Evaluation of ¹²³I-orthoiodohippurate single kidney clearance rate by renal sequential scintigraphy in a large cohort of likely normal subjects aged between 0 and 18 years

Abstract

Purpose

Age-related values of ¹²³I-orthoiodohippurate (OIH) single kidney clearance rate (Cl) were estimated in a large cohort of likely normal children aged between 0 and 18 years.

Methods

Among 4,111 children examined in the past 10 years, 917 were selected with the following inclusion criteria: (a) mild ultrasonographic hydronephrosis with right differential renal function (DRF) <53% and >47% (498 pts), (b) known or suspected urinary tract infection with normal ultrasound, serum creatinine and DMSA and DRF <53% and >47% (419 pts). ¹²³I-OIH-Cl was assessed using a validated gamma camera method. Children were divided into 21 age classes: from 0 to 2 years, eight 3-month classes; from 2 to 14 years, twelve 1-year classes; from 14 to 18 years, one 4-year class.

Results

Cl, plotted against age, was fitted using an increasing function ($y = a - be - cx$). Mean ¹²³I-OIH-Cl of 1,834 kidneys was 306±22 ml/min/1.73 m² BSA. Mean ¹²³I-OIH-Cl of the right and left kidneys was 307±23 and 305±22 ml/min/1.73 m² BSA, respectively (p<0.002). The best-fitting ¹²³I-OIH-Cl growing function was: Cl=311−230e−0.69×Age (months). ¹²³I-OIH-Cl improved progressively starting from birth, reaching 96% and 98% of the mature value at 1 and 1.5 years, respectively. ¹²³I-OIH-Cl at birth (age=0) was 81 ml/min/1.73 m² BSA. After 18.6 days of life, the renal function had doubled its starting value, and it reached a plateau of 311 ml/min/1.73 m² BSA at 2 years.

Conclusion

This work represents a systematic evaluation of ERPF by a gamma camera method in a large cohort of selected likely normal paediatric subjects.

Appendices

Appendix 1

Cross-calibration procedures; acquisition protocol and start delay

• 1. Syringe dose

  The dose was prepared in a 5 ml syringe and the contained volume was brought to 2 ml with saline. Before starting the examination, the syringe was measured by the gamma camera, after it had been positioned on a specially prepared 5-cm-thick plexiglas support placed on the examination table, the latter being in close contact with the collimator surface.

  The choice of a plexiglas support as a soft tissue-like attenuation material was made to:

  • a) Reduce the count rate of the dose in a range comparable to the kidney count rate and so avoid inaccuracy due to possible lack of counting linearity of the instrumentation

  • b) Expose the gamma camera to an emission spectrum with a Compton component comparable to the Compton emission of the patient.

  The choice of a 5-cm distance from the acquisition surface was made on the basis of the physiological range of the distance of the kidney from the posterior abdominal wall in the previously evaluated population [24].

  The activity in the syringe (S) was measured by means of a 50-s static image, using a 64×64 matrix with a 20% window centered over the 159-keV ¹²³I energy peak. The injected dose was calculated using a 10% automatic threshold region of interest (ROI) on the syringe static image. Thereafter, to compare the dose activity with the accumulated activity in the kidney under conditions of the same efficiency of measurement, the measured activity was corrected for the syringe ROI dimensions and for the thickness of the plexiglas support. For this purpose the algorithm reported in subsection 2 was used.

• 2. Camera depth and spatial response kidney ROI

  One lateral view of the kidney (3-min static image, 64×64 matrix) was acquired at the end of the dynamic scan using a ⁵⁷Co marker support placed in line with the acquisition surface, with the aim of determining the kidney depth of a single patient. The distance between the middle of the renal image and the collimator surface was measured in pixels, multiplied by the pixel length (6 mm) and expressed in centimetres.

  The influence on the system counting efficiency produced by kidney surface dimensions and depth was previously determined using phantoms (¹³¹I, ¹²³I and ^99mTc sources of different diameters positioned at increasing depth in a tissue-like attenuation medium, i.e. water) and appropriate correction factors were calculated and introduced into the computing algorithm.

  In the processing of the single patient study, the kidney surface dimensions were evaluated from the number of pixels included in the renal ROIs. Hence, for each patient the appropriate correction factor was automatically selected by the computer by choosing from amongst those previously determined with phantom studies on the basis of the following correction formula:

  $$E{\left( {d{\text{,}}s} \right)} = e^{{{\left( {{\text{ $ -\mu $ }}d} \right)}}} s^{{{\left( {{\text{0}}{\text{.06 + 0}}{\text{.0057}}d} \right)}}} $$

  where E=system efficiency, μ=soft tissue linear attenuation coefficient for energy of the employed isotope, d=kidney depth in cm and s=kidney area in pixels [3].

  These data were in agreement with the Siegel results in computing depth and source size effect on cardiac volume determination in MUGA [25].

• 3. Acquisition protocol and start delay

  Patient data were acquired by a dynamic acquisition (80×5 s frames, followed by 65×20 s frames), beginning 20 s before the radiotracer injection (bolus) to allow the determination of time zero of both input function (It₀) (precordial curve) and nephrographic curves (Rt₀). Tracer arrival delay (Δt) was estimate as Rt₀−It₀ by means of the Slope method [26].

