Graham also neglects the evidence from randomised controlled trials that the addition of antibiotics to therapy with acid-suppressant drugs and/or bismuth improves ulcerhealing rates and reduces relapse rates. To cite two examples: a trial of amoxycillin, metronidazole, and ranitidine versus ranitidine alone showed improved duodenal ulcer-healing rates at 6 weeks (92% vs 75%) and greatly improved relapse rates (9% vs 93%); and in a trial of bismuth, amoxicillin, metronidazole, and ranitidine versus bismuth and rantitidine together4 the ulcer-healing rates at 8 weeks were 89% and 81 %, respectively, and the relapse rates at 1 year were 6% and 76% respectively. Graham disregards these benefits, and in doing so he ignores completely the clinical significance of H pylori: successful eradication of this infection fundamentally changes the natural history of peptic ulceration from a relapsing, remitting condition to a
curable one. Graham dismisses the reduction in ulcer relapse rates as not indicating a causal relation between H pylori and peptic ulceration by postulating that ulcers arise in a background of mucosal inflammation that is opportunistically infected by H pylori. He presents no evidence to support this postulate and no mechanism whereby this background mucosal inflammation might occur. Yet is is known that H pylori can damage mucosal epithelial cells by several mechanisms, including the production of phosphilipases, ammonia, and a vacuolating toxin. Furthermore, on the basis of Graham’s thinking, an explanation of why peptic ulcer relapse rates fall when antibiotics are added to regimens containing acidsuppressant drugs and/or bismuth would require antibiotics to possess direct ability to repair mucosal damage. There exists no evidence to support such a possibility. Graham correctly says that H pylori does not fulfil Koch’s postulates for the causation of duodenal ulceration, but Koch’s postulates are applicable to conditions of only single aetiology. It is incorrect to attempt to translate these criteria to conditions, such as duodenal ulceration, whose aetiology involves several factors. It is widely accepted that there are important risk factors other than H pylori colonisation for duodenal ulcer, Therefore it is not unexpected that Koch’s third postulate ("when a pure culture is inoculated into a susceptible animal species the typical disease must result") has not been fulfilled for H pylori and duodenal ulcerationit is unlikely that any animal model can simulate both the conditions within the human stomach and the simultaneous existence of other risk factors. The aetiological importance of the risk factors for peptic ulceration is not known but the clinical (and public health) importance of H pylori stems from the fact that possession of this particular risk factor is amenable to a simple, shortterm, intervention and that the effect of this intervention is sustained. Graham’s view may divert the attention of clinicians, researchers, and funding bodies away from the important issues surrounding the use of H pylori eradication therapy in the management of peptic ulceration and
dyspepsia. *Liam
Murray, Ian Harvey
Department of Social Medicine, University of Bristol, Bristol BS8 2PR, UK 1
2
3
4
Van der Voort LHM, Van den Bos AP, Kamsteeg H. In vitro bactericidal effects of CBS on Helicobacter pylori. Rev Esp Enferm Dig 1990; 78 (suppl 1): A103. Huesca M, Gold B, Sherman P, Lingwood C. Colloidal bismuth subcitrate (CBS) blocks Helicobacter pylori adhesion to glycerolipid receptors. Gastroenterol 1992; 102: A639. Hentschel E, Brandstatter G, Dragosics B, et al. Effect of ranitidine and amoxycillin plus metronidazole on the eradication of Helicobacter pylori and the recurrence of duodenal ulceration. N Engl J Med 1993; 328: 308-12. Rauws EJ, Tytgat GNJ. Cure of duodenal ulcer disease associated with eradication of Helicobacter pylori. Lancet 1990; 335: 1233-35.
1580
String test for
Helicobacter pylori
SiR-Perez-Trallero and colleagues (March 11, p 622) describe a non-endoscopic method of obtaining Helicobacter pylori for culture. A nylon string, when swallowed, absorbs gastric secretions which are cultured for H pylori. This test is likely to be simpler and better tolerated than endoscopic biopsy but the claim that the results of culture for both the string test and biopsy were similar is not supported by the data. There was agreement between the two tests in 75% of cases; however, a substantial amount of this may be accounted for by chance alone. Indeed, when analysed by the kappa statistic,’ which measures concordance while taking chance into account, intertest agreement was only moderate (x=0-47, 95% CI 0-15-0-79). It is not clear whether the results from endoscopic biopsy were obtained from culture of a single antral specimen or from multiple biopsy samples of antrum and corpus. Gastric colonisation is often patchy, and an antral biopsy may be negative in the presence of infection elsewhere in the stomach, especially when there is coexisting gastric atrophy or metaplasia or after omeprazole therapy. We agree that, in general, culture is the gold standard for diagnosis of infection, but this is not so for H pylori, which is more accurately detected by the "C or ’4C urea breath test.2 Although the string test is a novel concept, its value in the detection of H pylori is likely to be limited.
