ClinicalScience (1981) 61,385-389
385
A non-invasive y-camera technique for the measurement of
intrarenal flow distribution in man
S . M. G R U E N E W A L D , C. C . N I M M O N , M . K . N A W A Z
AND
K . E. B R I T T O N
Department of Nuclear Medicine and Radioisotopes, St Bartholomew’s Hospital, London
(Received 4 September 1980126 January 1981; accepted 8 April 1981)
Summary
Abbreviation: ROI, regions of interest.
1. A new method, based on the transit time of
o-iodohippurate sodium (Hippuran) through the
kidney, is proposed as an accurate non-invasive
means of measuring the intrarenal flow distribution in man.
2. Data from [1231]Hippuran y-camera
renography are utilized in this method which
employs region of interest selection, deconvolution, cross-correlation and curve subtraction
to obtain the spectrum of transit times through
the cortical and juxtamedullary nephrons.
3. In 12 normal subjects the mean percentage
cortical flow was 83.9% (SEM 0.7%) which is
approximately the anatomical proportion of
cortical nephrons in the human kidney.
4. Cortical flow as a percentage of total was
significantly reduced in 2 1 hypertensive patients,
all of whom had no evidence of primary renal
disease (mean 74.6%, SEM 1.5%).
5. In both the normotensive and hypertensive
groups there was a good correlation between the
results obtained from the left and right kidneys of
the same patient showing the parallel physiological response of the two kidneys (mean
difference 4%, P < 0.001).
6. Reduction in the distribution of flow to the
cortical nephrons in the essential hypertensive
patients supports the hypothesis that renal
autoregulation is important in this syndrome.
Key words: cross-correlation, deconvolution,
y-camera renography, Hippuran, intrarenal flow
distribution, transit time.
Correspondence: Dr K. E. Britton, Department of
Nuclear
Medicine
and
Radioisotopes,
St
Bartholomew’s Hospital, West Smithfield, London
EC 1A 7BE.
25
0143-5221/81/100385-05$01.50/1
Introduction
There is to date no generally accepted method of
determining the intrarenal flow distribution in
man. The importance of this measurement lies in
the fact that overall renal blood flow may remain
relatively constant while marked alterations are
occurring in renal physiology. The intrarenal
haemodynamic changes which alter the flow to
the cortical and juxtamedullary nephrons, may be
the physiological determinant of salt and water
homeostasis 111, may have an importaat role in
the pathogenesis of renal failure 121 and in the
kidney’s response to hypertension 131. Meny of
the proposed techniques are not readily applicable in man [4,51, and the only two claiming
some support are the inert-gas washout [61 and
the [1311]Hippurantransit-time methods [7-91.
Validation of the 133Xe-washouttechnique has
been frequently attempted [lo, 111, but only
Schuler et al. [ 121 have shown a good correlation
between the curve components and the distribution of microspheres in the kidney. They,
however, studied a situation in which no change
in intrarenal blood flow distribution was found.
Other workers in the field have failed to show this
correlation and the method has been subject to
much criticism [4, 13-171. Although requiring an
arterial catheter, it is a reasonable method for
measuring the total renal blood flow [ l l l , but
133Xe-washoutcurves given the above criticism
should not be used to estimate the intrarenal flow
distribution with any justification.
The Hippuran transit-time method has been
validated in man [18, 191. The results of this
method have been reproduced by Reeve &
@ 1981 The Biochemical Society and the Medical Research Society
386
S . M . Gruenewald et al.
Crawley [201 using a different deconvolution
technique avoiding albumin injection and
computer-assisted blood background subtraction. In experimental animals there was a
highly significant correlation between the results
of the Hippuran transit-time method and the
distribution of radioactive microspheres [81.
Truniger [51 felt that the agreement betwen
microsphere distribution data and the results of
the deconvolution analysis of Rritton et al. 1211
was remarkable, but still not sufficient proof for
its validation.
