In young infants, excessive intestinal gas is frequently thought to be the cause of abdominal discomfort (1). However, except for hydrogen (H 2 ), little is known about intestinal gases, their production, and their physiologic and pathologic significance. The quantitatively most important volumes of flatus are carbon dioxide (CO 2 ), H 2 , and methane (CH 4 ) (2). H 2 is not metabolized and is quantitatively excreted in expired air. Measurement of breath H 2 has been used extensively for all age groups as a quantitative indicator of colonic fermentation (3–6). In addition, colonic bacteria produce a variety of other volatiles that are present in only trace concentrations in flatus. One subset of these trace gases, the sulfur-containing gases such as hydrogen sulfide (H 2 S), methanethiol (CH 3 SH), and dimethyl sulfide (CH 3 SCH 3 ), recently has received attention (7,8). These gases have been shown to be the major source of fecal malodor (9). Most important, these extremely toxic gases could irritate or damage the colonic mucosa. Roediger et al. (10) have postulated that excessive sulfide production could play a pathogenetic role in ulcerative colitis.
Although the physiology of the quantitatively important gases has been extensively studied, little is known about the factors that influence the release of the trace gases. Colonic bacteria release H 2 S and CH 3 SH during the process of metabolizing sulfur-containing compounds, which may be of dietary (e.g., sulfate) or endogenous (e.g., mucin) origin (11). The rate of production of these gases presumably is a function of the number (or activity) of various flora and the colonic availability of sulfur-containing substrates. In infants, different feeding modes (i.e., breast feeding, formula feeding) are associated with differences in the colonic microflora and provide different types and amounts of precursors (12,13). Thus, infants provide a unique opportunity for the study of the possible influence of dietary composition on sulfur gas production by feces.
In this pilot study, the objectives were to obtain initial data on gas production by infant feces, the effect of feeding mode on gas production, and the possible relation between gas production and stool characteristics. The influence of age was also studied. The production of H 2 , CO 2 , CH 4 , H 2 S, CH 3 SH, and CH 3 SCH 3 by the feces of breast-fed infants and infants fed milk-based or soy-based formula was studied in infants at approximately 1, 2, and 3 months of age.
The study used a longitudinal design whereby feces were collected from each infant on three occasions (i.e., at approximately 1, 2, and 3 months of age). Infants were either breast-fed or were fed one of two formulas. The study protocol was reviewed and approved by the University of Iowa Committee on Human Experimentation and informed written consent was obtained from both parents of each participating infant.
Subjects and feedings
Normal-term infants of either gender (birth weight >2500 g and gestational age >37 weeks) who had not received antibiotics since birth were enrolled before 1 month of age. Parents were requested to report the use of antibiotics, if any, during the study. Ten infants (four female, six male) were exclusively breast-fed at the time of enrollment; none of them received iron or other nutritional supplements. One of the infants was later reported to have received formula on four occasions, each time a single feeding. Because the formula feedings occurred more than 7 days before the next stool collection, data for this infant are included.
Eight infants were fed formula during the study. Before entering the study two of these infants were breast-fed for less than 2 weeks, and all infants were fed other formulas for at least a few days before starting on the study formulas. The study formulas were fed for at least 7 days before the first stool collection. Five infants (two female, three male) were fed a soy-based formula. The formula was provided in powder form and contained no carrageenan. Its protein (19.1 g/L) was isolated soy protein and its carbohydrate was corn syrup. It contained added L-methionine (52 mg/L) and taurine (85 mg/L). Three infants (one female, two male) were fed a milk-based formula that was also provided in powder form and contained no carrageenan. Its protein (14.4 g/L) was derived from cow milk solids and bovine whey protein concentrate and it contained added taurine (85 mg/L). It contained lactose as the sole carbohydrate. Neither formula contained nucleotides nor nonabsorbable carbohydrates and both formulas contained added ferrous sulfate and zinc sulfate. The final concentration of iron was approximately 11 mg/L and that of zinc was 6.6 mg/L in both formulas. Parents were asked to feed their children exclusively with the study formulas, which were provided free of charge, and not to provide any other foods.
