Journal of Chromatographic Science, Vol. 40, March 2002
`
`HPLC and LC–MS Studies of the Transesterification
`Reaction of Methylparaben with Twelve 3- to 6-Carbon
`Sugar Alcohols and Propylene Glycol and the
`Isomerization of the Reaction Products by
`Acyl Migration
`
`Minhui Ma*, Antonio DiLollo, Robert Mercuri, Tony Lee, Mark Bundang, and Elizabeth Kwong†
`Pharmaceutical Research & Development, Merck Frosst Canada & Co., 16711 Trans Canada Highway, Kirkland,
`Quebec, H9H 3L1, Canada
`
`Abstract
`
`Sugar alcohols and parabens are commonly used ingredients in
`oral suspension formulations. However, their possible
`incompatibility because of transesterification reaction is a concern
`during formulation development. In order to gain more knowledge
`about the reaction, a high-performance liquid chromatographic
`(HPLC) method is developed to separate the transesterification
`reaction products of methylparaben preservative with twelve 3- to
`6-carbon sugar alcohols and propylene glycol. It is found that the
`number of peaks separated or partially separated correlate well
`with the number of distinct hydroxyl groups present in the sugar
`alcohol molecules. This means that all the hydroxyl groups in a
`sugar alcohol molecule can react with methylparaben to form
`transesterification reaction products. These products are positional
`isomers that have identical UV spectra with a maximum at
`255 nm and the same m/z ratio for molecular ions by liquid
`chromatography–mass spectrometry. When isolated individually,
`they can isomerize (interconvert) under suitable conditions to
`form other positional isomers by intramolecular acyl migration.
`The acyl migration pathway for each of the isolated positional
`isomers from the transesterification reaction of methylparaben
`with sorbitol, ribitol, and xylitol is followed by HPLC. Based on
`the information, a tentative assignment of the six isomer peaks
`generated from the transesterification reaction between
`methylparaben and sorbitol is proposed.
`
`Introduction
`
`Sugar alcohols (also known as polyols) such as sorbitol and
`mannitol are among the most widely used excipients in the
`pharmaceutical industry. Parabens (a group of alkyl esters of
`p-hydroxybenzoic acid) are commonly used antimicrobial
`agents in pharmaceuticals, food, and cosmetics. During the sta-
`
`* Present address: Analytical Sciences, Amgen Inc., One Amgen Center Drive, Thousand Oaks,
`CA 91320.
`† Corresponding author: email elizabeth_kwong@merck.com.
`
`bility study of an experimental batch of an oral suspension
`formulation that contained sorbitol and paraben preservatives
`(methylparaben and propylparaben), extra peaks at a level of
`~1% of the area of the main drug peak were found near the sol-
`vent front of the high-performance liquid-chromatographic
`(HPLC) chromatograms for samples stored at 30°C for one
`year. In a subsequent investigation, it was demonstrated that
`similar peaks could be generated in a mixture of sorbitol and
`the parabens and that the formation of the extra peaks was
`affected most significantly by pH and temperature. In a com-
`mercially available antacid oral suspension that contains sor-
`bitol and the parabens at pH ~8, similar peaks were found at a
`much higher level.
`It has been proposed by others (1–5) that the extra peaks are
`the transesterification reaction products between paraben and
`sugar alcohols. Because of the concern about the adverse effect
`of the reaction on the effectiveness of the paraben preservatives
`in pharmaceutical formulation, the kinetics of the transester-
`ification reaction has been studied for parabens with sorbitol,
`xylitol, erythritol, and glycerol (2). The identities of the reac-
`tion products have been investigated for methylparaben and
`xylitol by liquid chromatography (LC)–mass spectrometry
`(MS), gas chromatography (GC)–MS, and nuclear magnetic
`resonance (NMR) (3). However, the attempt to identify the
`reaction products of methylparaben with sorbitol by MS and
`NMR after isolation by preparative LC was much less a success
`because of the fact that only three peaks (one major and two
`minor) were separated and the isolated products were unstable
`(2). In theory, six reaction products are possible for sorbitol,
`but these reaction products are positional isomers and difficult
`to separate. Others also reported only three (3–5) or four (1)
`peaks. In our investigation, up to six peaks were separated
`depending on the chromatographic conditions. The difficulty
`in separating all possible reaction products may be contributed
`to the incorrect conclusion made by Hensel et al. (3) that
`three isomers (or three pairs of diastereomers) were expected
`
`170
`
`Reproduction (photocopying) of editorial content of this journal is prohibited without publisher’s permission.
