Pediatric Nephrology
Pediatr Nephrol (1992) 6:323-327 9 IPNA 1992
Original article A reappraisal of the measurement of glomerular filtration rate in pre-term infants Barry H. Wilkins* Department of Child Health, Bristol University and Neonatal Intensive Care Unit, Southmead Hospital, Bristol, UK Received April 25, 1991; received in revised form December 23, 1991; accepted January 15, 1992
Abstract. Thirty-seven single-injection polyfructosan-S (PF-S, Inutest| and 98 continuous-infusion PF-S/creatinine clearance studies were performed in 39 sick very low birth weight infants. The single-injection clearance method for measuring glomernlar filtration rate has been shown to be a reliable technique if sampling is continued for 8 h or more and the PF-S (Inutest| assay is sensitive, accurate and precise. The continuous-infusion clearance method is also valid if the infusion is continued for more than 24 ta and preceded by a loading dose in the form of a double-rate infusion for 8 h. Creatinine clearance is usually less than PF-S clearance, the mean ratio being 0.91, suggesting that there is some creatinine reabsorption in the renal tubule in sick very low birth weight infants. Key words: Glomerular filtration rate - Pre-term infant Newborn - Inulin - Polyfructosan-S - Creatinine - Clearance
Introduction Glomernlar filtration rate (GFR) has been measured in newborn infants using either inulin as an exogenous marker or endogenous creatinine, and using either standard clearances or indirect methods which avoid urine collection. The single-injection and continuous-infusion methods should give accurate measurements of GFR if inulin distributes only in extracellular fluid, is freely filtered, not reabsorbed, not metabolised and excreted by no other route. This is so in the adult, where inulin has long been regarded as the best marker of GFR, and it has been shown [1, 2] that inulin and polyfructosan-S (PF-S, Inutest| (Laevosan-Gesellschaft, Linz, Austria), a polyfructoside now used as an inulin substitute, are freely filtered in pre-
*Present address: Neonatal Unit, Rush Green Hospital, Romford,
Essex, UK
term newborn infants. Using inulin without urine collection by the constant-infusion technique [3-6] requires that the equilibration of the marker throughout the extracellular fluid space is complete. This has been shown not to be true when the inulin is infused for only 2 or 3 h [3, 4], and the technique is only valid with a longer infusion time. For this reason the continuous-infusion method cannot be used in the first I or 2 post-natal days. However, the single-injection technique has been found to overestimate the GFR in the newborn period, even when sampling is continued for 5 h after the injection [3, 7, 8]. The sickest and most immature infants are liable to have a lower GFR and higher extracellular fluid volume, especially when oedematous, which makes the single-injection and continuous-infusion methods less reliable. The inherent difficulties of urine collection preclude standard techniques. The purpose of the present investigation was to re-evaluate the single-injection method and compare it with the continuous-infusion method using PF-S (Inutest| and also to investigate whether creatinine clearance is an accurate measure of GFR in such pre-term infants. Laboratory methods for measuring polyfrnctoside and creatinine were first improved so that they were applicable to small samples with increased precision and accuracy. The study had approval from the local medical ethics committee and parental consent was obtained.
Patients and methods This study was part of a wider study of renal glomemlar and tubular function in very low birth weight infants. Thirty-seven single-injection and 98 continuous-infusion measurements of PF-S clearance were made in 39 infants, between the ages of 0.5 and 33 days, Gestation at birth was between 25.5 and 33 weeks and birth weight between 720 and 2,000 g. No patient had a progressive rise in plasma creatinine as evidence of acute renal failure either before or during the studies which were only performed when i.v. therapy was being given. No subject was studied within 24 h of receiving occasional doses of fmsemide, aminophylline, indomethacin, digoxin or vancomycin, and volume replacement was never given during single-injection studies or within 12 h of sampling in continuous-infusion experiments. Diuretics were not routinely used. All
324 close to the end of the cannula at approximately 10, 40, 80, 120, 240, 360 and 480 min after the injection; in some cases sampling was longer. Two millilitres of blood was always removed before, and replaced after, sampling.
8.07.5 R
7.0 e-
Single injection followed by constant infusion. A single-injection GFR
._o
and a constant-infusion GFR cannot be obtained at the same time in the same subject so the two methods cannot be compared simultaneously. On 16 occasions in 16 babies, the single-injection experiment was immediately followed by a continuous-infusion experiment, the double-infusion rate being continued for 4 h in this case. It was thus possible to compare the two techniques 2 4 - 3 6 h apart.
t-0
~: 6.0 0 0
5.5 It. D..
'- 5.0
-J
~.5
Analysis of single-injection PF-S disappearance curves. PF-S concentraJ
i
t
i
/
i
100
200
300 Time (rain)
t, O0
500
600
Fig. 1. Actual experimental points from 2 single-injection experiments in 2 patients to which two-exponential (lower) and three-exponential (upper) curves have been fitted by manual curve-stripping. There were insufficient early points in the lowergraph to derive an intermediate third exponential which also explains why the third point does not lie on the curve. The 95% confidence interval for polyfructosan-S (PF-S, Inutest| measurements in duplicate is approximately the height of the symbols at all levels babies received ethamsylate, a vascular stabilising drug which has no known renal effects. One baby was subject to intrauterine steroid exposure. The infants were evenly distributed about the 50th weight per-centile according to the intrauterine growth charts of Keen and Pearse [9]. Six subjects did not have any respiratory distress, 16 had severe respiratory distress syndrome. Only 2 did not receive assisted ventilation.
