Determination of Glucosamine in Raw materials and Dietary
Supplements Containing Glucosamine Sulfate and/or Glucosamine Hydrochloride by
HPLC with FMOC-Su Derivatization: Collaborative Study
Joseph Zhou, Ted Waszkuc, and Felicia
Mohammed
NOW Natural Foods, Inc., Methods Development
Laboratory, 395 S. Glen Ellyn Rd, Bloomingdale, IL 60108
Collaborators: Chuck Ray; Randy Buren; Wendi
Wang; Hao Nguyen; Darryl Sullivan; Jack Jabusch; Xiaolan Kou; Qiuping Yang;
Aniko Solyom; Jonathan Wang; Tang S. Peng; Mike Blumhorst; Mythili Nagarajan;
Brandon Podhola; Li Huang; Cathy Shevchuk; Rupa Das; Kevin Orellana; Klaus
Reif.
A collaborative study
was conducted on the method for the determination of glucosamine in raw
materials and dietary supplements containing glucosamine sulfate and/or
glucosamine hydrochloride by HPLC with N-(9-fluorenylmethoxycarbonyloxy) succinimide (FMOC-Su) derivatization.
Thirteen blind duplicates of materials consisting of various commercial
products including tablets, capsules, drink mix and liquid products as well as
raw materials, blanks, and spike recovery products were tested by twelve
collaborating laboratories. The tests with the blank samples and the samples
with glucosamine spiked showed good specificity of the method. The average spike
recoveries at the spike levels of 100% and 150% of the declared amount were 99.0% with an RSD of 2.1% and 101% with
an RSD of 2.3%, respectively. The test results between laboratories on each
commercial product were reproducible with all of RSD no more than 4.0%, and the
results were repeatable in the same laboratory with an average RSD of 0.7%.HORRAT values ranged from 0.5 to 1.7 on both tests of spike recovery and
reproducibility between laboratories on commercial products. The average
determination coefficient of the calibration curves from the laboratories was
0.9995 with an RSD of 0.03%. None of the results from the collaborating
laboratories was outlier, partly indicating the robustness of the method. It is recommended that the method be
accepted by AOAC as Official First Action.
The
glucosamine HPLC method with FMOC-Su derivatization (1-3) was selected by an
AOAC expert review panel (ERP) as the most appropriate method to recommend for
further laboratory validation. The
results of the subsequent single lab validation study (SLV) when subjected to
peer review by the ERP and selected members of the AOAC Methods committee on
Dietary Supplements indicated the method was suitable for a full collaborative
study (4).
The
collaborative study is to evaluate the method’s accuracy and precision based on
its intra- and inter- laboratory performance (5-6). In this study, thirteen
blind duplicates of glucosamine test materials consisting of various commercial
products as well as blanks and spike recovery samples were analyzed by twelve
collaborating laboratories.
Glucosamine
product is one of the most popular dietary supplements in the United States,
and its effectiveness has been clinically proved for the treatment of
osteoarthritis (7). Glucosamine HCl and glucosamine sulfate are the two most
important glucosamine salt forms used in these products and also claimed on the
product labels. The establishment of an
official glucosamine method will facilitate its product quality control and
regulatory compliance.
Collaborative Study
Study Design
The study was conducted on twelve different test materials. One of them was split into two identical samples to test repeatability of the analytical results in the same laboratory. The identity or content of these thirteen samples was not released to the collaborators, and random identification numbers were assigned to each of the test materials. The blind duplicates of these materials and glucosamine reference standard were supplied to each of the laboratories for the study. Practice samples were also provided to ensure that each participant could successfully follow the method and to optimize their instruments before proceeding with the actual tests. The study director was available for consultation. Table 1 lists these thirteen test samples and their ingredients and potencies.
Collaborators
Twelve laboratories participated in the study. Three of them were industrial dietary supplement finished product manufacturers, five were raw material vendors, three were commercial testing laboratories, and one was the university laboratory. Geographically, ten of them were from the United States, one from Canada, and one from Germany.
