Log in Register

Login to your account

Password *
Remember Me

Create an account

Fields marked with an asterisk (*) are required.
Password *
Verify password *
Email *
Verify email *
Captcha *

2005–2006 Series: Update Sessions from ACP's 2005 Annual Session

Annals of Medicine


Update in Gastroenterology and Hepatology

Ronald L. Koretz, MD, and Timothy O. Lipman, MD

6 September 2005 | Volume 143 Issue 5 | Pages 347-354


We have several goals for this update, the most important one being to review the pertinent literature in gastroenterology and hepatology that appeared in 2004. Some of these papers contain important resource utilization implications that need debate. Others will provide opportunities to consider topics about critical reading of the literature.

Several of the papers included in this update are systematic reviews. Systematic reviews differ from the usual review articles we read, known as narrative reviews. The systematic review is analogous to a research project. Before anything is written, a protocol is created. A very specific question is posed, and the authors decide how they are going to use the literature to find the answer to that question. The conclusion must be driven by the data, consequently avoiding many of the biases that sneak into the narrative review.

Some systematic reviews combine data from randomized, controlled trials (RCTs) using a process called meta-analysis. The advantages of meta-analysis are increased power to see a difference and a more precise definition of the magnitude of any effect that is present. The downside of combining data from different studies is that studies often differ in ways that influence the results, a problem called heterogeneity. One must keep this in mind when reading any meta-analysis; no 2 studies are created exactly the same.

The result of a meta-analysis ("estimated effect") can be presented in several ways. The absolute risk difference (ARD) is the easiest to understand. The ARD represents the difference between the frequency of the event in the treated group and the frequency of the event in the control group.

The relative risk (RR) is another way to present the estimated effect; it is the ratio between the rate of the outcome in the treated group and the rate of the outcome in the control group. For adverse outcomes, an RR less than 1 favors the treated group. The odds ratio is the ratio of the odds of the outcome in the treated group and the odds in the control group. The odds ratio is more difficult to understand intuitively.

In meta-analysis, a significant effect exists when the 95% CI does not overlap the line of equivalence. The line of equivalence for the ARD is 0, meaning that the rate is the same in both groups. The line of equivalence for the RR and for the odds ratio is 1.0 because the numerator equals the denominator. The ARD can be easily translated into a number needed to treat for benefit (NNTB): Divide 1 by the ARD. When an intervention reduces the incidence of a complication by 5%, for example, we have to treat 20 patients (1/0.05) to prevent 1 such complication.

The NNTB cannot be directly computed from the RR. An RR reduction of 50% does not mean that for 2 patients receiving treatment, 1 will avoid a complication. To calculate the NNTB from the RR requires the baseline incidence of the complication. For example, if an adverse event occurs 1% of the time and the RR reduction from the intervention is 50% (the incidence is reduced to 0.5%), the NNTB is 200 (because the ARD is only 0.5%). If the intervention is expensive or is associated with substantial harms, a large NNTB might very well make us decide not to use it.

The critical reader will pay attention to the quality of an RCT and give more weight to high-quality trials. As one can usually glean from the methods section of the published report, different RCTs employ different degrees of methodologic rigor to avoid bias (for example, blinding the patient and the researcher to the randomly assigned treatment). High-quality studies are more rigorous and tend to show smaller treatment effects, suggesting that low-quality trials are more subject to bias.

Upper Gastrointestinal Issues

Proton-Pump Inhibitors Are Effective in Relieving the Symptoms of Nonulcer Dyspepsia

Moayyedi P, Delaney BC, Vakil N, et al. The efficacy of proton pump inhibitors in nonulcer dyspepsia: a systematic review and economic analysis. Gastroenterology. 2004;127:1329-37. [PMID: 15521002] [Medline]

Proton-pump inhibitors (PPIs) are commonly used for patients with nonulcer dyspepsia. Several RCTs have provided conflicting results. This systematic review, meta-analysis, and economic analysis was undertaken to determine the size of the beneficial effect (if one exists) and, if present, to calculate the resource allocation that would be required to undertake a policy of using PPIs in this condition. The authors systematically reviewed the literature, looking for trials that compared a PPI with placebo or with other therapy for nonulcer dyspepsia. The authors then used a Markov model to determine the cost to keep 1 patient symptom-free for a month.