Constant-infusion method. A 25 mg/ml solution of PF-S (Inutest| in water was infused i.v. through a 0.22-1xm anti-bacterial filter, using a digital electronic syringe pump (Graseby MS2000) accurate to _ 1%. For the first 8 h, the infusion rate was 1.2 ml/kg per hour, then continued at 0.6 ml/kg per hour for 24 h or more before blood sampling. Venous sites were inspected frequently and the cannula was replaced if there was any suspicion of the infusion becoming extravascnlar, and an additional 6 h was added to the infusion time. At the end of the infusion, blood (0.6 ml) and urine samples were obtained for PF-S and creatinine assay. Blood was sampled either from an arterial line or from another vein and was timed where possible to coincide with clinical blood samples. Urine samples were obtained from spontaneous voidings into a wad of cotton wool placed at the perineum (enclosed in a plastic napkin liner). Voidings were detected using an electronic detector with a probe embedded in the cotton wool and aspirated immediately using a plastic syringe. Contact of urine with cotton wool made no difference to the concentrations of PF-S and creatinine. The fact that plasma is only approximately 95% water was ignored. Duplicate 13-~tl aliquots from each plasma (separated immediately), diluted urine and diluted infusate sample were dispensed with 67 ~tl water and frozen at -20* C until analysed for polyfructoside. Sixty-microlitre aliquots were dispensed with 150 I.tl of water and frozen for later creatinine analysis. GFR (PF-S clearance) was calculated by: GFR = PF-S infusion rate/plasma PF-S concentration. Single-injection method. This method was only used when the infant had an i.v. cannula already inserted for fluid therapy and an arterial cannnla for blood sampling. PF-S, 200 mg/kg (2 ml/kg), was injected over 3 6 min, to allow smooth distribution throughout the central plasma compartment, into an injection port on the end of the i.v. cannula and followed by 1 ml 5% glucose solution. A small quantity was separately diluted ( • 200) and frozen for later analysis. The syringe was weighed immediately before and after the injection and the volume thus calculated (relative density 1.035). Blood samples (0.1 rrd), timed from the mid-point of the injection, were from the arterial cannula and were obtained fro~ a three-way tap
tions versus time profiles were displayed graphically on semi-logarithmic plots (Fig. 1). They were analysed in two ways. First, a straight line was fitted by eye to the terminal portion of the curve (without forcing occasional obvious outliers), and then the semi-logarithmic plot was subjected to manual graphical curve-stripping using a commonly described method [3-13] to derive the exponential curve formula. In most cases a triple-exponential formula was obtained, otherwise a double-exponential: PF-S concentration = A.e-Xa.t+B.e-Xb.t+C.e -xc-t, where t = time, A, B, C and La, ~ , ~ are the exponential constants, C and ~e being for the final mono-exponential part of the curve. In all cases a straight line was fitted to this without difficulty. A, B and C have the dimension mass volume (mg/l); ~a, ~,b and ~c have the dimension t -1 (min). A and ~a are the constants for the first exponent, derived from the earliest and steepest part of the concentration-time curve. Secondly, a computer programme (PKCalc, run on a microcomputer) performed the same function and derives a two- or three-exponential formula. PKCalc fits a straight least squares regression line to the experimental points forming the terminal portion of the curve. The advantage of the computer method is that it saves much time but it carries the disadvantage that outlying points (probably caused by laboratory errors) are not intelligently dealt with. Points were classed as outliers if they were more than three residual standard deviations (SDs) from the regression line. GFR (PF-S clearance) is the quantity injected, I, divided by the area under the concentration-time curve and calculated from: GFR = I/(A/~a+B/'Ab+CI~e), where I is in milligrams per kilogram birth weight (or study weight if this was greater). Others have shown that GFR is best related to weight, surface area being inappropriate in infants, and that a recent weight less than birth weight is also best not used because it gives a false over-estimation of GFR per kilogram [8].
Inulin assay. Inulin concentration was measured using a modification of the cysteine/tryptophan/sulphuric acid method of Waugh [2, 14]. A 13-I.tl sample (i. e. plasma, serum, diluted urine, diluted infusion sample) was dispensed with 67 ~1 water. The samples were first subjected to deproteinisation with cadmium hydroxide and then an alkali denaturing step to remove interfering sugars. The sensitivity of this method is 0.25 optical density (OD) units (at 515 nm)/gg per ml per cm. A 0.5 mg/ml original sample gives an OD of 0.72. The coefficient of variation (CV) was 1 % - 2 % for a sample concentration of 0.1 mg/ml and 1% at 0.5 mg/ml. The yield for PF-S added to plasma was 9 4 % - 105%. Plasma and urine blanks were usually 0.01 mg/ml PF-S equivalent or less (up to 0.07 without an alkali denaturing step).
Creatinine assay. Creatinine was measured in plasma and urine by a modification of a resin adsorption method [2, 15, 16]. The resin adsorbs creatinine but not substances which normally interfere with the Jaff6 creatinine assay. Approximately 6 mg Dowex 50W-X8(H) 200-400 mesh ion-exchange resin was added to the samples as a single drop of an 80% suspension in distilled water. After 15 min the resin was washed twice and the creatinine eluted with 600 I.tl alkaline picrate. The OD of the supernatant was measured at 515 nm. The reaction is linear and has a sensitivity of 0.0068 OD tmits/nmol per ml per cm. For a plasma creatinine of 100 I.tmol/1, the OD is 0.068. CV was 4.4% at 69 ~tmol/1 and 2.4% at 499 ~mol/l. The yield for creatinine added to plasma or urine was 9 5 % - 105%. Bilirubin, glucose, proteins and ketones did not interfere.
325 2.0
2.0-
O
1.s1 ................................................ ;--~;-;,.....-~-~......................... ~ 1.5-
~
/
t~
,n
A
m~A
,
o
1.0 -
~
t-
~, ~.c 9=
~,
.............................
, ' /"
1.0-
a" 9 " ' " ' ~ " "
"',i ......
'~ ...........
~'""~"
.............................
~o 0.513O.
'--E-- 0 . 5 -
0
I
0
80 0
~ 0.5
0
1.0
15
Fig. 2. Sixteen experiments showing continuous-infusion PF-S clearance which followed a single-injection PF-S clearance. The diagonal line is the line of identity
25 ,> 20 .c_
&
&
c-
o
0
I
I
5
10
I
I
I
15 20 25 Urine/plasma PF-S
I
I
30
35
Fig. 3. Comparison of urine/plasma ratios for creatinine and PF-S (Inutest| (n = 111). The diagonal line is the line of identity
Results
I
I
1
I
/,0 60 80 100 Plasma creatinine (~mol/I)
t
120
I
1/~0
Fig. 4. PF-S/creatinine clearance ratio plotted as a function of plasma creatinine (n = 111). The dotted lines show the mean and 95% tolerance interval for the ordinate. The diagonal line is the linear regression
2.0
injection
Single
20
"
In all single-injection experiments a smoothplasma disappearance curv e was obtained with no acute inflections, and outlying points were rare. The manual curve-stripping and the computer methods yielded very close results, The statistic GFR (manual) - GFR (computer) was never more than 6% of the mean of the two results. In some cases curve-stripping was difficult because the exponential slopes were not of sufficiently different magnitudes to avoid small errors. Figure 1 shows two typical PF-S disappearance curves from 2 patients, the symbols representing the experimental data and the continuous lines multiple exponential curves fitted by curve-stripping.: The curves do not become monoexponential until at least 2 and often 3 or 4 h after the injection. The range of ~,a was 0.0352-0.27 min -1, of )~b 0.00965-0.0732 rain -1, of )~c 0.00089-0.00353 rain-1 and
of the terminal half-time, loge2/~c, 196-779 rain. The wide ranges of these values were probably caused by the variable clinical states and ages of the study subjects. GFR and distribution volume vary widely, the latter presumably reflecting oedema. The range of GFR was 0.351.52 ml/kg per min, of distribution volume 259- 518 ml/kg and of mean residence time, distribution volume/GFR, 276-1042 min. These values are presented and discussed, together with other aspects of renal function in sick preterm infants, elsewhere [17]. Mean distribution volume was 366 ml/kg (SD 68, n = 37). This is higher than, but not significantly different from, the value obtained by Coulthard [3] (316 ml/kg, SD 43, n = 5) and higher still than Brion et al~ [18], possibly because the present study includes younger infants. In the 161experiments where a single injection was followedby a 24 h constant infusion, there was a close agreement between the two measurements, although in most cases the single-injection clearance was slightly less than the constant-infusion clearance (Fig. 2). The mean difference was 0.0673 ml/min per kg (range-0.079 to +0.206, SD 0.079, 95% confidence interval for difference = 0.0250.109). Figure 3 shows the urine/plasma creatinine concentration ratio plotted against the urine/plasma PF-S concentration ratio for the continuous-infusion experiments. The arithmetic mean ratio between PF-S and creatinine clearance for all the points in Fig. 3 is 1.101 (inverse 0.91). This clearance :ratio is approximately normally distributed and is symmetrical about the mean. The 95% tolerance interval (mean 1.101 + 2 SDs) is 0.770-1.433. The clearance ratio was slightly higher with higher plasma creatinine (Fig. 4), the regression being: clearance ratio = 0.872+0.00282 • creatinine. The regression slope has 95% confidence limits 0.00282_ 0.00166. There was no variation of this clearance ratio vr the degree of tubular water reabsorption as measured by urine/plasma PF-S (confidence limits for the regression slope 0.00285___ 0.00471). The ratio between log PF-S clearance and log creatinine clearance (geometric mean 1.075) was not used because the distribution is asymmetric and not normal.
