The effect of sucrose or starch-based diet on short-chain fatty acids and faecal microflora in rats


Dr A. Cresci, Department of Hygiene and Environmental Sciences, Via E. Betti 3, 62032 Camerino, Italy.


An investigation was carried out to determine whether variations of dietary carbohydrates could modify the colonic flora in rats. Sprague-Dawley rats were fed with two equicaloric diets based on the AIN-76 diet ( American Institute of Nutrition 1977) but differing from that diet in content of carbohydrates, i.e. high sucrose (64%) or high corn starch (64%). Feeding was continued for 9 months ad libitum and no variation in weight gain was recorded among the different diets. A prevalence of aerobes, and a significant reduction in the ratio anaerobes/aerobes in the faeces of rats on the high starch diet compared with the high sucrose diet, was observed. The anaerobe genera identified included Actinomyces, Bacteroides, Bifidobacterium, Clostridium, Eubacterium, Lactobacillus and Propionibacterium. Bacteroides was the most prevalent genus in both dietary groups (51·2 and 29·5% in the faeces of rats fed the sucrose and starch diets, respectively). In contrast, clostridia were prevalent in the starch-fed group (23·8%) and less so in the sucrose diet (11·5%), as propionibacteria were prevalent in faeces of rats fed the starch diet (15·5%), and low in the sucrose diet (3·9%). The remaining genera were scarce in faeces from rats on either diet. Total short-chain fatty acids (SCFA) were significantly higher in the faeces of animals fed the starch diet compared with those fed the sucrose diet. The relative concentrations of acetic, propionic and butyric acids were not significantly different between the two dietary groups. In conclusion, high starch diet can markedly modify the composition of faecal flora and alter considerably the faecal concentration of SCFAs, compounds which might have a health-promoting effect.


Epidemiologic and experimental studies show that diets high in sucrose promote colon cancer ( Bristol et al. 1985 ; Tuyns et al. 1987 ; Caderni et al. 1993, 1997 ). On the other hand, studies linking the consumption of complex carbohydrates to colon cancer find starch to be protective ( Macquart-Moulin et al. 1987 ; Bartram et al. 1991 ; Cassidy et al. 1994 ). It has been proposed that these correlations of dietary habit and cancer risk are a consequence of some dietary components acting as tumour promoters and colleagues as protective factors. It has been demonstrated that diets deficient in starch, cellulose and calcium increase colon mucosa proliferation, which is considered to be a marker for increased risk of colon cancer ( Caderni et al. 1988 ; Lipkin 1988). Furthermore, rats given boluses of sucrose exhibit a burst of cell proliferation in the colon ( Stamp et al. 1993 ). Such observations may have implications for cancer risk in man as most western diets incorporate relatively high amounts of sucrose and processed foods are often supplemented with sucrose and consumed on an empty stomach.

There is, therefore, preliminary evidence that the type of carbohydrate in the diet, particularly whether it is a simple sugar or a complex carbohydrate, can have marked effects on the colonic mucosa function, possibly influencing colon cancer. Dietary carbohydrates likely to have the greatest effect on the colon are those that are poorly digested in the small intestine and hence, pass intact into the large bowel ( Rowland et al. 1998 ).

It has been proposed that many of the effects of starch on the colon are a consequence of the fermentative activities of the gut microflora. Fermentation of carbohydrates yields short-chain fatty acids (SCFA), primarily acetic, propionic and butyric acids, which may directly influence the colonic mucosa, resulting in changes in cell proliferation rates, apoptosis (programmed cell death of damaged cells) and differentiation ( Roediger 1982; Cummings & Macfarlane 1991). In addition, carbohydrates may alter directly, or indirectly via SCFA formation, the physico-chemical conditions in the gut lumen, e.g. pH, and modify the composition of the microflora and the bacterial synthesis of carcinogens and promoters ( Rowland 1992; Cresci et al. 1998 ). As the various SCFA have very different metabolic fates (e.g. acetate is absorbed and reaches the liver and muscles where it is used as an energy source, whereas butyrate is a preferred energy source of colonocytes and induces cellular differentiation in colon cell lines), the extent of fermentation and pattern of SCFA is likely to be of crucial importance in determining the physiological effects of a particular carbohydrate ( Scheppach et al. 1995 ; Cummings & Macfarlane 1997).

Studies performed on the influence of dietary carbohydrates on composition of intestinal microflora have shown contrasting results ( Hentges 1980; Gorbach & Woods 1986). This disagreement could be due to the fact that changes in bacterial populations are evident only after drastic changes in diet composition, or after long-term feeding on different diets ( Moore et al. 1981 ; Canzi et al. 1994 ).

