregnancy, which magnified during pregnancy. The mean GFR in our study group was 115 ± 18 mL/min·1.73 m⁻² prior to pregnancy, increasing to 150 ± 23 mL/min·1.73 m⁻² by 8 weeks gestation. To our opinion, the hyperfiltration in pregnancy interferes with the performance of these prediction equations, as does the hyperfiltration in diabetes.^7,8,14

A limitation of this study is that it is not possible to extrapolate our results to patients with a low GFR. Particularly in that subgroup, it is clinically relevant to monitor renal function with advancing pregnancy. We designed this study to evaluate the accuracy of 3 indirect methods over a wide range of GFR values, and for this reason we included a heterogeneous group of women with previous uneventful and complicated pregnancies. Unfortunately, the GFR in most women in our study population was within the normal range.

These inferences raise the question how to evaluate renal function in pregnancy, as the prediction equations are inaccurate. It could be worthwhile to develop a specific prediction equation for GFR in pregnant women. However, the development of a new formula is difficult for a number of reasons. First, performing inulin clearance to validate a new formula in a large cohort of pregnant women is invasive, time-consuming, and expensive. Second, a prediction equation in pregnancy not only should take into account the physiological hyperfiltration that develops in pregnancy, but also the typical weight gain, which consists of both the growth of the conceptus and the accretion of maternal protein, fat, and body water. Third, the measurement of renal function in pregnancy is particularly clinically relevant for women with preexistent renal disease or preeclampsia and those women may have instead a reduced renal function and less hyperfiltration relative to normal pregnancy.

Recently, cystatin C, another serum marker for estimating GFR, has been evaluated as a serum marker to estimate renal function, especially normal renal function.^15,16 Serum cystatin C has the advantage of not being influenced by height, gender, age, and muscle mass,¹⁷ but there are other variables that affect cystatin C. Cystatin C is a positively charged molecule and its filtriation may be reduced during pregnancy because of an alteration in glomerular filtration barrier. This barrier is normally negatively charged. During pregnancy, there is a loss of this negative charge, which results in a reduced filtration of the positively charged cystatin C. This effect is likely to raise serum cystatin C levels.^{18

–21} Till now, the results in pregnancy for GFR estimated by cystatin C are conflicting.^{19
–21} It is conceivable that the accuracy of the GFR estimation in pregnancy may improve by combining serum cystatin C and creatinine in a single equation.²²

In conclusion, our data confirm 2 previous reports on the underestimation and limited accuracy of the Cockroft-Gault and MDRD formulas in estimating GFR in pregnancy. Besides, we showed longitudinally that the degree of underestimation and the measurement error increased in early pregnancy compared with the pre-pregnant state. Our study provides evidence that hyperfiltration of pregnancy decreases the accuracy of the prediction equations to estimate GFR.

Footnotes

Acknowledgments

The authors would like to thank T. Ekhart and I. Schreij for the recruitment of the participants and the collection of the data.

The authors declared no potential conflicts of interests with respect to the authorship and/or publication of this article.

The authors received no financial support for the research and/or authorship of this article.

References

Duvekot

Peeters

. Maternal cardiovascular hemodynamic adaptation to pregnancy. Obstet Gynecol Surv. 1994;49(12 suppl):S1–S14.

Duvekot

Peeters

. Renal hemodynamics and volume homeostasis in pregnancy. Obstet Gynecol Surv. 1994;49(12):830–839.

Cockcroft

Gault

. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16(1):31–41.

Levey

Bosch

Lewis

Greene

Rogers

Roth

. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med. 1999;130(6):461–470.

Smith

Moran

Ward

Davison

. Assessment of glomerular filtration rate during pregnancy using the MDRD formula. BJOG. 2008;115(1):109–112.

Alper

Webber

. Estimation of glomerular filtration rate in preeclamptic patients. Am J Perinatol. 2007;24(10):569–574.

Fontsere

Bonal

Salinas

. Is the new Mayo Clinic Quadratic equation useful for the estimation of glomerular filtration rate in type 2 diabetic patients? Diabetes Care. 2008;31(12):2265–2267.

Fontsere

Salinas

Bonal

. Are prediction equations for glomerular filtration rate useful for the long-term monitoring of type 2 diabetic patients? Nephrol Dial Transplant. 2006;21(8):2152–2158.

Sturgiss

Dunlop

Davison

. Renal haemodynamics and tubular function in human pregnancy. Baillieres Clin Obstet Gynaecol. 1994;8(2):209–234.

10.

Spaanderman

Ekhart

van Eyck

Cheriex

de Leeuw

Peeters

. Latent hemodynamic abnormalities in symptom-free women with a history of preeclampsia. Am J Obstet Gynecol. 2000;182(1 pt 1):101–107.

11.

Bland

Altman

. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8(2):135–160.

12.

Ahmed

Bentley-Lewis

Hollenberg

Graves

Seely

. A comparison of prediction equations for estimating glomerular filtration rate in pregnancy. Hypertens Pregnancy. 2009;28(3):243–255.

13.

Cote

Lam

von Dadelszen

Mattman

Magee

. Monitoring renal function in hypertensive pregnancy. Hypertens Pregnancy. 2010;29(3):318–329.

14.

Chudleigh

Dunseath

Evans

. How reliable is estimation of glomerular filtration rate at diagnosis of type 2 diabetes? Diabetes Care. 2007;30(2):300–305.

15.

Herget-Rosenthal

Bokenkamp

Hofmann

. How to estimate GFR-serum creatinine, serum cystatin C or equations? Clin Biochem. 2007;40(3-4):153–161.

16.

Toffaletti

McDonnell

. Variation of serum creatinine, cystatin C, and creatinine clearance tests in persons with normal renal function. Clin Chim Acta. 2008;395(1-2):115–119.

17.

Filler

Bokenkamp

Hofmann

Le Bricon

Martinez-Bru

Grubb

. Cystatin C as a marker of GFR—history, indications, and future research. Clin Biochem. 2005;38(1):1–8.

18.

Stevens

Schmid

Greene

. Factors other than glomerular filtration rate affect serum cystatin C levels. Kidney Int. 2009;75(6):652–660.

19.

Strevens

Wide-Swensson

Grubb

. Serum cystatin C is a better marker for preeclampsia than serum creatinine or serum urate. Scand J Clin Lab Invest. 2001;61(7):575–580.

20.

Strevens

Wide-Swensson

Torffvit

Grubb

. Serum cystatin C for assessment of glomerular filtration rate in pregnant and non-pregnant women. Indications of altered filtration process in pregnancy. Scand J Clin Lab Invest. 2002;62(2):141–147.

21.

Moodley

Gangaram

Khanyile

Ojwang

. Serum cystatin C for assessment of glomerular filtration rate in hypertensive disorders of pregnancy. Hypertens Pregnancy. 2004;23(3):309–317.

22.

Tidman

Sjostrom

Jones

. A comparison of GFR estimating formulae based upon s-cystatin C and s-creatinine and a combination of the two. Nephrol Dial Transplant. 2008;23(1):154–160.