Determination of carbohydrates in foods. I.-Available

DETERMINATION OF CARBOHYDRATES IN FOODS 1.-Available carbohydrate By D. A. T. SOUTHGATE A procedure is described for measuring the available carbohydrate in foods. It permits the measurement of the separate carbohydrate species which make up the available carbohydrates. The recovery of carbohydrates added to natural mixtures is within 3 % of the theoretical value.

Introduction The standard method for the determination of carbohydrates in foods is ‘by difference’, that is by deducting the sum of the measured moisture, ash, protein (calculated from the total nitrogen) and fat from the total weight. The value obtained in this way is modified by the determination of ‘crude fibre’.l This procedure provides a reasonably rapid and reproducible method, and in many circumstances the results obtained are sufficiently reliable. There are, however, many objections to its general use, particularly when applied to food materials. The value for carbohydrate ‘by difference’ includes all types of carbohydrate, from simple sugars to complex heteropolysaccharides, in addition, to other substances such as organic acids and lignin. It also includes any errors associated with the measurement of the other four constituents. In particular the use of a factor to convert total nitrogen to protein may produce considerable inaccuracy unless the true percentage of nitrogen actually present in the protein is known.2 To these technical objections must be added the physiological one that the different polysaccharides are utilised by Man to different extents. Furthermore, other studies3-G show that it is not sufficient to regard all available carbohydrates in the human diet as being metabolically or physiologically identical. Several groups of workers have proposed schemes of analysis for measuring the different carbohydrates in some foods.7-12 All the previous studies were, however, limited to certain classes of foodstuff and are not generally applicable to all foods or to homogenates of a mixed diet. The procedures described here were developed during studies on the availability of different nutrients in the human diet13-l5 and were designed to measure ‘available carbohydrate’16 i.e. the sum of glucose, fructose, sucrose, maltose, lactose, dextrins, starches and glycogens in the sample. The methods were used for the analysis of the foods whose composition were reported by McCance & Widdowson for the first time.17

Reducing sugars These were measured by the Munson-Walker method.lg The copper oxide was collected on asbestos in a Gooch crucible, dissolved in ferric sulphate solution and titrated with standard KMn04, using ferrous phenanthroline as indicator. The sugar values were calculated on the basis of the paper chromatographic examination of the sample using the revised Hammond table.19

Experimental Analytical All reagents were A.R. where obtainable. The sugars used as reference substances were dried under reduced pressure over Pzo5 at 37“ and used as solutions in saturated benzoic acid. Maize starch was purified by exhaustive extraction with hot 85 % (by vol.) methanol;l8 dilute acid hydrolysis of the extracted material produced glucose as the only sugar in theoretical yield.

Fractionation scheme The scheme used is described diagrammatically in Fig. 1.

Hexoses Hexoses were measured by the anthrone procedure20 modified by using a 0.2% (wt./vol.) solution of anthrone (recrystallised from ethanol) and standard glucose solutions containing 10, 20 and 30 pg/ml. When galactose was being measured the values obtained from the glucose standard curve were corrected by multiplying by 1 .85.20 Glucose Glucose was measured by the glucose-oxidase method of Marks.21 Paper chromatography This was carried out by the descending method using Whatman No. 1 paper and either ethyl acetate-acetic acidwater (3 : 1 : 3 by vol.) or butanol-acetic acid-water (4 : 1 : 1 by vol.). Aniline phthalateZ2 was used to detect the sugars. Examination of the paper was made under U.V. light. Removal of fermentable sugars Fermentable sugars were removed by taking a portion (50 ml) of the solution neutralised to pH 7-7.5 and adding 10 ml of a 10%(wt./vol.) suspension of washed baker’s yeast. The mixture was covered and incubated at 37“ for 18 h, after which time the mixture was centrifuged and sugars in the supernatant were measured.

Extraction of sugars and preparation of polysaccharide-containing fraction Two alternative and equivalent procedures may be usedthe choice depending on the state of the sample. The first procedure is applicable to most foodstuffs; however, if the J. Sci. Fd Agric., 1969, Vol. 20, June

Southgate: Determination of Carbohydrates in Foods.

