We conducted a two-group, parallel, randomized, double-blind controlled trial of rifaximin for poorly responsive celiac disease [Chang et al.
] to determine whether rifaximin would improve gastrointestinal symptoms when measured using the Gastrointestinal Symptom Rating Scale (GSRS), a validated seven-point questionnaire that evaluates five symptom areas: abdominal pain, reflux, indigestion, diarrhea, and constipation [Midhagen and Hallert, 2003
; Svedlund et al. 1988
]. Lactulose–hydrogen breath test improvement was measured as a secondary outcome. Our study included patients who were 18 years or older with a biopsy-proven diagnosis of celiac disease, and persistent symptoms while on a GFD. Potential causes of persistent gastrointestinal symptoms, including infectious, inflammatory, medication-related, and other similar causes were excluded; a complete list of exclusion criteria can be found in our full paper [Chang et al. 2011
]. Patients were randomized to either treatment with rifaximin 1200
mg daily for 10 days (n
25) or placebo (n
25) and were assessed at weeks 0, 2, and 12 using the GSRS and breath testing. Owing to the lack of standardized diagnostic criteria for what constituted a positive breath test and the fact that patients with celiac disease can have elevated baseline hydrogen levels, [Corazza et al. 1987
] we chose to evaluate our results using two different definitions adapted from the literature. Criteria 1 was defined as a rise in hydrogen of ≥20
ppm within 100 minutes [Sharara et al. 2006
; Pimentel et al. 2003
] and criteria 2 was defined using a more traditional double-peak interpretation of two peaks ≥20
ppm over the baseline hydrogen level within 100 minutes [Walters and Vanner, 2005
; Pimentel et al. 2000
GSRS scores were unaffected by treatment with rifaximin, regardless of baseline breath test results (). In a multivariable regression model, the duration of patients’ gastrointestinal symptoms significantly predicted their overall GSRS scores (estimate 0.029, p
0.006), whereas the duration that a patient had been on a GFD predicted lowered GSRS reflux and constipation subscores. This suggests that a GFD provides some incomplete degree of gastrointestinal symptom improvement in patients with poorly responsive celiac disease. In terms of the breath test results, our prevalence of SIBO varied depending on the criteria used to define a positive hydrogen breath, ranging from 8% to 55% of patients at baseline, and intermittently present in 12–28%. According to criteria 1, SIBO was present in 55% at baseline, intermittently present in 28% given placebo, and in 28% given rifaximin. Using criteria 2, SIBO was not found as commonly as according to criteria 1, in only 8% of patients at baseline, while intermittently present in 20% given placebo and in 12% given rifaximin, over 12 weeks. There was no difference in the prevalence of SIBO between placebo and treatment groups at weeks 2 and 12 ().
Summary of Gastrointestinal Symptom Rating Scale (GSRS) scores and hydrogen breath test results according to placebo and rifaximin groups.
There were a number of limitations to our study. Primarily, the study population was small and suffered from a relatively high rate of dropouts (18%, including one patient that left after randomization, but prior to the start of the study), although poorly responsive celiac disease itself is an uncommon entity. Also, breath testing can vary widely in its sensitivity and specificity in diagnosing SIBO, which could have limited the accuracy of our findings. In addition, we did not assess for methane production. However, our study clearly demonstrated that an empiric trial of rifaximin is unlikely to improve gastrointestinal symptoms in patients with poorly responsive celiac disease, regardless of breath testing results. This may be in part due to an overestimation of the prevalence of SIBO in celiac disease, for which the use of antibiotics for gastrointestinal symptoms would obviously be ineffective. Our patients had lower rates of SIBO than previously reported, particularly when using criteria 2, which is more consistent with prevalence rates seen when using intestinal aspirate cultures to diagnose SIBO (8% in our study compared with 11% when using aspirate cultures) [Rubio-Tapia et al. 2009
]. It may also be that patients with celiac disease have more resistant SIBO or multiple etiologies for their gastrointestinal symptoms, such that the treatment of SIBO alone would not be sufficient for symptom relief. It is very likely that a small proportion of patients with bloating and abdominal pain have intussusception and other alternate etiologies instead, as seen in our recent study [Gonda et al. 2010
], and would benefit from evaluation with ultrasound, CT, or MR enterography. Unfortunately, we did not systematically incorporate imaging findings in our study and were unable to draw any conclusions regarding this. Of the patients who did test positively for SIBO, we found that it was both intermittent and difficult to eradicate with a conventional antibiotic course. In addition to exploring other causes of persistent gastrointestinal symptoms, one future approach for the treatment of SIBO in patients with celiac disease might be to combine rifaximin with another antibiotic, either concomitantly or sequentially.