![Graph shows receiver operating characteristic curves for prediction of acute rejection according to percentage changes in PBV () and of Doppler US composite index calculated by combining the percentage changes in PBV and SPI (‚). Both parameters (change in PBV and change in SPI) had good accuracy for prediction of acute rejection. Best point of curve was achieved with a percentage decrease cutoff of SPI of 25% (specificity, 89% [16 of 18]; sensitivity, 86% [six of seven]). Dashed line is the line of no diagnostic accuracy.](https://www.researchgate.net/profile/Roberto-Merenda/publication/7964891/figure/fig1/AS:601637211107356@1520452897988/Graph-shows-receiver-operating-characteristic-curves-for-prediction-of-acute-rejection_Q320.jpg)
Content uploaded by Roberto Merenda
Author content
All content in this area was uploaded by Roberto Merenda on Jan 10, 2016
Content may be subject to copyright.
Massimo Bolognesi, MD,
PhD
David Sacerdoti, MD
Claudia Mescoli, MD
Valeria Nava, MD
Giancarlo Bombonato, MD
Carlo Merkel, MD
Roberto Merenda, MD
Paolo Angeli, MD, PhD
Massimo Rugge, MD
Angelo Gatta, MD
Published online before print
10.1148/radiol.2352040506
Radiology 2005; 235:651–658
Abbreviations:
OLT ⫽orthotopic liver
transplantation
PBV ⫽portal blood velocity
RAI ⫽rejection activity index
SPI ⫽splenic pulsatility index
1
From the Department of Clinical and
Experimental Medicine (M.B., D.S.,
V.N., G.B., C. Merkel, P.A., A.G.), De-
partment of Oncology and Surgical
Sciences (C. Mescoli, M.R.), and Insti-
tute of General Surgery (R.M.), Clinica
Medica 5, Dipartimento di Medicina
Clinica e Sperimentale, Policlinico Uni-
versitario, University of Padova, Via
Giustiniani 2, 35128 Padova, Italy. Re-
ceived March 17, 2004; revision re-
quested May 25; revision received
June 8; accepted July 20. Address
correspondence to M.B. (e-mail:
massimo.bolognesi@unipd.it).
Authors stated no financial relation-
ship to disclose.
Author contributions:
Guarantors of integrity of entire study,
A.G., M.B.; study concepts, M.B., D.S.;
study design, M.B., D.S., G.B., M.R.;
literature research, M.B.; clinical stud-
ies, M.B., D.S., V.N., G.B., C. Mescoli,
R.M., P.A.; data acquisition, M.B.,
D.S., C. Mescoli., V.N., G.B., R.M.,
P.A., M.R.; data analysis/interpreta-
tion, M.B., D.S., C. Merkel, M.R., A.G.;
statistical analysis, M.B., C. Merkel,
M.R.; manuscript preparation, M.B.,
V.N., C. Mescoli; manuscript defini-
tion of intellectual content, M.B., D.S.,
M.R., C. Merkel; manuscript editing,
M.B., D.S.; manuscript revision/re-
view, M.B., D.S., C. Merkel, M.R.,
A.G.; manuscript final version ap-
proval, all authors
©
RSNA, 2005
Acute Liver Rejection:
Accuracy and Predictive
Values of Doppler US
Measurements—Initial
Experience
1
PURPOSE: To prospectively evaluate accuracy and predictive values of Doppler
ultrasonographic (US) measurement of portal blood velocity (PBV) and splenic
pulsatility index (SPI) in diagnosis of clinically relevant acute rejection in patients
with clinicobiochemical hepatic dysfunction after orthotopic liver transplantation
(OLT).
MATERIALS AND METHODS: Study was approved by the institutional review
board, and protocol conformed to ethical guidelines of Declaration of Helsinki.
Patient informed consent was obtained. In 27 patients with OLT (23 men, four
women; mean age, 48 years; range, 27–64 years), PBV and SPI were measured at
Doppler US within 48 hours before or after liver biopsy for clinically suspected acute
rejection. Biopsy specimens were assigned scores according to Banff method, and
rejection activity index (RAI) was calculated. RAI score of 4 or greater was considered
clinically relevant acute rejection. Doppler US parameters were analyzed as absolute
values and as percentage point changes with respect to values obtained at last
examination before rejection was suspected. Information from two Doppler US
parameters was combined; Doppler US composite index was calculated. Statistical
tests were conducted to assess accuracy, sensitivity, specificity, and predictive values
of Doppler US parameters in diagnosis of graft rejection.
RESULTS: Clinically relevant acute rejection was diagnosed in nine patients. Me-
dian time from OLT until histologic diagnosis of acute rejection was 8 days (range,
5–20 days). Rejection was associated with a marked reduction in mean PBV (⫺43% ⫾
5 [standard error of the mean]) and a slight increase in SPI (⫹12% ⫾16). The
calculated Doppler US composite index was strictly related to severity of rejection
(P⬍.001). When applied retrospectively, this index had good accuracy (88%) for
prediction of rejection (specificity, 89%; sensitivity, 86%; negative predictive value,
94%).
CONCLUSION: During the first weeks after OLT, a marked decrease in PBV
associated with increased SPI supports suspicion of clinically relevant acute rejection.
©
RSNA, 2005
Acute rejection, an early complication of orthotopic liver transplantation (OLT), occurs in
more than 50% of patients (1,2) and can be diagnosed only by means of a liver biopsy
(3–5).
Acute rejection is characterized by a number of histologic abnormalities, and most of
them are potentially able to increase portal venous resistance and to obstruct portal blood
flow. Packing of the portal tract by inflammatory cells (ie, portal inflammation) (6– 8),
supraendothelial and subendothelial lymphocytic infiltration in either portal or central
veins (ie, endotheliitis) (8,9), centrilobular hepatocyte necrosis (10), and central hepato-
cellular ballooning (swelling) (11) all may increase portal venous resistance. Impairment of
Ultrasonography
651
R
adiology
sinusoidal blood flow caused by mechan-
ical obstruction, local precipitation of co-
agulation, disintegration of endothelial
integrity with an increase in interstitial
pressure (12), sinusoidal infiltration, and
arteritis (11) also may participate in the
increase in portal venous resistance.