Appendix 2

Kinetic assumptions and clearance calculation

• 1. Kinetic assumptions

  ¹²³I-OIH-Cl was then determined using a method and a proprietary program previously validated in our institution, based on time-activity curves generated from the heart and kidney areas using the ROI technique [3]. The kidney background was subtracted taking into account the intravascular radioactivity included in the renal ROI, which was significant in the first 2 min after injection, i.e. before any appreciable amount of cleared radioactivity left the kidney, according to the following formula valid only within about 0–2 min:

  $$R{\left( t \right)} = gP{\left( t \right)} + h{\int\limits_0^t {P{\left( t \right)}dt} }$$

  (1)

  where R(t) is the renal time-activity curve, P(t) is the precordial curve, g is a constant related to kidney ROI background volume, and h is the coefficient used to normalise the integral of the precordial curve to the renogram.

  Hence, h included the kidney influx constant (K _i) and geometrical efficiency of measurement of both precordial and kidney ROIs. In this context, gP(t) represented the background contribution to the renogram R(t) in the early minutes, and $h{\int\limits_0^t {P{\left( t \right)}dt} }$ was the activity accumulated in the kidney at time t.

  This theoretical approach, known as the Patlak-Rutland plot [27, 28], was empirically proposed in our institution as long ago as 1973 by Meldolesi et al. [29] and was subsequently validated for gamma camera study by Formiconi et al. [24].

  This postulate is valid if the input function is considered in the kidney; thus corrections have to be made for heart–kidney time delay and bolus dispersion in order to quantify renal uptake. The time delay and the deformation of the radioactive bolus during the transit from the heart to the kidney were taken into account, as also suggested by Meyer for cerebral blood flow measurements in dynamic PET [30]. To this end we simultaneously shifted by the arrival time difference, Δt, the precordial time-activity curve along the time axis and convolved it with a random walk function [W(t)] [31]. The choice of this dispersion function is justified by the fact that the function can be linearised and expressed by the two parameters A and K [31], representing the time peak delay and the asymmetry of the function, respectively. Thus, Eq. (1) becomes:

  $$R{\left( t \right)} = gP{\left( t \right)} * W{\left( t \right)}dt + h{\int\limits_0^t {P{\left( t \right)} * W{\left( t \right)}dt.} }$$

  (2)

  After subtracting the convolved background gP(t)*W(t)dt from R(t), $h{\int\limits_0^t {P{\left( t \right)} * W{\left( t \right)}dt} }$ represented renal uptake at time t.

  Hence parameters h, g, A and K were calculated by the least square method on the basis of the best fitting of the second term of Eq. 2 with respect to R(t) between time zero (Rt₀) and the first two minutes of the nephrographic curve, excluding the first 30 s (five early frames) to reduce the inaccuracy of parameter (h, g, A, K) values closely influenced by the high extra renal activity and statistical bias of renographic curves at this early stage. The choice of the first two minutes as the fitting time was also underlined by Piepsz et al. [32]: “When using Patlak-Rutland methodology, it is advisable to restrict the fitting procedure to the second minute of the test and check visually that this fixed time interval gives rise to a slope that is well adapted to the plot”. Moreover, “the final point of the fit should never exceed the time of the peak of the renogram minus one minute (T _max−1), which is always less than 5 min”. Another factor which may contribute to impairing the accuracy of the plot after two minutes is tracer escape into the extravascular space [32]; this may modify the shape of the precordial curve, making it not strictly representative of the temporal behaviour of the plasmatic concentration. Thus, the obtained integral of the precordial curve might not be proportional to the renal uptake. Our recent reports about dynamic SPECT RSS with temporal sampling of 0.76 s for nephrographic curves and a revolution time of 13 s agree with this hypothesis and highlight the fact that the K _i to the kidney is already stable at 30 s. Furthermore, the g(t) background volume component is substantially stable from the 30th to the 60th second [33, 34].

• 2. Clearance calculation

  The fraction of tracer accumulated in the kidney during the first minute following administration (D ₁) can be calculated by dividing the renal uptake by the injected dose (S) [3]. The former (i.e. renal uptake) is expressed as the integral of the convolved precordial curve and is corrected for kidney size and depth:

  $$D_{1} = \frac{{{\int\limits_0^t {P{\left( t \right)} * W{\left( t \right)}dt} }}} {S}$$

  (3)

  The decision to consider the value of the integral of the convolved precordial curve during the first minute was based on the best accordance with clearance reference values [3, 24, 29] calculated by the Sapirstein method [4]. The method was later validated in two different groups of 25 patients by means of calculation of the ERPF according to Sapirstein et al. [4]: blood samples were taken from the opposite arm to that used for tracer injection at 2, 4, 6, 8, 10, 17, 24, 34, 44 and 54 min after ¹²³I-OIH administration, during and after the end of the gamma-camera study [3]. The results of the sample counting were interpolated on the concentration curve using two-exponential fitting. Hence, ERPF (left and right ¹²³I-OIH-Cl sum) was calculated according to the known formula:

  $${\text{ERPF}} = \frac{{I\alpha \beta }}{{\alpha \beta + \beta A}}$$

  where I is the injected dose, A and B are the y intercepts and α and β are the respective exponential coefficients. The values obtained were normalised to 1.73 m² body surface area.

  Normalised ERPF was closely linearly related (r=0.97, p<0.00001) to the sum of D ₁ of each kidney (D _1s). The linear regression equation was: ${\text{ERPF}} = 10 + 3, \kern-2pt 182\;D_{1} s$. Thus D ₁ multiplied by the obtained proportionality factor (i.e. 3,182) provided the single kidney ¹²³I-OIH-Cl [3].