Study Procedures
Product Types.æ There were basically five types of the test materials analyzed: i) tablets; ii) capsules; iii) drink mix powder; iv) raw materials; and v) liquid products. At least one product of each type was chosen to represent the group. These representative samples also covered different material sources, e.g., glucosamine from shellfish or glucosamine from vegetable source; different salt forms: glucosamine hydrochloride and glucosamine sulfate; and different combination products with chondroitin, methylsulfonylmethane (MSM), S-adenosyl-L-methionine (SAMe), Vitamin C, etc.
Sample Potency Range.æ Because of the large variety of test materials covered in this study, the glucosamine concentrations differed greatly. As shown in Table 1, for example, 12.6 g of the commercial drink-mix powder contains only 1.5 g of glucosamine HCl. However 12.6 g of glucosamine HCl raw material with min 98.5% purity contains 12.4 g of glucosamine HCl. In addition, between glucosamine HCl and glucosamine sulfate, because the industrial glucosamine sulfate raw material has a complex empirical formula as 2GlucosamineFreeBase·H2SO4·2KCl, 1 g of such a material only contains 0.59 g of glucosamine free base (GFB), but 1 g of glucosamine HCl contains 0.83 g of GFB. As result, depending on which salt form to use and how much to use (limited by total physical sizes of the tablets or capsules, etc.), the GFB potency in the final products varies significantly. These differences in reality make it difficult in the method to use the same sampling size in weight for all the samples. Therefore, to ensure the best accuracy of tests, the method has classified different sample categories based on the material types for sampling.
Preparation of Test Samples.
ii) Blank Powder: A combination blank powder for tablets and capsules was prepared based on the weight percentages shown in Table 1, and treated as a tablet powder for analysis.
iii) Spike Recovery Samples: The two samples for spike recovery levels I&II were initially made as a liquid using the blank powder and glucosamine standard reference to avoid homogeneity problems, but those products decomposed from microbial action. It was later found that 1% sodium benzoate can be used as a preservative. Because of time limitation and stability concerns on the other products, the collaborators were asked to prepare G1 and G2 (the two spike recovery samples) with glucosamine HCl standard and G12 (the blank powder) provided. But the identity of G1, G2 and content of G12 (marked as a tablet powder) were kept unknown to collaborators all time. The parent products were made by accurately mixing 360 ± 10 mg of glucosamine HCl standard with 70 ± 10 mg of G12, and mixing 240 ± 10 mg of glucosamine HCl standard with 190 ± 10 mg of G12, respectively. All of the samples were processed under a very clean and controlled condition to avoid contamination. Glucosamine in solid state is, in general, chemically stable, but under special cases such as hot temperature and chemical reactions as well as microorganism, it may degrade.
Shipment.æ Samples and standard were shipped to collaborators at ambient temperature. Each of the test bags or bottles was labeled for identification (G1, G2…) and product type, e.g., tablet powder or capsule powder to assist collaborators to differentiate the products and find right categories. Collaborators were required to return a receipt acknowledgement forms to indicate receipt and condition of the shipped items. They were also directed to store samples and standard at room temperature.
Analysis.æ Collaborators are required to prepare new calibration solutions and curve each test day. For test material analysis, single preparation and single injection are required.
Data Reporting.æ Collaborators were asked to report the linearity of each calibration curve, the corresponding concentrations of the calibration solutions, and the percentages of glucosamine free base (GFB) found in each of the thirteen test samples using the data reporting sheets. They were also asked to report any important observations and significant deviations to the method.
Expected Values and Validation Data of the Test Materials.æ Table 2 lists the expected values (e.g., the GFB values calculated from the label claims) and validation data of the test materials. The validation data were obtained by the Study Director’s laboratory at NOW Foods using the same method and on the same materials as sent to the collaborating laboratories in this study.
The Method
Determination
of Glucosamine in Raw Materials and Dietary Supplements Containing Glucosamine
Sulfate and/or Glucosamine Hydrochloride by
HighPerformance Liquid Chromatography of FMOC-Su Derivatives
A. Principle
Glucosamine sulfate/hydrochloride finished products or raw
materials are dissolved in water. The
glucosamine free base is released by adding triethylamine to the solution to
neutralize the H2SO4/HCl salts, and derivatized with
9-fluorenylmethoxycarbonyl succinimide (FMOC-Su). The derivative is separated
by HPLC and measured with UV detection. Glucosamine has two natural
steroisomers (a and b), and the interconversion of these two in aqueous solution is not
preventable, resulting in two peaks in the chromatogram. The sum of the areas
of these two peaks is used for the quantification of the glucosamine free base.
