Nephrol Dial Transplant (1998) 13: 165-172
Nephrology
Dialysis
Transplantation
Original Article
On the influence of sample size on the prognostic accuracy and
reproducibility of renal transplant biopsy
Huijian J. Wang, Carl M. Kjellstraud', Sandra M. Cockfield and Kim Solez
Departments of Pathology and Medicine, Faculty of Medicine, University of Alberta
Correspondence and offprint requests to: Dr C. Kjellstrand, VP
Medical Affairs Aksys Ltd, Two Marriott Drive, Lincolnshire, IL
60069, USA.
than has generally been assumed in order for glomerulosclerosis rates to be reproducible or predictive of
outcome. Arteriolar hyalinosis is prognostically
important and shows good reproducibility independent
of sample size. Interstitial fibrosis/tubular atrophy
appear useless as predictors, being of no prognostic
importance and lacking reproducibility. Our finding
clarifies some of the discrepancies found by different
investigators regarding the importance of renal biopsy
in predicting prognosis. Preliminary, our data indicate
that samples containing fewer than 25 glomeruli are
unreliable in determining outcome based on glomerulosclerosis. The importance of our findings which are
based only on chronic lesions, with respect to acute
changes, is unknown.
Key words: sample size; renal biopsy; prognosis; transplant outcome; reproducibility
Introduction
The introduction of renal biopsy into clinical medicine
by Alwall in 1944 [1] has greatly advanced the understanding of renal diseases. While an accurate diagnosis
can sometimes be rendered in a biopsy containing only
a single glomerulus [2], the overall sample size needed
to generate accurate prognostic information is more
controversial. While several authors find a fairly good
correlation between the biopsy and natural outcome
in a variety of renal diseases [3-7], others find it much
less useful [8-10], and in some instances the renal
biopsy did not provide more information than could
be obtained by simple routine clinical investigation
[11].
The size of the renal biopsy is obviously of importance as some lesions can be quite localized, particularly those involving the glomeruli [12]. Although there
are statements in the literature as to how many glomeruli are needed for a satisfactory biopsy [13-16], this
information is mainly personal impression. We found
two articles [17,18] that carefully analysed the size
needed, either for prognostic information, or for reproducibility of findings in successive biopsies: Kellow
© 1998 European Renal Association-European Dialysis and Transplant Association
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
Abstract
Introduction. The minimal specimen size necessary for
accurate interpretation of a renal biopsy has not been
identified. We attempted such a determination by three
different analyses of a collection of biopsies performed
in renal transplants.
Methods. First, we studied the influence of three
lesions (glomerulosclerosis, arteriolar hyalinosis, interstitial fibrosis/tubular atrophy) in 199 baseline biopsies,
obtained at time of transplantation, on transplant
outcome. Secondly, we compared the results from the
three lesions in baseline biopsy with those from 114
subsequent core biopsies in the same patients. Thirdly,
we compared the two baseline biopsies obtained in 118
paired kidneys in cadaver transplantation where both
kidneys were used.
Results. For statistically significant prediction of outcome from glomerulosclerosis, we found that a specimen containing at least 25 glomeruli was needed in the
baseline biopsy. Arteriolar hyalinosis predicted outcome independent of sample size, but became less
important than percentage glomerulosclerosis in predicting outcome if only samples containing more than
25 glomeruli were considered. Interstitial fibrosis/tubular atrophy did not predict the outcome of a kidney,
independent of sample size.
When comparing baseline with subsequent core
biopsies, or with paired baseline biopsies, at least 14
glomeruli were necessary to allow even moderate reproducibility of glomerulosclerosis (Cohen's kappa
> 0.25) and to allow statistical significance (P < 0.05).
The reproducibility of arteriolar hyalinosis was not
dependent on sample size but was reproducible in 80%
of paired baseline biopsies, and in 67% of the comparison of the baseline with core biopsy. Both precision
and significance was lost if sample numbers were
reduced by including only larger samples. There was
no reproducibility in any study of interstitial fibrosis/
tubular atrophy when comparing either baseline with
subsequent biopsy, or paired baseline biopsies.
Summary. Much larger biopsy samples are necessary
H. J. Wang et al.
166
Subjectsandmethods
From July 1991 to April 1995, renal transplants (n = 199),
complete with baseline wedge biopsy, were performed at
University of Alberta Hospital, Canada. The observation
period ended on November 1, 1995; hence, all transplants
were observed for 6 months or until loss of transplant
function (defined as a return to dialysis or death of the
patient).
