Association of Helicobacter species with hepatitis C cirrhosis with or without hepatocellular carcinoma

Abstract

Background and aims: Recent studies have suggested that bacterial coinfection with Helicobacter species in patients already infected with hepatitis C virus (HCV) could be involved in the development of cirrhosis and hepatocellular carcinoma (HCC). A retrospective cross sectional study was performed in order to explore the association between Helicobacter species and HCV associated liver diseases.

Methods: The presence of Helicobacter species was tested by polymerase chain reaction on liver samples from four groups of patients.

Results: Helicobacter 16S rDNA was found in only 4.2% of liver samples from control patients (n = 24) and in 3.5% of liver samples from patients with non-cirrhotic chronic hepatitis C (n = 29) while it was found in 68.0% of liver samples from patients with HCV positive cirrhosis without HCC (n = 25) as well as in 61.3% of cirrhotic liver samples from patients with HCV positive cirrhosis and HCC (n = 31). In addition, when the HCC tumour tissue was tested (n = 21), 90.5% of samples were positive. DNA from Helicobacter pylori- and Helicobacter pullorum-like organisms was found.

Conclusions: There is an association between the presence of Helicobacter species DNA in the liver and hepatitis C cirrhosis, with or without HCC. Indeed, the presence of these bacteria could be the result of structural changes in the liver. Alternatively, Helicobacter species could be a co-risk factor in HCV chronic liver diseases. This result warrants prospective studies to determine the possible causal role of these bacteria in the progression of chronic hepatitis C.

There are extensive data indicating that chronic infection can lead to cancer in various organs. Parasites, such as schistosomes and liver flukes, and a bacterium, Helicobacter pylori, were classified as type I carcinogens by the International Agency for Research on Cancer in 1994.¹ The most compelling evidence comes from hepatocellular carcinomas (HCC) where chronic hepatitis B virus (HBV) infection is considered to be a direct aetiological factor and hepatitis C virus (HCV) infection a major risk factor for this disease.¹

Most HCC in developed countries are linked to chronic HCV infection, alcoholic cirrhosis, and sometimes HBV infection.² The liver fluke, Opisthorchis viverrini, is also a risk factor for the development of cholangiocarcinoma.³ While a small proportion of the general population is infected by HCV, chronic hepatitis occurs in up to 75% of cases; 20% will develop cirrhosis, and in 3–5% of patients, liver cirrhosis will evolve into HCC each year.⁴ The reasons for this evolution are not well understood and may involve co-risk factors due to HCV per se, such as host genetic factors and environmental factors. Current risk factors cannot fully explain this evolution.⁵

A bacterium, Helicobacter hepaticus, from the mouse gut flora colonises the bile canaliculi and liver in mice, induces chronic hepatitis, and its persistence leads to the development of a carcinoma.^6,7 In this model, as well as in the human stomach infected by H pylori, chronic inflammation induced by the pathogen is considered to be a key factor in carcinoma development.

In the past few years, the emergence of novel and diverse Helicobacter species associated with the pathogenesis of human enterohepatic diseases has been observed.^8–11Helicobacter species can also be present in the liver of HCV negative patients and have been associated with HCC development in the non-cirrhotic liver.^12,13

Concerning HCV liver diseases, H pylori and H pullorum DNA were detected in the liver tissue of a large majority of HCC patients, suggesting that these bacteria are implicated in the progression of chronic hepatitis C to cirrhosis and to HCC.¹⁴ Recently, Dore et al confirmed these results¹⁵ while Fan et al found Helicobacter species in HBV positive HCC.¹⁶

Based on these data, our aim was to develop new tools to study the prevalence of Helicobacter species in the liver of a large series of patients with HCV infection in order to evaluate the association between the presence of these bacteria in the liver and stage of disease.

MATERIALS AND METHODS

Patients and tissue sample analysis

A retrospective study was carried out on liver specimens collected from 109 patients (Bordeaux University Hospital) divided into four different groups according to liver pathology (table 1 ▶). Tissues not used for diagnostic purposes were immediately snap frozen in liquid nitrogen and kept frozen at −80°C until use, according to the safety and ethics rules followed by our university hospital.