All the previously mentioned morpho-
logic alterations are included in a well-
codified histologic grading scheme (ie,
Banff grading scheme for acute rejection)
used in the histologic assessment of acute
rejection after OLT (7). After we com-
bined the values obtained with assign-
ment of scores to portal inflammation,
bile duct inflammation damage, and ve-
nous endothelial inflammation, a rejec-
tion activity index (RAI) score was ob-
tained; this score was used to distinguish
clinically irrelevant mild rejection (RAI
score ⬍4; range, 0–9) from clinically rel-
evant acute rejection (RAI score ⱖ4;
range, 0–9). Mild clinically irrelevant re-
jection usually does not require additional
immunosuppressive therapy, whereas
clinically relevant moderate-severe rejec-
tion does. Therefore, a noninvasive
method for selection of patients at in-
creased risk for clinically relevant acute
rejection would be useful, but until now
the methods proposed have not demon-
strated sufficient accuracy (4,13,14).
Acute rejection after OLT results in in-
creased portal pressure (15,16), probably
caused by an increase in portal vascular
resistance. Portal blood velocity (PBV)
and arterial splenic resistance indexes are
two Doppler US parameters that are mod-
ified by an increase in portal venous re-
sistance. In this condition, PBV decreases
(17,18), while splenic pulsatility index
(SPI) increases (19). Indeed, SPI probably
reflects not only the resistance in the ar-
terial and capillary bed of the spleen but,
more likely, also the sum of downstream
resistance, including the splenic arterial
and capillary bed resistance, as well as the
splenic and portal venous and hepatic
vascular resistance (18).
After OLT, the sudden normalization
of portal vascular resistance provokes an
increase in PBV, with values greater than
those in the normal range. The early in-
crease, however, slowly subsides in the
following months (20). SPI, which is re-
lated to portal venous resistance (19), de-
creases to a normal level immediately af-
ter OLT, and it further progressively
decreases in the following months (20). It
remains increased if there is a complica-
tion that affects portal venous resistance
(21).
We hypothesized that if acute rejec-
tion modifies portal venous resistance, a
modification in the time course of PBV
and SPI may be expected, and the detec-
tion of such a modification might suggest
the presence of acute rejection, which
would prompt the physician to perform a
liver biopsy. Thus, the purpose of our
study was to prospectively evaluate the
accuracy and predictive values of Dopp-
ler ultrasonographic (US) measurement
of PBV and SPI in the diagnosis of clini-
cally relevant acute rejection in patients
with clinicobiochemical hepatic dysfunc-
tion after OLT.
MATERIALS AND METHODS
Patients and Protocol Design
This prospective pilot study was con-
ducted from September 1995 to March
2002 and was approved by our institu-
tional review board. The study protocol
conformed to the ethical guidelines of
the Declaration of Helsinki. Patient in-
formed consent was obtained after the
nature of the procedures had been fully
explained.
PBV and SPI were measured in a group
of 27 patients (23 men, four women;
mean age, 48 years; range, 27–64 years)
who had undergone OLT and were re-
ferred to our department. In these pa-
tients, a liver biopsy was performed for
clinically suspected acute rejection.
Liver biopsy was performed whenever
there was clinical and biochemical dete-
rioration suggestive of liver dysfunction:
increased transaminase, bilirubin, alka-
line phosphatase, and ␥-glutamyltrans-
ferase levels. Independent of suspected
rejection, all patients who had under-
gone OLT and who were referred to our
department were evaluated with duplex
Doppler US at least once within the first 3
days after OLT, then 7–15 days after, and
every 1–3 months thereafter during the
1st year. Duplex Doppler US examination
also was performed every time the clini-
cal situation (ie, malaise, fever, or bio-
chemical deterioration suggestive of liver
dysfunction) necessitated it.
Patients were included who were
18– 65 years old and had alcoholic, virus-
related, or primary biliary cirrhosis or pri-
mary sclerosing cholangitis. Also in-
cluded were patients who underwent
OLT and a liver biopsy, because they were
suspected of having acute rejection.
Within 48 hours either before or after
biopsy, a splanchnic Doppler US exami-
nation was performed. Only patients in
whom liver biopsy was performed within
the first 5 months after OLT were in-
cluded, since this is the period during
which almost all cellular rejections arise
(2,11,22). Only the first instance of a pos-
sible rejection episode was included in
each patient; repeated biopsies in the
same patients were not considered.
Patients were excluded on the basis of
the following: a biopsy was performed in
the first 3 days after OLT, because cellular
rejection is less frequent in the first days
after OLT (7) and because after the 4th
day all patients had undergone at least
one basal Doppler US examination after
OLT; low quality of visualization (which
was due to meteorism, obesity, surgical
dressing, inability of the patients to hold
their breath) at Doppler US, which was
insufficient to obtain a reliable measure-
ment of at least PBV; and vascular alter-
ations, such as portal or arterial throm-
bosis or stenosis, which are known to
modify Doppler US parameters (23,24).
Indeed, Doppler US is the primary
screening technique for detection of vas-
cular complications of hepatic transplan-
tation (23,24). Moreover, patients were
excluded a posteriori if fewer than five
portal tracts were present in the consid-
ered biopsy samples.
Of 94 patients who underwent OLT
and who were referred to our department
during the study period, 85 underwent
transplantation for cirrhosis or sclerosing
cholangitis (65 men, 20 women; mean
age, 50 years; range, 27–64 years). Fifty-
five of 85 patients underwent liver biopsy
after the 4th day after OLT for the onset
of signs of liver dysfunction. Among
them, 28 were excluded from the study:
in 13, visualization was insufficient; in
13, a vascular complication was detected
(three with hepatic artery thrombosis,
five with hepatic artery stenosis, one
with portal thrombosis, and four with
portal stenosis); in two, fewer than five
portal tracts were detectable in speci-
mens, which resulted in their a posteriori
exclusion from the final analysis. Thus,
27 patients were included for the final
analysis. Two surgical techniques were
used. The “piggyback” technique (25)
was used in five patients, whereas tempo-
rary portocaval anastomosis with preser-
vation of inferior vena caval flow accord-
ing to Belghiti et al (26) was used in the
other 22 patients. The hepatic artery was
reconstructed with end-to-end anasto-
mosis in 23 patients and with aortohe-
patic bypass grafting by using the infra-
renal technique in four patients. The
portal vein was reconstructed with end-
to-end anastomosis in all patients. Acute
rejection was treated with high-dose
methylprednisolone (Solu-Medrol; Phar-
macia Italia, Milan, Italy) (27).