We investigated the influence of sample size on outcome
in a study of 199 patients, all of whom are included in the
analysis. We had follow-up biopsies on 114 of the 199
transplanted kidneys, and these cases were used in a second
study of the influence of size on reproducibility, comparing
baseline with subsequent core biopsies. In 59 pairs of kidneys
we had baseline biopsies on both kidneys, and these 118
kidney biopsies form the basis for a third comparison of
reproducibility between the biopsies from the paired kidneys.
Patients
There were 70 female patients and 129 males. The mean age
was 42± 13 years (range 11-69). Of 199 transplants, 165
were first transplants, 29 second, and five third. The mean
age of the donors was 37 ± 15 years (range 4-71 years).
There were 145 cadavers and 54 living donors.
Preparation ofbiopsy
Intra-operative baseline biopsies were performed by
obtaining a small wedge from the renal cortex during the
time between vascular anastomosis and ureteral anastomosis.
Follow-up needle biopsies were obtained under ultrasound
guidance using a Truecutf (Baxter Healthcare Corp.,
Deerfield, IL, USA) needle in a biopsy gun. The needle size
was #16 and usually three cores were obtained.
The renal biopsy specimens for light microscopy were fixed
in 10% buffered formalin and then embedded in paraffin.
Sections of 3 urn thickness were cut and then stained with
haematoxylin and eosin, periodic acid Schiff (PAS), silverstain, and Masson trichrome stain.
Scoring ofbiopsies
The number of globally sclerosed glomeruli were counted
and then compared with the total number of glomeruli and
reported as a percentage. We also scored interstitial fibrosis/
tubular atrophy and arteriolar hyaline thickening as binary
data (absent/present). According to the Banff schema [19],
the presence of interstitial fibrous and tubular atrophy was
defined as > 55% of area involved in the process. Multiple
sections of each biopsy were always studied.
Treatment
All patients received triple therapy with prednisone, azathioprine, and cyclosporine. After transplantation, daily cyclosporine levels were determined and the dose regulated
according to blood levels, aiming at a concentration of
300 J.lg/l in the first 2 weeks and then slowly decreasing the
dose, aiming at a concentration of 100 J.lg/l by the end of
the first year. In 49 cases, cyclosporine was not used early
because of delayed graft function. Rejection episodes were
treated with i.v. solu-medrol, and if resistant, with i.v.
OKT-3.
Outcome ofrenal transplant
In total, 33 kidneys were lost, 11 due to patient death, 14
from rejection, five from technical complications, and three
from other causes. In all, 20 kidneys were lost within 6
months. There was no difference in long-term outcome
between related and cadaver kidneys. The 4 year cumulative
survival of living donors was 90 versus 80% for cadaver
donors (P=0.120). Data were therefore pooled for all
donors.
Statistical methods
The data were entered on Statview IV Statistical Program,
version 4.5 and SPSS 6.1 for the Macintosh computer.
Analyses used were Fisher's exact probability test, linear
regression analysis, Cohen's kappa coefficient and Cox proportional hazards analysis. P < 0.05 was regarded to be
significant. All data are mean ± SD. The number of glomeruli
per biopsy was used to decide the biopsy sample size.
Influence of lesion and sample size on outcome of renal
transplants
This was studied in Cox proportional hazards analysis by
successively excluding cases depending on the number of
glomeruli in the biopsy. The first analysis included all biopsies, the second only those with more than six glomeruli, and
then, more than nine, 14, 24 and 29 glomeruli successively.
In this study, the three lesions (percentage glomerulosclerosis,
interstitial fibrosis/tubular atrophy, and arteriolar hyaline
changes) were co-variates. The dependent outcome variable
in this study was transplant function defined as patient being
alive and not on dialysis.
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
et al. [17] in 1959 found reproducibility in 76%,
comparing results from core biopsies with those
obtained at autopsy, but did not specifically look at
biopsy size. Corwin et al. [18] deduced the minimal
size of biopsy from the mathematical model of binomial distribution.
Biopsies of transplanted kidneys offer a unique
opportunity to study this problem. If baseline biopsies
are obtained, it has been shown that chronic lesions,
e.g. glomerulosclerosis, can prognosticate the outcome
in such kidneys. Secondly, if subsequent biopsies are
obtained in the transplanted kidney within a reasonably short time, one can compare the results from
those biopsies with the baseline, and if paired kidneys
are used and baseline biopsies are obtained from both
kidneys in the pair, one has an opportunity of studying
reproducibility between the two biopsies.
