Keywords
upper GI endoscopy - gastroesophageal varices - cirrhosis of liver
Introduction
Variceal bleed is one of the most commonly encountered emergencies in patients with
cirrhosis. In-hospital mortality remains as high as 20% in spite of advances in management.[1] Baveno VI consensus statement and the National Institute for Health and Care Excellence
(NICE) guidelines on acute upper gastrointestinal (UGI) bleeding are widely used in
managing patients of variceal bleed.[2]
[3]
There is a discrepancy in the epidemiological profile, management protocols, and outcomes
worldwide. A nationwide survey on management and outcomes on variceal bleed in the
UK considered prevalence of variceal bleed as a cause for UGI bleed, timing of presentation
to hospital, etiology of varices, patients in whom endoscopy could be done within
24 hours of presentation, percentage of patients requiring anesthesia for UGI endoscopy,
high risk stigmata, need for endoscopic therapy, etc.[4] The National Confidential Enquiry into Patient Outcome and Death (NCEPOD) report
in the UK pointed out the delays in endoscopy, clinical as well as organizational
care, and need for out of hours dedicated gastrointestinal (GI) bleed services.[5] In-hospital mortality of patients with cirrhosis and variceal bleeding decreased
threefold over the past two decades, in concurrence with an early and combined use
of pharmacological and endoscopic therapies and short-term antibiotic prophylaxis.[6]
In spite of improved outcomes after variceal bleed, the necessity to optimize the
management strategies persists. There are gray areas in managing variceal bleed such
as duration of terlipressin therapy after variceal band ligation (VBL), use of proton
pump inhibitors, referral for transjugular intrahepatic portosystemic shunt (TIPSS)
after VBL, primary prophylaxis for gastric variceal bleed, antibiotics in variceal
bleed, etc.[7] A tool kit developed to provide safer services in UGI bleed also recommends that
all hospitals must collect a minimum dataset to measure service provision against
auditable outcomes.[8] This underscores the importance of reporting data from all over the world. Several
studies have done an audit of management, outcomes, and prognostic indicators.[9]
[10] Indian data in this regard, particularly variceal bleed management and outcomes,
are scarce. We decided to conduct an audit of cases with variceal bleed who have undergone
endoscopic therapy.
Materials and Methods
Study Design
It was a retrospective study of patients presenting with variceal bleed in a tertiary
care center. Data were generated through a computerized electronic record system.
Data of patients admitted for acute variceal bleed during August 2018 to December
2018 were collected and considered for analysis. All patients underwent endotherapy
within 24 hours of presentation. As a departmental protocol, all patients received
terlipressin on admission and every sixth hourly, which was continued for 24 hours
after endotherapy. All patients received antibiotics before and after endotherapy
till discharge or death. Patients with diagnosis of hepatorenal syndrome, spontaneous
bacterial peritonitis, multiorgan dysfunction, sepsis, hepatic encephalopathy, etc.,
were excluded from analysis to avoid bias in mortality results. Esophageal varices
were classified into small (<5 mm) and large varices (>5 mm), which is approved by
the American Association for the Study of Liver Diseases. Gastric varices were classified
according to Sarin’s classification.
Variceal hemorrhage was defined as bleeding from an esophageal or gastric varix at
the time of endoscopy or the presence of large esophageal varices with blood in the
stomach and no other recognizable cause of bleeding.
An episode of bleeding was clinically significant when there is a transfusion requirement
for two units of blood or more within 24 hours of time zero, together with a drop
in systolic blood pressure of 20 mm Hg and/or pulse rate >100 beats per minute at
time zero (time zero is the time of admission to the first hospital to which the patient
was taken).
The acute bleeding episode is defined by an interval of 120 hours (5 days) from time
zero. Evidence of any bleeding after 120 hours was the first rebleeding episode. Variceal
rebleeding was defined as the occurrence of a single episode of clinically significant
rebleeding from portal hypertensive sources from day 5. Clinically significant rebleeding
was defined as recurrent melena or hematemesis in any of the following settings: (1)
hospital admission; (2) blood transfusion; and (3) 30 g/L drop in hemoglobin.
Complete hemogram, lever function test, renal function test, and prothrombin time/international
normalized ratio (PT/INR) were done using standard methods. UGI endoscopy was done
by trained gastroenterologists.
