318 C. J. Edmonds

be carried out on the distal colon and rectum andsome factors which influence the pd have beenexamined. The pd is of significance both because itconsiderably influences the flows of charged particlesacross the epithelium and also because it is a measur-able variable which reflects something of the function-al state of the tissue. Since, however, it is affectedby a variety of factors, these must be defined andconsidered if any meaningful interpretation is to bemade.

References

Andersson, S., and Grossman, M. I. (1965). Profile of pH, pressureand potential difference at astroduodenal junction in man.Gastroenterology, 49, 364-371.

Archampong, E. Q., and Edmonds, C. J. (1972). Effect of luminalions on the transepithelial electrical potential difference ofhuman rectum. Gut, 13, 559-565.

Archampong, E. Q., Harris, J., and Clark, C. G. (1972). The absorptionand secretion of water and electrolytes across the healthy andthe diseased human colonic mucosa measured in vitro. Gut, 13,880-886.

Barry, R. J. C. (1967). Electrical changes in relation to transport. Brit.med. Bull., 23, 266-269.

Dalmark, M. (1970). The transmucosal electrical potential differenceacross human rectum in vivo following perfusion of differentelectrolyte solutions. Scand. J. Gastroent., 5, 421-426.

Diamond, J. M., and Harrison, S. C. (1966). The effect of membranefixed charges on diffusion potentials and streaming potentials.J. Physiol. (Lond.), 183, 37-57.

Edmonds, C. J. (1970). Electrical potentials of the sigmoid colon andrectum in irritable bowel syndrome and ulcerative colitis. Gut,11, 867-874.

Edmonds, C. J., and Marriott, J. (1967). The effect of aldosteroneand adrenalectomy on the electrical potential difference of ratcolon and on the transport of sodium, potassium, chloride andbicarbonate. J. Endocr., 39, 517-531.

Edmonds, C. J., and Marriott, J. C. (1968). Electrical potential andshort circuit current of an in vitro preparation of rat colonmucosa. J. Physiol. (Lond.), 194, 479-494.

Edmonds, C. J., and Nielsen, 0. E. (1968). Transmembrane electricpotential differences and ionic composition of mucosal cells ofrat colon. Acta physiol. scand., 72, 338-349.

Edmonds, C. J., and Pilcher, D. (1973). Electrical potentialdifference and sodium and potassium fluxes across rectalmucosa in ulcerative colitis. Gut, 14, 784-789.

Edmonds, C. J., and Richards, P. (1970). Measurement of rectalelectrical potential difference as an instant screening test forhyperaldosteronism. Lancet, 2, 624-627.

Finkelstein, A., and Mauro, A. (1963). Equivalent circuits as relatedto ionic systems. Biophys. J., 3, 215-237.

Geall, M. G., Code, C. F., Mcllrath, D. C., and Summerskill, W. H. J.(1969). Measurement of gastrointestinal transmural electricpotential difference in man. Gut, 11, 34-37.

Grantham, R. N., Code, C. F., and Schlegel, J. F. (1970). Referenceelectrode sites in determination of potential difference acrossthe gastrooesophageal mucosal junction. Mayo Clin. Proc., 45,265-274.

Rask-Madsen, J., and Brix Jensen, P. (1973). Electrolyte transportcapacity and electrical potentials of the normal and theinflamed human rectum in vivo. Scand. J. Gastroent., 8, 169-175.

Schultz, S. G. (1972). Electrical potential difference and electromotiveforces in epithelial tissues. J. gen. Physiol., 59, 794-798.

Smyth, D. H., and Wright, E. M. (1966). Streaming potentials in therat small intestine. J. Physiol. (Lond.), 182, 591-602.

Thompson, B. D., and Edmonds, C. J. (1974). Aldosterone, sodiumdepletion and hypothroidism on the ATPase activity of ratcolonic epithelium. J. Endocr., 62, 489-496.

Tomkins, A., and Edmonds, C. J. (1975). Effect of carbenoxolone onthe electrical potential difference and sodium and potassiumtransport across rectal mucosa. (In press).

The normal colonic bacterial flora

M. J. HILL AND B. S. DRASAR

From the Bacterial Metabolism Research Laboratory, Central Public Health Laboratory,Colindale Hospital, London

Interest in the human intestinal bacterial flora hasincreased greatly in recent years. To a large extentthis is because techniques have now been developedwhich permit the study of the dominant members ofthe gut flora, the non-sporing strictly anaerobicbacteria.There are two major approaches to the study of

the gut flora. The first is the classical bacteriologicalapproach, the identification and the enumeration ofthe major groups of bacteria. The second is the studyof the biochemical activities of the flora. These twoaspects of the flora will be discussed in turn.