B. Apparatus
(b)
LC
column.æ Phenomenex Prodigy
(MidBore™) ODS-3 100 Å, 5m, 150 x 3.2mm, Phenomenex order #: 00F-4097-R0 (Phenomenex, Torrance,
CA; www.phenomenex.com).
(c) LC
guard column.æ Phenomenex Prodigy SecurityGuard™ Cartridges ODS-3 100 Å, 4 x
3.0mm, Phenomenex order No: AJO-4287 (Phenomenex,).
(d) Analytical
balance.æ Ohaus AS60; readability, ± 0.0001 g (Ohaus, Florham Park, NJ; www.ohaus.co).
(e) Sonicator
(As described or equivalent).æ Branson 8210 ultrasonic cleaner (Branson Ultrasonic Corporation,
Danbury, CT; www.bransonultrasonics.com).
(f) Vortex
(As described or equivalent).æ Type 16700 mixer (Barnstead International, Dubuque, Iowa;
www.barnsteadthermolyne.com).
(g) pH
meter.æ Beckman f40 (Beckman Instruments, Inc., Irvine, CA; www.beckman.com).
(h) Grinder.æ One-Touch Coffee Grinder (General Electric Company, Fairfield,
CT; www.gehousewares.com) Model No. 106854.
(i) Volumetric flasks.æ 5 and 100 mL, class A.
(j) Volumetric
pipettors.æ 10 mL, class A.
(This is needed only for liquid sample analysis.)
(k) LC
solvent filters.æ
(l) Syringe
filters.æ PTFE,
0.45μm x 13mm (Restek, Bellefonte, PA; www.restekcorp.com), and
0.45μm x 25mm (Fisher Scientific Pittsburgh, PA; www.fisherscientific.com).
(m) Syringe.æ Luer-Lok™, 3 mL, from Becton, Dickinson and Company through VWR
International (South Planfield, NJ; www.vwrsp.com).
(n) Eppendorf variable volume
pipettors and tips.æ
50-200μL (accuracy: ± 1.0-0.6%, precision: £ 0.3-0.2%) and
500-2500μL (accuracy: ± 1.5-0.6%, precision: £ 0.3-0.2%). Both are available from VWR International (South Planfield, NJ;
www.vwrsp.com). Note: make sure both pipettors are properly calibrated.
(o) HPLC
injection vials.æ Screw cap vials
with Teflon coated caps (Agilent Technologies, Inc., Palo Alto, CA). (As described or equivalent).
C. Reagents
(a)
(b) Derivatization
reagent.æ N-(9-Fluorenylmethoxycarbonyloxy)succinimide (FMOC-Su), 97% pure, available from Lancaster (Windham,
NH; www.lancastersynthesis.com) Cat. No: 6908.
(c) Solvents.æ Acetonitrile, HPLC Grade; Trifluoroacetic Acid (TFA), min 99.0%
pure; Triethylamine (TEA), min 99% pure; Water, HPLC grade. All from Fisher
Scientific (Pittsburgh, PA).
(d) FMOC-Su
derivatization solution. –– 15 mM. Dissolve 50±1.0 mg of FMOC-Su in 10 mL of
acetonitrile. Prepare this solution fresh for each test.
(f) Derivatization
quench solution.æ Mixture of mobile phases A/B (1/1, v/v).
D. Preparation of Test Solutions
Accurately weigh
or measure the amount, as indicated in Table 3, into separate 100 mL volumetric flasks. For tablets,
find and record the mean weight of 20 tablets, grind, mix and weigh. For
capsules, empty and record the mean fill weight of 20 capsules, grind
the contents, mix and weigh. For liquid
products: shake well before taking the test portion.
Add 90 mL of
water to the test
portion, vortex for 1 min and sonicate for 5 min or until all solids dissolve
(Note: some of the products’ excipients, e.g., silicon dioxide, may never
dissolve). Pipette 750 mL of triethylamine into each of the flasks to neutralize HCl or H2SO4,
and dilute to volume with water. Filter about 1.5 mL of each solution through a0.45 mm x 25 mm PTFE filter into an
HPLC injection vial.
