1. Sugar profile method by ion chromatography (HPAEC-PAD)
Andrew Ruosch
Madison, WI
Covance Food Solutions
06June 2018

2. New method developed at Covance
Multi-lab validation study for a High Performance Anion Exchange Chromatography with Pulsed Amperometric Detection (HPAEC-PAD) sugar profile method applicable to a wide variety of matrices
New method developed at Covance
Co-authors: David Ellingson, Thomas Vennard, Scott Wejrowski, Dan Berg
Existing methodology for sugar testing GC
HPLC IC
Intended to improve the quality and performance of sugar profile testing
Seven analytes included in the sugar profile
Galactose, glucose, sucrose, fructose, lactose, isomaltulose, and maltose
Problems with other methods:
GC: outdated methodology requiring derivatization
HPLC: RI detector, so high level of interferences – salt, proteins do not work. Also these detectors lack specifity
IC: methods that run well for specific analytes or analyte groupings
Wanted a single, accurate, precise method to analyze a common sugar profile
Need for this to be applicable to a wide variety of sample types
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

3. Applicable Matrices 1 2 3 4 5 6 7 8 9 Food products Infant formulas
Drinks
Adult nutritionals
Pet and animal feeds
Meats
Dietary supplements
Oils
100% Fat 1 2 3 4 5 6 7 8 9
67% Fat
33% Carbohydrates
67% Fat
33% Protein
33% Fat
67% Carbohydrates
33% Fat
67% Protein
Samples high in fat, protein, and carbohydrates have all proven to be successful with this method
At least one matrix from each section of this triangle was fulfilled during validation.
The results of the validation demonstrate the performance of this method for sugar analysis across a wide variety of sugar testing needs.
But before we go further into the validation, I’d like to go over the method itself
100% Carbohydrates
100% Protein
67% Carbs
33% Protein
67% Protein
33% Carbs
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

4. Method Outline Samples are weighed into a 500-mL Erlenmeyer flask
Internal standard is added and samples are extracted in ethanol:water
Samples are qualitatively filtered
Optional protein clean-up with Carrez reagents
10 KDa molecular weight filtration
Evaporation and reconstitution with water
Diluted as needed and 0.45 µm filtration
Measurement with HPAEC-PAD
2.5 g needed primarily for homogeneity reasons. That larger weight reduced precision concerns.
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

5. Calibration Standard Chromatogram
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

6. Evolution of the Method
Initially work collaboration with AAFCO
Pet food and animal feed method
Need for a single, uniform method
High variability between methodologies due to instrumentation and extraction differences
Published in Journal of AOAC International in 2016
Pet Food method performance
Acceptable accuracy
Precision fell outside of target precision criteria for most matrices
Pet feed method was created because there were variabilities between different analysis types. A single, rugged method was needed for accurate and precise testing.
The pet feed method showed promise but the precision was a concern.
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

7. Key Items Adapted from Previous Method
New sugar method was built on the foundation of the pet and animal feed method
Critical information from previous method:
Identified the best choice for extraction solution
Equal parts ethanol:water
Solution tested better than methanol in extraction efficiency, but similarly to acetonitrile
50:50 ethanol:water showed optimum balance of analyte solubility and bacterial and enzymatic inhibition
Proved molecular weight filter aided in extraction without influencing the sugar recoveries
Identified the best choice for column (CarboPac PA20)
Developed a gradient to adequately separate all mono- and disaccharides
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

8. Areas to Improve Upon from Pet Feed Method
Precision
Evaluate sample preparation for better homogeneity
Introduce an internal standard
Evaluate the sample weight to solvent ratio
Instrument injection time
Shorten from 45 minutes
Expand the scope of the method beyond pet and animal feeds
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

9. Precision Optimization
Sample homogenization evaluation
Began with pet and animal feeds
Dry cat food
Dry dog food
Horse Feed
Swine feed
Milk replacer
Wet cat food
Assessed two sample preparation versions
Dry blend mill
Liquid nitrogen preparation
Finer sample grinds with liquid nitrogen milling
Showed improved precision data
Milk replacer and wet cat food did not require liquid nitrogen
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

10. Dry Blend vs Liquid Nitrogen
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

11. Internal Standard Evaluation
After optimizing the sample grinding, IS was evaluated
Reviewed specificity chromatograms for best IS options
Arabinose
Glucosamine
Ribose
Glucosamine has been used as IS for other similar methods, but its popularity in the pet food industry eliminated it.
Arabinose was chosen because it was more fully resolved from any of the core analytes
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