The Normal Colonic Flora

The normal colonic flora is usually inferred from thecomposition of the normal faecal flora, sincesuitable techniques for sampling various levels ofthe colon have yet to be developed. Data from animalstudies would support the assumption that the floradoes not alter during defaecation, indicating that thefaecal flora adequately represent that of the recto-sigmoid. We know that the flora of the recto-sigmoiddiffers from that at the ileocaecal junction (table I)and can infer, but no more, that the change takes

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from

Symposium on colonic function

Species Mean Loglo Viable Count/GramIntestinal Material

Terminal Caecum FaecesIleum

Enterobacteria 3-3 6-2 7-4Enterococci 2-2 3-6 5-6Lactobacilli <2 6-4 6-5Clostridia <2 3 0 5 4Bacteroides 5 7 7-8 9-8Gram-positive non-sporing

anaerobes (Eubacteriaand Bifidobacteria) 5 8 8-4 10-0

Number of samples studied 6 2 100

Table I The bacterial flora of the lower intestine in man

place in the upper rather than the lower parts of thecolon.

Non-sporing anaerobic, rod-shaped organismsoverwhelmingly predominate in the human faecalflora, accounting for more than 99% of the totalorganisms. The major species present are Bacteroidesfragilis, Bifidobacterium adoloscentis, and Eubac-terium aerofaciens (Peach, Fernandez, Johnson, andDrasar, 1974). The remainder are mainly Escherichiacoli, Streptococcus viridans, and Streptococcussalivarius and lactobacilli, but there are in additiona host of minor components of the flora includingveillonellae, bacilli, peptococci and peptostrepto-cocci, clostridia, Strep faecalis, and coliformorganisms other than Esch. coli (eg, Klebsiella spp,Proteus spp). It must be stressed that, although theseare minor components of the average normal flora,they may be of major importance in an individualperson.There are difficulties in interpreting the data on the

normal flora. For example, the enumeration ofclostridia is very difficult due to the lack of a suitableselective or diagnostic medium. Consequently,although the numbers of clostridia producinglecithinase can readily be determined there is nomeans of counting vegetative clostridia which do notproduce this enzyme. In the case of the latterorganisms only the numbers of spores can becounted and this may be totally unrelated to thevegetative count; it is possible, therefore, thatclostridia may be far more numerous in the colonicflora than is realized at present. Within the genusBacteroides there are more than 20 species. Morethan 95% of faecal bacteroides belong to the speciesB. fragilis and consequently the enumeration of theminor species is extremely time consuming. It isimportant to recognize, therefore, that a species ofnon-sporing anaerobe that represents only 01 %of the organisms of that genus would neverthelessbe present in numbers comparable to those of thecoliforms, streptococci, or lactobacilli. Until good

319

selective media are available for the individualspecies of non-sporing anaerobes it will be necessaryto pick 1000 colonies in order to be able to detectan organism present in considerable numbers. Thusthere are still many problems remaining to besolved before we can characterize the normal faecalflora.

It must be noted, too, that the faecal flora repre-sents only the luminal flora of the recto-sigmoidregion and that the mucosal flora (the bacteriagrowing in close association with the mucosalsurface and in the villous crypts) may differ markedlyfrom this. This criticism also applies to the data onthe small intestinal flora. Determination of themucosal flora presents great difficulties which willnot readily be overcome; however, it is suspectedthat the mucosal flora is more important than theluminal flora in, for example, urea metabolism.

