Mixing of the Intestinal Content and Variations ofFermentation Capacity Do Not Affect the Results ofHydrogen Breath TestMichele Di Stefano, M.D., Emanuela Miceli, M.D., Simona Malservisi, M.D., Antonio Missanelli, M.D.,Alessandra Strocchi, M.D., and Gino Roberto Corazza, M.D.Gastroenterology Unit, Istituto di Ricovero e Cura a Carattere Scientifico “S. Matteo” Hospital, Universityof Pavia, Pavia, Italy

OBJECTIVES: Although the hydrogen (H2) breath test hasbeen in use for many years for diagnosis of sugar malab-sorption, research is still underway to improve its diagnosticaccuracy. In this study, we investigated whether possibleconfusing factors caused by the ingestion of the test solutionitself (such as the delivery to the colon of other fermentablesubstrates pre-existing in the small bowel lumen, the releaseof preformed H2 trapped in the feces, or differences in thefermenting capacity of the colonic bacteria) may interferewith the increase of breath H2 concentration, an expressionof malabsorption of the test substrate.

METHODS: In 25 patients with untreated celiac disease and23 sex- and age-matched healthy volunteers, breath H2

excretion was measured after ingestion of a 250-ml solutioncontaining sorbitol, a poorly absorbed alcohol sugar. On 2other separate days, 12 randomly selected subjects in eachgroup underwent breath H2 excretion measurement afteringestion of 250 ml of a sugar free, nonabsorbable electro-lyte solution and 250 ml of a solution containing lactulose,a nonabsorbable disaccharide.

RESULTS: After sorbitol ingestion, celiac disease patientsshowed a significantly higher breath H2 excretion than didhealthy volunteers. Otherwise, breath H2 responses to elec-trolyte solution and lactulose showed no difference betweenthe two groups of subjects.

CONCLUSIONS: In a group of patients with sugar malabsorp-tion, increased breath H2 excretion does reflect malabsorp-tion. The washout or the mixing of the intestinal content orintergroup difference of fermenting activity of the colonicbacteria do not represent interfering factors and do notmodify the accuracy of the H2 breath test in day-to-dayclinical practice. (Am J Gastroenterol 2003;98:1584–1587.© 2003 by Am. Coll. of Gastroenterology)

INTRODUCTION

The hydrogen (H2) breath test is a simple, noninvasive,repeatable, inexpensive test that is commonly used for di-agnosis of sugar malabsorption (1). Although its introduc-tion in clinical practice dates back almost 30 yr, there is still

concern regarding the diagnostic accuracy of the H2 breathtest because of incomplete knowledge of the pathophysiol-ogy of consumption and of intestinal gas production (2). Theincreased production of intestinal gas that occurs when amalabsorbed carbohydrate reaches intestinal flora is de-tected by increased gas excretion in breath. This pathway isconsequent to absorption at the intestinal level of an amountof gas that is then vehiculated through the blood vessels tothe lung for breath excretion. Several steps of this phenom-enon still need to be clarified and are the objects of debate.On the basis of an early article by Levitt et al. (3), it hasalways been considered that a constant amount of gas (about14%) reaches the blood vessels, but there is also evidencethat gas absorption may be dependent on the rate of itsproduction (4). The role of methane (CH4) production in thephysiology of intestinal gas and, therefore, in the pathophys-iology of sugar malabsorption is largely unknown. Even ifcarbohydrate malabsorption can be detected in a subgroupof patients through the detection of an isolated increase inbreath CH4 excretion (5), and perhaps the combined eval-uation of breath CH4 and H2 may improve diagnosis (6),accurate diagnostic criteria based on breath CH4 excretionare not available, yet. Similarly, little information is avail-able about the role of other intestinal gases such as sulfide(7), and therefore we do not know whether their evaluationmay improve diagnosis. Moreover, practical aspects of thetest are not yet standardized: for example, carbohydrateload, duration of the test, timing of sampling, and positivitycriteria (8). So, it is evident that a lot of work is still neededto understand what happens in the human intestine aftercarbohydrate loads and, as a consequence, which is the bestway to perform the H2 breath test.

