- Email: [email protected]
Evaluation of a hydrophilic interaction liquid chromatography design space for sugars and sugar alcohols
Journal of Chromatography A, 1489 (2017) 65–74
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Evaluation of a hydrophilic interaction liquid chromatography design space for sugars and sugar alcohols Evan M. Hetrick a,∗ , Timothy T. Kramer b , Donald S. Risley a a b
Small Molecule Design and Development, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN, 46285, USA Discovery and Development Statistics, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN, 46285, USA
a r t i c l e
a b s t r a c t
i n f o
Article history: Received 21 November 2016 Received in revised form 20 January 2017 Accepted 26 January 2017 Available online 27 January 2017 Keywords: HILIC Sugar Sugar alcohols Carbohydrate CAD ELSD
Based on a column-screening exercise, a column ranking system was developed for sample mixtures containing any combination of 26 sugar and sugar alcohol analytes using 16 polar stationary phases in the HILIC mode with acetonitrile/water or acetone/water mobile phases. Each analyte was evaluated on the HILIC columns with gradient elution and the subsequent chromatography data was compiled into a statistical software package where any subset of the analytes can be selected and the columns are then ranked by the greatest separation. Since these analytes lack chromophores, aerosol-based detectors, including an evaporative light scattering detector (ELSD) and a charged aerosol detector (CAD) were employed for qualitative and quantitative detection. Example qualitative applications are provided to illustrate the practicality and efficiency of this HILIC column ranking. Furthermore, the design-space approach was used as a starting point for a quantitative method for the trace analysis of glucose in trehalose samples in a complex matrix. Knowledge gained from evaluating the design-space led to rapid development of a capable method as demonstrated through validation of the following parameters: specificity, accuracy, precision, linearity, limit of quantitation, limit of detection, and range. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Sugars and sugar alcohols can be found naturally in plant tissues and the fiber of fruits and vegetables and are frequently used by the food industry as thickeners or sweeteners as well as by the pharmaceutical industry as inactive ingredients in drug products. Analysis of these compounds is challenging due to their polar nature and lack of a suitable chromophore, rendering the commonly employed reversed-phase (RP) chromatography with ultraviolet (UV) or fluorescence detection impractical without some form of sample derivatization. Chromatographic techniques such as capillary electrophoresis (CE), gas chromatography (GC), and high performance liquid chromatography (HPLC) have been used in the analysis of sugars and sugar alcohols [1–6]. One popular approach for carbohydrate analysis is HPLC using anion-exchange chromatography with pulsed amperometic detection, which has been applied to various matrices [7–12]. Refractive index detectors (RID) have also been used for the direct detection of sugars and sugar alcohols [13,14]; however, RID is often plagued by poor sensitivity and incompatibility with gradient elution. In contrast, applying HPLC in the HILIC
∗ Corresponding author. E-mail address: [email protected] (E.M. Hetrick). http://dx.doi.org/10.1016/j.chroma.2017.01.072 0021-9673/© 2017 Elsevier B.V. All rights reserved.
mode with aerosol-based detectors is one of the better options for the separation and determination of sugars and sugar alcohols to achieve retention and selectivity as well as accurate and low level direct detection. The HILIC concept was first applied to the HPLC retention of sugars using polar stationary phases [15,16] although the HILIC terminology was later coined by Andrew Alpert for the separation of proteins, nucleic acids and polar molecules in 1990 [17]. Aerosol-based detectors, such as the evaporative light scattering detector (ELSD), charged aerosol detector (CAD) and nano quantity analyte detector (NQAD), have a broad application base and the concept and operation have been previously described [18–20]. A comparison of the three types of aerosol detectors has also recently been reported [21]. Several researchers have used HPLC-ELSD or HPLC-CAD for the analysis of a few sugars and sugar alcohol mixtures in various matrices [22,23]. In this report, we investigated 16 polar stationary phases and mobile phase combinations in the HILIC mode for the retention and separation of 26 sugar and sugar alcohol analytes. Because all of the analytes tested here lack a sufficient chromophore, aerosol-based detectors were used for the different analyses. A straightforward ranking algorithm was then developed to facilitate simple screening of the stationary phase/mobile phase combinations and identifying those that give the best separation for any given subset of analytes. The algorithm greatly speeds development of an analytical method for such
66
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
compounds by generating the best stationary phase and mobile phase combination for any combination of analytes making method development more efficient. The value of the ranking algorithm is demonstrated here for both qualitative and quantitative applications.
mobile phase A to 50% at 6.1 min. Mobile phase A was held at 50% until 10.5 min to quickly elute trehalose from the column. To facilitate re-equilibration, mobile phase B was returned to 8% at 10.6 min and held at 8% until 20.0 min for a total run time of 20.0 min. 3. Results and discussion
2. Materials and methods 3.1. Design space investigation 2.1. Chemicals Acetonitrile and acetone were purchased from EMD Sciences Inc. (Gibbstown, NJ). Deionized water and nitrogen were from an inhouse system. The analytes (i.e., sugars, sugar alcohols, and artificial sweetener) were obtained from the vendors listed in Table 1. 2.2. Equipment − design space investigation The HPLC system consisted of an Agilent 1100 pump and auto sampler (Santa Clara, CA). Detection was performed with an Alltech 3300 ELSD from Buchi Corporation (New Castle, DE) or an ESA Corona CAD from Thermo Scientific (Waltham, MA). The operating conditions for the ELSD were a temperature of 50 ◦ C, nitrogen gas flow of 1.4 L/min and gain setting of 2. The operating conditions for the CAD were a nebulizer temperature of 30 ◦ C, the range set to 200 pA and the nitrogen pressure set to 35 psi. The HPLC columns evaluated are listed in Table 2. The column temperature was maintained at 25 ◦ C throughout the analysis. The injection volume was 10 L while the mobile phase flow rate was 1.0 mL/min. For the design space investigation, each analyte was tested on each column using a gradient starting at 10% water and 90% organic (either acetonitrile or acetone) and then a linear gradient to 60% organic in 30 min, followed by a 4 min hold and a return to the starting condition in 0.5 min and then equilibration for 5.5 min for a total run time of 40 min. Single replicates were used to obtain retention data. 2.3. Sample preparation for design space investigation An individual stock solution of each sugar and sugar alcohol was prepared at 5 mg/mL in water. With the exception of lactitol, raffinose, and maltotriose, samples for HPLC analysis were prepared from the stock by transferring 100 L of the stock solution into an HPLC vial and then adding 900 L of either acetonitrile or acetone to be consistent with the mobile phase organic modifier used during the analysis. Due to the solubility of lactitol, raffinose and maltotriose, 100 L of the stock solution was added to an HPLC vial followed by 100 L of water and then 800 L of the appropriate organic solvent. All of these samples were used to acquire the design space investigation data. 2.4. Trehalose analysis method conditions for validation experiments Final method conditions for glucose analysis in trehalose employed a 150 × 4.6 mm, 2.6-m particle size AccucoreTM HILIC HPLC column with an ESA CAD. Mobile phase A consisted of Milli-Q water and mobile phase B consisted of acetonitrile (HPLC grade). The diluent was a 60/40 (v/v) mixture of acetonitrile and Milli-Q water. For calibration, five glucose standards were prepared over a range of approximately 0.00072 mg/mL–0.0072 mg/mL in diluent. Trehalose samples (USP grade) were prepared at approximately 40 mg/mL on an anhydrous basis. The glucose samples used were USP grade. For HPLC analysis, the column temperature was set to 55 ◦ C with a flow rate of 1.0 mL/min. The injection volume was 10 L. Mobile phase A was held at 8% for 6.0 min (92% mobile phase B) during which time glucose eluted, followed by an increase in
The design space investigation across 26 different analytes (Table 1) with 16 different chromatography columns (Table 2) and two different mobile phase compositions (using the gradient described in the Materials and methods section) produced the results in Tables 3 and 4. The columns used were selected to represent the broad range of commercially-available HILIC stationary phases available, including zwitterionic, amide, diol, triazole, amino, silica, and hydroxyl ligand. Other columns with stationary phases of the same type may exhibit similar selectivity but specific columns not included in this evaluation should be evaluated to assess similarity. Data presented in Tables 3 and 4 include the retention time of each analyte analyzed on all of the stationary phases with both the acetone/water (Table 3) and acetonitrile/water (Table 4) mobile phases. Based on these results, a unique algorithm ® was developed in JMP statistical software (SAS Institute Inc.) to subset the data for the analytes of interest, calculate the intervals between each analyte’s retention time for the stationary and mobile phases, and then create a table and graphic highlighting the minimum intervals for each stationary and mobile phase. Note that a similar algorithm could be developed in other software such as Excel. These results enable the scientist to identify method conditions (i.e., combination of stationary phase and mobile phase) that have better separations (based upon both resolution and total run time) relative to the others. At any time, additional columns or additional analytes not described here could be analyzed as described above to expand the design space algorithm. 3.2. Application of the HILIC design space screen and algorithm: qualitative and quantitative applications ®
The value of the JMP algorithm is evident in its ability to quickly provide a combination of stationary phase and mobile phase that will separate mixtures of known sugars and/or sugar alcohols. These initial method conditions can be used for qualitative analysis or as starting conditions for quantitative analysis (see sections below for relevant qualitative and quantitative case studies). 3.2.1. Qualitative analysis case study Erythritol, xylitol, dextrose, sucrose, and trehalose are all used as sweeteners in various beverages including flavored waters, energy drinks and teas. As such, it is valuable to have a method that separates all of those components that can be used in a qualitative or quantitative manner to rapidly identify which components are present in a given sample. A subset of these five sugars and sugar alcohols was created from the complete set of 26 sugars and sugar ® alcohols dataset and then the JMP algorithm applied. Based on the ® HILIC screen described above and the associated JMP algorithm, the plot in Fig. 1 was created. Fig. 1 shows the retention time of each analyte across each of the 16 columns with the acetonitrile/water mobile phase system. Note for this scenario, while there are five analytes of interest, there are actually six peaks of interest given that dextrose exists as a mixture of two anomers. For this case, it was desirable to separate the two anomers of dextrose. As a result, some stationary phases in Fig. 1 demonstrate six peaks (indicating resolution of the dextrose anomers) while others demonstrate only five peaks (indicating that the dextrose anomers co-elute under those conditions)
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
67
Table 1 Test analytes. Sugar
Sugar Alcohol
Artificial Sweetener
Analyte
Supplier
Analyte
Supplier
Analyte
Supplier
Arabinose
Oakwood Chemical (West Columbia, SC)
Sigma Aldrich (St. Louis, MO)
Sucralose
Spectrum Chemical (New Brunswick, NJ)
Lactose Mannose Rhaminose Xylose Fructose Raffinose Maltose Galactose Sucrose Trehalose Maltotriose Dextrose
Acros Organics (Geel, Belgium) TCI Ltd. (Tokyo, Japan)
Sorbitol Xylitol Threitol Inositol Lactitol
Alfa Aesar (Ward Hill, MA)
Erythritol Arabitol Maltitol Ribitol Mannitol
Sigma Aldrich (St. Louis, MO)
Inditol Ducitol
Fisher Scientific (Fairlawn, NJ)
Alfa Aesar (Ward Hill, MA) Oakwood Chemical (West Columbia, SC) TCI Ltd. (Tokyo, Japan)
Fisher Scientific (Fairlawn, NJ)
Acros Organics (Geel, Belgium)
Table 2 Columns evaluated in the HILIC design space study. Columna
Phase Type
Particle Size
Supplier
Zwitterionic
5 m
Zwitterionic
5 m
ZIC -cHILIC
Zwiterionic
5 m
TSKgel Amide-80
Amide
3 m
AccucoreTM Amide
Amide
2.6 m
Amide
3.5 m
Kromasil 60-5-Diol
Diol
5 m
Cosmosil HILIC
Triazole
5 m
AccucoreTM Urea HILIC
Amino
2.6 m
Amino
3 m
Luna NH2
Amino
3 m
Sepax Polar-Silica
Silica
3 m
AccucoreTM HILIC
Silica
2.6 m
Synchronis HILIC
Silica
3 m
XBridge BEH HILIC
Silica-Hybrid
3.5 m
Penta-Halo HILIC
Penta-Hydroxy-Ligand
2.7 m
Macherey-Nagel (Bethlehem, PA) Merck Sequant (Umeå, Sweden) Merck Sequant (Umeå, Sweden) Tosoh Bioscience, LLC (Montgomery, PA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Kromasil (Brewster, NY) Nacalai Tesque (San Diego, CA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Phenomenex (Torrance, CA) Sepax (Newark, DE) Thermo Scientific (Waltham, MA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Mac-Mod (Chadds Ford, PA)
®
Nuceodur HILIC ®
ZIC -HILIC ®
®
XBridge BEH Amide
®
®
Sperisorb NH2
®
®
a
Column dimensions 150 × 4.6 mm.
