8/9/2019 J. Nutr.-1981-Slavin-287-97

1/11

Neutra l Dete rgen t F iber Hemicellu lo se and Ce llu lo seDiges tib ility in Human Sub jects1 2

JOANNE L. SLAVIN, PAULA M. BRAUER ANDJUDITH A. M ARLETTD ep ar tm en t o f N utr itio na l S cie nc esU niv er sity o f Wis co ns in 1 3 00 L in de n D rMadison WI 53706

ABSTRACT The fate of dietary fiber and its components was examined inseven women consuming low cellu lo se (LC) and h igh cellu lo se (HC) d ie ts, eachfor about 1 m onth. The diets w ere of constant daily com position and differedonly in that the H C diet contained an additional 16 g/day non-nutritive fiber(Solka F loe), which increased the neu tra l dete rgen t fiber (NDF) of the d ie t from9.5 to 23.5 g/day and C ram pton and M aynard cellulose from 5.4 to 19.3 g/day.When apparent fiber d iges tib ilities during 5-day periods were determined , bothN DF and cellulose digestibilities varied greatly and inconsistently in eachsubje ct throu ghout both diet periods. T herefo re, sam ples w ere po oled to forma single 20-30 day composite for each subject during each diet. Meanapparent ND F digestibility, after correcting for protein contam inationin feca l NDF, was 70.4 ±7.3 during the LC die t and decreased to 23.0 ±15.0during the H C diet. C ellulose digestibility w as 69.7  ±10.7 w ithout and15.7 ±17.4 w ith the added cellulose. H em icellulose was calculated asNDF minus cellu lo se . When the feca l NDF was corrected fo rp ro te in con tamination, hemicellu lo se d igestib ility ave raged 71.7 ±5 .4 during the LC die t and51 .0  ±7 .9 during the H C d iet. In a separate ex perim ent, 16 g/day S olka F loewas ingested with a semi-purified liquid diet and only 8 of the

cellulose w as digested. These results suggest that m ore than half of the fiberin a LC die t con ta in ing fru its, vegetable s and re fined gra ins is deg raded, whilethe apparent digestibility of refined cellulose is m inim al. J. N utr. Ill: 287-297, 1981.INDEXING KEY W ORDS neutral detergent fiber •hemicellulose •cel lulose •dietary fiber digest ibil ity

Information on the fate of dietary fiber (6-8) but decreases when bran (6) oin the gut is essential for evaluating the cellulose (8) supplements are added to trole of fiber in gastrointestinal function diet. Hemicellulose appears to be eand disease. Early determinations of ap- tensively degraded on a low fiber dieparent fiber digestibility in m an, probably 99 (6) and 96 (7), while cellulosebecause of inadequate fiber methods (1), digestion is less, 75 (6) and 80 (7).

yielded estimates of fiber digestibility Although these data suggest that fiberanging from 2 to 100 (2-4). Recently,the detergent procedures of Van Soest ——,,_., i.ii . i m Received for publication 12 May 1980.(5), which have gained acceptance in fiber , Supportedin part by NIH G™.AM 21712,The collegresearch have been USed tO quantitate Agricultural and Life Sciences University of W isconsin M adisonr»i • i.'l_'l'i_ 1C r\\ T\* i'l_'l-«_ r an

8/9/2019 J. Nutr.-1981-Slavin-287-97

2/11

288 SL AV IN , BR AU ER A ND M AR L ET T

is extensively degraded in the humangut, problems with the design and methods used in these studies leave this conclusion in question. The original neutraldetergent fiber (NDF) procedure does notadequately remove starch from humanfoodstuffs (10); yet, in only one of thesestudies (8) was an amylase modification(11) of the NDF procedure (5) employed.Sequentially determining acid-detergentfiber (ADF) on an NDF residue, in aneffort to obtain a truer estimate of hemi-cellulose (10), was only done in theCornell study (8). Finally, the amount oftime needed for any adaptation to a particular fiber intake is unknown, and todate, apparent fiber digestibility hasbeen assessed only during short-termstudies.

The purpose of the reported experiments was to estimate N DF, celluloseand hemicellulose digestibilities in human subjects consuming low and highcellulose diets. To overcome limitationsfound in previous fiber digestibi li ty studies, residual starch in NDF food residueswas removed by an a-amylase treatmentof the NDF residue; cellulose wasmeasured directly in food and feces toavoid problems3 with the sequentialNDF-ADF procedure. Each diet was consumed for about 1 month (20-30 days)to examine changes in fiber digestibilityas subjects adapted to a particular fiberintake, and the amount of undigestiblenitrogen in fecal NDF was quantitatedas a possible source of error in NDFdigestibility measurements.