Eight RCTs compared PPIs with placebo in 3293 patients. The RR of a patient still having dyspeptic symptoms after treatment with a PPI (compared with no treatment) was 0.86 (95% CI, 0.78 to 0.95); the number needed to treat to improve symptoms in 1 patient was 9. The effect was better if the dyspeptic symptom was predominantly heartburn (RR, 0.76 [CI, 0.66 to 0.88]), but the benefit was still present if the primary symptom was epigastric pain (RR, 0.85 [CI, 0.79 to 0.92]). For patients with dysmotility-like dyspepsia (such as nausea or bloating), PPIs had no effect (RR, 1.02 [CI 0.92 to 1.13]). Two trials compared histamine-2 blockers with PPIs; the calculated RR was 0.93 (CI, 0.84 to 1.02), a statistically nonsignificant trend toward an effect. The resource acceptability of the intervention depended on the cost of the PPIs. At $90 per month for a brand-name agent, the cost for keeping 1 patient symptom-free for 1 month was $278; the cost decreased to $57 when the agent cost $20 per month (the price for a generic PPI).

The authors concluded that a generic PPI is a cost-effective intervention for patients with nonulcer dyspepsia, especially those with predominant heartburn or epigastric pain. Economic analyses are most reliable when the effect of the intervention is known from RCTs because such data provide a less biased estimate. Employing observational data or data from nonrandomized trials is often done in decision analysis but is not ideal practice. Sensitivity analysis (recalculating the effect after substituting a range of values) can show if uncertainty in the effect of the intervention significantly affects the conclusions.

Eradicating Helicobacter pylori in Infected Patients with Nonulcer Dyspepsia Provided a Small Benefit

Moayyedi P, Deeks J, Talley NJ, et al. An update of the Cochrane systematic review of Helicobacter pylori eradication therapy in nonulcer dyspepsia: resolving the discrepancy between systematic reviews. Am J Gastroenterol. 2003;98:2621-6. [PMID: 14687807] [Medline]

This systematic review was undertaken to determine if eradication of Helicobacter pylori would relieve the symptoms of nonulcer dyspepsia in patients who were infected. Twelve RCTs were identified, involving 2903 patients. The RR of a patient still having dyspeptic symptoms after treatment (compared with no treatment) 1 year later was 0.91 (CI, 0.86 to 0.95) (NNTB, 15). The authors concluded that eradicating H. pylori in infected patients with nonulcerdyspepsia has a small but statistically significant effect.

Although eradication therapy for H. pylori infection did have a modest effect (as was true for PPI therapy in the preceding paper), most of the patients did not become symptom-free. One possible explanation for this outcome is that the therapy is not effective in nonulcer dyspepsia, and the small number of patients who improved may have had ulcer disease that was missed by the index endoscopy but then responded to PPIs or H. pylori eradication.

The downside of treating patients with H. pylori is antibiotic resistance. Not only will the H. pylori potentially become resistant, but all the other bacteria that happen to be living in the patient's body at the time of exposure to the antibiotic will also tend to become resistant. When an intervention only has a small benefit, one must weigh that benefit against all of the risks and costs.

Helicobacter pylori Eradication Did Not Reduce Gastric Cancer Incidence

Wong BC, Lam SK, Wong WM, et al. Helicobacter pylori eradication to prevent gastric cancer in a high-risk region of China: a randomized controlled trial. JAMA. 2004;291:187-94. [PMID: 14722144] [Abstract/Free Full Text]

Infection with H. pylori is associated with gastric cancer, leading the World Health Organization to declare H. pylori to be a Group 1 carcinogen. If H. pylori causes gastric cancer, eradicating it should theoretically prevent cancer. Because the effect of H. pylori treatment on the prevention of gastric cancer in infected individuals is unknown, these authors undertook a double-blind RCT.

The study enrolled 1630 asymptomatic participants living in Fujian Province, China, who were identified as carriers of H. pylori. The 817 participants who were randomly assigned to receive H. pylori eradication therapy received a 2-week course of omeprazole, 20 mg; a combination product of amoxicillin and clavulanate potassium, 750 mg; and metronidazole, 400 mg (all agents were taken twice daily). The 813 control patients received placebo. The primary outcome measure was the incidence of gastric cancer over the next 7.5 years.

During the follow-up, 18 new cases of cancerous gastric lesions were diagnosed: 7 in the treatment group and 11 in the control group. Although this difference was not statistically significant (P = 0.33), fewer lesions were found in the treated group. Furthermore, the Kaplan–Meier plot suggested that the lesions in the treated group occurred later in the follow-up period. In the subgroup of patients with no precancerous lesions on presentation, none were found to have gastric cancer after H. pylori eradication, compared with 6 lesions found in patients who received placebo (P = 0.02).

Because the differences were not statistically significant, the authors concluded that H. pylori eradication did not reduce the incidence of gastric cancer. However, we cannot exclude the presence of a type 2 error (namely that an inadequate number of patients were enrolled to demonstrate, with statistical signficance, a true effect). Although these data do not allow us to advocate the institution of a program of H. pylori eradication (regardless of the expert opinion of the World Health Organization), it is fair to say that additional research is warranted.


