Hall 2000
Hall 2000
Abstract: The objectives of this work were to evaluate the ef®cacies of commercial starch analyses and
of starch analysis extraction and gelatinisation procedures. In Study 1, accuracy and speci®city of
commercially available starch analyses were evaluated with six co-operating laboratories (®ve
commercial, one university). Results from 11 test samples showed three laboratories with recoveries of
puri®ed starch of 92 g kgÿ1 or less. Three and four laboratories had in¯ated values when samples
contained glucose or sucrose, respectively. Analyses appeared to have good speci®city for glucose.
Incompleteness of starch detection and interference by non-starch carbohydrates can affect
commercially available analyses. In Study 2, extraction with 80:20 ethanol/water (v/v; 80EtOH) or
90:10 ethanol/water (v/v; 90EtOH) to remove low-molecular-weight carbohydrates, and gelatinisation
with heat, alkali (KOH), 6 M urea or 8 M urea were evaluated. Extraction with 80EtOH or 90EtOH
reduced interference from non-starch carbohydrates. Gelatinisation with heat was adequate for good
recoveries of starch glucose for both control (non-extracted) and 80EtOH-extracted samples;
gelatinisation with alkali was required for 90EtOH-extracted samples. Recoveries of pure starch
samples were greatest with no extraction and heat gelatinisation. 80EtOH extraction with heat
gelatinisation appears to be an adequate preparation method when removal of low-molecular-weight
carbohydrates is desired.
# 2000 Society of Chemical Industry
INTRODUCTION EXPERIMENTAL
Analysis of starch in feedstuffs is often accomplished Study 1. Laboratory comparison
by enzymatic methods in order to distinguish the Six feed analysis laboratories (®ve commercial, one
glucose in starch from that in other carbohydrates. university) collaborated in this study. Each laboratory
Complete hydrolysis of starch to glucose and speci®c was sent 11 samples identi®ed only by sample number
measurement of the released glucose are required to (Table 1), for starch analysis according to their usual
obtain accurate starch values. However, endogenous methods (Table 2). The samples included pure
glucose in the sample, glucose released by non- samples and feed samples. Pure samples included
amylolytic enzymes, or interference of mono- corn starch (CS; Sigma Chemical Co, St Louis, MO,
saccharides other than glucose can overestimate the USA), potato starch (isolated in our laboratory), a-D-
starch content. Pre-extraction with aqueous ethanol glucose (Glc; Sigma Chemical Co, St Louis, MO,
can reduce this interference through removal of low- USA), D-fructose (Fru; Aldrich Chemical Co, Inc,
molecular-weight carbohydrates. However, it has Milwaukee, WI, USA), cellobiose (Cb; Acros Organ-
been reported that ethanol pre-extraction of samples ics, New Jersey, USA) and confectioners' sugar (Dixie
can make starch resistant to enzymatic hydrolysis and Crystals, Savannah, GA, USA). The moist total mixed
consequently to analysis.1 Incomplete hydrolysis of ration containing citrus pulp was dried in a 55 °C
starch to glucose or incomplete assay of released forced-air oven. All feed samples were ground to pass
glucose will result in low starch values. The aims of the 1 mm screen of a Wiley mill (Arthur H Thomas,
this study were to evaluate the accuracy of starch Philadelphia, PA, USA).
analyses performed by feed analysis laboratories and Collaborating laboratories used various methods of
to evaluate the effectiveness of aqueous ethanol gelatinisation followed by enzymatic hydrolysis in
extraction and starch gelatinisation methods in starch sodium acetate buffer (Table 2). Glucose detection
analysis. methods varied. Laboratories A, B, C and D analysed
* Correspondence to: Mary B Hall, Department of Animal Sciences, University of Florida, PO Box 110920, Gainesville, FL 32611, USA
†
Florida Agricultural Experiment Station, Journal Series No R-07473
(Received 31 March 2000; accepted 20 July 2000)
Laboratory
Sample a A B C D E F Mean SE b
Soybean meal (48%) 36 53 35 31 22 69 41 7.0
Citrus pulp 51 50 58 0 39 139 56 18.6
Distillers' grains 81 80 96 33 72 85 75 8.9
Hominy feed 624 652 628 683 637 619 641 9.7
CS Glc 955 926 894 826 992 778 895 32.9
CS Fru 854 838 822 845 896 809 844 12.4
CS Cb 857 847 870 964 915 799 875 23.4
Confectioners' sugar 204 207 184 129 42 8 129 35.0
Corn starch 947 885 921 987 984 894 936 17.9
Citrus pulp TMR 138 126 153 121 135 167 140 7.0
Potato starch 972 913 881 1014 1015 916 952 23.1
a
Table 3. Starch analysis results from six CS Glc, corn starch glucose; CS Fru, corn starch fructose; CS Cb, corn starch cellobiose.
b
laboratories (starch g kgÿ1 sample dry matter) Standard error.
that their analysis methods did not correct for pure starch samples raise questions about the com-
endogenous glucose. Free fructose in the CS Fru pleteness of gelatinisation or hydrolysis, or the ef®cacy
sample did not appear to in¯ate the starch values for of enzymes. Evaluation of puri®ed samples of the type
any laboratory (Table 3), suggesting good speci®city included in this study allows laboratories to assess the
for glucose. Laboratory F used a cold water extract accuracy and ef®cacy of their starch method.