At our centre, surgical baseline wedge biopsies are
regularly performed in transplanted kidneys, and
follow-up needle biopsies are frequent. Furthermore,
clinical follow-up of these patients is almost complete.
We felt this offered a unique opportunity to study the
influence of renal biopsy sample size, both on ability
to predict the outcome of transplant, and on
reproducibility.
167
Reliability of renal biopsy
Influence ofsample size on reproducibility offindings
°
Results
General
In the baseline biopsy we identified four lesions: glomerulosclerosis, arteriolar hyalinosis and interstitial fibrosis/tubular atrophy. Since the last two lesions always
overlapped, they were analysed together. The number
of glomeruli in the baseline wedge biopsy was 17.8 ± 12
(range 1-72). In the follow-up biopsy the number of
glomeruli was 1O.4±5.6 (range 1-34) (see Figure 1) .
In 60/199 (30%) of the baseline and 29/114 (25%) of
the follow-up needle biopsies, there was some glomerulosclerosis. The mean percentage of sclerotic glomeruli was 4±7 (range 0-43) and 4±9 (range 0-58) in
baseline and follow-up biopsies respectively. Arteriolar
hyalinosis was present in 31% of baseline biopsies and
Influence ofsample size on prognostic reliability
Percentage glomerulosclerosis
When all baseline biopsies were related to outcome in
Cox proportional hazards analysis, percentage glomerulosclerosis was of no predictive value [relative risk
(RR)=0.998, confidence interval (CI) 0.959-1.038,
P=0.915) . However, as the biopsy size was increased
by excluding cases with few glomeruli, accuracy
increased and reached a maximum when samples
including more than 25 glomeruli were used (RR =
1.056, CI 1.010-1.105, P=0.017). These relations are
demonstrated in Table I and Figure 2. Thus, in order
to predict outcome from glomerular changes, our data
Table I, Relative risk (RR), confidence intervals (CI) and significance of percentage glomerulosclerosis in predicting long-term graft
function related to sample size, expressed as number of glomeruli in
baseline biopsy (Cox proportional hazard analyses)
Included biopsies
n"
P
RR
CI
All cases
> 6 glomeruli
> 9 glomeruli
> 14 glomeruli
> 19 glomeruli
> 24 glomeruli
197
0.915
0.822
0.699
0.173
0.062
0.017
0.998
1.005
1.008
1.031
1.043
1.056
0.959-1 .038
0.965-1.046
0.967-1.051
0.987-1.078
0.998-1.091
1.010-1.105
172
142
97
69
49
"The larger the sample size, the higher is the relative risk of graftloss, as a function of percentage glomerulosclerosis. Statistical
significance (P <0.05j, is first reached only when samples with more
than 24 glomeruli are included.
45
30
40
25
35
30
20
25
1:
g 15
1:
::l
0
0
0
20
15
10
20
30
40
50
lII1JMBER GI.O~n:R UJ.lIN IIASEJ.lNE BIOPSIF$
MF..AN 16.4 :t 11.1. RANGE 1 • 72.
60
70
80
NUMBER GI.OMERUJ.l IN T RANSrl..Al'(J'
Mt:AN IU :t S.6. RANGE 1 • .l4
Fig. 1. Histogram of number of glomeruli per biopsy in baseline and subsequent transplant biopsy.
810PS IF_~
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
In the study of reproducibility of the morphological changes,
we studied the influence of the size of biopsy by successively
excluding cases as described above, and doing repeated l
or Fisher's exact analyses, Cohen 's kappa determinations for
association and linear regression analyses. Percentage sclerotic glomeruli was always studied as a continuous variable
but we studied interstitial fibrosis and arteriolar hyaline
to 3
changes both as continuous variables graded from
and as categorical variables, using either four groupings (0,
I, 2 and 3), or as binary variables (absent or present). Best
accuracy always occurred in the latter two lesions with the
binary absence/presence analyses, and only those results will
be reported.
As new lesions potentially could appear or old ones
disappear in subsequent biopsies, we also carried out all
analyses comparing the baseline with a subsequent core
biopsy in two ways: first we included all biopsies independent
of the time between the baseline and subsequent biopsy, then
we included biopsies only if performed within 60 days of the
baseline biopsy.