652 䡠Radiology 䡠May 2005 Bolognesi et al
R
adiology
Doppler US Examination
Doppler US parameters were measured
after overnight fasting. Initially, color
Doppler US (Sonolayer SSA-270A; Toshiba,
Tokyo, Japan) was performed with a 3.75-
MHz sector electronic probe. After October
1998, color Dopper US (HDI 5000; ATL,
Seattle, Wash) was performed with a
broadband curved-array transducer (C5-2
40R).
Parameters were measured in patients
during suspended normal respiration.
PBV was evaluated as time-averaged max-
imum velocity multiplied by the coeffi-
cient 0.57, with the assumption that the
portal velocity profile was parabolic
(19,28). The Doppler sample volume cov-
ered 50% of the diameter of the portal
vein, and care was taken to maintain the
Doppler angle at 55° or just smaller than
that. Doppler waveform calculation was
obtained with imaging during two car-
diac cycles, and a manual tracing of the
Doppler waveform, which followed the
outside borderline of the maximum fre-
quency shift, was obtained. The wall fil-
ter was fixed at 100 Hz.
The splenic arterial Doppler resistance
index was measured as follows: SPI ⫽(V
ps
⫺
V
ed
)/V
m
(29), where V
ps
is peak systolic
velocity, V
ed
is end-diastolic velocity, and
V
m
is mean velocity. SPI values were mea-
sured near the hilum of the spleen (19,30).
The sample volume of the Doppler system
was placed in one of the main branches of
the splenic artery, about 0.5–1.0 cm inside
the organ. To minimize sampling error, the
Doppler spectrum was increased by using
the lowest frequency-shift range possible
without aliasing, and the wall filter was set
at 100 Hz.
Doppler US operators (M.B., D.S., and
G.B., with 8–10 years of experience) were
blinded to the results of the biochemical
tests and of the biopsies in the patients.
The images were interpreted prospec-
tively by one of the operators at the time
of evaluation. To decrease interobserver
variability in the Doppler US results, all
three US operators participated in a co-
operative training program according to
Sabba` et al (28) and Sacerdoti et al (30)
before the beginning of the study.
Histologic Assessment of Acute
Rejection
The diagnosis of acute (ie, cellular) re-
jection was considered on the basis of
clinical data (abnormal results of liver
function tests, increase in cholestatic
markers, and presence of eosinophilia)
and was established with analysis of a
core-needle biopsy specimen (7,8,10,11,
22,31) that was obtained with a modified
Menghini aspiration biopsy set and a 16-
gauge needle (Surecut; TSK Laboratory,
Dublin, Ireland). Not fewer than five por-
tal tracts were present in all the consid-
ered biopsy samples (range, five to 13).
Although the number of five portal tracts
accepted in this study is fewer than that
required for a reliable assessment of
chronic liver diseases, it can be consid-
ered reliable in the clinicopathologic set-
ting of acute rejection of the transplant
(32). All biopsy specimens were simulta-
neously examined by two pathologists
(M.R. and C. Mescoli, with 25 and 3 years
of experience, respectively), who were
unaware of Doppler US results. They de-
termined a diagnosis and a histologic
score (RAI score) in consensus. The RAI
was calculated by attributing a separate
score (RAI score of 1–3) to portal inflam-
mation, bile duct inflammation damage,
and venous endothelial inflammation,
according to the severity of the involve-
ment. The total score is the sum of the
scores for those single components (7).
Patient Grouping and Statistical
Analysis
On the basis of the histologic results,
patients were grouped according to the
presence or absence of clinically relevant
acute rejection, defined as an RAI score of
4 or greater. Patients without histologi-
cally proved acute rejection or with an
RAI score of less than 4 were arbitrarily
assumed to be controls. Results were re-
ported as the mean ⫾standard error of
the mean. Time of biopsy after OLT was
reported as median and range. Differ-
ences between patients with and without
clinically relevant acute rejection were
evaluated with the Student ttest for un-
paired samples. Differences between the
values obtained before and at the time of
biopsy were evaluated with the Student t
test for paired samples. Correlations were
evaluated with the Spearman nonpara-
metric correlation coefficient. A differ-
ence with a Pvalue less than .05 was
considered significant.
PBV and SPI recorded at the time of the
episode of suspected acute rejection were
analyzed as absolute values and as per-
centage point changes with regard to val-
ues obtained in the previous evaluation
(the latest evaluation performed before
the episode of suspected acute rejection).
To compare percentage changes in PBV
and in SPI, the data were treated as
though they were continuous, and a ttest
was applied.
Data were analyzed for comparison of
patients with clinically relevant acute re-
jection (RAI score ⱖ4) with all the other
patients. A further analysis also was per-
formed for comparison of the modifica-
tions of Doppler US parameters with the
severity of the single histologic parame-
ters of the RAI score, that is, with the
degree of portal inflammation, bile duct
inflammation damage, or venous endo-
thelial inflammation.
A new index was calculated in which
the modification of both PBV and SPI was
combined and was compared with the RAI
score. This original composite index (Dopp-
ler US composite index) was calculated by
subtracting the percentage change in SPI
from the percentage change in PBV.
To better analyze the relationships be-
tween change in Doppler US parameters
and RAI, patients were classified into
three groups according to the degree of
RAI as follows: those with no histologic
activity of rejection, those with mild ac-
tivity, and those with clinically relevant
acute rejection. One-way analysis of vari-
ance with the posttest for linear trend
between the mean and the column num-
ber was then performed. This test verifies
whether there is a trend among groups
arranged in a natural order so that the
values increase (or decrease) as one
moves from left to right across the col-
umns.
