Received: 5 August 2018 | First decision: 3 September 2018 | Accepted: 9 October 2018

DOI: 10.1111/apt.15042

Ascitic cholesterol is superior to serum‐ascites albumin

gradient in the detection of non‐portal hypertensive ascites
and the diagnosis of mixed ascites

Li Du1 | Shenghua Zhu1 | Zhiwen Lu1 | Ting Xu1 | Tao Bai1 | Dong Xu2 |
Xiuqi Wei3 | Jin Li4 | Keshu Xu1 | Jin Ye1 | Xiaohua Hou1 | Yuhu Song1

1
Division of Gastroenterology, Union
Hospital, Tongji Medical College Huazhong Summary
University of Science and Technology, Background: The diagnostic value of ascitic cholesterol in the differential diagnosis
Wuhan, China
2 of ascites is controversial.
Department of Infectious Diseases, Tongji
Hospital, Tongji Medical College, Huazhong Aim: To investigate the diagnostic performance of ascitic cholesterol in the differ-
University of Science and Technology,
ential diagnosis of ascites.
Wuhan, China
3
Department of Clinical Laboratory, Union Methods: Consecutive patients with new‐onset ascites were enrolled prospectively.
Hospital, Tongji Medical College, Huazhong The pertinent data were collected from 629 patients with all forms of ascites.
University of Science and Technology,
Wuhan, China Results: In the training cohort, determination of the ascitic cholesterol level was a
4
Department of Clinical Laboratory, highly effective method of distinguishing non‐portal hypertension (NPH) from portal
Zhongnan Hospital of Wuhan University,
hypertension (PH). At the pre‐determined cut‐off value of 45 mg/dL, the sensitivity
Wuhan, China
of ascitic cholesterol was superior to the serum‐ascites albumin gradient (SAAG) in
Correspondence
identifying NPH‐related ascites; the area under the receiver operating characteristic
Yuhu Song and Xiaohua Hou, Division of
Gastroenterology, Union Hospital, Tongji curve was 0.945. In the patients misdiagnosed based on SAAG classification, the
Medical College, Huazhong University of
diagnostic accuracy of ascitic cholesterol was 69%. The ascitic cholesterol level
Science and Technology, Wuhan, China.
Emails: yuhusong@yahoo.com or showed excellent performance in identifying peritoneal lesions in patients with
yuhusong@163.com; houxh@hust.edu.cn
mixed ascites.
Funding information Conclusion: Ascitic cholesterol is an excellent measure for detecting NPH ascites
National Natural Science Foundation of
and for identifying peritoneal lesions in mixed ascites. Thus, this simple and cost‐ef-
China, Grant/Award Number: 81270506,
81570555; Clinical Research Physician fective measure should be determined in patients with new‐onset ascites (www.chic
Program of Tongji Medical College, HUST
tr.org.cn; ChiCTR‐DCD‐15006907).

Li Du and Shenghua Zhu contributed equally to this study.

The Handling Editor for this article was Professor Stephen Harrison, and it was accepted
for publication after full peer‐review.

Aliment Pharmacol Ther. 2018;1–8. wileyonlinelibrary.com/journal/apt © 2018 John Wiley & Sons Ltd | 1
2 | DU ET AL.

1 | INTRODUCTION enrolled a larger number of the patients with all forms of ascites,
and then investigated the diagnostic value of ascitic cholesterol in
Ascites is the pathological accumulation of fluid in the peritoneal the differential diagnosis of ascites.
cavity. The aetiology of ascites include liver cirrhosis, peritoneal car-
cinomatosis, congestive heart failure, tuberculous peritonitis, renal
diseases, hepatic vascular diseases, pancreatitis, etc.1 In some cases, 2 | METHODS
it is challenging to determine the cause of ascites. Appropriate ascitic
fluid analysis is the most rapid and cost‐effective method in defining 2.1 | Patient selection
the aetiology of ascites.1 Several biochemical parameters such as
In this prospective cohort study, all the patients with new‐onset
total protein, the serum‐ascites albumin gradient (SAAG), lactate
ascites (training cohort) who were admitted to Union Hospital of
dehydrogenase, glucose, triglyceride, and various tumour markers
Huazhong University of Science and Technology (Wuhan, China)
have proven to be useful in diagnosing the cause of ascites.2–10 Pre-
were assessed for eligibility from May 2015 to December 2016. The
vious studies demonstrated that ascitic cholesterol exhibited excel-
following inclusion criteria were used: (a) patients with new‐onset
lent diagnostic performance in the differentiation between cirrhosis
ascites; (b) patients who had received diagnostic paracentesis; and
and malignant ascites, and the diagnostic value of cholesterol was
(c) patients who consented to participate. The exclusion criteria were
superior to that of the SAAG7,9,11,12; thus, ascitic cholesterol has
the following: (a) patients with renal failure who had received peri-
been used as a marker of malignant ascites.9,13,14 As a result, the
toneal dialysis; (b) bloody ascites caused by trauma or surgery; (c)
measurement of ascitic cholesterol is considered a useful adjunct to
ascites with unknown origin; and (d) inability to adjudicate the cause
cytology in diagnosing peritoneal carcinomatosis due to its simplicity
of ascites due to insufficient data (Figure 1). All patients underwent
and cost effectiveness. However, the elevation of ascitic cholesterol
a stringent protocol that included clinical history, physical examina-
was also observed in patients with ascites who had purulent peri-
tion, blood examination, ascitic fluid analysis, imaging modalities,
tonitis, congestive heart failure and tuberculous ascites, which dimin-
peritoneal biopsy, etc. The study was conducted according to the
ished the specificity of ascitic cholesterol in diagnosing peritoneal
principles of the Declaration of Helsinki, and the protocol was
carcinomatosis.9,15 Thus, these findings revealed that the cholesterol
approved by the institutional ethics board review and registered at
concentration of ascitic fluid was not specific in the diagnosis of
www.chictr.org.cn (ChiCTR‐DCD‐15006907).
peritoneal carcinomatosis. In consider of this, the measurement of
ascitic cholesterol concentration was classified as unhelpful for the
diagnosis of ascites according to AASLD guidelines on the manage-
2.2 | Diagnostic criteria
ment of adult patients with ascites (2009 and 2012).1 The diagnostic
value of ascitic cholesterol is still debated because a relatively small The pertinent data of enrolled patients were collected, and the aeti-
sample of the patients confirmed its clinical value. In this study, we ology of ascites was identified by the well‐established clinical