Table 1.

 Pathological features in the four patient groups

	Group I	Group II	Group III	Group IV
No	24	29	25	31
Mean age (y)	53	40.1	53.8	57.8
Sex (M/F)	15/9	19/10	14/11	25/6
METAVIR score
Fibrosis
F0–F1	24	15	0	0
F2	0	12	0	0
F3	0	2	0	0
F4	0	0	25	31
Activity
A0	23	0	0	0
A1	1	12	7	6
A2	0	15	12	19
A3	0	2	6	5
Dysplasia
Absent	24	29	7	2
Present	0	0	16	28

Biopsy samples from control patients (group I) were taken from the non-tumoral part of hepatectomy specimens after resection for hepatic benign tumours (n = 12; that is, seven focal nodular hyperplasias, two cavernous haemangiomas, one liver cell adenoma, one angiomyolipoma, and one solitary fibrous tumour) or metastatic tumours (n = 12; that is, nine colorectal, one mammary, one oesophageal, and one ovarian).

Patients in group II presented with chronic active hepatitis C without cirrhosis. Three patients also had a moderately excessive alcohol intake (>20 g of alcohol/day for women and 40 g for men) and in one case serological sequelae of hepatitis B infection.

Group III consisted of samples from patients who underwent liver transplantation for terminal stage cirrhosis due to hepatitis C. Two patients also had a past history of excessive alcohol intake and four had sequelae of hepatitis B infection (6/25).

Patients in group IV presented with cirrhotic hepatitis C and hepatocellular carcinoma. Most samples from this last group were obtained after total (n = 28) or partial (n = 3) hepatectomy. In addition to HCV infection, there was also a medical history of excessive alcohol intake in six cases, past hepatitis B in two cases, and both risk factors were present in one case (9/31). In group IV, tumour material was only available in 21 cases.

The diagnosis of HCV infection was based on a third generation test (Ortho HCV 3.0 ELISA, Monolisa anti-HCV; Sanofi Diagnostics Pasteur Inc., France) and a positive HCV serum RNA (Cobas amplicor HCV 2.0; Roche Diagnostics, Branchburg, New Jersey, USA).

Pathological study

For all patients, conventional histology was performed on formalin fixed liver tissues. Sections were stained with haematein-eosin-safran, Masson’s trichromic stain, and reticulin stain.

HCV chronic hepatitis was scored using the METAVIR system.¹⁷ The diagnosis of HCC and other hepatic tumours was based on accepted criteria.¹⁸

Control strains used to determine polymerase chain reaction specificity

A range of bacterial strains were used to test the specificity of different primers used in this study. These strains included a large panel of Campylobacter strains and other enteric bacteria commonly isolated from patients, as well as the following Helicobacter species: H pylori (ATCC 700392 and 700824), H felis (CCUG 28539 T), H bilis (ATCC 51630), H hepaticus (ATCC 51448), H muridarum (ATCC 49282), H pullorum (CCUG 33842 and CCUG 33839), and F rappini (CCUG 29176).

DNA extraction

DNA from frozen liver material (20–25 mg/specimen) was extracted using the QIAamp Kit (Qiagen Inc., Chatsworth, California, USA) according to the manufacturer’s recommendations, with minor modifications.¹⁹ DNA was stored at −20°C.

PCR conditions

Polymerase chain reaction (PCR) amplification was carried out in a final volume of 10 μl containing 1× buffer, 1.5 mM MgCl₂, 100 μM deoxynucleoside triphosphates, 0.5 μM each of the four primers, 0.4 U of Taq polymerase (Eurobio, Les Ulis, France), and 10 ng of DNA in a Perkin Elmer Cetus 9600 thermocycler under the conditions listed in table 2 ▶. PCR products were analysed on a 1–4% agarose gel or on a 12% polyacrylamide gel depending on amplicon size, and stained with ethidium bromide.

Table 2.

