Journal of Chromatographic Science, Vol.

48, February 2010

Quantitative HPLC Analysis of Active Pharmaceutical

Ingredients in Syrup Vehicle Using Centrifugal
Filter Devices and Determination of Xanthan
Gum in Syrup Vehicle Using Rheometry
Yong Chen*,†, Tanya Tadey, Mougang Hu†, Geoff Carr, and Junan Guo
Patheon Inc., 2100 Syntex Court, Mississauga, ON, Canada, L5N 7K9

Abstract samples into analytical instruments (1).

Liquid–liquid extraction (LLE) and solid-phase extraction
Using rapid centrifugal filtration (≤ 30 min) of diluted samples, the (SPE) are two popular sample preparation methods used in most
filter membrane prevented compounds with molecular weight analytical laboratories. LLE utilizes a solvent that is immiscible
higher than the nominal molecular weight limit (NMWL) from with sample matrices to extract target analytes while SPE uti-
transporting through the membrane, thus separating them from lizes an extraction phase adsorbed on fine particles packed in a
compounds with molecular weight smaller than NMWL, which cartridge to retain target analytes, which are eluted separately
would pass through the membrane. The purpose of this study aims from interfering components. However, both of these methods
to remove high molecular weight matrix (such as xanthan gum)
are time-consuming (2).
interferences while achieving a quantitative analysis of the active
pharmaceutical ingradients of interest. Two model active
Ultra-filtration is a pressure-driven membrane separation and
pharmaceutical ingredients, L-arginine and amphotericin B, were purification process. It is being widely used in pharmaceutical
quantitatively recovered from the diluted syrup vehicle after and food industries and in water and waste water treatment pro-
centrifugation with the filter devices. The reproducibility [% cesses (3–11). Ultra-filtration membranes may be considered
relative standard deviation (RSD), peak area] of the filtered samples porous membranes that are characterized by their nominal
was less than 0.5%. For amphotericin B samples. The linear range molecular weight limit (NMWL) (12). Molecules with molecular
was 0.28 µg/mL to 28.2 µg/mL. The limit of detection was 0.06 weights larger that NMWL are retained while those with molec-
µg/mL. The limit of quantification was 0.28 µg/mL. ular weight smaller than NMWL pass through the membrane.
The viscosity of a syrup vehicle changed linearly with the Separations of molecules with different molecular weights are
concentration of xanthan gum. A method was thus developed to thus achievable. Cross-flow filtration is widely used in the
determine xanthan gum in the syrup vehicle. The accuracy was
industry to provide continuous separation and purification of
within 95.0% to 105.0% at different concentration levels.
various products (13). However, centrifugal ultra-filtration is
easier to operate in analytical laboratories for small sample sizes.
Syrup vehicles are complex mixtures, consisting of a variety of
Introduction excipients such as diluters, solubilizers, stabilizers, and preserva-
tives. Direct injection of diluted syrup vehicle into an HPLC
system resulted in rapid reduction of column efficiency, split
Quantification of active pharmaceutical ingredients (APIs) in
peaks, and severely reduced column lifetimes. It was suspected
complex pharmaceutical matrixes, such as syrup vehicles, is very
that high molecular weight polymers caused these problems by
challenging due to the incompatibility of the matrices with
adsorbing onto the stationary phase of the HPLC column. This
modern instruments [e.g., high-performance liquid chromatog-
work investigated the use of centrifugal filter devices to remove
raphy (HPLC)]. Direct injection of such samples into analytical
high molecular weight interfering ingredients from the matrix
systems often contaminates and even damages the systems.
to allow quantification of APIs in the syrup by HPLC. Unlike the
Moreover, the concentrations of target analytes are often too low
traditional use of ultra-filtration to retain and concentrate high
to allow direct detection of analytes even with advanced detec-
molecular compounds like proteins for subsequent analysis, this
tion systems. A sample preparation step, including clean up and
work was focused on the analysis of small molecular compounds
possible enrichment, is usually required prior to introduction of
in the filtrate after removing high molecular weight
interferences with the filter devices. This work was extended to
* Current Address: Supelco, 595 North Harrison Road, Bellefonte, PA 16823
† Author to whom correspondence should be addressed: Y. Chen, e-mail [email protected]; and
develop a method for the determination of xanthan gum based
M. Hu, e-mail [email protected]. on measurement of viscosity.