Collection and incubation of fecal samples
Infants were brought to the Fomon Infant Nutrition Unit at the University of Iowa for collection of stool specimens during the weeks in which they reached 1, 2, and 3 months of age, respectively. Their actual ages ranged from 27 to 38 days, 54 to 81 days, and 83 to 113 days, respectively. Two of the breast-fed infants participated only at 1 month of age. To obtain the desired two fecal specimens, it was at times necessary for an infant to return on a second or even a third day during the same week. At times only one specimen was obtained, and in one case no specimen was obtained in spite of three attempts. For the collection of fresh fecal specimens, infants were placed on metabolic beds (14), and urine collection bags were placed over the genital area to avoid urine admixture to fecal samples. Stools were collected directly into heat-treated pans.
Within 15 minutes of passage, 5 to 10 g feces was transferred into a 750 mL preweighed gas-impermeable bag made of aluminum polyester (Quintron, Milwaukee, WI). When more than 10 g stool was obtained, the stool was divided into two portions and each transferred into a bag. After flushing the bag with argon for 2 minutes and leaving approximately 200 mL argon in the bag, the bag was sealed and incubated in a water bath at 37°C for 4 hours. After incubation the bags were placed in a refrigerator until later that day when they were shipped on ice by overnight carrier to the laboratory in Minneapolis.
For the day before and the day after a stool collection, parents were asked to complete a 24-hour stool record. They were provided with forms on which they marked for each stool the appropriate box for color (black/green, brown, or yellow), consistency (hard, firm/formed, soft/loose, or watery) and odor (normal/usual or foul). The consistency and odor of stools that were incubated for gas production were not recorded.
The total volume of gas evolved was determined by aspiration into a calibrated syringe. With the exception of CO 2 , which was measured using an infrared detector (Beckman LB-2, Fullerton, CA), the concentrations of gases were determined by gas chromatography. The sulfur-containing gases (H 2 S, CH 3 SH, and CH 3 SCH 3 ) were quantitated using an HP5890A gas chromatograph (GC; Hewlett-Packard, Palo Alto, CA) equipped with a flame photometric detector specific for sulfur. The HP5890A GC was equipped with a flame ionization detector for CH 4 determination and a thermal conductivity detector for oxygen analysis. H 2 was measured using a Beckman GC 72-5 (Fullerton, CA) equipped with a reduction detector (Trace Analytical RGD 2, Menlo Park, CA). Gas concentrations were determined via comparison of the peak area or height of the unknown versus that of authentic standards. After aspiration of the gas, the bags were dried to constant weight for determination of the dry weight of the fecal sample. Gas production was expressed as amount of gas per gram of dry stool.
Where stools were divided into two portions before incubation, or where more than one stool was obtained on the same day, values were averaged. Because the gas values were markedly skewed, values were transformed prior to statistical analysis using square root. The data were examined by repeated analysis of variance measures. When significant differences were found (P < 0.05, Fisher test), the Fisher least significant difference (LSD) test was used to compare means pair-wise. Data analyses were performed using SYSTAT statistical software (SYSTAT for the Macintosh, version 5.2, SYSTAT, Evanston, IL) and SAS (version 6.12, SAS Institute, Cary, NC). Data are presented as means with standard error.
Oxygen was detected in low concentrations (<1.0%) in eight samples but not in the remaining samples, which ruled out a significant leakage problem. Among duplicate samples (divided samples or two samples collected on the same day), agreement was generally good, with coefficients of variation ranging from 23% for CO 2 to 45% for H 2 S. CH 3 SCH 3 could not be detected in any sample.