The need for a pretest dinner meal free of fiber and wheatis suggested in a previous article by Anderson et al. (9)demonstrating the incomplete absorption of carbohydratesubstrates of different foods containing wheat flour. Thissimple measure allows reduction of fasting breath H2 levelson the day of the test by reducing the ingestion of nonab-sorbable and fermentable substrates. In addition, other fac-tors could theoretically interfere with the diagnostic accu-racy of the test. Among them, the ingestion of the sugar test

THE AMERICAN JOURNAL OF GASTROENTEROLOGY Vol. 98, No. 7, 2003© 2003 by Am. Coll. of Gastroenterology ISSN 0002-9270/03/$30.00Published by Elsevier Inc. doi:10.1016/S0002-9270(03)00373-3

solution by itself may trigger intraluminal mechanisms thatcause an increase in the H2 breath excretion unrelated to thetest substrate fermentation. On these grounds, the ileal mo-tor response to the ingestion of the test solution and theconsequent delivery to the colon of fermentable substratespre-existing in the small bowel lumen (10), and the releaseof preformed H2 trapped in the feces (11), could induce afalse positive test result. Mere intergroup differences in thefermenting capacity of the colonic bacteria can represent afurther source of reduction of the test accuracy.

In an attempt to clarify some potential pitfalls affectingthe accuracy of the test, we checked to what extent themixing of intestinal content and the variation in the colonicfermentation capacity may represent, respectively, causes ofa reduction in specificity and sensitivity of the H2 breath testin the diagnosis of sugar malabsorption.

MATERIALS AND METHODS

A total of 25 patients including 13 women and 12 men witha mean age of 35 yr (range 20–55 yr) with untreated celiacdisease (CD) diagnosed on the basis of malabsorption symp-toms, and 23 sex- and age-matched healthy volunteers, tookpart in the study. The decision to enroll CD patients wasbased on the need to study a group of subjects whoseintestinal malabsorption was established. This group of pa-tients allowed us to study the H2 breath excretion responseto the administration of different substrates when malab-sorption is present in comparison to normal conditions.Moreover, our preliminary results indicated that samples ofjejunal juice of CD patients incubated in an in vitro fecalsystem with a stool aliquot of H2 and CH4 producing healthyvolunteer produce a significant amount of H2 and CH4,suggesting the presence of pre-existing substrates in thelumen (12).

All subjects underwent breath H2 measurement after ad-ministration of 250 ml of a solution containing 5 g ofsorbitol, a poorly absorbed alcohol sugar (13). The dose andthe concentration of the solution was chosen on the basis ofprevious results (14) demonstrating that the vast majority ofnormal subjects show no malabsorption after the ingestionof a 2% sorbitol solution. Breath samples were taken atfasting and every 30 min for a 4-h period. The test wasconsidered positive if an increase in breath H2 excretion�20 ppm over fasting value was evident (14–16).

In addition, on 2 separate days, 12 CD patients and 12healthy volunteers (both groups randomly selected) under-went two further evaluations: 1) breath H2 measurementafter the administration of 250 ml of a sugar free, nonab-sorbable, electrolyte solution (Colyte; Reed & Carnrick,Piscataway, NJ) to exclude the possible interfering effect ofthe delivery to the colon of pre-existing fermentable sub-strates (10), or of the release of preformed H2 entrapped inthe feces (11); 2) breath H2 measurement after the admin-istration of a 250 ml solution containing 10 g of lactulose,a nonabsorbable disaccharide, to exclude the possible inter-

fering effect of differences between the two groups in thebacterial fermenting capacity. On both occasions, breathsamples were collected at fasting and every 30 min afteringestion of the test substrate for a 4-h period.

For each subject, the cumulative breath H2 excretion,which was evaluated by calculating the area under thetime-concentration curve (17), and the peak of breath H2

excretion (i.e., the maximal breath H2 excretion increaseover baseline) were recorded. Breath CH4 excretion, themost important metabolic pathway of H2 consumption (11),was also measured. Subjects with mean breath CH4 excre-tion �5 ppm were considered as producers of CH4 (5).

The day before each breath test procedure, a pretestdinner meal consisting only of rice, meat, and olive oil wasprescribed (18, 19). Ingestion of a dinner containing carbo-hydrates in the form of rice flour only on the eveningpreceding the test reliably reduces fasting breath H2 levels inhealthy volunteers. This meal was then followed by a 12-hfasting period before the test. Breath testing started between8:30 and 9:30 AM, after thoroughly washing the mouth with40 ml of 1% chlorhexidine solution (20). Smoking (21) andphysical exercise (22) were not allowed for 1 h before andthroughout the test.

