1848
Letters to the Editor
Response to Collin et al. Jennifer L. Vande Voort, MD1 and Joseph A. Murray, MD1 doi:10.1038/ajg.2009.217; published online 19 May 2009
To the Editor: We are pleased with the interest that Drs Collin, Kaukinen, and Maki have in our article as shown in their thoughtful and helpful commentary (1). We completely concur that lymphocytic duodenosis may presage full-blown symptomatic celiac disease and that there is potential usefulness of serological tests for detecting those patients in whom the pathological findings represent mild gluten-sensitive enteropathy. The power of the human leukocyte antigen type is in its negative predictive value and, hence, it is a simple, logical way of eliminating celiac disease as a cause in a substantial proportion of patients. Unfortunately, serological tests, including endomysial antibodies, are not particularly sensitive for celiac disease with mild enteropathy. The development of newer techniques, including immunoglobulin A deposition in the mucosa, may provide clinicians and pathologists with a tool to identify again those individuals in whom celiac disease is a cause of the problem (2). CONFLICT OF INTEREST
The authors declare no conflict of interest. REFERENCES 1. Collin P, Kaukinen K, Mäki M. Duodenal intraepithelial lymphocytosis: Celiac disease or not? Am J Gastroenterol; published online 19 May 2009 [e-pub ahead of print]. 2. Salmi TT, Collin P, Jarvinen O et al. Immunoglobulin A autoantibodies against transglutaminase 2 in the small intestinal mucosa predict forthcoming celiac disease. Aliment Pharmacol Ther 2006;24: 541–52.
1
Mayo Clinic, Rochester, Minnesota, USA. Correspondence: Joseph A. Murray, MD, Mayo Clinic, Gastroenterology W19A, 200 First Street SW, Rochester, Minnesota 55905, USA. E-mail:
[email protected]
The American Journal of GASTROENTEROLOGY
The Genetics of Alcoholic Liver Disease: Better Patient Group Definition Is Required Eric Nguyen-Khac, MD, PhD1,2, Hakim Houchi, MSc1, Martine Daoust, PhD1 and Mickaël Naassïla, PhD1 doi:10.1038/ajg.2009.255
To the Editor: We were most interested to read the results of the study by Gleeson et al. (1) on the involvement of proinflammatory cytokine gene polymorphisms in alcoholic liver disease. Gleeson et al. (1) report the allelic frequencies of single nucleoside polymorphisms (SNPs) of interleukin-6 (IL6 − 174) and interleukin-10 (IL10 − 592) in Child–Pugh class C alcoholic cirrhotics. We would like to raise a number of issues. Firstly, do the statistically significant differences reported in the study persist after Bonferroni correction (which is usually strongly recommended when so many statistical tests are used)? Furthermore, the choice of a control group is always a delicate matter: Gleeson et al. (1) selected a group of alcoholic liver disease–free heavy drinkers with a PGA (TP, apolipoprotein A1, gamma–glutamyl transferase) score ≤6, which enables cirrhosis to be ruled out in 90% of cases (2). However, this approach does not exclude hepatic fibrosis (at varying stages of progression), steatosis or alcoholic hepatitis (AH), and thus does not eliminate patients who are very similar to cirrhotic subjects in physiopathological terms. Today, a number of tests (such as the Fibroscan (Echosens, Paris, France) based on non-invasive hepatic ultrasound elastography) facilitate screening for alcoholic fibrosis without requiring a liver biopsy. In a population of asymptomatic heavy drinkers (such as that studied by Gleeson et al. (1)), the Fibroscan is capable of diagnosing 32% of previously unsuspected cirrhoses and 19% of cases of severe fibrosis and have better diagnostic performance