37% of follow-up biopsies. There was no relationship
between the findings of arteriolar hyalinosis and the
number of glomeruli (R=0.075 , P=0.303). Interstitial
fibrosis was present in 40 and 35% respectively.
H. J. Wang et al.
168
-r------------------.......,,......--...
••
••
••
0.9
••
••
•• C
1
1.08
•
~
V)
OV)
0.8
I
~V)
~o
•
..:l..:l
U·
V)~
.~
0.7
0<
..:l=:
;:H"
~=:
~O
S~
0.6
..:lZ
~o
~E::
ZU
~ ....
0.5
u~
=:~
~=:
~~
=:~
O~
••.p
••
••
••
••
••
1.04
•
•
0.4
~O
U
~~
••
••
••
••
0.3
> ....
~
0.2
C
0.1
C
••
••
••
••
••
••
••
••
••
••
C
••
•
1.02
1
0 - t - - ' - - ' - - ' - - ' - - + - - ' - - ' - - ' - - ' - - + - - ' - - ' - - ' - - ' - - - r , r - ' - - ' - - ' - - ' - - + - 0.98
40
0
10
20
30
NUMBER OF GLOMERULI
Fig. 2. The probability value for glomerulosclerosis (open squares, hatched line) and relative risk (closed circles, solid line) of subsequent
transplant failure plotted versus the number of glomeruli per biopsy included in the Cox proportional hazard analyses. As only biopsies
with more glomeruli are included, the P value decreases and the relative risk increases. The data indicate that at least 25 glomeruli are
necessary for reliable prediction (P < 0.05). P < 0.05 is reached first when samples containing over 24 glomeruli are included. See text for details.
indicate that samples with fewer than 25 glomeruli are
unreliable.
Comparison ofglomerulosclerosis and arteriolar
hyalinosis in predicting outcome
Arteriolar hyalinosis
In stepwise Cox proportional hazards analysis, arteriolar hyalinosis was more important in predicting outcome, until only samples with more than 25 glomeruli
were included, at which point percentage glomerulosclerosis became the better predictor of outcome.
Arteriolar hyalinosis was the best predictor of
transplant failure when all biopsies were included.
In kidneys without hyalinosis the RR = 0.435,
CI =0.219-0.863, P=0.017. No improvement in prognostic accuracy was obtained when including only
larger samples. This was done by excluding samples
with retrospectively fewer than seven, 10, 15, and 20
glomeruli, as for the analyses of the number of
glomeruli.
Interstitial fibrosis/ tubular atrophy
There was no correlation between interstitial fibrosis/
tubular atrophy and long-term outcome, independent
of sample size. Statistical significance was not reached
in any of the Cox proportional hazards analyses.
Sample size and reproducibility ofbaseline to subsequent
transplant core biopsies
Glomerulosclerosis
The reproducibility of glomerulosclerosis analysed by
linear regression between baseline and subsequent
biopsies is demonstrated in Table 2 and Figure 3. We
successively included larger and larger samples by
excluding cases with too few glomeruli, and also
divided our material to include all biopsies, independent of the length of time between the two biopsies, or
to include only those biopsies that were done within
60 days in order to exclude new lesions on the glomeruli. Better correlation was found in the analysis includ-
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
;::l;::l
..:l0
<z
1.06
PVALUE
Reliability ofrenal biopsy
169
N .41
R • 0.46. P •
30
•
•:I
2S
20
10
•
•
•
•
:
15
0 .003
•
•
5
o --~- ...__•••- • .• -. - -.-----1•5
.J,....- -.---!-
..---r---r---,.-
o
5
,---,---.----,.-
10
15
20
.----,--.--,.- . - --f"
2S
30
35
PERCENT GLOMERULOSCLEROSIS IN BASELINE BIOPSY
Fig. 3. Percentage glomerulosclerosis in transplant core biopsies, analysed by linear regression as the dependent variable of the percentage
glomerulosclerosis in the baseline biopsy. Only 41 samples without time exclusion and all containing more than nine glomeruli per biopsy
are included. Although there is statistical significance, in 16 (39%) of the samples > 10% glomerulosclerosis was found in one of the
biopsies, while there was none in the other paired biopsy.
ing only biopsies done within 60 days, and shows that
precision, demonstrated both by increasing R values
and decreasing P values, increased in all samples until
only samples with 10 or more glomeruli were included.