or less than 5 peaks indicating additional co-eluting peaks. The two best stationary phases from the standpoint of resolving the dextrose anomers and having suitable resolution between the other ® ® analytes were the ZIC -HILIC and ZIC -cHILIC columns at the top of Fig. 1. Based on its ability to provide a shorter method (i.e., the ® last peak to elute has the shortest retention time), the ZIC -HILIC column with acetonitrile-water was identified as the leading candidate for qualitative or quantitative analysis of the target analytes. ® Another way to analyze the design space screen via the JMP algorithm is presented in Fig. 2. Fig. 2 shows a plot of the max-
imum retention time (Y-axis) vs. the minimum retention time gap (X-axis) for each stationary phase with both mobile phase systems. The optimum separation would have a large retention time gap (X-axis) to maximize resolution with a small maximum retention time (Y-axis) to minimize total run time. This translates to the lower right corner of the plot in Fig. 2. With the ideal location being the lower right corner of Fig. 2, the group of stationary phases/mobile phases that best fit this region are the ® ® ZIC -cHILIC and ZIC -HILIC columns with both acetone and acetonitrile mobile phases. This supports the observation from Fig. 1
68
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Table 3 Retention Time Data for the Test Analytes with Acetone/Water Mobile Phase. 1 and 2 refer to anomers of the same analyte. Sugar Type
Sugar or Sugar Alcohol
Accucore
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
2.267 2.348 2.338 2.338 2.150 2.437 2.446 2.498 2.764 2.764 2.528 2.528 2.782 2.782 2.263 2.436 2.967 2.051 2.171 1.812 2.346 2.811 2.165 2.165 2.347 2.474 2.262 2.483 3.078 3.097 2.717 2.412 2.295 2.429 2.302 2.362
Sugar Type
Sugar or Sugar Alcohol
Sepax Silica
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
2.963 2.963 3.130 3.130 3.018 3.018 3.270 3.407 3.907 3.907 3.471 3.471 3.959 3.959 2.997 2.997 4.249 2.647 2.885 2.229 3.353 3.996 2.815 2.815 3.134 3.365 2.966 3.363 4.331 4.408 3.822 3.295 3.044 3.289 3.010 3.187
Accucore Urea
Accucore-Amide
Cosmosil
4.917 6.353 7.009 7.461 4.677 7.061 7.500 9.150 13.159 13.687 11.327 12.093 15.809 16.479 5.663 8.196 16.423 3.037 4.221 1.978 9.809 13.165 4.149 4.796 4.798 6.754 3.546 6.727 15.383 13.432 11.306 6.396 4.445 6.434 3.714 4.915
4.620 4.620 6.491 6.491 5.278 5.278 6.408 6.408 11.073 11.073 10.155 10.155 13.331 13.331 5.599 5.599 14.261 3.583 3.583 3.156 8.975 11.640 4.530 4.530 4.863 6.242 3.847 5.885 10.216 11.155 9.756 6.198 4.489 6.102 3.886 4.861
Syncronis
Tosoh-Amide
Waters-Amino
4.822 5.947 6.996 6.404 5.520 5.520 6.539 7.781 10.498 11.286 9.459 10.226 12.378 13.053 5.747 5.747 13.820 3.720 4.449 3.105 9.316 11.377 4.517 5.227 5.268 6.664 4.278 6.346 10.323 11.119 10.008 6.425 4.984 6.253 4.362 5.248
7.668 9.402 10.700 10.214 7.467 10.256 10.705 12.409 16.581 17.020 14.898 15.560 19.149 19.685 8.752 11.472 19.753 5.100 6.929 2.950 13.536 16.579 6.729 7.580 7.599 9.931 5.773 9.895 17.982 16.825 14.862 9.588 7.096 9.610 5.990 7.776
3.031 3.474 3.619 3.845 3.311 3.311 3.817 4.401 5.734 6.043 5.232 5.607 7.208 7.557 3.151 3.151 7.047 2.397 3.063 1.933 4.283 5.596 2.699 2.963 2.919 3.492 2.495 3.476 6.255 5.643 4.987 3.373 2.738 3.381 2.564 2.962
4.037 4.037 4.762 4.762 4.517 4.517 5.240 5.240 8.464 8.464 6.864 6.864 8.791 8.791 4.702 4.702 9.736 3.050 3.050 2.234 6.043 7.827 3.543 3.543 4.103 5.215 3.334 5.129 7.868 9.193 7.231 5.029 3.760 5.082 3.430 4.192
Kromasil-Diol 4.100 4.489 5.039 5.159 4.005 4.984 5.163 5.651 7.648 7.913 6.973 7.195 9.139 9.282 4.614 5.351 9.703 3.344 3.821 2.363 6.280 8.305 3.815 4.052 4.314 5.169 3.698 5.017 8.047 8.278 7.244 4.982 4.109 4.916 3.787 4.316 XBridge-BEH-Amide 5.700 7.075 7.874 8.308 5.631 7.841 8.251 9.820 13.789 14.267 12.227 12.889 16.560 17.136 6.650 8.954 17.210 3.841 5.125 2.512 10.914 13.938 5.027 5.670 5.653 7.585 4.281 7.502 15.167 14.088 12.171 7.267 5.272 7.275 4.445 5.774
Luna-Amino
Nucleodur
Penta-Halo
4.708 4.708 6.106 6.106 4.934 4.934 6.166 6.166 8.780 8.780 8.195 8.195 10.765 10.765 5.316 5.316 9.817 3.306 3.306 2.497 6.890 8.748 4.238 4.238 4.583 5.824 3.770 5.499 9.712 8.708 7.760 5.540 4.367 5.323 3.914 4.562
4.636 5.727 6.214 6.803 5.287 5.287 6.354 7.605 10.254 11.040 9.303 10.065 12.355 13.009 5.542 6.721 13.699 3.578 4.278 2.966 9.174 11.171 4.338 5.025 5.094 6.442 4.089 6.152 10.135 10.956 9.843 6.214 4.794 6.053 4.189 5.054
3.444 3.987 4.457 4.643 3.763 3.763 4.683 5.377 7.879 8.249 7.248 7.573 10.325 10.580 3.908 3.908 10.788 2.631 3.284 2.161 6.280 8.814 3.082 3.358 3.613 4.726 2.881 4.553 8.475 8.720 7.577 4.530 3.353 4.434 2.965 3.634
XBridge-HILIC
ZIC-cHILIC
ZIC-HILIC
7.658 9.916 11.221 12.430 8.995 9.984 11.177 13.400 16.800 17.876 15.796 16.871 19.662 20.516 9.972 11.924 20.727 5.391 6.747 4.034 15.336 18.060 7.369 8.825 8.282 10.991 6.131 10.267 17.021 16.967 15.805 10.775 7.842 10.174 6.271 8.027
6.618 8.790 9.077 10.373 6.747 8.398 9.423 11.712 14.142 15.474 12.528 13.848 15.448 16.413 7.742 10.296 16.390 3.881 5.325 2.325 11.651 14.497 5.847 7.483 6.199 8.404 4.646 8.126 16.440 13.798 12.261 8.049 5.861 7.884 4.797 6.256
2.592 2.592 2.636 2.636 2.571 2.571 2.771 2.771 3.106 3.106 2.885 2.885 3.209 3.209 2.567 2.567 3.394 2.376 2.376 2.125 2.809 3.177 2.459 2.459 2.627 2.754 2.534 2.747 3.233 3.363 3.053 2.700 2.569 2.710 2.556 2.641
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
69
Table 4 Retention Time Data for the Test Analytes with Acetonitrile/Water Mobile Phase. 1 and 2 refer to anomers of the same analyte. Sugar Type
Sugar or Sugar Alcohol
Accucore
Accucore Urea
Accucore-Amide
Cosmosil
Kromasil-Diol
Luna-Amino
Nucleodur
Penta-Halo
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
3.219 3.485 3.811 3.811 3.199 3.867 4.032 4.234 6.604 6.604 5.791 5.791 8.321 8.321 3.594 4.010 9.114 2.721 2.985 2.007 5.536 6.987 2.918 2.997 3.899 4.639 3.341 4.600 5.951 8.240 7.044 4.531 3.738 4.535 3.406 3.957
3.753 4.221 5.095 5.400 4.514 4.514 5.152 5.737 8.687 8.988 8.458 8.905 11.886 12.268 4.646 4.646 11.664 3.132 3.624 2.467 7.332 8.976 3.529 3.856 4.089 5.249 3.229 5.030 7.726 9.017 8.406 5.089 3.871 4.996 3.280 4.064
7.197 8.764 11.019 11.566 10.048 10.048 11.264 12.610 18.357 18.625 17.214 17.778 22.303 22.699 9.705 11.378 22.247 4.876 5.872 2.463 15.726 18.393 6.496 7.383 8.286 11.464 5.723 11.205 18.120 19.016 17.482 11.117 7.957 11.073 5.898 8.333
7.218 7.218 10.659 10.659 8.633 8.633 10.819 10.819 17.720 17.720 17.084 17.084 21.239 21.239 9.636 9.636 21.830 5.596 5.596 4.311 15.599 18.047 7.043 7.043 8.236 10.979 5.953 10.301 15.395 18.180 16.892 10.970 7.781 10.696 5.905 8.090
6.156 6.869 9.100 9.472 7.677 7.677 9.150 9.952 15.312 15.646 14.768 15.173 19.153 19.420 8.528 8.528 19.673 5.240 5.473 3.481 13.854 16.269 6.007 6.464 7.654 9.886 5.889 9.455 13.734 16.681 15.470 9.769 7.334 9.466 5.891 7.533
6.450 6.450 8.887 8.887 7.223 7.223 9.122 9.122 13.091 13.091 12.918 12.918 16.378 16.378 8.251 8.251 15.594 4.792 4.792 3.292 11.492 13.392 6.045 6.045 6.953 8.974 5.355 8.412 12.548 13.505 12.661 8.745 6.696 8.355 5.454 6.781
6.011 7.268 9.031 9.821 7.649 7.649 9.052 10.339 14.933 15.687 14.322 15.110 18.605 19.221 8.329 9.049 19.588 4.918 5.293 3.851 13.973 16.097 5.898 6.803 7.498 9.828 5.653 9.301 13.657 16.168 15.141 9.633 7.218 9.284 5.677 7.315
5.544 5.544 8.017 8.427 6.710 6.710 8.043 9.010 14.155 14.513 13.608 14.016 18.028 18.288 7.444 7.444 18.265 4.214 4.214 2.612 12.462 14.964 5.314 5.314 6.485 8.798 4.642 8.476 12.990 15.251 14.199 8.686 6.171 8.461 4.704 6.398
Sugar Type
Sugar or Sugar Alcohol
Sepax Silica
Syncronis
Tosoh-Amide
Waters-Amino
XBridge-BEH-Amide
XBridge-HILIC
ZIC-cHILIC
ZIC-HILIC
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
3.954 4.230 4.763 4.763 4.745 4.745 5.023 5.238 7.363 7.363 6.779 6.779 9.324 9.324 4.481 4.899 9.940 3.418 3.710 2.490 6.527 8.016 3.674 3.745 4.892 5.740 4.233 5.665 7.169 9.195 8.069 5.572 4.714 5.545 4.324 4.913
6.179 7.437 9.048 9.821 7.715 7.715 9.091 10.385 15.107 15.794 14.293 15.108 18.432 19.037 8.469 8.469 19.498 5.051 5.405 3.981 13.994 15.991 6.044 6.953 7.579 9.826 5.782 9.310 13.585 16.132 15.032 9.675 7.301 9.376 5.786 7.460
11.150 12.796 15.353 15.893 12.497 14.339 15.650 16.893 22.342 22.580 21.515 22.000 26.091 26.421 14.079 15.671 26.098 8.394 9.681 4.475 20.189 22.514 10.434 11.499 12.570 15.797 9.416 15.560 21.774 22.930 21.715 15.515 12.181 15.420 9.583 12.574
7.999 7.999 9.664 9.664 9.121 9.121 10.758 10.758 17.595 17.595 15.697 15.697 19.254 19.254 9.933 9.933 20.514 5.606 5.606 3.281 14.116 16.414 6.711 6.711 8.348 10.844 6.101 10.428 14.591 18.598 16.264 10.730 7.662 11.049 6.066 8.517
8.807 9.525 12.038 12.566 9.317 10.999 12.223 13.488 19.092 19.365 18.270 18.789 23.107 23.477 10.841 12.264 23.070 5.995 6.896 3.402 16.934 19.393 7.603 8.470 9.315 12.367 6.663 12.059 18.428 19.715 18.465 12.099 8.934 11.924 6.785 9.262
10.029 11.954 14.511 15.543 12.547 12.547 14.356 15.911 20.252 20.905 19.609 20.412 23.477 24.068 13.644 14.355 23.996 7.998 8.438 5.258 19.022 21.111 9.992 11.465 11.772 14.849 8.869 14.046 19.457 20.696 19.816 14.681 11.534 14.126 8.838 11.408
7.950 9.753 11.327 12.406 9.743 9.743 11.440 13.055 16.717 17.545 15.685 16.708 19.059 19.871 10.447 11.625 19.896 5.653 6.389 3.263 15.042 17.149 7.601 9.067 8.902 11.497 6.654 11.021 16.948 17.086 15.941 11.266 8.642 11.007 6.684 8.787