We propose a new method, based on the
frequency distribution of Hippuran transit time,
for determining intrarenal haemodynamics. The
data from [ 12311Hippurany-camera renography is
utilized to determine the fractional cortical flow
of the total renal flow by means of the mathematical processes of region of interest selection,
deconvolution, cross-correlation, normalization
and curve subtraction. The use of the y-camera
avoids the disadvantages of probe renography
which include the difficulty of renal localization
and the inability to separate renal pelvis from
parenchyma. In the course of these studies
evidence has been obtained that calyceal and/or
pelvic retention of tracer may distort the apparent
distribution of transit-times and can contribute an
additional mode to this distribution.
Methods
The patient is seated reclining in a modified
dentists’ chair with the ycamera (Ohio Nuclear
series 100) fitted with a diverging collimator set
up over his back so that both kidneys and the
lower chest are in the field of view. After a rapid
intravenous injection of [12311Hippuran1221 into
a medial antecubital vein, 64 x 64 resolution
frames are recorded initially at 1 s interval for
100 s followed by 90 frames at 10 s intervals to
complete the 20 min study with an on-line Varian
V76 computer programmed to record and
process the data.
An image of the data at 2 min is reconstructed
and displayed and from this image regions of
interest (ROI) are defined for the whole left and
right kidneys, background regions and the heart
region.
To outline the pelvi-calyceal area, the computer constructs a mean time picture which is a
functional image displaying the mean tracer time
for each pixel of the 64 x 64 matrix so that pixels
with long mean times, i.e. calyces and pelvis, have
high intensities. It is now possible to draw an
ROI parallel to the outer lateral edge of the
kidney so as to include all areas of high intensity
leaving an outer zone of cortex. A third ROI is
constructed from the 15 min image corresponding
to the renal pelvis (inner zone).
Thus these ROI form three zones in each
kidney. These zones are not specifically anatomical, but help to emphasize the dominant
component in each zone to aid subsequent
mathematical analysis. The outer zone may be
taken as representative of the cortical nephrons;
the middle zone contains mainly medullary and
calyceal region, but in addition cortical components from over- and under-lying cortex; and
the inner zone is representative of the renal pelvis
with only a small renal parenchymal contribution.
With the heart curve as the renal input and the
curves from the outer and outer-and-middle
zones separately, deconvolution analyses by the
direct matrix-algorithm method are performed to
give the retention of activity in each zone
(impulse retention function), which would have
resulted from a theoretical spike injection into the
renal artery without recirculation. The early
vascular component now has to be removed from
these impulse retention functions by detecting the
subsequent plateau level of the retention function
followed by back extrapolation from the shoulder
of this curve.
Differentiation of these corrected retention
functions gives the spectra of transit time through
the outer and outer-and-middle zones respectively, but there is a high statistical noise
contribution. To reduce this noise and recover the
signal, the transit time spectra are crosscorrelated with a Gaussian function. The
assumptions and mathematical basis of this
procedure have been discussed by Nimmon et
al., 1231.
If one makes the assumption (see the Discussion section) that the transit-time spectrum of
the cortical nephrons is not appreciably lengthened during flow through the collecting system
then tracer transit through the outer zone will
represent flow mainly through the cortical
nephrons alone. This outer zone cross-correlation
curve is normalized to the value of the outer and
middle zone cross-correlation curve by means of
a proportionality factor arrived at by comparing
the leading edges of the two curves. Subtraction
of the two cross-correlation curves results in a
new curve, which represents the second mode
originating from the middle zone. In the absence
of pelvi-calyceal holdup, this curve is due to
tracer transit through the juxtamedullary
nephrons. By comparing the heights of these
cortical and juxtamedullary cross-correlation
curves, one obtains the percentage cortical flow
from the total flow through the nephrons [211 and
387
Intrarenaljow distribution in man
simulation studies have shown that the standard
deviation for these results is 6.8% when [12311Hippuran is used in a dose of 1.5-2.0 mCi,
provided that the renal function is near normal.
TABLE 1. Intrarenal flow distribution in normotensive
patients
Patient no.