326 Discussion
Curve-stripping Manual stripping of single-injection curves requires that the successive exponential slopes, ~,a, )~b,etc. are of different orders of magnitude, in practice different by successive factors of 7 or more, and that, if sampling were to have continued further, the same final mono-exponential curve would have continued to time infinity. This is because the method in fact integrates the curve to infinity. This process is very much operator dependent [3], but in the present study the mono-exponential part of the curve was defined accurately with three or more points. With computer methods only the experimenter will be able to judge whether outlying points are due to experimental error. If GFR is calculated using only the final mono-exponential part of the curve and ignoring the initial short half-life exponents due to the initial equilibration phase and using only C and ~ , it is always overestimated by between 7 and 24%, 47% in 1 patient where there was gross oedema and ascites. This compares with a mean value of 20% for the adult subjects studied with chromium-labelled ehtylene diamine tetra-acetic acid (Cr-EDTA) [11] and a mean of 4.4% in the healthy pre-term infants studied by Coulthard [3]. It is clear from the present study that ignoring the initial equilibration phase may sometimes result in unacceptable errors in calculating GFR. The curve does not become mono-exponential until up to 4 h, considerably longer,than in adults using Cr-EDTA [11]. The cause of this long equilibration period is presumably the slower penetration of PFS into the deeper extracellular compartments. It is not particularly a function of the low GFR or the high distribution volume, although the latter may contribute to this slow equilibration if the extracellular fluid is expanded by compartments such as oedema or ascites. Both Chantler et al. [11] and Veall and Gibbs [19] emphasise the need for the subject not to be oedematous. Strictly speaking this does not invalidate the method, but as these authors point out a much longer sampling period is required. The differences between this study and others [3, 11, 17, 18] reflect the different populations studied as well as methodological differences, including the longer sampling times here. )~ais similar to other studies using two-compartment analysis [3, 18], its wide range reflecting differences in distribution throughout extracellular fluid. Sampling arterial blood is ideal because it minimises the unknown delay between injection of tracer i. v. and its appearance in renal artery blood [11]. Computer methods do not in any way eliminate this requirement [19]. The use of arterial blood sampling is much easier now that arterial cannulas are more commonly used in neonatal practice, although sampling venous or capillary blood would not falsely underestimate GFR by more than 5% [11]. In one study, four samples of capillary blood were taken in the early phase of the plasma disappearance curve, in addition to the arterial samples. It was difficult to time the samples because blood drips only slowly and irregularly from a heel stick, as opposed to the rapid and easily timed sample obtained from arterial catheters. The time was taken as the mid-point of the bleeding period. In this case the curve for the capillary
samples was, when smoothed, 90s delayed compared with the arterial curve. This would have caused an underestimation error of 2% in calculating GFR. The error from venous sampling may be a little more than this, but circulation times are not known in these infants. Ethical considerations prevented further investigation of this. Sampling from nonvascular sites cannot give accurate results because other extracellular compartments are not, like the vascular compartment, a well-stirred pool. Single-injection methods require exact timing, accurate sampling and highly reliable, accurate, sensitive and precise laboratory methods before any curves can be analysed.
Convolution analysis of curves The continuous-infusion method is only valid if a steadystate plasma PF-S value is achieved, and the time required to achieve this cannot be predicted at the start of a continuous-infusion experiment. The plasma PF-S disappearance curves were mathematically convolved, using computer routines written in BBC BASIC by and available from the author, to predict theoretical PF-S concentration-time curves for a continuous infusion. Plasma PF-S would not reach 97% of the steady-state value until up to 48 h. This can be much shortened by preceding the infusion with a bolus of 1 / ~ times the infusion rate. The range of 1/~ is 283-1124 rain (nearly four-fold) in the 37 patients studied. Alternatively, an initial double-rate infusion for 200-790 min achieves the steady-statue value, but of the 8 values greater than 480 min 7 were in infants less than 2 days old. The present continuous-infusion experiments continued for 24 h after an initial double-rate infusion for 480 min. In the cases where the ideal time is only 200 rain, the plasma PF-S will overshoot, but the equilibration period is much shorter so steady state will still be achieved within 24 h. A steady state cannot be guaranteed within 24 h in any infant less than 3 days old or in infants who are grossly oedematous.
Continuous infusion versus single injection The relationship between the single-injection and constantinfusion clearances (Fig. 2) is the converse of that found by other authors [3]. This is likely to be because in the present study sampling has been continued for 8 h or more after the single injection, and the improved precision of the PF-S assay has resulted in a more accurately definable plasma disappearance curve. The tendency of previous studies to overestimate GFR with the single-injection method [3] has therefore been eliminated. The higher values for the constant-infusion clearance may be in part caused by the fact that they are inevitably 36 h later than the single-injection clearances, but the circulation delay between injection and PF-S appearing in renal arteries may contribute.