Prophylactic Antibiotics Provided No Benefit to Patients with Severe Acute Pancreatitis

Isenmann R, Runzi M, Kron M, et al. Prophylactic antibiotic treatment in patients with predicted severe acute pancreatitis: a placebo-controlled, double-blind trial. Gastroenterology. 2004;126:997-1004. [PMID: 15057739][Medline]

Several RCTs have concluded that prophylactic antibiotics benefit patients with severe pancreatitis (1, 2). As a result, physicians administer antibiotics to such patients when they are first admitted to the hospital. However, none of these studies blinded the researcher and the patient to the randomly assigned treatment. These investigators undertook a double-blind RCT of prophylactic antibiotics in patients with severe acute pancreatitis.

Severe pancreatitis was defined as either the presence of necrosis on computed tomography or a serum C-reactive protein level greater than 150 mg/L. Fifty-eight patients received intravenous ciprofloxacin (400 mg twice daily) and metronidazole (500 mg twice daily), and 56 patients received placebo antibiotics. Although the authors planned to enter 100 patients into each arm, they stopped the trial because an interim analysis found that the rate of infectionwas arithmetically higher in the treated group; 7 patients in the treatment group and 5 receiving placebos had infected pancreatic necrosis. Extrapancreatic infections occurred in 13 of the 58 patients receiving treatment compared with 14 of 56 patients in the placebo group (P = 0.585). Ten patients in the treatment group and 6 patients receiving placebo required surgery. There was 1 death in each group.

The authors concluded that prophylactic antibiotics did not reduce the incidence of infected necrosis in patients with a diagnosis of severe pancreatitis. The investigators believed that the diagnosis of infected pancreatic necrosis in previous unblinded studies was influenced by knowledge of the treatment group to which the patient had been assigned. Furthermore, patients in the treated group might have had a lower incidence of infection because specimen cultures that otherwise would have been positive were negative because of residual antibiotics. On the other hand, the combination of metronidazole and a quinolone may not be as effective as other antibiotic combinations used in previous studies. Nevertheless, this blinded RCT's failure to confirm the findings from unblinded studies has left the utility of prophylactic antibiotics in this setting more open to debate, especially given that lower-quality trials tend to show larger treatment effects. At this time, the value of giving antibiotics to patients with severe pancreatitis remains unresolved.

The Benefits of Nutritional Support in Patients with Acute Pancreatitis Were Unclear

Marik PE, Zaloga GP. Meta-analysis of parenteral nutrition versus enteral nutrition in patients with acute pancreatitis. BMJ. 2004;328:1407. [PMID: 15175229] [Abstract/Free Full Text]

Many health care professionals believe that nutritional support is an important adjunctive therapy in hospitalized patients with a variety of underlying diseases. However, many RCTs that have assessed parenteral nutrition have not only failed to show benefit but have even suggested that it does harm by increasing the rate of infections (3). In an attempt to avoid such complications, physicians have learned to pass a feeding tube into the proximal small intestine of patients with acute pancreatitis and infuse nutrient solutions. Marik and Zaloga systematically searched the literature for RCTs that compared postpyloric enteral nutrition with parenteral nutrition in patients with acute pancreatitis and combined the data in a meta-analysis. They identified 6 RCTs with 263 participants, abstracted the data, and compared the effect of these 2 types of nutrition support on infections, complications other than infections, operative interventions, length of hospital stay, and mortality.

The RR for death for patients receiving enteral versus parenteral nutrition was 0.66 (CI, 0.32 to 1.37). Enteral nutrition significantly reduced infections (RR, 0.45 [CI, 0.26 to 0.78]), the need for subsequent surgery (RR, 0.48 [CI, 0.22 to 0.99]), and the length of hospitalization (–2.9 days [CI, –1.6 to – 4.3 days]). The RR for noninfectious complications was 0.61 (CI, 0.31 to 1.22).

Marik and Zaloga concluded that enteral nutrition was the preferred route of nutrition support in patients with acute pancreatitis and went on to make some strong recommendations. The authors advised physicians to start these jejunal infusions at the time of the patient's admission to the hospital when 3 or more Ransom criteria are present. Withholding nutritional support for 48 hours in a patient with less severe disease is a reasonable strategy according to the authors, but a jejunal infusion should be started at that point if the patient is not eating. They also recommended that malnourished patients receive enteral nutrition immediately.