Sampling of alveolar air was performed by means of acommercial device (Gasampler; Quintron, Milwaukee, WI)that allows the first 500 ml of dead space air to be separatedand discarded and the remaining 700 ml of end-alveolar airto be collected in a gas-tight bag. Subjects were instructed toavoid deep inspiration and not to hyperventilate before exha-lation. The variability of H2 determinations using this collect-ing method was shown to be approximately 25% (23).

A gas chromatograph dedicated to the detection of H2 inair samples was used for breath sample analysis (model DP;Quintron). The accuracy of the detector was � 2 ppm of H2

with a linear response range between 2 and 150 ppm of H2.

RESULTS

All of the celiac patients proved to have malabsorption of5 g of sorbitol, suggesting that none of them could beconsidered as nonproducers of H2 (24). In the healthy vol-unteer group, none of the subjects malabsorbed the admin-istered dose of sorbitol. As expected, cumulative H2 excre-tion after sorbitol was higher in CD patients than in thehealthy volunteers (Fig. 1). In particular, CD patientsshowed an average area under time–H2 concentration curve(6435 � 148 ppm/min) and maximal breath H2 increaseover baseline (44 � 6 ppm) that was significantly higher (p� 0.001 for both comparisons) than those of healthy vol-unteers (2054 � 99 ppm/min and 13 � 3 ppm, respectively).

Figure 2 shows the breath H2 excretion response to Colyteand lactulose administration in CD patients and healthyvolunteers. No significant differences were evident in termsof cumulative breath H2 excretion after lactulose (2646 �188 ppm/min vs 2265 � 154 ppm/min, respectively) and

1585AJG – July, 2003 Intraluminal Factors and Hydrogen Breath Test

after Colyte (5824 � 144 ppm/min vs 4845 � 112 ppm/min)between these two groups.

With regard to breath CH4 excretion, the prevalence ofbreath CH4 excretion was similar between patients (three of25; 12%) and controls (three of 23; 13%). Because ofsample scantiness no statistical evaluation of results wasallowed.

DISCUSSION

The first description of the presence of hydrogen gas tracesin expired air dates back to the early 1960s, in an article byNielsen (25). Several years later, Calloway suggested theuse of breath H2 excretion in expired air as a possiblemethod for monitoring intestinal gas production (26). How-ever, it was the great work of Levitt (3) and Bond (27) thatmade it possible to introduce the H2 breath test into clinicalpractice. Because it is an inexpensive, noninvasive andrepeatable assay, the H2 breath test was highly successfulboth in clinical and in research settings. It is performed inthe vast majority of gastroenterology units, and several

hundreds of papers dealing with this test have been pub-lished in the world literature during the last 40 yr. However,to improve this method, a series of both methodological andpathophysiological aspects still require clarification. Be-sides the presence of several methodological differences inthe day-by-day practical execution of the test, includingamong gastroenterological reference centers, for which astandardization is needed (8), it is crucial to analyze themajor source of variability represented by possible luminalfactors that might interfere with the diagnostic accuracy ofthe test.

On these grounds, the aim of the present study was toevaluate whether increased breath H2 excretion after theingestion of the test substrate solution could be caused byfactors not related to the fermentation of the substrate. Inparticular, our aim was to exclude the possibility that thedelivery to the colon of other fermentable substrates pre-existing in the small bowel lumen or the release of pre-formed H2 trapped in the feces, or differences betweennormal and disease states in fermenting capacity of thecolonic bacteria, might represent interfering factors. In thisstudy, confirming earlier results (14), we have shown thatthe administration of 5 g of sorbitol allows a clear separationbetween CD patients and healthy volunteers. All of thepatients proved to have sorbitol malabsorption. In contrast,all of the healthy volunteers were able to absorb this sugar.