Patients with new onset ascites

(n = 497)

Exclusion (n = 56)
1. Peritoneal dialysis patients (n = 25)
2. Bloody ascites caused by trauma or
surgery (n = 8)
3. Ascites of unkown origin (n = 11)
4. Inability to adjudicate the cause of ascites
due to insufficient data (n = 12)

Included in the study

(n = 441)

Cirrhosis Cardiac ascites Miscellaneous Mixed ascites Malignant ascites NPH benign
(n = 126) (n = 18) PH (n = 48) (n = 69) (n = 106) ascites (n = 74)

FIGURE 1 Enrolment of the patients with new‐onset ascites in training cohort. PH, portal hypertension; NPH, non‐portal hypertension
DU ET AL. | 3

T A B L E 1 The aetiology of ascites in the patients enrolled in our study

Training cohort Validation cohort Total Ascitic cholesterol (mg/
Aetiology (n = 441) (n = 188) (n = 629) dL)
Portal hypertension 192 (43.5) 89 (47.3) 281 (44.7) 22.04 ± 13.92
Sterile cirrhotic 110 (24.9) 60 (31.9) 170 (27.0) 20.88 ± 12.37
Infected cirrhotic 16 (3.6) 5 (2.7) 21(3.3) 20.11 ± 11.6
Cardiac ascites 18 (4.1) 7 (3.7) 25 (4.0) 31.32 ± 18.56
Miscellaneous portal hypertensive 48 (10.8) 17 (9.0) 65 (10.33) 22.82 ± 15.08
Fulminant hepatic failure 7 (1.6) 4 (2.1) 11 (1.7) 29 ± 19.72
Acute hepatitis superimposed on cirrhosis 14 (3.2) 8 (4.3) 22 (3.5) 10.83 ± 8.51
Hepatic sinusoidal obstruction syndrome 25 (5.7) 4 (2.1) 29 (4.6) 29.78 ± 11.21
Chylous cirrhotic ascites 0 (0) 1 (0.5) 1 (0.2) 11.6
Budd‐Chiari syndrome 1 (0.2) 0 (0) 1 (0.2) 36.74
Unexplained portal hypertension 1 (0.2) 0 (0) 1 (0.2) 7.35
Non‐portal hypertension 180 (40.8) 72 (38.3) 252 (40.1) 82.75 ± 35.19
Malignant ascites 106 (24.0) 46 (24.5) 152 (24.2) 87.01 ± 35.96
Miscellaneous non‐portal hypertensive 74 (16.8) 26 (13.8) 100 (15.9) 76.57 ± 33.26
Tuberculous peritonitis 28 (6.3) 14 (7.4) 42 (6.7) 88.94 ± 17.01
Dialysis‐related ascites 2 (0.5) 0 (0) 2 (0.3) 72.7 ± 11.6
Pancreatic ascites 17 (3.9) 3 (1.6) 20 (3.2) 59.55 ± 34.8
Secondary bacterial peritonitis 8 (1.8) 3 (1.6) 11 (1.7) 60.03 ± 31.32
Chylous malignant ascites 2 (0.5) 0 (0) 2 (0.3) 127.61 ± 85.46
Connective tissue disease 6 (1.4) 2 (1.1) 8 (1.3) 81.59 ± 21.27
Eosinophilic gastroenteritis 5 (1.1) 2 (1.1) 7 (1.1) 111.37 ± 15.85
Carcinoma and perforation 1 (0.2) 0 (0) 1 (0.2) 26.68
Connective tissue disease and nephrotic 2 (0.5) 0 (0) 2 (0.3) 16.63 ± 8.12
syndrome
Carcinoma and tuberculosis 1 (0.2) 0 (0) 1 (0.2 74.63
Nephrotic syndrome 2 (0.5) 2 (1.1) 4 (0.6) 17.4 ± 18.17
Mixed ascites 69 (15.6) 27 (14.4) 96 (15.3) 32.48 ± 23.59
Carcinoma and cirrhosis 49 (11.1) 21(11.2) 70 (11.1) 29.78 ± 19.72
Carcinoma and heart failure 1 (0.2) 0 (0) 1 (0.2) 88.94
Carcinoma and miscellaneous portal 7 (1.6) 3 (1.6) 10 (1.6) 46.79 ± 34.42
hypertension
Pancreatitis and pancreatic portal hypertension 1 (0.2) 0 (0) 1 (0.2) 76.18
Connective tissue disease and cirrhosis 3 (0.7) 2 (1.1) 5 (0.8) 19.72 ± 20.88
Tuberculosis and cardiac 1 (0.2) 0 (0) 1 (0.2) 51.82
Tuberculosis and cirrhosis 5 (1.1) 0 (0) 5 (0.8) 48.34 ± 17.4
Cirrhosis and nephrotic syndrome 2 (0.5) 1 (0.5) 3 (0.5) 9.28 ± 6.57