12. Internal Standard Precision Evaluation
Results of all six feeds were analyzed with and without the arabinose internal standard
Internal standard added to both types of sample grinds
Lowest variability found with samples that had the liquid nitrogen grinding and internal standard
Acceptable precision in all samples using liquid nitrogen grinding and IS
Dry blend grinding with IS did not provide acceptable precision
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

13. | Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

14. | Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

15. | Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

16. Sample Weight Analysis
Initial method used 5 grams into 200-mL of extraction solution (except for milk replacer)
Optimized sample weight:extract volume ratio
Tested with CRMs
NIST 3233 (breakfast cereal)
BCR-644 (artificial foodstuff)
Best recovery with 1 gram into 200-mL of extraction solution
Same recovery at 1 gram with higher volumes (500-mL and 1000-mL)
Since sample homogeneity was a concern at 1 gram, both the sample mass and volume were increased 2.5x
The method extraction scheme is 2.5 grams into 500-mL extraction solution
By addressing each of the three issues, precision has vastly improved for this method
However, at this point we have decided not to use liquid nitrogen for all sample types. As we moved forward into expanded food testing, we found that it was unfeasible and not necessary to do. The introduction of the internal standard with the higher sample weight of 2.5g combined to minimize homogeneity concerns.
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

17. Injection Time Modification
Another goal to shorten the run time (45 minutes originally)
Comparative GC and LC sugar methods are 30 and 21 minutes, respectively
Gradient adjustments to elute sugars faster were unsuccessful
Successful monosaccharide separation is delicate
Added a post column hydroxide infusion to minimize pH changes at the cell
Allowed elimination of 15 minutes of re-equilibration time from each injection
Injection time is 30 minutes, which is on par with the GC method
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

18. Recovery, repeatability and reproducibility parameters
Multi-lab Validation
Samples performed across four Covance sites
Madison, WI, USA (MSN)
Battle Creek, MI, USA (BC)
Harrogate, UK (HAR)
Singapore (SGP)
Same standard and certified reference materials analyzed at all sites
Each site also analyzed additional matrices that would be common to their site-specific sugar testing needs
All matrices were analyzed for accuracy and precision
Spike recoveries were used to measure accuracy for matrices that did not have certified or reference values
Recovery, repeatability and reproducibility parameters
Analytical Range
0.1% - 5%
5 % - 50 %
50 % - 100%
Each Individual Sugar
Recovery, %
95-105
RSDr, % ≤7 ≤5 ≤3
RSDR, %
≤10 ≤8 ≤4
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

19. Infant formula/adult nutritional Pet food and animal feed
Validation Samples
Category
Matrix
Material ID
Plan
Sites
Food Products
Cereal
NIST 3233
CRM
BC, HAR, MSN, SGP
Baby Food
NIST 2383a
MSN
Juice
NIST 3282
HAR
Vegetable
Spinach
Innate, spikes
Oil/Fat
Palm Oil
Baking Chocolate
Lemon juice
High fat
Peanut butter
SGP
Artificial Foodstuff
BCR-644
BC, HAR, SGP
Infant formula/adult nutritional
Infant formula
NIST 1849a
HAR, MSN, SGP
Dietary Supplements
Gummy
Tablet
Premix
Drink
Pet food and animal feed
Dry dog food
Dry cat food
Horse feed
Swine feed
Supplement
Milk replacement supplement
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