 Primers and polymerase chain reaction (PCR) conditions used for amplification of the different DNA targets

Name	Sequence	Organism	Gene	Cycling conditions	Amplicon size (bp)	Reference
HS-1	AACGATGAAGCTTCTAGCTTGCTAG	Helicobacter genus	16S rRNA	(94°C; 30 sec, 60°C; 1 min 72°C; 30 sec)×40	399	Germani21
HS-2	GTGCTTATTCGTTAGATACCGTCAT	Campylobacter genus				Avenaud12
C97	GCTATGACGGGTATCC	Helicobacter genus	16S rRNA	(94°C; 30 sec, 54°C; 90 sec 72°C; 60 sec)×35	398	Fox20
C98	GATTTTACCCCTACACCA
F2-16S-CHPEC	ATCCTGGCTCAGAGTGAACG	Pseudo-universal	16S rRNA	(94°C; 30 sec, 60°C; 30 sec, 72°C; 80 sec)×40	1456	This study
R4-16S-CHPEC	CCTACGGTTACCTTGTTACGAC
C96	ACGGGTGAGTAACGCATAGG	Helicobacter genus	16S rRNA	(94°C; 30 sec, 60°C; 30 sec, 72°C; 80 sec)×40	1374	This study
R4-16S-CHPEC	CCTACGGTTACCTTGTTACGAC
F1-mdh-coli	TTGTCGAATGTGCCTACGTT	Escherichia coli*	mdh	(94°C; 1 min, 60°C; 1 min, 72°C; 1 min)×35	130	This study
R1-mdh-coli	ATACCTTCCAGCGCGTTCT
No name	ATGAATGCTAGTTGTTGTGAG	Helicobacter pullorum	16S rRNA	(94°C; 30 sec, 60°C; 1 min, 72°C; 80 sec)×40	467	Stanley24
No name	GATTGGCTCCACTTCACA	Helicobacter canadensis
HPY S	AGGTTAAGAGGATGCGTCAGTC	Helicobacter pylori*	23S rRNA	(94°C; 1 min, 55°C; 1 min, 72°C; 1 min)×40	267	Ménard33
HPY A	CGCATGATATTCCCATTAGCAGT
F1-glmM-HP	AACGCCCTTTCTTCTCAAGC	Helicobacter pylori*	glmM	(94°C; 30 sec, 60°C; 30 sec, 72°C; 10 sec)×40	130	This study
R1-glmM-HP	GCATGCATTCGCTCACAA
F1-cdtB-pullorum	GTCTTTTGAGTGGATTGGATTCT	Helicobacter pullorum*	cdtB	(94°C; 30 sec, 60°C; 1 min, 72°C; 20 sec)×40	140	This study
F2-cdtB-pullorum	GCCAGCAAGTCTTTTGAGTG				148
R2-cdtB-pullorum	CACTCCGGGTGCTTGTGTAT
F4-cdtB	TCAGCTAGCACAGAAAATAAATGG	Non-specific Helicobacter pullorum	cdtB	(94°C; 30 sec, 55°C; 60 sec, 72°C; 30 sec)×40	719	This study
R3-cdtB	ACTAGCTGCTGCCAAAATCG
F2-cdtB-bilis	CGAATCTATTATCCGGGCTTG	Helicobacter bilis*	cdtB	(94°C; 30 sec, 60°C; 1 min, 72°C; 30 sec)×40	151	This study
R2-cdtB-bilis	GCCAAGCGAGTTCTATCATTAG
F1-ureB-rappini	GATGATTAGGGCGACACAGC	Helicobacter rappini*	ureB	(94°C; 30 sec, 60°C; 30 sec, 72°C; 10 sec)×40	101	This study
R2-ureB-rappini	CCCCAGATTCTATCTGCTTACTC

Cycling conditions: after four minutes of initial denaturation at 94°C, each reaction mixture was amplified for 35–40 cycles under the conditions indicated, depending on the PCR. After the last cycle, a final extension was continued for another seven minutes.

F, forward; R, reverse.

Partial Genbank available cdtB sequences of H pullorum (AF220065), H bilis (AF243077), and H hepaticus (AF163667 and AF243076) were aligned in order to design the most possible specific H pullorum and H bilis PCR on the cdtB gene. Forward F1- and F2-cdtB-pullorum were designed to be used with the R2-cdtB-pullorum reverse primer.