The aim of this study was to verify whether long-term feeding on a high starch diet could affect colon function by altering the faecal flora composition and SCFA production in rats. The rat was chosen as a model as most of the previous studies in this field were carried out using rats.

Materials and methods

Animals and diets

In all experiments, 11–12-week-old female Sprague-Dawley rats (Morini, Reggio Emilia, Italy) were used. The animals were housed in plastic cages with wire tops and bottoms and maintained at constant environmental temperature (20–22 °C), on a 12 h light–dark cycle, according to internationally accepted guidelines. The animals were divided into two groups (n = 8) and fed a starch- or sucrose-based diet ( Table 1) for 9 months. Before beginning the dietary experiments, all rats were maintained for 2 weeks on a standard diet (AIN-76; American Institute of Nutrition 1977). The sucrose and starch diets were based on the AIN-76 diet, differing from that diet in content of carbohydrate, having a slightly lower casein content and a slightly higher fat content, and being composed of both corn and olive oil. Dietary components were purchased from Piccioni Laboratories, Gessate, Milano, Italy. Food and water were provided ad libitum.

Table 1. Composition of experimental diets
  • *

    70% amylopectin and 30% amylose.

  • American Institute of Nutrition 1977.

dl-Methionine 0·30·3
Maize starch *64·0
Olive oil8·08·0
Corn oil4·04·0
Mineral mix AIN 4·04·0
Vitamin mix AIN 1·01·0
Choline bitartrate0·20·2

Sample collection

Fresh faecal samples were taken from single rats which had been isolated in plastic cages. Each sample was used to determine faecal microflora and SCFAs.

Faecal flora

Samples of approximately 1 g (wet weight) were immediately placed in an anaerobic cabinet (Don Whitley Scientific Limited, Shipley, UK) and suspended in 10 ml reducing solution ( Holdeman et al. 1977 ). After homogenization with a Stomacher Lab Blender Model 80-BA 7020 (Seward Medical, London, UK), 1 ml of the faecal homogenate was suspended in 9 ml reducing solution and a series of 10-fold dilutions (10−1–10−10) prepared. A given amount of each dilution (50 μl) was inoculated onto two non-selective media, Peptone yeast extract glucose agar (PYG) ( Holdeman et al. 1977 ) and Columbia blood agar (bioMérieux) used in duplicate for both aerobic and anaerobic bacterial counts, and three selective media, Rogosa agar (Oxoid) for Lactobacillus counts, Sulphite Polymyxin Milk Agar (SPM) ( Mevissen-Verhage et al. 1982 ) for Clostridium and Brain Heart Infusion agar (Oxoid) containing vancomycin (0·75 mg ml−1) and kanamycin (10 mg ml−1) (bioMérieux) for Bacteroides.

Fifty colonies were picked randomly from the 10−7 or 10−8 dilution of non-selective agar plates from each rat, representing 15–80% of the colonies on the plate. The colonies were identified by Gram stain, colony morphology and use of Rapid ID32 A system ( Van Winkelhoff et al. 1988 ; Looney et al. 1990 ). The organisms identified comprised the genera Eubacterium, Propionibacterium, Bifidobacterium and Actinomyces. The counts reported for these groups were obtained by multiplying the total count on the non-selective plate by the proportion of the identified colony type.

Short-chain fatty acid analysis

The general methods, preparation and incorporation of standards were essentially as described by Holdeman et al. (1977) and Peladan et al. (1984) . Faecal samples (0·5 g) were supplemented with caproic acid as internal standard and acidified with 0·5 ml 50% (w/v) H2SO4. The acidic solution was then extracted with 1 ml diethyl ether, centrifuged for 10 min at 4500 rev min−1 (3334 g) (224 Universal Sealed Swinging Bucket Rotor; MP4R model IEC Centrifuge, Needham Heights, MA, USA) and analysed with a Perkin Elmer-Sigma 3B gas-chromatograph (Perkin Elmer, Norwalk, CT, USA) equipped with a flame ionization detector (FID) and a glass column, approximately 2 m × 4 mm i.d., packed with GP 10% SP1000 1% H3PO4 treated chromosorb W, AW 100–120 mesh (Supelco, Inc., Supelco Park Bellefonte, PA, USA). The operating conditions for SCFA determinations have been described elsewhere ( Cresci et al. 1988 ). Retention times for the individual fatty acids were determined by injecting each SCFA standard separately onto the column. Peaks were recorded on a 1022 integrator (Perkin Elmer), which gave concentrations of individual peaks calculated from the relative response factors obtained for the standard SCFA mixture (Supelco, Inc.).