3.221 3.395 3.815 3.815 3.628 3.628 3.892 4.047 5.992 5.992 5.509 5.509 7.717 7.717 3.655 3.877 8.254 2.928 3.069 2.345 5.325 6.506 3.126 3.126 3.794 4.396 3.326 4.298 5.241 7.239 6.412 4.314 3.682 4.263 3.392 3.799
70
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Fig. 1. Bubble plot of dextrose anomers, erythritol, sucrose, trehalose, and xylitol separation based on retention time and the type of HPLC column.
Fig. 2. Application of the column/mobile phase selection algorithm to a mixture of dextrose anomers, erythritol, xylitol, sucrose, and trehalose.
above. The best combination graphically (i.e., largest minimum retention time gap and shortest maximum retention time) is the ® ZIC -HILIC column with the acetone/water mobile phase system. Based on the observations in Figs. 1 and 2, isocratic conditions ® were experimentally applied using the ZIC -HILIC column with the acetone (82.5%)/water and acetonitrile (80%)/water mobile phases for the separation of erythritol, xylitol, dextrose (anomers), sucrose and trehalose. Additionally, three beverages were evaluated which included Hetley’s Organic Black Tea (containing xylitol and sucrose), Halo Trehalose Infusion Water Blackberry/Plum (containing trehalose and sugars) and BAI Antioxidant Infusion Molokai Coconut (containing erythritol). It was determined that acetoni® trile/water mobile phase with the ZIC -HILIC column provided a better overall separation of erythritol, xylitol, dextrose (anomers), sucrose and trehalose (due to the added matrix peaks in the HALO beverage). The resulting chromatogram overlay for the separation is shown in Fig. 3 with the sweeteners found in the different beverages matching the corresponding sugar or sugar alcohol in the standard preparation (analyte concentration = 0.5 mg/mL). The pro® cess described above illustrates how the user of this JMP algorithm can quickly achieve an efficient separation and method for a subset
mixture from the 26 sugars and sugar alcohols shown in Table 1. This example with beverage analysis demonstrates that the HILIC design space can be used as a guide. Complicated sample matrix components may require further development and/or method conditions may need to be modified.
3.2.2. Quantitative analysis case study: controlling a potential impurity in a pharmaceutical excipient The value of the HILIC screen for quantitative analysis of sugars and sugar alcohols is best illustrated with a case study where the screen has been applied to develop a control strategy for a pharmaceutical development project. The application described here involves a lyophilized drug product where the active pharmaceutical ingredient (API) is a small molecule containing a primary amine. The development team recognized that the presence of a primary amine presented the risk for the Maillard reaction occurring between the API and reducing sugars, leading to potential impurity formation [24]. For this reason, non-reducing carbohydrates were chosen as potential excipients to be evaluated during formulation screening studies, with trehalose (Fig. 4) as the leading candidate. However, during initial screening studies with trehalose
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
71
®
Fig. 3. Chromatogram overlays using the ZIC -HILIC column with a mobile phase consisting of 80% acetonitrile/20% water at a flow rate of 1.0 mL/min using ELSD detection. Chromatogram A = blank, chromatogram B = standard mixture, chromatogram C = Hetley’s Organic Black Tea, chromatogram D = Halo Trehalose Infusion Water (Blackberry/Plum), and chromatogram E = BAI Antioxidant Infusion (Molokai Coconut). Peak identification: peak 1 = erythritol, peak 2 = xylitol, peak 3 and 4 = dextrose anomers, peak 5 = sucrose, and peak 6 = trehalose.
Fig. 4. Structure of trehalose, a non-reducing sugar evaluated as a potential excipient.
as the primary excipient, an unexpected impurity was observed to grow during stability studies. The impurity was identified as a glucose adduct formed via the Maillard reaction between glucose (a reducing sugar) and the primary amine of the API. The structure of the glucose adduct was confirmed via NMR, mass spectrometry, and RP-HPLC retention time match with an authentic sample (data not shown). Glucose was not intentionally added to the formulation and it was determined that the source of the glucose was as a residual impurity in the trehalose excipient. Trehalose is a disaccharide with one molecule of trehalose consisting of two molecules of glucose bound at the anomeric carbon. Thus, while glucose is a reducing sugar (and thus able to participate in the Maillaird reaction), trehalose itself is not a reducing sugar. As part of a robust control
strategy, a HILIC-HPLC method was pursued to detect glucose in trehalose as an upstream control to prevent glucose adduct formation in the drug product. Based on the trehalose:API ratio in the formulation, the desire to control the glucose adduct to limits specified by the International Conference on Harmonisation (ICH) [25], and the reaction stoichiometry observed during stability studies, a limit of not more than 0.02% (wt/wt) glucose in trehalose was calculated. The excipient manufacturer controlled reducing sugars with a method and to limits specified in the United States Pharmacopeia; however, tighter control was required based on the desired control described above. Thus, a new method needed to possess the selectivity, sensitivity, linearity, precision, and accuracy necessary to quantitate glucose in trehalose at the 0.02% (wt/wt) level and below. The HILIC screen developed and described in the previous sections, along with previously-published HILIC method development work [26], proved valuable for rapidly identifying chromatographic conditions that would separate glucose from trehalose and serve as the basis for a method for controlling glucose in trehalose. Mannitol was also being considered as a secondary excipient in the ® formulation and as a result, the JMP algorithm described above was executed with glucose, trehalose, and mannitol as the constituents. However, the list of columns was limited to the ones that did not separate the glucose anomers as one peak was desired for glucose (7 of the 16 columns met this criteria). The output of the ® JMP algorithm including these three analytes is summarized in Fig. 5.
72
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Fig. 5. Application of the column/mobile phase selection algorithm to a mixture of glucose, trehalose, and mannitol.
Fig. 6. Separation in order of elution between glucose (red), mannitol (black), and trehalose (blue) on the AccucoreTM HILIC column with acetonitrile/water mobile phase and the gradient described in the Trehalose Analysis section with CAD detection. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 5 Method conditions for quantitation of glucose in trehalose. Parameter
Description
Analytical Column Column Temperature Autosampler Temperature Mobile Phase A Mobile Phase B Flow Rate Injection Volume Run Time Gradient
Thermo Scientific, AccucoreTM HILIC, 4.6 mm x 150 mm, 2.6 m 55 ◦ C Ambient 100% Milli-Q Water 100% Acetonitrile 1.0 mL/minute 10 L 20 min %A Time (minutes) 8 0.00 6.00 8 50 6.10 50 10.50 8 10.60 8 20.00
CAD Parameters CAD Detector Range Filter Nebulizer Heater Nitrogen Pressure
ESA Corona VEO 100 pA 5.0 s 35 ◦ C 35 psi
%B 92 92 50 50 92 92
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74 Table 6 Accuracy and precision of trehalose spiked with glucose. Level
Number of Preparations
Average Recovery
%RSD
0.009% 0.02% 0.03%
N=3 N=6 N=3
113% 109% 112%
6.7 1.2 0.5
Fig. 5 is a plot of maximum retention time (Y-axis) vs. minimum retention time gap (X-axis) for various column and mobile phase systems for the analytes glucose, trehalose, and mannitol. The optimum separation would have a large minimum retention time gap (X-axis) to maximize resolution with a small maximum retention time (Y-axis) to minimize total run time. This translates to the lower right corner of the plot in Fig. 5. This figure illustrates that ® the top 3 columns are the Waters Spherisorb Amino, Sepax Silica, and AccucoreTM HILIC with acetonitrile/water as the mobile phase. Each of these systems exhibits the desirable characteristic of not resolving anomers of glucose (to maximize sensitivity for glucose ® and facilitate single-peak quantitation). The Waters Spherisorb Amino has the best minimum retention time gap (resolution) but suffers from a longer run time. The AccucoreTM HILIC and Sepax Silica had similar minimum retention time gaps and similar total run times. Based on its slightly shorter analysis time, the Accucore was selected and the separation on this system is shown in Fig. 6. Note that this column/mobile phase combination also exhibits the desirable characteristic of glucose eluting before trehalose. Since glucose is expected to be present at lower levels than trehalose, this prevents the need to integrate a small glucose peak on the tail of a large trehalose peak, which is expected to improve sensitivity for glucose and lead to a more precise, robust method. All subsequent work was conducted with the AccucoreTM HILIC column. Based on further method development experiments, certain conditions were modified slightly from the initial screen described above, including modifications to the gradient and the column temperature. These modifications were made to improve resolution and decrease run time. However, the fundamentals of the separation (HILIC stationary phase and mobile phase components) remained the same. Specific details of the method are shown in Table 5.