Beyond this, statistical precision is lost because there
are only few cases that include more than 10 glomeruli
in both biopsies. The correlation coefficient was only
moderate, always less than 0.5, in statistically significant analyses.
Arteriolar hyalinosis
We analysed arteriolar hyalinosis by both sample size
and time between biopsies with Cohen 's kappa and
Fisher's exact probability test (Table 3). The best
statistical significance was obtained when all samples
were included, while agreement increased only moderately when sample size was increased. There was no
increase in agreement , and always less statistical significance when a time limit of < 60 days between biopsies
was used; there was reproducibility in only two-thirds
of the cases; in one-third , a lesion was found in only
one of the two biopsies.
Interstitialfibrosis/tubular atrophy
There was no significant relationship in interstitial
fibrosis/tubular atrophy between baseline and subsequent biopsies in any combination of time limit or
sample size.
Comparison ofbaseline biopsies in paired kidneys
Percentage glomerulosclerosis
Table 2. Reproducibility of glomerulosclerosis in baseline and subsequent biopsies and its relation to sample size and time between
biopsies (linear regression analysis)
Cases included
All
> 6 glomeruli
> 9 glomeruli
> 14 glomeruli
> 19 glomeruli
No time limit
n
Ra
p
108
66
41
12
2
0.17
0.33
0.46
0.57
0.073
0.006
0.003
0.053
The results of agreement of glomerulosclerosis in
paired baseline biopsies studied by linear regression
analysis, are demonstrated in Table 4. There was no
< 60 days between biopsies correlation when all biopsies were included (r=0 .217,
P=O.l081), however, the precision increased with
p
n
Ra
sample size and became statistically significant when
samples with > 14 glomeruli were used (r=0.83,
80
0.24
0.034
P<O.OOI). However, only in 16cases did both biopsies
51
0.40
0.003
include over 14 glomeruli, and the results were greatly
31
0.49
0.005
influenced by a single outlier; if this were excluded
10
0.56
0.095
r=0.025, P=0.858 (Figure 4) .
2
"Although the R value increases with an increasing number of
glomeruli in the biopsies, this becomes statistically insignificant in
samples with over 14 glomeruli as statistical precision is lost due to
the small number of samples with large number of glomeruli.
Arteriolar hyalinosis
Including all 54 paired biopsies with arterioli there was
agreement in 80%, (Cohen's kappa=0.55) of the
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
-5
H. J. Wang et at.
170
Table 3. Reproducibilit y of arteriolar hyalinosis in baseline and subsequent biopsies, and its relation to sample size and time between
biopsies, analyses by Cohen' s kappa
No time limit
Cases
included
All
> 6 glomeruli
> 9 glomeruli
> 14 glomeruli
Less than 60 days between biopsies
11
Percentage
agreement
Kapp a
pa
11
Percentage
agreement
Kappa
pa
104
69
46
15
67
68
67
73
0.34
0.33
0.31
0.46
0.002
0.012
0.041
0.080
76
53
31
12
68
66
67
75
0.38
0.29
0.31
0.45
0.003
0.043
0.132
0.076
' Probability calculated using Fisher' s exact test.
Table 4. Reproducibility of glomerul osclerosis between paired baseline biopsies and its relation ship to sample size ( linear regression
an alysis)
Cases included
11
R
P
All
> 6 glomeru li
> 9 glomeruli
> 14 glomeruli
56
42
31
16
0.22
0.26
0.34
0.83
0.108
0.091
0.061
< 0.001
Interstitial fib rosis/tubular atrophy
Percentage agreement
Kappa
p
All
>6
>9
> 14
54
41
31
17
80
78
71
71
0.55
0.53
0.32
0.47
0.0001
0.004
0.185
0.162
I
•
60
...
50
~
40
.~
For glomerulosclerosis, there was much bett er agreement between the two base line biopsies (Table 4) than
between the baseline and subsequent core biopsy
(Table 2). Thus, the R value for more than 14 glomeruli
between the two baseline biopsies is 0.83, compared
with 0.56 for baseline to core biopsy. On the contrary,
there is no great difference in the reproducibility of
arteriolar hyalinosis. Cohen's kappa is 0.45 (Table 5)
versus 0.47 (Table 3) for baseline/baseline and
N .31
R . 0.34. p . 0.061
•
1!l
s
a
s
30
.
~
e.
