Alterations in colonic mucosal lesions in patients with portal hypertension

Introduction 

Portal hypertension (PHT) is a frequent complication of chronic liver disease, detected not only in schistosomiasis but also in cirrhosis of any aetiology [1]. Various vascular abnormalities have been observed in the upper GI tract of cirrhotic patients, including gastro-oesophageal varices and gastric antral vascular ectasia. These vascular lesions account for most of the upper GI bleeding in cirrhotic patients [2], [3]. Other parts of the GI mucosa such as the duodenum, jejunum [4], [5] and the colon [6], [7] have also been noted to be affected by PHT. Colonic mucosal erythema, vascular ectasia, colorectal varices and haemorrhoids have been described as source of lower GI bleeding in patients with PHT [8]. The features of PHC are not well defined but include vascular lesions, colitis-like abnormalities, rectal varices, or a combination of these findings [3]. The diagnostic criteria and clinical significance of this condition is confusing. This may be partially due to imprecise terminology, lack of uniform endoscopic descriptions, interobserver variability and the absence of distinctive histopathologic features [3], [9]. Major histopathological changes of PHC include dilated tortuous mucosal capillaries with irregular wall thickening, oedema of lamina propria and mild chronic inflammatory infiltrate [10], [7]. These mucosal lesions are generally diagnosed endoscopically but histological confirmation is needed when other similar conditions such as undefined colonic lesions are suspected.

Our study aimed to evaluate the prevalence of colonic mucosal changes in patients with liver cirrhosis and their clinical significance.

Patients and methods 

This is a prospective study including 40 patients with post-viral liver cirrhosis and PHT admitted to the Tropical Medicine Department and referred to Gastrointestinal Endoscopy Unit, Kasr-El-Aini Hospital, Cairo University and Ahmed Maher Teaching Hospital in the period from June 2007 to July 2008. PHT was diagnosed on the basis of suggestive clinical data (past variceal bleeding, ascites), presence of portosystemic collaterals, splenomegaly, ascites on abdominal ultrasound and lastly the diagnosis of oesophageal or gastric varices by upper GI endoscopy. The study was approved by the ethical committee of the Tropical Medicine Department Kasr-El-Aini Hospital, Cairo University. Patients on medical treatment of PHT, those with active bilharzial and other parasitic colitis (by stools analysis and rectal snip), active alcohol ingestion and diseases causing vascular abnormalities, i.e., chronic nephropathy, aortic stenosis, colostomy or inflammatory bowel disease were excluded from the study.

After obtaining an informed consent, upper GI endoscopy was performed in all patients using an Olympus GF-240 videoscope or a Pentax EG 2940 videoscope. The presence and grades of oesophageal varices, gastric varices and portal congestive gastropathy were assessed according to de Franchis [11]. Liver disease severity was assessed according to Child-Pugh’s classification. All patients underwent a full length colonoscopy (up to the caecum) using a Pentax EC3440F colonoscope after preparation with magnesium citrate solution, clear oral fluids for 24h before the procedure and tap water enemas 2h before the procedure. Diazepam 10mg or midazolam 5mg IV either alone or in combination with pethidine (25–100mg) was used for conscious sedation. PHC was diagnosed endoscopically by the presence of vascular ectasia, hyperaemic colonic mucosa, angiodysplastic lesions and rectal varices (the latter were defined endoscopically according to Thakeb et al. [12]). According to Bini et al. [6] PHC was classified into mild (erythema of the colonic mucosa), moderate (presence of vascular ectasia) and severe (presence of rectal varices).

Multiple mucosal biopsies were taken from the rectum and sigmoid colon (including rectal snip). Furthermore, a biopsy was taken from any abnormal lesion in the colon other than angiodysplastic areas. Biopsies were fixed using formalin and embedded in paraffin blocks, from which, at least four 5 micron-thick sections were cut and stained with haematoxylin and eosin for histopathological examination. Stained sections were assessed for the presence of vascular ectasia and wall thickness, oedema and inflammatory cellular infiltrate. Histopathological features suggestive of PHC included the presence of dilated tortuous mucosal capillaries with irregular wall thickening, oedema of the lamina propria and mild chronic inflammatory infiltrate [10], [7], [13].

Data are expressed as mean±SD. Statistical comparisons were made with the χ2 test. P value less than 0.05 is considered significant. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) were calculated.

Results 

Forty patients with liver cirrhosis and PHT were included in this study. These were 60% males and 40% female. The age ranged from 30 to 80years (mean±SD: 51±11years). The main cause of liver cirrhosis was related to hepatitis C (85%) or B (10%) infection and combined in 5% of cases. Most of the patients fit into Child-Pugh C class 70%, Child B in 22.5%. History of lower GI bleeding (eight cases) and regular sessions of sclerotherapy or band ligation (six cases) were reported. Laboratory investigations of the studied patients are shown in Table 1. Abdominal ultrasound revealed criteria of liver cirrhosis, splenomegaly in 95% of patients and no evidence of focal hepatic lesions.