73
the first analyst was 109% (see Table 6) and the average recovery from the second analyst was 113% (data not shown). The overall RSD of all 12 preparations was 2.4%, demonstrating acceptable intermediate precision. 3.2.2.1.3. Linearity. Linearity was evaluated with a series of six solutions spanning a range from 0.009% to 0.04% (wt/wt). This range spans the approximate limit of detection of 0.009% (wt/wt) to 2x the expected specification limit of 0.02% (wt/wt). The peak area of glucose vs. the concentration of glucose was plotted and a linear regression was performed. The r2 value of 0.999 along with the insignificant Y-intercept (-3.4) when compared to the slope (5510.7) demonstrated acceptable method linearity. 3.2.2.1.4. Limit of quantitation and limit of detection. Due to the intended purpose of the method, which was to control glucose in trehalose to not more than 0.02% (wt/wt), the level of the lowest standard (0.009% wt/wt) was evaluated as a practical limit of quantitation. Based on the data collected during evaluation of precision described above, a signal-to-noise (S/N) ratio was calculated at the 0.009% level. Based on the n = 3 measurements described above, the S/N ratio was 73 ± 3 (average ± standard deviation), well above the conventional definition of LOQ which is typically the concentration which provides an S/N ratio of 10. Assuming that the S/N scales in a linear fashion with analyte concentration, a calculated LOQ (S/N = 10) would be 0.001% (wt/wt) glucose in trehalose and a calculated LOD (S/N = 3) would be 0.0004% (wt/wt) glucose in trehalose. 3.2.2.1.5. Range. Based on results from linearity, accuracy, and precision determinations, the range of the method was 0.009% (wt/wt) to 0.03% (wt/wt). Although the LOQ was determined to be approximately 0.001% (wt/wt) with a calculated LOD of 0.0004% (wt/wt), ICH Q2(R1) indicates that range is derived from linearity, accuracy, and precision studies [27]. In this case, linearity was evaluated over the range of 0.009% to 0.04% (wt/wt) and accuracy was evaluated over the range of 0.009% to 0.03% (wt/wt). Thus, the range of the method is reported as 0.009% (wt/wt) to 0.03% (wt/wt) which is supported by linearity, accuracy, and precision studies. 4. Conclusions ®
3.2.2.1. Method validation. Using the optimized conditions summarized in Table 5, a series of experiments was conducted to validate the method. Method validation parameters were selected from ICH Q2(R1) [27] and the results are summarized below. 3.2.2.1.1. Specificity. Specificity was evaluated by performing a blank injection consisting of only sample diluent, as well as injecting a “clean” sample of trehalose not spiked with glucose. Absence of interference from the diluent blank and the trehalose matrix combined with the precise and accurate recovery of glucose from spiked trehalose samples (see below Section 3.2.2.1.2) demonstrated the specificity of the method. 3.2.2.1.2. Accuracy and precision. Accuracy and precision were evaluated from the same set of experiments. To demonstrate both accuracy and precision, trehalose was spiked with glucose at the following levels (wt/wt): 0.009%, 0.02%, and 0.03%. These levels spanned the expected range of the method, from the approximate LOQ to 1.5x the intended specification limit. Three preparations were made at both the 0.009% and 0.03% levels while six preparations were made at the 0.02% level and each preparation was individually injected. The recovery (accuracy) and%RSD (precision) are summarized in Table 6 and were suitable for the intended purpose of the method considering the extremely low target levels of the analyte. Intermediate precision was evaluated by having a different analyst prepare a second set of six samples spiked at the 0.02% level on a different day using a different column, different HPLC, and different reagent solutions. The average recovery from
An algorithm was developed using JMP statistical software through the evaluation of 26 sugars and sugar alcohols across 16 different HILIC stationary phases using either acetone/water or acetonitrile/water mobile phases with aerosol-based detectors. This approach to method development for mixtures of sugars and/or sugar alcohols provides a simple and efficient selection process of the stationary phase and mobile phase to rapidly obtain an optimum separation. This strategy was demonstrated in two scenarios through the separation of a mixture of five sugars and sugar alcohols found in different beverages and for the separation of trehalose, glucose, and mannitol in a pharmaceutical formulation. Furthermore, method validation parameters for the trace analysis of glucose using CAD demonstrated excellent linearity, accuracy, precision, and specificity. Acknowledgements The authors wish to acknowledge Matt W. Switzer, David P. Myers, Brian W. Pack, and Mark Strege for insightful technical contributions related to HILIC, chromatography, method validation, and impurity control strategy. References [1] X. Wang, Y. Chen, Determination of carbohydrates as their p-sulfophenylhdrazones by capillary zone electrophoresis, Carbohydr. Res. 332 (2001) 191–196.
74
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
[2] Y. Cao, Y. Wang, X. Chen, J. Ye, Study on sugar profile of rice during ageing by capillary electrophoresis with electrochemical detection, Food Chem. 86 (1) (2004) 131–136. [3] P.M. Medeiros, B.R.T. Simoneit, Analysis of sugars in environmental samples by gas chromatography-mass spectrometry, J. Chromatogr. A 1141 (2007) 271–278. [4] Zs. F. Katona, P. Sass, I. Molnar-Perl, Simultaneous determination of sugars, sugar alcohols, acids and amino acids in apricots by gas chromatography-mass spectrometry, J. Chromatogr. A 847 (1+2) (1999) 91–102. [5] M.A. Adams, Z. Chan, P. Landman, T.D. Colmer, Simulataneous determination by capillary gas chromatography of organic acids, sugars, and sugar alcohols in plant tissue extracts as their trimethylsilyl derivatives, Anal. Biochem. 266 (1) (1999) 77–84. [6] R. Andersen, A. Sorensen, Separation and determination of alditols and sugars by high-pH anion-exchange chromatography with pulsed amperometric detection, J. Chromatogr. A 897 (2000) 195–204. [7] Y. Ding, H. Yu, S. Mou, Direct determination of free amino acids and sugars in green tea by anion-exchange chromatography with integrated pulsed amperometric detection, J. Chromatogr. A 982 (2) (2002) 237–244. [8] S. Ouchemoukh, P. Schweitzer, M. Bachir Bey, H. Djoudad-Kadji, H. Louaileche, HPLC sugar profiles of Algerian honeys, Food Chem. 121 (2) (2010) 561–568. [9] M. Davis, A rapid modified method of compositional carbohydrate analysis of lignocellulosics by high-pH anion-exchange chromatography with pulsed amperometric detection (HPAEC/PAD), J. Wood Chem. Technol. 18 (2) (1998) 235–252. [10] R.I. Cataldi Tommaso, G. Margiotta, L. Lasi, B. Di Chio, C. Xiloyannis, S. Bufo Sabino, Determination of sugar compounds in olive plant extracts by anion-exchange chromatography with pulsed amperometric detection, Anal. Chem. 72 (16) (2000) 3902–3907. [11] A. Caseorp, L. Marr, M. Claeys, A. Kasper-Giebl, H. Puxbaum, A. Pio Casimiro, Determination of saccharides in atmospheric aerosol using anion-exchange high-performance liquid chromatography and pulsed-amperometric detection, J. Chromatogr. A 1171 (1-2) (2007) 37–45. [12] K. Kaiser, R. Benner, Determination of amino sugars in environmental samples with high salt content by high-performance anion-exchange chromatography and pulsed amperometric detection, Anal. Chem. 72 (11) (2000) 2566–2572. [13] W. Xu, L. Liang, M. Zhu, Determination of sugars in molasses by HPLC following solid-phase extraction, Int. J. Food Prop. 18 (3) (2015) 547–557. [14] E. Moro, R. Majocchi, C. Ballabio, S. Molfino, P. Restani, A rapid and simple method for simultaneous determination of glycerol, fructose, and glucose in wine, Am. J. Enol. Vitic. 58 (2) (2007) 279–282.
[15] L.A.Th. Verhaar, B.F.M. Kuster, Contribuion to the elucidation of the mechanism of sugar retention on amine-modified silica in liquid chromatography, J. Chromatogr. 234 (1982) 57–64. [16] R. Orth, H. Englehardt, Separation of sugars on chemically modified silica gel, Chromatographia 15 (1982) 91–96. [17] A.J. Alpert, Hydrophilic-interaction chromatography for the separation of peptides, nucleic acids and other polar compounds, J. Chromatogr. A 499 (1990) 177–196. [18] L.B. Allen, J.A. Koropchak, Condensation nucleation light scattering: a new approach to development of high-sensitivity universal detectors for separations, Anal. Chem. 65 (1993) 841–844. [19] R.W. Dixon, D.S. Peterson, Development and testing of a detection method for liquid chromatography based on aerosol charging, Anal. Chem. 74 (2002) 2930–2937. [20] A. Stolyhwo, H. Colin, G. Guiochon, Use of light scattering as a detector principle in liquid chromatography, J. Chromatogr. A 265 (1983) 1–18. [21] L.-E. Magnusson, D.S. Risley, J.A. Koropchak, Aerosol-based detector for liquid chromatography, J. Chromatogr. A 1421 (2015) 68–81. [22] C. Ma, Z. Sun, C. Chen, L. Zhang, S. Zhu, Simultaneous separation and determination of fructose sorbitol, glucose and sucrose in fruits by HPLC-ELSD, Food Chem. 145 (2014) 784–788. [23] M. Grembecka, A. Lebiedzinska, P. Szefer, Simultaneous separation and determination of erythritol, xylitol, sorbitol, mannitol, maltitol, fructose, glucose, sucrose and maltose in food products by high performance liquid chromatography coupled to charged aerosol detector, Microchem. J. 117 (2014) 77–82. [24] S.W. Baertschi, K.M. Alsante, D. Santafianos, Stress testing: the chemistry of drug degradation, in: S.W. Baertschi, K.M. Alsante, R.A. Reed (Eds.), Pharmaceutical Stress Testing: Predicting Drug Degradation, second ed., Informa Healthcare, London, 2011, pp. 49–141. [25] Impurities in New Drug Products, Q3B(R2), International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2006, www.ich.org. [26] B.A. Olsen, D.S. Risley, V.S. Sharp, B.W. Pack, M.L. Lytle, Pharmaceutical applications of hydrophilic interaction chromatography, in: B.A. Olsen, B.W. Pack (Eds.), Hydrophilic Interaction Chromatography: A Guide for Practitioners, Wiley, Hoboken, NJ, 2013, pp. 111–168. [27] Validation of Analytical Procedures: Text and Methodology, Q2(R1), International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005, www.ich.org.