Of importance, the reader should realize that these recommendations actually go beyond the scope of the RCTs, which compared 2 forms of nutritional support rather than 1 form of support versus no intervention. Although enteral nutrition is better than parenteral, there is no evidence that it is better than doing nothing. We have only a few RCTs that compare nutritional support with no nutritional support. Two trials have compared parenteral nutrition with no nutritional support. One, reported only in abstract form, found no difference between the treated and control groups (4), and the other found that patients in the treatment group had a worse outcome than the controls (longer duration of hospitalization and probably more infections) (5). Only 1 small trial has compared enteral nutrition versus no nutrition support; an organ failure score was the same in the 2 groups (6). Thus, this current systematic review may only be telling us that enteral nutrition is not as harmful as parenteral nutrition. This field needs a large multicenter trial that compares enteral nutrition with no intervention.

Cancer Prevention

Antioxidant Supplements Did Not Prevent Gastrointestinal Cancer and May Increase Mortality

Bjelakovic G, Nikolova D, Simonetti RG, et al. Antioxidant supplements for prevention of gastrointestinal cancers: a systematic review and meta-analysis. Lancet. 2004;364:1219-28. [PMID: 15464182] [Medline]

Ever since oxidative stress has been associated with cancer, researchers have speculated that antioxidant supplements might reduce the incidence and mortality of cancer. The investigators undertook a systematic review and meta-analysis to address this issue. They sought RCTs that compared the incidence of gastrointestinal cancer in people taking antioxidant supplements with the incidence in those individuals not taking such supplements. This systematic review also included an exploratory analysis of the effect of supplements on all-cause mortality.

The authors identified 14 RCTs (170 525 participants). Antioxidants (ß-carotene; vitamins A, C, and E; and selenium, alone or in combination) did not prevent all cancer (RR, 0.96 [CI, 0.88 to 1.04]) or esophageal, gastric, colorectal, or pancreatic malignancies. Selenium appeared to reduce the risk for liver cancer, but the trials were of low quality.

The authors used 2 different mathematical models to perform the meta-analyses: a fixed-effects model and a random-effects model. In general, the fixed-effects model is employed when the studies to be combined appear to be homogeneous; such models usually have narrower confidence intervals (and are thus more likely to find significant differences). Nine of the RCTs, 7 of which were of high quality, provided data regarding all-cause mortality. These 7 trials appeared to be heterogeneous, but, in keeping with the planned protocol, the investigators employed both models in the data combination. The fixed-effects model found a significant difference in mortality (RR, 1.06 [CI, 1.02 to 1.10]); the random-effects model did not show this increase. Because the RR was greater than 1.0, the authors raised concern that antioxidants may cause harm.

Thus, this systematic review found no evidence that antioxidant supplements prevented gastrointestinal cancer and found limited evidence of harm. The increase in all-cause mortality requires careful interpretation. The investigators undertook the analysis almost as an afterthought and did not include many other RCTs that assessed the utility of antioxidants in preventing other disease processes, such as coronary artery disease (although the investigators are currently conducting this broader analysis). They also were able to demonstrate this harm only with a mathematical model that would not usually be employed. On the other hand, shortly after this paper appeared, a meta-analysis of high-dose vitamin E use in elderly patients (in trials assessing the ability of vitamin E to prevent a variety of diseases) also suggested that the recipients of that antioxidant vitamin had a higher mortality rate (7).

An important message from this article is that we are not omniscient. Even agents that seem to us to be safe may not be. The discovery that antioxidants are ineffective is not the first time that high-quality RCTs disproved accepted dogma.

Fecal DNA Screening Detected More Colorectal Neoplasia than Fecal Occult Blood Testing; Neither Test Identified the Majority of Neoplastic Lesions

Imperiale TF, Ransohoff DF, Itzkowitz SH, et al. Fecal DNA versus fecal occult blood for colorectal-cancer screening in an average-risk population. N Engl J Med. 2004;351:2704-14. [PMID: 15616205] [Abstract/Free Full Text]

When used for colon cancer screening, fecal occult blood testing is both insensitive and nonspecific because not all cancerous lesions bleed and other noncancerous lesions do. Techniques can now test stool for DNA containing mutations that are associated with cancer. This study compared the Hemoccult II fecal occult blood test (Beckman-Coulter, Inc., Fullerton, California) with fecal DNA testing as a screening modality for colon cancer.

The study sample was average-risk, asymptomatic persons who were 50 years of age or older and who were about to undergo colonoscopy screening. Before endoscopy, they submitted stool samples that were tested for occult blood and for abnormal DNA base sequences. Using colonoscopy as the gold standard, the investigators compared the results of the 2 stool tests in all of the individuals who had colon cancer, polyps with high-grade dysplasia, and advanced neoplasia (an adenomatous polyp that contained villous elements or high-grade dysplasia or was >10 mm in size) identified. Originally, 4404 participants had both stool tests performed and underwent colonoscopy; from this population, the authors also tested subgroups of patients with smaller polyps (n = 648) or no polyps (n = 1423).