I

327

Sample

Extraction with aqueous alcohol

I

Residue

Exiract

Extract with diethyl ether; air-dry

Examine by paper chromatography

I

I

~~

I

I

II

II

I

Remove alcohol Treat with lead acetate

Starch and dextrins only present

I

Measure reducing sugars

Dilute acid hydrolysis neutralise. Measure reducing sugars or hexoses by anthrone method = Starch (as ducose) .

Y

I

I

Other golykaccharides present

Incubate with Takadiastase. 18 h, 37"c at pH 4.5. Add 4 vols ethanol Filter Filtrate

Measure hexoses

E

Starch (as glucose)

Measure glucose by oxidase method

I = Glucose

If lactose is

present treat with yeast. Measure residual sugars =Lactose

I

Fructose = Total(glucose + lactose) Invert with llCl

Dilute, 1 asure anthrone method

= Total sugars

If lactose is

present treat with yeast. Measure residual hexose =Lactose Fig. 1 . Diagrammatic description of the scheme of analysis

sample contains a large amount of water (over 90%) the amounts of methanol that have to be added to the sample may make the procedure technically cumbersome and in these cases the second procedure should be used. Procedure I (for samples containing less than 90 % moisture)

A sample of 4-5 g was weighed into a 100 ml conical flask, extracted with four 25 ml portions of boiling 85% (by vol.) methanol, and was then filtered hot into a volumetric flask. The filtrate was made up to 100 ml and the residue was washed with diethyl ether and allowed to dry in air. When the sample contained more than 25 % moisture, the concentration of methanol in the first extraction should be increased so that the concentration of the methanol is maintained at 85 % (by vol.) and not reduced by the moisture in the sample. Procedure 2 (for samples containing more than 90 % moisture)

A sample of between 10-20 g was weighed out into a 250 ml flask marked at 100 ml. Ethanol and water were added to bring the concentration of ethanol to 50% (by vol.) in a

-

J. Sci. Fd Agric., 1969, Vol. 20, June

total volume of 100 ml. The flask was then heated on a boiling water bath for 1 hour with a small funnel in the neck of the flask. The contents were allowed to cool and the volume was adjusted to 100 with 50% (by vol.) ethanol. Ethanol (100 ml) was. then added and the mixture was thoroughly mixed. The supernatant was filtered on a Buchner funnel and the residue was washed with several small portions of 80% (by vol.) ethanol. The combined filtrate and washings were made up to volume. The residue was washed with diethyl ether and allowed to dry in air. The residue obtained by either procedure was heated in an air-oven to remove the last traces of solvent, and allowed to come into equilibrium with the laboratory atmosphere before being weighed. It was then finely ground and mixed. Measurement of sugars in the alcoholic extract Two alternative procedures are described. The first is more time-consuming but gives greater precision and provides information on the proportion of sucrose present in the extract. The second is very much more rapid but only provides values for total sugars.

328

Southgate: Determination of Carbohydrates in Foods. I

The extract was first examined by paper chromatography to determine the sugars present, since the subsequent procedure and interpretation of the results could only be determined adequately in the light of this information. Procedure I A portion of the extract was evaporated by gently heating to remove alcohol in a rotary evaporator. The aqueous residue was transferred with water to a volumetric flask, 0 - 1 ml of a saturated solution of lead acetate was added, and the solution was made up to 100 ml and mixed. After 15 minutes the precipitate was filtered off (using a dry paper) and the filtrate was tested for lead. If any lead was present it was precipitated with solid Na2C03 and the precipitate was removed by filtration. The lead-free filtrate was used for the determination of reducing sugars before and after inversion, and for glucose if required. Inversion was carried out by adding 5 ml conc. HCl to 50 ml of the filtrate and leaving it overnight in a room maintained at 20" or above. The inverted solution was neutralised with N a ~ C 0 3before reducing sugars were measured. If paper chromatography showed that lactose was present in addition to reducing monosaccharides, a portion of the inverted solution was treated with yeast to remove fermentable sugars, and the unfermentable sugars were measured. Glucose was measured in the lead-free filtrate before inversion, using the glucose-oxidase procedure of Marksz1 and the fructose was then estimated by difference. Procedure 2 The alcoholic extracts were diluted to give concentrations within the range described above for the anthrone method, and hexose sugars were measured directly on the dilutions. If lactose is present the hexose in a yeast-treated extract must also be measured. When the concentration of sugar in the extract was very low it was desirable to prepare a series of glucose standards for the anthrone method which contained alcohol at the same concentration as present in the diluted extracts. Extracts which are coloured must be treated with lead acetate solution during the course of their dilution and filtered before being carried through the method.