Creatinine clearance Creatinine clearance methods are still frequently used in renal function studies in infants and it has been claimed that creatinine clearance is a valid estimate of GFR in low birth
327
weight infants [20]. In pre-term infants variable results have been obtained by investigators comparing polyfructoside clearance with creatinine clearance, most finding fairly good agreement [ 18, 2 0 - 25]. The significant correlation coefficient found between creatinine clearance and polyfructoside clearance [20] hides large differences found between these two measurements in individual cases, the scatter being greater than in the present study. Correlation [20, 23] is not an appropriate statistical approach to comparing two measures of the same variable [26]. Problems with creatinine clearance are related to difficulties in measuring the true concentration of creatinine in plasma and also to an uncertain degree of creatinine reabsorption or secretion by the renal tubule and difficulties with timed urine collection. There is some evidence for tubular transfer of creatinine in adult patients [27, 28], but it is secretion not reabsorption and occurs especially in those in renal failure with high plasma creatinine concentrations. Evidence from lambs and newborn rabbits suggests that creatinine is significantly reabsorbed by the renal tubule especially in the immature kidney [29], so that creatinine clearance underestimates PF-S clearance more so in the most immature animals. This may apply to very low birth weight infants [ 18, 23] and in the present study there is evidence of some degree of creatinine absorption in the renal tubule (Fig. 3); hence creatinine clearance may not be valid in the sick immature infant. An alternative explanation for the diminished creatinine clearance compared with PF-S clearance is that the plasma creatinine is falsely high owing to the presence of non-creatinine chromogen, despite the resin adsorption method being used which is designed to measure only true creatinine. However, the PF-S/creatinine clearance ratio is higher with higher creatinines. The opposite would apply if non-creatinine chromogen were involved, because there is no reason why it should vary with true plasma creatinine [23]. The opposite did apply in one previous study which did not use a resin adsorption method [20]. Conclusion
The single-injection and continuous-infusion methods are different but both should in theory give accurate and similar results. The present experiments support their use in sick pre-term infants, but the continuous-infusion method cannot be used in infants less than 3 days old or in those who are Very oedematous, because of very slow equilibration. The single-injection method can be used in such patients provided that sampling continues for much longer than previously advocated. Acknowledgements. This work was funded by the Children Nationwide Medical Research Fund, UK, and the Southmead Research Foundation, Southmead Hospital, Bristol, UK. The author is grateful to Mrs. N. Kingston and Mrs. M. Williams for performing some of the laboratory work.
References 1. Coulthard MG, Ruddock V (1983) Validation of inulin as a marker for glomerular filtration in pre-term babies. Kidney Int 23:407-409 2. Wilkins BH (1992) The glomerular filterability of polyfructosan-S in immature infants. Pediatr Nephrol 6: 323- 327
3. Coulthard MG (1983) Comparison of methods of measuring renal function in preterm babies using inulin. J Pediatr 102:923-930 4. Guignard J-P (1977) Assessment of renal function without urine collection. Arch Dis Child 52:424 5. Rose GA (1969) Measurement of glomerular filtration rate by inulin clearance without urine collection. BMJ 2:91 - 9 3 6. Cole BR, Giangiacomo J, Ingelfinger JR, Robson AM (1972) Measurement of renal function without urine collection. A critical evaluation of the constant infusion technic for determination of inulin and para-aminohippurate. N Engl J Med 287:1109-1114 7. Fawer C-L, Torrado A, Guignard J-P (1979) Single injection clearance in the neonate. Biol Neonate 35:321-324 8. Coulthard MG (1985) Maturation of glomerular filtration in pre-term and mature babies. Early Hum Dev 11:281 - 292 9. Keen DV, Pearse RG (1988) Weight, length and head circumference curves for boys and girls between 20 and 42 weeks gestation. Arch Dis Child 63:1170-1172 10. Sapirstein LA, Widt DC, Mandel MJ, Hanusek G (1955) Volumes of distribution and clearances of intravenously injected creatinine in the dog. Am J Physiol 181: 330-336 11. Chantler C, Garnett ES, Parsons V, Veall N (1969) Glomerular filtration rate measurement in man by the single injection method using 5aCr-EDTA. Clin Sci 37: 169-180 12. Nosslin B (1965) Determination of clearance and distribution volume with the single injection technique. Acta Med Scand 179 [Supp1442]: 97-101 13. Gibaldi M, Perrier D (1975) Pharmacokinetics, 2nd edn. Dekker, New York, pp 89-95 14. Waugh WH (1977) Photometry of inulin and polyfructosan by use of a cysteine/tryptophan reaction. Clin Chem 23:639-645 15. Mitchell RJ (1973) Improved method for specific determination of creatinine in serum and urine. Clin Chem 19:408-410 16. Stoten A (1968) A micro method for creatinine using resin to remove interfering substances. J Med Lab Techno125:240-247 17. Wilkins B (1992) Renal function in sick pre-term infants. 1. Glomerular filtration rate. Arch Dis Child (in press) 18. Brion LP, Fleischman AR, McCarton C, Schwartz GJ (1986) A simple estimate of glomerular filtration rate in low birthweight infants during the first year of life: non-invasive assessment of body composition and growth. J Pediatr 109:698-707 19. Veall N, Gibbs GP (1982) The accurate determination of tracer clearance rates and equilibrium distribution volumes from single injection plasma measurements using numerical analysis. In" Joekes AM, Constable AR, Brown NJG, Tauxe WN (eds) Radionuclides in nephrology. Academic Press, New York, pp 125-130 20. Stonestreet BS, Bell EF, Oh W (1979) Validity of endogenous creatinine clearance in low birthweight infants. Pediatr Res 13: 1012-1014 21. Dean RFA, McCance RA (1947) Inulin, diodone, creatinine and urea clearances in newborn infants. J Physiol 106:431 -439 22. Barnett HL, Hare WK, McNamara H, Hare RS (1948) Influence of postnatal age on kidney function of premature infants. Proc Soc Exp Biol Med 69:55-57 23. Coulthard MG, Hey EN, Ruddock V (1985) Creatinine and urea clearances compared to inulin clearance in pre-term and mature babies. Early Hum Dev 11: 11 - 19 24. Aperia A, Broberger O, Elinder G, Herin P, Zetterstrrm R (1981) Postnatal development of renal function in pre-terrn and full-term infants. Acta Paediatr Scand 70: 183-187 25. Arant BS (1984) Estimating glomerular filtration rate in infants. J Pediatr 104:890-893 26. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet I: 307-310 27. Bauer JH, Brooks CS, Burch RN (1982) Clinical appraisal of creatmine clearance as a measurement of glomerular filtration rate. Am J Kidney Dis 2:337-346 28. Blythe WB (1982) The endogenous creatinine clearance. Am J Kidney Dis 2:321-323 29. Duarte-Silva M, Guignard J-P (1985) Creatinine transport by the maturing rabbit kidney. Kidney Int 28:595