levels than the PGA score (3). We believe that lack of this type of screening introduces significant bias into the study by Gleeson et al. (1). Lastly, Gleeson et al. (1) report that they did not find any difference for the tumor necrosis factor-alpha (TNFα)308 A/G SNP. We have recently showed that significantly lower expression of the TNF-308 A allele in patients with severe, acute AH defined according to a Maddrey’s discriminant function of ≥32 and compatible liver histology results (4). Low frequencies of this polymorphism have only been found in studies with significant mortality (4,5), which could correspond to a favorable genetic background for more severe expression of alcoholic liver disease. This was not confirmed in the study by Gleeson et al. (1), as only 84 out of 223 patients (38%) had undergone a liver biopsy (51 (61%) of whom had AH). Furthermore, it was not clear whether or not the AH was severe because the discriminant function for this group was not specified. Better definition of the study population is thus required for optimal interpretation of the results, because proinflammatory cytokines are not necessarily involved in acute AH–free subjects with Child C cirrhosis to the same extent as in those with acute AH. CONFLICT OF INTEREST
Authors declare no conflict of interest. REFERENCES 1. Gleeson D, Bradley MP, Jones J et al. Cytokine gene polymorphisms in heavy drinkers with and without decompensated liver disease: a case–control study. Am J Gastroenterol 2008;103:3039–46. 2. Poynard T, Aubert A, Bedossa P et al. A simple biological index for detection of alcoholic liver disease in drinkers. Gastroenterology 1991;100:1397–402. 3. Nguyen-Khac E, Chatelain D, Tramier B et al. Assessment of asymptomatic liver fibrosis in alcoholic patients using fibroscan: prospective comparison with seven non-invasive laboratory tests. Aliment Pharmacol Ther 2008;28: 1188–98. 4. Nguyen-Khac E, Houchi H, Daoust M et al. The -308 TNFalpha gene polymorphism in severe acute alcoholic hepatitis: identification of a new susceptibility marker. Alcohol Clin Exp Res 2008;32:822–8. 5. Tsuchiya N, Tokushige K, Yamaguchi N et al. Influence of TNF gene polymorphism in
VOLUME 104 | JULY 2009 www.amjgastro.com
Letters to the Editor
patients with acute and fulminant hepatitis. J Gastroenterol 2004;39:859–66.
1 INSERM ERI 24, Groupe de Recherche sur I’Alcool et les Pharmacodépendances (GRAP), Jules Verne University of Picardy, rue des Louvels, Amiens, France; 2Hépatogastroenterology Department, Amiens University Medical Center, Place Victor Pauchet, Amiens cedex 1, France. Correspondence: Eric Nguyen-Khac, MD, PhD, HepatoGastroenterology Department, Amiens University Medical Center, Amiens University Hospital, Place Victor Pauchet, F-80054 Amiens cedex 1, France. E-mail:
[email protected]
Response to Nguyen-Khac et al. Dermot Gleeson, MD, BSc, FRCP1 doi:10.1038/ajg.2009.260
To the Editor: We thank Nguyen-Khac and colleagues for their letter (1) and for their interest in our paper (2) and we reply as follows. As discussed in the penultimate paragraph of the Discussion, we did not apply the Bonferroni correction and justify not so doing on the basis that plausible a priori hypotheses can be constructed for individual polymorphisms. If the Bonferroni correction is applied, none of the case–control differences remain significant. This question has arisen only because we choose to report data on all polymorphisms in a single paper, rather than separately. Our case–control classification is based on liver dysfunction, rather than histology. However, we present evidence that > 70% of the cases had cirrhosis, whereas the prevalence of cirrhosis in the controls is likely to be < 10%. Our aim was to define two cohorts of heavy drinkers with markedly different susceptibilities to alcoholic liver disease (ALD). We do not think that this strategy has lead to a significant bias. We accept (end of the first paragraph of Discussion) that our screening of the heavy drinking controls by liver chemistry, ultrasound, and PGA © 2009 by the American College of Gastroenterology
