Detection of Helicobacter pylori in the Setting of Acute Upper Gastrointestinal Bleeding

Of all the possible indications proposed for testing and treating Helicobacter pylori, peptic ulcer disease (PUD) is the one for which there is the most scientific evidence and for which consensus is unanimous. 1–4 Testing for H. pylori is problematic, however, in the setting of one of the major complications of PUD. Upper gastrointestinal bleeding (UGIB) is the most common complication of PUD and can also be life threatening. Detecting H. pylori infection during an acute bleed can be a challenge as the accuracy of several of the diagnostic methods for H. pylori is diminished in this setting. 5–8 False negative results seem to be particularly prevalent; an error one would want to avoid given that H. pylori eradication clearly reduces the rate of recurrent ulcer hemorrhage. 9–13 In this issue of the Journal of Clinical Gastroenterology, Winiarski et al attempt to address the challenge of H. pylori detection in acute UGIB by evaluating the urea breath test in this setting. 14 There are both invasive and noninvasive methods of H. pylori detection. Invasive methods require gastroscopy and include histology, culture, and rapid urease tests. Except, perhaps for the decreased sensitivity of culture due to the fastidious nature of this bacterium, these invasive methods have excellent diagnostic accuracy for patients not actively bleeding. 15 The urea breath test (UBT) is a noninvasive method that also has excellent accuracy. 16 The stool antigen test, another noninvasive technique, is a newer method with excellent results. 17 Other noninvasive techniques such as detecting salivary or whole blood antibodies have been more disappointing. 18,19 Serology has been extensively studied. Certain kits in certain hands have done quite well but a major problem with variability of results exists, not only between kits but also with the same kit in different laboratories. 15,20,21 Hence, serology should only be used when validated in that particular laboratory. All these methods have advantages and disadvantages. The invasive methods obviously have the disadvantage of requiring endoscopy. The urea breath test requires that the patient be off proton pump inhibitors prior to testing. 22 The stool test is not preferred in North America simply because it involves feces and serology is associated with a high false positive rate because antibodies persist even after infection has been eradicated. 23,24 The problem of detecting H. pylori during an acute bleed has been clearly demonstrated for the rapid urease test. Several studies have shown a reduced sensitivity with false negative rates of about 25% to 40%. 5–8,25 One study found a sensitivity of only 41%. 26 The reason for this is not clear. It had been suggested that blood plasma has a buffering effect 27 thus adversely affecting the urease activity, but others have found this not to be the case. 28 In fact, there was no statistical difference in urease test results when biopsy specimens were either tested immediately, or after proliferation of organisms was permitted with culture in blood for 5 minutes. Rather a more recent study reported a direct bactericidal effect of human plasma against H. pylori.29 Others have suggested that large volume gastric lavage for UGIB prior to endoscopy may result in temporary removal of the bacteria thus leading to false negative diagnostic tests. 30 Other invasive methods of detection have also been less performant in acute UGIB. Colin et al 26 found very poor test performance for both histology and culture with sensitivity of only 33% and 34% respectively. Some studies have found that histology fairs better than the urease test in this setting, but the performance is still not as good as in non-bleeders with a sensitivity rate of 86% in one study. 8,31 Noninvasive methods of detection have not been as extensively studied in acute UGIB. The accuracy of serology would not be expected to be different and in fact, it seems not to be. 5,31 One would question the accuracy of the stool antigen in patients having melena, but one study suggested this was not a problem; they obtained a high sensitivity (96%) but the specificity was quite poor (33%), highlighting the need for further study. 32 UBT has had very promising results with 95% sensitivity in one study. 31 Griñó et al 5 however found that although the UBT had a sensitivity of 91% the specificity was 78% in the setting of UGIB, substantially lower than in non-bleeders. In this issue Winiarski et al 14 reported that the test performance of the UBT was not inferior in acute UGIB than in non-bleeding patients. The sensitivity and specificity in the 258 non-bleeding patients as well as in the 81 bleeding patients was greater than 95% in all cases. The number of UGIB subjects in this study is still relatively small especially in the H. pylori negative group (n = 17), such that the confidence intervals around the test performance estimates are not narrow. Nonetheless, UBT appears to offer an unsurpassed accuracy in detection of H. pylori in UGIB patients. Larger studies should be performed to confirm this finding and to determine if the severity of the bleed and hence the amount of blood in the stomach influences the test performance. One of the challenges in evaluating H. pylori detection methods in UGIB is choosing the appropriate gold standard. Winiarski et al 14 chose to use serology and certainly this is understandable, as it is not affected by the presence of blood in the stomach. However, it is not a particularly accurate method as described above. Griñó et al chose to use positive histology or the combination of breath test and serology as their gold standard. 5 At first glance, a combination appears to be the more accurate standard, but it is not always clear how to label discordant results. Should a positive serology with negative histology and urea breath test, for example, be considered a false positive serology or false negative histology and urea breath test in the setting of an UGIB? The answer is not so clear. An accurate gold standard, however, must be used for the result to be credible and all studies should strive to achieve this and express the results in terms of sensitivity, specificity, positive, and negative predictive values with 95% confidence intervals. In the clinical setting, using 1 test rather than a combination to detect H. pylori would be ideal. Given the poor sensitivity of the urease test, this is obviously not the test to use. Given the false positive results, serology probably should also not be the test to use as one would not want to falsely attribute the cause of the PUD to H. pylori if it truly was not there. This could lead to the belief that the cause of the bleeding ulcer was subsequently treated and the patient not placed on maintenance acid suppression therapy. With regards to the UBT, one wonders how easily it would be to administer to a patient with an acute UGIB; drinking a urea-containing solution and blowing into a tube might not be that simple in someone with nausea and hematemesis. This seemed not to be a problem with Winiarski et al's 14 report as they claim that none of the patients had any problem completing the UBT. Perhaps this is due to their use of a capsule of 13 C rather than a solution. Three studies 5,14,31 have been able to obtain reasonably accurate results with UBT in the setting of acute UGIB. One study found a positive predictive value of 96%. 5 The negative predictive value in this study, however, was 54%. This is less than desirable given a significant occurrence of false negative results. Winiarski et al 14 obtained a much superior negative predictive value. Larger studies are however required prior to advocating a negative UBT test as a sufficiently accurate result in this setting. Hence, in the clinical setting of an UGIB, if the urease test, histology, or UBT result is positive, one can rely on that test being accurate. If any of these are negative, however, the result may not be reliable. Even a negative UBT, given the conflicting estimates of negative predictive value, 5,14 may be a false negative result. Another test should therefore subsequently be done once the acute episode is over, at least until further larger studies confirm the excellent negative predictive value of the UBT obtained by Winiarski et al 14 in their study. One can also argue that the search for H. pylori should, in fact, only begin to be undertaken once the acute bleed is over. However, this introduces the problem of having to hold acid suppression for 2 weeks prior to testing, something one may not wish to do in a patient who recently bled. Regardless of which of these approaches one uses, it is important to test for and treat H. pylori infection in a patient with PUD, whether the ulcer was complicated by bleeding or not.