Prognostic Factors Associated with Mortality in Cirrhotic Patients with Bleeding Varices

Introduction

Upper gastrointestinal bleeding (UGIB) is the loss of blood through the GI tract originating proximal to the Treitz angle.¹ Acute hemorrhage from varices is one of most menacing portal hypertension complications and related to high morbidity and mortality.² The prognosis of cirrhotic patients is related to severity of the hepatic condition liver which could be assessed using the Child–Turcotte–Pugh (CTP) classification, with higher scores having a significant impact on survival times.³

Portal hypertension leads to the development of portosystemic collateral venous vessels. Every year 5 to 10% of patients with cirrhosis will develop esophageal varices.⁴ This is more likely to occur in patients with progressive cirrhosis and may be present in up to 60% of patients with decompensated cirrhosis.⁵

Mortality rates following a first variceal bleed in cirrhotics increase with advancement of the Child score and range between 15 and 80%.⁶ The main causes of death are unstoppable hemorrhage, infection and kidney failure. There are multiple determinants linked to increased mortality and poor prognosis including high MELD scores, kidney failure, hepatic venous pressure gradients above 20 mm Hg, and evidence of an active bleed on endoscopy.^7,8

It is uncertain if the currently used prognostic scores, namely the CPT and MELD scores are reliable in determination of acute variceal hemorrhage risk.

The aim of the current study was to identify the prediction model most suitable for determining the outcome of acute variceal bleeding.

Patients and Methods

This study was a prospective study performed on 310 patients with upper gastrointestinal tract bleeding, 210 were excluded and 100 were included from October 2017 to December 2018. They were collected from Tropical Medicine Department and Internal Medicine Department, Tanta University Hospital.

Inclusion criteria:

• Cirrhotic patients with bleeding varices (esophageal, fundal, or both).

Exclusion criteria:

• Patients with hepatocellular carcinoma (HCC).

• Patients with upper GI hemorrhage due to causes other than ruptured varices (peptic ulcers and erosions, esophagitis, malignant masses, and vascular ectasia).

The study was approved by the ethical review board, Tanta University, Faculty of Medicine, with approval code 31779/09/2017. All patients participating in the study provided a signed informed consent. The study protocol abides by the guidelines of the Declaration of Helsinki 1975 guidelines and was approved by the faculty's human research committee.

A meticulous history was taken from all patients and they were subjected to thorough examination and monitoring. All relevant laboratory tests were performed including baseline and serial liver function tests and serum creatinine. Blood counts, electrolyte, and arterial blood gas levels, as well as amounts of blood transfused were recorded. An upper GI endoscopy was done to diagnose the cause of bleeding and enable making appropriate treatment decisions. All the patients underwent abdominal ultrasonography.

The recorded baseline data were used to calculate various prognostic scores including CTP, MELD, APACHE II (acute physiology and chronic health evaluation II), AIMS65, sepsis-associated organ failure assessment (SOFA) and the ROCKall scores.

The primary end point: survival and rebleeding within 6 weeks after the first variceal bleeding attack.

The secondary end point: survival for 6 months after the first variceal bleeding attack.

Statistical Analysis

Mean, standard deviation, Student's t-test, and chi-square test were performed by Statistical Package for Social Sciences (IBM SPSS StatisticsVersion 20) and receiver operating characteristic (ROC) curve.

Results

This study was a prospective study performed on 310 patients with upper GIT bleeding presented to Tanta University Hospital: 210 patients had been excluded (either refused or not met the inclusion criteria) and 100 patients had been included from October 2017 to December 2018. After the first attack of bleeding 85 patients lived and 15 patients died; after 6 months 64 patients were survivors and 21 did not survive. Total deaths after 6 months were 36 patients (►Fig. 1).

Close

Fig. 1

Study analysis population.

Export to PPT

Risk Factors for Death After the First Attack of Bleeding

After the first attack of bleeding 15 patients died out of total 100 patients enrolled in the study. Regarding age, sex, smoking, presence of ascites, diabetes mellitus, and hypertension there were no statistically significant differences between the patients.

There were significant differences between the studied patients as regards the number of blood units (p < 0.001). More the blood units needed, worse the prognosis. There were no significant differences between the studied patients as regards Sungestaken tube inflation before the endoscopy, associations (hepatic encephalopathy, spontaneous bacterial peritonitis, and hepatorenal syndrome), proton pump infusion before endoscopy, and beta blockers intake (►Table 1).