E. Derivatization Procedures
Note: Both standards and test solutions must be
derivatized simultaneously in 5 mL volumetric flasks.
Pipet the exact
amount, as specified below, of the filtered solutions from the HPLC injection
vials into separate 5 mL volumetric flasks: (i) Glucosamine Standard Working
Solutions (three-point calibration): 50, 125(or 100), and 200 mL respectively; and (ii) for all other products: 125(or 100) mL. Add 500 mL of 15 mM FMOC-Su solution to each flask. Cap the flasks tightly
with Teflon stoppers, mix well with vortex, and sonicate all the flasks in the
sonicator water bath at 50 °C for 30 min. Remove the
flasks from the bath, let cool to room temperature, and dilute the flasks to
volume with the mixture of mobile phases A/B (1/1, v/v). Mix well with vortex.
Filter each solution through 0.45 mm x 13 mm PTFE filter into an HPLC vial for injection.
F. Determination
(a) System
suitability tests.æ Equilibrate the HPLC system with the mobile phase for at least 30
min. Make 5 replicate injections of the second (mid-concentration) glucosamine
HCl working standard. The typical retention
time (t)
of glucosamine anomer peak 1(the earlier eluted peak) should not be less than 4
minutes, and the relative retention times (Rr) of glucosamine anomers peak
2 to peak 1 should be 1.2 (Rr = t2/t1).
The peak tailing factor (T) should not be more than 2.0 (T = W0.05/2f,
where W0.05
is width of the peak measured at a point 5% of the peak height
from the baseline; and f is horizontal distance from the vertical
line at the peak maximum to the point on leading edge of the peak at 5%
height). The relative standard deviation (RSD) of the sums of peak area
of glucosamine peaks 1 and 2 from the 5 injections should not be more than
2.0%.
(b) Mobile
phase gradient program.æ Elute the analytes with the following gradient mode of mobile
phases A and B. 0.0-6.0 min: held isocratic at 70A:30B; 6.0-11.0 min: change to
0A:100B; 11.0-13.0 min: to 70A:30B; and 13.0-15.0 min: held isocratic at
70A:30B.
(c) Run
time.æ 15 min.
(d) Injection.æ Make single injection of each standard working solution and
unknown solution.
G. Calculation
(a) Concentrations
of glucosamine working standard solutions.æCalculate the concentrations of glucosamine free base (GFB) in
working standard solutions, after FMOC derivatization:
STD n, GFB, mg/mL = 0.83091x d x W x
F
Where: n =1, 2,
3 for three different standard working solutions; 0.83091 is the conversion
factor from glucosamine HCl to glucosamine free base: 179.17/215.63; d = the
dilution factor: v/(100 mL x 5 mL), with v = 0.050, 0.125 (or 0.100), and 0.200
mL for STD1, STD2, STD3, respectively; W = the amount of glucosamine HClstandard weighed, mg; and F = the purity factor of glucosamine HCl
standard used.
(b) Percentage of Glucosamine Free Base in All
Glucosamine Contained Materials.æ Calculate the % glucosamine free base in all glucosamine
contained materials:
% mg/mg GFB = (P – b) x 100 / (a x D x W)
Where: P =
the sum of peak area of glucosamine peaks 1 and 2 of the unknown test sample; a
= slope of the calibration curve; b = intercept of the calibration curve; D is
the dilution factor: v/(100 mL x 5 mL), with v = 0.125 (or 0.100) mL; W =
the amount of unknown test portion weighed, mg.
For liquid sample:
mg/mL GFB = (P – b) / (a x
D x V)
Where: V = the amount of liquid unknown
product used, mL.