*All of these PCR were species specific, with control strains listed in materials and methods. F4/R3-cdtB primers were used to determine the nucleotide sequence of H pullorum cdtB gene.

Escherichia coli PCR amplification

Escherichia coli malate dehydrogenase (mdh) gene was selected to develop an E coli species specific PCR. The design of the primer was based on the alignment of 205 mdh sequences from E coli as well as from closely related species such as Salmonella typhimurium and Salmonella enteritica (table 2 ▶). This PCR proved to be specific for E coli.

Helicobacter genus and species specific PCR amplification

Helicobacter genus specific primer pairs C97/C98²⁰ and HS1/HS2^12,21 were used to generate 16S rDNA amplicons of approximately 400 bp (table 2 ▶).

Samples generating a positive result with the Helicobacter genus specific PCR were subsequently analysed with six different sets of primers for detection of four of the species previously found in human liver—that is, H bilis, H pullorum, H pylori, and F rappini (table 2 ▶).

Purification and cloning of PCR products for 16S rDNA and cdtB gene sequencing

Different H pylori strains have been found in human gastric biopsies,²² and therefore the likelihood of mixed strains in hepatic biopsies cannot be excluded. Moreover, the presence of several copies of 16S and 23S rRNA genes in H pylori also indicates the possibility of heterozygosity.²³ For these reasons, PCR products were cloned prior to sequencing. Helicobacter species 16S rDNA sequences were amplified with F2/R4-16S-CHPEC or C96/R4-16S-CHPEC primers, H pylori 23S rDNA sequences were amplified with HPY S/A primers, and H pullorum cdtB sequences were amplified with F4/R3-cdtB primers (table 2 ▶). PCR products were cloned into the pGEM-T easy vector System I (Promega, Madisson, Wisconsin, USA), according to the manufacturer’s recommendations. Plasmids containing the expected amplicon were purified using the Qiagen Miniprep Kit (Qiagen) and sequencing was achieved on both strands with pGEM-T specific primers (F1-pGEMT: 5′-CGG CCA GTG AAT TGT AAT ACG-3′ and R1-pGEMT: 5′-ATG ACC ATG ATT ACG CCA AG-3′). For Helicobacter species 16S rDNA amplicons, internal sequences were determined with pseudo-universal eubacterial primers designed for this study (F1-16S-HCES: 5′-ACA CGG TCC AGA CTC CTA CG-3′; F4-16S-CHPEC: 5′-CAA GCG GTG GAG CAT GTG-3′; R1-16S-HCES: 5′-CTT GCA CCC TCC GTA TTA CC-3′; and R2-16S-CHEPC: 5′-AAG GGC CAT GAT GAC TTG AC-3′).

RESULTS

Pathological findings (table 1 ▶)

Control tissues from group I were considered strictly normal or subnormal (mild steatosis, mild non-specific lobular or portal inflammatory infiltrate); only one showed obstructive cholestasis which was considered a consequence of tumour compression. None of the cases showed chronic hepatitis, liver cell dysplasia, or fibrosis. In group II, liver biopsies showed typical features of chronic hepatitis C, with no fibrosis or portal fibrosis, with or without septa. In group III, patients presented with chronic hepatitis C of varying activity. Liver dysplasia, of the large or small cell type, was found in 16 cases (particularly abundant in four). Macronodules, of the macroregenerative or dysplastic type, were found in four liver samples. In group IV, hepatitis C activity, as well as dysplasia, was present in most livers and was either limited or extensive. All 31 HCC were of the classical type: unifocal in eight cases, bifocal in six, and multifocal in 17. The size of the largest tumoral nodule measured 14 cm. Macroregenerative and dysplastic nodules were frequently associated (n = 12).

In the 21 HCC cases where the tumour could be analysed for Helicobacter detection (non-necrotised HCC with tumour size>15 mm), well differentiated (grade 2, n = 8) or moderately differentiated HCC (grade 3, n = 13) with a classical trabecular (n = 11), trabeculoglandular (n = 8), or macrotrabecular and compact (n = 2) architecture was observed.

Escherichia coli and Helicobacter genus detection (fig 1 ▶)

Figure 1.