M ATERI AL S A ND M ET HOD S

Eleven healthy women, ages 20-39years, of normal height and weightparticipated in two experiments. Informed consent was obtained from eachsubject and experiments had been approved by the College of A griculturaland Life Sciences' Committee on Research I nvolving H uman Beings, U niversity of Wisconsin-Madison. Experiment 1 was a 2-month study in whichseven women consumed a low cellulosediet for 1 month and then a high cellulosediet for an additional month. Experi

ment 2 was a 1-week study in which fourdifferent women ingested a semipurified,fiber-free diet to which 16 g/day non-nutri tive fiber (Solka Floe) was added.

Experiment 1

The two diets and experimental protocolof experiment 1 have been described (12).Briefly, the daily diet consisted of normal,typically-consumed food except for thebread which contained 38 g wheat gluten,40 g lactalbumin and 22 g casein. Because the low cellulose diet served as anexperimental diet of another study (13),it was high in protein (% total kcal:23% protein, 30% fat, 47% CH O). Fibersources of the low cellulose diet included three servings of fruits, four servings of vegetables and refined grains. Thediet contained the Recommended DietaryAllowance (RDA) of all nutrients exceptcalcium, which was 600 mg/day. The highcellulose diet was the same as the lowcellulose diet, except that 16 g/daySolka Floe (BW-200, Brown Co., Berlin,NH ) was added to the bread. Each dietwas ingested by the seven women for20-30 days.

Brilliant Blue dye was given orallybefore breakfast every 5 days as a markerfor separating feces into composites andfor estimating transit time. All stools werecollected, divided into 5-day composites,blended and freeze-dried. Four to six 5-day fecal composites were obtained fromeach subject during each diet. Gastrointestinal transit times, which have beenpreviously published (12), were determined as the time between ingestionand appearance of Brill iant Blue dye andas the time between ingestion and excretion of 80% of 20 small, radiopaquepellets. D ye was ingested by each subject 4-5 times (31 observations) duringthe low cellulose period and 5-6 times(40 observations) during the high celluloseperiod. Pellets were taken with the dye,twice during each diet period, exceptthat subject D ingested pellets onlyonce during the low cellulose diet period.

3 B ra uer , P . M . 1979) Ap pa ren t d igest ib ility of n eu tr al d eter gen t fib er in you n g an d eld er ly ad u lts. M aster s thesis,

U n iv er s i ty o f Wi s co n si n .

8/9/2019 J. Nutr.-1981-Slavin-287-97

3/11

DIETARY FIBER DIGESTIBILITY 289

Food composites for fiber analysis wereprepared during each dietary period andlyophilized to constant weight. Dietaryfiber in food was determined by a modification (14) of the Van Soest neutral-detergent fiber (NDF) procedure (5) andaveraged 9.5 g/day during the lowcellulose diet and 23.5 g/day during thehigh cellulose diet. The modified NDF(14) procedure includes glass wool as afiltering aid and an 18-hour incubationof N DF residue with hog a-amylase(#A6880, Sigma Chemical Co., St. Louis,M O.). Fecal NDF analyses were similarto the unmodified procedure (11), exceptthat glass wool was used as a filtering aid.

Cellulose was measured directly infood and feces by a modification (15)of the Crampton and M aynard procedure(16). Acetone extraction was substitutedfor the benzene, ether and ethanolextractions, a change which did notalter cellulose recovery. The low cellulosediet contained 5.4 g/day cellulose, thehigh cellulose diet, 19.3 g/day. To estimate hemicellulose in food and feces,cellulose values were subtracted fromthe respective NDF residue weights. Apparent digestibilities of ND F, celluloseand hemicellulose were calculated asintake minus output and expressed as apercentage of intake for four to six 5-dayperiods for each subject during eachdietary period.

In addition to the 5-day composites,aggregate composites to represent fecalexcretion throughout the two experimental diets were prepared for eachsubject by combining 2 g of each 5-day,dried composite and mixing. Hence, aggregate fecal composites represented thefecal excretions of each subject duringeach 20-30 day study. A pparent digestibilities of NDF, cellulose and hemicellulose were determined for the ag

gregate fecal composites. The nitrogencontent of the ND F and cellulose residuesof each aggregate composite was determined by the macro-Kjeldahl method(17). Fecal NDF and cellulose were corrected for protein [N x 4.3] (18) contamination and apparent digestibilitiesof NDF, hemicellulose and cellulosewere recalculated.