Values within parentheses are expressed as percentage.

criterion.5,10,16 All the data of enrolled patients were reviewed by
two independent staff hepatologists who were blinded to the results
of ascitic cholesterol and the SAAG. Any discrepant diagnoses were
jointly re‐evaluated to reach a final consensus. The patients with
ascites were divided into two groups: portal hypertension (PH) and
non‐portal hypertension (NPH). The aetiology of PH‐related ascites
included the following: cirrhosis, cardiac diseases, fulminant hepatic
failure, hepatic sinusoidal obstruction syndrome (HSOS), Budd‐Chiari
syndrome (BCS), and unexplained PH (clinical evidence of PH with
unidentified aetiology). The diagnosis of cirrhosis, spontaneous bac-
terial peritonitis, nephrotic syndrome, cardiac ascites, malignant
ascites, tuberculous peritonitis, pancreatic ascites, and secondary
bacterial peritonitis were described in previous studies.10,16–18 The
criterion of BCS and pyrrolizidine alkaloids‐induced HSOS were
described in previous studies.19–22 Eosinophilic gastroenteritis was
defined according to Talley's proposal.23 Mixed ascites was diag-
nosed when PH accompanied another aetiology of NPH.1,5,16 The
details were described in the Supporting Information.
4 | DU ET AL.

(A) Training set (B) Training set

200 40

150

100 30
Ascitic cholesterol (mg/dL)

80
20

SAAG (g/L)
60
10

40

0
20

–20
0
ic ic ac d nt n tic tic ac PH
d nt gn
ot ot di PH ixe na nig ho ho di ixe na ni
irr
h
irr
h ar is
c M lig be rr rr ar is
c M ig be
c c C M a ci ci C M al
il e ed M PH r ile t ed M PH
er t N N
St ec te ec
Inf S I nf

F I G U R E 2 Scatter dot plot showing the distribution of ascitic cholesterol (A), SAAG (B) in the patients with ascites enrolled in the training
cohort. Median with interquartile range is included, horizontal lines at 45 mg/dL for ascitic cholesterol (A), 11.0 g/L for SAAG (B). Infected
cirrhosis is liver cirrhosis with spontaneous bacterial peritonitis. Misc PH, miscellaneous portal hypertension; NPH, non‐portal hypertension;
SAAG, serum‐ascites albumin gradient. ▲ indicated the patients with malignant ascites had 285.77 mg/dL of ascitic cholesterol