Statistical analysis

Significance of difference between mean values was tested with the Student’s t test. A P of ≤0·05 was considered significant. Statistical analysis was performed with the MINITAB Statistical Software Package (MINITAB 10.2 Inc., State College, PA, USA).


There were no significant variations between groups in individual body weights recorded at the beginning of the experiment (220–240 g) and at various periods during the experiment ( Table 2). Body weight gain and food intake were similar in the two groups ( Table 2).

Table 2. Mean body weights of rats and total amount of diet consumed
DietsInitial body
weight *(g)
Body weight after
60 d * (g)
Body weight after
120 d * (g)
Body weight after
270 d * (g)
Body weight
gain (g)
Diet consumed
(g rat−1 d−1)
  • *

    Mean body weights of two groups of rats fed sucrose or starch diet were not significantly different (P ≤ 0·05), according to the Student’s t test.

  • †Means ±  s. d. of eight rats per group.

Sucrose237·7 ± 4·2 288·0 ± 13·3302·2 ± 22·6307·0 ± 23·569·3 ± 5·813·9 ± 6·3
Starch232·0 ± 9·6287·5 ± 15·9292·2 ± 30·9304·5 ± 33·772·5 ± 6·311·5 ± 5·2

Figure 1 shows that the faeces of rats fed diets high in sucrose or starch for 9 months had similar total counts of bacteria in their faeces. However, the starch-fed rats had a higher content of aerobes and hence, a lower anaerobe/aerobe ratio. The predominant anaerobic genera identified in the faeces of starch-fed rats comprised Actinomyces, Bacteroides, Bifidobacterium, Clostridium, Eubacterium, Lactobacillus and Propionibacterium. In the faeces of rats fed the sucrose diet, Actinomyces was absent ( Table 3 and Fig. 2). Bacteroides was the most prevalent genus in both dietary groups ( Fig. 2) but accounted for a significantly greater proportion of the total count in the sucrose-fed rats. In contrast, clostridia (percentage value) were more prevalent in the starch-fed group, as were propionibacteria. The remaining genera each accounted for less then 5% of the total anaerobic count, although about 23% of the organisms could not be assigned to any genera on the basis of the identification systems used ( Fig. 2). Total SCFA was substantially and significantly higher in faeces of rats fed the starch diet than in those given the sucrose diet ( Table 4). However, the percentages of acetic, propionic and butyric acid were not significantly different between the two dietary groups.

Figure 1.

Anaerobe and aerobe total bacterial count (and ratio) in two groups of rats fed a sucrose or starch diet. a–d Bars not sharing the same superscript are significantly different (P ≤ 0·05 Student’s t test). (▪), Anaerobes; (□), aerobes; (▓), anaerobes/aerobes

Table 3. Mean bacterial counts (log10 cfu g−1) identified in faeces of rats fed the sucrose or starch diet

x¯ ± s. d.
x¯ ± s. d.
  1. ND, not detected (<106).

  2. †Log 10 cfu g−1 wet weight of faeces.

  3. Values marked with an asterisk are significantly different from the sucrose diet, according to the Student’s t test (P ≤ 0·05).

Actinomyces ND8·68 ± 0·24
Bacteroides 10·42 ± 0·46 9·81 ± 0·41*
Bifidobacterium 9·16 ± 0·588·63 ± 0·38*
Clostridium 9·77 ± 0·969·71 ± 1·11
Eubacterium 9·40 ± 0·758·54 ± 1·05
Lactobacillus 9·20 ± 0·518·73 ± 0·35*
Propionibacterium 9·28 ± 0·619·53 ± 0·71
Unidentified genera10·08 ± 1·229·70 ± 1·06
Total anaerobic count10·71 ± 1·4310·34 ± 1·23
Figure 2.

Predominant bacterial genera found in faeces of rats fed a sucrose or starch diet. Values are expressed as percentage of total anaerobic count. Each value is the mean ± standard deviation of eight rats. Bar markers = s. d. (□), Sucrose; (▪), starch

Table 4. Mean faecal content of SCFA (μmol g−1) in two groups of rats fed sucrose or starch diet for 9 months
x¯ ± s. d,
x¯ ± s. d< br>(%)
x¯ ± s. d< br>(%)
x¯ ± s. d< br>(%)
  1. Values marked with an asterisk are significantly different from the sucrose diet, according to the Student’s t test (P ≤ 0·05).