Sensitivity, specificity, accuracy, and
predictive values of Doppler US parame-
ters for the presence of acute rejection
were determined according to standard
formulas, and receiver operating charac-
teristic curves were constructed. Areas
under the curve and standard errors of
areas under the curve were calculated and
compared according to Hanley and Mc-
Neil (33,34). The number needed to diag-
nose was also calculated according to
Batstone (35). This number, which is cal-
culated as 1/[sensitivity ⫺(1 ⫺specific-
ity)], indicates the number of tests that
need to be undertaken to gain a positive
response for the presence of disease and
gives a ready comparison between tests
(35). Statistical analysis was performed
with commercially available software
programs (BMDP Statistical Software, ver-
sion 7.0, 1993, BMDP, Berkeley, Calif;
Prism, version 2.01, 1996, GraphPad
Software, San Diego, Calif).
RESULTS
Table 1 shows the histologic findings.
With consideration of all 27 patients, the
median time between OLT and biopsy
Volume 235 䡠Number 2 Acute Liver Rejection 䡠653
R
adiology
was 10.0 days. Doppler US data collected
at the time of biopsy were compared with
values obtained 3–28 days before biopsy.
The characteristics of the included patients
are reported in Table 2. Measurements of
PBV were available in all patients. Because
of clinical and/or technical reasons, the SPI
measurement was not available in two pa-
tients (both with acute rejection).
Portal Blood Flow Velocity
PBV was not significantly different in
patients with and patients without clini-
cally relevant acute rejection (RAI score ⱖ
4) (21.8 cm/sec ⫾3.4 [standard error of
the mean] vs 24.9 cm/sec ⫾2.0, P⫽.39).
When one compares values before biopsy
with those at biopsy, PBV changed from
38.9 cm/sec ⫾4.5 to 21.8 cm/sec ⫾3.4
(P⬍.001) in patients with acute rejection
(RAI score ⱖ4) and from 31.3 cm/sec ⫾
2.8 to 24.9 cm/sec ⫾2.0 (P⫽.034) in
patients with an RAI score of less than 4.
The mean percentage decrease in PBV
was larger in patients with acute rejection
(RAI score ⱖ4): ⫺43% ⫾5 versus ⫺15% ⫾
7(P⫽.013). A mean difference also was
obtained when the patients were classi-
fied according to the presence of portal
infiltration (⫺38% ⫾6vs⫺14% ⫾8, P⫽
.029). When we considered all 27 pa-
tients, a correlation was present between
RAI score and percentage change in PBV
(r⫽⫺0.41, P⫽.035). A significant dif-
ference also was obtained when we con-
sidered the portal inflammation score or
the bile duct inflammation damage score
instead of the RAI score (r⫽⫺0.43, P⫽
.023 and r⫽⫺0.41, P⫽.033, respec-
tively).
SPI Value
Mean SPI values were not significantly
different in patients with and patients
without acute rejection (1.01 ⫾0.14 vs
0.84 ⫾0.04, respectively, P⫽.15). When
we compared mean values before biopsy
with those at biopsy, SPI changed from
0.92 ⫾0.10 to 1.01 ⫾0.14 (P⫽.56) in
patients with acute rejection (RAI score
of ⱖ4) and from 0.96 ⫾0.07 to 0.84 ⫾
0.04 (P⫽.058) in patients with an RAI
score of less than 4. The mean percentage
change between the values before and
those at biopsy was not significantly dif-
ferent between patients with and without
acute rejection (⫹12% ⫾16 vs ⫺9% ⫾5,
respectively, P⫽.12). On the contrary,
patients with portal inflammation (RAI
score of 1–3) or bile duct inflammation
damage (RAI score of 1 or 2) had a signif-
icant increase in mean SPI, whereas pa-
tients without these signs showed a de-
crease in mean SPI (⫹14% ⫾12 vs ⫺15% ⫾
4, P⫽.011; ⫹9% ⫾8vs⫺19% ⫾5, P⫽
.013, respectively). When we considered
all biopsies, a direct correlation was
present between RAI score and percent-
age change in SPI (r⫽0.49, P⫽.012).
The correlation was confirmed when we
considered the portal inflammation score
or the bile duct inflammation damage
score instead of the RAI score (r⫽0.43,
P⫽.030 and r⫽0.52, P⫽.007, respec-
tively).
TABLE 1
Histologic Findings at Biopsy
Patient*
Histologic Score
RAI
Score Histologic Diagnosis
Portal
Inflammation
Bile Duct
Inflammation
Damage Endotheliitis
1 0 0 0 0 Mild cholangitis or hepatocellular steatosis, no rejection
2 0 0 0 0 Minimal portal or lobular hepatitis, no rejection
3 0 0 0 0 Minimal lobular alterations, no rejection
4 0 0 0 0 Minimal lobular alterations, no rejection
5 0 0 0 0 Minimal lobular alterations, no rejection
6 0 0 0 0 Minimal lobular alterations, no rejection
7 0 0 0 0 Minimal lobular alterations, no rejection
8 0 0 0 0 Minimal portal or lobular hepatitis, no rejection
9 0 0 0 0 Minimal lobular alterations, no rejection
10 0 0 0 0 Minimal lobular alterations, no rejection
11 0 1 0 1 Minimal portal or lobular hepatitis consistent with relapse of hepatitis
C virus, no rejection
12 0 1 0 1 Consistent with mild rejection
13 0 0 1 1 Consistent with mild rejection
14 0 1 0 1 Consistent with mild rejection; minimal nodular portal inflammation,
suggestive of relapse of hepatitis C virus†
15 1 1 0 2 Consistent with mild rejection
16 0 1 1 2 Consistent with mild rejection
17 1 1 0 2 Mild nodular portal hepatitis; mild acute rejection vs hepatitis C virus
relapse
18 1 1 1 3 Consistent with mild rejection; relapse of hepatitis C virus also was
suspected
19 2 1 1 4 Acute OLT rejection, clinically relevant
20 2 2 0 4 Acute OLT rejection, clinically relevant
21 2 2 0 4 Acute OLT rejection, clinically relevant
22 1 1 2 4 Acute OLT rejection, clinically relevant
23 2 2 1 5 Acute OLT rejection, clinically relevant
24 3 2 0 5 Acute OLT rejection, clinically relevant
25 2 2 1 5 Acute OLT rejection, clinically relevant
26 2 2 1 5 Acute OLT rejection, clinically relevant
27 3 2 1 6 Acute OLT rejection, clinically relevant
* Patients are listed nonconsecutively.