The researchers found colon cancer in 31 patients; 16 of these individuals (51.6%) had positive fecal DNA test results, but only 4 (12.9%) had positive fecal blood test results. Of 40 patients found to have polyps with high-grade dysplasia, 13 (32.5%) had positive fecal DNA results and 6 (15%) had positive fecal occult blood test results. Both of these differences between tests were statistically significant. The sensitivities of the DNA and the fecal occult blood tests for 347 patients with polyps measuring greater than 10 mm or containing villous elements were 13.5% and 10.2% (not significantly different). The differences in the frequencies of positive test results for DNA and foroccult blood were not statistically significant for patients with small polyps (7.6% and 4.8%, respectively) or without polyps (5.6% and 4.8%).

The authors concluded that neither test identified most neoplastic colonic lesions; however, the fecal DNA test was more sensitive for cancer and polyps with high-grade dysplasia and was not less specific.

The sensitivity of fecal DNA tests described in this large study is lower than reported in smaller studies, even if this type of test is more sensitive than fecal occult blood testing. Although the differences in specificity were not statistically significant, the fecal DNA test was positive more often in patients without any lesions. Therefore, we should not be too hasty in concluding that the 2 tests have equivalent specificity.

Furthermore, the fecal DNA test costs $400 to $800 per patient. Whereas the trials of fecal occult blood testing have shown that this colon cancer screening method reduces the death rate from colon cancer, the all-cause mortality was not significantly reduced. If this is a true observation (and not a type 2 error), no life-years would be saved; as such, the resources required to save a life-year would be infinite. If we use the costly fecal DNA test as our first screening measure, we are possibly making the resource expenditure "more infinite."

Liver Disease and AIDS

Peginterferon-2a and Ribavirin Combination Therapy Was Most Effective in Patients Infected with Both HIV and Hepatitis C Virus

Torriani FJ, Rodriguez-Torres M, Rockstroh JK, et al. Peginterferon alfa-2a plus ribavirin for chronic hepatitis C virus infection in HIV-infected patients. N Engl J Med. 2004;351:438-50. [PMID: 15282351] [Abstract/Free Full Text]

Liver disease associated with hepatitis C virus (HCV) infection is believed to progress more rapidly to cirrhosis in patients who are co-infected with HIV than in patients who are only infected with HCV. Although this may be true, an alternative explanation for much of the data supporting this hypothesis is that patients with HIV infection are living long enough for the natural history of HCV infection to catch up with them. In either event, we are now reading more about treating hepatitis C in patients with HIV infection. This RCT compared 3 drug regimens in these patients. The investigators enrolled 868 co-infected people who had well-controlled HIV infection and who had not been treatedpreviously with interferon or ribavirin. The patients were randomly assigned to 48-week courses of peginterferon-2a, 180 µg/wk, plus ribavirin, 800 mg/d; peginterferon-2a plus placebo; or interferon-2a, 3 million IU 3 times weekly, plus ribavirin. They followed the patients for 24 weeks after treatment ended. The primary end point was a sustained virologic response (undetectable serum HCV RNA level at week 72).

The sustained viral response rates in the 3 groups were 40% for peginterferon-2a plus ribavirin, 20% for peginterferon-2a plus placebo, and 12% for interferon-2a plus ribavirin. The rates of sustained virologic response were lower in the patients infected with HCV genotype 1 (approximately 60% of the patients): 29%, 14%, and 7% in the 3 treatment groups, respectively. Twelve deaths occurred during the study, and 2 of these were attributed to the drug therapy.

The dropout rate was high (25% to 39%); dropouts were attributed to adverse effects, poor treatment response, unwillingness to continue, or failure to return. Neutropenia and thrombocytopenia were more common among recipients of peginterferon; anemia was more common among those given ribavirin.

The investigators concluded that the preferred treatment for HCV in patients co-infected with HIV is combination therapy using peginterferon-2a and ribavirin. Two other recent studies comparing combination peginterferon treatment with nonpegylated interferon–based combination therapy in co-infected patients came to the same conclusion, although the sustained viral response rates were lower (27% vs. 12% and 27% vs. 20%) (8, 9). Of note, even if peginterferon-combination therapy was superior, the actual response rates are lower than those reported in HCV patients without HIV co-infection even though the HIV disease was well controlled. Effectiveness rates (those seen in general practice) are typically lower than efficacy rates (those seen in RCTs), so the response rates in the community are likely to be lower than what clinical trials have reported. Finally, in 1 of these latter 2 studies, 1 treatment-related death occurred in 412 treated patients (9). Thus, 2 large trials have suggested that co-infected patients have a 0.25% risk for dying as a consequence of hepatitis C antiviral therapy.