Pediatr Nephrol (1992) 6:323-327 9 IPNA 1992
Original article A reappraisal of the measurement of glomerular filtration rate in pre-term infants Barry H. Wilkins* Department of Child Health, Bristol University and Neonatal Intensive Care Unit, Southmead Hospital, Bristol, UK Received April 25, 1991; received in revised form December 23, 1991; accepted January 15, 1992
Abstract. Thirty-seven single-injection polyfructosan-S (PF-S, Inutest| and 98 continuous-infusion PF-S/creatinine clearance studies were performed in 39 sick very low birth weight infants. The single-injection clearance method for measuring glomernlar filtration rate has been shown to be a reliable technique if sampling is continued for 8 h or more and the PF-S (Inutest| assay is sensitive, accurate and precise. The continuous-infusion clearance method is also valid if the infusion is continued for more than 24 ta and preceded by a loading dose in the form of a double-rate infusion for 8 h. Creatinine clearance is usually less than PF-S clearance, the mean ratio being 0.91, suggesting that there is some creatinine reabsorption in the renal tubule in sick very low birth weight infants. Key words: Glomerular filtration rate - Pre-term infant Newborn - Inulin - Polyfructosan-S - Creatinine - Clearance
Introduction Glomernlar filtration rate (GFR) has been measured in newborn infants using either inulin as an exogenous marker or endogenous creatinine, and using either standard clearances or indirect methods which avoid urine collection. The single-injection and continuous-infusion methods should give accurate measurements of GFR if inulin distributes only in extracellular fluid, is freely filtered, not reabsorbed, not metabolised and excreted by no other route. This is so in the adult, where inulin has long been regarded as the best marker of GFR, and it has been shown [1, 2] that inulin and polyfructosan-S (PF-S, Inutest| (Laevosan-Gesellschaft, Linz, Austria), a polyfructoside now used as an inulin substitute, are freely filtered in pre-
*Present address: Neonatal Unit, Rush Green Hospital, Romford,
Essex, UK
term newborn infants. Using inulin without urine collection by the constant-infusion technique [3-6] requires that the equilibration of the marker throughout the extracellular fluid space is complete. This has been shown not to be true when the inulin is infused for only 2 or 3 h [3, 4], and the technique is only valid with a longer infusion time. For this reason the continuous-infusion method cannot be used in the first I or 2 post-natal days. However, the single-injection technique has been found to overestimate the GFR in the newborn period, even when sampling is continued for 5 h after the injection [3, 7, 8]. The sickest and most immature infants are liable to have a lower GFR and higher extracellular fluid volume, especially when oedematous, which makes the single-injection and continuous-infusion methods less reliable. The inherent difficulties of urine collection preclude standard techniques. The purpose of the present investigation was to re-evaluate the single-injection method and compare it with the continuous-infusion method using PF-S (Inutest| and also to investigate whether creatinine clearance is an accurate measure of GFR in such pre-term infants. Laboratory methods for measuring polyfrnctoside and creatinine were first improved so that they were applicable to small samples with increased precision and accuracy. The study had approval from the local medical ethics committee and parental consent was obtained.
Patients and methods This study was part of a wider study of renal glomemlar and tubular function in very low birth weight infants. Thirty-seven single-injection and 98 continuous-infusion measurements of PF-S clearance were made in 39 infants, between the ages of 0.5 and 33 days, Gestation at birth was between 25.5 and 33 weeks and birth weight between 720 and 2,000 g. No patient had a progressive rise in plasma creatinine as evidence of acute renal failure either before or during the studies which were only performed when i.v. therapy was being given. No subject was studied within 24 h of receiving occasional doses of fmsemide, aminophylline, indomethacin, digoxin or vancomycin, and volume replacement was never given during single-injection studies or within 12 h of sampling in continuous-infusion experiments. Diuretics were not routinely used. All
324 close to the end of the cannula at approximately 10, 40, 80, 120, 240, 360 and 480 min after the injection; in some cases sampling was longer. Two millilitres of blood was always removed before, and replaced after, sampling.
8.07.5 R
7.0 e-
Single injection followed by constant infusion. A single-injection GFR
._o
and a constant-infusion GFR cannot be obtained at the same time in the same subject so the two methods cannot be compared simultaneously. On 16 occasions in 16 babies, the single-injection experiment was immediately followed by a continuous-infusion experiment, the double-infusion rate being continued for 4 h in this case. It was thus possible to compare the two techniques 2 4 - 3 6 h apart.
t-0
~: 6.0 0 0
5.5 It. D..
'- 5.0
-J
~.5
Analysis of single-injection PF-S disappearance curves. PF-S concentraJ
i
t
i
/
i
100
200
300 Time (rain)
t, O0
500
600
Fig. 1. Actual experimental points from 2 single-injection experiments in 2 patients to which two-exponential (lower) and three-exponential (upper) curves have been fitted by manual curve-stripping. There were insufficient early points in the lowergraph to derive an intermediate third exponential which also explains why the third point does not lie on the curve. The 95% confidence interval for polyfructosan-S (PF-S, Inutest| measurements in duplicate is approximately the height of the symbols at all levels babies received ethamsylate, a vascular stabilising drug which has no known renal effects. One baby was subject to intrauterine steroid exposure. The infants were evenly distributed about the 50th weight per-centile according to the intrauterine growth charts of Keen and Pearse [9]. Six subjects did not have any respiratory distress, 16 had severe respiratory distress syndrome. Only 2 did not receive assisted ventilation.