E xper imen t 2Four female subjects consumed a semi-

purified, liquid diet (Ensure, Ross Laboratories, Columbus, OH) along with16 g/day Solka Floe for 1 week. Averagedaily intake of the liquid diet was 1,200kcal (range 800-1,720 kcal/day). Intakeof non-fibrous carbohydrate sources (sodapop, fruit-flavored drinks, sugar, hardcandy) provided an additional mean 350kcal/day. Two of the four subjects consumed 6-12 ounces of a cola soda perday. Brilliant Blue dye was given on days2 and 8 of the study to permit collectionof one 5-day fecal composite from eachsubject. Fecal composites were lyophilized,fecal NDF and cellulose measured, andprotein contamination of fecal fiber residues determined, as described in experiment 1. The dietary fiber source, SolkaFloe, was analyzed for ND F and cellulosecontents which were 14.5 g and 13.5 gper day, respectively, and apparentdigestibilities of ND F and cellulose werecalculated.

Apparent fiber digestibility for eachsubject was calculated as the mean ofthe four to six 5-day composites collected during each of the two levels offiber intake. Intersubject and dietarydifferences were assessed by two-wayanalysis of variance with unweightedmeans (19). Between-diet and between-subject differences in fiber digestibilityof the aggregate composites also wereexamined by two-way analysis of variance.Correlations between fiber digestibilityand transit time were calculated. In all instances, a P value < 0.05 was consideredsignificant.

RESULTSNeutral detergent fiber wassignificantly

(P < 0.01) more degraded when the low

cellulose diet was consumed than whenSolka Floe was added to the diet (table 1).No differences in ND F digestibility amongsubjects were found. But, digestibilityvaried greatly among 5-day compositesin any subject (fig. 1). The variabilityin fiber digestibility among compositeswas less during the low cellulose period;coefficients of variation between 5-day

8/9/2019 J. Nutr.-1981-Slavin-287-97

4/11

290 SL AV I N, BRA UER A ND M A RL ET T

TABLE 1

Mea n digestibility ofND F in hea lthy women consuming low a nd high cellulose diets

Low cellulose High cellulose

Subject M ean1 Range M ean2 Range

ABCDEFG56.6 20.356.960.955.972.814.710.013.37.584.7

±.668.0±17.832.6-82.141.2-78.846.3-69.141.6-73.259.8-78.483.7-90.643.0-86.611.2

28.943.527.024.923.231.1-11.315.07.119.09.220.7

Mean ±SD 65.1 ±10.7 21.4 ±17.3

1 M ean  ±SD of five determinations except for Subjects A , C and D , where n = 4.of six determinations except for Subject C, where n = 5.

2 Mean ±so

measures of NDF digestibility rangedfrom 4.3% in Subject F to 35.7% inSubject A. More variation in apparentfiber digestibility among composites occurred when the high cellulose diet wasconsumed. Coefficients of variation ofgreater than 100% were seen in Subjects A and G, the two subjects inwhich negative digestibi lities of NDFwere observed (table 1).

No consistent trends in digestibilitywith time were seen (fig. 1). M eanapparent digestibility of NDF during thelow cellulose diet was 61% on day 5,after which it increased slightly butinsignificantly up to day 25. During thehigh cellulose diet mean digestibility ofNDF was 38% on day 5, decreasedto 10% on day 15 and appeared tostabilize at about 20% after day 20.

Digestibility of cellulose also decreasedsignificantly (P < 0.01) when Solka Floewas added to the diet while no significantdifferences among subjects were observed (table 2). Results were similarto those seen with NDF digestibility.Again, wide ranges of digestibility withina subject were observed. Negative digestibility of cellulose during the highcellulose diet was seen once in SubjectF, twice in Subjects A and D and 4 timesin Subject G; mean cellulose digestibilitywas negative only in Subject G.

Solka Floe ingestion had no significanteffect on apparent hemicellulose digestibility and no differences were seen

among subjects (table 3). Ranges ofdigestibility within a subject were againwide, but negative digestibility during a5-day period was seen only in SubjectG, during two of the six5-day HC periods.

To determine whether gastrointestinaltransit time was a factor responsiblefor the individual variability in fiber

( ME A N S O ) 3 8 3 0

8 0High

60 -

DA Y

Fig. 1 Changes in NDF digestibi li ty with timein seven women. The top graph shows NDFdigestibi li ty during the high cel lulose diet and thebottom graph, N DF digestibility during the lowcellulose diet. Mean ±SDvalues of NDF digestibilities (% ) for all subjects of each 5-day period aregiven at the top of each graph. Each subject isidenti fied by a letter at the right side.