To analyze whether the delivery to the colon of otherfermentable substrates pre-existing in the small bowel lu-men or whether the release of preformed H2 trapped in thefeces may interfere with test results, we administered anamount of nonabsorbable, nonfermentable electrolyte solu-tion equivalent to the amount of the sorbitol solution, which,passing through the bowel, helps the crossing of any intralu-minal substrates present and induces a mixing of intestinalcontent. Therefore, proving the absence of increments of H2

breath excretion is equal to excluding an interfering role ofthese two factors. This evaluation did not show any differ-ence between CD patients with malabsorption versushealthy volunteers in terms of cumulative breath H2 excre-tion, indicating that mixing of intestinal contents does notrepresent an interfering factor that would be able to reducethe specificity of the H2 breath test in the diagnosis of sugarmalabsorption.

To analyze whether differences between normal and dis-ease states in fermenting capacity of the colonic bacteria(which theoretically can affect the sensitivity of the test)could represent interfering factors, we administered 20 g oflactulose, a nonabsorbable disaccharide that is certainlyfermentable in the colon, to evaluate the intestinal capacityof gas production and to compare this parameter among thesubjects. In this way, we checked the possibility that apositive test might be the result of a tendency to producegreater than normal amounts of gases. The results of thisevaluation showed that no difference was evident betweenpatients and healthy volunteers. In conclusion, we can rea-sonably believe that we have excluded intergroup differ-

Figure 1. Breath hydrogen excretion after the administration of 250ml of a solution containing 5 g of sorbitol in patients with celiacdisease (●) and in healthy volunteers (f).

Figure 2. Breath hydrogen excretion after the administration of(left) 250 ml of Colyte, a nonabsorbable, nonfermentable electro-lyte solution and (right) 250 ml of a solution containing 10 g oflactulose in patients with celiac disease (●) and in healthy volun-teers (f).

1586 Di Stefano et al. AJG – Vol. 98, No. 7, 2003

ences in gas production as interfering factors with the di-agnostic accuracy of the H2 breath test.

It should also be argued that one cause of false negativesis represented by the presence of a non–H2-producer status,and a previous article by our group of investigators showedthat approximately 14% of subjects do not produce H2

within 4 h after lactulose administration (24). Therefore,differences in intergroup colonic H2 production capacitymay represent a cause of reduction in diagnostic accuracy ofthe test. However, all patients enrolled in this present studyproved to be H2 producers and therefore this status did notinterfere with our results.

Another possible cause of false negative results is repre-sented by a different CH4 production capacity at the coloniclevel. CH4 production is considered to be the main meta-bolic pathway of H2 consumption (11): H2 produced duringcolonic fermentation may be condensed into CH4 mole-cules, thus reducing the amount of H2 excreted in breath.However, in the present study, the prevalence of CH4-producing patients was low and it was not possible toperform statistical tests.

In conclusion, on the basis of our data, a negative breathtest after sorbitol administration is truly indicative of thenonarrival of this carbohydrate in the colon and therefore ofa physiological absorption, whereas a positive test indicatescolonic fermentation and therefore the pathological malab-sorption of the carbohydrate solution administered. Finally,it is conceivable that mixing of the intestinal content or theintergroup differences in fermenting activity of the colonicbacteria do not modify the accuracy of the H2 breath test inday-to-day clinical practice.

ACKNOWLEDGMENT

This work was supported in part by a grant from IRCCS S.Matteo Hospital, Pavia, Italy.

Reprint requests and correspondence: Gino Roberto Corazza,M.D., Gastroenterology Unit, University of Pavia, IRCCS “S.Matteo” Hospital, Viale Golgi 19, 27100, Pavia, Italy.

Received May 14, 2002; accepted Jan. 27, 2003.

REFERENCES

1. Levitt MD, Donaldson RM. Use of respiratory hydrogen (H2)excretion to detect carbohydrate malabsorption. J Lab ClinMed 1970;75:937–45.

2. Strocchi A, Corazza GR, Gasbarrini G. Recent advances onhydrogen metabolism in man. Ital J Gastroenterol 1992;24:207–11.

3. Levitt MD. Production and excretion of hydrogen gas in man.N Engl J Med 1969;281:122–7.

4. King TS, Elia M, Hunter JO. Abnormal colonic fermentationin irritable bowel syndrome. Lancet 1998;352:1187–9.

5. Corazza GR, Benati G, Strocchi A, et al. The possible role ofbreath methane measurement in detecting carbohydrate mal-absorption. J Lab Clin Med 1994;124:695–700.