20. Analyte
Matrix
Sample ID
RM value (%)
RM Range (%)
Lab Overall Mean (%)
% of CRM Value
Lab Range (%)
Fructose
Cereal
NIST 3233
0.81
0.727
89.8%
Baby Food
NIST 2383a
3.96
4.26
107.6%
4.09 – 4.45
Artificial Foodstuff
BCR-644
16.2
15.8
97.5%
Cranberry Juice
NIST 3282
2.08
1.89
90.9%
1.86 – 1.91
Glucose
1.04
0.922
88.7%
3.80
4.07
107.1%
3.92 – 4.22
0.850
0.834
98.1%
0.829 – 0.840
Sucrose
13.42
13.6
101.3%
3.57
3.35
93.8%
3.18 – 3.52
10.81
10.7
99.0%
Maltose
0.46
0.443
96.3%
Lactose
0.50
0.496
99.2%
0.474 – 0.521
Lactose Monohydrate
15.85
15.6
98.4%
Infant formula/Drink
NIST 1849a
47.6
49.5
104.0%
Total Sugar
15.7
99.4%
12.05
12.4
102.9%
11.8 – 12.9
2.862
2.808 – 2.916
2.72
95.0%
2.69 – 2.74
This is a table of all results compiled from the RM testing. I don’t want to overwhelm with statistics and tables, but I do want to call out that all values lie within the RM ranges. While most averages correspond very well with those of the RM values, we do recognize some that drift further away.
For NIST 3233, the RM ranges are quite large, given the concentrations. This is from data collected from various methodologies that are driving inconsistent results (as seen in the initial pet feed work). We have confidence in our values.
Likewise for the baby food (NIST 2383a), we observed elevated results for glucose and fructose, while also a lowered result for sucrose. This may indicate sucrose degradation. To confirm this we analyzed this sample by the HPLC method for sugars. The results confirmed our findings.
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018
Sucrose (%)
Glucose (%)
Fructose (%)
IC mean (n=9)
3.20%
4.07%
4.26%
HPLC mean (n=3)
3.28%
4.33%
4.03%
NIST Range
3.45% %
3.69% %
3.87% %

21. CRM and SRM Analysis
Previous table compiled all results from the reference material (RM) testing
All averages lie within the RM ranges with the exception of Cranberry Juice (NIST 3282)
Likely from a small amount of individual labs creating the reference range
Juice matrix was confirmed with spiking analysis
Most averages correspond very well with those of the RM values
Differences observed were likely due to methodology differences
Evident with large RM ranges (glucose and fructose for NIST 3233)
Baby food (NIST 2383a)
Observed elevated results for glucose and fructose, as well as a lowered result for sucrose
Sucrose degradation confirmed with HPLC
Sucrose (%)
Glucose (%)
Fructose (%)
IC Method mean (n=9)
3.20%
4.07%
4.26%
HPLC Method mean (n=3)
3.28%
4.33%
4.03%
NIST Range
3.45% %
3.69% %
3.87% %
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

22. Summary of Spiking Analysis
Spikes were performed at multiple levels on the samples without references values
Although a few matrices yielded recoveries outside of acceptable criteria, the data as a whole was acceptable
All analytes recoveries were compared with AOAC SMPR criteria
Minority that did not meet requirements were due to interferences in chromatography that did not allow for accurate quantification (mainly LOQ spikes)
These were also noted in the validation report for this method
Matrices spiked
Lemon juice
Horse feed
Baking Chocolate
Peanut butter
Swine feed
Palm Oil
Gummy
Dry dog food
Spinach
Tablet
Dry cat food
Premix
Milk replacer
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

23. Spike Recovery Analysis: Lemon Juice
Testing Site
Analyte
Approximate Spike Level Added (%)
Average Innate Level (%)
Average Spike Recovery (%)
%RSDr
Target Recovery Range (%)
MSN BC HAR SGP
Galactose 
0.5
0.00
97.5
1.4
90-110
0.75
97.0
2.3
Glucose 
1.0
0.47
98.8
1.7
1.25
99.1
2.4
Sucrose 
98.4
2.0
1.6
Fructose 
0.9
0.39
99.2
1.2
2.1
Lactose 
97.8
98.3
1.5
Isomaltulose
103.6
2.9
102.9
Maltose 
1.9
1.8
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

24. Recovery, repeatability and reproducibility parameters
Precision
%RSD were much lower than the initial pet food and animal feed method
The %RSD values for all analytes across all matrices would meet AOAC SMPR criteria, with only one exception (lactose in milk replacer)
Includes all reference and non-reference materials for both repeatability and reproducibility standards
Combined total of 70 analytes (including total sugar)
Recovery, repeatability and reproducibility parameters
Analytical Range
0.1% - 5%
5 % - 50 %
50 % - 100%
Each Individual Sugar*
Recovery, %
95-105
RSDr, % ≤7 ≤5 ≤3
RSDR, %
≤10 ≤8 ≤4
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

25. Validation Summary The multi-lab validation was completed
Accuracy and precision were deemed acceptable for all matrices listed
Linearity, specificity, range, LOQ, stability were also acceptable
This method was considered acceptable to be ran on most matrices for sugar analysis
High protein samples out of scope of method
Tuna, wet cat food, ground beef
Protein interference with internal standard
Suppression and inconsistent peak responses
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