20
5
f:l
0
'"
•
Co:
10
0
- 10
·10
0
10
20
30
40
50
60
PERCENTGlm lEUlOSCLE ROSIS BASEll'IE I
Fig. 4. Percentage glomerulosclerosis in paired baseline biop sies, studied by linear regression, and including only biopsies with more than
nine glomeruli. There is poor agreement , and the R value decreases and the P value increases greatly , by excluding the single outlier with
44 and 32% sclerosis.
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
11
There was no repr odu cibility in interstitial fibrosis/
tubular at rophy between paired baseli ne biops ies, independent of sample or time.
Comparison offindings in the baseline/baseline and
baseline/core analyses
Table S. Reproducibility of ar teriolar hyalinosis between paired
baseline biopsies and its relation to sample size. Analysis calculated
using Cohen 's kappa and probability by Fisher's exact test
Cases included
biopsies (P =O .OOOI) . Both precision and probability
decreased as the sample number decreased by excluding
samples based on nu mb er of glome rul i ( Ta ble 5).
Reliability of renal biopsy
baseline/core biopsies respectively, when samples
including more than 14 glomeruli are considered.
Discussion
Wehrmann's patients. There were no correlation's
between interstitial fibrosis/tubular atrophy either
between baseline and subsequent biopsies, nor in paired
baseline biopsies.
We believe that this evaluation of the importance of
size of the renal biopsy clarifies the many differences
that have been found between morphological changes
in the kidney, and outcome [3-15]. Our findings agree
well with the conclusions of Corwin et al. [18] and
Madaio [21]. Based on a mathematical model of
binomial distribution, they thought at least 20 glomeruli were necessary for reasonable clinical prediction of
focal glomerulosclerosis and of the severity of lupus
nephritis. They did not discuss other findings such as
hyalinosis or fibrosis. Our findings are quantitatively
different from those of Gaber et al. [22] who found
samples containing more than 10 glomeruli to be
prognostically important. Their analysis is different
from ours, in that time to graft loss was not considered
and the sample size much smaller.
Each of the three analyses performed by us has
certain strengths and weaknesses. In our trial to relate
outcome to morphological changes there are many
other factors that influence the fate of a transplanted
kidney, and these clinical factors such as tissue-type,
drug toxicity, and acute renal failure, make this analysis
weak. In the comparison of baseline to core transplant
biopsy, obviously new lesions may appear secondary
to insults against the kidney after transplantation or
old lesions perhaps disappear. However, as we focused
only on chronic changes and also used a 60 day timelimited analysis, we believe that the data here are
reliable. It appears to us unlikely that any of these
chronic changes should either form or disappear within
the 60 day time limit of our analysis [14,15]. The
comparison of a superficial wedge biopsy with the
deeper core biopsy may be affected by quantitative
differences. These may include a higher percentage of
sclerosed glomeruli in the superficial wedge biopsy
than in the deeper core biopsy obtained with a needle
[23]. In fact, we found that glomerulosclerosis was less
reproducible in the comparison of baseline wedge
biopsy to core needle biopsy than hyalinosis.
The easiest study to interpret is when two paired
baseline biopsies are compared. However, although
the latter is morphologically the most sound, it suffers
from the disadvantage of containing the fewest number
of biopsies, and therefore statistical instability markedly increased as increasing sample sizes were included
with subsequent decrease in the number of biopsies.
Perhaps the greatest strength in our three analyses
is, that although there may be some quantitative
differences, all of them essentially showed the same
findings: for judgment of glomerular changes a large
number of glomeruli (15-25) are necessary; for prognostication and reliability of arteriolar hyalinosis,
almost any size of renal biopsy will suffice; and finally,
that interstitial fibrosis/tubular atrophy has little value
either for prognosis or reproducibility.
The second limitation of our study is that only
chronic changes in the kidney were studied. Naturally,
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
In this study, we examined the importance of sample
size in renal transplant biopsy in three different ways.
The succinct findings are that, to make any prognostic
conclusions from glomerulosclerosis, more than 25
glomeruli are necessary, but for arteriolar hyalinosis
sample size is not important. Interstitial fibrosis/
tubular atrophy does not seem to carry prognostic information.
All three separate analyses of glomerulosclerosis, its
predictability on outcome, its reproducibility between
baseline and subsequent biopsies, as well as the comparisons of the two baseline biopsies, are in agreement.