Blood pressure
(mmHg)
Cortical flow in each kidney
(96)
~
Results
1
2
3
Twelve normotensive patients on no medication
and referred for renography for various reasons
(e.g. transplant donor) were studied by the above
method if their routine renogram analysis
revealed a normal result. They were found to
have a mean arterial blood pressure (diastolic and
f pulse pressure) of 93 mmHg with a range of
83-100 mmHg and their percentage cortical flow
(Table 1) had a mean value of 83.9% (SEM
0.7%).
Twenty-one patients with mild essential hypertension, on no treatment and without pelvicalyceal tracer retention on renography, were
also studied (Table 2) and their mean blood
pressure was 118 mmHg (range 105-150
mmHg) and percentage cortical flow was 74.6%
1.5%).
The difference between these means is
statistically significant ( P < 0.002). Comparing
the percentage cortical flow between the left and
right kidneys of each patient in the total group of
33 patients, there was a good correlation between
the two sides ( r = +0.88, P < 0.001).
4
5
6
7
120165
120170
120170
130/80
120180
115180
10
11
120180
120185
125180
125180
115175
I2
120180
8
9
Patient no.
1
2
3
4
5
6
7
8
9
10
11
12
Assumptions inherent in the Hippuran transittime method are that the mean rate of Hippuran
uptake into the kidney, transfer through the cells
of the proximal tubule and rate of removal from
the renal pelvis must be rapid relative to the mean
rate of Hippuran transit along the lumens of the
nephrons of each population. In the normal
kidney with a normal pelvis these assumptions
have been shown to be true [71.
What then accounts for the difference between
the transit times of Hippuran through the cortical
nephrons and juxtamedullary nephrons? In both
populations of nephrons the Hippuran is secreted
into the proximal tubules and thereafter is not
reabsorbed. The greatest contribution to the
transit time is the time taken for the Hippuran to
pass through the respective loops of Henle. The
collecting ducts, calyces and pelvis are common
to both cortical nephrons and juxtamedullary
nephrons and, because transit times through
these parts of the nephron are of the same order
Right
82
88
85
83
83
85
87
81
83
90
80
81
87
78
85
76
82
82
84
88
85
89
86
80
85
91
TABLE2. Intrarenal flow distribution in untreated essential
hypertensive patients
Blood pressure
(mmHd
(SEM
Discussion
Left
13
I4
15
16
17
18
19
20
21
140/100
13 5 / LOO
140190
150190
140195
170/110
1551100
160195
2001125
140195
I50/100
150195
18011 10
145190
1901120
140/100
145195
1501105
150/100
145/100
l50/100
Cortical flow in each kidney
(%)
Left
Right
81
79
76
63
53
81
75
76
73
61
84
82
83
70
52
82
76
82
61
65
80
74
84
71
60
75
80
71
68
55
80
74
88
76
90
71
78
74
92
77
89
71
as the sampling time (10-20 s), they will cause no
appreciable delay to the passage of tracer.
Thus it might be expected that the cortical
nephrons with their short loops would have a
faster transit time than the juxtamedullary
nephrons whose long loops penetrate deeply into
the renal medulla, where the high osmolality
causes fluid reabsorption and therefore slows
movement of the intratubular non-reabsorbable
Hippuran. A high osmolality is required in the
388
S . M. Gruenewald et al.
renal medulla in order for the kidneys to
concentrate solutes and conserve water. This can
only be maintained if flow through the nonautoregulating juxtamedullary nephrons is slow.
The method described here uses initial regions
of interest to separate cortical, medullary and
pelvic zones so that the outer zone contains
predominantly cortical nephron glomeruli and
loops of Henle, but excludes juxtamedullary
glomeruli and the long loops of their nephrons.
The transit time measured through this outer
zone represents predominantly cortical nephron
transit. About 85% of the total nephron
population are cortical nephrons [241 and in the
normal subject one would expect that approximately this percentage of the total flow will be
passing through the cortical nephrons at any time
under resting conditions. These expectations have
been borne out by this pilot study of left and right
kidneys of the 12 normotensive patients (mean
83-9%, SEM 0.7%).