†Some morphologic lesions were consistent with relapse of hepatitis C virus.
654 䡠Radiology 䡠May 2005 Bolognesi et al
R
adiology
Evaluation of Doppler US
Composite Index
Acute rejection results in both a de-
crease in PBV and a slight increase in SPI,
and the mean Doppler US composite in-
dex was significantly different when we
compared subjects with and subjects
without acute rejection (⫺53% ⫾16 vs
⫺6% ⫾8, P⫽.008). The Doppler US
composite index mean difference in-
creased significantly when we compared
(a) patients with versus patients without
portal infiltration (⫺50% ⫾13 vs ⫺1% ⫾
7, P⬍.001) and (b) patients with versus
patients without bile duct inflammatory
infiltration (⫺38% ⫾11 vs ⫹5% ⫾7, P⫽
.007). Moreover, there was a significant
correlation between the Doppler US com-
posite index and the RAI score (r⫽
⫺0.59, P⬍.001), the portal inflamma-
tion score (r⫽⫺0.57, P⬍.002), and the
bile duct inflammation damage score (r⫽
⫺0.61, P⬍.001).
A decreasing trend was present in the
percentage change in the Doppler US
composite index when we classified the
patients according to the absence of his-
tologic activity, the presence of mild ac-
tivity, and the presence of clinically rele-
vant acute rejection (analysis of variance:
F⫽5.53, P⫽.011; posttest for linear
trend: r⫽⫺0.58, P⫽.003). The trend
was more evident in respect to the two
Doppler US parameters analyzed sepa-
rately (percentage change in PBV: r⫽
⫺0.38, P⫽.055; percentage change in
SPI: r⫽0.43, P⫽.032) (Fig 1).
Prediction of Acute Rejection
To evaluate sensitivity and specificity,
a receiver operating characteristic curve
was drawn with percentage changes in
PBV. Then, we tried to improve the accu-
racy of the single parameter by drawing a
receiver operating characteristic curve
with the calculated Doppler US compos-
ite index (Fig 2). With percentage changes
in PBV, the best accuracy for the predic-
tion of acute rejection was 81% (22 of
27), with a percentage decrease cutoff of
⫺40% (specificity, 89% [16 of 18]; sensi-
tivity, 67% [six of nine]; positive predic-
tive value, 75% [six of eight]; negative
predictive value, 84% [16 of 19]; number
needed to diagnose, 1.79).
When we considered the percentage
changes in the Doppler US composite in-
dex, the best accuracy (88% [22 of 25])
was obtained with a percentage decrease
cutoff of ⫺25% (specificity, 89% [16 of
18]; sensitivity, 86% [six of seven]; posi-
tive predictive value, 75% [six of eight];
negative predictive value, 94% [16 of
17]); number needed to diagnose, 1.33).
The Doppler US composite index allowed
us to obtain a higher accuracy in the pre-
diction of the presence of acute rejection
in respect to PBV changes alone (Fig 2),
but the Pvalue for the mean area under
the curve calculated from the Doppler US
composite index was not statistically dif-
ferent (0.91 ⫾0.06 [Doppler US compos-
ite index] vs 0.85 ⫾0.08 [PBV], zscore ⫽
1.11).
DISCUSSION
Our prospective pilot study shows that,
in patients who have undergone liver
transplantation, acute rejection results in
a pattern of hemodynamic modifications
that involves both PBV and SPI. In par-
ticular, in acute rejection, a distinct de-
crease in PBV (significantly larger than
that which occurred in patients with no
acute rejection) is coupled with a slightly
increased SPI. The evaluation of splanch-
nic hemodynamic changes could provide
an additional parameter in the evalua-
tion of patients who are suspected of hav-
ing acute rejection (a) by helping physi-
cians in the interpretation of clinical-
biochemical liver dysfunction and (b) by
helping physicians to better select sub-
TABLE 2
Characteristics of Patients in Study
Characteristic
With Clinically Relevant
Acute Rejection*
Without Clinically Relevant
Acute Rejection
Time from transplantation (d)
Median 8 10.5
Range 5–20 4–97
Sex
No. of men 6 17
No. of women 3 1
Age (y)
Mean ⫾standard error 47 ⫾449⫾2
Range 27–64 27–63
Cause of liver disease
Alcoholic cirrhosis 2 4
Virus-related cirrhosis 3 11
Viral-related cirrhosis with
hepatocellular carcinoma
22
Primary biliary cirrhosis 0 1
Primary sclerosing
cholangitis
20
Immunosuppression regimen
Cyclosporine†510
Tacrolimus‡24
Tacrolimus, mycophenolate
mofetil§
24
* RAI score of 4 or more.
†Sandimmun; Novartis Farma, Origgio, Varese, Italy.
‡Prograf; Fujisawa, Milan, Italy.
§Tacrolimus, Prograf; Fujisawa, Milan, Italy. Mycophenolate mofetil, Cellcept; Roche, Milan,
Italy.
Figure 1. Bar graph shows percentage
changes in the duplex Doppler US parameters
detected at liver biopsy. Patients were classified
according to RAI score. Graph demonstrates
that these Doppler US parameters, particularly
the Doppler US composite index, are related to
severity of rejection. Although theoretical
maximal score for the RAI is 9, patients with
clinically relevant acute rejection were as-
signed scores of 4–6 on the horizontal axis,
because 6 was the highest RAI score among
patients in this study (Table 1). White bars ⫽
percentage changes in PBV, gray bars ⫽per-
centage changes in SPI, black bars ⫽percent-
age changes in Doppler US composite index.
Results with one-way analysis of variance with
posttest for linear trend test each parameter
follow: * indicates P⫽.032; **, P⫽.055; §,P⫽
.003.
Volume 235 䡠Number 2 Acute Liver Rejection 䡠655
R
adiology
jects in whom the invasive histologic test
has to be performed.