index < 6 (not ≤6), will exclude cirrhosis with at least 90% certainty (3), but will not exclude fibrosis short of cirrhosis. However, the clinical significance of lesser degrees of fibrosis is not established in ALD. On the basis of our reported experience (4), 85% of patients presenting with decompensated ALD for the first time have alcoholic hepatitis on biopsies performed within a month of presentation. We believe that it is this alcoholic hepatitis, rather than the fibrosis, which is the cause of the decompensation. Histological features of ALD change with time, and we do not believe that the rigid classification of patients on the basis of the appearances of (usually) one liver biopsy is valuable. We note that in the Nguyen-Khac study the Fibroscan was superior to the PGA index in distinguishing between milder degrees of fibrosis but not between cirrhosis and milder degrees of fibrosis (5). The Fibroscan was not available when we commenced the study in 1998, and we look forward to other genetic studies classifying patients with ALD using this technique. In patients with severe alcoholic hepatitis, Nguyen-Khac et al. found lower tumor necrosis factor (TNF)-308 A allele carriage (6). In contrast, we found a (non-significant) trend toward a higher A allele carriage rate for this polymorphism in decompensated ALD, compared with that in heavy drinking controls. As argued above, we believe that most of these patients had alcoholic hepatitis when they first presented with decompensation. We have, as suggested, calculated TNF-308 genotype distribution in the 81 patients in our study who had a Maddrey’s discriminant function of ≥32 when they presented, and it is essentially identical to that in our decompensated ALD patient group as a whole (31% A allele carriage rate). We cannot explain the (as far as we are aware, unique) finding by Nguyen-Khac et al. of a lower A allele frequency in patients with alcoholic hepatitis compared with controls. Given that the A allele is associated with increased transcription (7), their
observation is difficult to reconcile with the putative role of TNF in alcoholic liver injury. Could it be relevant that the number of controls they studied was modest? As we discuss (paragraph two, TNF section of Discussion), much of the published data on ALD and TNF-308 genotype distribution (and indeed, that of many candidate genes linked to ALD) are contradictory. Further studies and meta-analyses are clearly needed. In this context, we agree with Nguyen-Khac’s basic point, which is that better definitions of ALD patient and control categories that are agreed on by researchers in this area, would help to clarify many areas of confusion. CONFLICT OF INTEREST
The authors declare no conflict of interest. REFERENCES 1. Nguyen-Khac E, Hoichi H, Daoust M et al. The genetics of alcoholic liver disease: better patient group definition is required. Am J Gastroenterol 2009 (this issue). 2. Gleeson D, Bradley M, Jones J et al. Cytokine gene polymorphisms in heavy drinkers with and without liver disease: a casecontrol study. Am J Gastroenterol 2008;103:3039–46. 3. Poynard T, Aubert A, Bedossa P et al. A simple biological index for detection of alcoholic liver disease in drinkers. Gastroenterol 1991;100(5 Pt 1):1397–402. 4. Elphick DA, Dube AK, Farlane E et al. Spectrum of liver histology in presumed decompensated alcoholic liver disease. Am J Gastroenterol 2007;102:780–8. 5. Nguyen-Khac E, Chatelain D, Tramier B, et al. Assessment of asymptomic liver fibrosis in alcoholic patients using fibroscan: prospective comparison with seven non-invasive laboratory tests. Alimentary pharmacology and Therapeutics 2008;28:1188–98. 6. Nguyen-Khac E, Hoichi H, Daoust M et al. The -308 TNF alpha polymorphism in severe alcoholic hepatitis: identification of a new susceptibility marker. Alcohol Clin Exp Res 2008;32:822–8. 7. Wilson AG, Symons JA, McDowell TL, et al. Effects of a polymorphism in the human tumor necrosis factor alpha promoter on transcriptional activation. Proc Natl Acad Sci USA 1997;94:3195–9. 1 Liver Unit, Sheffield Teaching Hospitals, Royal Hallamshire Hospital, Sheffield, UK; Correspondence: Dermot Gleeson, MD BSc FRCP, Sheffield Teaching Hospitals, Liver Unit, P Floor, Royal Hallamshire Hospital, Glossop Rd, Sheffield S10 3PA, UK. E-mail:
[email protected]
The American Journal of GASTROENTEROLOGY
1849