104 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.
Journal of Chromatographic Science, Vol. 48, February 2010

Experimental Centrifugal ultra-filtration

Centrifugal filter devices were cleaned by centrifugation with
Materials and chemicals 10 mL of deionized water prior to immediate use. 10 mL of stan-
L-arginine, amphotericin B, HPLC-grade methanol, HPLC- dard and sample solutions were transferred into centrifugal filter
grade triethylamine, and 1-hexanesulfonic acid sodium salt were devices, and centrifugation was performed at 5000 rpm for 30
purchased from Aldrich (Oakville, Ontario, Canada). Sodium min.
phosphate monobasic, monohydrate, was purchased from EMD
Chemicals (Gibbstown, NJ). Syrup vehicle (Ora Sweet SF) was Determination of viscosity
purchased from Paddock Laboratories (Minneapolis, MN). The viscosities of xanthan gum solutions were determined
Centrifugal filter devices (Amicon Ultra-15, 10K NMWL) were with Brookfield Digital Rheometer model DV-III Ultra
purchased from Millipore (Billerica, MA). (Brookfield Engineering Laboratories, Middleboro, MA). 8.0 mL
Sucrose, sorbital, sodium saccharin, strawberry flavor, citric of solutions were transferred into the sample chamber (13 RP)
acid, sodium citrate, methylparaben, propylparaben, and xan- which was maintained at 25ºC. A SC4-18 spindle was used, and
than gum were provided by The Pharmaceutical Development its speed was set to 50 rpm.
Service Department of Patheon (Toronto, Canada).
HPLC analysis
Preparation of standard and sample solutions Agilent (Santa Clara, CA) 1100 HPLC system with data acqui-
Preparation of standard stock solutions sition was utilized for the analysis. For the analysis of L-arginine,
the separation was carried out on a YMC Pack C8 column (S-5
µm, 150 mm × 4.6 mm), purchased from YMC (Milford, MA)
51.50 mg of L-arginine was accurately weighed into a 50-mL
volumetric flask, diluted to volume with the mobile phase, and
mixed well. with a total analysis time of 10 min for standards and 30 min for
88.15 mg of amphotericin B was accurately weighed into a samples. The column was maintained at 20ºC. A mixture of
250-mL volumetric flask, and about 50 mL of methanol was buffer solution (50 mM 1-hexanesulfonic acid sodium salt, 25
added. The mixture was sonicated until the amphotericin B was mM sodium phosphate and 0.2% triethylamine, pH = 2.3), and
dissolved, diluted to volume with methanol, and mixed well. methanol with the volume ratio of 75:25 was used as the mobile
phase. The flow rate of the mobile phase was 1.0 mL/min, and the
Preparation of standard working solutions sample injection volume was 20 µL. The effluent was monitored
5.0 mL of standard stock solutions were pipetted into a 50-mL by UV detection at 210 nm. For the analysis of amphotericin B,
the separation was carried out on a YMC Pack C8 column (S-5
µm, 12 nm, 150 mm × 4.6 mm) with a total analysis time of 20
volumetric flask and diluted to volume with the mobile phase for
L-arginine or with methanol for amphotericin B.
min. The column was maintained at 25ºC. A mixture of buffer
Preparation of sample solutions solution (0.1 M sodium phosphate, pH = 3.0) and methanol with
5.0 mL of standard stock solutions and 3.0 mL of the syrup the volume ratio of 30:70 was used as the mobile phase. The flow
vehicle (Ora Sweet SF) were pipetted into a 50-mL volumetric rate of the mobile phase was 1.0 mL/min, and the sample injec-
flask and diluted to volume with the mobile phase for L-arginine tion volume was 20 µL. The effluent was monitored at 405 nm.
or with methanol for amphotericin B.