Contents lists available at ScienceDirect
Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma
Evaluation of a hydrophilic interaction liquid chromatography design space for sugars and sugar alcohols Evan M. Hetrick a,∗ , Timothy T. Kramer b , Donald S. Risley a a b
Small Molecule Design and Development, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN, 46285, USA Discovery and Development Statistics, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN, 46285, USA
a r t i c l e
a b s t r a c t
i n f o
Article history: Received 21 November 2016 Received in revised form 20 January 2017 Accepted 26 January 2017 Available online 27 January 2017 Keywords: HILIC Sugar Sugar alcohols Carbohydrate CAD ELSD
Based on a column-screening exercise, a column ranking system was developed for sample mixtures containing any combination of 26 sugar and sugar alcohol analytes using 16 polar stationary phases in the HILIC mode with acetonitrile/water or acetone/water mobile phases. Each analyte was evaluated on the HILIC columns with gradient elution and the subsequent chromatography data was compiled into a statistical software package where any subset of the analytes can be selected and the columns are then ranked by the greatest separation. Since these analytes lack chromophores, aerosol-based detectors, including an evaporative light scattering detector (ELSD) and a charged aerosol detector (CAD) were employed for qualitative and quantitative detection. Example qualitative applications are provided to illustrate the practicality and efficiency of this HILIC column ranking. Furthermore, the design-space approach was used as a starting point for a quantitative method for the trace analysis of glucose in trehalose samples in a complex matrix. Knowledge gained from evaluating the design-space led to rapid development of a capable method as demonstrated through validation of the following parameters: specificity, accuracy, precision, linearity, limit of quantitation, limit of detection, and range. © 2017 Elsevier B.V. All rights reserved.
1. Introduction Sugars and sugar alcohols can be found naturally in plant tissues and the fiber of fruits and vegetables and are frequently used by the food industry as thickeners or sweeteners as well as by the pharmaceutical industry as inactive ingredients in drug products. Analysis of these compounds is challenging due to their polar nature and lack of a suitable chromophore, rendering the commonly employed reversed-phase (RP) chromatography with ultraviolet (UV) or fluorescence detection impractical without some form of sample derivatization. Chromatographic techniques such as capillary electrophoresis (CE), gas chromatography (GC), and high performance liquid chromatography (HPLC) have been used in the analysis of sugars and sugar alcohols [1–6]. One popular approach for carbohydrate analysis is HPLC using anion-exchange chromatography with pulsed amperometic detection, which has been applied to various matrices [7–12]. Refractive index detectors (RID) have also been used for the direct detection of sugars and sugar alcohols [13,14]; however, RID is often plagued by poor sensitivity and incompatibility with gradient elution. In contrast, applying HPLC in the HILIC
∗ Corresponding author. E-mail address: [email protected] (E.M. Hetrick). http://dx.doi.org/10.1016/j.chroma.2017.01.072 0021-9673/© 2017 Elsevier B.V. All rights reserved.
mode with aerosol-based detectors is one of the better options for the separation and determination of sugars and sugar alcohols to achieve retention and selectivity as well as accurate and low level direct detection. The HILIC concept was first applied to the HPLC retention of sugars using polar stationary phases [15,16] although the HILIC terminology was later coined by Andrew Alpert for the separation of proteins, nucleic acids and polar molecules in 1990 [17]. Aerosol-based detectors, such as the evaporative light scattering detector (ELSD), charged aerosol detector (CAD) and nano quantity analyte detector (NQAD), have a broad application base and the concept and operation have been previously described [18–20]. A comparison of the three types of aerosol detectors has also recently been reported [21]. Several researchers have used HPLC-ELSD or HPLC-CAD for the analysis of a few sugars and sugar alcohol mixtures in various matrices [22,23]. In this report, we investigated 16 polar stationary phases and mobile phase combinations in the HILIC mode for the retention and separation of 26 sugar and sugar alcohol analytes. Because all of the analytes tested here lack a sufficient chromophore, aerosol-based detectors were used for the different analyses. A straightforward ranking algorithm was then developed to facilitate simple screening of the stationary phase/mobile phase combinations and identifying those that give the best separation for any given subset of analytes. The algorithm greatly speeds development of an analytical method for such
66
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
compounds by generating the best stationary phase and mobile phase combination for any combination of analytes making method development more efficient. The value of the ranking algorithm is demonstrated here for both qualitative and quantitative applications.
mobile phase A to 50% at 6.1 min. Mobile phase A was held at 50% until 10.5 min to quickly elute trehalose from the column. To facilitate re-equilibration, mobile phase B was returned to 8% at 10.6 min and held at 8% until 20.0 min for a total run time of 20.0 min. 3. Results and discussion
2. Materials and methods 3.1. Design space investigation 2.1. Chemicals Acetonitrile and acetone were purchased from EMD Sciences Inc. (Gibbstown, NJ). Deionized water and nitrogen were from an inhouse system. The analytes (i.e., sugars, sugar alcohols, and artificial sweetener) were obtained from the vendors listed in Table 1. 2.2. Equipment − design space investigation The HPLC system consisted of an Agilent 1100 pump and auto sampler (Santa Clara, CA). Detection was performed with an Alltech 3300 ELSD from Buchi Corporation (New Castle, DE) or an ESA Corona CAD from Thermo Scientific (Waltham, MA). The operating conditions for the ELSD were a temperature of 50 ◦ C, nitrogen gas flow of 1.4 L/min and gain setting of 2. The operating conditions for the CAD were a nebulizer temperature of 30 ◦ C, the range set to 200 pA and the nitrogen pressure set to 35 psi. The HPLC columns evaluated are listed in Table 2. The column temperature was maintained at 25 ◦ C throughout the analysis. The injection volume was 10 L while the mobile phase flow rate was 1.0 mL/min. For the design space investigation, each analyte was tested on each column using a gradient starting at 10% water and 90% organic (either acetonitrile or acetone) and then a linear gradient to 60% organic in 30 min, followed by a 4 min hold and a return to the starting condition in 0.5 min and then equilibration for 5.5 min for a total run time of 40 min. Single replicates were used to obtain retention data. 2.3. Sample preparation for design space investigation An individual stock solution of each sugar and sugar alcohol was prepared at 5 mg/mL in water. With the exception of lactitol, raffinose, and maltotriose, samples for HPLC analysis were prepared from the stock by transferring 100 L of the stock solution into an HPLC vial and then adding 900 L of either acetonitrile or acetone to be consistent with the mobile phase organic modifier used during the analysis. Due to the solubility of lactitol, raffinose and maltotriose, 100 L of the stock solution was added to an HPLC vial followed by 100 L of water and then 800 L of the appropriate organic solvent. All of these samples were used to acquire the design space investigation data. 2.4. Trehalose analysis method conditions for validation experiments Final method conditions for glucose analysis in trehalose employed a 150 × 4.6 mm, 2.6-m particle size AccucoreTM HILIC HPLC column with an ESA CAD. Mobile phase A consisted of Milli-Q water and mobile phase B consisted of acetonitrile (HPLC grade). The diluent was a 60/40 (v/v) mixture of acetonitrile and Milli-Q water. For calibration, five glucose standards were prepared over a range of approximately 0.00072 mg/mL–0.0072 mg/mL in diluent. Trehalose samples (USP grade) were prepared at approximately 40 mg/mL on an anhydrous basis. The glucose samples used were USP grade. For HPLC analysis, the column temperature was set to 55 ◦ C with a flow rate of 1.0 mL/min. The injection volume was 10 L. Mobile phase A was held at 8% for 6.0 min (92% mobile phase B) during which time glucose eluted, followed by an increase in
The design space investigation across 26 different analytes (Table 1) with 16 different chromatography columns (Table 2) and two different mobile phase compositions (using the gradient described in the Materials and methods section) produced the results in Tables 3 and 4. The columns used were selected to represent the broad range of commercially-available HILIC stationary phases available, including zwitterionic, amide, diol, triazole, amino, silica, and hydroxyl ligand. Other columns with stationary phases of the same type may exhibit similar selectivity but specific columns not included in this evaluation should be evaluated to assess similarity. Data presented in Tables 3 and 4 include the retention time of each analyte analyzed on all of the stationary phases with both the acetone/water (Table 3) and acetonitrile/water (Table 4) mobile phases. Based on these results, a unique algorithm ® was developed in JMP statistical software (SAS Institute Inc.) to subset the data for the analytes of interest, calculate the intervals between each analyte’s retention time for the stationary and mobile phases, and then create a table and graphic highlighting the minimum intervals for each stationary and mobile phase. Note that a similar algorithm could be developed in other software such as Excel. These results enable the scientist to identify method conditions (i.e., combination of stationary phase and mobile phase) that have better separations (based upon both resolution and total run time) relative to the others. At any time, additional columns or additional analytes not described here could be analyzed as described above to expand the design space algorithm. 3.2. Application of the HILIC design space screen and algorithm: qualitative and quantitative applications ®
The value of the JMP algorithm is evident in its ability to quickly provide a combination of stationary phase and mobile phase that will separate mixtures of known sugars and/or sugar alcohols. These initial method conditions can be used for qualitative analysis or as starting conditions for quantitative analysis (see sections below for relevant qualitative and quantitative case studies). 3.2.1. Qualitative analysis case study Erythritol, xylitol, dextrose, sucrose, and trehalose are all used as sweeteners in various beverages including flavored waters, energy drinks and teas. As such, it is valuable to have a method that separates all of those components that can be used in a qualitative or quantitative manner to rapidly identify which components are present in a given sample. A subset of these five sugars and sugar alcohols was created from the complete set of 26 sugars and sugar ® alcohols dataset and then the JMP algorithm applied. Based on the ® HILIC screen described above and the associated JMP algorithm, the plot in Fig. 1 was created. Fig. 1 shows the retention time of each analyte across each of the 16 columns with the acetonitrile/water mobile phase system. Note for this scenario, while there are five analytes of interest, there are actually six peaks of interest given that dextrose exists as a mixture of two anomers. For this case, it was desirable to separate the two anomers of dextrose. As a result, some stationary phases in Fig. 1 demonstrate six peaks (indicating resolution of the dextrose anomers) while others demonstrate only five peaks (indicating that the dextrose anomers co-elute under those conditions)
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
67
Table 1 Test analytes. Sugar
Sugar Alcohol
Artificial Sweetener
Analyte
Supplier
Analyte
Supplier
Analyte
Supplier
Arabinose
Oakwood Chemical (West Columbia, SC)
Sigma Aldrich (St. Louis, MO)
Sucralose
Spectrum Chemical (New Brunswick, NJ)
Lactose Mannose Rhaminose Xylose Fructose Raffinose Maltose Galactose Sucrose Trehalose Maltotriose Dextrose
Acros Organics (Geel, Belgium) TCI Ltd. (Tokyo, Japan)
Sorbitol Xylitol Threitol Inositol Lactitol
Alfa Aesar (Ward Hill, MA)
Erythritol Arabitol Maltitol Ribitol Mannitol
Sigma Aldrich (St. Louis, MO)
Inditol Ducitol
Fisher Scientific (Fairlawn, NJ)
Alfa Aesar (Ward Hill, MA) Oakwood Chemical (West Columbia, SC) TCI Ltd. (Tokyo, Japan)
Fisher Scientific (Fairlawn, NJ)
Acros Organics (Geel, Belgium)
Table 2 Columns evaluated in the HILIC design space study. Columna
Phase Type
Particle Size
Supplier
Zwitterionic
5 m
Zwitterionic
5 m
ZIC -cHILIC
Zwiterionic
5 m
TSKgel Amide-80
Amide
3 m
AccucoreTM Amide
Amide
2.6 m
Amide
3.5 m
Kromasil 60-5-Diol
Diol
5 m
Cosmosil HILIC
Triazole
5 m
AccucoreTM Urea HILIC
Amino
2.6 m
Amino
3 m
Luna NH2
Amino
3 m
Sepax Polar-Silica
Silica
3 m
AccucoreTM HILIC
Silica
2.6 m
Synchronis HILIC
Silica
3 m
XBridge BEH HILIC
Silica-Hybrid
3.5 m
Penta-Halo HILIC
Penta-Hydroxy-Ligand
2.7 m
Macherey-Nagel (Bethlehem, PA) Merck Sequant (Umeå, Sweden) Merck Sequant (Umeå, Sweden) Tosoh Bioscience, LLC (Montgomery, PA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Kromasil (Brewster, NY) Nacalai Tesque (San Diego, CA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Phenomenex (Torrance, CA) Sepax (Newark, DE) Thermo Scientific (Waltham, MA) Thermo Scientific (Waltham, MA) Waters Corporation (Milford, MA) Mac-Mod (Chadds Ford, PA)
®
Nuceodur HILIC ®
ZIC -HILIC ®
®
XBridge BEH Amide
®
®
Sperisorb NH2
®
®
a
Column dimensions 150 × 4.6 mm.