8/9/2019 J. Nutr.-1981-Slavin-287-97

5/11

D IETARY FI BER D IGEST IBI LI TY 291

TABL E 2M ea n a p pa r en t c ellu lo se d ig es tib ility in h ea lth y wo me n c on su min g lo w a n d h ig h c ellu lo se d ie ts

Low cellulose High cellulose

Subject M ean1 Range M ean2 Range

ABCD EFG62.158.169.059.274.088.773.021.716.27.619.19.02.811.832.6-84.442.4-76.157.8-74.430.7-69±3.137.7±8.719.8±.915.3±2.516.1±.525.5±9.6-18.0±23.0-40.8-55.813.9-64.210.6-30.4-4.5-51.7

Mean ±SD 69.2 ±10.7 14.4 ±17.5

1 M ean  ±so of five determinations except for Subjects A , C and D , where n = 4.of six determinations except for Subject C, where n = 5.

2 Mean ±SD

digestibility, individual measurementsof transit time and fiber digestibilitydetermined during the same 5-day periodwere correlated (table 4). During thelow cellulose diet, NDF, cellulose andhemicellulose digestibilities correlatedsignificantly with dye transit time, whileNDF and cellulose digestibilities weresignificantly correlated to pellet transittime. N either dye nor pellet transittimes correlated with apparent fiberdigestibilities during the high cellulosediet.

To examine gastrointestinal transit timeas a possible cause of the variabilityin apparent f iber digestibility amongsubjects, mean transit times of the pelletsor dye of each subject were correlatedwith mean N DF, cellulose and hemi

cellulose digestibilities of each subject.Only mean cellulose digestibi lity andmean pellet transit time during thelow cellulose diet were signif icantlycorrelated (table 4).

The wide variability between repeatedfiber digestibility measurements in anindividual suggested that Brilliant Bluedye was an ineffective aid for dividingfecal samples into 5-day compositesrepresentative of the residue from fivedays of intake. Hence, aggregate fecalsamples, which represent 20-30 days ofexcretion for each subject were prepared.As expected, NDF digestibi lity determined from the aggregate fecal composites was similar to the mean NDFdigestibility of 5-day composites of eachsubject ( table 1, f ig. 2) . During the low

TABL E 3M ea n a p pa r e nt h em ic ellu lo se d ig es tib ili ty in h ea lth y w om en c on su min g lo w a n d h ig h c el lu lo se d ie ts 1

Low cellulose High cellulose

Subject M ean2 Range M ean3 Range

ABCDEFGMean

±SD47.3

23.964.618.350.114.751.534.570.89.678.07.761.526.160.5

±11.524.8-79.039.5-82.431.2-66.14.6-77.659.8-81.971.2-91.221.9-86.133.3

14.765.911.855.27.964.915.650.414.750.2

±9.48.7±2.746.9±20.118.1-51.147.8-78.145.1-67.055.1-88.928

1 Fecal and food hemicellulose determined as N DF minus cellulose. 2 M ean  ±SD of five determinations except for Subjects A , C and D , where n = 4. s M ean  ±SD of six determinations except forSubject C, where n = 5.

8/9/2019 J. Nutr.-1981-Slavin-287-97

6/11

292 SL AV I N, BRA UER A ND M A RL ET T

TABLE 4Rela ti onsh ip s between ap pa re nt fi be r d iges ti bi li ti es

a nd g astr oin testin al tr a ns it time s

Low cellulose diet High cellulose diet80% 80%

pellets pelletsTT1 dye TT TT dye TT

c or r el a ti on c oe ffi ci en ts o f a l l o bs er va t wn s

Fiber fraction (n = 13) (n = 3 1) (n = 14) (n = 40)NDF 0.559- 0.459 0.008 0.139Cellulose 0.573' 0.362- 0.045 0.165Hemicellulose

ND FCellulose em i ellulose

0.383 0.399- -0.158 -0.055

co rr ela tio n co effic ien ts o f mea n d ata n = 7 )

0.6990.765-0.436

0.7390.7410.625

0.188 0.2820.207 0.330

-0.094 0.023

1T T = transit time. •ignif icant at P < 0.05. Signif icant atP < 0.0 1.

cellulose diet, NDF digestibility was59.4 ±11.2% when aggregate fecal composites were used to calculate digestibility, compared to 65.1 ±10.7% (table 1)with 5-day composites. W hen Solka Floewas ingested, NDF digestibility averaged 19.7 ±16.2% with aggregate composites and 21.4 ±17.3% with 5-daycomposites. Again, no significant difference in digestibility among subjects wasseen.