Constant-infusion method. A 25 mg/ml solution of PF-S (Inutest| in water was infused i.v. through a 0.22-1xm anti-bacterial filter, using a digital electronic syringe pump (Graseby MS2000) accurate to _ 1%. For the first 8 h, the infusion rate was 1.2 ml/kg per hour, then continued at 0.6 ml/kg per hour for 24 h or more before blood sampling. Venous sites were inspected frequently and the cannula was replaced if there was any suspicion of the infusion becoming extravascnlar, and an additional 6 h was added to the infusion time. At the end of the infusion, blood (0.6 ml) and urine samples were obtained for PF-S and creatinine assay. Blood was sampled either from an arterial line or from another vein and was timed where possible to coincide with clinical blood samples. Urine samples were obtained from spontaneous voidings into a wad of cotton wool placed at the perineum (enclosed in a plastic napkin liner). Voidings were detected using an electronic detector with a probe embedded in the cotton wool and aspirated immediately using a plastic syringe. Contact of urine with cotton wool made no difference to the concentrations of PF-S and creatinine. The fact that plasma is only approximately 95% water was ignored. Duplicate 13-~tl aliquots from each plasma (separated immediately), diluted urine and diluted infusate sample were dispensed with 67 ~tl water and frozen at -20* C until analysed for polyfructoside. Sixty-microlitre aliquots were dispensed with 150 I.tl of water and frozen for later creatinine analysis. GFR (PF-S clearance) was calculated by: GFR = PF-S infusion rate/plasma PF-S concentration. Single-injection method. This method was only used when the infant had an i.v. cannula already inserted for fluid therapy and an arterial cannnla for blood sampling. PF-S, 200 mg/kg (2 ml/kg), was injected over 3 6 min, to allow smooth distribution throughout the central plasma compartment, into an injection port on the end of the i.v. cannula and followed by 1 ml 5% glucose solution. A small quantity was separately diluted ( • 200) and frozen for later analysis. The syringe was weighed immediately before and after the injection and the volume thus calculated (relative density 1.035). Blood samples (0.1 rrd), timed from the mid-point of the injection, were from the arterial cannula and were obtained fro~ a three-way tap
tions versus time profiles were displayed graphically on semi-logarithmic plots (Fig. 1). They were analysed in two ways. First, a straight line was fitted by eye to the terminal portion of the curve (without forcing occasional obvious outliers), and then the semi-logarithmic plot was subjected to manual graphical curve-stripping using a commonly described method [3-13] to derive the exponential curve formula. In most cases a triple-exponential formula was obtained, otherwise a double-exponential: PF-S concentration = A.e-Xa.t+B.e-Xb.t+C.e -xc-t, where t = time, A, B, C and La, ~ , ~ are the exponential constants, C and ~e being for the final mono-exponential part of the curve. In all cases a straight line was fitted to this without difficulty. A, B and C have the dimension mass volume (mg/l); ~a, ~,b and ~c have the dimension t -1 (min). A and ~a are the constants for the first exponent, derived from the earliest and steepest part of the concentration-time curve. Secondly, a computer programme (PKCalc, run on a microcomputer) performed the same function and derives a two- or three-exponential formula. PKCalc fits a straight least squares regression line to the experimental points forming the terminal portion of the curve. The advantage of the computer method is that it saves much time but it carries the disadvantage that outlying points (probably caused by laboratory errors) are not intelligently dealt with. Points were classed as outliers if they were more than three residual standard deviations (SDs) from the regression line. GFR (PF-S clearance) is the quantity injected, I, divided by the area under the concentration-time curve and calculated from: GFR = I/(A/~a+B/'Ab+CI~e), where I is in milligrams per kilogram birth weight (or study weight if this was greater). Others have shown that GFR is best related to weight, surface area being inappropriate in infants, and that a recent weight less than birth weight is also best not used because it gives a false over-estimation of GFR per kilogram [8].
Inulin assay. Inulin concentration was measured using a modification of the cysteine/tryptophan/sulphuric acid method of Waugh [2, 14]. A 13-I.tl sample (i. e. plasma, serum, diluted urine, diluted infusion sample) was dispensed with 67 ~1 water. The samples were first subjected to deproteinisation with cadmium hydroxide and then an alkali denaturing step to remove interfering sugars. The sensitivity of this method is 0.25 optical density (OD) units (at 515 nm)/gg per ml per cm. A 0.5 mg/ml original sample gives an OD of 0.72. The coefficient of variation (CV) was 1 % - 2 % for a sample concentration of 0.1 mg/ml and 1% at 0.5 mg/ml. The yield for PF-S added to plasma was 9 4 % - 105%. Plasma and urine blanks were usually 0.01 mg/ml PF-S equivalent or less (up to 0.07 without an alkali denaturing step).
Creatinine assay. Creatinine was measured in plasma and urine by a modification of a resin adsorption method [2, 15, 16]. The resin adsorbs creatinine but not substances which normally interfere with the Jaff6 creatinine assay. Approximately 6 mg Dowex 50W-X8(H) 200-400 mesh ion-exchange resin was added to the samples as a single drop of an 80% suspension in distilled water. After 15 min the resin was washed twice and the creatinine eluted with 600 I.tl alkaline picrate. The OD of the supernatant was measured at 515 nm. The reaction is linear and has a sensitivity of 0.0068 OD tmits/nmol per ml per cm. For a plasma creatinine of 100 I.tmol/1, the OD is 0.068. CV was 4.4% at 69 ~tmol/1 and 2.4% at 499 ~mol/l. The yield for creatinine added to plasma or urine was 9 5 % - 105%. Bilirubin, glucose, proteins and ketones did not interfere.
325 2.0
2.0-
O
1.s1 ................................................ ;--~;-;,.....-~-~......................... ~ 1.5-
~
/
t~
,n
A
m~A
,
o
1.0 -
~
t-
~, ~.c 9=
~,
.............................
, ' /"
1.0-
a" 9 " ' " ' ~ " "
"',i ......
'~ ...........
~'""~"
.............................
~o 0.513O.
'--E-- 0 . 5 -
0
I
0
80 0
~ 0.5
0
1.0
15
Fig. 2. Sixteen experiments showing continuous-infusion PF-S clearance which followed a single-injection PF-S clearance. The diagonal line is the line of identity
25 ,> 20 .c_
&
&
c-
o
0
I
I
5
10
I
I
I
15 20 25 Urine/plasma PF-S
I
I
30
35
Fig. 3. Comparison of urine/plasma ratios for creatinine and PF-S (Inutest| (n = 111). The diagonal line is the line of identity
Results
I
I
1
I
/,0 60 80 100 Plasma creatinine (~mol/I)
t
120
I
1/~0
Fig. 4. PF-S/creatinine clearance ratio plotted as a function of plasma creatinine (n = 111). The dotted lines show the mean and 95% tolerance interval for the ordinate. The diagonal line is the linear regression
2.0
injection
Single
20
"
In all single-injection experiments a smoothplasma disappearance curv e was obtained with no acute inflections, and outlying points were rare. The manual curve-stripping and the computer methods yielded very close results, The statistic GFR (manual) - GFR (computer) was never more than 6% of the mean of the two results. In some cases curve-stripping was difficult because the exponential slopes were not of sufficiently different magnitudes to avoid small errors. Figure 1 shows two typical PF-S disappearance curves from 2 patients, the symbols representing the experimental data and the continuous lines multiple exponential curves fitted by curve-stripping.: The curves do not become monoexponential until at least 2 and often 3 or 4 h after the injection. The range of ~,a was 0.0352-0.27 min -1, of )~b 0.00965-0.0732 rain -1, of )~c 0.00089-0.00353 rain-1 and
of the terminal half-time, loge2/~c, 196-779 rain. The wide ranges of these values were probably caused by the variable clinical states and ages of the study subjects. GFR and distribution volume vary widely, the latter presumably reflecting oedema. The range of GFR was 0.351.52 ml/kg per min, of distribution volume 259- 518 ml/kg and of mean residence time, distribution volume/GFR, 276-1042 min. These values are presented and discussed, together with other aspects of renal function in sick preterm infants, elsewhere [17]. Mean distribution volume was 366 ml/kg (SD 68, n = 37). This is higher than, but not significantly different from, the value obtained by Coulthard [3] (316 ml/kg, SD 43, n = 5) and higher still than Brion et al~ [18], possibly because the present study includes younger infants. In the 161experiments where a single injection was followedby a 24 h constant infusion, there was a close agreement between the two measurements, although in most cases the single-injection clearance was slightly less than the constant-infusion clearance (Fig. 2). The mean difference was 0.0673 ml/min per kg (range-0.079 to +0.206, SD 0.079, 95% confidence interval for difference = 0.0250.109). Figure 3 shows the urine/plasma creatinine concentration ratio plotted against the urine/plasma PF-S concentration ratio for the continuous-infusion experiments. The arithmetic mean ratio between PF-S and creatinine clearance for all the points in Fig. 3 is 1.101 (inverse 0.91). This clearance :ratio is approximately normally distributed and is symmetrical about the mean. The 95% tolerance interval (mean 1.101 + 2 SDs) is 0.770-1.433. The clearance ratio was slightly higher with higher plasma creatinine (Fig. 4), the regression being: clearance ratio = 0.872+0.00282 • creatinine. The regression slope has 95% confidence limits 0.00282_ 0.00166. There was no variation of this clearance ratio vr the degree of tubular water reabsorption as measured by urine/plasma PF-S (confidence limits for the regression slope 0.00285___ 0.00471). The ratio between log PF-S clearance and log creatinine clearance (geometric mean 1.075) was not used because the distribution is asymmetric and not normal.