THE EFFECT OF DIET ON THE FLORAPeople living in Uganda on a diet of matoke (boiledmashed banana) have a faecal flora which differsmarkedly from that of English people living on anormal western diet (Aries, Crowther, Drasar, Hill,and Williams, 1969) and this difference has beenassumed to be due to the difference in diet. Altera-tion of the flora of adults by modifying their diet hasproved to be extremely difficult to demonstrate,however. The flora of people on a low-fat diet forfour weeks was not detectably different from that ofthe same people on a control diet; a diet rich inbagasse (fibre from sugar cane) had no detectableeffect even in 12 weeks. However, differences in theflora of three dietary groups living in Hong Kongcould be demonstrated. Our conclusions from thesedata are that (a) the effects of dietary alteration areonly slowly manifest, and in controlled studiesprolonged periods on a diet (more than a year) maybe necessary before the effects can be demonstratedusing the classical bacteriological techniques; (b)classical bacteriological techniques are unsuitablefor the study of faecal micro-ecology. If total faecalP-glucuronidase is measured then the effects of dieton the flora can be demonstrated readily in fourweeks (Wynder and Reddy, 1974). Using a batteryof such tests, together with the results of screeningstudies to determine the mean enzymic activity ofstrains of each genus, such finer alterations in theflora might be detected.

Metabolic Activity of the Flora

In the colon there is approximately 1-5 kg wet weightof bacteria and therefore it would be amazing if theirmetabolic activity did not play a significant role inthe overall economy of the human body. There has

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from

M. J. Hill and B. S. Drasar

been considerable interest in this subject in recentyears and it would be impossible to cover thissubject adequately. In the subsequent section threeselected topics will be discussed in some depthrather than an attempt to cover the full range of thesubject.

HYDROLYSIS OF GLYCOSIDES ANDGLUCURONIDESA high proportion of taxonomic tests in bacteriologyare fermentation reactions that involve an initialhydrolysis of a glycosidic bond followed by fermen-tation of the released monosaccharide. Examples ofsuch reactions are thefermentation ofaesculin, salicin,and amygdalin. All three examples given are of fi-glucosides and such compounds are poorly metabo-lized by intestinal enzymes. Many plant products areglucosides of toxic compounds and it is the abilityof the colonic flora to hydrolyse these compounds torelease the toxic aglycone that renders an item offood dangerous. Thus, untreated cassava containsamygdalin which is hydrolysed by the flora to releasecyanide. Untreated cassava is highly toxic and inareas where cassava lis a staple source of starch theraw product is extracted exhaustively to removethe amygdalin before converting the 'purified'cassava into flour.

In other parts of the world cycad nuts are a staplesource of starch but, again, untreated cycads areextremely hepatotoxic because they contain theglucoside cycasin, which is hydrolysed by the gutflora to release methylazoxy-methanol. In germ-freerats cycasin administered orally is harmless at a dosewhich would kill a conventional rat in 48 hours(Spatz, Smith, McDaniel, and Laqueur, 1967). Torender them harmless the cycads are extractedrepeatedly with water to remove the cycasin.The ability of the flora to hydrolyse plant gluco-

sides has been used extensively in medicine. Theglucoside cathartics senna and cascara are hydro-lysed to release their active agents. It has beendemonstrated (Hardcastle and Wilkins, 1970) thatwhereas cascara sagrada has no effect on the colonthe aglycone increases colic motility of rabbits. Ifthe free aglycone is fed to the animals it has nocathartic action, indicating that it is sensitive to eithergastric or small intestinal secretions. Thus theglucose moiety protects the aglycone during itstransit through the upper digestive tract allowing itto be released in an active form in the colon.The ability of the bacterial flora to hydrolyse

glucuronides is important in the enterohepaticcirculation of compounds (Smith, 1966). This hasboth beneficial and harmful effects. For example,chloramphenicol, which undergoes extensive entero-hepatic circulation, is retained within the body for a

prolonged period of time and therefore has a betteropportunity to exert its antibacterial effect. Similarlyphenolphthalein and morphine are also retainedthereby improving their therapeutic effect. However,retention within the enterohepatic circulation alsoallows more time for bacterial degradative enzymesto be induced. If we use chloramphenicol as theexample there is a wide range of bacterial modifica-tions that can occur (Smith, 1966), all of whichresult in loss of antibacterial activity and some ofwhich may result in the formation of toxic products.

Lactose is readily metabolized by bacterial fi-galactosidase but this is normally of no consequencein western human adults since the intestinal ,B-galactosidase degrades any dietary lactose longbefore it reaches the large bowel. In most parts ofthe world, however, intestinal P-galactosidase is lostduring childhood and ingestion of lactose results infermentative diarrhoea. This has been reviewed byNeale (1971). Bacterial ,-galactosidase is utilized inlactulose therapy. Lactulose is the ketoisomer oflactose; it is not hydrolysed by intestinal P-galactosi-dase but is readily degraded by the bacterial enzyme.By carefully controlling the dose of lactulose theamount of acid produced in the colon, and thereforethe colonic pH, can be controlled.