 Percentage of positive biopsies for Helicobacter genus by DNA detection. Groups I–IV correspond to control patients (n = 24), patients with chronic hepatitis C (n = 29), patients with post hepatitis C cirrhosis (n = 25), and patients with cirrhosis and HCC (n = 31), respectively. Concerning group IV, biopsies were obtained in cirrhotic tissue for all (Ci) and in tumoral (Tu) tissue for 21 patients. Helicobacter genus positivity and negativity was compared using Fisher’s exact test. *Cases positive in cirrhotic liver tissue were also positive in corresponding tumoral tissue.

Among the 130 specimens from 109 patients, none was positive for E coli. Helicobacter genus DNA was detected in 38 cases: 36 using HS1/HS2 primers and the same 36 plus two additional cases using C97/C98 primers. These positive cases were found mainly in group III (17 cases, 68%) and group IV (19/31 (61.3%) cases in cirrhotic liver; 19/21 (90.5%) cases in the tumour). In contrast, only one of 24 (4.2%) and one of 29 (3.5%) samples were positive in groups I and II, respectively, and the difference between groups I and II compared with groups III and IV was statistically significant (p<10⁻⁴). Moreover, all 19 patients in group IV who were positive at the cirrhotic liver tissue level were also positive in the corresponding tumoral tissue.

As some of the patients, mainly from groups III and IV, had other aetiological factors, we examined if the presence of Helicobacter was associated with these factors, and found that it was evenly distributed.

Helicobacter species identification

Material from 38 Helicobacter genus positive patients was then tested with species specific primers (table 3 ▶). None reacted with primers for cdtB of H bilis or for ureB of F rappini by PCR. Three patients were positive using specific primers for 16S rDNA of H pullorum.²⁴ However, this PCR did not react with our two H pullorum control strains which can amplify Helicobacter canadensis, another enterohepatic Helicobacter.²⁵ For these reasons a new PCR, designed on the cdtB gene, was developed to detect H pullorum strains which possibly remained undetected using the 16S rDNA PCR. F1- and R2-cdtB-pullorum specific primers were used; only the same three samples were positive, in addition to the control strains. The 35 remaining Helicobacter genus positive patients were positive for H pylori 23S rDNA and 22 were also positive for H pylori glmM (61.1%). No H pullorum DNA sequence was detected in liver specimens from these 35 patients and no H pylori DNA sequence was detected in liver specimens from the three patients positive for H pullorum DNA sequences.

Table 3.

 Identification at the species level of Helicobacter DNA present in liver material

Histological group (n)	Helicobacter genus	Helicobacter pylori		Helicobacter pullorum
		23S rDNA	glmM	16S rDNA	cdtB
I-Control (24)	1	1	1*	0	0
II-Hepatitis C virus (29)	1	1	1*	0	0
III-Cirrhosis C without HCC (25)	17	16	12*	1	1§
IV-HCC and cirrhosis C
Cirrhotic tissue (31)	17†	16†	8*	1†	1§
Tumoral tissue (21)	19	17	10*	2	2§

HCC, hepatocellular carcinoma

*These cases were all positive using the 23S rDNA primers.

†Cases positive in cirrhotic liver tissue were also positive in the corresponding tumoral tissue.

§These cases were all positive using the 16S rDNA primers.

Sequencing was performed on 16S rDNA amplicons of 1456 bp from patients C5 and C45 (positive for H pullorum PCR) and of 1374 bp from patient C51 (positive for H pylori PCR) (table 4 ▶). The 16S rDNA sequences from biopsies C5 and C45 were 99% similar to the 16S rDNA of H pullorum for the entire length of the available sequence in GenBank. Moreover, these two sequences were different from each other as they had 10 mismatches. The 16S rDNA sequence from the C51 biopsy had 99% sequence homology with H pylori 16S rDNA. Another Blast program search was performed on a shorter fragment (398 bp) of C51 corresponding to the C97/98 amplicon usually used for such experiments. Interestingly, higher sequence homologies (99%) were found with 59 similar sequences of H pylori 16S rDNA, including previously reported sequences amplified from liver specimens from patients with HCC—that is, Helicobacter species ”liver 3”, Helicobacter species ”liver 2”, and Helicobacter species ”liver 1”.¹²

Table 4.