(c) Percentage of Glucosamine HCl in Finished Products or
Raw Materials.æ Calculate the % glucosamine HCl in glucosamine HCl finished
products or raw materials:
% mg/mg G×HCl = (P – b) x 100 / (0.83091 x a x D x W)
For liquids:
mg/mL G×HCl = (P – b) / (0.83091 x
a x D x V)
(d) Percentage
of Glucosamine Sulfate (2GlucosamineFreeBase·H2SO4)
in Finished Products or Raw Materials.æ Calculate the % glucosamine sulfate in glucosamine sulfate
finished products or raw materials:
% mg/mg G×Sulfate = (P – b) x 100
/ (0.78511 x a x D x W)
Where: 0.78511 is the conversion factor
from glucosamine sulfate (2GlucosamineFreeBase·H2SO4) to glucosamine free base:
(2x179.17)/456.418
For liquid samples:
mg/mL G
(e) Amount of Glucosamine Sulfate (or HCl)
Per Product Unit.æCalculate Glucosamine Sulfate (or HCl) in mg per Tablet (or
Capsule), or Glucosamine Sulfate (or HCl) in mg per Serving Volume (mL):
mg G×Sulfate (HCl) per Tablet (Capsule) = %w/w G×Sulfate (HCl) x 100 x mgOneTab (Cap)
or: mg G×Sulfate (HCl) per Serving Volume = mg/mL G×Sulfate (HCl) x mLOneServing
Where: mgOneTab (Cap) = the average weight of one tablet or the average
fill weight of one capsule, in mg; and mLOneServing = the serving volume for
liquid products, in mL.

Results and Discussion
Collaborative
Study Results
The analyses were completed in all twelve collaborating laboratories in two weeks. The sample identification, which was randomly assigned to the test samples, was decoded after the test results were received, and the names of the participating laboratories were coded from L1 to L12 for their data presentation in this report.
Table 4 shows the complete set of data submitted from the collaborating laboratories in the percentages of glucosamine free base for eight commercial products and two different blanks. These data indicate the method performance on reproducibility of the analytical results between laboratories. Table 4 also shows the results on two identical test samples (G10 and G13), that not only indicate the method performance on the repeatability of the analytical results in the same lab, also reproducibility between laboratories. Table 5 shows the concentrations of calibration solutions and linearity of calibration curves reported from each laboratory. Table 6shows the spike recovery results of two spike levels from all the laboratories. The percentages of spike recovery were calculated by dividing the percent glucosamine free base found in the test in each test sample by the percent GFB fortified in the test sample and multiplying by 100. The data indicate both the method accuracy and method precision between-laboratories.
The results in Table 4 were used to generate the statistical data presented in Tables 7 and 8 respectively for reproducibility between laboratories and repeatability in the same laboratory.
The HORRAT value is the ratio of the relative standard deviation, expressed as a percent (%RSD) to the predicated relative standard deviation, expressed as a percent (%PRSD), i.e., HORRAT = %RSD / %PRSD, where %PRSD = 2C-0.1505 and C = the estimated mean concentration. For the spike recovery results presented in Table 6, the true concentrations of glucosamine free base (%GFB spiked) was used to calculate %PRSD for HORRAT values. But for the statistical data shown in Tables 7 and 8, the experimental concentration values of GFB were used to calculate %PRSD.
Collaborators’
Comments
Most of the collaborators found the method was easy to follow and
perform. Reasonable modifications of the method (as shown below) without
significantly affecting the test results and also with the Study Director’s
permission were allowed to show the method’s robustness.
Laboratory 2 increased the post gradient
equilibration time from 2 to 7 min before next injection for a total run time
of 20 min. The results from the other laboratories were obtained using the
conditions per method.
Laboratory 5 observed a small peak around a
retention time of 4.1 min when the analyst uses water as a blank for test. The
laboratory also observed this little peak in some glucosamine sample analysis,
and suspected it had merged with peak 1 of glucosamine in other tests. The
analyst also found in some tests another small peak between peaks 1 and 2 of
glucosamine. It is not sure that the peaks were due to some impurities in the
particular lot of the reagents. However since the size of these peaks were tiny
the quantitation results from the laboratory were not affected as shown in
Tables 4-8.
Laboratory 6 reported that during a replicate
injection of 5, the area of glucosamine peak 1 slightly increased while the
area of peak 2 slightly decreased, but the total area of peaks 1 and 2 remained
the same (RSD = 0.19% and 0.33% with 5 injections for each of 2 trials).