Fecal NDF from the aggregate samples contained 6.23 ±1.09%nitrogen during the low cellulose diet and 0.92±0.15% nitrogen during the high cellulosediet. When nitrogen was converted toprotein with the factor 4.3 (18), the protein content of the fecal NDF fractionsfrom the low cellulose diet ranged from

AFDUncorrected 4911 5238 5625 4924 6822 7930 63-12Corrected 6614 6340 6928 6426 7726 8333 71-6

28 i-APPARENTDFNECESg/

8/9/2019 J. Nutr.-1981-Slavin-287-97

7/11

DIETARY FIBER DIGESTIBILITY 293

TABLE 5D ig estibility o f c ellu lo se a nd h emice llu lo se d ete rmin ed fr om a gg re ga te c omp os ites

Subject

M eans:

Cellulose Hemicellulose

LC HC LC2 LC3 HC2 HC3

ABCDEFC65.658.165.662.074.190.771.94.535.523.419.418.326.5-17.627.644.443.731.060.263.251.266.170

69.7 ±10.7 45.9 ±13.5 71.7 ±5.4 30.7 ±12.8 51.0 ±7.9

1Fiber analysis done on aggregate fecal composites representing 20-30 days of excretion by eachsubject during each diet. 2Hemicellulose determined by difference (NDF minus cellulose). 3Fecalhemicellulose determined by difference (ND F corrected for protein contamination minus cellulose).

21 (Subject F) to 33% (Subject A). Protein contamination of fecal ND F residuesfrom the high cellulose period was less(about 3-4%), although the absoluteamount of protein in the fecal NDF fraction was the same for both diets (~1g/day). W hen the fecal N DF resultsfrom the low cellulose diet were corrected for protein (fig. 2), digestibilitiesof ND F increased significantly (P < 0.01).Digestibility increased the most in Subject A and the least in Subject F (fig. 2) ,while mean apparent NDF digestibility

of the group increased from 59.4 ±11.2%to 70.4 ±7.3% with the correction. Correcting the fecal NDF residues from thehigh cellulose diet for protein contenthad little effect on digestibility, whichwas19.7±16.2%without and23.0 ±15.0%with correction (fig. 2).

Apparent cellulose digestibility determined from the aggregate fecal samples (table 5) agreed well with the meansof cellulose analyses performed on the 5-day composites (table 2). The effect ofdiet on cellulose digestibility was significant (P

8/9/2019 J. Nutr.-1981-Slavin-287-97

8/11

294 SL AV IN , BRA UER A ND M A RL ET T

16

1 4

*? .o 1 2

O C IOL

u . 8

3 eU

- 15.311.6 7 .1 0 .5 15 .5 14 .1 2 .06 .0 Appa re nt Dig e stib ility )

NDF

CELLULOSE

1 2 3 4

S U B J E C TFig. 3 Fecal fiber excretion per day when Solka Floe was the sole fiber source in a semipurified diet

(experiment 2). T he left bar represents N D F excretion (0) and the right bar represents cellulose excretion (D). Apparent digestibi li ties (%) of NDF and cel lulose are given above each bar. Daily fecal f iberwas determined as f iber excreted during a 5-day col lection period divided by 5.

DISCUSSION

Cummings (20) has suggested that atleast half of the fiber in a low fiber,Westernized diet is degraded in thehuman gut. With our low cellulose diet,digestibility of NDF calculated fromaggregate composites was 59% and 70%when fecal NDF residues were correctedfor protein. M ean N DF digestibility calculated from 5-day composites was 65%during the low cellulose period. Otherresearchers (6-8) have reported digestibility of NDF to be about 80% when subjects consumed a low fiber diet. FoodNDF residues were not treated withamylase in two of these studies (6, 7);hence, digestibi lity estimates are probably inflated. Indeed, Farrell et al. (6)recognized that the food NDF was probably contaminated with starch and analyzed the food with an enzymaticmodification of the NDF procedure (21).

Calculating fiber digestibi li ty using thislower food fiber value yields an NDFdigestibility of 73%, as compared to80% when no amylase modification forfood NDF was used. Fiber digestibilityduring a low fiber diet, as measuredwith the Southgate method (22), whichmeasures both water-soluble and insoluble fiber, ranges from 70 to 80%(23). Hence, fiber digestibility data generated using available fiber methodsagree that more than half of ingestedfiber is degraded when subjects consume a low fiber diet.

Digestibility of NDF decreased significantly when Solka Floe was added tothe diet. Fiber digestibility also decreases when bran is added to a lowfiber diet (6). N either bran nor refinedcellulose appear to be highly fermentedin human subjects (24), so this apparentdecrease in fiber digestibility is notunexpected. Whether the digestibi li ty of a

8/9/2019 J. Nutr.-1981-Slavin-287-97

9/11

D IETARY FI BER D IGEST IBI LI TY 295

more fermentable fiber source decreasesalso as it is consumed in increasingamounts has never been examined. Further, it is unclear what effect either afermentable or an unfermentable fibersupplement would have on the apparentdigestibility of the food-derived fiber.