Statistical Analysis
In the present study, the collected data were subjected to following statistical analysis.
Descriptive statistics used were mean, standard deviation, frequency, and proportion.
Inferential statistics including chi-squared test, Cramer’s V, and independent-samples
t-test were used for analyzing the various constraints among the study population.
All the statistical methods were performed through the Statistical Package for the
Social Sciences (SPSS) for windows package (version 22.0).
Results
During the study period, 107 patients were satisfying inclusion criteria and their
data were recorded. Descriptive parameters are tabulated in [Table 1].[2]
Table 1
Baseline characteristics
Parameter
|
Mean ± SD (CI)
|
Abbreviations: CI, confidence interval; INR, international normalized ratio; SD, standard
deviation.
|
Age (y)
|
48 ± 14.85 (11–81)
|
Hospital stay (d)
|
3.83 ± 2.08 (1–17)
|
Door to endoscopy time (d)
|
12.21 ± 5.91 (1–25)
|
Mean arterial pressure (mm Hg)
|
90.73 ± 15.36 (50–123.33)
|
Hemoglobin (g/dL)
|
7.71 ± 2.18 (2.6–13)
|
Platelet count (lacs per mL)
|
1.05 ± 0.66 (0.27–3.34)
|
Serum creatinine level (mg/mL)
|
0.97 ± 0.47 (0.70–4.13)
|
Serum total bilirubin (mg/mL)
|
3.06 ± 3.30 (0.30–1.92)
|
INR
|
1.63 ± 0.52 (0.96–2.90)
|
Results clearly revealed ([Table 2]) that majority of the cases (89.7%, n = 96) diagnosed were due to cirrhosis of liver. Etiological distribution of cirrhosis
of liver is shown in [Fig. 1]. Alcohol (75 out of 96, i.e., 78.1%) was the major contributor for cirrhosis. Most
of the cases (77.6%, n = 86) of esophageal varices were found to be large (>5 mm). Glue injection was required
in five (4.7%) cases. Only one patient (0.9%) rebled. Majority (90.7%, n = 97) of the cases were discharged. Majority of patients were in B and C of the Child–Turcotte–Pugh
(CTP) class (84.1%, n = 90) and very few of the cases were in CTP class A (15.9%, n = 17). Gastric varices were seen in 35 patients (37.7%) with gastroesophageal varices
1 (GOV1) being the most common subtype (14%, n = 15) followed by combination of GOV1 and GOV2 (10.3%, n = 11). GOV2 was the least common subtype (1.9%, n = 2). Type 2 isolated gastric varices and ectopic varices were not seen in any of
the cases.
Table 2
Percent distribution of diagnostic and treatment variables
Variables
|
|
No of patients, n = 107 (%)
|
Abbreviations: CTP, Child–Turcotte–Pugh; EHPVO, extrahepatic portal venous obstruction;,
GLUE-INJ, glue injection; GOV, gastroesophageal varices; GV, gastric varices; IGV,
isolated gastric varices; NCPF, noncirrhotic portal fibrosis.
|
Diagnosis
|
Cirrhosis of liver
|
96 (89.7)
|
EHPVO
|
7 (6.5)
|
NCPF
|
4 (3.7)
|
Esophageal varices
|
Large
|
83 (77.6)
|
Small
|
24 (22.4)
|
Outcome
|
Discharged
|
97 (90.7)
|
|
Died
|
10 (9.3)
|
CTP
|
A
|
17 (15.9)
|
|
B
|
39 (36.4)
|
|
C
|
51 (47.7)
|
GV type in positive cases
|
GOV1
|
15 (14.0)
|
|
GOV2
|
2 (1.9)
|
|
GOV1 + GOV2
|
11 (10.3)
|
|
IGV1
|
7 (6.5)
|
Gastric varices
|
35 (32.7)
|
GLUE-INJ
|
5 (4.7)
|
Diabetes Mellitus
|
20 (18.7)
|
Rebleed
|
1 (0.9)
|
Fig. 1 Etiological distribution of cirrhosis. Alcohol, 75; HBV, 7; NAFLD, 7; Alcohol + HBV,
4; HCV, 3. HBV, hepatitis B virus; HCV, hepatitis C virus; MAP, mean arterial pressure;
MELD, model for end-stage liver disease; NAFLD, nonalcoholic fatty liver disease;
PLT, platelet; TB, total bilirubin.