or less than 5 peaks indicating additional co-eluting peaks. The two best stationary phases from the standpoint of resolving the dextrose anomers and having suitable resolution between the other ® ® analytes were the ZIC -HILIC and ZIC -cHILIC columns at the top of Fig. 1. Based on its ability to provide a shorter method (i.e., the ® last peak to elute has the shortest retention time), the ZIC -HILIC column with acetonitrile-water was identified as the leading candidate for qualitative or quantitative analysis of the target analytes. ® Another way to analyze the design space screen via the JMP algorithm is presented in Fig. 2. Fig. 2 shows a plot of the max-
imum retention time (Y-axis) vs. the minimum retention time gap (X-axis) for each stationary phase with both mobile phase systems. The optimum separation would have a large retention time gap (X-axis) to maximize resolution with a small maximum retention time (Y-axis) to minimize total run time. This translates to the lower right corner of the plot in Fig. 2. With the ideal location being the lower right corner of Fig. 2, the group of stationary phases/mobile phases that best fit this region are the ® ® ZIC -cHILIC and ZIC -HILIC columns with both acetone and acetonitrile mobile phases. This supports the observation from Fig. 1
68
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Table 3 Retention Time Data for the Test Analytes with Acetone/Water Mobile Phase. 1 and 2 refer to anomers of the same analyte. Sugar Type
Sugar or Sugar Alcohol
Accucore
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
2.267 2.348 2.338 2.338 2.150 2.437 2.446 2.498 2.764 2.764 2.528 2.528 2.782 2.782 2.263 2.436 2.967 2.051 2.171 1.812 2.346 2.811 2.165 2.165 2.347 2.474 2.262 2.483 3.078 3.097 2.717 2.412 2.295 2.429 2.302 2.362
Sugar Type
Sugar or Sugar Alcohol
Sepax Silica
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
2.963 2.963 3.130 3.130 3.018 3.018 3.270 3.407 3.907 3.907 3.471 3.471 3.959 3.959 2.997 2.997 4.249 2.647 2.885 2.229 3.353 3.996 2.815 2.815 3.134 3.365 2.966 3.363 4.331 4.408 3.822 3.295 3.044 3.289 3.010 3.187
Accucore Urea
Accucore-Amide
Cosmosil
4.917 6.353 7.009 7.461 4.677 7.061 7.500 9.150 13.159 13.687 11.327 12.093 15.809 16.479 5.663 8.196 16.423 3.037 4.221 1.978 9.809 13.165 4.149 4.796 4.798 6.754 3.546 6.727 15.383 13.432 11.306 6.396 4.445 6.434 3.714 4.915
4.620 4.620 6.491 6.491 5.278 5.278 6.408 6.408 11.073 11.073 10.155 10.155 13.331 13.331 5.599 5.599 14.261 3.583 3.583 3.156 8.975 11.640 4.530 4.530 4.863 6.242 3.847 5.885 10.216 11.155 9.756 6.198 4.489 6.102 3.886 4.861
Syncronis
Tosoh-Amide
Waters-Amino
4.822 5.947 6.996 6.404 5.520 5.520 6.539 7.781 10.498 11.286 9.459 10.226 12.378 13.053 5.747 5.747 13.820 3.720 4.449 3.105 9.316 11.377 4.517 5.227 5.268 6.664 4.278 6.346 10.323 11.119 10.008 6.425 4.984 6.253 4.362 5.248
7.668 9.402 10.700 10.214 7.467 10.256 10.705 12.409 16.581 17.020 14.898 15.560 19.149 19.685 8.752 11.472 19.753 5.100 6.929 2.950 13.536 16.579 6.729 7.580 7.599 9.931 5.773 9.895 17.982 16.825 14.862 9.588 7.096 9.610 5.990 7.776
3.031 3.474 3.619 3.845 3.311 3.311 3.817 4.401 5.734 6.043 5.232 5.607 7.208 7.557 3.151 3.151 7.047 2.397 3.063 1.933 4.283 5.596 2.699 2.963 2.919 3.492 2.495 3.476 6.255 5.643 4.987 3.373 2.738 3.381 2.564 2.962
4.037 4.037 4.762 4.762 4.517 4.517 5.240 5.240 8.464 8.464 6.864 6.864 8.791 8.791 4.702 4.702 9.736 3.050 3.050 2.234 6.043 7.827 3.543 3.543 4.103 5.215 3.334 5.129 7.868 9.193 7.231 5.029 3.760 5.082 3.430 4.192
Kromasil-Diol 4.100 4.489 5.039 5.159 4.005 4.984 5.163 5.651 7.648 7.913 6.973 7.195 9.139 9.282 4.614 5.351 9.703 3.344 3.821 2.363 6.280 8.305 3.815 4.052 4.314 5.169 3.698 5.017 8.047 8.278 7.244 4.982 4.109 4.916 3.787 4.316 XBridge-BEH-Amide 5.700 7.075 7.874 8.308 5.631 7.841 8.251 9.820 13.789 14.267 12.227 12.889 16.560 17.136 6.650 8.954 17.210 3.841 5.125 2.512 10.914 13.938 5.027 5.670 5.653 7.585 4.281 7.502 15.167 14.088 12.171 7.267 5.272 7.275 4.445 5.774
Luna-Amino
Nucleodur
Penta-Halo
4.708 4.708 6.106 6.106 4.934 4.934 6.166 6.166 8.780 8.780 8.195 8.195 10.765 10.765 5.316 5.316 9.817 3.306 3.306 2.497 6.890 8.748 4.238 4.238 4.583 5.824 3.770 5.499 9.712 8.708 7.760 5.540 4.367 5.323 3.914 4.562
4.636 5.727 6.214 6.803 5.287 5.287 6.354 7.605 10.254 11.040 9.303 10.065 12.355 13.009 5.542 6.721 13.699 3.578 4.278 2.966 9.174 11.171 4.338 5.025 5.094 6.442 4.089 6.152 10.135 10.956 9.843 6.214 4.794 6.053 4.189 5.054
3.444 3.987 4.457 4.643 3.763 3.763 4.683 5.377 7.879 8.249 7.248 7.573 10.325 10.580 3.908 3.908 10.788 2.631 3.284 2.161 6.280 8.814 3.082 3.358 3.613 4.726 2.881 4.553 8.475 8.720 7.577 4.530 3.353 4.434 2.965 3.634
XBridge-HILIC
ZIC-cHILIC
ZIC-HILIC
7.658 9.916 11.221 12.430 8.995 9.984 11.177 13.400 16.800 17.876 15.796 16.871 19.662 20.516 9.972 11.924 20.727 5.391 6.747 4.034 15.336 18.060 7.369 8.825 8.282 10.991 6.131 10.267 17.021 16.967 15.805 10.775 7.842 10.174 6.271 8.027
6.618 8.790 9.077 10.373 6.747 8.398 9.423 11.712 14.142 15.474 12.528 13.848 15.448 16.413 7.742 10.296 16.390 3.881 5.325 2.325 11.651 14.497 5.847 7.483 6.199 8.404 4.646 8.126 16.440 13.798 12.261 8.049 5.861 7.884 4.797 6.256
2.592 2.592 2.636 2.636 2.571 2.571 2.771 2.771 3.106 3.106 2.885 2.885 3.209 3.209 2.567 2.567 3.394 2.376 2.376 2.125 2.809 3.177 2.459 2.459 2.627 2.754 2.534 2.747 3.233 3.363 3.053 2.700 2.569 2.710 2.556 2.641
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
69
Table 4 Retention Time Data for the Test Analytes with Acetonitrile/Water Mobile Phase. 1 and 2 refer to anomers of the same analyte. Sugar Type
Sugar or Sugar Alcohol
Accucore
Accucore Urea
Accucore-Amide
Cosmosil
Kromasil-Diol
Luna-Amino
Nucleodur
Penta-Halo
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
3.219 3.485 3.811 3.811 3.199 3.867 4.032 4.234 6.604 6.604 5.791 5.791 8.321 8.321 3.594 4.010 9.114 2.721 2.985 2.007 5.536 6.987 2.918 2.997 3.899 4.639 3.341 4.600 5.951 8.240 7.044 4.531 3.738 4.535 3.406 3.957
3.753 4.221 5.095 5.400 4.514 4.514 5.152 5.737 8.687 8.988 8.458 8.905 11.886 12.268 4.646 4.646 11.664 3.132 3.624 2.467 7.332 8.976 3.529 3.856 4.089 5.249 3.229 5.030 7.726 9.017 8.406 5.089 3.871 4.996 3.280 4.064
7.197 8.764 11.019 11.566 10.048 10.048 11.264 12.610 18.357 18.625 17.214 17.778 22.303 22.699 9.705 11.378 22.247 4.876 5.872 2.463 15.726 18.393 6.496 7.383 8.286 11.464 5.723 11.205 18.120 19.016 17.482 11.117 7.957 11.073 5.898 8.333
7.218 7.218 10.659 10.659 8.633 8.633 10.819 10.819 17.720 17.720 17.084 17.084 21.239 21.239 9.636 9.636 21.830 5.596 5.596 4.311 15.599 18.047 7.043 7.043 8.236 10.979 5.953 10.301 15.395 18.180 16.892 10.970 7.781 10.696 5.905 8.090
6.156 6.869 9.100 9.472 7.677 7.677 9.150 9.952 15.312 15.646 14.768 15.173 19.153 19.420 8.528 8.528 19.673 5.240 5.473 3.481 13.854 16.269 6.007 6.464 7.654 9.886 5.889 9.455 13.734 16.681 15.470 9.769 7.334 9.466 5.891 7.533
6.450 6.450 8.887 8.887 7.223 7.223 9.122 9.122 13.091 13.091 12.918 12.918 16.378 16.378 8.251 8.251 15.594 4.792 4.792 3.292 11.492 13.392 6.045 6.045 6.953 8.974 5.355 8.412 12.548 13.505 12.661 8.745 6.696 8.355 5.454 6.781
6.011 7.268 9.031 9.821 7.649 7.649 9.052 10.339 14.933 15.687 14.322 15.110 18.605 19.221 8.329 9.049 19.588 4.918 5.293 3.851 13.973 16.097 5.898 6.803 7.498 9.828 5.653 9.301 13.657 16.168 15.141 9.633 7.218 9.284 5.677 7.315
5.544 5.544 8.017 8.427 6.710 6.710 8.043 9.010 14.155 14.513 13.608 14.016 18.028 18.288 7.444 7.444 18.265 4.214 4.214 2.612 12.462 14.964 5.314 5.314 6.485 8.798 4.642 8.476 12.990 15.251 14.199 8.686 6.171 8.461 4.704 6.398
Sugar Type
Sugar or Sugar Alcohol
Sepax Silica
Syncronis
Tosoh-Amide
Waters-Amino
XBridge-BEH-Amide
XBridge-HILIC
ZIC-cHILIC
ZIC-HILIC
Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol Sugar Alcohol
Arabinose-1 Arabinose-2 Dextrose-1 Dextrose-2 Fructose-1 Fructose-2 Galactose-1 Galactose-2 Lactose-1 Lactose-2 Maltose-1 Maltose-2 Maltotriose-1 Maltotriose-2 Mannose-1 Mannose-2 Raffinose Rhaminose-1 Rhaminose-2 Sucralose Sucrose Trehalose Xylose-1 Xylose-2 Arabitol Ducitol Erythritol Inditol Inositol Lactitol Maltitol Mannitol Ribitol Sorbitol Threitol Xylitol
3.954 4.230 4.763 4.763 4.745 4.745 5.023 5.238 7.363 7.363 6.779 6.779 9.324 9.324 4.481 4.899 9.940 3.418 3.710 2.490 6.527 8.016 3.674 3.745 4.892 5.740 4.233 5.665 7.169 9.195 8.069 5.572 4.714 5.545 4.324 4.913
6.179 7.437 9.048 9.821 7.715 7.715 9.091 10.385 15.107 15.794 14.293 15.108 18.432 19.037 8.469 8.469 19.498 5.051 5.405 3.981 13.994 15.991 6.044 6.953 7.579 9.826 5.782 9.310 13.585 16.132 15.032 9.675 7.301 9.376 5.786 7.460
11.150 12.796 15.353 15.893 12.497 14.339 15.650 16.893 22.342 22.580 21.515 22.000 26.091 26.421 14.079 15.671 26.098 8.394 9.681 4.475 20.189 22.514 10.434 11.499 12.570 15.797 9.416 15.560 21.774 22.930 21.715 15.515 12.181 15.420 9.583 12.574
7.999 7.999 9.664 9.664 9.121 9.121 10.758 10.758 17.595 17.595 15.697 15.697 19.254 19.254 9.933 9.933 20.514 5.606 5.606 3.281 14.116 16.414 6.711 6.711 8.348 10.844 6.101 10.428 14.591 18.598 16.264 10.730 7.662 11.049 6.066 8.517
8.807 9.525 12.038 12.566 9.317 10.999 12.223 13.488 19.092 19.365 18.270 18.789 23.107 23.477 10.841 12.264 23.070 5.995 6.896 3.402 16.934 19.393 7.603 8.470 9.315 12.367 6.663 12.059 18.428 19.715 18.465 12.099 8.934 11.924 6.785 9.262
10.029 11.954 14.511 15.543 12.547 12.547 14.356 15.911 20.252 20.905 19.609 20.412 23.477 24.068 13.644 14.355 23.996 7.998 8.438 5.258 19.022 21.111 9.992 11.465 11.772 14.849 8.869 14.046 19.457 20.696 19.816 14.681 11.534 14.126 8.838 11.408
7.950 9.753 11.327 12.406 9.743 9.743 11.440 13.055 16.717 17.545 15.685 16.708 19.059 19.871 10.447 11.625 19.896 5.653 6.389 3.263 15.042 17.149 7.601 9.067 8.902 11.497 6.654 11.021 16.948 17.086 15.941 11.266 8.642 11.007 6.684 8.787