Measurement of starch, dextrins and glycogen in the residue insoluble in aqueous alcohol Two alternative procedures can be used depending on the types of polysaccharide present in the sample. Procedure I When only starch and dextrins are present, for example when highly refined cereal products are being analysed, this more rapid method may be used. It may also be used when other polysaccharides are present if a lower accuracy can be tolerated. A portion of the air-dried sugar and fat-free residue (300400 mg) was weighed into a round-bottomed 1 1 flask and 450 ml of 1 % (by vol.) HzS04 were added. The mixture was boiled under reflux for 4 h, and the flask was swirled at intervals to carry down particles of sample adhering to the flask above the level of the acid. The hydrolysate was cooled, partly neutralised with 50% (wt./vol.) NaOH and made up to 500 ml. A portion was then filtered and completely neutral-

ised with solid Na~C03,and the reducing sugars were measured. If chromatography shows that pentoses are present in the hydrolysate a portion should be treated with yeast before the pentoses are measured. Procedure 2 When polysaccharides other than starch, dextrins or glycogen were present, as for example in most plant products, a portion of the sugar and fat-free residue containing 100-200 mg of starch was weighed into a flask and the sample was gelatinised with hot water (10 ml), acetate buffer, 2 M pH 4 . 5 (0.3 ml) was added, followed when cool by 5 ml of 10% (wt./ vol.) Takadiastase (Parke Davis) and a few drops of toluene. The mixture was incubated at 37" overnight (18 h). Ethanol (60 ml) was added to the flask and the contents were mixed. The mixture was then transferred to a volumetric flask (100 ml) with 80% (by vol.) ethanol and made up to volume. The precipitate was allowed to settle and a portion of the supernatant was filtered and used for measurement of reducing sugars after removal of ethanol as described above or of hexoses by the anthrone method after suitable dilution. The filtrate was also examined by paper chromatography. If the extract contained any sugars in addition to glucose and traces of maltose the determinations should be repeated with a different batch of Takadiastase. N

Results Where the alternative procedures were applicable, comparison showed that identical results were obtained provided that usual analytical precautions were taken. The procedures have been used with nearly all classes of foodstuff, including mixed diets, and have been found to be technically applicable to all so far examined. Completeness of extraction Chromatographic studies of aqueous extracts of the residue after extraction by both procedures shows that the residue contains no simple sugars. The alcoholic extracts contain all the mono- and di-saccharides present in the sample; recovery tests described below show that very little inversion of sucrose occurs during the extraction, but when the sample contains organic acids it may be desirable to adjust the pH of the mixture of sample and alcohol to between 6 and 7, before heating them together. Specificity of the methods The Munson-Walker method for reducing sugars appears to be the least affected of all such methods by non-sugar reducing substances, and this appears to be the general experience of other workers.19 When extracts from a range of foods were treated with ion-exchange resins to remove all polar substances, only those from coffee showed any significant lowering of its reducing value. Meat products known from chromatographic evidence to contain no sugars gave no reduction when carried through the entire procedure. Interpretation of the reduction values in terms of sugars is aided by the knowledge of the qualitative composition of the solutions being analysed, and visual estimation of the proportion of glucose to fructose is adequate in most cases. In situations where it is important to know the ratio of glucose to fructose the glucose is determined by the glucose-oxidase