aetiology of liver cirrhosis was hepatitis B (n = 47), hepatitis C

2.3 | Validation cohort
(n = 7), alcoholic (n = 22), autoimmune diseases (n = 9), nonalcoholic
A second set of in‐patients with new‐onset ascites were recruited fatty liver disease (n = 3), and others (n = 38). The characteristics of
from three university hospitals in China (Union Hospital, Tongji Med- the enrolled patients are shown in Tables S1 and S2. First, the indi-
ical College, HUST; Tongji Hospital, Tongji Medical College, HUST; vidual result of ascitic cholesterol was evaluated in different types of
and Zhongnan Hospital, Wuhan University) for the validation cohort. ascites, and a predetermined cut‐off value of 45 mg/dL4,14 was used
There were 188 patients with new‐onset ascites prospectively in our study. The results (Figure 2A) showed that low levels of ascitic
enrolled between January and May 2017. The protocol used for the cholesterol (<45 mg/dL) were observed in most patients with PH‐re-
training cohort was conducted in the validation cohort. The distribu- lated ascites (93%); most (86%) of the patients with non‐portal
tion of the enrolled patients with ascites is shown in Table 1. hypertensive ascites had higher concentrations of ascitic cholesterol
(≥45 mg/dL). The receiver operating characteristic (ROC) curve of
ascitic cholesterol was calculated to confirm its diagnostic efficacy.
2.4 | Ascitic fluid cholesterol assay The AUC (area under curve) for the ROC curve was 0.945 in the
Ascites samples obtained by paracentesis were collected in tubes, training cohort. The cut‐off value of 45 mg/dL yielded the following
and then sent for routine biochemical assays and cytological exami- results: sensitivity of 86%, specificity of 93%, negative predictive
nation. The ascitic fluid cholesterol was determined by the CHOD‐ value (NPV) of 87%, positive predictive value (PPV) of 92%, and
PAP method using a cholesterol determination assay kit (Shanghai accuracy of 90% (data not shown). Thus, ascitic cholesterol is an
Fosun Long March, Shanghai, China). excellent discriminator for differentiating PH from other causes.
Simultaneously, the AUC of the ROC curve was 0.692 for serum
cholesterol and 0.721 for serum‐ascites cholesterol gradient (data
not shown); therefore, these two parameters were not considered in
3 | RESULTS
the diagnosis of ascites. Finally, the diagnostic performance of the
SAAG was also determined in the training cohort. The AUC under
3.1 | Training cohort
ROC curve was 0.954 in the training cohort; at the cut‐off value of
During the study period, 497 consecutive patients with new‐onset 11.0 g/L, it yielded the following results: sensitivity of 81%, speci-
ascites were recruited for the study (Figure 1). The classification of ficity of 95%, NPV of 85%, PPV of 94%, and accuracy of 89% (data
the aetiology for the 441 included patients is shown in Table 1. The not shown). These findings revealed that ascitic cholesterol exhibited
DU ET AL. | 5

T A B L E 2 Diagnostic performance of SAAG and ascitic cholesterol in the whole cohort

Positive Negative
predictive predictive
Sensitivity Specificity value value Accuracy
Variables (%) 95% CI (%) 95% CI (PPV) (%) 95% CI (NPV) (%) 95% CI (%)
SAAG (<11.0 g/L) 79.51 73.89‐84.39 97.14 94.45‐98.76 96.04 92.35‐98.27 84.47 80.05‐88.25 88.93
Ascitic cholesterol 86.11 81.22‐90.13 93.59 90.06‐96.16 92.34 88.16‐95.40 88.26 84.05‐91.68 90.06
(≥45 mg/dL)
Combined SAAG 73.36 67.35‐78.80 99.29 97.45‐99.91 98.90 96.07‐99.87 81.10 76.56‐85.10 87.24
+ascitic cholesterol
(Ascitic cholesterol
≥45 mg/dL and SAAG
<11.0 g/L)
SAAG +ascitic 92.06 88.01‐95.08 91.43 87.51‐94.43 90.62 86.37‐93.90 92.75 89.03‐95.52 91.73
cholesterol (ascitic
cholesterol ≥45 mg/dL
or SAAG <11.0 g/L)