Large within-subject variations in fiberdigestibility were observed in the present study and that of Hummell (3), butno consistent trends in digestibility withtime were evident in either study. Thenegative fiber digestibilities during thehigh cellulose diet, though, suggest thatwide within-subject variations may becaused in part by the inability of dyeto adequately divide fecal samples intocomposites corresponding to a 5-dayintake. During the low cellulose diet,Subject F's feces were divided withoutthe aid of a marker. Yet the most consistent fiber digestibility among composites was observed in Subject F during the low cellulose period (mean:84.7 ±3.6%).

Reasons for the consistently negativefiber digestibility in Subject G areobscure. During the high cellulose intake Subject G's feces were about 50%dry matter, while the feces of all othersubjects were about 30% dry matter

(12). Yet, approximately 45% of the drymatter of all feces from all subjects,including G, during the high cellulosediet was NDF. Thus, Subject G appearedto excrete almost twice as much NDFas the other subjects during the highcellulose diet.

Another source of within-subject variation in fiber digestibility may be theability to adapt to a particular diet. M eandigestibility of NDF with the cellulosesupplement was 38% on day 5, decreasedto 10% on day 15 and then stabilizedat about 20%. An individual's microflora

react differently to the ingestion of anovel fiber source (23) and thus, theamount of time needed for any adaptation to a particular fiber intake is notknown. Recent evidence (25, 26), though,implies that the microflora are not influenced by the fiber composition of thediet.

Although statistically we could not

detect differences in fiber digestibilityamong subjects, mean ND F digestibilitycalculated from 5-day composites, rangedfrom 56 to 85% during the low cellulosediet and from -11 to 43% during thehigh cellulose diet. Wide ranges of fiberdigestibility have been previously described (3, 8, 27) and may be due todifferences in response of an individual's microflora to a fiber source (23) orto differences in gastrointestinal rateamong individuals (8). Others, however,find no relationship between fiber digestion and transit time when pectin is thefiber source (28). I n our study whenapparent fiber digestibility was low (HCperiod), fiber digestibility and transittime were not correlated, whereas whenmore than half of the fiber was digested(LC diet), fiber digestibility and transittime may have been related. T hus, itappears that the relationship betweengut transit and fiber digestibil ity maydepend on either the level or the kindof fiber.

Recent work (6, 7, 9) lends support tothe early observation by Williams andOlmstead (4) that hemicellulose is moredigestible than cellulose in human subjects. Because these investigators (6, 7, 9)did not sequentially determine ADF

on an NDF sample to eliminate pectins(10) and also did not treat the foodN DF with amylase, it is unlikely thathemicellulose and cellulose measures infood and feces are reliable. Van Soest(8) found that cellulose in a low fiberdiet and also cellulose in a cabbage dietwere more digestible than the hemicellulose. Our results suggest that whenfood NDF starch and fecal NDF nitrogenare considered in the calculation, cellulose and hemicellulose have comparabledigestibilities, about 70% (table 5) whena low fiber diet is ingested.

Estimates of hemicellulose in thepresent study also may not be reliable.Because NDF contains hemicellulose,cellulose and lignin and the Cramptonand M aynard method isolates only cellulose, calculated hemicellulose valuesinclude lignin. Human foodstuffs arelow in lignin and Solka Floe containsless than 1% lignin so the error would

8/9/2019 J. Nutr.-1981-Slavin-287-97

10/11

296 SL AV I N, BRA UER A ND M A RL ET T

be small. But, correcting for the protein in NDF before calculating hemi-cellulose had a marked effect on hemi-cellulose digestibi li ty, increasing it from46 to 71% on the low cellulose dietand from 31 to 51% with the addedcellulose. These conflicting results couldbe caused by the crude gravimetricschemes for quantitating hemicelluloseand cellulose and could illustrate theneed for a better means of separatingand characterizing the components offiber.

Comparison of apparent cellulose digestibility determined when a solidfood diet was consumed, 14% (experiment 1, table 2) versus the consumption of a liquid diet, 8% (experiment 2,figure 3), suggests that little Solka Floeis degraded, independent of the form ofthe rest of the experimental diet. Otherstudies (8, 29) also have reported thatdigestion of Solka Floe is much less thanthat of fiber in foods (6-8). Refinedcelluloses are chemically processed andhence differ markedly from natural cellulose in physical and biological properties(24). Van Soest's preliminary data (8)suggest that Solka Floe may have a negative and inhibitory effect upon fecalmicrobial action. Prolonged feeding ofSolka Floe did not induce cellulolyticfermentation but appeared to depressfiber digestibi li ty during the sequentialdiet (8).