Cellulose and hemicelluloses are complex glyco-sides which are susceptible to bacterial attack; it hasbeen shown that normally 20% of a dose of hemi-cellulose is degraded during transit through the gut.This degradation will result in the formation of short-chain fatty acids which by their osmotic effect mightwell explain the bulking effect of fibre.

AMMONIA METABOLISMIntestinal bacteria produce ammonia by deaminationof amino acids but the principal route is by ureahydrolysis. Urea undergoes an enterohepatic circula-tion, being produced in the liver then entering thecolon by passive diffusion, where it is hydrolysedto ammonia and carbon dioxide. The ammonia re-enters the portal blood, again by diffusion, andreturns to the liver to be converted back to urea orto be incorporated into protein. Approximately 15-30% of the total urea pool, ie, about 7 g, is hydro-lysed each day (Walser and Bodenlos, 1959; Jones,Smallwood, Craigie, and Rosenoer, 1969). Themajor source of colonic urease is the gut bacterialflora. Treatment of patients with neomycin or otherantibiotics reduces the faecal urease activity (Evans,Aoyagi, and Summerskill, 1966) and germ-freeanimals fail to catabolize urea to any significantextent (Levenson, Crowley, Horowitz, and Malm,1959). Interestingly, it appears that it is the mucosalflora that is important in urea metabolism rather thanthe luminal flora. This conclusion was reached by

320

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from

Symposium on colonic function

Wolpert, Phillips, and Summerskill (1971) followingthe observation that urea delivered luminally wasmuch less efficiently hydrolysed by the flora than thatadministered systematically. Hardly any urea isdetectable in faeces so that virtually all of the ureaentering the gut is hydrolysed to ammonia.

In the normal healthy person, the ammoniagenerated in the gut is restricted to the portal bloodsystem, but in hepatic disease the liver fails todetoxify the ammonia and it enters the generalcirculation (White, Phear, Summerskill, and Sher-lock, 1955). This is said to result in the chronic orrecurrent neuropsychiatric disorders related to pro-tein intolerance, ultimately resulting in hepaticcoma due to the direct cerebrotoxic action ofammonia (Schenker, McCandless, Brophy, andLewis, 1967). Similar symptoms arise, for the samereasons, in protein intolerance due to portal sys-temic anastomosis (Riddell, 1955). Hyperam-monaemia has been reviewed by Summerskill (1970).The major course of therapy of hepatic coma is the

control of intestinal ammonia production andnitrogen metabolism. This has been attemptedeither by (a) reducing the supply of nitrogen com-pounds to the gut, (b) reducing the number ofammonia-producing bacteria, (c) reducing the timeavailable for bacteria to produce ammonia, and (d)preventing the escape of ammonia from the gut.Attempts to reduce the numbers of ammonia-producing bacteria take two forms: antibiotictherapy and replacement therapy. A number ofantibiotics have been shown to be effective for a shorttime but these have the usual side effects whichaccompany disturbance of the gut flora, namely,diarrhoea, steatorrhoea, and staphylococcal entero-colitis (Walker, Emlyn-Williams, Craigie, Rosenoer,Agnew, and Sherlock, 1965). Replacement therapyis based on the concept that, since some bacteria areureolytic while others are not, replacement of thenormal (ureolytic) flora by one which fails tohydrolyse urea (eg, lactobacilli) will be beneficial. Itmight well be so if such a replacement were possible;as it is this method of therapy has little success, andonly serves to emphasize the remarkable stability ofthe gut flora. The amount of ammonia producedshould be reduced if the time available for itsproduction is reduced, and this is the rationalebehind the use of enemas and purgatives. There isstill some question regarding the amount of agentto be used, and the ultimate effectiveness of thistreatment which would inevitably appeartobeashort-term measure.

Prevention of the escape of ammonia from the gutis the basis of lactulose therapy. Ammonia is readilyabsorbed from the gut in its free form. When thecolonic pH is reduced the ammonia will be in an

321

ionized form and will be on a pH gradient; underthese conditions the ammonia should theoreticallymove from the blood to the colon rather than thereverse, resulting in reduced blood ammonia levelsand increased faecal excretion of ammonia. This isthe rationalization of the use of lactulose, the dose ofwhich is carefully controlled to give a faecal pH of4 to 5 (Bircher, Muller, Guggenheim, and Haem-merli, 1966; Elkington, Floch, and Conn, 1969).Unfortunately for the theory, under conditions inwhich lactulose was being used successfully, ie,when the blood ammonia level was reduced, therewas no increase in faecal ammonia levels. It isapparent that some rethinking is necessary.