Table 1 Clinical data in patients first 6 weeks after first attack of bleeding

		Outcome after first bleeding attack						Chi-square
		Nonsurvivor		Survivor		Total
		N	%	N	%	N	%	χ2	p-Value
Complaint	Hematemesis	7	46.67	34	40.00	41	41.00	1.194	0.550
	Melena	3	20.00	29	34.12	32	32.00
	Both	5	33.33	22	25.88	27	27.00
Number of blood units	No	7	46.67	22	25.88	29	29.00	11.291	0.046a
	One	0	0.00	30	35.29	30	30.00
	Two	6	40.00	14	16.47	20	20.00
	Three	1	6.67	9	10.59	10	10.00
	Four	1	6.67	9	10.59	10	10.00
	Five	0	0.00	1	1.18	1	1.00
Sungestaken tube usage	No	14	93.33	84	98.82	98	98.00	1.961	0.161
	Yes	1	6.67	1	1.18	2	2.00
Association (SBP, HRS, HE)	No	5	33.33	35	41.18	40	40.00	0.327	0.568
	Yes	10	66.67	50	58.82	60	60.00
Proton pump infusion	No	0	0.00	3	3.53	3	3.00	0.546	0.460
	Yes	15	100.00	82	96.47	97	97.00
Beta blockers	No	13	86.67	67	78.82	80	80.00	0.490	0.484
	Yes	2	13.33	18	21.18	20	20.00

Abbreviations: HE, hepatic encephalopathy; HRS, hepatorenal syndrome; SBP, spontaneous bacterial peritonitis.

^aStatistically significant.

There was no significant difference between the studied patients as regards the endoscopic findings, variceal grades, and method of intervention.

There was significant difference between the studied patients as regards systolic blood pressure–the lower the systolic blood pressure, the worse the prognosis. Also thrombocytopenia, increased serum creatinine, and INR are risk factors for increased death rate (►Table 2).

Table 2 Patients scores and laboratory data in first attack of bleeding

	Outcome after first bleeding attack		t-test
	Nonsurvivors (mean ± SD)	Survivors (mean ± SD)	T	p-Value
Hemoglobin	8.413 ± 1.215	8.482 ± 1.031	0.233	0.817
WBCs	6.160 ± 4.577	6.206 ± 3.904	0.041	0.967
Platelets	135.000 ± 55.554	104.188 ± 49.957	2.166	0.033a
ALT	31.400 ± 11.306	35.800 ± 36.419	0.462	0.645
AST	52.533 ± 17.067	56.812 ± 28.533	0.562	0.576
Bilirubin	3.467 ± 4.667	2.339 ± 2.558	1.364	0.176
Albumin	2.733 ± 0.623	2.965 ± 0.474	1.660	0.100
INR	1.593 ± 0.616	1.339 ± 0.396	2.090	0.039a
Creatinine	1.313 ± 0.558	1.037 ± 0.352	2.546	0.012a
Urea	57.333 ± 27.105	45.882 ± 32.298	1.294	0.199
Child Score	8.867 ± 3.067	8.000 ± 2.345	1.257	0.212
MELD score	17.600 ± 7.935	12.682 ± 5.583	2.938	0.004a
AIMS56	2.133 ± 1.552	1.071 ± 0.961	3.561	0.001a
APACHE II	7.267 ± 6.147	4.459 ± 3.194	2.666	0.009a
SOFA score	3.467 ± 1.767	3.376 ± 1.739	0.185	0.854
ROCKall	4.667 ± 1.839	3.212 ± 1.544	3.269	0.001a

Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; INR, international normalized ratio.

^aStatistically significant.

Significant differences were noted between the studied patients as regards the MELD score, AIMS56 score, APACHE II score, and ROCKall score (p ≤ 0.001). Increased MELD, AIMS56, APACHE II, and ROCKall scores were risk factors for mortality. Child score wasn't associated with mortality (►Table 2; ►Fig. 2).

Close

Fig. 2

Receiver operating characteristic curve for scores for outcome after first bleeding attack.

Export to PPT

AIMS56 score has 92.94 sensitivity, 26.67 specificity, and 57.7% accuracy.

ROCKall score has 80.00 sensitivity, 73.33 specificity, and 74.4% accuracy. ROCKall score was the most suitable score for predicting mortality after the first attack of bleeding.

Factors Affecting Outcome 6 Months After Bleeding

The studied patients were matched as regards age or sex (►Table 4). Presence of ascites was associated with decreased survival.

