Can glomerular filtration rate blood sample acquisitions be truncated reliably?

Introduction: Iodine-125 iothalamate is frequently used to measure glomerular filtration rate (GFR). Current protocols involve patient blood sampling over several hours. Our objective of this retrospective investigation was to determine if GFR calculations using only 2 blood samples agreed with the standard protocol requiring 6 samples.

Methods: Twenty-three GFR studies were performed on 14 pts [9 males & 5 females aged 1-73 years (mean = 17±22 years; median = 5 years)] with a variety of malignancies, prior to &/or after administration of nephrotoxic drugs. Eight pts had 1 study, 3 pts had 2 studies & 3 pts had 3 studies, with a maximum of 3 months between first & last studies. After intravenous administration of 0.9-2.0 MBq (25-30 µCi) I-125 sodium iothalamate, 6 plasma samples were obtained at approximately 5 min, 10 min, 15 min, 3.0 hr, 3.5 hr & 4.0 hr after injection, then counted in a calibrated well counter. Four different methods of calculating GFR were applied to each set of blood sample measurements. Method 1: A widely available algorithm ("Cleveland Clinic Plasma Disappearance Protocol") implemented a published approach (Pediatr Nephrol, 1996;10(1),25-8) to compute plasma clearance rate (PCR) & GFR values from the 6 plasma counts. The algorithm performs a bi-exponential fit to a 2-compartment model & assumes a non-equilibrium early phase component followed by an equilibrium late phase reflecting renal clearance of tracer. Method 2: ("Denver GFR v2.02e") also performed a bi-exponential fit applied to the 6 plasma counts. Method 3: A variant 2-sample algorithm available from the University of Denver that uses somewhat diffferent curve modeling methods to perform a mono-exponential fit to the equilibrium late phase data using only the 3.0 hr & 4.0 hr counts. Method 4: The same algorithm as for Method 3 but applied to only the 3.5 hr & 4.0 hr counts. For all 4 Methods, pt height & weight were used to estimate body surface area, which was used to index GFR’s. Statistical analyses were performed of PCR, GFR & biological half-life of the late phase (T_1/2) calculations of the 4 Methods.

Results: PCR values were not normally distributed (range: 20 – 164 ml/min, Shapiro-Wilk p = .01), nor were T_1/2 values (range: 29 – 365 min, p < .0001). GFR values were normally distributed (range: 68 - 170 ml/min/1.73 m², p = .07). ANOVA demonstrated no significant differences among the 4 methods for GFR, PCR or T_1/2 values (Table). Among the 23 studies, Method 2 agreed exactly with Method 1 on all 3 studies with abnormally low GFR (<74 ml/min) & all 20 cases with normal GFR (κ = 1.00 for "very good agreement" (Biometrics. 1977;33:159–74)). The 2-sample methods (3 & 4) had "moderate agreement" & "good agreement" respectively with the 6-sample methods (1 & 2) (Table). Two studies that were abnormal by Methods 1 & 2 were normal by Method 3, while 2 studies that were normal by Methods 1 & 2 were abnormal by Method 4.

Conclusions: While average measurements among the 4 methods were statistically similar, detection of abnormally reduced GFR can be compromised by use of only 2 blood samples versus use of all 6 blood samples.

• Download figure
• Open in new tab
• Download powerpoint