This study involved only patients in
whom liver biopsy was performed be-
cause of the clinical and biochemical
findings that led the physician to suspect
acute rejection. The rationale for such a
choice lies in the following fact: From a
clinical viewpoint, clinically relevant
acute rejection in patients who might
benefit from a change in the drug treat-
ment has to be properly identified in
those patients in whom biochemical ab-
normalities are consistent with clinically
suspected acute rejection. This identifica-
tion is important because, in a general
population of patients who have under-
gone transplantation, any modification
in the therapeutic strategy may be ex-
cluded on the basis of nearly normal liver
test results.
Researchers in previous studies demon-
strated that acute rejection causes an
acute transient increase in portal pressure
(15), which is significantly higher in pa-
tients with severe acute rejection than in
those with moderate or mild acute rejec-
tion (16). After effective antirejection
therapy, portal pressure normalizes, and
this normalization supports the hypoth-
esis that intrahepatic resistance is increased
by the structural and/or functional events
associated with acute rejection (15). The
increase in intrahepatic resistance results
in a decrease in both liver microperfusion
(12) and total hepatic blood flow (36). As a
consequence, changes in portal-hepatic
Doppler US parameters might be expected
during acute rejection (37). The issues
about whether these changes are suffi-
ciently large to be detected with Doppler
US and whether they are specific to acute
rejection are not clear.
Findings in previous studies about the
use of Doppler US in the assessment of
liver graft hemodynamics during acute
rejection have been controversial. No re-
liability has been reported as far as he-
patic resistance indexes are concerned
(38– 41). Results in a few studies have
suggested that the loss of the normal
triphasic waveform pattern of hepatic
vein blood flow could be a parameter that
is indicative of impending rejection
(37,40,42–44), but this hypothesis has
been rejected by Kok et al (41) and by
Zalasin et al (45).
The effect of acute rejection on portal
hemodynamics has been evaluated as
change in PBV and as change in damping
index (calculated as minimum velocity
shift divided by maximum velocity
shift). Harms et al (40) found a decrease
in the damping index of the portal vein
and an increase in the damping index of
hepatic veins in patients with acute rejec-
tion. The effect of acute rejection on PBV
has been reported in two studies (37,46).
Mohr et al (37) reported a decrease in
PBV in four patients with acute rejection,
and PBV returned to normal levels after
effective antirejection therapy. On the
contrary, Kok et al (46) compared a group
of nine patients with acute rejection and
a group of 14 patients without acute re-
jection and did not find any significant
difference in PBV behavior. A possible
explanation for the difference between
the results of our study and those in the
study of Kok et al (46) may be attributed
to a difference in the selected patients
who were included in the study. In the
study of Kok et al (46), biopsies were per-
formed regardless of the clinical condi-
tion of the patients. On the contrary, in
our study, only patients with signs of
liver dysfunction underwent a liver bi-
opsy.
In addition, a different classification
for acute rejection was used in our study:
The use of the RAI score (7) allowed a
better estimate of the seriousness of liver
involvement. Moreover, in the study of
Kok et al (46), both pediatric and adult
patients were examined, and the etiology
of the liver disease was very heteroge-
neous: It included not only cirrhosis but
also biliary atresia, acute hepatic failure,
and metabolic disorders. Etiology of liver
disease is not without effect on the time
course of PBV after OLT, and when one
analyzes the changes in portal hemody-
namics provoked by acute rejection,
changes normally occurring in the first
weeks should be kept in mind. In pa-
tients who undergo transplantation for
cirrhosis, PBV and portal blood flow in-
crease dramatically immediately after
surgery, and then they progressively de-
crease (20). Therefore, a hypothetical de-
crease in PBV provoked by acute rejec-
tion could only accelerate the decrease
already present in the early posttrans-
plantation phase. On the contrary, in pa-
tients who undergo OLT for acute liver
failure, PBV and portal blood flow are
significantly decreased in relation to the
values in patients who undergo trans-
plantation for cirrhosis (16,20). In our
study, patients with acute hepatic failure
or metabolic disorders and pediatric pa-
tients were not included.
In the group of patients included in
our study, PBV decreased during acute
rejection. Also, in patients without acute
rejection, PBV decreased in respect to val-
ues before biopsy, but in patients with
acute rejection, the decrease was signifi-
cantly larger.
On the contrary, in regard to SPI, the
normal tendency for a decrease in this
index is inverted in the weeks after OLT;
it increases with the RAI score. An in-
creased SPI is a nonspecific marker of in-
creased portal vascular resistance (21).
Findings of the evaluation of the abso-
lute values of PBV and SPI were not useful
in helping us to distinguish between pa-
tients with acute rejection and those
without acute rejection. Indeed, as pa-
tients with OLT undergo dramatic modi-
fication in portal-hepatic hemodynamics
early after surgery, it is more important
to evaluate percentage changes over time
than to consider parameters as absolute
values.
When the information provided by
PBV and SPI was combined in an index, a
good receiver operating characteristic
curve was obtained (an area under the
curve of 0.91 indicates a good level of
diagnostic performance).
The possible value of serial US in the
monitoring of a patient’s response to an-
tirejection therapy was not addressed in
this study, but Mohr et al (37) reported
that the decrease in PBV detected during
rejection subsided after effective antire-
jection therapy.
Three limitations of our study should
be emphasized. First, an important draw-
back is certainly the not so trivial per-
Figure 2. Graph shows receiver operating
characteristic curves for prediction of acute re-
jection according to percentage changes in
PBV (⽧) and of Doppler US composite index
calculated by combining the percentage
changes in PBV and SPI (‚). Both parameters
(change in PBV and change in SPI) had good
accuracy for prediction of acute rejection. Best
point of curve was achieved with a percentage
decrease cutoff of SPI of ⫺25% (specificity,
89% [16 of 18]; sensitivity, 86% [six of seven]).
Dashed line is the line of no diagnostic accu-
racy.