In hepatic failure the production of ammonia bythe gut flora is detrimental to the patient; in renalfailure the reverse is true. Here the blood ureaincreases to toxic levels unless the nitrogen intakeis controlled. As the blood urea level increases sodoes the faecal urease (Brown, Hill, and Richards,1971), resulting in the release of ammonia whichreturns via the portal system to the liver where aproportion is incorporated into protein. Theincreased faecal urease could be the result either ofan increased proportion of organisms producingurease, or to an increased output of enzyme from aconstant number of active organisms. In fact, theformer is true, with the proportion of urease-positiveorganisms in the faecal flora increasing with theblood urea level. There is no apparent reason forthis; the production of urease is not known toconfer any ecological advantage in the presence ofhigh urea levels.

METABOLISM OF BILE ACIDSThe metabolism of bile acids is of interest from twopoints of view: these are (1) the major metabolicreactions which give rise to the secondary bile acidsin bile and faeces; (2) the minor reactionswhich maygive rise to carcinogens and be implicated in theaetiology of colon cancer.The major metabolic reactions carried out by the

gut bacteria are (1) hydrolysis of bile conjugates torelease the free bile acids; (2) hydroxyl oxidation toa keto group, followed under certain circumstancesby reduction to a ,B-hydroxyl group; (3) removal ofthe 7ac-hydroxyl group to yield the major secondarybile acids deoxycholic acid (from cholic acid) andlithocholic acid (from chenodeoxycholic acid).Although the deconjugation reaction occurs

extensively in the ileum the other reactions takeplace mainly in the caecum and colon. The dehy-droxylation reaction yields products which are lesswell absorbed than their substrates and conse-quently contributes to the total faecal loss of bileacids. Although the products of these major reac-

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from

M. J. Hill and B. S. Drasar

tions may play a role in small intestinal disease, theydo not contribute in any large measure to largeintestinal disorders.The minor reactions of bacteria upon bile acids

are of more interest to us because we have pro-duced the hypothesis that they are important in theaetiology of large bowel cancer. We have postulated(Aries et al, 1969) that colon cancer is caused by a

carcinogen or cocarcinogen produced in situ in thecolon from some otherwise benign substrate; wehave also postulated that the substrate is the bile acidmoiety. We have shown (Hill, Drasar, Aries,Crowther, Hawksworth, and Williams, 1971; Hill,1974) that in comparisons between a number ofcountries, some with high and some with lowincidences of colon cancer, the incidence of coloncancer correlates with the total faecal bile acidconcentration and with the faecal concentration ofdihydroxycholanic acids. The latter correlation wasremarkably good.The question which then arises is, What is the nature

of the carcinogen or cocarcinogen produced fromthe bile acids? We have postulated that the carcino-gen is a polyunsaturated bile acid, and have suggesteda pathway by which the bacteria could theoreticallyproduce a polycyclic aromatic compound from thebile acids using only four types of nuclear dehydro-genation reaction (Hill, 1971a). These have all beendemonstrated in vitro using human gut bacteria andwe have produced in vitro a bile acid with rings A andB of the steroid nucleus fully aromatic. One of thefour reactions is the dehydroxylation which proceedsvia an unsaturated intermediate (Samuelsson, 1960;Goddard and Hill, 1973); this reaction is carried outby a wide range of anaerobic organisms, the propor-tion of active strains being much higher in areas

with a high incidence of colon cancer than in thosewith a low incidence. The other three reactions areonly carried out by certain lecithinase-negativeclostridia, principally Cl. paraputrificum, which willsubsequently be referred to as nuclear-dehydro-genating (or NDH) clostridia. These organisms arerare in the faeces of people living in areas with a lowincidence of colon cancer but much more commonin faeces from people living in high incidence areas.Thus in support of our hypothesis that the gut floraproduce an unsaturated carcinogen from the bileacids and that this is important in the aetiology ofcolon cancer, we find that in areas with a highincidence of colon cancer the concentration of thepostulated substrate is high and the relevant bacteriaare common, whilst in areas of low incidence thesubstrate concentration is low and the relevantbacteria are rare.During the last year we have been testing our