a similar performance to that of the SAAG in the differentiation of
PH from NPH.
3.2 | Validation cohort
The validation cohort consisted of 217 consecutive in‐patients with
new‐onset ascites recruited from three medical institutions; 29
patients were excluded, leaving 188 for analysis. Of these patients,
89 with PH, 72 with NPH, and 27 with mixed ascites. Demographic
characteristics and biochemical parameters were similar in the train-
ing cohort and the validation cohort (Table S1). At a predetermined
cut‐off value of 45 mg/dL, ascitic cholesterol had a sensitivity of
86%, specificity of 96%, NPV of 90%, and PPV of 94% in validation
cohort (data not shown); the AUC of the ROC curve was 0.954.
These results confirmed the findings of the training cohort. There-
fore, ascitic cholesterol provides an excellent parameter for distin-
guishing PH‐related ascites from other causes.
3.3 | Diagnostic performance of the SAAG and
ascitic cholesterol in the whole cohort
The SAAG is extensively used in the diagnostic workup of ascites
because of high sensitivity for a high SAAG (≥11.0 g/L) in diagnosing
PH‐related ascites. Thus, diagnostic performance of the SAAG and
ascitic cholesterol was determined in the whole cohort. At a cut‐off
point of 11.0 g/L, the SAAG had a sensitivity of 80%, specificity of
97%, diagnostic accuracy of 89%, NPV of 84%, and PPV of 96%;
while ascitic cholesterol at a cut‐off value of 45 mg/dL had a sensi-
tivity of 86%, specificity of 94%, diagnostic accuracy of 90%, NPV
of 88%, and PPV of 92% (Table 2). It indicated that a high SAAG
had higher sensitivity for the detection of PH‐related ascites. While
high ascitic cholesterol possessed higher sensitivity in diagnosing
NPH‐related ascites. In addition, we also investigated diagnostic per-
formance of the SAAG and/or ascitic cholesterol. As shown in
Table 2, combining the SAAG (<11.0 g/L) and ascitic cholesterol
(≥45 mg/dL) increased the specificity to 99% in diagnosing non‐por-
tal hypertensive ascites; the SAAG (<11.0 g/L) or ascitic cholesterol
(≥45 mg/dL) yielded positive result of 92% in non‐portal hyperten-
sive ascites. These findings indicate that the combination of the
SAAG (<11.0 g/L) and ascitic cholesterol (≥45 mg/dL) exhibits better
performance than that of individual index in differential diagnosis of
ascites.
Interestingly, 89% (135/152) of malignant ascites cases had asci-
tic cholesterol higher than 45 mg/dL, 74% (109/147) of malignant
ascites had an SAAG below 11.0 g/L. In the patients with tubercu-
lous peritonitis, 100% (42/42) of the patients had ascitic cholesterol
≥45 mg/dL, 90% (36/40) of the patients had an SAAG <11.0 g/L.
The results revealed that ascitic cholesterol exhibited better efficacy
in diagnosing malignant ascites and tuberculous peritonitis than the
SAAG. In addition, 20% of cardiac ascites, 60% of pancreatic ascites,
64% of secondary bacterial peritonitis had ascitic cholesterol
≥45 mg/dL. Finally, we also examined mean values of ascitic choles-
terol in different types of ascites (Table 1). The results indicated that
ascitic cholesterol might assist in differentiating the aetiologies of
non‐portal hypertensive ascites through mean values of ascitic
cholesterol.
3.4 | Diagnostic value of ascitic cholesterol in the
misdiagnosed patients based on SAAG classification
We further investigated the diagnostic efficacy of ascitic cholesterol
in the patients misdiagnosed according to SAAG classification. In the
52 patients misdiagnosed according to SAAG classification, 60% (9/
15) of benign ascites (tuberculous peritonitis excluded) cases were
correctly diagnosed by ascitic cholesterol. Additionally, 70% (23/33)
of malignant ascites and 100% (4/4) of the tuberculous peritonitis
could be correctly diagnosed (Table S3). Thus, we concluded that
ascitic cholesterol provides an excellent parameter for determining
patients misdiagnosed according to SAAG classification, especially
among patients with malignant ascites and tuberculous peritonitis.
6 | DU ET AL.

(A) Mixed ascites (B) Mixed ascites

110 40
105
100
Ascitic cholesterol (mg/dL)

30
80

SAAG (g/L)
60 20

40
10
20

0 0
is t, sis
, H, sis t, sis H, n H,
t,
n st as an sta N PH sion NP sion ,
nt sta an sta NP esio NP sion
a
n a n n le n a n n l n
ig et ig ta ig l le n a ig ta
al m al e ig al
en ea ig et al e ig al ig l le
m eal m lm en ne b al m m lm en ne en ea
b n m eal d a b o b
d ton
d d a
an one d rito d to d an ne d rit
an iton t an pe an eri an iton H rito an pe an eri
r H i
H
P pe P
pe
r H
P ou
t P ith p
H H
P pe
r P
pe
H
P ou
t
P ith p
H
u t h i t h w t h i th w
o i t w o u i t w
w ith w ith w
w

F I G U R E 3 Scatter dot plot showing the distribution of ascitic cholesterol (A), SAAG (B) in mixed group. PH, portal hypertension; NPH, non‐
portal hypertension; SAAG, serum‐ascites albumin gradient