that was acid hydrolysed as a sample blank with each
assay. Because of this correction, interference from Study 2. Comparison of extraction and gelatinisation
endogenous glucose, fructose or sucrose was elimi- methods
nated. Starch recovery differed with extraction type and
The pure carbohydrates that required hydrolysis to gelatinisation method (Table 4). `Assay run' was not
present interference were cellobiose and sucrose. a signi®cant factor for any of the treatments. Heat,
Laboratories D and E reported starch values greater KOH and 6 M urea gelatinisation methods provided
than 104% for the CS Cb sample, suggesting equally good recoveries for unextracted corn starch,
cellobiase contamination or technical errors. Results whereas gelatinisation with 8 M urea gave a lower
of the analyses of confectioners' sugar suggest that value. The KOH and Heat gelatinisation treatments
enzyme preparations used by four laboratories con- provided higher starch values than both urea treat-
tained invertase (sucrase) activity. Confectioners' ments for 90EtOH-extracted corn starch. Further
sugar contains only 33 g kgÿ1 starch and 14 g kgÿ1 free evaluation of Heat and KOH methods on 90EtOH-
glucose, yet starch values for four laboratories ranged extracted pure samples showed a higher average starch
from 129 to 207 g kgÿ1. Again, laboratory F's water value for KOH (772 g kgÿ1) than for Heat (763 g kgÿ1;
extract/acid hydrolysis correction prevented interfer- P = 0.0001). The four gelatinisation methods did not
ence by sucrose, exemplifying the advantage of differ in their starch recoveries for 80EtOH-extracted
including appropriate sample blanks. hominy feed according to the results of the Student±
Among the feed samples tested, the standard errors Newman±Keuls test. However, the orthogonal con-
across laboratories for soybean meal, distillers' grains, trast of Heat KOH vs 6 M urea 8 M urea for
hominy feed and citrus pulp TMR were small (Table 80EtOH-extracted samples indicated that the two sets
3). In contrast, the standard error for citrus pulp was
more than twice as large. This may be due to the
presence in citrus pulp of approximately 360 g kgÿ1 of Table 4. Starch analysis results for extraction and gelatinisation methods
DM of mono- and oligosaccharides, including glucose (starch g kgÿ1 sample dry matter, mean standard error)
and sucrose, which may interfere with the starch
Gelatinisation method Control a 90EtOH b 80EtOH c
analysis.6 In addition to the analytical issues previously
discussed, variation in results across laboratories may Heat 954 4.8a 915 5.3a,x 608 0.15a
be due to errors in subsampling, or other factors. Dry KOH 945 7.0a 929 4.2a,y 609 0.08a
matter estimates for samples were similar among the 6 M urea 947 1.8a 879 13.2b 591 0.11b
8 M urea 927 6.0b 869 10.2b 580 0.18b
laboratories (data not shown).
The differences in results among laboratories a
Unextracted corn starch.
b
indicate the importance of speci®city and complete- c
Corn starch extracted with 90:10 ethanol/water (v/v) solution.
Corn hominy feed extracted with 80:20 ethanol/water (v/v) solution.
ness of recovery in starch analysis. Interference can be
a,b Comparison of all gelatinisation methods. Values in same column with
minimised through use of adequate sample blanks, different letters differ, P < 0.05.
purer enzyme reagents and assays with high speci®city x,y Comparison of Heat and KOH gelatinisation. Values in same column with
for starch glucose. Starch recoveries below 95% for different letters differ, P < 0.05.
of gelatinisation methods did differ. The urea gelati- ment of released glucose, pre-extraction with aqueous
nisation methods were less effective than either Heat ethanol of samples containing substantial amounts of
or KOH, particularly with the aqueous ethanol- low-molecular-weight carbohydrates, and use of a
extracted samples. The need for KOH gelatinisation gelatinisation method appropriate to the sample. Use
to achieve the greatest starch recovery for the of 80:20 ethanol/water (v/v) as an extractant and
90EtOH-extracted samples is in agreement with gelatinisation by heating appears to be an adequate
reports that ethanol extraction makes samples more method when removal of low-molecular-weight car-
resistant to starch analysis.1 However, the 80EtOH- bohydrates is needed, whilst heat gelatinisation ap-
extracted samples did not re¯ect this resistance. pears to suf®ce for unextracted samples.
The unextracted feed samples gave higher starch
values than did the aqueous ethanol-extracted samples
(Table 5). This may be at least partially due to
REFERENCES
interfering carbohydrates in the samples, particularly 1 Selvendran RR, Discussion, in The Analysis of Dietary Fiber in
in citrus pulp and citrus pulp TMR. The gelatinisation Food, Ed by James WPT and Theander O, Marcel Dekker, New
of 80EtOH-extracted samples with Heat gave similar York, p 171 (1981).
results to the treatment of 90EtOH-extracted samples 2 Englyst H, Wiggins HS and Cummings JH, Determination of the
non-starch polysaccharides in plant foods by gas±liquid chroma-
with KOH. Heat gelatinisation of 90EtOH-extracted
tography of constituent sugars as alditol acetates. Analyst
samples did not give good starch recovery. 107:307 (1982).
3 Holm J, Bjorck I, Drews A and Asp NG, A rapid method for the
analysis of starch. Starch/Starke 38:224 (1986).
CONCLUSIONS 4 Karkalas JJ, An improved enzymatic method for the determination
Inaccuracies in starch analysis methods re¯ecting of native and modi®ed starch. J Sci Food Agric 36:1019 (1985).
incomplete measurement of starch or measurment of 5 SAS, The SAS System for Windows, Release 6.12. SAS Institute,
Cary, NC (1996).
interfering substances as starch can affect analytical 6 Hall MB, Hoover WH, Jennings JP and Miller Webster TK, A
results from commercial laboratories. Reductions in method for partitioning neutral detergent-soluble carbo-
these errors may be accomplished by speci®c measure- hydrates. J Sci Food Agric 79:2079 (1999).