We know that only 10% to 20% of patients infected with HCV will ever experience liver failure. Treatment trials have assessed only an intermediate outcome (viral response). Thus, we have no data from RCTs that treatment actually reduces the incidence of end-stage liver disease, a fact recently noted by the U.S. Preventive Services Task Force in their discussion of screening for HCV (10). Because the treatment is expensive (a 48-week course of peginterferon-2a and ribavirin costs more than $20 000), unproven in its effect on clinical outcomes, and toxic, one might question the decision to use this regimen to treat any patients with hepatitis C, especially those with other serious comorbid conditions.

A key point for the critical reader: You cannot assume that improving a surrogate outcome will improve a clinical end point.

Should HIV Infection Be a Contraindication to Transplantation?

Ragni MV, Belle SH, Im K, et al. Survival of human immunodeficiency virus-infected liver transplant recipients. J Infect Dis. 2003;188:1412-20. [PMID: 14624365] [Medline]

In the past, HIV seropositivity was an absolute contraindication to liver transplantation. However, the advent of effective antiretroviral therapy has encouraged people to question this dictum. The investigators reported a retrospective case series comparing the outcomes of 24 HIV-seropositive liver transplant recipients (21 with end-stage hepatitis B or C and 3 with fulminant hepatic failure) with those of 5225 HIV-seronegative recipients.

At 12, 24, 36, 48, and 60 months after liver transplantation, survival was, respectively, 87%, 73%, 73%, 73%, and 36% in the HIV-seropositive group and 87%, 82%, 78%, 74%, and 71% in the HIV-seronegative group. The authors concluded that, because there was no difference in survival, HIV infection should not be a contraindication to transplantation.

Before accepting this conclusion at face value, we should consider several problems. First, these 24 HIV-seropositive patients were carefully selected for transplantation; their average Model for End-Stage Liver Disease (MELD) score was 15, whereas the average MELD score for transplanted patients in the United States is 22. (The MELD score is a prognostic index that is calculated from the international normalized ratio, serum bilirubin level, and serum creatinine concentration; it is used to set priorities among patients on liver transplant lists.) Second, publication bias is a concern. What if these authors, having a particularly good experience, published their data, while other centers with less impressive outcomes chose not to publish theirs? Third, the small quantity of patients who received long-term follow-up means that considerable statistical uncertainty surrounds the survival rates. Among these 24 patients, only 6 were followed for more than 2 years and only 3 were followed for more than 3 years. Because the 5-year survival rate was arithmetically much lower in the HIV-seropositive group, we cannot assume that the survival rates in the 2 groups are equivalent just because the difference between them was not statistically significant. Finally, from the perspective of resource utilization, the overall health care costs of HIV-seropositive patients after transplant will be higher than those of HIV-seronegative patients (because of the costs of the comorbid disease).

The resource expenditure is not just economic. Transplantable livers are limited resources. Using one in an HIV-seropositive patient denies its use in an HIV-seronegative patient. While we can make efforts not to exclude HIV-seropositive patients from mainstream medicine, these data do not provide a compelling argument to change our policy of making HIV infection a contraindication to liver transplantation.

Capsule Endoscopy

Capsule Endoscopy Usefulness Was Assessed in Patients with Obscure Bleeding

Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology. 2004;126:643-53. [PMID: 14988816][Medline]

The latest endoscopic advance marketed to gastroenterologists is the wireless capsule endoscope. The mechanism is contained within a capsule that the patient swallows. Peristalsis carries the capsule through the gastrointestinal tract; during that transit, the capsule visualizes the mucosa and transmits 2 pictures per second to a receiver strapped to the patient's body. This retrospective case series examined the utility of wireless capsule endoscopy in identifying the source of bleeding in patients with gastrointestinal bleeding from an unknown source.

The authors enrolled 100 consecutive patients who had ongoing overt gastrointestinal bleeding and no abnormalities on upper and lower endoscopy (26 cases), previous overt bleeding with unremarkable endoscopic findings (31 cases), or occult bleeding manifesting as an iron deficiency with fecal occult blood–positive stools (43 cases). The capsule successfully identified the source of the bleeding in 24 of the 26 patients with active bleeding. Most of the lesions were vascular malformations; 2 other patients had jejunal varices, and 1 patient had cancer. The diagnostic success rate in those with a history of overt bleeding was much poorer: Only 4 lesions were identified in 31 cases. The capsule identified 19 lesions in the 43 iron-deficient patients. Most were vascular malformations or small intestinal ulcerations. Of note, the capsule became impacted in previously unrecognized small intestinal strictures of 5 patients; 1 capsule was removed with push enteroscopy, and 4 patients required surgery. The authors concluded that capsule endoscopy was useful in patients with obscure bleeding.