 Results of DNA sequencing on a subgroup of cases found positive for Helicobacter pylori or Helicobacter pullorum by polymerase chain reaction (PCR)

Biopsy No	Group	PCR				Sequence alignment results (Strain—Genbank accession number)						Conclusion
		H pylori		H pullorum
		23S rDNA	glmM	16S rDNA	cdtB	16S rDNA (universal)		H pylori 23S rDNA		H pullorum cdtB
C5	III	–	–	+	+	UB 3659	L36147	/	NA	35818*	AF220065	“H pullorum”-like
C45	IV	–	–	+	+	UB 151	L36144	/	NA	35818*	AF220065	“H pullorum”-like
C51	IV	+	+	–	–	MC 238	U01329	210§	AF550406	/	NA	“H pylori”-like

NA, not amplified by the corresponding PCR: HPYS/A and cdtB for H pylori and H pullorum detection, respectively.

*H pullorum taxon: 35818. §H pylori taxon: 210.

For 16S rDNA, 23 rDNA, and cdtB sequences, amplified primer less sequences of 1414 or 1332, 222, and 675 bp were blasted and alignments were done using GenBank database.³⁴ The 16S rDNA sequences for biopsies C5, C45, and C51 were submitted to GenBank and assigned accession numbers AY394474, AY394473 and AY394476, respectively. The cdtB sequences and the deduced protein sequences for biopsies C5 and C45 were assigned accession numbers AY394475 and AY394477, respectively. The 23S rRNA sequence for biopsies C51 was assigned accession number AY394478.

Four other 16S rDNA C97/C98 amplicons of the 35 samples positive for H pylori PCR were also sequenced for both DNA strands. Sequences obtained were shown to be similar to H pylori 16S rRNA gene (data not shown).

Furthermore, the species specific amplicons of biopsies C5, C45, and C51 were also sequenced. The 675 bp cdtB sequences from C5 and C45 biopsies were similar and were shown to have 98% homology with the only two cdtB gene sequences of H pullorum available in GenBank (table 4 ▶), with several residue changes in the corresponding nucleotide and protein sequences. For the 23S rDNA sequence from biopsy C51, the higher sequence homologies (100%) were found with 17 similar sequences of H pylori 23S rDNA.

DISCUSSION

The discovery of the presence of Helicobacter species DNA in liver material from patients with liver disease has led to the challenging hypothesis that these bacteria may play a role in the evolution of hepatic lesions from chronic viral hepatitis to cirrhosis and HCC. Determinants of this evolution are not yet fully understood, including those occurring in HCV positive patients.

A similar evolution of stomach lesions from gastritis to atrophy, intestinal metaplasia, dysplasia, and adenocarcinoma has been linked to infection by H pylori.²⁶ In an attempt to detect an association between Helicobacter species infection of the liver and liver pathology, we studied HCV positive patients. Our results showed an association: virtually all patients with HCC were Helicobacter species positive in their HCC tumour material and 61–68% of those with cirrhosis were positive in liver tissue compared with 4.5% and 3.2% of hepatitis patients and controls, respectively. The results were also highly consistent. For Helicobacter genus detection by PCR, tumour material was always positive when cirrhotic tissue was positive. The difference in age between patients with cirrhosis C and those with cirrhotic hepatitis C with HCC could explain the higher prevalence of Helicobacter in the group of patients with cirrhosis C. However, the control group, which was similar in age to this group, did not have a higher prevalence, indicating that in contrast with Helicobacter in the stomach, the prevalence in the liver does not increase with age.

These data are in agreement with several other studies which showed the presence of Helicobacter species in the liver of patients with and without non-cirrhotic HCC.^12–16 Only two of these reports examined HCV positive patients. In the study of Ponzetto et al, 23 of 25 HCC HCV positive frozen liver specimens contained Helicobacter species DNA and sequencing of the 16S rDNA amplicons also revealed the presence of H pylori- and H pullorum-like organisms.¹⁴ More recently, a similar study was performed on formalin fixed paraffin embedded liver tissues.¹⁵ Despite the fact that DNA is not well conserved in such tissue, tests showed that six of 11 patients with HCV positive HCC contained Helicobacter species DNA in the liver, and three corresponded to H pylori as they contained vacA sequences.