Significant associations ([Table 3]) were observed between diagnosis and outcome (p = 0.013), where majority of the cases with cirrhosis of liver discharged; however,
more of mortality cases had noncirrhotic portal fibrosis (p = 0.001). Low mean arterial pressure (MAP), low platelet count, higher serum creatinine,
higher total bilirubin (TB), higher INR, and higher model for end-stage liver disease
(MELD) were associated with increased risk of mortality ([Fig. 2] [Table 4]).
Table 3
Analysis of variables in discharged and expired patients
Variable
|
Outcomes, n = 107 (%)
|
p-Valuea
|
|
Discharged, n = 97
|
Died, n = 10
|
|
Abbreviations: CTP, Child–Turcotte–Pugh; EHPVO, extra hepatic portal venous obstruction;
NCPF, noncirrhotic portal fibrosis; SD, standard deviation.
a
p-Value calculated using chi-squared test.
|
Age
|
|
47.90 ± 15.15
|
49.00 ± 12.24
|
0.824
|
(Mean ± SD)
|
(Mean ± SD)
|
<30 y
|
11 (11.2)
|
0 (0)
|
0.414
|
31–60 y
|
70 (72.2)
|
9 (90.0)
|
|
>60 y
|
16 (16.5)
|
1 (10.0)
|
|
Gender
|
Male
|
74 (76.3)
|
9 (90)
|
0.322
|
Female
|
23 (23.7)
|
1 (10.1)
|
|
Diagnosis
|
Cirrhosis of liver
|
88 (90.7)
|
8 (80.0)
|
0.013
|
EHPVO
|
7 (7.2)
|
0
|
|
NCPF
|
2 (2.1)
|
2 (20.0%)
|
|
Esophageal varices
|
Large
|
74 (76.3)
|
9 (90.0)
|
0.322
|
Small
|
23 (23.7)
|
1 (10.0)
|
|
Door to endoscopy
|
<12 h
|
45 (46.4)
|
4 (40.0)
|
0.699
|
>12 h
|
52 (53.6)
|
6 (60.0)
|
|
Rebleed
|
Yes
|
0 (0.0)
|
1 (10.0)
|
0.002
|
No
|
97 (100.0)
|
9 (90.0)
|
|
CTP
|
A
|
17 (17.5)
|
0 (0.0)
|
0.082
|
B
|
37 (38.1)
|
2 (20.0)
|
|
C
|
43 (44.3)
|
8 (80.0)
|
|
Fig. 2 Parameters associated significantly with increased risk of mortality due to variceal
bleed in present study. INR, international normalized ratio; MAP, mean arterial pressure;
MELD, model for end-stage liver disease; PLT, platelet; TB, total bilirubin.
Table 4
Analysis of investigations in discharged and expired patients
Variable
|
Outcomes, n = 107 (Mean ± SD)
|
p-Value
|
|
Discharged, n = 97
|
Died, n = 10
|
|
Abbreviations: INR, international normalized ratio; MAP, mean arterial pressure; MELD,
model for end stage liver disease; SD, standard deviation.
|
MAP (mm Hg)
|
93.90 ± 12.09
|
60.00 ± 7.86
|
0.001
|
Hemoglobin (g/dL)
|
7.81 ± 2.24
|
6.85 ± 1.40
|
0.189
|
Serum platelet (lacs per mL)
|
1.13 ± 0.66
|
0.41 ± 0.07
|
0.001
|
Serum creatinine
|
0.85 ± 0.15
|
2.15 ± 0.85
|
0.001
|
Total bilirubin
|
2.25 ± 1.69
|
10.94 ± 4.71
|
0.001
|
INR
|
1.53 ± 0.44
|
2.61 ± 0.30
|
0.001
|
Hospital stay
|
3.91 ± 2.15
|
3.10 ± 1.29
|
0.247
|
MELD
|
13.47 ± 3.96
|
32.60 ± 4.50
|
0.001
|
[Table 3] also revealed nonsignificant association between CTP class and outcome as well as
no difference between discharged and mortality cases in their Hb values.