Results and Discussion

Attempts at quantification of a drug substance in a commer-

cialized syrup focused on diluting the syrup to 10 to 20 times and
subsequently injecting the diluted syrup solution into an HPLC
system. The first injection resulted in good separation and quan-
titation. However, peak shape (tailing factor, T, and peak width,
W1/2) and column efficiency (theoretical plate number, N) deteri-
orated rapidly from the second injection and afterward. Figure 1
illustrates the deterioration of the peak shape and column effi-
ciency. With the injection of less diluted syrup solution (four
times dilution), a brand new column was destroyed after the first
injection. Because the injected solution had been filtered
through a 0.45-µm filter and the column pressure did not
Figure 1. Chromatograms of consecutive injections of a diluted syrup sample
increase significantly, it was unlikely that particulates had
spiked with amphotericin B (retention time: 5.1 min): (A) First injection
(T = 1.0, N = 2613, W1/ = 0.228 for the major peak amphotericin B
blocked the frits. After examining the composition of the syrup,
2
unless otherwise specified), (B) second injection (T = 0.95, N = 1537, W1/ = attention was drawn to the presence of xanthan gum, a water-sol-
2
0.302), (C) third injection (T = 0.96, N = 946, W1/ = 0.413), and (D) fourth uble polymer widely used as stabilizer and rheology modifier in
2
the pharmaceutical industry. The typical molecular weight range
of xanthan gum is 2 × 106 to 5 × 107 (v). Adsorption of xanthan
injection (peak split).