Table 1. Laboratory profile of the studied patients.

	Laboratory profile	Mean	SD
	Total bilirubin (mg/dl)	2.2	2.3
	ALT (IU/ML)	41	27
	AST (IU/ML)	48	24
	Albumin (g/dl)	2.6	0.6
	INR	1.6	3.8
	PLT (150–450×103/mm3	114	79
	Hb (13–17.5g/dl)	11	2

Upper GI endoscopy revealed different grades of oesophageal, gastric varices and PHG (Table 2). Colonoscopic examination revealed normal colonic mucosa in 11 patients (27.5%). Different colonoscopic lesions were reported: haemorrhoids in 17 patients (42.5%), diffuse hyperaemic colonic mucosa in 16 patients (40%), angiodysplastic lesions in 13 patients (32.5%), rectal varices in seven patients (17.5%), colonic polyps (inflammatory by histopathology) in four patients (10%) and multiple rectal ulcers in one patient (2.5%) (Fig. 1). Features of PHC were found in the descending and recto-sigmoid colon. Mild colopathy was diagnosed in 10 patients (25%), moderate in 12 patients (30%) while seven patients (17.5%) had severe colopathy [6].

Table 2. Upper endoscopic findings of the studied patients.

	Upper endoscopy	N	%
	Oesophageal varices	33	82.5%
	F1/F2/F3	5/13/15	12.5/32.5/37.5
	Gastric varices	8	20%
	GEV-1/GEV-2	4/1	10/2.5
	IGV-1/IGV-2	3/0	7.5/0
	PHG	35	87.5%
	Mild/severe PHG	20/15	50/37.5

AST (aspartate transaminase), ALT (alanine transaminase), INR (international normalized ratio), gastro-oesophageal varices type1 (GEV-1), gastro-oesophageal varices type2 (GEV-2), isolated gastric varix type 1 (IGV-1), isolated gastric varix type 2 (IGV-2); portal hypertensive gastropathy (PHG).

View Large Image Download to PowerPoint

Fig. 1. The colonoscopic findings of the studied patients.

PHC significantly correlated with Child–Pugh score and the mere presence of oesophageal varices (p value=0.01, <0.001, respectively). On the other hand, no statistically significant correlation was found between the grading of varices, presence of gastropathy or repeated sessions of sclerotherapy or band ligation and the presence of PHC.

Regarding laboratory investigations, platelet count significantly correlated with the presence of angiodysplasia (p value 0.001). Other laboratory investigations (ALT, Albumin, INR) had no significant correlation.

History of lower GI bleeding (that was reported by eight patients) was significantly associated with the presence of haemorrhoids and rectal varices (Fig. 2) (p value 0.001 and <0.001, respectively), but not with angiodysplasia or hyperaemic mucosa.

View Large Image Download to PowerPoint

Fig. 2. Rectal varix with a nipple on top in a case of bleeding per rectum.

Histopathological examination of endoscopic biopsies from rectum and sigmoid colon revealed normal colonic mucosa in 15% and 17.5% of cases, respectively while histopathological evidence of PHC (Fig. 3) was found in 85% of patients and 82.5%, respectively.

View Large Image Download to PowerPoint

Fig. 3. Colonic biopsy with evident histopathological features of PHC. The lamina propria shows multiple ectatic blood vessels with irregular wall thickening together with oedema and mild non-specific chronic inflammatory cellular infiltrate. The colonic crypts are normal (haematoxylin & eosin 200×).

Table 3 summarizes the correlation between the histopathological evidence and colonoscopic features of PHC. Thirty four patients (85%) had histopathological evidence of PHC, 27 patients of these had coexisting colonoscopic features of PHC. Evaluation of colonoscopic features of PHC revealed a sensitivity of 79%; specificity of 66.6%, PPV of 93% and a NPV of 36.4% taking histopathology as the gold standard for diagnosis.

Table 3. Relation between the histopathological evidence and colonoscopic features of PHC.

	Histopathological evidence of PHC	Colonoscopic features of PHC		P value
		Absent (11)	Present (29)
	Absent (6)	4	2	0.02
	Present (34)	7	27

Discussion 

The features of PHC are not well defined but these include vascular lesions, colitis-like abnormalities, rectal varices or a combination of these findings [3]. There is confusion regarding the diagnostic criteria of PHC and clinical significance of this condition. This might be due to inaccurate terminology, lack of uniform endoscopic descriptions, interobserver variability and the absence of distinctive histopathologic features [3], [9].