and 100% (5/5) of mixed ascites with benign peritoneal involvement

3.5 | Diagnostic value of ascitic cholesterol for the
mixed ascites
In total, 15% of the patients (96 cases) included in our study had
mixed ascites. Thus, it is important to diagnose mixed ascites in clini-
cal practice. Since imaging modalities fail to detect peritoneal carci-
nomatosis at an early stage, it is difficult to identify the exact
aetiology of ascites when the patients with ascites have PH and car-
cinoma. Thus, the patients with ascites who had PH and carcinoma
were classified as mixed ascites. A similar classification of mixed
ascites was described previously.5 In the patients with mixed ascites,
a peritoneal tumour metastasis and peritoneal benign lesions were
defined by thickening, adhesion, nodular changes or abnor-
mal metabolism of peritoneum based on image results (CT, MRI,
PET‐CT). The images obtained from picture archiving and communi-
cation systems (PACS) were reviewed independently and in consen-
sus by two radiologists. Unfortunately, 25 cases were excluded
because the patients had received the examination of imaging (CT,
MRI or PET‐CT) in other hospitals, and the image data were not
obtained in our PACS. Of mixed ascites with the data of the images,
the causes were PH and carcinoma without peritoneal metastasis
(n = 41), PH and carcinoma with peritoneal metastasis (n = 16), PH
and benign NPH (connective tissue disease, pancreatic ascites, tuber-
culosis) without peritoneal involvement (n = 9), PH and benign NPH
with peritoneal involvement (n = 5). In the patients with mixed
ascites, 90% (37/41) of PH and carcinoma without peritoneal metas-
tasis and 89% (8/9) of PH and benign NPH without peritoneal
involvement had low ascitic cholesterol (<45 mg/dL); meanwhile,
69% (11/16) of mixed ascites with peritoneal malignant metastasis
had an ascitic cholesterol level ≥45 mg/dL (Figure 3A). These find-
ings revealed that mixed ascites with high ascitic cholesterol
(≥45 mg/dL) probably had peritoneal lesions. Simultaneously, most of
the patients with mixed ascites had a high SAAG (≥11.0 g/L) (Fig-
ure 3B), which was consistent with previous studies.1,5 These results
indicated that the SAAG is not helpful for identifying peritoneal
lesions in mixed ascites patients, or for distinguishing PH‐related
ascites from mixed ascites. In conclusion, the diagnostic value of
ascitic cholesterol was superior to that of SAAG in diagnosing mixed
ascites.
4 | DISCUSSION
Abdominal paracentesis is likely the most rapid and cost‐effective
method of diagnosing the cause of ascites. Previous studies demon-
strated that ascitic cholesterol was a useful marker for discriminating
between malignancy‐related and benign ascites.3,7,9,13,24 Our
prospective cohort study validated ascitic cholesterol as an excellent
marker for the differentiation of PH from other causes. Several fac-
tors contributed to contradictory conclusions between previous
reports and our study. The conclusion was drawn from a small sam-
ple size in previous studies3,7,9,13,24; furthermore, the variety of
underlying causes that led to ascites was limited, which confined
their scientific value.
A large number of patients with all forms of ascites were
enrolled in our prospective cohort study, and the sample size and
the variety of underlying causes guarantee the reliability of our
DU ET AL.
results. Thus, the conclusion derived from our study is more accurate
compared with previous studies. Several studies explored the mecha-
nism underlying the increased concentration of cholesterol in malig-
nant ascites. The elevations in ascitic cholesterol in peritoneal
carcinomatosis compared to cirrhosis were mainly caused by the
increased movement of plasma high‐density lipoprotein and low‐den-
sity lipoprotein into the peritoneal cavity.25
The SAAG has been used extensively in the diagnostic workup
of patients with ascites. In our study, the diagnostic accuracy of
SAAG is inferior to the findings reported in previous studies.1,5 The
difference between our study and previous studies may be attribu-
ted to the constituent ratio of underlying causes. The causes of
ascites in the previous study included liver cirrhosis (84.1%), cardiac
insufficiency (2.7%), malignant neoplasia (2.4%), miscellaneous PH
(3.8%), and mixed ascites (4.7%).5 Both previous studies and our
research showed that the SAAG had high sensitivity in diagnosing
PH‐related ascites.5 In developed countries, more than 80% of
patients with ascites have cirrhosis. Therefore, the SAAG is widely
used to distinguish PH from other causes of ascites. However, the
aetiologies of the enrolled patients in our study included liver cirrho-
sis (30.3%), malignant neoplasia (24.2%), tuberculous peritonitis
(6.7%), cardiac insufficiency (4.0%), miscellaneous PH (10.3%), and
mixed ascites (15.3%). The results showed the percentage of miscel-
laneous PH, NPH and mixed ascites in our study (developing coun-
try) was greater than that in previous studies. Therefore, it is a
challenge to define the causes of ascites in developing countries. In
our study, the SAAG possessed better sensitivity in detecting PH‐re-