The problem of measuring the distribution of
transit times is not straightforward. The spectra
obtained by deconvolution with matrix algorithm
are the sum of the true solution and an oscillatory
noise component. It is the oscillatory nature of
this component which makes quantitation and
recognition of statistically significant modes
difficult, when simple smoothing algorithms are
applied either pre- or post-deconvolution. An
alternative scheme of utilizing only mean time
and variance of spectra (Reeve et al. [251)
suffers from two main disadvantages. The
variance is heavily dependent, as shown by the
authors, on the state of diuresis; it fails to give a
measurement of percentage ‘modal’ contribution.
Further any errors in judging the point of
truncation of the spectra will give a significant
contribution to the variance and possible erroneous identification of further ‘modes’. A comparison of deconvolution methods for model
resolution of spectra has been made [231. The
correlation technique used in the present study
was found to give more accurate results than
either the matrix algorithm with simple 1-2-1
smoothing or a more sophisticated iterative
technique. The simulation data used in these
studies had a statistical noise level comparable
with that obtained from patient data in this
clinical study.
The main limitation of analysis of probe
renograms is that it is not possible to separate out
the effect of calyceal-pelvic transport on the
transit spectra. Further no information is obtainable on the site of origin within the kidney of a
particular mode. Probe methods for renographic
analysis have been used for a long time 171. It has
been found wanting because the renal pelvis is
included in the field of view of a probe and may
well be contributing to a multimodal distribution
of transit time. By using a y-camera and the
technique described, the pelvis is excluded as a
possible source of the extra modes described by
Reeve et al. [25] in hypertensive patients for,
in the present study, a slightly dilated calyx
or renal pelvis was the commonest cause of
multimodal data. The use of regional renography
with the y-camera overcomes the above shortcomings of probe renography which is essential
for a serious attempt at measuring nephron
transit times.
A reduction in renal blood flow is common in
patients with essential hypertension [261. It has
been shown angiographically that this decrease is
not uniform in distribution, but that cortical flow
is reduced out of proportion to the other parts of
the kidney [3]. In the present study the mildly
hypertensive patients likewise showed a reduction
in the percentage cortical flow (mean 74.6%, SEM
1.5%).
This can be looked at in two ways. If the
essential hypertension comes first, the reduction
in cortical nephron flow represents a normal
autoregulatory response reducing perfusion pressure to the glomeruli of the outer cortex to
maintain a normal filtration rate 1271. At the
same time the non-autoregulating juxtamedullary nephrons receive an increased perfusion pressure. Both actions tend to reduce
distribution to cortical nephrons and increase that
to juxtamedullary nephrons. Alternatively, alteration of autoregulatory control might be a primary
phenomenon in essential hypertension leading to
increased salt and water retention increasing
cardiac output, after the hypothesis of
Ledingham & Cohen [281 or else the autoregulatory control disturbance would be
associated with local renin release subserving
autoregulation [271 and also systemic renin
release initiating hypertension or both. Some of
the patients with essential hypertension have a
normal intrarenal flow distribution (Table 2),
making a primary role less likely. However, there
is evidence that there is a subgroup of essential
hypertensive patients where a normal or high
percentage distribution to cortical nephrons is
associated with a high cardiac output and normal
effective renal plasma flow [291.
In normal subjects one would expect the two
kidneys to differ very little in their distribution of
blood flow. Similarly in hypertensive patients
without renal disease one would expect a parallel
response by each of the two kidneys to the raised
blood pressure. The good correlation between
Intrarenaljlow distribution in man
percentage cortical flow in the left and right
kidney of the total group confirms these expectations and provides essential evidence of the
validity of the technique.
With this method the contributions of cortical
nephrons and juxtamedullary nephrons to total
nephron function can be measured with only a
small standard deviation which was also found
with simulation studies [231. The method is
readily applicable to the data available from
routine [1231]Hippuran pcamera renography.
