Should we exclude individuals from endoscopy based exclusively on the absence of alarm symptoms?
M. Dinis-Ribeiro,H. Lomba-Viana,R. Silva,N. Fernandes,N. Abreu,C. Brandão,L. Moreira‐Dias,A. da Costa‐Pereira
DOI: https://doi.org/10.1080/00365520410003317
2004-01-01
Scandinavian Journal of Gastroenterology
Abstract:TO THE EDITOR: We have read with great interest the paper by Voutilainen et al. (1), which reports on the relation between dyspepsia, alarm symptoms and several endoscopic findings, including gastric cancer. As the authors note, patient’s history and physical examination alone may be unreliable in diagnosing the aetiology of dyspepsia (2). After the exclusion of individuals undergoing follow-up for several reasons, such as those with lesions associated with gastric cancer, chronic atrophic gastritis or dysplasia, they analysed the data of 3378 patients. In this set, considering cancer diagnosis on the basis of the presence of alarm symptoms, the validity was 67.0% (2281 being those individuals with no alarm symptoms and no cancer and those with cancer correctly identified by the alarm symptoms/3378), with a positive predictive value of 1.1% (12/1104) but a very high negative predictive value, 99.8% (2269/2274). These estimates, even taking into consideration the biases of selection referred to by the authors, allowed them to conclude that alarm symptoms are a major indication for endoscopy. But, as Voutilainen et al. did stress, these results are the opposite of those in other studies, where alarm symptoms did not predict significant endoscopic findings (3). In fact, we would be very cautious about excluding patients without alarm symptoms from endoscopic examination. Even in this study, 30% (5 in 17) of cancer cases did not have alarm symptoms (false-negatives); and most symptomatic gastric cancers had metastasized at the time of endoscopy, with a short survival thereafter. Furthermore, we should also consider the diagnosis of other gastric lesions defined in this paper as gastropathy. Endoscopic observation shows a very high interobserver variability for several kinds of gastropathy, which may include lesions such as atrophy and intestinal metaplasia. In this study, 23 endoscopists performed endoscopic examinations, and gastropathy was identified in 1311 individuals. Considering again the accuracy of symptoms in the diagnosis of gastric lesions as gastropathy or cancer, validity decreases to 55.9% (1890/3378). The negative predictive value is now 62.5% (856/2274), with 41% falsely evaluated has having no gastric lesions (n = 632). This in fact could have a major impact on considering alarm symptoms as the only indication for prompt endoscopy, as gastropathy, as defined endoscopically, could reveal histological lesions for which diagnosis or follow-up might be worthwhile. We are following 136 patients (median follow-up time 3 years) with gastric lesions as severe as atrophic chronic gastritis, such as atrophy, metaplasia and dysplasia. In all of these patients, magnification chromoendoscopy (4) and serologic evaluation with pepsinogen I and II (5) were performed, as well as clinical questioning on symptoms. Not all of our patients were taking medication at the time of their visit, and none of our patients presented with alarm symptoms. Dyspepsia was the most commonly reported complaint. In patients with lesions as severe as low-grade dysplasia (n = 23), only 11% complained of dyspepsia. In this small sample, the negative predictive value of dyspepsia for prediction of endoscopic lesions was 66%. On the other hand, even in such a selective population, we found validity measures for diagnosis with serum levels of PGI and PGII, especially the PGI/II ratio, similar to others, including those in population-based studies (5). Thus, we must stress that alarm symptoms should not be over considered as the sole indication for endoscopy in a general population, as they do not accurately correlate with cancer or other lesions defined as gastropathy. Furthermore, although excluded from this study, also in patients with lesions associated with cancer, symptoms did not correlate at all with progression to cancer, at least not in the early forms. Other non-invasive ways to exclude patients from endoscopy, such as serum pepsinogen I and II, must be developed and implemented in this setting and in other settings such as in young patients without alarm symptoms in whom endoscopic examination was until now neither cost-effective nor readily acceptable, as Voutilainen et al. have stressed.