Southgate: Determination of Carbohydrates in Foods. I methodzl and the fructose calculated by difference. Table I shows that the procedure gives accurate recovery of glucose in the presence of sucrose and fructose. Only oxidase preparations without invertase activity should be used. The anthrone method for hexoses has provided reliable in routine use. Glucose, fructose, sucrose and maltose give identical colour yields after correcting for the hydrolysis of the disaccharides, galactose gives 54% of the colour yield of the glucose, and lactose gives a colour yield equivalent to an equimolar mixture of glucose and galactose. The enzyme preparation (Takadiastase for analysis on talc, Parke Davis; preparations on lactose should not be used) has shown consistent behaviour over the past 10 years in convertTABLE I Recovery of glucose from mixtures of sucrose and fructose using the glucose-oxidase method ~~

Mixture Sucrose and fructose (20 mg/ml; 5 mg/ml) Fructose, 50 rng/rnl Fructose, 5 rng/ml Fructose, 0 . 5 mg/ml

Glucose added, mg/ml

Glucose recovered,

5

5.04 4.89 4.97 5.05

5

5 5

mg/ml

329

ing starch, dextrins and glycogen to glucose, giving about 98 % conversion during overnight incubation at 37” in 0.04 M acetate buffer (pH 4.5). Most of the remaining 2 % is maltose. The preparation does not produce free sugars from samples of apple and orange pectin and various plant gums or inulin under the conditions described. Furthermore no sugars other than glucose and maltose have been observed in hydrolysates of the mixed polysaccharides from a wide range of foods.

Recovery of added sugars It is extremely difficult to devise a rigorous test for a procedure such as that described. Recovery tests have the disadvantage that the added substances are of necessity purified and may be in a different physical condition from those present in the sample. Table I1 presents the results of some recovery experiments where various carbohydrates were added to mixtures chosen to contain possible interfering substances. In all cases except one, when an unextracted sample of starch was used, recovery was within f 3%. Table 111 shows that when the results obtained with the additions to the mixture of soya flour and milk powder are expressed as a proportion of the total available carbohydrate there is good agreement between analytical and theoretical values.

TABLE 11 Recovery of added carbohydrates

In each case the carbohydrates shown were added to the basal mixture and the two mixtures analysed - the difference between the values obtained on the basal mixture alone and with added carbohydrates is taken as the recovery of the added constituent. Basal mixture Rye flour, soya flour, ground nuts and milk powder

Soya flour and milk powder

Carbohydrate added Found by Amount R e c y w , A analysis added arabinose sucrose inulin starch (unextracted) available carbohydrate’ glucose starch ‘available carbohydrate’ lactose

Soya flour and milk powder (3 different mixtures) glucose and fructose sucrose

Rye flour

9.3

I .o

1.0 8.7 8.6

14.4

15.7

92*

23.7 4.9

24.4 5.0

97 97

9.9

9.8

101

14.8 5.2 4.0 3.2

14.8 5.2 3.9 3.1

100 100 103 102

9.9 13.5 15.0

9.7 13.6 14.5 15.2 15.8 14.9 30-1 31 .O 34.5

102 99 104 102 103 102 100 100 98 103 99 96

15.5

starch

16.2 15.0 30.0 30.9 33.6

lactose sucrose starch

5.1

5.0

14.9 23.2

15.0 24.1

* This starch was later found to contain free sugars J. Sci. Fd Agric., 1969, Vol. 20, June

100 106

330

I

Southgate: Determination of Carbohydrates in Foods.

TABLE 111

Composition of the total available carbohydrate Comparison of the distribution as found in recovery experiments with theoretical values, i.e. as added to the basal mixture of soya and milk powder Total available carbohydrate, % -