Patients with gross pelvi-calyceal tracer retention
and/or moderate-to-severe impairment of renal
function have to be excluded, since their data are
not susceptible to this analysis.
The present non-invasive method of determining the percentage cortical flow by deconvolution,
cross-correlation and curve subtraction forms a
basis for the further investigation of human
intrarenal blood flow distribution.
389
review of analytical factors and their implications in man.
Seminars in Nuclear Medicine, 6 , 193-215.
G., ZILLIG,B., EGLOFF,B. & TRUNIGER,
B. (1980)
I121 SCHULER,
lntrarenal haemodynamics in hypoxic and hypercapnic rats.
In: Radionuclides in Nephrology, pp. 119-124. Proceedings of
the 4th International Symposium, Boston, U.S.A. Ed. Hollenberg, N.K. & Lange, S. Georg Thieme, Stuttgart.
1131 BRITTON, K.E., BROWN,N.J.G. & BLUHM,M.M. (1971)
Xenon washout. Lancer, ii, 822-823.
I141 GLASS,H.I.& DE GARRETA,A.C. (1971) The quantitative
limitations of experimental curve fitting. Physics in Medicine
and Biology, 16,119-130.
1151 SLOTKOFF,L.M., LOGAN, A,, JOSE, P., D’AVELLA,J. &
EISNER,G.M. (197 1) Microsphere measurement of intrarenal
circulation of the dog. Circulafion Research, 28, 158-166.
1161 MOWAT, P., LUPU, A.N. & MAXWELL, M.H. (1972)
Limitations of lllXe washout technique in estimation of renal
blood Row. American Journal of Physiology, 223,682-688.
S., WILSON,C.G. & FERRIS,
T.F.
I171 STEIN,J.H., BOONJARERN,
(1973) Alterations in intrarenal blood Row distribution.
Methods of measurement and relationship to sodium balance.
Circulation Research, 32 (Suppl. I), 61-67.
1181 BROWN,N.J.G. & BRITTON,K.E. (1972) The theory of
renography and analysis of results. In: Radionuclides in
Nephrology, pp. 315-324. Ed. Blaufox. M.D. & FunckBrentano, J.L. Grune and Stratton, New York.
1191 WILKINSON,S.P., SMITH, I.K., CLARKE,M., ARROYO,V.,
RICHARDSON,J., MOODIE,H. & WILLIAMS,R. (1977) Intrarenal distribution of plasma flow in cirrhosis as measured by
transit renography : relationship with plasma renin activity and
sodium and water excretion. Clinical Science and Molecular
Medicine, 52,469-475.
References
1201 REEVE,J. & CRAWLEY,
J.C.W. (1974) Quantitative radioisotope renography : the derivation of physiological data by
I I I BARGER,A.C. (1966) Renal hemodynamic factors in condeconvolution analysis using a single injection technique.
gestive heart failure. Annals of fhe New York Academy of
Clinical Science and Molecular Medicine, 41,3 17-330.
Science, 139,276-284.
1211
BRITTON,K.E., BERNARDI,M., WILKINSONS.P., BROWN,
121 CARRIERE,S., THORBURN,G.D., O’MORCHOE, C.C. &
N.J.G., PEARCE,P.C. & JENNER,R. (1980) Validation of a
BARGER,A. (1966) InIrarenal distribution of blood Row in
non-invasive method of estimating intrarenal plasma flow
dogs during haemorrhagic hypotension. Circulafion Research,
distribution. In: Radionuclides in Nephrology, pp. 204-208.
19,167-179.
Proceedings of the 4th International Symposium, Boston,
I31 HOLLENBERG,
N.K., EPSTEIN,M., BASCH,R.I. & MERRILL.
U.S.A. Ed. Hollenberg, N.K. , & Lange, S. Georg Thieme,
J.P. (1969) No mans land of the renal vasculature. American
Stuttgart.
I
Journal of Medicine, 41,845-854.