More than 25 glomeruli are necessary to predict outcome from glomerulosclerosis, and more than 14 to
result in a coefficient of determination (R 2 ) of > 0.25
in the comparison of baseline to baseline or to subsequent biopsies. To us, these data indicate that glomerulosclerosis is a localized lesion in the kidney, and
that many more glomeruli are necessary for evaluation
of lesions in the glomeruli than generally assumed.
Arteriolar hyalinosis is different. In all three analyses, the highest precision was reached statistically
when all samples were included, and the statistical
significance rapidly declined as the number of samples
was reduced, subsequent to the increase in sample size.
We interpret these findings as showing that arteriolar
hyalinosis is a diffuse lesion, and that almost any renal
biopsy in which a blood vessel is found is diagnostic
for this lesion. There were discordant findings in 10%
of the comparison of baseline with baseline and in
20-30% of baseline with subsequent core biopsy. These
findings at first appears to be in disagreement with the
observation by Bell [20] who found renal arteriolar
hyalinosis to be unevenly distributed in autopsies.
However, he studied only hypertensive patients, and
such patients are often excluded by transplant teams.
Thus, the age-related lesions seen by us may be more
diffuse, and the agreement between baseline to baseline
and baseline to core biopsy, was better. The 50%
discrepancy in the findings between paired baseline
and core biopsy is compatible with sampling error. It
may also be due to the qualitative difference between
the more superficial baseline wedge biopsy and the
deeper core biopsy or to factors which cause new
arteriolar hyaline change (cyclosporine/hypertension)
or resolution of lesion of donor organs [14,15].
Interstitial fibrosis/tubular atrophy was of no value
in predicting outcome of the kidneys, independent of
sample size. This observation is at variance with previous observations, particularly those by Wehrmann
et al. [3]. These different findings probably reflect the
fact that scarring, on a vascular atherosclerotic basis,
as present in our patients, is quite patchy and therefore
highly prone to sampling error, while it may be more
diffuse in glomeulonephritis, the disease present in
171
H. J. Wang et al.
172
References
1. Alwall N. Aspiration biopsy. Acta Med Scand J Int Med 1952;
143: 430
2. Brewer DA. Renal Biopsy. 2nd edn. Williams and Wilkins Co,
Baltimore 1973; 3
3. Wehrmann M, Bohle A, Held H, Schumm G, Kendziorra H,
Pressler H. Long-term prognosis of focal sclerosing glomerulonephritis. An analysis of 250 cases with particular regard to
tubulointerstitial changes. Clin Nephrol1990; 33: 1I5-122
4. Mbakop A, Chatelenat F. Renal puncture biopsy in the aged
subject: a propos of 119 cases. Ann Pathol 1985; 5:101-105
5. Whiting-O'Keefe Q, Henke JE, Shearn MA, Hopper J, Biava
CG, Epstein WV. The information content from renal biopsy
in systemic lupus erythematosus. Ann Int Med 1982;96: 718-723
6. Waltzer WC, Miller F, Arnold A, Jao S, Anaise D, Rapaport
FT. Value of percutaneous core needle biopsy in the differential
diagnosis of renal transplant dysfunction. J Urol 1987; 137:
1I17-1I21
7. Richards NT, Darby S, Howie AJ, Adu D, Michael
J. Knowledge of renal histology alters patient management in
over 40% of cases. Nephrol Dial Transplant 1994; 9: 1255-1259
8. Delaney V, Ling BN, Campbell WG et al. Comparison of fineneedle aspiration biopsy, Doppler ultrasound, and radionuclide
scintigraphy in the diagnosis of acute allograft dysfunction in
renal transplant recipients: sensitivity, specificity, and cost analysis. Nephron 1993; 63: 263-272
9. Gladman DD, Urowitz MB, Cole E, Ritchie S, Chang CH,
Churg J. Kidney biopsy in SLE; 1. A clinical-morphologic
evaluation. Q J Med 1989; 73: 1I25-1133
10. Hogan SL, Nachman PH, Wilkman AS, Jennette JC, Falk
RJ. Prognostic markers in patients with antineutrophil cytoplasmic autoantibody-associated microscopic polyangitis and
glomerulonephritis. J Am Soc Nephrol1996; 7: 23-32
11. Malleson PN. The role of the renal biopsy in childhood onset
systemic lupus erythematosus: a viewpoint. Clin Exp Rheumatol