3.221 3.395 3.815 3.815 3.628 3.628 3.892 4.047 5.992 5.992 5.509 5.509 7.717 7.717 3.655 3.877 8.254 2.928 3.069 2.345 5.325 6.506 3.126 3.126 3.794 4.396 3.326 4.298 5.241 7.239 6.412 4.314 3.682 4.263 3.392 3.799
70
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Fig. 1. Bubble plot of dextrose anomers, erythritol, sucrose, trehalose, and xylitol separation based on retention time and the type of HPLC column.
Fig. 2. Application of the column/mobile phase selection algorithm to a mixture of dextrose anomers, erythritol, xylitol, sucrose, and trehalose.
above. The best combination graphically (i.e., largest minimum retention time gap and shortest maximum retention time) is the ® ZIC -HILIC column with the acetone/water mobile phase system. Based on the observations in Figs. 1 and 2, isocratic conditions ® were experimentally applied using the ZIC -HILIC column with the acetone (82.5%)/water and acetonitrile (80%)/water mobile phases for the separation of erythritol, xylitol, dextrose (anomers), sucrose and trehalose. Additionally, three beverages were evaluated which included Hetley’s Organic Black Tea (containing xylitol and sucrose), Halo Trehalose Infusion Water Blackberry/Plum (containing trehalose and sugars) and BAI Antioxidant Infusion Molokai Coconut (containing erythritol). It was determined that acetoni® trile/water mobile phase with the ZIC -HILIC column provided a better overall separation of erythritol, xylitol, dextrose (anomers), sucrose and trehalose (due to the added matrix peaks in the HALO beverage). The resulting chromatogram overlay for the separation is shown in Fig. 3 with the sweeteners found in the different beverages matching the corresponding sugar or sugar alcohol in the standard preparation (analyte concentration = 0.5 mg/mL). The pro® cess described above illustrates how the user of this JMP algorithm can quickly achieve an efficient separation and method for a subset
mixture from the 26 sugars and sugar alcohols shown in Table 1. This example with beverage analysis demonstrates that the HILIC design space can be used as a guide. Complicated sample matrix components may require further development and/or method conditions may need to be modified.
3.2.2. Quantitative analysis case study: controlling a potential impurity in a pharmaceutical excipient The value of the HILIC screen for quantitative analysis of sugars and sugar alcohols is best illustrated with a case study where the screen has been applied to develop a control strategy for a pharmaceutical development project. The application described here involves a lyophilized drug product where the active pharmaceutical ingredient (API) is a small molecule containing a primary amine. The development team recognized that the presence of a primary amine presented the risk for the Maillard reaction occurring between the API and reducing sugars, leading to potential impurity formation [24]. For this reason, non-reducing carbohydrates were chosen as potential excipients to be evaluated during formulation screening studies, with trehalose (Fig. 4) as the leading candidate. However, during initial screening studies with trehalose
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
71
®
Fig. 3. Chromatogram overlays using the ZIC -HILIC column with a mobile phase consisting of 80% acetonitrile/20% water at a flow rate of 1.0 mL/min using ELSD detection. Chromatogram A = blank, chromatogram B = standard mixture, chromatogram C = Hetley’s Organic Black Tea, chromatogram D = Halo Trehalose Infusion Water (Blackberry/Plum), and chromatogram E = BAI Antioxidant Infusion (Molokai Coconut). Peak identification: peak 1 = erythritol, peak 2 = xylitol, peak 3 and 4 = dextrose anomers, peak 5 = sucrose, and peak 6 = trehalose.
Fig. 4. Structure of trehalose, a non-reducing sugar evaluated as a potential excipient.
as the primary excipient, an unexpected impurity was observed to grow during stability studies. The impurity was identified as a glucose adduct formed via the Maillard reaction between glucose (a reducing sugar) and the primary amine of the API. The structure of the glucose adduct was confirmed via NMR, mass spectrometry, and RP-HPLC retention time match with an authentic sample (data not shown). Glucose was not intentionally added to the formulation and it was determined that the source of the glucose was as a residual impurity in the trehalose excipient. Trehalose is a disaccharide with one molecule of trehalose consisting of two molecules of glucose bound at the anomeric carbon. Thus, while glucose is a reducing sugar (and thus able to participate in the Maillaird reaction), trehalose itself is not a reducing sugar. As part of a robust control
strategy, a HILIC-HPLC method was pursued to detect glucose in trehalose as an upstream control to prevent glucose adduct formation in the drug product. Based on the trehalose:API ratio in the formulation, the desire to control the glucose adduct to limits specified by the International Conference on Harmonisation (ICH) [25], and the reaction stoichiometry observed during stability studies, a limit of not more than 0.02% (wt/wt) glucose in trehalose was calculated. The excipient manufacturer controlled reducing sugars with a method and to limits specified in the United States Pharmacopeia; however, tighter control was required based on the desired control described above. Thus, a new method needed to possess the selectivity, sensitivity, linearity, precision, and accuracy necessary to quantitate glucose in trehalose at the 0.02% (wt/wt) level and below. The HILIC screen developed and described in the previous sections, along with previously-published HILIC method development work [26], proved valuable for rapidly identifying chromatographic conditions that would separate glucose from trehalose and serve as the basis for a method for controlling glucose in trehalose. Mannitol was also being considered as a secondary excipient in the ® formulation and as a result, the JMP algorithm described above was executed with glucose, trehalose, and mannitol as the constituents. However, the list of columns was limited to the ones that did not separate the glucose anomers as one peak was desired for glucose (7 of the 16 columns met this criteria). The output of the ® JMP algorithm including these three analytes is summarized in Fig. 5.
72
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
Fig. 5. Application of the column/mobile phase selection algorithm to a mixture of glucose, trehalose, and mannitol.
Fig. 6. Separation in order of elution between glucose (red), mannitol (black), and trehalose (blue) on the AccucoreTM HILIC column with acetonitrile/water mobile phase and the gradient described in the Trehalose Analysis section with CAD detection. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Table 5 Method conditions for quantitation of glucose in trehalose. Parameter
Description
Analytical Column Column Temperature Autosampler Temperature Mobile Phase A Mobile Phase B Flow Rate Injection Volume Run Time Gradient
Thermo Scientific, AccucoreTM HILIC, 4.6 mm x 150 mm, 2.6 m 55 ◦ C Ambient 100% Milli-Q Water 100% Acetonitrile 1.0 mL/minute 10 L 20 min %A Time (minutes) 8 0.00 6.00 8 50 6.10 50 10.50 8 10.60 8 20.00
CAD Parameters CAD Detector Range Filter Nebulizer Heater Nitrogen Pressure
ESA Corona VEO 100 pA 5.0 s 35 ◦ C 35 psi
%B 92 92 50 50 92 92
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74 Table 6 Accuracy and precision of trehalose spiked with glucose. Level
Number of Preparations
Average Recovery
%RSD
0.009% 0.02% 0.03%
N=3 N=6 N=3
113% 109% 112%
6.7 1.2 0.5
Fig. 5 is a plot of maximum retention time (Y-axis) vs. minimum retention time gap (X-axis) for various column and mobile phase systems for the analytes glucose, trehalose, and mannitol. The optimum separation would have a large minimum retention time gap (X-axis) to maximize resolution with a small maximum retention time (Y-axis) to minimize total run time. This translates to the lower right corner of the plot in Fig. 5. This figure illustrates that ® the top 3 columns are the Waters Spherisorb Amino, Sepax Silica, and AccucoreTM HILIC with acetonitrile/water as the mobile phase. Each of these systems exhibits the desirable characteristic of not resolving anomers of glucose (to maximize sensitivity for glucose ® and facilitate single-peak quantitation). The Waters Spherisorb Amino has the best minimum retention time gap (resolution) but suffers from a longer run time. The AccucoreTM HILIC and Sepax Silica had similar minimum retention time gaps and similar total run times. Based on its slightly shorter analysis time, the Accucore was selected and the separation on this system is shown in Fig. 6. Note that this column/mobile phase combination also exhibits the desirable characteristic of glucose eluting before trehalose. Since glucose is expected to be present at lower levels than trehalose, this prevents the need to integrate a small glucose peak on the tail of a large trehalose peak, which is expected to improve sensitivity for glucose and lead to a more precise, robust method. All subsequent work was conducted with the AccucoreTM HILIC column. Based on further method development experiments, certain conditions were modified slightly from the initial screen described above, including modifications to the gradient and the column temperature. These modifications were made to improve resolution and decrease run time. However, the fundamentals of the separation (HILIC stationary phase and mobile phase components) remained the same. Specific details of the method are shown in Table 5.