To interpret these results, it is important to remember that the study sample was heterogeneous. Many patients with overt bleeding have a life-threatening process; presumably a diagnosis could be life-saving (even if the bleeding has stopped for a period of time, because the lesion may bleed again). On the other hand, patients with iron deficiency from occult bleeding are losing only very small amounts of blood each day and are not at risk for shock; the threat to their well-being is iron-deficiency anemia and the long-term prognosis of the underlying lesion itself. We can successfully treat anemia with iron therapy even if the source of the bleeding is unknown.

The reason to do a work-up in patients with iron deficiency is to identify the few who have cancer. This type of cancer is usually colonic and is found with colonoscopy; the capsule does not even reliably visualize the colon. The small intestinal lesions are usually vascular; if treatment is going to be considered, it will likely be surgery. The risk associated with surgery is probably at least as high as the risk from an occultly bleeding lesion suddenly beginning to bleed overtly. We know that iron-deficient patients have good prognoses if upper and lower endoscopies show no abnormalities (11). Thus, given the 5% incidence of capsule-induced obstruction, the risk associated with the procedure may outweigh the gain in iron-deficient patients with occult bleeding (because this condition can be successfully managed with iron therapy alone). We would conclude that capsule endoscopy is useful, however, for patients with ongoing major bleeding if upper and lower endoscopy is unrevealing.

Complementary and Alternative Medicine

An Herbal Preparation Alleviated the Symptoms of Irritable Bowel Syndrome

Madisch A, Holtmann G, Plein K, et al. Treatment of irritable bowel syndrome with herbal preparations: results of a double-blind, randomized, placebo-controlled, multi-centre trial. Aliment Pharmacol Ther. 2004;19:271-9. [PMID: 14984373] [Medline]

Many physicians believe that complementary and alternative medicine cannot treat diseases effectively. Numerous RCTs of such interventions have shown a beneficial effect (12); however, most of these trials are of relatively low quality, so the perceived benefit may be due to the introduction of bias. This RCT assessed the efficacy of a commercially available herbal preparation, STW 5, (composed of 9 plant extracts); a related preparation, STW 5-II (composed of 6 extracts); and a bitter candytuft monoextract for treating irritable bowel syndrome.

The investigators randomly assigned 208 patients with irritable bowel syndrome (defined as abdominal discomfort associated with disturbed bowel habits and lasting for 3 months or more) to receive 1 of 4 regimens: 1) STW 5 (n = 51), 2) STW 5-II (n = 52), 3) bitter candytuft monoextract (n = 53), or 4) placebo (n = 52). Participants in each of the 4 groups received 20 drops of the assigned preparation 3 times daily for 4 weeks. The primary outcomes were changes in scores of total abdominal pain and irritable bowel syndrome symptoms.

At the end of the study, the patients receiving either STW 5 or STW 5-II had significantly better scores on both scales than did those receiving the monoextract or placebo. There were no significant differences between the bitter candytuft monoextract group and the placebo group. The patients tolerated all of the interventions well with no serious adverse effects.

Both STW 5 and STW 5-II alleviated the symptoms of irritable bowel syndrome. This trial was a high-quality study; the randomization scheme was adequate, concealment of allocation occurred, a power calculation was performed, and the trial was double-blind. Only 5 of the 208 patients dropped out, and 4 of them had adequate data to be included in the final analyses. Both STW 5 and STW 5-II contain peppermint and caraway fruit, agents that have been shown to be useful in RCTs of patients with nonulcer dyspepsia (13). The commercial preparation costs approximately $15 to $20 for a 1-month supply. Consequently, practitioners should consider adding this herbal preparation to their therapeutic armamentarium for irritable bowel syndrome.

Author & Article Info

To summarize the findings contained in this update, we conclude the following:


  1. For nonulcer dyspepsia, PPIs appear to be useful, especially if they are available at generic prices; H. pylori eradication therapy may be marginally useful (NNTB, 15) but entails the risk for antibiotic resistance.
  2. Therapy to eradicate H. pylori did not prevent gastric cancer in a large (but perhaps still underpowered) RCT.
  3. For severe, acute pancreatitis, new evidence raises questions about the utility of antibiotic prophylaxis.
  4. Current evidence does not support the claim that patients with acute pancreatitis should receive nutritional support.
  5. Antioxidants have not been shown to prevent gastrointestinal cancer and may be harmful.
  6. For colon cancer screening, fecal DNA testing is probably too expensive for routine use, even though it appears to be more sensitive than fecal occult blood testing.
  7. In patients with hepatitis C and HIV infection, the most effective regimen for achieving a sustained response of HCV viremia is combination therapy with peginterferon-2a and ribavirin. However, no RCT comparing treatment with no treatment is available to assess the hypothesis that antiviral therapy prevents end-stage liver disease from hepatitis C.
  8. Liver transplantation may be less effective and will probably require more post-transplant resource utilization in patients who are also infected with HIV compared with HIV-seronegative recipients.
  9. Capsule endoscopy is useful for patients with overt bleeding in whom upper and lower standard endoscopy has not provided a diagnosis; its utility in patients with iron deficiency is debatable.
  10. The herbal preparation STW 5 appears to be effective in treating irritable bowel symptoms.