1989; 7: 563-566
12. Fuiano G, Comi N, Magri P et al. Serial morphometric analysis
of sclerotic lesions in primary 'focal' segmental glomerulosclerosis. J Am Soc Nephrol1996; 7: 49-44
13. Oberho1zer M, Torhorst J, Perret E, Mihatsch MJ. Minimum
sample size of kidney biopsies for semiquantitative and quantitative evaluation. Nephrology 1983; 34: 192-195
14. Morozumi K, Thiel G, Albert FW, Banfi G, Gudet F, Mihatsch
MJ. Studies on morphological outcome of cyclosporineassociated arteriolopathy after discontinuation of cyclosporine
in renal allografts. Clin NephroI1992; 38: 1-8
15. Solez K, Racusen LC, Marcussen N et al. Morphology of
ischemic acute renal failure, normal function, and cyclosporine
toxicity in cyclosporine-treated renal allograft recipients. Kidney
Int 1993; 43: 1058-1067
16. Pirani CL. Evaluation of kidney biopsy specimens. In: Tisher
CC, Brenner BM (eds). Renal Pathology with Clinical and
Functional Correlation. Vol 1. J.B.Lippincott Co, Philadelphia;
1989
17. Kellow WF, Cotsonas NJ, Chomet B, Zimmerman
MJ. Evaluation of the adequacy of needle-biopsy specimens of
the kidney. Arch Int Med 1959; 104: 353-359
18. Corwin HL; Schwartz MM; Lewis EJ. The importance of sample
size in the interpretation of the renal biopsy Am J Nephrol1988;
8: 85-89
19. Solez K, Axelsen RA, Benediktsson H et al. International
standardization of criteria for the histologic diagnosis of renal
allograft rejection: the Banff working classification of kidney
transplant pathology. Kidney Int 1993; 44: 41I-422
20. Bell ET. Renal Diseases. l st edn. Lea and Febiger,
Philadelphia, 1946
21. Madaio MP. Renal biopsy. Kidney Int 1990; 38: 529-543
22. Gaber LW, Moore LW, Alloway RR, Amiri H, Ver SR, Gaber
AO. Glomerulosclerosis as a determinant of posttransplant
function of older donor renal allografts. Transplantation 1995;
60: 334-339
23. Schmidt A, Baker R. Renal biopsy in children: analysis of 61
cases of open wedge biopsy and comparison with percutaneous
biopsy. J Uro11976; 116: 79-80
24. Olsen S. Renal histopathology in various forms of acute anuria
in man. Kidney Int 1976: 10: S2-S8
25. Houghton DC, Hartnett M, Campbell-Boswell M, Porter G,
Bennett W. A light and electro1 microscopic analysis of gentamicin nephrotoxicity in rats. Am J Patho11976: 83: 589-612
26. Solez K, Morel-Maroger L, Sraer J-D. The morphology of
'acute tubular necrosis' in man: analysis of 57 renal biopsies and
a comparison with the glycerol model. Medicine (Baltimore)
1979; 58: 362-376
Received for publication: 3.1.97
Accepted in revised form: 5.9.97
Downloaded from http://ndt.oxfordjournals.org/ by guest on October 6, 2014
there was no acute disease found in the baseline
biopsies of the transplanted kidneys, therefore, one
cannot extrapolate our findings to biopsies of acute
ongoing disease in the kidney. It has been shown that
a single glomerulus may suffice to make diagnosis [2].
However similar quantitative influences found by us
may potentially be present in acute diseases of the
kidney. Thus, the correlation between histology and
renal function and acute tubular necrosis has always
been poor [24-26]. This poor correlation and some of
the confusion of the prognostic reliability and the
controversy if different lesions of lupus nephritis can
change into each other [4,5,9,11], may be partially
explained by too small sample size in analyses. Since
the earliest countings of the number of glomeruli
obtained in a needle biopsy, the total number found,
as a mean, is 10 [2]. This has remained unchanged
and is the same as we found our in our core biopsies.
These 10 glomeruli represent only 0.001% of the 1 x 106
glomeruli in the kidney.
A practical conclusion from our findings is that the
average size of baseline biopsies should be increased
so that better representation is obtained. The size of
the biopsy needle and the number of cores that should
be obtained in a needle biopsy of the kidney, also
needs reconsideration. Based on our observations in
transplants and Corwin's general mathematical analyses, we conclude that there must be a minimum of
20-25 glomeruli for reliable prognosis and probably
at least 15 for true representation of chronic changes.