105
 Journal of Chromatographic Science, Vol. 48, February 2010

gum on the packed particles of the column would be extremely quickly. To investigate the efficiency of removing the interfering
difficult to eliminate under normal HPLC conditions, thus dete- xanthan gum from the matrix by the use of ultra-filtration, a few
riorating the column efficiency or even damaging the column. It experiments were carried out.
was therefore critical to separate xanthan gum from sample solu- The first experiment was to determine the viscosity of xanthan
tions prior to injection. gum standard aqueous solution before and after ultra-filtration.
The experiment was performed at four concentration levels:
Determination of xanthan gum in the syrup 1261.5 µg/mL, 267.5 µg/mL, 126.15 µg/mL, and 26.75 µg/mL
Xanthan gum is a long chain polysaccharide composed of the labeled as 1, 2, 3, and 4 in Figure 3. For each concentration level,
sugars glucose, mannose, and glucuronic acid. Xanthan gum is the first column represents the viscosity of the xanthan gum
widely used as a thickener, stabilizer, emulsifier, and foaming standard aqueous solution; the second column represents the
agent in food and pharmaceutical industry. As demonstrated pre- viscosity of the filtrate of the xanthan gum standard aqueous
viously, adsorption of xanthan gum onto the stationary phase solution after filtration; the third column represents the
destroys columns. In order to perform quantitative analyses of viscosity of deionized waster (see Figure 3 legend). The results
active pharmaceutical ingredients of interest in a syrup sample, demonstrate that the viscosity of the filtrates of the xanthan gum
it is important to remove xanthan gum from the syrup and to standard aqueous solutions after ultra-filtration decreased to the
analyze the concentration of xanthan gum in the syrup. same value of deionized water, which means that most xanthan
A series of xanthan gum aqueous standard solutions were pre- gum was efficiently retained in the filter after ultra-filtration The
pared by serial dilution of a stock standard solution and the vis- second experiment was to determine the viscosity of diluted
cosities of each standard solution were determined. It was found syrup (20 times dilution) and the same diluted syrup spiked with
that the viscosity changed linearly with the xanthan gum con- xanthan gum standard before and after ultra-filtration. The
centration from 12.6 µg/mL to 1.26 mg/mL (linear correlation results from this experiment demonstrated that the viscosity of
coefficient: 0.995, a = 0.0232, b =1.2999). To estimate the con- both solutions decreased significantly to the same value after
centration of xanthan gum in the syrup, two calibration methods ultra-filtration. This suggested that even with the presence of
were used. The first one was standard addition. 5 mL of the syrup other matrix components, xanthan gum could be separated from
was diluted to 100 mL, and the solution was used as the sample. the matrix with the filter device. Unfortunately, the concentra-
The sample was spiked with a standard solution with 5 mL incre- tions of xanthan gum in the filtrates were much lower than the
ments. The viscosity of the sample and the spiked samples was limit of quantitation of the previous two methods developed for
determined. Figure 2 presents the results from which the con- the analysis of xanthan gum. Highly sensitive techniques, such
centration of xanthan gum was estimated as ca. 3.8 mg/mL as chromatography, are required to determine the concentra-
(equivalent concentration, assuming the xanthan gum used in tions of xanthan gum in the filtrate.
the syrup is the same as the xanthan gum used as the standard).
The second calibration method was external calibration. An ana- Analysis of L-arginine and amphotericin B in the syrup
lytically prepared syrup vehicle, consisting of sucrose, sorbital, Dialysis has long been used for elimination of proteins from
sodium saccharin, strawberry flavor, citric acid, sodium citrate, bio-samples. Dialysis membranes separate molecules according
methylparaben, propylparaben, and deionized water, was used as to their shape and size (14–16). However, hours and even days
the matrix. The viscosities of the standard xanthan gum solu- are required to reach an equilibrium. Ultra-filtration accelerates
tions prepared in the syrup vehicle were determined. The the separation in minutes by the use of high pressure. Special
repeatability from the six standard solutions (0.08%, w/w) was designed centrifugal ultra-filtration filter devices are widely used
1%. The accuracy was within 95.0% to 105.0% at two concen- to concentrate high molecular weight compounds (e.g., pro-
tration levels (0.04% and 0.2%, w/w). The linear correlation coef- teins) after filtering low molecular compounds and solvents. In
ficient was 0.994 from concentrations of 0.02% to 0.4% (w/w). It this study, the devices were investigated for the elimination of
was, therefore, demonstrated that both methods are suitable for xanthan gum, while filtrates were collected for analysis.
quantitative analysis of xanthan gum in the syrup vehicle. L-arginine is an amino acid. It is a small molecule compound
With such a high concentration of xanthan gum in the syrup,
injection of even diluted samples would destroy the column

Figure 3. Investigation of the efficiency of removal of xanthan gum from the

Figure 2. Determination of xanthan gum concentration by standard addition. matrixes.