lated ascites. However, the sensitivity of ascitic cholesterol was
superior to the SAAG in detecting NPH‐related ascites. More impor-
tantly, our results demonstrated that ascitic cholesterol exhibited
excellent efficacy in diagnosing malignant ascites and tuberculous
peritonitis. In addition, we showed that ascitic cholesterol may aid in
differentiating the aetiologies in non‐portal hypertensive ascites,
which was revealed by mean values of ascitic cholesterol in different
types of ascites. Since there are similarities in the pathophysiology
and the response to diuretics observed for nephrotic syndrome and
PH, nephrotic syndrome should be classified as PH. In this case, asci-
tic cholesterol possesses better diagnostic performance. In conclu-
sion, the SAAG was an excellent parameter for diagnosing PH ‐
related ascites, while ascitic cholesterol was highly effective in
detecting NPH‐related ascites. Thus, the SAAG and ascitic choles-
terol should both be determined in laboratory investigation of ascitic
fluid. Calculating the SAAG involves measuring the albumin concen-
tration of serum and ascitic fluid specimens obtained on the same
day and subtracting the ascitic fluid value from the serum value.
Thus, an infusion of albumin results in errors when calculating the
SAAG. In addition, the cost of determining ascitic cholesterol ($1 per
sample) is cheaper than that of determining SAAG ($ 0.9 per sample,
$ 0.9 × 2 test = $1.8 per assay) in our hospital. Thus, it is easy and
cost effective to determine the level of ascitic cholesterol in clinical
practice.
The combined SAAG (<11.0 g/L) and ascitic cholesterol (≥45 mg/
dL) increased specificity to 99%; the SAAG (<11.0 g/L) or ascitic
| 7
cholesterol (≥45 mg/dL) yielded sensitivity to 92% in diagnosing
NPH‐related ascites. The results indicated that the combination of
SAAG (<11.0 g/L) and ascitic cholesterol (≥45 mg/dL) improved diag-
nostic performance. El‐Serag et al identified patients with cirrhosis
and a low SAAG, and demonstrated that evaluation of an SAAG
<1.1 g/dL in patients with known cirrhosis had low yield and was
less likely to be helpful than in patients without cirrhosis.16 Thus, it
is a challenging problem for clinicians to deal with patients misdiag-
nosed based on SAAG classification. In our study, the diagnoses of
60% of patients with benign ascites (tuberculous peritonitis
excluded), 70% of patients with malignant ascites and 100% of
tuberculous peritonitis could be corrected by ascitic cholesterol.
These findings suggest that ascitic cholesterol should be determined
in patients misdiagnosed according to SAAG classification, especially
the patients with malignant ascites and tuberculous peritonitis.
In our study, 15.3% of the patients with mixed ascites were iden-
tified. However, the utility of diagnosing mixed ascites has not been
adequately established. In the study conducted by Runyon et al,
most of the mixed ascites patients had a high SAAG. While El‐Serag
et al demonstrated that some of the patients with mixed ascites,
such as cirrhosis plus malignant ascites, cirrhosis plus tuberculous
peritonitis had a low SAAG.16 Our studies showed similar results in
patients with mixed ascites (Figure 3). The results indicated that the
SAAG is not helpful for the diagnosis of mixed ascites. Most of the
patients with mixed ascites who had peritoneal lesions had high
levels of ascitic cholesterol. Thus, the diagnostic value of ascitic
cholesterol was superior to that of the SAAG in diagnosing mixed
ascites.
This prospective study had potential limitations. First, few
patients under 16 years of age were recruited in our study. Second,
the definition of peritoneal lesions was based on the image results,
which may have led to a false diagnosis compared with pathological
examination. Finally, we did not explore the mechanism underlying
the increased concentration of cholesterol in NPH‐related ascites.
In summary, high ascitic cholesterol (≥45 mg/dL) has a high sensi-
tivity for detecting NPH‐related ascites; ascitic cholesterol is a valu-
able parameter for patients misdiagnosed according to SAAG
classification. In mixed ascites, ascitic cholesterol is useful in identify-
ing peritoneal lesions, such as peritoneal carcinomatosis and tubercu-
lous peritonitis. Thus, this simple and cost‐effective parameter
should be determined in all patients with new‐onset ascites in clini-
cal practice.
ACKNOWLEDGEMENT
We are grateful to Dr Xin Li and Drs Tingting Guo (Department of
Radiology, Union Hospital, Tongji Medical College) for their kind
assistances in analysing imaging data.
Declaration of personal interests: None.
Declaration of funding interests: This work was partially supported
by National Natural Science Foundation of China (no. 81570555,
81270506) and Clinical Research Physician Program of Tongji Medi-
cal College, HUST (2017).
8 |
AUTHORSHIP
Guarantor of the article: Yuhu Song and Xiaohua Hou.
Author contributions: Yuhu Song and Xiaohua Hou: conception of
study, grant application; Li Du, Shenghua Zhu, Zhiwen Lu, Ting Xu,
Dong Xu, Xiuqi Wei, Jin Li: data collection and data analysis; Keshu
Xu, Jin Ye, Xiaohua Hou: review of chart and data interpretation; Li