The fecal NDF fractions collectedduring the low and high cellulose dietsin experiment 1 contained a significantquantity of nitrogen while the fecal NDFfractions collected when Solka Floe wasingested with a semi-purified diet contained minimal nitrogen. These resultssuggest that the nitrogen found in fecalNDF residues originated from foodsources rather than from endogenous gutsecretions. Probably, the nitrogen is theM aillard product, an indigestible compound formed during the non-enzymaticbrowning reaction between amino acidsand the degradation products of sugars(30). A lthough M aillard products are asmall part of most diets, they are indigestible and hence become concentrated in the feces. The Maillard producthas a high nitrogen content, about 11%

(31); thus, if the nitrogen in the fecalN DF fraction is the M aillard product,the factor used to convert nitrogen toprotein, 4.3, would underestimate non-

fibrous, nitrogenous contamination ofthe fecal NDF samples.W hen fecal NDF values from the ag

gregate samples during the low celluloseperiod we corrected for protein, NDFdigestibility was significantly increasedfrom 60 to 70%. Because of the substantial ly larger proportion of fecal NDFduring the high cellulose period, correction for fecal NDF protein had virtually no effect on apparent NDF digestibility, which for the group was 20%without and 23% with correction. Thevalidity of correcting for the nitrogen inNDF, however, can be argued. The Mail-lard product is indigestible and is isolated in the lignin fraction during fiberanalysis (31). T he nitrogenous matrixisolated with fiber in the form of Maillardproducts (1) or as indigestible protein(32) may be responsible for some of theeffects associated with fiber in the gut.True lignin contains no nitrogen, though,and is in low concentration in mosthuman foodstuffs, so other fiber chemistsfeel that artifact lignin should not beisolated as dietary fiber (33).

Independent of this controversy, however, the findings of this study suggestthat substantial quantities of unprocessedfood fiber are digested during transitthrough the human gastrointestinal tractwhereas little or no digestion of SolkaFloe occurs.

ACKNOWLEDGMENTS

The authors acknowledge with appreciation D ee K iefer for technical assistance, Professor Norman Draper forstatistical advice and the 11 women whoparticipated in the study.

L IT ERATU RE CI TED1. Van Soest, P. J. & Robertson, J. B. (1977)

Analytical problems of fiber. In: Carbohydratesand Health (Hood, L. F., ed.), pp. 69-83,Avis, Westport, CT.

2. M angold, E. (1934) T he digestion and utilization of crude fiber. N utr. A bs. Rev. 3, 647-656.

3. H ummell, F. C., Shepherd, M . L . & M acy, I . G.(1943) Disappearance of cel lulose and hemi-

8/9/2019 J. Nutr.-1981-Slavin-287-97

11/11

D IETARY FI BER D IGEST IBI LI TY 297

cellulose from the digestive tracts of children.J. Nutr. 25, 59-70.

4. W illiams, R. D. & Olmstead, W. H. (1936)The effect of cellulose, hemicellulose and ligninon the weight of the stool: a contributionto the study of taxation in man. J. N utr. 11,433-449.

5. Goering, H . K . & Van Soest, P. J. (1970) Forage fiber analysis. Agriculture Handbook #379,A gr. Res. Serv., U SD A, U .S. Govt. PrintingOffice, Washington, DC.

6. Farrell, D. J., Girle, L. & Arthur, J. (1978)Effects of dietary fibre on the apparent digestibi li ty of major food components and on bloodlipids in man. Aust. J. Exp. Biol. M ed. 56,469-479.

7. Holloway, W. D., Tasman-Jones, C. & Lee, S. P.(1978) Digestion of certain fractions of dietaryfiber in humans. Am. J. Clin. Nutr. 31, 927-930.

8. Van Soest, P. J., Robertson, J. B., Roe, D . A .,Rivers, J. & Hackler, L. R. (1978) Therole of dietary f iber in human nutrition. Proc.Cornel l Nutr. Conf. 5-12.9. Gramstorff-Fetzer, S., K ies, C. & Fox, H . M .(1979) Gastric disappearance of dietary fiberby adolescent boys. Cereal Chem. 56, 34-47.

10. Bailey, R. W., Chesson, A . & M onro, J. (1978)Plant cel l wal l f ractionation and structuralanalysis. Am. J. Clin. Nutr. 31, S77-S81.

11. Robertson,] .B . (1978) The detergent systemof fiber analysis. I n: Topics in D ietary FiberResearch (Spiller, G. A., ed.), pp. 1-42,Plenum Press, New York.

12. Slavin, J. L. & M arie«, J. A. (1980) Inf luence of ref ined cellulose on human bowelfunction and calcium and magnesium balance.Am J. Clin. Nutr. 33, 1932-1939.