*Stephen E Patchett, Hugh E Mulcahy, Michael J G Farthing Digestive Diseases Research Centre, Medical College of St Bartholomew’s Hospital, London EC1M 6BQ, UK
1 2
Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Measurement 1960; 20: 37. Bell D. Urea breath tests. In: Northfield TC, Mendal M, Goggin PM, eds. Helicobacter pylori infection. Lancaster: Kluwer, 1993: 127-39.
Authors’
reply
SiR-Our small series lacks the statistical power of a larger one but we felt that publication of our preliminary results was justified by the need for a simpler alternative to endoscopy. For instance, if only H pylori is sought, endoscopy can be avoided in all patients with positive stringtest results. Our short report did not contain the statistical analysis: McNemar’s paired test (with Yates’ correction) yielded X2=044 (p=0-5) whereas the published article only noted that the results of both tests were similar (not
different). patients have been evaluated simultaneously by endoscopic biopsy and the string test. 2 patients were double negative; 4 were double positive; 1 was biopsy positive and string-test negative; and 1 was biopsy negative and stringtest positive. The addition of these few patients to the series does not change our conclusion. It does reduce the confidence interval, although we do not consider intertest agreement to be of prime importance (x=0-47, 95% CI 0-20-0-73). In the original series and in the augmented series, the p for the kappa test was less than 0-005 (on a null hypothesis ofx=0). We compared culturing the distal part of the string with culturing two antral biopsy specimens per patient. Cultures of antral biopsy material may yield false negatives in the presence of H pylori infection but the probability of a falsenegative result of antral biopsy when compared with biopsy material obtained from multiple sites in the stomach is low, less than 3% in a recent report.’ 8
new
I
We do
doubt that the breath test is more sensitive in detecting pylori infection than either biopsy culture or string culture, but the breath test requires facilities not not
H
and it is expensive. Its most is that it does not yield H pylori. The important deficiency for further is needed study (eg, detection of microorganism DNA and fingerprinting) and in routine clinical cytotoxins so that antibiotic management susceptibility can be assessed. National Institutes of Health consensus development A US H in conference statement on pylori peptic ulcer disease (Feb 8, 1994) lamented the lack of a "readily available, inexpensive, and accurate noninvasive method to monitor the eradication of H pylori. In the absence of such an assay, routine monitoring for relapse, reinfection, or treatment failure cannot be recommended". The string test should now be thoroughly evaluated to see whether it satisfies those
available
to
district
hospitals
requirements. *Emilio Perez-Trallero, Milagrosa Montes Microbiología y Unidad de Epidemiología Infecciosa,
Servicio de
Facultad de
Medicina, Hospital NS Aránzazu, Apartado 477, 20080 San Sebastián, Spain
1 Genta RM, Graham DY.
Comparison of biopsy sites for the histopathologic diagnosis of Helicobacter pylori: a topographic study of H pylori density and distribution. Gastrointest Endosc 1994; 40: 342-45.