1221 HAWKINS,L.A., ELLIOTT,A.T., DYKE,L.J. & BARKER,F.
141 AUKLAND,
K. (1975) Intrarenal distribution of blood Row. Are
(1981)
A
rapid
quantitative
method
for the production of
reliable methods available for measurement in man?
1231-Hippuran.Journal of Labelled Compounds and RadioScandinavian Journal of Clinical and Laboratory Investigapharmaceuficals, 18, 126.
tion, 35,481486.
1231 NIMMON,
C.C., LEE, T.Y., BRITTON,K.E., GRANOWSKA,
M. &
151 TRUNIGER,B. (1980) Editorial review: intrarenal haemoGRUENEWALD,
S.M. (1980) The practical application of
dynamics. In: Radionuclides in Nephrology, pp. 76-79.
deconvolution techniaues to dvnamic studies. In: Proceedinns
Proceedings of the 4th International Symposium, Boston,
of the International ‘Atomic Energy Agency Symposium i n
U S A . Ed. Hollenberg, N.K. & Lange, S. Georg Thieme,
Medical Radionuclide Imaging. Heidelberg, Federal Republic
Stuttgart. .
of
Germany. IAEA, Vienna (In press).
161 THORBURN,
G.D., KOPALD,H.H., HERD,J.A., HOLLENBERG,
1241 VALTIN,H. (1977) Structural and functional heterogeneity of
M., O’MORCHOE, C. & BARGER, A.C. (1963) lntrarenal
mammalian nephrons. American Journal of Physiology, 233,
distribution of nutrient blood flow determined with krypton 85
F49 1-F501.
in the unanaesthetised dog. Circulafion Research, 13, 290J.C.W., GOLDBERG,
A.D., KENNARD,
1251 REEVE,J., CRAWLEY,
307.
C., KENNY.D. & SMITH,I.K. (1978) Abnormalities of renal
171 BRITTON,
K.E. & BROWN,N.J.G. (1971) Clinical Renography,
transport of sodium 0-1 1311liodohippurate (Hippuran) in
pp. 207-2 19, Lloyd-Luke, London.
essential hypertension. Clinical Science and Molecular
181 WILKINSON,S.P., BERNARDI,M., PEARCE,P.C., BRITTON,
Medicine, 55,24 1-247.
K.E., BROWN,N.J.G., POSTON,L., CLARKE,M., JENNER,
R. &
SMITH,H.W. (1951) The Kidney; Sfrucfure and Function in
1261
WILLIAMS,
R. (1978) Validation of ‘transit renography’ for the
Healfh and Disease. Oxford University Press.
determination of the intrarenal distribution of plasma Row:
1271 BRITTON, K.E. (1968) Renin and renal autoregulation, Lancel,
comparison with the microsphere method in the anaesthetised
ii, 329-333.
rabbit and pig. Clinical Science and Molecular Medicine, 55,
1281 LEDINGHAM,W.P. & COHEN, R.D. (1962) Circulatory
277-283.
changes during the reversal of experimental hypertension.
191 BRITTON,K.E. (1979) The measurement of intrarenal flow
Clinical Science, 22,69-77.
distribution in man. Clinical Science, 56, 101-104.
1291 BRITTON, K.E., GRUENEWALD,
S.M. & NIMMON,C.C. (1980)
I101 BLAUFOX,
M.D., FROMOWITZ,
A,, MENG,C.-H. & ELKIN,M.
The
effect of nadolol on cardiac output, total and intrarenal
(1970) Validation of the use of the xenon-I33 to measure
blood Row distribution. In: International Experience wifh
intrarenal distribution of blood Row. American Journal of
Nadolol: a Long-Acfing /HHocking Agent. International
Physiology, 249,440-444.
Congress and Symposium Series, No. 37, pp. 77-86. Royal
1111 HOLLENBERG,
N.K., MANGEL,R. & FUNG,H.Y.M. (1976)
Society of Medicine, London.
Assessment of intrarenal perfusion with radio xenon: a critical