13. Foley, P. & Linkswiler, H. M . (1977) Theeffects of level of protein intake on urinarycalcium in the premenopausal women. Fed.Proc. 36, 1122.

14. Schaller, D . (1977) Enzyme-modified N DFmethod. I n: T he N utritional Significance ofD ietary Fiber (K imura, K . K ., ed), Bureau ofFood, FDA, Washington, DC.15. O'Shea, J., do V alle Riberio, M . A. &M oran, M . A . (1968) Relationship betw eendigestibility (in vitro), crude fiber, and cellulosecontent of some animal feeds. I r. J. A grie.Res. 7, 173-181.

16. Crampton, E. W. & M aynard, L. A. (1938)T he relation of cellulose and lignin contentto the nutri tive value of animal feeds. J. Nutr.1 5 3 83 †” 395

17. Fleck, A. & M unro, H. N. (1965) The determination of organic ni trogen in biologicalmaterials. A review. Clin. Chem. A cta. 11,2-12.

18. Dintzis, F. R., M cBrien, J. B., Baker, F. L.,Inglett, G. E., Jacob, R. A ., M uñoz, J. M .,Klevay, L. M ., Sandstead, H. H. & Shuey,N. C. (1979) Some effects of baking andhuman gastrointestinal action upon a hardred wheat bran. In: D ietary Fibers: Chemistry and Nutri tion (Inglett, G. E. & Falkehad,S. L, à ©d.),pp. 157-171, A cademic Press,New York.

19. Winer, B. J. (1971) Statistic Principles inExperimental Design, 2nd ed., McGraw-Hil l,New York.

20. Cummings, J. H . (1978) N utritional implications of dietary fiber. A m. J. Clin. N utr. 31,S21-S29.

21. A ngus, R., Sutherland, T. M . & Farrell, D . J.(1977) A simpl if ied method for determiningfibre in foods. Proc. Nutr. Soc. Aust. 2, 90.

22. Southgate, D. A.T. (1976) The measurementof unavailable carbohydrates: structural poly-saccharides. In: Determination of Food Carbohydrates, pp. 61-74, Applied Science, London.

23. Prynne, C. J. & Southgate, D. A. T. (1979)T he effects of a supplement of dietary fibreon faecal excretion by human subjects. Br. J.Nutr. 41, 495-503.

24. Van Soest, P. J. (1978) D ietary fibers: theirdef ini tion and nutritional properti es. Am. J.Clin. Nutr. 31, S12- S20.

25. D rasar, B. S., Jenkins, D . J. A . & Cummings,J. H. (1976) The influence of a diet rich inwheat fibre on the human fecal flora. J. M ed.Microbio . 9, 423-431.

26. M cL ean-Baird, R. L ., Walters, R. L ., D avies,P. S., Hill, M . J., Draser, B. S. & South-gate, D. A.T. (1977) The effects of two dietaryfiber supplements on gastrointestinal transit,stool weight and frequency, and bacterialf lora and fecal bile acids in normal subjects.Metabolism 26, 117-127.

27. I smail-Beigi, F., Reinhold, J. G., Faraju, B. &A badi, P. (1977) Effects of cellulose addedto diets of low and high fiber content uponthe metaboli sm of calcium, magnesium, zincand phosphorus by man. J. Nutr. 107, 510-518.

28. Cummings, J. H., Southgate, D. A. T., Branch,W. J., Wiggins, H . S., H ouston, H ., Jenkins,D. J. A., Jiuraj, T. & Hill, M . J. (1979) Thedigestion of pectin in the human gut and itseffect on calcium absorption and large bowelfunction. Br. J. Nutr. 41, 477-485.

29. M ickelson, O., M akdami, D . D ., Cotton, R. H .,Titcomb, S.T. , Colmey, J. C .& Gatty, R. (1979)Effects of a high fiber bread diet on weightloss in college-age males. A m. J. Clin. N utr.32, 1703-1709.

30. Van Soest, P. J. (1965) U se of detergents inthe analysis of fibrous feeds. II. Study ofeffects of heating and drying on yield of f iberand lignin in forages. J. Assoc. Off. Agr.Chem. 48, 785-790.

31. Van Soest, P. J. & Robertson, J. B. (1976)

Chemical and physical properti es of dietaryfibre. M iles Symposium on D ietary Fibre,Nutrition Society of Canada.

32. Saunders, R. M . & Betschart, A. A. (1980)The signi ficance of protein as a componentof dietary fiber (letter). A m. J. Clin. N utr.33, 960-961.

33. Southgate, D. A.T., Hudson, G. J. & Englyst, H.(1978) T he analysis of dietary fibre—thechoices for the analyst. J. Sci. Fd. Agrie.29, 279-288.