106
Journal of Chromatographic Science, Vol. 48, February 2010

(molecular weight = 174) with high hydrophilicity (Figure 4). Its maybe the retaining of amphotericin B with high molecular
syrup samples were prepared by spiking L-arginine standard weight polymers. Because amphotericin B is a yellow compound,
solutions into the syrup. The prepared samples were then diluted the change of the white color of the membrane of the filter
10 times with the mobile phase prior to ultra-filtration. The fil- devices to yellow clearly demonstrated the adsorption of ampho-
trates were collected for HPLC analysis. Figure 5 shows the chro- tericin B on the membrane. To completely recover amphotericin
matograms of the syrup, L-arginine standard, and spiked B, the strength of the diluent was further increased. Only when
samples. There were no significant interfering peaks from the 100% methanol was used as the diluent did the recovery of
syrup at the retention time of L-arginine, so quantitation for L- amphotericin B reach from 98.5% to 100.6% at different concen-
arginine was unambiguous. The mean recovery of three L-argi- tration levels. Further investigation demonstrated that the repro-
nine samples was 100.3 ± 0.8%, which implied that the filter ducibility (%RSD, peak area) for the filtered samples was 0.3% for
devices neither retained nor adsorbed L-arginine. The repro- 30 injections. The linear range was 0.28–28.2 µg/mL. The limit of
ducibility (%RSD, peak area) of L-arginine in the filtered samples detection (LOD) was estimated to be 0.06 µg/mL [signal-to-noise
was less than 0.5% for 20 injections, and the peak shape was ratio (S/N) = 4.1]. The limit of quantification (LOQ) was estimated
maintained through out the whole analysis. The constant injec- to be 0.28 µg/mL (S/N = 22.7). The %RSD of the recovery of three
tion and peak shape could not be achieved without the removal samples at the concentration of LOQ was 1.7%. These results sug-
of high molecular interfering compounds such as xanthan gum gested that the developed method was suitable for quantifying
from the samples by ultra-filtration. The preliminary study sug- amphotericin B in the syrup.
gested that the use of ultra-filtration as the sample preparation Finally, it was interesting to notice that precipitates were
tool was feasible for quantifying L-arginine in the syrup. observed when the samples were diluted with methanol. It was
The analysis of L-arginine appeared to be a simple case because assumed that the precipitates were salts, which are not soluble in
the interaction of L-arginine with the matrix and the filter devices methanol. But we learned later that xanthan gum is not soluble
was negligible. A more complicated situation is the analysis of in methanol. This phenomenon suggested an alternative
amphotericin B in the syrup. Amphotericin B is a larger molec- approach to quantifying amphotericin B in the syrup: That is,
ular compound (molecular weight = 924) with high hydropho- dilute the samples with methanol, centrifuge the diluted samples
bicity. The interaction of amphotericin B with the matrix and the in centrifuge tubes, dilute the supernatant with water to
filter devices would be expected to be higher than that of L-argi- decrease the strength before injection, and inject into HPLC.
nine with the matrix and the filter devices. It was observed that It was found that amphotericin B was completely recovered
amphotericin B was not detected when a mixture of methanol (recovery = 98.8%, n = 3) using this method while the column
and water (20:80, v/v) was used as the diluent. The recovery of efficiency and the peak shape were not jeopardized. This seemed
amphotericin B increased to only 26% when the volume ratio of to be simpler than the ultra-filtration method. However, it has to
methanol to water in the diluent was increased to 70:30. The low be realized that this is a case-by-case situation. For example, for
recovery of amphotericin B was easily ascribed to the adsorption the case of L-arginine, ultra-filtration might be simpler because
of amphotericin B on the membranes of the filter devices and dilution prior to injection was not required. In addition, some
other polymers in other syrups may not be precipitated by
methanol, etc. Nevertheless, this is still an interesting method
for sample preparation.

Conclusion

Figure 4. Structures of (A) L-arginine and (B) amphotericin B. The preliminary results demonstrated that the developed
sample preparation method was suitable to remove interfering
high molecular weight components from syrup vehicle by the use
of the filter devices, making possible quantitative analysis of APIs
in the complex matrix by HPLC. The sample preparation proce-
dure was simple and efficient. The recoveries of target analytes are
mostly within 98–102% under optimized extraction and centrifu-
gation conditions. The drawback of the method is that analytes in
the filtrate are not concentrated, which might prevent its use for
the analysis of analytes with extremely low concentrations in the
samples. In these cases, enrichment of analytes are generally
required to meet instrumental detection limits.
A viscosity method was developed and validated for the deter-
mination of xanthan gum. The validation challenged the repro-
Figure 5. Determination of L-arginine in a syrup. Chromatograms of (A)
diluted syrup after ultra-filtration, (B) L-arginine standard, and (C) filtered
ducibility, accuracy, and linearity, and it demonstrated that the
diluted syrup spiked with L-arginine (retention time: 4.2 min). method is suitable to determine the xanthan gum in a syrup with
the concentration of xanthan gum from 0.02% to 0.4% (w/w).