these agents concentrations
had to
achieved large enough systemic produce effects, the initial local reaction of
hypoaesthesia would exclude all except tetrodotoxin. Hypotension and possibly headache subsequent to vasodilation, plus the slight visual disturbance, might occur with tetrodotoxin; nevertheless this agent or one with similar cellular action seems excluded by the absence of muscular weakness. Some of the symptoms (hypotension, headache, and vomiting) point to a possible involvement of histamine, locally manifested by histamine-like actions such as pain, erythema, and swelling, produced directly or via release of histamine at the point of contact. Because these symptoms were lacking, we are unable to explain the pattern of toxicity other than to suggest an allergic reaction of an unusual type. The suggestion is based on the nature of the exposure to the frog and the unlikelihood that any toxic secretion could have achieved substantial blood concentrations. Whether potent contact allergens are present in these secretions would be an interesting topic for further study. M J Forbes, *M J
Tyler,
I S de la Lande
*Departments of Zoology and Clinical and Experimental Pharmacology, University of Adelaide, South Australia 5005, Australia
Allergic response to Brunei frog Staurois natator
1
highly toxic skin secretions, which are fatal when ingested by human beings and other mammals.’-3 We are unaware of any previous report of frog skin secretions having an effect through unbroken skin. The incident reported here occurred on Sept 1, 1993, when a 14-year-old boy was climbing up a waterfall at Labi in Brunei. He brushed the outside of his right hand against a frog. Moments later he complained of numbness on the outer aspect of his hand where the frog had been in contact. Very soon after, he felt faint and nauseated, and remarked that the peripheral vision of his right visual field was blurred. He developed a left-sided headache and fell asleep in the car. The trip home took about 90 min, and he retired to bed. 4 h after the incident he felt well enough to go out for a meal. Next morning the numbness had returned to his hand; the left-sided headache returned, and he spent the morning in bed. By midday, he felt worse, and started to vomit profusely for about an hour before he fell asleep. By about 1600 h he felt better and was able to get up. Although the headache and nausea left after about 36 h, he reported that the numbness in his hand recurred for 3 days. After this he
SiR-Many frogs
made
possess
full recovery. He had no relevant past medical other than a tendency to show prolonged skin reactions to scratches and stinging stimuli, especially microscopic jellyfish stings, manifested by prolonged itching and a rash lasting a month but no systemic symptoms. The frogs on the face of the waterfall are the "blackspotted rock frog" Staurois natator, which are widely distributed in Brunei and Sabah and are known to produce a skin toxin considerably stronger than that of the average frog species. One suggestion was that the symptoms were caused by brushing against the toxic urticarial plant Laportia which grows along the edge of streams. Subsequently, the route taken by the boy was followed; no urticarial plants were encountered. Histological secretions of the skin of an adult Staurois natator collected at the site revealed hypertrophy of both mucous and granular glands. The physiologically active agents in frog skin secretions are diverse, including simple amines such as tryptamines and catecholamines, more complex amines such as tetrodotoxin and batrachotoxin, and cardioactive steroids such as bufogenines.4 There are also various peptides including the caerulein and bombesin families; others (magainins) are antimicrobial. Even if any of a
history,
2
Allen ER, Neill WT. Effect of marine toad toxins on man. Herpetologica 1956; 12: 150-51. Licht LE. Death following possible ingestion of toad eggs. Toxicon
1967; 5: 141-42. 3
4
5
Myers CW, Daly JW. Preliminary evaluation of skin toxins and vocalisations in taxonomic and evolutionary studies of poison-dart frogs (Dendrobatidae). Bull Am Mus Nat Hist 1976; 157: 173-262. Russell FE, Dart RC. Toxic effects of animal venoms. In: Amdur M, O’Doull J, Klaassen CD, eds. Casarett and Doull’s toxicology. 4th ed. New York: Pergamon Press, 1991: 753-803. Bevins CL, Zasloff M. Peptides from frog skin. Annu Rev Biochem 1990; 59: 395-414.
Listeriosis from cheese
consumption of raw-milk
SIR-On April 28, 1995, the French National Reference Centre (NRC) for listeriosis (Institut Pasteur, Paris) notified health authorities of a cluster of six human cases of Listeria monocytogenes caused by an unusual phagovar
(2389:3552:2425:1444:1317:3274:2671:47:52:108:340:31), subsequently called the epidemic strain. The six strains were isolated between April 2 and April 19 in hospitals located in different regions of France. In addition, the NRC reported four epidemic strains among the 2500 food isolates received since Jan 1. These strains were isolated from dairy products. Preliminary investigations showed that the four dairy products that were contaminated with the epidemic strain were all samples of Brie de Meaux (a raw-milk soft cheese) and that all ten cases notified by May 10 had eaten brie cheese. However, the brand name of the cheese consumed by cases could not be obtained. The brie had been bought either in a supermarket or at a local market and had been cut on the counter. Further investigations were then conducted by the Ministry of Agriculture to identify the origin of the brie sold at the places at which the patient bought it. Concomitantly, a case-control study was conducted; controls (2-4 per case) were matched to cases for underlying condition, area of residence, and age. The results of investigations implicated one specific chain of Brie de Meaux production. Batches of brie incriminated by the investigation were then recalled. Disinfection and control measures were reinforced at the production level. Interview of 16 patients and 26 controls showed an increased risk for listeriosis among consumers of brie (odds ratio 7-0; 95% CI On May 18, a letter was sent to health 1-1-56-2; p<005). 1581