656 䡠Radiology 䡠May 2005 Bolognesi et al
R
adiology
centage of patients in whom a reliable
complete series of Doppler US data could
not be obtained, because of the particular
clinical condition present in the first
weeks after OLT. This drawback may
limit the clinical application of the
method. Second, the method was tested
only in patients who underwent OLT for
cirrhosis or sclerosing cholangitis, and it
probably cannot be used in patients with
vascular thrombosis or stenosis. Third,
this study is only a pilot study; the index
value had good accuracy when it was ap-
plied retrospectively, and confirmation
of the results in other groups of patients,
possibly in other centers, is needed be-
fore the method can be recommended
for extensive clinical use.
Another possible limitation of the
study is the reproducibility of Doppler US
measurements. In this study, Doppler US
operators participated in a cooperative
training program to decrease the interob-
server variability of the measurements
(28,30), and this program was demon-
strated to decrease the variance of the
measurements caused by the operators to
a nonsignificant level (28,30), with no
systematic effect related to the different
measurement times (28). The problem of
the variability may be particularly impor-
tant for the SPI, which takes into account
the average velocity over time. Measure-
ment of the SPI has a greater variability
than that of the resistive index, which is
calculated as: (V
ps
⫺V
ed
)/V
ps
(30). None-
theless, we decided to use the SPI because
we think that, thanks to the inclusion of
the mean velocity in the equation for SPI,
which was included earlier in this article,
this index is probably more sensitive for
the detection of waveform abnormalities.
On the other hand, the percentage
changes of the Doppler US parameters
investigated in the study are greater than
the intrinsic variability of the measure-
ments, and, therefore, they can be con-
sidered significant.
According to our results, we think that
the analysis of changes in PBV and in SPI
values, in the first weeks after OLT, could
be included among the parameters al-
ready used to identify patients who are
suspected of having acute rejection (fever
and increased liver enzyme and bilirubin
values) and therefore can help the physi-
cian in making the decisions about
whether to perform biopsy and when to
do so.
In conclusion, if the results of this pilot
study are confirmed with results in fur-
ther studies in centers committed to the
follow-up of patients who have under-
gone OLT, a prescheduled protocol of
Doppler US measurements (PBV and SPI)
could represent an additional noninva-
sive test that is useful in the identifica-
tion of patients with acute rejection:
Acute rejection should be suspected
when, during the first weeks after OLT, a
marked unexpected decrease in PBV and
an increase in SPI are found.
References
1. Wiesner RH, Demetris AJ, Belle SH, et al.
Acute hepatic allograft rejection: inci-
dence, risk factors, and impact on out-
come. Hepatology 1998; 28:638–645.
2. Henley KS, Lucey MR, Appelman HD, et
al. Biochemical and histopathological
correlation in liver transplant: the first
180 days. Hepatology 1992; 16:688–693.
3. Williams JW, Foster PF, Sankary HN. Role
of liver allograft biopsy in patient man-
agement. Semin Liver Dis 1992; 12:60–
72.
4. Wiesner RH. Advances in diagnosis, pre-
vention, and management of hepatic al-
lograft rejection. Clin Chem 1994; 40:
2174–2185.
5. Neuberger J. Liver allograft rejection: cur-
rent concepts on diagnosis and treat-
ment. J Hepatol 1995; 23(suppl 1):54–61.
6. Snover DC, Sibley RK, Freese DK, et al.
Orthotopic liver transplantation: a patho-
logical study of 63 serial liver biopsies
from 17 patients with special reference to
the diagnostic features and natural his-
tory of rejection. Hepatology 1984; 4:
1212–1222.
7. Demetris AJ, Batts KP, Dhillon AP, et al.
Banff schema for grading liver allograft
rejection: an international consensus
document. Hepatology 1997; 25:658–663.
8. Datta Gupta S, Hudson M, Burroughs AK,
et al. Grading of cellular rejection after
orthotopic liver transplantation. Hepatol-
ogy 1995; 21:46–57.
9. Batts KP. Acute and chronic hepatic allo-
graft rejection: pathology and classifica-
tion. Liver Transpl Surg 1999; 5(4 suppl
1):S21–S29.
10. de Groen PC, Kephart GM, Gleich GJ,
Ludwig J. The eosinophil as an effector
cell of the immune response during he-
patic allograft rejection. Hepatology
1994; 20:654–662.
11. Snover DC, Freese DK, Sharp HL,
Bloomber JR, Najarian JS, Ascher NL.
Liver allograft rejection: an analysis of
the use of biopsy in determining out-
come of rejection. Am J Surg Pathol 1987;
11:1–10.
12. Klar E, Angelescu M, Zapletal C, Kraus T,
Herfarth C. Impairment of hepatic micro-
circulation as an early manifestation of
acute rejection after clinical liver trans-
plantation. Transplant Proc 1999; 31:385–
387.
13. Nagral A, Quaglia A, Sabin CA, et al.
Blood and graft eosinophils in acute cel-
lular rejection of liver allografts. Trans-
plant Proc 2001; 33:2588–2593.
14. Warle MC, Metselaar HJ, Hop WC, et al.
Early differentiation between rejection in
liver transplant patients by serum and
biliary cytokine patterns. Transplanta-
tion 2003; 75:146–151.
15. Hadengue A, Lebrec D, Moreau R, et al.
Persistence of systemic and splanchnic
hyperkinetic circulation in liver trans-
plant patients. Hepatology 1993; 17:175–
178.
16. Gadano A, Hadengue A, Widmann JJ, et
al. Hemodynamics after orthotopic liver
transplantation: study of associated fac-
tors and long-term effects. Hepatology
1995; 22:458–465.
17. Zoli M, Marchesini G, Cordiani MR, et al.
Echo-Doppler measurement of splanch-
nic blood flow in control and cirrhotic
subjects. J Clin Ultrasound 1986; 14:429–
437.
18. Taourel P, Blanc P, Dauzat M, et al. Dopp-
ler study of mesenteric, hepatic, and por-
tal circulation in alcoholic cirrhosis: rela-
tionship between quantitative Doppler
measurements and the severity of portal
hypertension and hepatic failure. Hepa-
tology 1998; 28:932–936.
19. Bolognesi M, Sacerdoti D, Merkel C, et al.
Splenic Doppler impedance indices: in-
fluence of different portal hemodynamic
conditions. Hepatology 1996; 23:1035–
1040.