326 Discussion
Curve-stripping Manual stripping of single-injection curves requires that the successive exponential slopes, ~,a, )~b,etc. are of different orders of magnitude, in practice different by successive factors of 7 or more, and that, if sampling were to have continued further, the same final mono-exponential curve would have continued to time infinity. This is because the method in fact integrates the curve to infinity. This process is very much operator dependent [3], but in the present study the mono-exponential part of the curve was defined accurately with three or more points. With computer methods only the experimenter will be able to judge whether outlying points are due to experimental error. If GFR is calculated using only the final mono-exponential part of the curve and ignoring the initial short half-life exponents due to the initial equilibration phase and using only C and ~ , it is always overestimated by between 7 and 24%, 47% in 1 patient where there was gross oedema and ascites. This compares with a mean value of 20% for the adult subjects studied with chromium-labelled ehtylene diamine tetra-acetic acid (Cr-EDTA) [11] and a mean of 4.4% in the healthy pre-term infants studied by Coulthard [3]. It is clear from the present study that ignoring the initial equilibration phase may sometimes result in unacceptable errors in calculating GFR. The curve does not become mono-exponential until up to 4 h, considerably longer,than in adults using Cr-EDTA [11]. The cause of this long equilibration period is presumably the slower penetration of PFS into the deeper extracellular compartments. It is not particularly a function of the low GFR or the high distribution volume, although the latter may contribute to this slow equilibration if the extracellular fluid is expanded by compartments such as oedema or ascites. Both Chantler et al. [11] and Veall and Gibbs [19] emphasise the need for the subject not to be oedematous. Strictly speaking this does not invalidate the method, but as these authors point out a much longer sampling period is required. The differences between this study and others [3, 11, 17, 18] reflect the different populations studied as well as methodological differences, including the longer sampling times here. )~ais similar to other studies using two-compartment analysis [3, 18], its wide range reflecting differences in distribution throughout extracellular fluid. Sampling arterial blood is ideal because it minimises the unknown delay between injection of tracer i. v. and its appearance in renal artery blood [11]. Computer methods do not in any way eliminate this requirement [19]. The use of arterial blood sampling is much easier now that arterial cannulas are more commonly used in neonatal practice, although sampling venous or capillary blood would not falsely underestimate GFR by more than 5% [11]. In one study, four samples of capillary blood were taken in the early phase of the plasma disappearance curve, in addition to the arterial samples. It was difficult to time the samples because blood drips only slowly and irregularly from a heel stick, as opposed to the rapid and easily timed sample obtained from arterial catheters. The time was taken as the mid-point of the bleeding period. In this case the curve for the capillary
samples was, when smoothed, 90s delayed compared with the arterial curve. This would have caused an underestimation error of 2% in calculating GFR. The error from venous sampling may be a little more than this, but circulation times are not known in these infants. Ethical considerations prevented further investigation of this. Sampling from nonvascular sites cannot give accurate results because other extracellular compartments are not, like the vascular compartment, a well-stirred pool. Single-injection methods require exact timing, accurate sampling and highly reliable, accurate, sensitive and precise laboratory methods before any curves can be analysed.
Convolution analysis of curves The continuous-infusion method is only valid if a steadystate plasma PF-S value is achieved, and the time required to achieve this cannot be predicted at the start of a continuous-infusion experiment. The plasma PF-S disappearance curves were mathematically convolved, using computer routines written in BBC BASIC by and available from the author, to predict theoretical PF-S concentration-time curves for a continuous infusion. Plasma PF-S would not reach 97% of the steady-state value until up to 48 h. This can be much shortened by preceding the infusion with a bolus of 1 / ~ times the infusion rate. The range of 1/~ is 283-1124 rain (nearly four-fold) in the 37 patients studied. Alternatively, an initial double-rate infusion for 200-790 min achieves the steady-statue value, but of the 8 values greater than 480 min 7 were in infants less than 2 days old. The present continuous-infusion experiments continued for 24 h after an initial double-rate infusion for 480 min. In the cases where the ideal time is only 200 rain, the plasma PF-S will overshoot, but the equilibration period is much shorter so steady state will still be achieved within 24 h. A steady state cannot be guaranteed within 24 h in any infant less than 3 days old or in infants who are grossly oedematous.
Continuous infusion versus single injection The relationship between the single-injection and constantinfusion clearances (Fig. 2) is the converse of that found by other authors [3]. This is likely to be because in the present study sampling has been continued for 8 h or more after the single injection, and the improved precision of the PF-S assay has resulted in a more accurately definable plasma disappearance curve. The tendency of previous studies to overestimate GFR with the single-injection method [3] has therefore been eliminated. The higher values for the constant-infusion clearance may be in part caused by the fact that they are inevitably 36 h later than the single-injection clearances, but the circulation delay between injection and PF-S appearing in renal arteries may contribute.