Lactose Glucose and fructose Sucrose Starch

Mixture 1

Mixture 2

Mixture 3

-

-

Found

Added

Found

Added

Found

Added

10.3 19.5 30.6 59.3

10.3 19.3 30.3 59.9

6.2 20.9 25.1 47.8

6.1 21.1 24.6 48.2

4.8 22.5 22.5 50.3

4.1 21.6 22.2 51.5

Comparison with other methods Few studies of the measurement of available carbohydrate lend themselves to comparison with the procedures described, but with cereal products they gave results exactly similar to those obtained by Fraser & H0lmes.~3 Discussion Any analytical procedure for available carbohydrates must of necessity represent a compromise between the ‘ideal’ procedure based on the known properties of the carbohydrates and a practical laboratory procedure. The procedures described in the present paper have been used with a wide variety of foods and appear to be most applicable to all foodstuffs, although, as new types of food are analysed, improvements might be necessary. Any method that can be applied as a routine procedure for carbohydrates is faced with two difficulties, First, the qualitative difficulties between foods and secondly these differences may affect the interpretation of what are basically apparently similar analytical results. It is probable, therefore, that any procedure for carbohydrate, whilst being technically easy (although time-consuming), can only be made so if the procedure is controlled at the various stages by qualitative information, obtained for instance from paper or thin-layer chromatography. While recovery tests provide some evidence of the reliability of a method, these tests cannot be regarded as providing a

rigorous proof of the method. The added constituent is of necessity pure and is in a different form from the related constituent present in the natural product. Plant materials present special problems because of the structural carbohydrates which enclose the sugars and starches. The evidence from ‘complete’ analysis is of doubtful value because ‘complete’ is a relative term : but comparisons between carbohydrate ‘by difference’ and ‘by analysis’ in the studies of Durnin & Southgate (unpublished) suggest that the present procedures measure available carbohydrate with reasonable accuracy. ‘Complete’ analyses of foods require the measurement of all the constituents, and of these the measurement of unavailable carbohydrates present additional problems which will be discussed in a subsequent paper. Acknowledgments Mrs. J. A. Finch and Mrs. G. A. Till played a valuable part in the original studies of the methods reported here and their assistance is gratefully acknowledged. Most of the experimental work reported here was carried out in the former Department of Experimental Medicine in Cambridge. Dunn Nutritional Laboratory, University of Cambridge and Medical Research Council, Milton Road, Cambridge Received 3 December, 1968

References 1. Henneberg, W., & Stohman, F., Beitr. Begrundung einer rationellen Futterung Wiederhauer, 1860, I ; 1864, I1 (Brauns2. 3. 4. 5. 6. 7.

weig) Jones, D. B., US.Dep. Agric. Circ., 1941, No. 183 McDonald, I., & Braithwaite, D. M., Clin. Sci., 1964, 27, 23 McDonald, I., Proc. Nutr. Soc., 1965, 24, iv Yudkin, J., Proc. Nutr. Soc., 1964, 23, 149 Yudkin, J., & Roddy, J., Lancet, 1964, ii, 6 Widdowson, E. M., & McCance, R. A., Biochem. J . , 1935, 29, 151

8. McCance, R. A., Widdowson, E. M., & Shackleton, L. R. B., Spec. Rep. Ser. med. Res. Coun., 1936, No. 213 9. Williams, R. D., & Olmsted, W. H., J. biol. Chem., 1935, 103, 653

10. Williams, R. D., Wicks, L., Bierman, H. R., & Olmsted, W. H., J. Nutr., 1940, 19, 593 11. Clegg, K. M., J. Sci.Fd Agric., 1956, 7,40

12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

Bolton, W., Analyst, Lond., 1960, 85, 189 Durnin, J. V. G . A., Proc. Nutr. SOC.,1961, 20, ii Southgate, D. A. T., Proc. Nutr. Soc., 1961, 20, iii Southgate, D. A. T., Thesis, Univ. London, 1964 McCance, R. A., & Lawrence, R. D., Spec. Rep. Ser. med. Res. Coun., 1929, No. 135 McCance, R. A., & Widdowson, E. M., Spec. Rep. Ser. med. Rex Coun., 1960, No. 297 Schoch, T. J., J. Am. chem. Soc., 1942, 64,214 ‘Official Methods of Analysis’, (7th ed.), 1950, p. 506 (Washington: A.O.A.C.) Dische, Z . , ‘Methods of biochemical analysis’, 1955, Vol. 2, p. 313 (New York: Interscience) Marks, V., Clinica. chim.Acta, 1959, 4, 395 Partridge, S. M., Nature, Lond., 1949, 164, 443 Fraser, J. R., & Holmes, D. C.,J. Sci. Fd Agric., 1956, 7, 589