20. Bolognesi M, Sacerdoti D, Bombonato G,
et al. Change in portal flow after liver
transplantation: effect on hepatic arterial
resistance indices and role of spleen size.
Hepatology 2002; 35:601–608.
21. Bolognesi M, Sacerdoti D, Bombonato G,
et al. Splenic impedance indices: a useful
method to monitor patients after liver
transplantation? Hepatology 1998; 27:
674–678.
22. Mor E, Solomon H, Gibbs JF, et al. Acute
cellular rejection following liver trans-
plantation: clinical pathologic features
and effect on outcome. Semin Liver Dis
1992; 12:28–40.
23. Dodd GD III, Memel DS, Zajko AB, Baron
RL, Santaguida LA. Hepatic artery steno-
sis and thrombosis in transplant recipi-
ents: Doppler diagnosis with resistive in-
dex and systolic acceleration time.
Radiology 1994; 192:657–661.
24. Kok T, Slooff MJ, Thijn CJ, et al. Routine
Doppler ultrasound for the detection of
clinically unsuspected vascular complica-
tions in the early postoperative phase
after orthotopic liver transplantation.
Transpl Int 1998; 11:272–276.
25. Tzakis A, Todo S, Starzl TE. Orthotopic
liver transplantation with preservation of
the inferior vena cava. Ann Surg 1989;
210:649–652.
26. Belghiti J, Noun R, Sauvanet A. Tempo-
rary portocaval anastomosis with preser-
vation of caval flow during orthotopic
liver transplantation. Am J Surg 1995;
169:277–279.
27. Volpin R, Angeli P, Galioto A, et al. Com-
parison between two high-dose methyl-
prednisolone schedules in the treatment
of acute hepatic cellular rejection in liver
transplant recipients: a controlled clinical
trial. Liver Transpl 2002; 8:527–534.
28. Sabba` C, Merkel C, Zoli M, et al. Inter-
observer and inter-equipment variability
of echo-Doppler examination of the por-
tal vein: effect of a cooperative training
program. Hepatology 1995; 21:428–433.
29. Taylor KJ, Burns PN, Woodcock JP, Wells
PN. Blood flow in deep abdominal and
pelvic vessels: ultrasonic pulsed Doppler
analysis. Radiology 1985; 154:487–493.
30. Sacerdoti D, Gaiani S, Buonamico P, et al.
Interobserver and interequipment vari-
Volume 235 䡠Number 2 Acute Liver Rejection 䡠657
R
adiology
ability of hepatic, splenic, and renal arte-
rial Doppler resistance indices in normal
subjects and in patients with cirrhosis.
J Hepatol 1997; 27:986–992.
31. Demetris AJ, Qian SG, Sun H, Fung JJ.
Liver allograft rejection: an overview of
morphologic findings. Am J Surg Pathol
1990; 14(suppl 1):49–63.
32. Guido M, Rugge M. Liver biopsy sam-
pling in chronic viral hepatitis. Semin
Liver Dis 2004; 24:89–97.
33. Hanley JA, McNeil BJ. A method of com-
paring the areas under receiver operating
characteristic curves derived from the
same cases. Radiology 1983; 148:839–
843.
34. Hanley JA, McNeil BJ. The meaning and
use of the area under a receiver operating
characteristic (ROC) curve. Radiology
1982; 143:29–36.
35. Batstone G. Practising by the evidence:
the role of pathology. J Clin Pathol 1997;
50:447–448.
36. Gadano A, Durand F, Degott C, et al.
Studies of portal hemodynamics and he-
patic oxygen consumption during acute
liver allograft rejection. Transplantation
1997; 64:1188–1192.
37. Mohr H, Godderz W, Lohse AW, Gerken
G, Meyer zum Buschenfelde KH. Duplex
ultrasound in diagnosis and monitoring
of liver transplant rejection. Lancet 1995;
346:315–316.
38. Longley DG, Skolnick ML, Sheahan DG.
Acute allograft rejection in liver trans-
plant recipients: lack of correlation with
loss of hepatic artery diastolic flow. Radi-
ology 1988; 169:417–420.
39. Marder DM, DeMarino GB, Sumkin JH,
Sheahan DG. Liver transplant rejection:
value of the resistive index in Doppler US
of hepatic arteries. Radiology 1989; 173:
127–129.
40. Harms J, Ringe B, Pichlmayr R. Postoper-
ative liver allograft dysfunction: the use
of quantitative duplex Doppler signal
analysis in adult liver transplant patients.
Bildgebung 1995; 62:124–131.
41. Kok T, Haagsma EB, Klompmaker IJ, et al.
Doppler ultrasound of the hepatic artery
and vein performed daily in the first 2
weeks after orthotopic liver transplanta-
tion: useful for the diagnosis of acute re-
jection? Invest Radiol 1996; 31:173–179.
42. Coulden RA, Britton PD, Farman P, No-
ble-Jamieson G, Wight DG. Preliminary
report: hepatic vein Doppler in the early
diagnosis of acute liver transplant rejec-
tion. Lancet 1990; 336:273–275.
43. Britton PD, Lomas DJ, Coulden RA,
Farman P, Revell S. The role of hepatic
vein Doppler in diagnosing acute rejec-
tion following paediatric liver trans-
plantation. Clin Radiol 1992; 45:228–
332.
44. Jequier S, Jequier JC, Hanquinet S, Le
Coultre C, Belli DC. Orthotopic liver
transplants in children: change in he-
patic venous Doppler wave pattern as an
indicator of acute rejection. Radiology
2003; 226:105–112.
45. Zalasin S, Shapiro RS, Glajchen N, Stan-
cato-Pasik A. Liver transplant rejection:
value of hepatic vein Doppler waveform
analysis. Abdom Imaging 1998; 23:427–
430.
46. Kok T, Slooff MJ, Peeters PM, et al.
Changes in portal hemodynamics and
acute rejection in the first 2 weeks after
orthotopic liver transplantation: a pro-
spective Doppler ultrasound study. Invest
Radiol 1996; 31:774–780.
658 䡠Radiology 䡠May 2005 Bolognesi et al
R
adiology