Effect of feeding
The feces of breast-fed infants showed significantly (P < 0.001) greater release of H 2 than did the feces of infants from either formula group (Table 1, Fig. 1). However, the feces of infants fed soy-based formula produced more CH 4 than did the feces of infants in the other feeding groups (P = 0.001) (Fig. 2). No significant feeding-related differences in CO 2 production were present. The production of H 2 S by feces of breast-fed infants was significantly lower (P = 0.002) than that by feces of soy-fed infants (Fig. 3). The feces of some breast-fed infants actually produced little or no H 2 S. The fecal production of CH 3 SH (Fig. 4) was lower (P < 0.001) in breast-fed infants than in infants from either formula group. The fecal production of total sulfur gases (H 2 S plus CH 3 SH) was significantly lower in breast-fed infants than in soy-fed infants (P < 0.001) but was not significantly different between breast-fed and milk-fed infants.
Across all feeding groups and all ages, there was a significant inverse correlation between CH 3 SH and H 2 production (r = -0.357, P = 0.0024; data not shown) and a positive correlation between CH 4 and H 2 S production (r = 0.573, P < 0.001).
Effect of age
Although the extent of age-related changes was generally modest (Figs. 1–4), some changes reached statistical significance. In the feces of breast-fed infants, production of H 2 was borderline lower (P = 0.054) at 3 months compared with at 1 and 2 months of age. A significant (P < 0.005) age-related increase of CH 4 production was observed in soy-formula–fed infants. Production of H 2 S tended to increase with age in breast-fed infants. Production of CH 3 SH in the feces of soy-formula–fed infants increased significantly (P = 0.027) with age.
Data are summarized in Table 2. Breast-fed infants produced a significantly (P < 0.05) greater number of stools than did infants in either formula group. There was a significant decrease in stool frequency with age. The stools of soy-fed infants were recorded significantly (P < 0.001) more often as firm (43% of stools) than were stools of all other infants. Five infants (one breast-fed, two milk-fed, two soy-fed) had foul-smelling stools recorded on at least one occasion. On a percentage basis, stools of soy-fed infants were significantly (P < 0.01) more often recorded as foul smelling (25% of stools) than were stools of breast-fed infants (0%) or of milk formula-fed infants (11%).
Six of the stools studied for gas production were collected in proximity to a foul-smelling stool (i.e., where the records indicated a foul-smelling stool on the day before or on the day after the study stool was collected). Each of these six stools produced total sulfur gases in excess of 1000 nmol/g of stool. In contrast, without a foul-smelling stool in temporal vicinity, total sulfur gases in excess of 1000 nmol/g were observed only in 12 of 43 stools. Also, among breast-fed and milk-formula–fed infants, 6/9 stools of infants who ever had a foul-smelling stool produced total sulfur gases in excess of 750 nmol/g stool, whereas only 2/26 stools from infants who never had a foul-smelling stool produced this much total sulfur gases.
Gas production by feces of infants has not previously been quantitated. This pilot study was designed to generate initial data that would enable the design of more definitive studies of the relation between diet and gas production and, especially, the relation between infant behavior and gas production. Despite the limited objective and the small number of infants participating, some firm conclusions may be drawn. First, associations between diet and fecal gas production: 1) with breast feeding, there is low production of CH 3 SH and H 2 S and high production of H 2 ; 2) with feeding of milk-based formula, there is low production of CH 4 ; and 3) with feeding of soy-based formula, there is high production of CH 4 and H 2 S. Second, age-related changes in gas production were modest. Fecal H 2 production in breast-fed infants decreased somewhat with age and CH 4 and CH 3 SH production increased in soy-formula–fed infants. The lack of a major influence of age on fecal gas production suggests that a relatively static relationship between the colonic flora and the diet is established by 4 weeks of age.
The volume of gas released by a fecal sample reflects the end result of a complex interaction of a variety of factors. First, the production of a gas requires the presence of an adequate number of fecal organisms that are capable of carrying out the metabolic reactions that result in that gas. For some gases, only selective bacteria possess the required metabolic pathways. For example, CH 4 is produced mainly by one species of bacteria, Methanobrevibacter smithii(15). Also, only selective bacteria, the so-called sulfate-reducing bacteria, have the ability to convert sulfate to H 2 S (16). The fecal concentrations of these organisms have been shown to vary widely in healthy adults (17). In contrast, the feces of virtually all individuals contain large numbers of the various organisms capable of releasing H 2 during carbohydrate fermentation.