Creatinine clearance Creatinine clearance methods are still frequently used in renal function studies in infants and it has been claimed that creatinine clearance is a valid estimate of GFR in low birth
327
weight infants [20]. In pre-term infants variable results have been obtained by investigators comparing polyfructoside clearance with creatinine clearance, most finding fairly good agreement [ 18, 2 0 - 25]. The significant correlation coefficient found between creatinine clearance and polyfructoside clearance [20] hides large differences found between these two measurements in individual cases, the scatter being greater than in the present study. Correlation [20, 23] is not an appropriate statistical approach to comparing two measures of the same variable [26]. Problems with creatinine clearance are related to difficulties in measuring the true concentration of creatinine in plasma and also to an uncertain degree of creatinine reabsorption or secretion by the renal tubule and difficulties with timed urine collection. There is some evidence for tubular transfer of creatinine in adult patients [27, 28], but it is secretion not reabsorption and occurs especially in those in renal failure with high plasma creatinine concentrations. Evidence from lambs and newborn rabbits suggests that creatinine is significantly reabsorbed by the renal tubule especially in the immature kidney [29], so that creatinine clearance underestimates PF-S clearance more so in the most immature animals. This may apply to very low birth weight infants [ 18, 23] and in the present study there is evidence of some degree of creatinine absorption in the renal tubule (Fig. 3); hence creatinine clearance may not be valid in the sick immature infant. An alternative explanation for the diminished creatinine clearance compared with PF-S clearance is that the plasma creatinine is falsely high owing to the presence of non-creatinine chromogen, despite the resin adsorption method being used which is designed to measure only true creatinine. However, the PF-S/creatinine clearance ratio is higher with higher creatinines. The opposite would apply if non-creatinine chromogen were involved, because there is no reason why it should vary with true plasma creatinine [23]. The opposite did apply in one previous study which did not use a resin adsorption method [20]. Conclusion
The single-injection and continuous-infusion methods are different but both should in theory give accurate and similar results. The present experiments support their use in sick pre-term infants, but the continuous-infusion method cannot be used in infants less than 3 days old or in those who are Very oedematous, because of very slow equilibration. The single-injection method can be used in such patients provided that sampling continues for much longer than previously advocated. Acknowledgements. This work was funded by the Children Nationwide Medical Research Fund, UK, and the Southmead Research Foundation, Southmead Hospital, Bristol, UK. The author is grateful to Mrs. N. Kingston and Mrs. M. Williams for performing some of the laboratory work.
References 1. Coulthard MG, Ruddock V (1983) Validation of inulin as a marker for glomerular filtration in pre-term babies. Kidney Int 23:407-409 2. Wilkins BH (1992) The glomerular filterability of polyfructosan-S in immature infants. Pediatr Nephrol 6: 323- 327
3. Coulthard MG (1983) Comparison of methods of measuring renal function in preterm babies using inulin. J Pediatr 102:923-930 4. Guignard J-P (1977) Assessment of renal function without urine collection. Arch Dis Child 52:424 5. Rose GA (1969) Measurement of glomerular filtration rate by inulin clearance without urine collection. BMJ 2:91 - 9 3 6. Cole BR, Giangiacomo J, Ingelfinger JR, Robson AM (1972) Measurement of renal function without urine collection. A critical evaluation of the constant infusion technic for determination of inulin and para-aminohippurate. N Engl J Med 287:1109-1114 7. Fawer C-L, Torrado A, Guignard J-P (1979) Single injection clearance in the neonate. Biol Neonate 35:321-324 8. Coulthard MG (1985) Maturation of glomerular filtration in pre-term and mature babies. Early Hum Dev 11:281 - 292 9. Keen DV, Pearse RG (1988) Weight, length and head circumference curves for boys and girls between 20 and 42 weeks gestation. Arch Dis Child 63:1170-1172 10. Sapirstein LA, Widt DC, Mandel MJ, Hanusek G (1955) Volumes of distribution and clearances of intravenously injected creatinine in the dog. Am J Physiol 181: 330-336 11. Chantler C, Garnett ES, Parsons V, Veall N (1969) Glomerular filtration rate measurement in man by the single injection method using 5aCr-EDTA. Clin Sci 37: 169-180 12. Nosslin B (1965) Determination of clearance and distribution volume with the single injection technique. Acta Med Scand 179 [Supp1442]: 97-101 13. Gibaldi M, Perrier D (1975) Pharmacokinetics, 2nd edn. Dekker, New York, pp 89-95 14. Waugh WH (1977) Photometry of inulin and polyfructosan by use of a cysteine/tryptophan reaction. Clin Chem 23:639-645 15. Mitchell RJ (1973) Improved method for specific determination of creatinine in serum and urine. Clin Chem 19:408-410 16. Stoten A (1968) A micro method for creatinine using resin to remove interfering substances. J Med Lab Techno125:240-247 17. Wilkins B (1992) Renal function in sick pre-term infants. 1. Glomerular filtration rate. Arch Dis Child (in press) 18. Brion LP, Fleischman AR, McCarton C, Schwartz GJ (1986) A simple estimate of glomerular filtration rate in low birthweight infants during the first year of life: non-invasive assessment of body composition and growth. J Pediatr 109:698-707 19. Veall N, Gibbs GP (1982) The accurate determination of tracer clearance rates and equilibrium distribution volumes from single injection plasma measurements using numerical analysis. In" Joekes AM, Constable AR, Brown NJG, Tauxe WN (eds) Radionuclides in nephrology. Academic Press, New York, pp 125-130 20. Stonestreet BS, Bell EF, Oh W (1979) Validity of endogenous creatinine clearance in low birthweight infants. Pediatr Res 13: 1012-1014 21. Dean RFA, McCance RA (1947) Inulin, diodone, creatinine and urea clearances in newborn infants. J Physiol 106:431 -439 22. Barnett HL, Hare WK, McNamara H, Hare RS (1948) Influence of postnatal age on kidney function of premature infants. Proc Soc Exp Biol Med 69:55-57 23. Coulthard MG, Hey EN, Ruddock V (1985) Creatinine and urea clearances compared to inulin clearance in pre-term and mature babies. Early Hum Dev 11: 11 - 19 24. Aperia A, Broberger O, Elinder G, Herin P, Zetterstrrm R (1981) Postnatal development of renal function in pre-terrn and full-term infants. Acta Paediatr Scand 70: 183-187 25. Arant BS (1984) Estimating glomerular filtration rate in infants. J Pediatr 104:890-893 26. Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet I: 307-310 27. Bauer JH, Brooks CS, Burch RN (1982) Clinical appraisal of creatmine clearance as a measurement of glomerular filtration rate. Am J Kidney Dis 2:337-346 28. Blythe WB (1982) The endogenous creatinine clearance. Am J Kidney Dis 2:321-323 29. Duarte-Silva M, Guignard J-P (1985) Creatinine transport by the maturing rabbit kidney. Kidney Int 28:595