Interestingly, and as previously reported in liver specimens from patients with HCC,^13,16 the highest homology of our 398 bp 16S rDNA sequences was obtained with H pylori sequences previously reported in liver specimens from patients with non-cirrhotic HCC.¹²

This study is the first in which both tumour and cirrhotic liver tissue samples from patients with HCV positive HCC were tested. When comparing the results of cirrhotic liver tissue from patients with only cirrhosis (61.3%) with those with both cirrhosis and HCC (68%), a similar proportion of Helicobacter species positive specimens was obtained from patients with HCC.

However, at least one study did not find any Helicobacter species, raising the possibility that the results of other studies may be false positives.²⁷ However, there are a number of arguments against this interpretation. Firstly, we did not find any specimens positive for E coli. This bacterium, which is one of the gut’s main inhabitants, would most likely be present in the liver as a contaminant, either via the portal circulation or by retrograde transfer from the duodenum. It is indeed an argument against the possibility of extrahepatic DNA sequestered in the liver via the portal circulation because in such cases E coli DNA would also be found. Secondly, we are confident that intra-laboratory contaminants cannot explain these results as we did not use nested PCR and, moreover, there was variability in the PCR products, as proven by sequencing. Finally, Helicobacter species cannot be considered as environmental contaminants. Therefore, we came to the same conclusion as Wadström et al²⁸ that false negative results may have occurred in the study in which no Helicobacter species DNA were found in the liver.²⁷ Indeed, the PCR positive control was not appropriate (gastric tissue) and not enough technical information on the PCR was given in the paper to determine whether there might have been another problem.

It may be argued that if a Helicobacter species is present in the liver, one should be able to culture it. Indeed, one case of positive culture was reported in a patient suffering from Wilson’s disease.¹⁰ The retrospective nature of our study, based on frozen material, made culture attempts difficult. In any case, it is most likely that these bacteria are in low quantities and in a special physiological state, rendering them difficult to culture under standard conditions. Even in the case of mouse liver disease due to H hepaticus, it has proved very difficult to culture the bacteria.²⁹ Moreover, there are Helicobacter species such as Helicobacter heilmannii which, despite their presence in high numbers (for example, in the pig stomach) do not grow on currently available media but only under very special conditions such as in the mouse stomach.³⁰

Among the different Helicobacter species previously found in hepatobiliary diseases²⁰ and tested by specific PCR in this study, neither H bilis nor H rappini was found but H pullorum-like DNA was detected in the liver of three patients. DNA from H pylori-like organisms was found in 35 of 38 Helicobacter genus positive patients but only 61.1% of these 35 cases were amplified by glmM PCR, suggesting that in the remaining cases either this gene is absent or nucleotide variations occur in this region impairing its detection by PCR. Another possibility is that an unknown Helicobacter species close to H pylori is present.

A limitation of our study is obviously its retrospective nature, which did not enable us (1) to determine whether H pylori was present in the stomach of these patients and (2) to gather all of the clinical and biological information needed for a more thorough analysis (for example, virus genotype, duration of viral infection). However, we believe that this study, which included a large series of HCC patients with specimens obtained both in tumour and cirrhotic tissue and for which a number of new tools were developed, offers a strong argument in favour of an association between Helicobacter species and both HCV cirrhosis and hepatitis C cirrhosis with HCC.

In conclusion, based on this study we have no argument for a causal association and it may well be that the presence of Helicobacter is the consequence of structural changes in the liver (namely, intrahepatic shunts) when cirrhosis occurs.^31,32 However, the possibility that Helicobacter infection is a risk factor for the evolution towards cirrhosis remains and deserves further exploration, given that these bacteria can produce toxins which may interfere with hepatic cells. A prospective study, including bacterial culture from material obtained from the stomach in addition to liver tissue, is warranted to determine the H pylori status of the stomach in these patients.

Abbreviations

• HCC, hepatocellular carcinoma

• HCV, hepatitis C virus

• HBV, hepatitis B virus

• PCR, polymerase chain reaction