73
the first analyst was 109% (see Table 6) and the average recovery from the second analyst was 113% (data not shown). The overall RSD of all 12 preparations was 2.4%, demonstrating acceptable intermediate precision. 3.2.2.1.3. Linearity. Linearity was evaluated with a series of six solutions spanning a range from 0.009% to 0.04% (wt/wt). This range spans the approximate limit of detection of 0.009% (wt/wt) to 2x the expected specification limit of 0.02% (wt/wt). The peak area of glucose vs. the concentration of glucose was plotted and a linear regression was performed. The r2 value of 0.999 along with the insignificant Y-intercept (-3.4) when compared to the slope (5510.7) demonstrated acceptable method linearity. 3.2.2.1.4. Limit of quantitation and limit of detection. Due to the intended purpose of the method, which was to control glucose in trehalose to not more than 0.02% (wt/wt), the level of the lowest standard (0.009% wt/wt) was evaluated as a practical limit of quantitation. Based on the data collected during evaluation of precision described above, a signal-to-noise (S/N) ratio was calculated at the 0.009% level. Based on the n = 3 measurements described above, the S/N ratio was 73 ± 3 (average ± standard deviation), well above the conventional definition of LOQ which is typically the concentration which provides an S/N ratio of 10. Assuming that the S/N scales in a linear fashion with analyte concentration, a calculated LOQ (S/N = 10) would be 0.001% (wt/wt) glucose in trehalose and a calculated LOD (S/N = 3) would be 0.0004% (wt/wt) glucose in trehalose. 3.2.2.1.5. Range. Based on results from linearity, accuracy, and precision determinations, the range of the method was 0.009% (wt/wt) to 0.03% (wt/wt). Although the LOQ was determined to be approximately 0.001% (wt/wt) with a calculated LOD of 0.0004% (wt/wt), ICH Q2(R1) indicates that range is derived from linearity, accuracy, and precision studies [27]. In this case, linearity was evaluated over the range of 0.009% to 0.04% (wt/wt) and accuracy was evaluated over the range of 0.009% to 0.03% (wt/wt). Thus, the range of the method is reported as 0.009% (wt/wt) to 0.03% (wt/wt) which is supported by linearity, accuracy, and precision studies. 4. Conclusions ®
3.2.2.1. Method validation. Using the optimized conditions summarized in Table 5, a series of experiments was conducted to validate the method. Method validation parameters were selected from ICH Q2(R1) [27] and the results are summarized below. 3.2.2.1.1. Specificity. Specificity was evaluated by performing a blank injection consisting of only sample diluent, as well as injecting a “clean” sample of trehalose not spiked with glucose. Absence of interference from the diluent blank and the trehalose matrix combined with the precise and accurate recovery of glucose from spiked trehalose samples (see below Section 3.2.2.1.2) demonstrated the specificity of the method. 3.2.2.1.2. Accuracy and precision. Accuracy and precision were evaluated from the same set of experiments. To demonstrate both accuracy and precision, trehalose was spiked with glucose at the following levels (wt/wt): 0.009%, 0.02%, and 0.03%. These levels spanned the expected range of the method, from the approximate LOQ to 1.5x the intended specification limit. Three preparations were made at both the 0.009% and 0.03% levels while six preparations were made at the 0.02% level and each preparation was individually injected. The recovery (accuracy) and%RSD (precision) are summarized in Table 6 and were suitable for the intended purpose of the method considering the extremely low target levels of the analyte. Intermediate precision was evaluated by having a different analyst prepare a second set of six samples spiked at the 0.02% level on a different day using a different column, different HPLC, and different reagent solutions. The average recovery from
An algorithm was developed using JMP statistical software through the evaluation of 26 sugars and sugar alcohols across 16 different HILIC stationary phases using either acetone/water or acetonitrile/water mobile phases with aerosol-based detectors. This approach to method development for mixtures of sugars and/or sugar alcohols provides a simple and efficient selection process of the stationary phase and mobile phase to rapidly obtain an optimum separation. This strategy was demonstrated in two scenarios through the separation of a mixture of five sugars and sugar alcohols found in different beverages and for the separation of trehalose, glucose, and mannitol in a pharmaceutical formulation. Furthermore, method validation parameters for the trace analysis of glucose using CAD demonstrated excellent linearity, accuracy, precision, and specificity. Acknowledgements The authors wish to acknowledge Matt W. Switzer, David P. Myers, Brian W. Pack, and Mark Strege for insightful technical contributions related to HILIC, chromatography, method validation, and impurity control strategy. References [1] X. Wang, Y. Chen, Determination of carbohydrates as their p-sulfophenylhdrazones by capillary zone electrophoresis, Carbohydr. Res. 332 (2001) 191–196.
74
E.M. Hetrick et al. / J. Chromatogr. A 1489 (2017) 65–74
[2] Y. Cao, Y. Wang, X. Chen, J. Ye, Study on sugar profile of rice during ageing by capillary electrophoresis with electrochemical detection, Food Chem. 86 (1) (2004) 131–136. [3] P.M. Medeiros, B.R.T. Simoneit, Analysis of sugars in environmental samples by gas chromatography-mass spectrometry, J. Chromatogr. A 1141 (2007) 271–278. [4] Zs. F. Katona, P. Sass, I. Molnar-Perl, Simultaneous determination of sugars, sugar alcohols, acids and amino acids in apricots by gas chromatography-mass spectrometry, J. Chromatogr. A 847 (1+2) (1999) 91–102. [5] M.A. Adams, Z. Chan, P. Landman, T.D. Colmer, Simulataneous determination by capillary gas chromatography of organic acids, sugars, and sugar alcohols in plant tissue extracts as their trimethylsilyl derivatives, Anal. Biochem. 266 (1) (1999) 77–84. [6] R. Andersen, A. Sorensen, Separation and determination of alditols and sugars by high-pH anion-exchange chromatography with pulsed amperometric detection, J. Chromatogr. A 897 (2000) 195–204. [7] Y. Ding, H. Yu, S. Mou, Direct determination of free amino acids and sugars in green tea by anion-exchange chromatography with integrated pulsed amperometric detection, J. Chromatogr. A 982 (2) (2002) 237–244. [8] S. Ouchemoukh, P. Schweitzer, M. Bachir Bey, H. Djoudad-Kadji, H. Louaileche, HPLC sugar profiles of Algerian honeys, Food Chem. 121 (2) (2010) 561–568. [9] M. Davis, A rapid modified method of compositional carbohydrate analysis of lignocellulosics by high-pH anion-exchange chromatography with pulsed amperometric detection (HPAEC/PAD), J. Wood Chem. Technol. 18 (2) (1998) 235–252. [10] R.I. Cataldi Tommaso, G. Margiotta, L. Lasi, B. Di Chio, C. Xiloyannis, S. Bufo Sabino, Determination of sugar compounds in olive plant extracts by anion-exchange chromatography with pulsed amperometric detection, Anal. Chem. 72 (16) (2000) 3902–3907. [11] A. Caseorp, L. Marr, M. Claeys, A. Kasper-Giebl, H. Puxbaum, A. Pio Casimiro, Determination of saccharides in atmospheric aerosol using anion-exchange high-performance liquid chromatography and pulsed-amperometric detection, J. Chromatogr. A 1171 (1-2) (2007) 37–45. [12] K. Kaiser, R. Benner, Determination of amino sugars in environmental samples with high salt content by high-performance anion-exchange chromatography and pulsed amperometric detection, Anal. Chem. 72 (11) (2000) 2566–2572. [13] W. Xu, L. Liang, M. Zhu, Determination of sugars in molasses by HPLC following solid-phase extraction, Int. J. Food Prop. 18 (3) (2015) 547–557. [14] E. Moro, R. Majocchi, C. Ballabio, S. Molfino, P. Restani, A rapid and simple method for simultaneous determination of glycerol, fructose, and glucose in wine, Am. J. Enol. Vitic. 58 (2) (2007) 279–282.
[15] L.A.Th. Verhaar, B.F.M. Kuster, Contribuion to the elucidation of the mechanism of sugar retention on amine-modified silica in liquid chromatography, J. Chromatogr. 234 (1982) 57–64. [16] R. Orth, H. Englehardt, Separation of sugars on chemically modified silica gel, Chromatographia 15 (1982) 91–96. [17] A.J. Alpert, Hydrophilic-interaction chromatography for the separation of peptides, nucleic acids and other polar compounds, J. Chromatogr. A 499 (1990) 177–196. [18] L.B. Allen, J.A. Koropchak, Condensation nucleation light scattering: a new approach to development of high-sensitivity universal detectors for separations, Anal. Chem. 65 (1993) 841–844. [19] R.W. Dixon, D.S. Peterson, Development and testing of a detection method for liquid chromatography based on aerosol charging, Anal. Chem. 74 (2002) 2930–2937. [20] A. Stolyhwo, H. Colin, G. Guiochon, Use of light scattering as a detector principle in liquid chromatography, J. Chromatogr. A 265 (1983) 1–18. [21] L.-E. Magnusson, D.S. Risley, J.A. Koropchak, Aerosol-based detector for liquid chromatography, J. Chromatogr. A 1421 (2015) 68–81. [22] C. Ma, Z. Sun, C. Chen, L. Zhang, S. Zhu, Simultaneous separation and determination of fructose sorbitol, glucose and sucrose in fruits by HPLC-ELSD, Food Chem. 145 (2014) 784–788. [23] M. Grembecka, A. Lebiedzinska, P. Szefer, Simultaneous separation and determination of erythritol, xylitol, sorbitol, mannitol, maltitol, fructose, glucose, sucrose and maltose in food products by high performance liquid chromatography coupled to charged aerosol detector, Microchem. J. 117 (2014) 77–82. [24] S.W. Baertschi, K.M. Alsante, D. Santafianos, Stress testing: the chemistry of drug degradation, in: S.W. Baertschi, K.M. Alsante, R.A. Reed (Eds.), Pharmaceutical Stress Testing: Predicting Drug Degradation, second ed., Informa Healthcare, London, 2011, pp. 49–141. [25] Impurities in New Drug Products, Q3B(R2), International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2006, www.ich.org. [26] B.A. Olsen, D.S. Risley, V.S. Sharp, B.W. Pack, M.L. Lytle, Pharmaceutical applications of hydrophilic interaction chromatography, in: B.A. Olsen, B.W. Pack (Eds.), Hydrophilic Interaction Chromatography: A Guide for Practitioners, Wiley, Hoboken, NJ, 2013, pp. 111–168. [27] Validation of Analytical Procedures: Text and Methodology, Q2(R1), International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, 2005, www.ich.org.