The study shows an unfavourable effect on symptoms in subjects with IBS after intake of L. plantarum MF1298 compared to placebo. To our knowledge, similar unfavourable effects of probiotics have not been reported in subjects with IBS. Other studies with probiotics in subjects with IBS show either no effect or a favourable effect . The divergent results could be related to different probiotic properties and health effects of the genera, strains, and species in use. Quigley assumed the possible superiority of Bifidobacterium spp for treatment in IBS . Bifidobacterium (B.) animalis DN-173010 increased stool frequency in subjects with constipation at entry in a large study . O'Mahony et al. compared the symptomatic effect of L. salivarius UCC4331 and B. infantis 35624 in subjects with IBS . B. infantis 35624 induced a favourable effect on IBS symptoms. In a second study by the same researchers, the beneficial effect of B. infantis 35624 was confirmed . Lactobacilli have been evaluated in several trials with inconsistent results, but no deleterious effects have been reported. One trial of L. reuteri ATCC 55730 showed no significant effect on gastrointestinal symptoms in patients with IBS, while another trial also in patients with IBS showed no effect of L. casei strain GG . L. casei strain GG in combination with other probiotics showed a positive effect on IBS symptoms in one study by Kajander . Two trials with L. plantarum 299v showed a reduction of abdominal pain and flatulence, while a small crossover study found no effect on symptoms of IBS .
Reports of unfavourable effects of probiotics are rare and probiotics have until recently been regarded as safe . Untoward effects were reported in only three out of 185 human studies . A strain of L. acidophilus increased faecal protein catabolites in healthy volunteers in one study, while Saccharomyces cerevisiae increased disease activity in patients with stable Crohn's disease in one study, and in another study increased serum glucose in healthy volunteers . Sepsis has been reported in some subjects using probiotics . The most alarming report was published in 2008, showing increased mortality of severe acute pancreatitis following treatment with a multispecies probiotic preparation .
The doses of probiotics used for the treatment of IBS in other trials vary from 2 × 108 to 2 × 1010 CFU per day [7, 15, 19, 23, 24]. In a dose-finding study, the optimal dose of B. infantis 35624 was 1 × 108 CFU which was superior to placebo, 1 × 106 CFU and 1 × 1010 CFU. However, the 1 × 1010 CFU dose was associated with significant formulation problems . The only previous study in humans with L. plantarum MF1298 is a study of the survival and persistence of the strain in the gastrointestinal tract in 17 healthy volunteers. They were given 6 × 109 CFU per day of L. plantarum MF1298 either as a freeze-dried preparation or present in 15 g fermented sausage. No gastrointestinal symptoms or other adverse events were spontaneously reported, but such symptoms were not systematically recorded . The dose of 1 × 1010 CFU L. plantarum MF1298 selected for this study was in the same order as the doses used in other studies with lactobacilli . Despite the lack of evidence for a reduced or detrimental effect of high doses, we cannot exclude that a too high dose might have contributed to the unfavourable outcome of this study.
It has been proposed that the most potent probiotics may have increased pathogenicity . The multispecies probiotic preparation used in the study of acute, severe pancreatitis where mortality was increased, was composed of six strains . These strains, selected from 69 different probiotic bacteria, had better probiotic properties in combination than the individual components. L. plantarum MF1298 had the best in vitro probiotic properties of 22 strains [8–10]. The probiotic with the best probiotic properties as determined in vitro is not necessarily the best one to "confer a health benefit on the host".
The possibility of contamination of capsules by pathogens was excluded in our study, but the presence of endotoxins in the L. plantarum MF1298 and placebo preparations was not checked. This is, however, unlikely to be the reason for the unfavourable effect, because the company providing the capsules is a reliable producer of food supplements.
The serious adverse event and three minor adverse events reported were judged to be unrelated to the treatment.
Strengths and weaknesses
Probably due to the heterogeneity of the sample in terms of bowel habit predominance, we cannot point to aggravation of a specific symptom. But all outcomes, both the primary outcome (treatment preference) and the secondary outcomes (number of weeks with satisfactory relief of symptoms and IBS sum score), show the same unfavourable direction for active treatment. This strengthens the internal validity, but a type I error cannot be excluded.
With an IBS sum score difference of 2 chosen as clinically significant, the number needed to harm was 3.7. A 2-point difference on a scale with a range of 15 means 13%, and a change of 10% is often regarded as significant on such scales. However, the IBS sum score was not validated for responsiveness and clinically significant differences. Considering that the mean IBS sum score in the run-in period was 5.97, the score difference of 2 chosen as clinically significant might be rather high.
In this study the mean age of subjects was 50 years and the proportions of subjects with diarrhoea predominant, constipation predominant, and alternating IBS were 38%, 6%, and 56%, respectively. In corresponding studies the participants were younger and the proportions of subjects in the subgroups were more balanced [1, 25]. The older age and the somewhat different distribution of subgroups in our study raise the question of external validity. A beneficial effect in younger subjects, in subjects with more or less symptoms compared with our participants, in subgroups of subjects (such as constipation predominant), or in populations with other dietary habits and gut microflora cannot be excluded. Furthermore, a longer period of intervention would have strengthened the internal validity, but increased the drop-out rates.
The advantage of the crossover design used in this study is the increase in power of within-participant comparisons, and thus its requirement for fewer participants. For ethical reasons, the number of participants and the study period should be reduced as much as possible in a phase II study like this one. The design is fitted for stable chronic diseases . Although IBS is a fluctuating disease, Figure 2 shows that the prerequisites for the use of this design were fulfilled. The IBS sum scores in the two periods were not significantly different. The detection of L. plantarum MF1298 in one faecal sample at the end of the washout period indicates that the washout period was too short in this subject. However, because the amount of L. plantarum MF1298 detected was small and the recording of symptoms took place in the last week of the three-week treatment period, the possibility for a carryover effect is negligible. In summary, the crossover design turned out to be appropriate.