Sucrose 39·4 ± 10·3
 11·1 ± 3·1
 3·2 ± 1·5
 53·7 ± 11·1
Starch 78·5 ± 27·3*
 16·4 ± 6·2*
 6·5 ± 3·0*
 101·4 ± 28·2*


Analysis and cultivation of intestinal micro-organisms is very complex ( Tannock 1983; Mallett et al. 1987 ), rendering diet-related changes in microflora composition difficult to demonstrate ( Mallett et al. 1987 ). However, in recent years, probably due to methodological improvements, dietary-induced variations in faecal microflora composition have been detected, particularly when indigestible carbohydrates such as dietary fibre or non-digestible oligosaccharide have been added to the diet ( Wells et al. 1992 ; Maczulak et al. 1993 ; Rowland & Tanaka 1993; Canzi et al. 1994 ; Cresci et al. 1995 ; Gibson & Roberfroid 1995).

Our data demonstrate that not only indigestible, but also predominantly digestible carbohydrate, such as the standard corn starch used in this study ( Stephen et al. 1983 ; Mallett et al. 1988 ; Englyst et al. 1992 ), can affect the faecal flora composition over a long feeding schedule. In particular, they show that a high corn starch diet markedly increased the total aerobe count and decreased the anaerobe/aerobe ratio. There was a higher concentration of clostridia and propionibacteria and a lower concentration of bacteroides in the gut flora of rats fed the high starch diet than in those on the sucrose diet.

In addition to faecal flora variations, we demonstrated a marked difference (about twofold) in faecal SCFA content in the rats fed a high sucrose diet compared with those on a high starch diet.

It seems likely that the sucrose in the diet was rapidly hydrolysed to glucose and fructose in the upper small intestine and absorbed, whereas at least some of the starch escaped digestion in the upper gut and reached the colon, resulting in the increased SCFA concentration in faeces. It is noteworthy that there was such a large difference in SCFA concentration in faeces of rats on the two dietary groups, as it is known that SCFA are absorbed from the colon (acetate and propionate) or utilized by the colonic mucosa (butyrate) ( Cummings & Macfarlane 1991, 1997). It is likely, therefore, that even greater differences in SCFA concentrations occurred in the colon.

In spite of the large differences in faecal SCFA concentration, the relative proportions of the major acids did not change with diet, suggesting that the microfloral changes ( Fig. 2) did not result in an increased proportion of butyrate-producing strains. This was surprising in view of the known tendency for starch fermentation in the gut to favour butyrate production. On the other hand, production rates and ratios of SCFA may be more important than luminal concentrations because absorption rates, which is the proportion of SCFA which is offered to the mucosal cells, probably parallel production rather than luminal concentrations. However, this may reflect the extended dietary exposure used in the present study in contrast to the short-term intervention used by other workers. Most investigations of dietary effects have been conducted over periods of less than 4 weeks ( Mallet & Rowland 1988), allowing little time for the long-term adaptation of the microflora to dietary supplements.

Increased concentrations of SCFA in the colon, particularly n-butyrate, are usually considered to be beneficial to the host animal ( Roediger 1996; Cumming & Macfarlane 1997). Consequently the approximate twofold difference in SCFA concentration seen in the present study may be indicative of a beneficial influence of the high starch diet, and be consistent with effects on gut mucosa reported in previous starch intervention studies in rats ( Caderni et al. 1993, 1997 ).

The significance of the bacteriological modifications associated with the change from a simple to a complex carbohydrate source in the diet is difficult to interpret. Current studies of diet and microfloral composition indicate that increases in lactic acid-producing bacteria may be associated with a beneficial effect towards the host ( Rowland & Tanaka 1993; Gibson & Beaumont 1996; Rowland 1996).

In our study, the numbers of bifidobacteria and lactobacilli were significantly lower in the rats fed starch compared with the sucrose-fed animals, suggesting that a dietary change from simple to complex carbohydrate cannot be considered as beneficial in terms of microfloral composition. However, it is noteworthy that the anaerobe/aerobe ratio in the starch-fed rats was markedly lower than that in the sucrose-fed animals. Studies on populations exhibiting major differences in colon cancer rates have indicated an association between low anaerobe/aerobe ratio and low colon cancer risk ( Hill et al. 1971 ).


The authors thank Mary Forrest for linguistic revision of the text and Ian Rowland for helpful discussions. This work has been supported by grants from European Community Project AIR 2 CT 94–0933 and partially with funds from Consiglio Nazionale delle Ricerche (progetto finalizzato Fat.Ma n. 9500692. PF41) and MURST (Ministero Università e Ricerca Scientifica).