Author and Article Information
Author & Article Info

From Olive View-UCLA Medical Center, Sylmar, California, and the Veterans Affairs Medical Center, Washington, DC.

Potential Financial Conflicts of Interest: None disclosed.

Requests for Single Reprints: Ronald Koretz, MD, Department of Medicine, Olive View-UCLA Medical Center, 14445 Olive View Drive, Sylmar, CA 91342; e-mail, This email address is being protected from spambots. You need JavaScript enabled to view it. .

Current Author Addresses: Dr. Koretz: Department of Medicine, Olive View-UCLA Medical Center, 14445 Olive View Drive, Sylmar, CA 91342.

Dr. Lipman: Gastroenterology/Hepatology/Nutrition Section, Department of Medicine, Veterans Affairs Medical Center, 50 Irving Street NW, Washington, DC 20422.

Author & Article Info

1. Golub RSiddiqi FPohl D. Role of antibiotics in acute pancreatitis: a meta-analysis. J Gastrointest Surg. 1998;2:496-503. [PMID: 10457308].[Medline]

2. Sharma VKHowden CW. Prophylactic antibiotic administration reduces sepsis and mortality in acute necrotizing pancreatitis: a meta-analysis. Pancreas. 2001;22:28-31. [PMID: 11138967].[Medline]

3. Koretz RLLipman TOKlein S. AGA technical review on parenteral nutrition. Gastroenterology. 2001;121:970-1001. [PMID: 11606512].[Medline]

4. Hyde DFloch MH. The effect of peripheral nutritional support and nitrogen balance in acute pancreatitis [Abstract]. Gastroenterology. 1984;86:1119.

5. Sax HCWarner BWTalamini MAHamilton FNBell RH JrFischer JEet al. Early total parenteral nutrition in acute pancreatitis: lack of beneficial effects. Am J Surg. 1987;153:117-24. [PMID: 3099588].[Medline]

6. Powell JJMurchison JTFearon KCRoss JASiriwardena AK. Randomized controlled trial of the effect of early enteral nutrition on markers of the inflammatory response in predicted severe acute pancreatitis. Br J Surg. 2000;87:1375-81. [PMID: 11044164].[Medline]

7. Miller ER 3rdPastor-Barriuso RDalal DRiemersma RAAppel LJGuallar E. Meta-analysis: high-dosage vitamin E supplementation may increase all-cause mortality. Ann Intern Med. 2005;142:37-46. [PMID: 15537682].[Abstract/Free Full Text]

8. Chung RTAndersen JVolberding PRobbins GKLiu TSherman KEet al. Peginterferon alfa-2a plus ribavirin versus interferon alfa-2a plus ribavirin for chronic hepatitis C in HIV-coinfected persons. N Engl J Med. 2004;351:451-9. [PMID: 15282352].[Abstract/Free Full Text]

9. Carrat FBani-Sadr FPol SRosenthal ELunel-Fabiani FBenzekri Aet al. Pegylated interferon alfa-2b vs standard interferon alfa-2b, plus ribavirin, for chronic hepatitis C in HIV-infected patients: a randomized controlled trial. JAMA. 2004;292:2839-48. [PMID: 15598915].[Abstract/Free Full Text]

10. Calonge NRandhawa G. The meaning of the U.S. Preventive Services Task Force grade I recommendation: screening for hepatitis C virus infection. Ann Intern Med. 2004;141:718-9. [PMID: 15520429].[Abstract/Free Full Text]

11. Rockey DCCello JP. Evaluation of the gastrointestinal tract in patients with iron-deficiency anemia. N Engl J Med. 1993;329:1691-5. [PMID: 8179652].[Abstract/Free Full Text]

12. Koretz RLRotblatt M. Complementary and alternative medicine in gastroenterology: the good, the bad, and the ugly. Clin Gastroenterol Hepatol. 2004;2:957-67. [PMID: 15551247].[Medline]

13. Thompson Coon JErnst E. Systematic review: herbal medicinal products for non-ulcer dyspepsia. Aliment Pharmacol Ther. 2002;16:1689-99. [PMID: 12269960].[Medline]