26. Additional Validation on Protein Matrices
Extracts required more thorough deproteinization
Use Carrez reagents to clarify extracts
Validation plan was created to evaluate matrices with protein clean-up
Analyzed a variety of matrices with high concentrations of protein
Re-analyzed some of the reference materials tested in initial validation to ensure consistency
Followed the same multi-lab format (four test sites)
Focus on accuracy and precision
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

27. Result without Carrez extraction (%) Result with Carrez extraction (%)
NIST 3233 Comparison Data
Reference material subjected to Carrez testing, with the results compared directly to the average values obtained in the initial validation
Carrez treatment did not have any negative effect on the individual or total sugar recoveries
Analyte
Result without Carrez extraction (%)
Result with Carrez extraction (%)
% Change
Fructose
0.727
0.712
-2.1
Sucrose
13.6
13.9
2.2
Glucose
0.923
0.921
-0.2
Maltose
0.443
0.429
-3.3
Total Sugar
15.7
15.4
-1.9
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

28. Validation Matrices: Protein
Category
Matrix
Material ID
Plan
Sites
Food Products
Cereal
NIST 3233
SRM
HAR, MSN, SGP
Baby Food
NIST 2383a
MSN
Lunch Meat
Ham
Innate, spikes BC
Turkey
Salami
Meat
Mince
HAR
Seafood
Tuna
Infant formula/adult nutritional
Infant formula
NIST 1849a
CRM
Dietary Supplements
Protein Powder
Drink
Pet food
Dry cat food
Wet cat food
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

29. Accuracy: Reference Materials
Analyte
Matrix
Sample ID
CRM value (%)
CRM Range (%)
Lab Overall Mean (%)
% of CRM Value
Lab Range (%)
Fructose
Cereal
NIST 3233
0.81
0.712
87.9%
Baby Food
NIST 2383a
3.96
3.99
100.8%
Glucose
1.04
0.921
88.6%
3.80
3.93
103.4%
Sucrose
13.42
13.9
103.6%
3.57
3.20
89.6%
Maltose
0.46
0.429
93.3%
Lactose
0.50
0.478
95.6%
Lactose Monohydrate
Infant formula/Drink
NIST 1849a
47.6
51.0
107.1%
Total Sugar
15.8
15.4
97.5%
12.05
11.8
97.9%
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

30. Accuracy Summary: Reference Materials
The protein clean-up results consistent with the initial validation
NIST 3233 once again trends lower than the reference value, but results are within the range
NIST 2383a shows the same evidence of sucrose degradation
Elevated glucose and fructose levels, with a lower sucrose value
Sucrose (%)
Glucose (%)
Fructose (%)
IC Method mean (n=9)
3.20%
3.93%
3.99%
HPLC Method mean (n=3)
3.28%
3.97%
4.13%
NIST Range
3.45% %
3.69% %
3.87% %
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

31. Accuracy: Spike Recoveries
Spikes were performed at multiple levels on the samples without references values
All average recoveries for all analytes met AOAC SMPR criteria
Matrices Spiked:
Wet cat food
Dry cat food
Tuna
Salami
Turkey
Ham
Protein powder
Mince
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

32. Precision
All repeatability and reproducibility parameters outlined by the AOAC SMPR were met, with the exception of maltose for the cereal matrix
A total of 33 analytes (including total sugar) were evaluated for precision, with only one instance not meeting criteria
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

33. Protein Validation Summary
The method may be run with or without the Carrez clarification step
This step must be used when analyzing matrices known to have high protein concentrations
~35% or more protein on a dry matter basis
Accuracy and precision met acceptance criteria and this optional step was added to the method
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

34. Expanded Scope
At this time, the scope has been expanded to support a full spectrum of matrices
Samples from each region of the AOAC triangle below were analyzed and passed SMPR criteria
This is a single, accurate, precise, and rugged method capable of analyzing a wide variety of matrices
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018

35. Collaboration with Thermo Fisher
This project was done in partnership between Covance and Thermo Fisher
The development and validation work was all performed at Covance sites
Thermo Fisher provided an ICS-5000 for use on this project, as well as guidance and troubleshooting assistance throughout the process.
| Sugar Profile Method by Ion Chromatography (HPAEC-PAD) June 06, 2018