6. Rumessen JJ, Nordgaard-Andersen I, Gudmand-Hoyer E. Car-

bohydrate malabsorption: quantification by methane and hy-drogen breath test. Scand J Gastroenterol 1994;29:826–32.

7. Strocchi A, Furne J, Ellis C, Levitt MD. Methanogens out-compete sulphate reducing bacteria for H2 in the human colon.Gut 1994;35:1098–101.

8. Strocchi A, Corazza GR, and Tenue Study Group. Qualitycontrol study of H2 breath testing for the diagnosis of carbo-hydrate malabsorption in Italy. Ital J Gastroenterol Hepatol1997;29:122–7.

9. Anderson IH, Levine AS, Levitt MD. Incomplete absorptionof the carbohydrate in all-purpose wheat flour. N Engl J Med1981;304:891–2.

10. Kerlin P, Zinsmeister A, Phillips S. Motor response to food ofthe ileum, proximal colon, and distal colon in healthy humans.Gastroenterology 1983;84:762–70.

11. Strocchi A, Levitt MD. Factors affecting hydrogen productionand consumption by human fecal flora. The critical roles ofhydrogen tension and methanogenesis. J Clin Invest 1992;89:1304–11.

12. Di Stefano M, Strocchi A, Veneto G, Corazza GR. Endoge-nous substrates are responsible for raised fasting breath H2

levels in coeliac disease. Gastroenterology 2000;118:A366.13. Jain NK, Rosemberg DB, Ulahannan MJ, et al. Sorbitol intol-

erance in adults. Am J Gastroenterol 1985;80:678–81.14. Corazza GR, Strocchi A, Rossi R, et al. Sorbitol malabsorption

in normal volunteers and in patients with coeliac disease. Gut1988;29:44–8.

15. Hyams JS. Sorbitol intolerance: An unappreciated cause offunctional gastrointestinal complaints. Gastroenterology 1983;84:30–3.

16. Hasler W, Tatum D, Owyang C. Fructose and sorbitol mal-absorption in normal volunteers. Gastroenterology 1986;90:1453.

17. Kotler OP, Holt PR, Rosensweig NS. Modification of thebreath hydrogen test increased sensitivity for the detection ofcarbohydrate malabsorption. J Lab Clin Med 1982;100:798–805.

18. Corazza GR, Strocchi A, Gasbarrini G. Fasting breath hydro-gen in celiac disease. Gastroenterology 1987;93:53–8.

19. Perman JA, Modier S, Barr RG, et al. Fasting breath hydrogenconcentration: normal values and clinical application. Gastro-enterology 1984;87:1358–63.

20. Thompson DG, O’Brien JD, Hardie JM. lnfluence of theoropharyngeal microflora on the measurements of exhaledbreath hydrogen. Gastroenterology 1986;191:853–60.

21. Rosenthal A, Solomons NW. Time-course of cigarette smokecontamination of clinical hydrogen breath-analysis tests. ClinChem 1983;29:1980–1.

22. Payne DL, Welsh JD, Claypool PL. Breath hydrogen (H2)response to carbohydrate malabsorption after exercise. J LabClin Med 1983;102:147–50.

23. Strocchi A, Ellis C, Levitt MD. Reproducibility of measure-ments of trace gas concentrations in expired air. Gastroenter-ology 1991;101:175–9.

24. Corazza GR, Strocchi A, Sorge M, et al. Prevalence andconsistency of low breath H2 excretion following lactuloseingestion. Possible implications for the clinical use of the H2

breath test. Dig Dis Sci 1993;38:2010–6.25. Nielsen JP. Trace constituents in breath as related to flatu-

lence. Proceedings of Fifth Annual Dry Bean Research Con-ference, Western Regional Research Laboratory, United StatesDepartment of Agriculture, Denver, CO, 1961:49.

26. Calloway DH. Respiratory hydrogen and methane as affectedby consumption of gas-forming foods. Gastroenterology 1966;51:383–9.

27. Bond JH, Levitt MD. Use of pulmonary hydrogen measure-ments to quantitate carbohydrate absorption. Study of partiallygastrectomized patients. J Clin Invest 1972;51:1219–25.

1587AJG – July, 2003 Intraluminal Factors and Hydrogen Breath Test