Discussion
Both fructose and lactose intolerance as shown by breath testing were common in this large group of FGID patients, one third of whom had an overlap of both intolerances. The important question is whether intolerances are an underlying mechanism or an epiphenomenon in FGID. The prevalence of intolerances was similar across all major types of FGID, except IBS-C. As the different FGID phenotypes were not related to distinct distributions of the intolerances, a causal relationship between FGID and the intolerances would have to be explained by divergent host responses to the saccharide ingestion. Such potential host factors include the enteric microbiome, intestinal permeability, nervous system and immune responses, all of which are interrelated and differ between FGID and controls. Hypersensitivity to distension and to ingested nutrients has been shown in different FGID, but we are not aware of comparisons between subgroups, except in IBS-C, where fermentation processes may be influenced by a prolonged transit time.
The induction of symptoms (intolerance) following sugar ingestion appears to be more relevant than malabsorption per se in FGID, as demonstrated by the following observations. Patients' main clinical GI and non-GI symptoms were significantly reproduced during breath testing, but there was no association with the markers of malabsorption. Malabsorption is similarly common after fructose or lactose loading in IBS and controls, but symptom induction is much higher in IBS. Furthermore, in patients with intolerance, effective symptom relief with dietary adaptation was independent of the presence of malabsorption. The relationship between malabsorption and intolerance therefore appears to be indirect. Intolerance without evidence of malabsorption occurred in approximately 20% of our patients and in 18–40% in earlier studies, whereas malabsorption rarely existed without intolerance. Potential explanations for this discrepancy are a microbiome not producing measurable levels of hydrogen or methane in some individuals or an alternative mechanism for symptom generation, such as increased chemosensitivity to fermentation products. Increased production of sulphide from carbohydrate fermentation has been shown in IBS. It should be noted that the relationship between symptom induction, and therefore intolerance rates, and malabsorption is dependent on the choice of symptom scoring, with a positive association between the number and intensity of induced symptoms and the percentage of positive hydrogen breath tests following lactose having been shown. In this study, a sensitive symptom index reflecting both intensity and symptom number was used very similar to the 9-item validated score used by Choi et al., with intolerance rates not greater than in other large comparative studies in Caucasians for lactose – we are not aware of any similar large studies for fructose intolerance. Only a small minority of these referred patients were able to correctly identify their food intolerance before the provocation testing, but this undoubtedly also partly reflects a referral bias to a GI practice.
Dietary modification based on fructose and lactose intolerance testing was clearly beneficial. Over 80% of all FGID patients attained adequate global symptom relief and average relief was 7 on the 10-point symptom scale. Adequate symptom relief was lower at about 50% in patients with constipation. These data are consistent with an earlier study where 85% of 48 IBS patients compliant with a similar diet reported an improvement of >5 on the 10-point scale for pooled symptoms after 14 months and also a reduced response rate in constipated patients. The impressive responses in this and smaller open trials in FGID are confirmed by a double-blinded, placebo-controlled fructose and fructan challenge study. Psychological and disease fluctuation components in our response rates could only be assessed in a blinded and placebo-controlled dietary study, which was not feasible in a study of this scale due to the inherent complexity.
Non-GI co-morbidity was common in FGID, especially in the central nervous, musculoskeletal and atopic categories. An elevated prevalence of non-GI functional syndromes, including fibromyalgia, migraine and chronic fatigue, is reported in FGID. Over 50% of our patients reported CNS symptoms, with fatigue being most frequent. Tiredness has previously been associated with a high FODMAP diet in IBS patients. Non-GI symptoms in FGID patients progressively increased with the number of experienced intolerances. The mechanisms underlying the association between GI and non-GI symptoms were not investigated, but they may be central (e.g. somatisation, central sensory processing), peripheral (e.g. afferent sensory sensitisation) or both (e.g. neuro-immune activation, disseminated abnormality in transporter proteins). The spectrum of non-GI symptoms was similar in fructose and lactose intolerances, except for more joint pain, prandial rhinitis and reactions to cosmetics with fructose intolerance. Earlier, small studies demonstrated increased depression, lethargy, decreased plasma tryptophan and various trace elements in fructose intolerance and a range of non-GI symptoms in lactose intolerance, including cardiac arrhythmias, musculoskeletal, atopic and nervous system symptoms. Intriguing potential links include toxic metabolites produced by anaerobic digestion, accumulation of advanced glycation end-products, abnormalities in fructose transporter proteins, increased intestinal permeability, actions of the phlorizin hydrolase moiety of lactase and enzyme homologies with inflammatory mediators, none of which have been confirmed. The short-chain fatty acids derived from colonic fermentation of fructose and lactose are similar. Further evaluation with specific tools is clearly warranted.