The current study was conducted on 40 patients with liver cirrhosis and PHT. Full length colonoscopy, revealed endoscopic features suggestive of PHC located in the recto-sigmoid and descending colon in 72.5% of patients while 27.5% of patients had normal colonic mucosa. None of the parameters (grade of oesophageal varices, endoscopic variceal eradication, presence of PHG or gastric varices) affected the occurrence of PHC. Previous studies by other authors [14], [1], [9] reported features of colopathy in 92%, 82% and 66% of cirrhotic patients, respectively.

With further classification of colopathy into grades as described by Bini et al [6], the current study revealed, mild, moderate and severe forms of colopathy in 40%, 32.5% and 17.5% of patients, respectively. Similarly, mild colopathy was demonstrated in 36.6%, 42% and 54%, respectively as reported by other investigators [15], [1], [9] and moderate colopathy in 36% of patients [1]. Severe colopathy was reported by authors [9], [16] in 31% and 40% of cases, respectively. On the other hand, lesser prevalence of severe colopathy (13% and 12%) was documented [6], [1]. These differences may be attributed to different aetiologies of liver disease (alcoholic or post-viral) or interobserver variability. Moreover, the wide range in the incidence of anorectal varices could be due to the absence of a clear grading system as well as the different aetiologies of liver cirrhosis or PHT (higher prevalence in extrahepatic portal hypertension) as reported by Thakeb et al. [12].

Our results showed that prevalence of PHC increased with worsening of Child-Pugh class. This could be due to increased haemodynamic dysfunction in those with more advanced liver disease [15], [1]. However, this relation was not confirmed by others [9].

Colonic manifestations of PHT often coexist with other features such as oesophageal varices and PHG and seem likely to be related, at least in part to the degree of elevation of portal venous pressure [17]. In the present study, a significant correlation was found between the mere presence of oesophageal varices and PHC. Similarly, other authors [8], [6] were able to demonstrate the same findings, others [15], [1], [9], however, did not verify this correlation in their study.

In the current study, none of parameters (grades of oesophageal varices, presence of gastric varices and congestive gastropathy or its severity) had a significant correlation with PHC. Similarly Misra et al. [8] and Bini et al. [6] found no clear relationship between large oesophageal varices or severity of portal gastropathy and PHC. On the other hand, predominant PHC (87.5%) was observed in patients with large oesophageal varices [18] and a significant association was verified between both PHC and gastropathy [19].

Variceal obliteration did not have a significant effect on the prevalence of mucosal abnormalities in the colon as concluded in the current study. This is in accordance with previous reports [15], [1], [9]. Similarly, variceal band ligation and sclerotherapy did not affect the incidence of haemorrhoids, anorectal varices or PHC [16], [19]. In contrast, Bini et al. [6] demonstrated increased prevalence of PHC post-band ligation and/or sclerotherapy and attributed this to the redistribution of blood flow in the stomach and colon with increased prevalence of gastropathy and colopathy. These different results may be due to different in the sample size and the percentage of patients who had done endoscopic interventions with achievement of variceal obliteration in previous studies.

The clinical significance of PHC as a potential source of clinically significant lower GI hemorrhage was emphasized by some investigators [8], [20]. Low frequency (1%) of bleeding from rectal varices was reported compared to oesophageal varices [17] .This low frequency may be related to the relatively thicker walls and deeper site of rectal varices compared to those in the oesophagus. Herein, 20% of patients had history of bleeding per rectum that significantly correlated with the presence of haemorrhoids and rectal varices but not with the presence of hyperaemic colonic mucosa or angiodysplastic lesions. In agreement with our results, Ghoshal et al. [15] concluded that haematochezia was associated with the detection of colorectal varices but not PHC.

In the current study, the results of histopathological evidence of PHC were in concordance with endoscopic features. The significant association between the endoscopic and histopathological findings of PHC underlines the importance of colonoscopic examination. Thus, PHC can be suggested on the basis of macroscopic appearance without the need for histopathological confirmation. In contrast to our data, Misra et al. [7] concluded that the major histopathological changes in colonic mucosa in PHT had no relation with the clinical or endoscopic findings. Like we did in this study, some studies have included the histopathological entity of mild chronic inflammatory infiltrate in the diagnosis of PHC [7], [6] .They stated that mucosal abnormalities in PHC is associated with non-significant inflammatory infiltrates that may be confused with colitis. Thus, a standardized grading system to classify the endoscopic appearance and severity of PHC should be adopted.

It should be put into consideration that there are limitations in this study as it included 40 patients only and a small percentage of patients with history of endoscopic interventions or lower GI bleeding. We think that a larger study is needed so the results can be generalized.