107
 Journal of Chromatographic Science, Vol. 48, February 2010

Future works will include further characterizing the filter duced by lipase-catalyzed acidolysis. Sep. Purif. Technol. 50:
devices, investigating robustness of the method, and applying 184–191 (2006).
8. O. Akoum, D. Richfield, M.Y. Jaffrin, L.H. Ding, and P. Swart. A
the method for the analysis of analytes in other complex study of selected herbicides retention by nanofiltration mem-
matrixes containing interfering ingredients of high molecular branes—The role of organic fouling. J. Membr. Sci. 279: 291–300
weight molecules. (2006).
9. M. Mecozzi and E. Pietrantonio. Carbohydrates proteins and lipids
in fulvic and humic acids of sediments and its relationships with
mucilaginous aggregates in the Italian seas. Mar. Chem. 101: 27–39
(2006).
References 10. M. Lolo, S. Pedreira, C. Fente, B.I. Vázquez, C.M. Franco, and A.
Cepeda. Use of the diphasic dialysis as a new extraction procedure
1. J. Pawliszyn. Sample Preparation: Quo Vadis? Anal. Chem. 75: in the determination of enrofloxacin and ciprofloxacin in egg. Anal.
2543–2558 (2003) Chim. Acta 480: 123–130 (2003).
2. L.R. Snyder, J.J. Kirkland, and J.L. Glajch, ed. Chapter 4 Sample 11. C. Lin, J. Sheu, H. Wu, and Y. Huang. Determination of hydro-
Preparation, in Practical HPLC Method Development, 2nd edition. quinone in cosmetic emulsion using microdialysis sampling cou-
John Wiley & Sons, Inc. Now York. 1997. pled with high-performance liquid chromatography. J. Pharm.
3. S. Lee, G. Park, G. Amy, S. Hong, S. Moon, D. Lee, and J. Cho. Biomed. Anal. 38: 414–419 (2005).
Comparison of ceramic and polymeric membranes for natural 12. G. Jonsson. Transport phenomena in ultra-filtration: membrane
organic matter (NOM) removal. J. Membr. Sci. 201: 191–201 selectivity and boundary layer phenomena. Pure & AppI. Chem. 58:
(2002). 1647–1656 (1986).
4. M. Liakopoulou-Kyriakides, E.S. Tzanakakis, C. Kiparissidis, L.V. 13. L. Ahmad, M.F. Chong, S. Bhatia. Ultra-filtration modeling of mul-
Ekaterianiadou, and D.A. Kyriakidis. Kinetics of xanthan gum pro- tiple solutes system for continuous cross-flow process. J. Chem. Eng.
duction from whey by constructed strains of Xanthomonas Sci. 61: 5057–5069 (2006).
campestris in batch fermentations. Chem. Eng. Technol. 20: 14. J.F. Maher. Principles of dialysis and dialysis of drugs. Am. J. Med.
354–360 (1997). 62: 475–481 (1977).
5. M. Papafianni, S.K. Psomas, L. Batsilas, S.V. Paras, D.A. Kyriakidis, 15. M.A.L. Eriksson, J. Gabrielsson, and L.B. Nilsson. Studies of drug
and M. Liakopoulou-Kyriakides. Characterization of alkaline kerati- binding to plasma proteins using a variant of equilibrium dialysis. J.
nase of Bacillus licheniformis strain HK-1 from poultry waste. Pharm. Biomed. Anal. 38: 381–389 (2005).
Process Biochem. 37: 73–83 (2001). 16. H. Wan and M. Rehngren. High-throughput screening of protein
6. Y. Li, A. Shahbazi, and C.T. Kadzere. Separation of cells and pro- binding by equilibrium dialysis combined with liquid chromatog-
teins from fermentation broth using ultra-filtration. J. Food Eng. 75: raphy and mass spectrometry. J. Chromatogr. A 1102: 125–134
574–580 (2006). (2006).
7. A.F. Vikbjerg, G. Jonsson, H. Mu, and X. Xu. Application of ultra-fil- Manuscript received February 26, 2008;
tration membranes for purification of structured phospholipids pro- Revision received August 4, 2008.

108