The epidemiology of fructose intolerance is poorly characterised. In our FGID patients, fructose was more common than lactose intolerance and more frequent in Northern than Southern Europeans, in contrast to lactose intolerance. Intolerances were more prevalent in females, significantly so for lactose, but confirmation is required in a population-based study. Interestingly, the overall prevalence of intolerances was higher in all FGID patients than in the more narrowly defined Rome III FGID subgroups. Consequently, studies employing Rome III criteria are likely to underestimate the overall relevance of intolerances.
The optimal breath test methodology remains unclear despite attempts at standardisation. In this study, average maximum symptom scores were attained 15 min earlier following fructose than lactose. Using the popular 3-h test duration, 16% of fructose and 23% of lactose intolerance results would have changed from positive to negative. H2 and CH4 concentrations peaked over 50 min earlier following fructose than lactose, and after 3 h in approximately 10% and 35% of tests respectively. The earlier exhaled gas and symptom peaks with fructose confirm an earlier report and may be due to differences in the location of absorption/digestion or of bacterial metabolism. The definitions of the gas concentration thresholds for a positive breath test are the subject of debate, demonstrably affecting test interpretation. We used the most widely reported dose of lactose (50 g) and a fructose dose (35 g) in the range of daily consumption in Europe, inducing GI symptoms in <10% of controls, and allowing comparison with seminal FODMAP studies. Both of these sugar doses are higher than the average amounts of free saccharide ingested during a single meal in normal life, but are used as provocation tests to identify patients likely to benefit from dietary manipulation. On the basis of the above, we recommend a 5-h test duration and, in case of discrepancy between symptom and malabsorption data, to base the final test interpretation on the intolerance results. This recommendation will result in effective treatment of 20% more patients compared with current practice in centres where clinical decisions are based exclusively on malabsorption. Furthermore, the correlation between clinical symptoms and those experienced during breath testing is an important confirmation for the patient. A breath collection duration longer than 3 h has been recommended in previous studies for improved diagnostics of lactose intolerance.
This study provides no corroboration for an association between exhaled methane and constipation using any of the common CH4 threshold definitions. Diarrhoea was more common than constipation in patients with elevated breath CH4. Conversely, increased CH4 was equally prevalent in constipation and diarrhoea. Furthermore, in IBS-C, breath tests were significantly more frequently CH4 negative than positive. This discrepancy with previous publications may be explained by variations in the definition of constipation (stool characteristics, transit, manometric changes, global evaluation), test methodology (CH4 threshold defined by very low baseline or peak concentrations, test duration, type of sugar) or patient selection (IBS, functional constipation, proportion with diarrhoea). Further study is required given this wide heterogeneity and the positive associations described mainly in smaller subgroups.
The relationship and overlap between small intestinal bacterial overgrowth (SIBO) and sugar malabsorption is extensive and it is acknowledged that currently a reliable distinction is not possible. Indeed, both conditions are intricately interrelated and the fermentation of sugars reaching bacteria in the colon (malabsorption) or in the small intestine (SIBO) can register similarly in breath tests, depending on the underlying rapidity of intestinal transit and the composition of the microbiome, among other factors. Different constructs for a distinction have been presented, such as early hydrogen peaks, gas thresholds and different substrates, but a true distinction remains elusive and we have chosen not to attempt one in our study. However, for a comparison with other papers, we have included the rather low percentage of patients with early rises in breath hydrogen excretion.
Study limitations are the absence of a blinded, placebo-controlled design with healthy controls. The choice of an inert placebo in this group of sensitive patients is challenging. Healthy controls were not included in this study aiming at comparisons within FGID. Comparisons with controls have been reported previously. An overlap between FGID subgroups is inevitable in all related studies. We chose the most prominent FGID for classifying each patient. Advantages of our single-centre study are the large number of successive patients with different FGID tested for both intolerance and malabsorption and treated in standardised fashion. Meta-analyses in this field are problematic due to the inherent variability in patient selection and in testing and treatment procedures.
In conclusion, fructose and lactose intolerance are common and frequently overlap in FGID, with no differences in prevalence between subgroups. Non-GI symptoms are more common in FGID patients with intolerances. Effective symptom relief is achieved with standardised dietary adaptation. Clinical FGID symptoms correlate with the symptoms induced during testing and not with malabsorption, which consequently does not appear to be the main driver of symptoms. Mechanisms relating to carbohydrate intolerance and accompanying non-GI symptoms are of special interest for future research.