Laboratory 10 prepared all the samples by
using a half of the recommended amount for sampling, and 50 mL volumetric
flasks instead of the 100 mL per method. For derivatization, they used 10 mL volumetric
flasks instead of 5 mL, and all recommended volumes for sample and reagents
were doubled. Some of the differences in sampling weight were shown in Table 5
(Amounts of the blank powder and glucosamine HCl standard used). However all of
their test results are acceptable.
Laboratory 11 performed the tests without a
guard column.
Laboratory 12 noticed insoluble residues in
the sample preparations of G1, G2, G8, G10, G12 and G13.
None of the collaborating laboratories
reported any system suitability problems in the study.
Performance
Characteristic of the Method
In
summary of the data presented in Tables 4-8, the tests with the blank samples
and the samples with glucosamine spiked showed good specificity of the method.
The average spike recoveries at the spike levels of 100% and 150% of the declared amount were 99.0% with an RSD of 2.1%
and 101% with an RSD of 2.3%, respectively. The test results between
laboratories on each commercial product were reproducible with all of RSD no
more than 4.0%. The results were repeatable in the same laboratory on two
identical samples, with an average RSD of 0.68% for all laboratories. The average determination coefficient of the
calibration curves from the laboratories was 0.9995 with an RSD of 0.03%.
HORRAT
values showed 1.1 and 0.9 for the spike recovery analysis at the levels of 100%
and 150% respectively. For the tests of reproducibility between laboratories on
each commercial product, HORRAT values ranged from 0.5 to 1.7. The tests of
repeatability in the same laboratory on two blind duplicates showed HORRAT
values of 0 - 1.2 for all laboratories.
The zero and near zero values for HORRAT are the results of some
laboratories finding and reporting the exactly same or similar %GFB on two
identical tablet samples (G10 and G13), which diminished RSD in the numerator
for calculation of HORRAT value. The fact that collaborating laboratories found
the same or very similar results on two blind duplicates among 13 test
materials demonstrated well the reliability of the method.
The
method is also rugged and robust (6, 8 for definitions). The method has been
tested by twelve collaborating laboratories on various test materials. Although
the same method was followed, the actual use conditions in different
laboratories may vary greatly. It is
important to note that all of the twelve laboratories succeed in the study, andnone of their results was outlier.
Since
glucosamine concentration in commercial products varies significantly with
product manufacturer, type (tablets, capsules, raw materials, etc.) and
glucosamine salt form used (glucosamine hydrochloride and glucosamine sulfate),
the tests may be more accurate if sampling amount is based on its label
claims.
Recommendations
Based upon the results of the collaborative
study it is recommended that the method be accepted by AOAC as Official First
Action.
Acknowledgments
We appreciate AOAC Methods
Committee K and ERP members for their valuable guidance, comments and
suggestions, as well as AOAC staff’s hard working and strong support to this
collaborative study. We thank the following collaborators for their
participation:
Chuck Ray and Randy Buren,
Cargill, Inc., Eddyville, IA
Wendi Wang, Hao Nguyen,
Advanced Botanical Consulting & Testing, Inc., Tustin, CA
Darryl Sullivan and Jack
Jabusch, Covance Laboratories, Inc., Madison, WI
Xiaolan Kou and Qiuping
Yang, Nature’s Sunshine Products, Inc., Provo, UT
Aniko Solyom, The University
of Arizona, Tucson, AZ
Jonathan Wang, Nature’s Way
Products, Inc., Springville, UT
Tang S. Peng, Pure World
Botanicals, South Hackensack, NJ
Mike Blumhorst, ArcherDaniels Midland Company, Decatur, IL
Mythili Nagarajan and
Brandon Podhola, Enzymatic Therapy, Inc., Green Bay, WI
Li Huang, Cathy Shevchuk, JR
Laboratories Inc., Burnaby, Canada
Rupa Das and Kevin Orellana,
BI Nutraceuticals, Long Beach, CA
Klaus Reif, PhytoLabs,
Vesternbergsgreuth, Germany
This study is the result of
a contract between the Center for Food Safety and Applied Nutrition, FDA and
the Office of Dietary Supplements, NIH with the AOAC INTERNATIONAL. The purpose
of the contract is to provide the FDA, as well as other government agencies,
and the dietary supplements industry with AOAC Official Methods, applicable to
commercial available dietary supplements and their raw materials.