The bacteriological aspect of gas release is further complicated by the ability of some fecal bacteria to consume gases produced by other bacteria. For example, greater than 90% of the H 2 produced in the colon is consumed by other bacteria, and H 2 release by feces represents the net of production minus consumption (18). Thus, strictly speaking, what was measured in the present study was gas release, not gas production. Lastly, the metabolism of bacteria that play no direct role in release or consumption of gas can influence the release of gas indirectly. For example, acidification of the colon via the bacterial production of organic acids by one group of organisms could affect gas release by other bacteria (19).
Beside the microbial make-up of the colon, the availability of appropriate substrates is the other crucial factor determining the production of the various gases. H 2 production in the colon has uniformly been shown to be a function of the malabsorption of carbohydrates, which provide the substrate for H 2 -producing fermentation reactions. The high release of H 2 by feces of breast-fed infants presumably reflects a greater availability of carbohydrates to the fecal flora. This carbohydrate could be lactose, which is present in human milk in a concentration of approximately 70 g/L. The concentration of lactose in the milk-based formula was approximately the same as in breast milk, yet fecal H 2 production was much less. Breast milk contains, in addition to lactose, considerable quantities (up to 18 g/L) of complex oligosaccharides (20), which are almost totally absent from cow milk. The chemical structure of human-milk oligosaccharides suggests that they are unlikely to be hydrolyzed by the digestive enzymes. Using the breath H 2 test, Brand-Miller et al. (21) compared lactulose with oligosaccharides extracted from the milk of the infant's own mother. The high H 2 excretion observed after feeding the oligosaccharides indicated that they escape digestion and absorption in the small intestine and undergo fermentation in the colon.
The most striking relationship between diet and fecal gas production observed in the present study was that between the low production of sulfur gases by the feces of breast-fed infants, compared with infants fed either type of formula. The difference was most pronounced for release of CH 3 SH, but H 2 S release was also low in the feces of some breast-fed infants. A plausible explanation lies in the availability of suitable substrates. Breast milk provides approximately 3.21 mmol/L of cysteine, methionine, and taurine to the 1-month-old infant, and approximately 2.87 mmol/L to the 3-month-old infant (22,23); approximately one fourth of this (<0.8 mmol/L) is provided by methionine. In contrast, both formulas provided greater amounts of sulfur-containing amino acids and also contained added sulfate. The milk-based formula provided 3.637 mmol/L total sulfur-containing amino acids, of which approximately one half (1.775 mmol/L) was provided by methionine. In addition, the formula contained 0.254 mmol sulfate in the form of zinc sulfate and ferrous sulfate. The soy-based formula, which contained 0.348 mmol/L added methionine, provided 4.734 mmol/L sulfur-containing amino acids, of which methionine accounted for 2.010 mmol/L. In addition, the soy-based formula contained 0.309 mmol/L sulfate in the form of zinc sulfate and ferrous sulfate. Thus, the formulas contained more than twice the amount of methionine present in breast milk and they contained sulfate additives.
Production of CH 3 SH appears to be largely dependent upon availability of methionine, the favored substrate for bacterial CH 3 SH production. In studies with human fecal homogenates the release of CH 3 SH was greatly enhanced by the addition of methionine (24). In contrast, H 2 S can be produced from a variety of sulfur-containing substrates including dietary compounds, such as sulfate and sulfur-containing amino acids (cysteine, taurine, methionine), and endogenous compounds, such as mucin and taurocholic acid (11). Human fecal homogenates release far more H 2 S when supplemented with appropriate amino acids than when supplemented with sulfate (11).
An additional explanation for the low production of CH 3 SH by feces of breast-fed infants may be the presence of unabsorbed carbohydrate. In fecal homogenates CH 3 SH release is inversely proportional to the amount of carbohydrate in the medium (25). In the present study H 2 release, an indicator of carbohydrate fermentation, was high in the feces of breast-fed infants and there was a significant inverse relationship between H 2 and CH 3 SH release. Thus, unabsorbed carbohydrate in the feces of breast-fed infants may have been responsible for inhibition of CH 3 SH production.