Discussion
Out of 107 cases analyzed, 9.3% (n = 10) died. This is comparable to other (10%) reported studies done on a larger sample
size (10–15%).[11] Overall percentage of patients with gastric varices was 32.7%, which is higher than
a usual of around 20% in patients with cirrhosis.[12] It is recommended that an acute variceal bleed should undergo endoscopic therapy
in less than 12 hours.[13]
[14] Forty-nine out of 107 (45.7%) patients in our study have undergone endoscopic variceal
ligation (EVL) within this time span. Patient’s poor general condition requiring treatment
and stabilization before endoscopy and delay on part of patient’s attenders to give
consent for the procedure were reasons for remaining patients in whom endoscopy was
done after 12 hours. All patients have undergone endoscopy within 24 hours of presentation.
Moreover, door to endoscopy time does not seem to affect the survival in our study
(p = 0.699).
With advancement of liver disease, chances of variceal bleed also increase from 0%
for CTP class A up to 30% for CTP class C.[15]
[16] We report that majority of our patients were of CTP class B and C (84.1%, n = 80). Proportion of CTP class C patients in expired group is 80% compared with 44%
in alive group. Relatively well-compensated groups (CTP A and B) were 55% compared
with 20% in expired group. p-Value being insignificant (0.08) might be due to smaller sample size. Larger sample
size is needed for better validation of the p-value.
Only one among the 107 patients had rebleed in hospital and the p-value was statistically significant for rebleed in expired patients
In a similar multicentered retrospective study done on a larger sample size by Chalasani
et al, age, blood pressure, hemoglobin, bilirubin, creatinine, INR, MELD score, etc.,
parameters were studied and were comparable to our results.[9]
[17] MAP, low platelet counts, high serum creatinine, high TB, high INR, higher MELD,
and rebleed emerged as predictors of in-hospital mortality in our study. Age, sex,
door to endoscopy time of 12 to 24 hours, CTP class, duration of hospital stay, and
hemoglobin on admission were not statistically different across survived and expired
cases.
Main strength of our study is that it shows that the mortality due to variceal bleed
can be minimized if patients are timely managed. Even though 5 days of intravenous
terlipressin is recommended, it has not shown to be beneficial in improving survival.[7] It also adds to the cost of treatment. This is particularly important in developing
countries like India. We have treated patients with terlipressin for 24 hours with
comparable survival rate. In this regard, our study highlights that treating patients
with terlipressin for 24 hours is not only cost effective but also proved to be equally
beneficial in reducing mortality.
This was a retrospective study done on a relatively small sample size and hence there
are limitations in extrapolating the inferences to all patients with variceal bleed
in whom endotherapy was done. Since there were limitations in accessing the data on
risk stratification for risk factors like hepatocellular carcinoma (HCC), portal vein
thrombosis, and other factors like obliteration or nonobliteration of varices after
EVL, evaluation of the development of post-EVL ulcers could not be done. Since our
study focused on mortality due to variceal bleed, it does not give any idea about
the all-cause mortality. Fibroscan has a limited sensitivity in patients with ascites
and obesity.[18] Moreover, the cut-off value of F2 and F4 fibrosis varies with the etiology.[19] Hence, we could not study the association of degree of fibrosis and mortality. Patients
in whom endotherapy was unsuccessful could not undergo TIPSS, balloon-occluded retrograde
transvenous obliteration, or endoscopic ultrasound-guided coiling due to unavailability
of these modalities in our institute, which possibly could have further reduced bleeding-related
mortality.
Owing to the dearth of parallel studies, particularly from India, our study throws
light on the need for a thorough assessment of patients, risk stratification, and
data recording and publication to set the standard operating procedure protocol suitable
in Indian context in future.
Conclusion
In conclusion, the mortality rate due to variceal bleed was 9.3% in our study. Rebleeding
occurred in as low as 0.9%. Alcohol-related cirrhosis of liver was the most common
etiology of variceal bleed. Low MAP, low platelet count, higher serum creatinine,
higher TB, higher INR, and MELD were associated with increased risk of mortality whereas
age, sex, door to endoscopy time, grade of varices, CTP grade, and hemoglobin on admission
did not emerge as predictors of mortality in the present study. However, we recommend
more number of prospective studies in this regard on a larger sample size.