Du, Tao Bai: statistical analysis; Li Du, Shenghua Zhu, and Yuhu
Song: preparation of the paper. Li Du, Shenghua Zhu, Xiaohua Hou,
Yuhu Song: revision of the manuscript. All the authors approved the
final version of the manuscript.
ORCID
Li Du http://orcid.org/0000-0001-8864-6380
REFERENCES
1. Runyon BA, APG Committee. Management of adult patients with
ascites due to cirrhosis: an update. Hepatology. 2009;49:2087‐2107.
2. Boyer TD, Kahn AM, Reynolds TB. Diagnostic value of ascitic fluid
lactic dehydrogenase, protein, and WBC levels. Arch Intern Med.
1978;138:1103‐1105.
3. Jungst D, Gerbes AL, Martin R, et al. Value of ascitic lipids in the dif-
ferentiation between cirrhotic and malignant ascites. Hepatology.
1986;6:239‐243.
4. Gerbes AL, Jungst D, Xie YN, et al. Ascitic fluid analysis for the dif-
ferentiation of malignancy‐related and nonmalignant ascites. Proposal
of a diagnostic sequence. Cancer. 1991;68:1808‐1814.
5. Runyon BA, Montano AA, Akriviadis EA, et al. The serum‐ascites
albumin gradient is superior to the exudate‐transudate concept in
the differential diagnosis of ascites. Ann Intern Med. 1992;117:215‐
220.
6. Chen SJ, Wang SS, Lu CW, et al. Clinical value of tumour markers
and serum‐ascites albumin gradient in the diagnosis of malignancy‐
related ascites. J Gastroenterol Hepatol. 1994;9:396‐400.
7. Gupta R, Misra SP, Dwivedi M, et al. Diagnosing ascites: value of
ascitic fluid total protein, albumin, cholesterol, their ratios, serum‐
ascites albumin and cholesterol gradient. J Gastroenterol Hepatol.
1995;10:295‐299.
8. Salvatore F, Sacchetti L, Castaldo G. Multivariate discriminant analy-
sis of biochemical parameters for the differentiation of clinically con-
founding liver diseases. Clin Chim Acta. 1997;257:41‐58.
9. Gulyas M, Kaposi AD, Elek G, et al. Value of carcinoembryonic anti-
gen (CEA) and cholesterol assays of ascitic fluid in cases of inconclu-
sive cytology. J Clin Pathol. 2001;54:831‐835.
10. Liu F, Kong X, Dou Q, et al. Evaluation of tumor markers for the dif-
ferential diagnosis of benign and malignant ascites. Ann Hepatol.
2014;13:357‐363.
11. Bijoor AR, Venkatesh T. Value of ascitic fluid cholesterol and serum‐
ascites albumin gradient in differentiating cirrhotic and malignancy
related ascites. Indian J Clin Biochem. 2001;16:106‐109.
12. Rana SV, Babu SG, Kocchar R Usefulness of ascitic fluid cholesterol
as a marker for malignant ascites. Med Sci Monit. 2005;11:CR12‐142.
DU ET AL.
13. Gerbes AL, Jungst D. Role of cholesterol determination in ascitic
fluid analysis. Hepatology. 2009;50:1320.
14. European Association for the Study of the Liver and European Asso-
ciation for the Study of the Liver. EASL Clinical Practice Guidelines
for the management of patients with decompensated cirrhosis. J
Hepatol. 2018;69:406–460.
15. Koch TR. New tools for the diagnosis of peritoneal carcinomatosis?
Am J Gastroenterol. 2002;97:2133‐2134.
16. Khandwalla HE, Fasakin Y, El‐Serag HB. The utility of evaluating low
serum albumin gradient ascites in patients with cirrhosis. Am J Gas-
troenterol. 2009;104:1401‐1405.
17. Wu SS, Lin OS, Chen YY, et al. Ascitic fluid carcinoembryonic anti-
gen and alkaline phosphatase levels for the differentiation of primary
from secondary bacterial peritonitis with intestinal perforation. J
Hepatol. 2001;34:215‐221.
18. Farias AQ, Silvestre OM, Garcia‐Tsao G, et al. Serum B‐type natri-
uretic peptide in the initial workup of patients with new onset
ascites: a diagnostic accuracy study. Hepatology. 2014;59:1043‐
1051.
19. DeLeve LD, Valla DC, Garcia‐Tsao G, et al. Vascular disorders of the
liver. Hepatology. 2009;49:1729‐1764.
20. Plessier A, Rautou PE, Valla DC. Management of hepatic vascular
diseases. J Hepatol. 2012;56(Suppl. 1):S25‐S38.
21. Kan X, Ye J, Rong X, et al. Diagnostic performance of Contrast‐en-
hanced CT in pyrrolizidine alkaloids‐induced hepatic sinusoidal
obstructive syndrome. Sci Rep. 2016;6:37998.
22. Li X, Yang X, Xu D, et al. Magnetic resonance imaging findings in
patients with pyrrolizidine alkaloid‐induced hepatic sinusoidal
obstruction syndrome. Clin Gastroenterol Hepatol. 2017;15:955‐957.
23. Zhang M, Li Y. Eosinophilic gastroenteritis: a state‐of‐the‐art review.
J Gastroenterol Hepatol. 2017;32:64‐72.
24. Castaldo G, Oriani G, Cimino L, et al. Total discrimination of peri-
toneal malignant ascites from cirrhosis‐ and hepatocarcinoma‐associ-
ated ascites by assays of ascitic cholesterol and lactate
dehydrogenase. Clin Chem. 1994;40:478‐483.
25. Jungst D, Xie Y, Gerbes AL. Pathophysiology of elevated ascites fluid
cholesterol in malignant ascites. Increased ascites to serum relation
of proteins and lipoproteins in patients with peritoneal carcinomato-
sis as compared to patients with cirrhosis of the liver. J Hepatol.
1992;14:244‐248.
SUPPORTING INFORMATION
Additional supporting information will be found online in the
Supporting Information section at the end of the article.
How to cite this article: Du L, Zhu S, Lu Z, et al. Ascitic
cholesterol is superior to serum‐ascites albumin gradient in
the detection of non‐portal hypertensive ascites and the
diagnosis of mixed ascites. Aliment Pharmacol Ther.
2018;00:1–8. https://doi.org/10.1111/apt.15042