The increased release of CH 3 SH from the feces of formula-fed infants provides at least a partial explanation for the difference in fecal odor between breast-fed and formula-fed infants. The odor of CH 3 SH is noxious and more intense per unit of gas than the odor of H 2 S. In the present study there was an association between the occurrence of malodorous stools and fecal release of sulfur gases.
The increased release of CH 4 observed with the soy-based formula is of some interest because approximately 35% of adults possess high fecal concentrations of methanogens, whereas the other 65% have very low concentrations of these organisms (26). Attempts to link the diet of adults to the presence of absence of methanogens have been unsuccessful (e.g., spouses who presumably consume relatively similar diets show no concordance for CH 4 production). It has been shown, however, that there is a strong concordance for the presence (or absence) of CH 4 production among siblings (27). It seems possible that dietary exposure during infancy, which would tend to be similar for siblings, plays a role in the development of a methanogenic flora, which persists into adulthood. Feeding of a soy-based formula, which provides no lactose, may be responsible for fostering the establishment of a methanogenic flora.
H 2 S and CH 3 SH are highly toxic gases that can damage or irritate the intestinal mucosa. Environmental concentrations of 100 ppm are irritating to membranes of the eyes and the lungs and concentrations of 1000 ppm are lethal within 1 hour (28). We have observed cecal concentrations of H 2 S and CH 3 SH of 2000 ppm and 500 ppm, respectively, in gas aspirated from the ceca of rats (8). The colonic mucosa of the adult rat (and human) has a specialized detoxifying system that rapidly converts these sulfur gases to thiosulfate (29). This detoxification presumably allows the mucosa to protect itself from what would otherwise be injurious concentrations of the sulfur gases. In our study, levels of fecal sulfur gases of formula fed infants were similar to the values reported in adults, in which fecal H 2 S release is 1400 ± 182 nmol/g dry weight and CH 3 SH is approximately 530 ± 75 nmol/g dry weight (30). It is possible that the detoxification system for the sulfur gases may not be fully developed in infants and that irritation by sulfur gases might cause discomfort and/or play a role in certain intestinal disorders of infants.
The major finding of this pilot study was that infants' diets, whether breast milk, milk-based formula, or soy-based formula, influence gas release by colonic bacteria. Further investigations are necessary to elucidate the mechanisms by which the diet produces these differences.
The authors thank Janice M Jeter, RN, and the other nurses of the Fomon Infant Nutrition Unit for their skillful services.
1. Barr RG. Colic. In: Wyllie R and Hyams JS, eds. Pediatric gastrointestinal diseases: pathophysiology, diagnosis, management.
St. Louis: Mosby; 1996:241–50.
2. Levitt MD. Volume and composition of human intestinal gas determined by means of an intestinal washout technique. N Engl J Med 1971; 284: 1394–98.
3. Bond JH, Levitt MD. Use of pulmonary hydrogen (H 2
) measurements to quantitate carbohydrate malabsorption: Study of partially gastrectomized patients. J Clin Invest 1972; 51: 1219–25.
4. Bond JH, Currier B, Buchwald H, Levitt MD. Colonic conservation of malabsorbed carbohydrates. Gastroenterol 1980; 78: 444–7.
5. Barr RG, Hanley J, Patterson DK, et al. Breath hydrogen excretion in normal newborn infants in response to usual feeding patterns: evidence for `functional lactase insufficiency` beyond the first month of life. J Pediatr 1984; 104: 527–33.
6. Levitt MD, Hirsh P, Fetzer CA, et al. H2 excretion after ingestion of complex carbohydrates. Gastroenterol 1987; 92: 383–9.
7. Suarez FL, Furne JK, Springfield J, et al. Insights into human colonic physiology obtained from study of flatus composition. Am J Physiol 1997; 272: G1028–33.
8. Suarez F, Furne J, Springfield J, et al. Production and elimination of sulfur-containing gases in the rat colon. Am J Physiol 1998; 274: G727–33.