Table 4 Clinical and laboratory data in patients after 6 months

		Outcome after 6 months		t-test
		Died	Alive	T	p-Value
Hemoglobin	Range	6.4–11.5	6.1–10.5	1.067	0.289
	Mean ± SD	8.690 ± 1.147	8.414 ± 0.990
WBCs	Range	2.5–23.9	2.3–24.8	2.108	0.038a
	Mean ± SD	7.733 ± 5.217	5.705 ± 3.263
Platelets	Range	40–150	40–430	−1.569	0.121
	Mean ± SD	89.476 ± 33.957	109.016 ± 53.533
ALT	Range	17–343	10–62	2.393	0.019a
	Mean ± SD	51.857 ± 68.623	30.531 ± 12.623
AST	Range	26–245	12–133	2.956	0.004a
	Mean ± SD	72.095 ± 44.050	51.797 ± 19.134
Total bilirubin	Range	0.9–21	0.6–8.8	3.434	0.001a
	Mean ± SD	3.905 ± 4.246	1.825 ± 1.385
Direct bilirubin	Range	0.2–15.1	0.2–5	3.117	0.003a
	Mean ± SD	2.319 ± 3.156	0.988 ± 0.801
Serum albumin	Range	2–3.4	2–4.1	−4.966	<0.001a
	Mean ± SD	2.571 ± 0.345	3.094 ± 0.439
INR	Range	1.03–3.4	1–2.2	3.511	0.001a
	Mean ± SD	1.586 ± 0.611	1.258 ± 0.252
Prothrombin activity	Range	24–95	34–100	−2.636	0.010a
	Mean ± SD	65.214 ± 24.094	77.438 ± 16.238
Serum creatinine	Range	0.6–2.6	0.6–2	2.991	0.004a
	Mean ± SD	1.227 ± 0.458	0.974 ± 0.287
Urea	Range	24–220	20–154	3.130	0.002a
	Mean ± SD	64.095 ± 50.075	39.906 ± 21.165

Abbreviations: ALT, alanine transaminase; AST, aspartate transaminase; DBP, diastolic blood pressure; INR, international normalized ratio; SBP, systolic blood pressure.

^aStatistically significant.

Associations as hepatic encephalopathy and spontaneous bacterial peritonitis were associated with decrease survival.

There were significant differences between the studied patients as regards WBCs, ALT, AST, total bilirubin, direct bilirubin, serum albumin, INR, prothrombin activity, serum creatinine, and urea. Leucocytosis, increased ALT, AST, bilirubin, INR, activity, serum creatinine, and urea decrease albumin associated with decrease survival.

Child scores were significantly different between studied patient groups on the contrary to the MELD score (►Fig. 3).

Close

Fig. 3

Receiver operating characteristic curve between MELD and Child scores for 6 months survival.

Export to PPT

ROC curve shows that Child score had 65.62 sensitivity, 95.24 specificity, and 88.7% accuracy, whereas MELD score had 100.00 sensitivity, 21.43 specificity, and 55.5% accuracy.

Discussion

Variceal hemorrhage is a complication of cirrhosis that denotes decompensation and that still has a high mortality rate.⁹

Even though management of variceal hemorrhage has improved in the last decades, 6-week mortality following a GI bleed remains high at around 10 to 20%, and rises with advanced cirrhosis.¹⁰

In this study, there was significant difference between the studied patients as regard blood pressure. Our findings agree with those of Gado et al,¹¹ and Jiménez et al.¹² They found that hemodynamic instability at admission, Child class C, blood in GI tract at the index endoscopy, rebleeding within five days of endoscopy, and in-hospital complications were independent predictors of mortality.

We have detected significant differences between the studied patients as regards associations (hepatic encephalopathy, spontaneous bacterial peritonitis, thrombocytopenia, increase serum creatinine, raised INR, and prolonged prothrombin time), as did Goldis et al.¹³

No significant differences were recorded as regards proton pump inhibitor (PPI) infusion. This finding was in disagreement with Komori et al,¹⁴ who found an association between regular PPI treatment before and after the onset of variceal bleeds and increments in short as well as long-term mortalities. This could be explained by the increased risk of infections such as bacterial infections in general and spontaneous bacterial peritonitis following regular PPI use.^15,16

We detected significant difference between the studied patients as regards MELD score, ROCKall score, and AIMS65 score. This finding was in agreement with Saltzman et al,¹⁷ and Gado et al.¹¹ This finding was in disagreement with Mohammad et al,¹⁸ who reported the superiority of the SOFA score in mortality prediction when compared with the MELD, APACHEII, and CPT scores and concluded that independently the AIMS65 score was the simplest and most applicable scoring system for mortality prediction among cirrhosis patients suffering from acute variceal hemorrhage.

Conclusion

High MELD, AIMS56, APACHE II, and ROCKall scores were risk factors for mortality after acute variceal bleeding. High death rate during the first 6 weeks is associated with anemia, hypotension, thrombocytopenia, increased serum creatinine, and INR. Decreased survival at 6 months is associated with increased Child score, presence of ascites, and associations such as hepatic encephalopathy and spontaneous bacterial peritonitis.