Reply

Abstract

We thank Drs Meyer and Vassar for their interest in our study1Pinto-Sanchez M.I. et al.Gastroenterology. 2017; 153: 448-459.e8Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar and comments on the use of the Fragility Index (FI). We agree that FI may be useful as an intuitive measure of the robustness of randomized, control trials2Walsh M. et al.J Clin Epidemiol. 2014; 67: 622-628Abstract Full Text Full Text PDF PubMed Scopus (355) Google Scholar; however, caution needs to be exercised when interpreting the FI.3Carter R. et al.Eur Heart J. 2017; 38: 346-348PubMed Google Scholar, 4Ahmed W. et al.Crit Care Med. 2016; 44: 1142-1143Crossref PubMed Scopus (58) Google Scholar Carter et al simulated 60,000 clinical trials with sample sizes ranging from 100 to 1000 per group and calculated the Fisher’s exact P-value and FI for each data study.3Carter R. et al.Eur Heart J. 2017; 38: 346-348PubMed Google Scholar They demonstrated that, similar to the P-value, the increase in the FI was not an indicator of the strength of the effect and recommended a broader approach, which would also encompass estimated effect size, study design, and mitigation of biases. Furthermore, the use of the standalone FI has been criticized, especially in the case of small sample size studies, which are often accompanied by low number of events and ultimately lead to a lower FI.4Ahmed W. et al.Crit Care Med. 2016; 44: 1142-1143Crossref PubMed Scopus (58) Google Scholar We feel we made it clear in our article1Pinto-Sanchez M.I. et al.Gastroenterology. 2017; 153: 448-459.e8Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar that the findings in our study are not robust and this is why we recommended a phase III randomized trial and do not advocate the use of Bifidobacterium longum in clinical practice at this time. As discussed in our article,1Pinto-Sanchez M.I. et al.Gastroenterology. 2017; 153: 448-459.e8Abstract Full Text Full Text PDF PubMed Scopus (410) Google Scholar we think it is unlikely that the beneficial effect of B longum NCC3001 on depression in this pilot study could be due to the higher baseline depression scores in the probiotic group and regression to the mean. The reason for this being that the treatment difference for Hospital Anxiety Depression score was reproduced by a robust analysis of covariance (F = -2.02; 95% confidence interval -3.60 to -0.43; P = .0126; multiplicity adjusted P = .0252). We also showed a clear separation between the linear decrease in the Hospital Anxiety Depression score of the probiotic and placebo groups. Furthermore, and most important, the effect of B longum NCC3001 on depression scores was associated with decreased brain activation patterns in multiple areas involved in mood regulation, such as the amygdala and the frontolimbic complex, and adequate relief of irritable bowel syndrome symptoms. We agree that the small sample size is a particular limitation of this study, but this is often the case with pilot studies. The results, however, are very encouraging and are accompanied by mechanistic insights particularly related to the findings on functional magnetic resonance imaging. As stated, we agree that these results should be validated in adequately powered phase III trial, where the FI could be applied. The Fragility of Probiotic Bifidobacterium longum NCC3001 Use for Depression in Patients With Irritable Bowel SyndromeGastroenterologyVol. 154Issue 3PreviewThe trial by Pinto-Sanchez et al1 published in a recent issue of Gastroenterology reports a relationship between probiotic Bifidobacterium longum NCC3001 and a decrease in depression scores in patients with irritable bowel syndrome (IBS). IBS is considered a gut–brain disorder and, thus, many patients with IBS experience depression and anxiety symptoms.2–4 Therefore, the results of this trial could be far reaching for people with IBS and lead to an increased use of probiotics with Bifidobacterium longum NCC3001. Full-Text PDF