9. Suarez FL, Springfield J, Levitt MD. Identification of gases responsible for the odour of human flatus and evaluation of a device purported to reduce this odour. Gut 1998; 43: 100–4.
10. Roediger WEW, Moore J, Babidge W. Colonic sulfide in pathogenesis and treatment of ulcerative colitis. Dig Dis Sci 1997; 42: 1571–9.
11. Levine J, Ellis CJ, Furne JK, et al. Fecal hydrogen sulfide production in ulcerative colitis. Am J Gastroenterol 1998; 93: 83–7.
12. Balmer ES, Scott PH, Wharton BA. Diet and fecal flora in the newborn: casein and whey proteins. Arch Dis Child 1989; 64: 1678–84.
13. Balmer ES, Wharton BA. Diet and fecal flora in the newborn: iron. Arch Dis Child 1991; 66: 1390–4.
14. Fomon SJ. Nutrition of normal infants.
St. Louis: Mosby, 1993.
15. Miller TL, Wollin MJ. Enumeration of Methanobrevibacter smithii
in human species. Arch Microbiol 1982; 131: 14–8.
16. Gibson GR. A review. Physiology and ecology of the sulphate-reducing bacteria. J Appl Bacteriol 1990; 69: 769–97.
17. Florin TG, Neale G, Gibson GR, et al. Metabolism of dietary sulphate: absorption and excretion in humans. Gut 1991; 32: 766–73.
18. Strocchi A, Levitt MD. Factors affecting hydrogen production and consumption by human fecal flora: the critical roles of hydrogen tension and methanogenesis. J Clin Invest 1992; 89: 1304–11.
19. Perman JA, Modler S, Olson AC. Role of pH in production of hydrogen from carbohydrates by colonic bacterial flora. J Clin Invest 1981; 67: 643.
20. Coppa GV, Gabrielli O, Pierani P, et al. Changes in carbohydrate composition in human milk over 4 months of lactation. Pediatrics 1993; 91: 637–41.
21. Brand-Miller JC, McVeagh P, McNeil Y, et al. Digestion of human milk oligosaccharides by healthy infants evaluated by the lactulose hydrogen breath test. J Pediatr 1998; 133: 95–8.
22. Sarwar G, Darling P, Ujiie M, et al. Use of amino acid profiles of preterm and term human milks in evaluating scoring patterns for routine protein quality assessment of infant formulas. J AOAC Int 1996; 79: 498–502.
23. Rassin DK, Sturman JA, Gaull GE. Taurine and other free amino acids in milk of man and other mammals. Early Hum Dev 1978; 2: 1–13.
24. Suarez F, Levitt MD. Clinical significance of sulfur-containing gases produced by colonic bacteria. Perspect Colon Rectal Surg 1999; 12: 91–9.
25. Hiele M, Ghoos Y, Rutgeers P, et al. Influence of nutritional substrates on the formation of volatiles by the fecal flora. Gastroenterology 1991; 100: 1597–602.
26. Bond JH, Engel RR, Levitt MD. Factors influencing pulmonary methane excretion in man. J Exp Med 1971; 133: 572–88.
27. Strocchi A, Levitt MD. Intestinal gas. In: Sleisenger MH, Fordtran JS, eds. Gastrointestinal disease: pathophysiology, diagnosis, management. Philadelphia: WB Saunders; 1993: 1035–42.
28. Ellenhorn MJ, Schonwald S, Ordog G, et al. Ellenhorn's medical toxicology: diagnosis and treatment of human poisoning
. Baltimore: Williams & Wilkins, 1997.
29. Levitt MD, Furne J, Springfield J, et al. Detoxification of hydrogen sulfide and methanethiol in cecal mucosa. J Clin Invest 1999; 104: 1107–14.
30. Suarez FL, Furne JK, Springfield J, et al. Bismuth subsalicylate markedly decreases hydrogen sulfide release in the human colon. Gastroenterology 1998; 114: 923–9.
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
Breast-fed; Formula-fed; Gas production; Infant feces; Sulfur gases