hypothesis in individuals, since those with a high

faecal bile acid concentration and large numbers ofNDH clostridia should be at high risk whilst thosewith low levels of only one or neither of these tworequirements should be at low risk. In the study,which is being conducted in collaboration withworkers at Northwick Park Hospital, Harrow, and atSt Mark's Hospital, London, we have examined thestools of patients with newly diagnosed large bowelcancer and patients admitted to the same ward withdiseases other than colon cancer. Only the firststool on entry to the ward was acceptable, sincelater stools would have been affected by the hospitaldiet, which is low in fat and would result in a lowfaecal bile acid level (Hill, 1971b). The study isbeing carried out blind with the code being brokenafter each 50 patients. The latest analysis (Hill,Drasar, Goddard, Peach, Williams, Meade, Cox,Simpson, and Morson, 1974) shows that in thelarge bowel cancer patients 22 of 29 belonged to thehigh-risk group compared with one of 60 of thecontrol group (table II).

Faecal Bile Nos. ofNDH Large Bowel ControlAcid Clostridia Cancer PatientsConcentration (per g wet wt) Patients(mg/g dry wt)

>6 >108 22 (76%) 1 (1-7%)<6 >10' 4 (14%) 25 (42%)>6 <103 2 (7%) 6 (10%)<6 <10' 1 (3%) 28 (47%)

Table II Bile acid concentration and numbers ofNDH clostridia in faeces from patients with large bowelcancer and controls admitted to the same gastro-enterology ward with diseases other than large bowelcancer

These findings support those obtained in theinternational study. The next steps are (a) to seewhether the faecal bile acids concentration togetherwith the numbers of NDH clostridia can be used topredict colon cancers in a prospective study; (b) tosee whether any of the products of nuclear dehydro-genation of the bile acids obtained in vitro is carcino-genic in animal tests; (c) to see which of theseunsaturated bile acids are produced in vivo in man;(d) to see the effect of dietary changes on the faecalbile acid concentration and on the numbers ofNDH clostridia produced.

Conclusions

Although our knowledge of the gut flora both interms of its composition and of its function hasincreased dramatically in recent years, the un-answered questions still greatly outnumber those

322

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from

Symposium on colonic function 323

answered. In this communication we have been ablemerely to outline the problems involved and theprogress made towards solving some of them. It isevident that this is still a new field of study which canmake a contribution to the understanding not onlyof gastroenterological but also of general medicaland pharmacological problems.

References

Aries, V., Crowther, J. S., Drasar, B. S., Hill, M. J., and Williams,R. E. 0. (1969). Bacteria and the aetiology of cancer of thelarge bowel. Gut, 10, 334-335.

Bircher, J., Mtiller, J., Guggenheim, P., and Haemmerli, U. P. (1966).Treatment of chronic portal-systemic encephalopathy withlactulose. Lancet, 1, 890-893.

Brown, C. L., Hill, M. J., and Richards, P. (1971). Bacterial ureasesin uraemic men. Lancet, 2, 406.408.

Elkington, S. G., Floch, M. H., and Conn, H. 0. (1969). Lactulose inthe treatment of chronic portal systemic encephalopathy. NewEngl. J. Med., 281, 408-412.

Evans, W. B., Aoyagi, T., and Summerskill, W. H. J. (1966). Gastro-intestinal urease in man. II. Urea hydrolysis and ammoniaabsorption in upper and lower gut lumen and the effect ofneomycin. Gut, 7, 635-639.

Goddard, P., and Hill, M. J. (1973). The dehydrogenation of thesteroid nucleus by human gut bacteria. Trans. biochem. Soc.,1, 1113-1115.

Hardcastle, J. D., and Wilkins, J. L. (1970). The action of sennosidesand related compounds on human colon and rectum. Gut, 11,1038-1042.

Hill, M. J. (1971a). Gut bacteria steroids and cancer of the largebowel. In Some Implications of Steroid Hormones in Cancer,edited by D. C. Williams and M. H. Briggs, pp. 94-106.Heinemann, London.

Hill, M. J. (1971b). The effect of some factors on the faecal concen-tration of acid steroids, neutral steroids and urobilins. J.Path., 104, 239-245.

Hill, M. J. (1974). Bacteria and the etiology of colon cancer. Cancer(Philad.), 34,815-818.

Hill, M. J., Drasar, B. S., Aries, V., Crowther, J. S., Hawksworth,G. and Williams, R. E. 0. (1971). Bacteria and aetiologyof cancer of the large bowel. Lancet, 1, 95-100.

Hill, M. J., Drasar, B. S., Goddard, P., Peach, S., Williams, R. E. O.,Meade, T. W., Cox, A., Simpson, J. E. P., and Morson, B.,(1974). Bacteria, bile acids, and the aetiology of large bowel.cancer. J. med. Microbiol., 7, vii-viii.

Jones, E. A., Smallwood, R. A., Craigie, A., and Rosenoer, V. M(1969). The enterohepatic circulation of urea nitrogen. Clin.Sci., 37, 825-836.

Levensen, S. M., Crowley, L. V., Herowitz, R. E., and Malm, 0. J.(1959). The metabolism of carbon-labelled urea in the germ-free rat. J. biol. Chem., 234, 2061-2062.

Neale, G. (1971). Disaccharidase deficiencies. J. clin. Path., 24, Suppl.(Roy. Coll. Path.) 5, 22-28.

Peach, S., Fernandez, F., Johnson, K., and Drasar, B. S. (1974). Thenon-sporing anaerobic bacteria in human facces. J. med.Microbiol., 7, 213-221

Riddell, A. G. (1955). Ammonia intoxication in surgical patients.Ann. roy. Coll. Surg. Engl., 17, 319-329.

Samuelsson, B. (1960). On the mechanism of the biological formationof deoxycholic acid from cholic acid. J. biol. Chem., 235,361-366.

Schenker, S., McCandless, D. W., Brophy, E., and Lewis, M. S.(1967). Studies on the intracerebral toxicity of ammonia.J. cin. Invest., 46, 838-848.

Smith, R. L. (1966). The biliary excretion and enterohepatic circula-tion of drugs and other organic compounds. Fortschr. Arug-neimittelforsch., 9, 300-359.

Spatz, M., Smith, D. W. E., McDaniel, E. G., and Laqueur, G. L.(1967). Role of intestinal micro-organisms in determiningcycasin toxicity. Proc. Soc. exp. Biol. (N. Y.), 124, 691-697.

Summerskill, W. H. J. (1970). Ammonia metabolism in the gastro-intestinal tract. Progr. Gastroent., 2, 276-287.

Walker, J. G., Emlyn-Williams, A., Craigie, A., Rosenoer, V. M.,Agnew, J., and Sherlock, S. (1965). Treatment of chronicportal-systemic encephalopathy by surgical exclusion of thecolon. Lancet, 2, 861-866.

Walser, M., and Bodenlos, L. J. (1959). Urea metabolism in man.J. clin. Invest., 38, 1617-1626.

White, L. P., Phear, E. A., Summerskill, W. H. J., and Sherlock, S.(1955). Ammonium tolerance in liver disease: observations basedon catheterization of the hepatic veins. J. clin. Invest., 34, 158-168.

Wolpert, E., Phillips, S. F., and Summerskill, W. H. J. (1971). Trans-port of urea and ammonia production in the human colon.Lancet, 2, 1387-1390.

Wynder, E. L., and Reddy, B. S. (1974). Metabolic epidemiology ofcolorectal cancer. Cancer (Philad.), 34, 801-806.

Absorption and secretion by the colon

JOHN H. CUMMINGS

From the Medical Research Council Gastroenterology Unit, Central Middlesex Hospital, London

Various aspects of colonic absorptive function inman have been dealt with in a number of reviews(Phillips, 1969; Turnberg, 1970; Wrong, 1971;Shields, 1972; Sladen, 1972; Edmonds and Pilcher,1972).

Electrolytes and Water

The role which the colon plays in conserving waterand electrolytes has been known for some time,

deduced largely from a comparison of the volumeand composition of ileostomy effluent with that offaeces and confirmed by perfusion of the colon invivo. The quantity of these substances absorbeddaily by the colon has, however, to be revised,since it has been shown that the flow of ilealcontents into the colon each day is three timesthe normal volume of ileostomy effluent (Kanaghinis,Lubran, and Coghill, 1963;Phillips and Giller, 1973).Fasting ileal flow rates, measured by slow intestinal

on 2 August 2018 by guest. P

rotected by copyright.http://gut.bm

j.com/

Gut: first published as 10.1136/gut.16.4.318 on 1 A

pril 1975. Dow

nloaded from