`
`Flat Line Capital Exhibit 1007
`Page 1
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 1
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`from the reaction between methylparaben and sorbitol. In
`order to better understand the reaction mechanism between
`parabens and sugar alcohols, the separation of all possible
`reaction products is necessary.
`In this study, LC and LC–MS methods were developed to sep-
`arate (or partially separate) and identify (as a group of iso-
`mers) all possible reaction products of methylparaben with
`twelve 3- to 6-carbon sugar alcohols and propylene glycol.
`During the study, it was noticed that the positional isomers,
`when isolated individually by fraction collection, could iso-
`merize (convert) reversibly from one to another. A similar iso-
`merization reaction (which involves intramolecular acyl
`migration on a glucuronic acid ring) has been studied recently
`by Nicholson and colleagues using LC–NMR (6–11) and also by
`others (12–14) for a number of glucuronides. However, an acyl
`migration reaction on linear sugar alcohol molecules have not
`been reported. In this study, we will show that useful infor-
`mation can be obtained by an HPLC study on the acyl migra-
`tion reactions of the transesterification reaction products of
`methylparaben with sugar alcohols.
`
`Experimental
`
`Reagents
`All chemicals used in this study were of reagent grade and
`used without further purification. D-Sorbitol (99%), D-man-
`nitol (+99%), L-iditol (99%), D-threitol (99%), glycerol
`(+99.5%), and p-hydroxyl benzoic acid (99%) were obtained
`from Aldrich (Milwaukee, WI); D-xylitol (99%), D-ribitol (99%),
`D-arabitol (99%), dulcitol (99%), erythritol (99%), 1,2-propane-
`diol (propylene glycol) (99.9%), and methylparaben (99%)
`were obtained from Sigma (St. Louis, MO); and D-Talitol
`(+98%) was obtained from TCI America (Portland, OR). Allitol
`was specially ordered from Omicron Biochemicals (South
`Bend, IN). Sodium methyl paraben was provided by Merck
`Research Laboratories (West Point, PA). Tetrahydrofuran (THF)
`(99.9%) and acetonitrile (HPLC grade) were supplied by EM
`Science (Gibbstown, NJ). Trifluoroacetic acid (TFA) was pur-
`chased from Pierce (Rockford, IL). The deionized water used for
`the preparation of all the solutions was purified with a Milli-Q
`water purification system (Millipore, Bedford, PA).
`
`Generation of transesterification reaction products
`The transesterification reaction products of methylparaben
`with the twelve 3- to 6-carbon sugar alcohols were generated as
`follows. Approximately 50 mg of sodium methylparaben was
`added to 5 mL of a 60-mg/mL sugar alcohol aqueous solution.
`After the pH was adjusted to ~11 with a few drops of 1M sodium
`hydroxide, the solution was heated in a 90°C oven for 2 h. The
`stress conditions were chosen so that the reaction could reach
`equilibrium in a short period of time. Similar reaction conditions
`have been employed by others (3). The sample solution was
`then cooled to room temperature and filtered through a 0.45-µm
`polytetrafluoroethylene syringe filter into an HPLC vial.
`The transesterification reaction of methylparaben with propy-
`lene glycol was generated in the same way except that a much
`higher concentration of propylene glycol (5.7M) was used and
`
`the solution was kept in the 90°C oven much longer (48 h).
`
`LC
`The HPLC system used in this study was an HP 1090 LC
`(Hewlett-Packard, Palo Alto, CA) equipped with an autosampler
`and a diode-array detector (DAD). HP Chemstation software
`was used to acquire and analyze the chromatographic data. In
`order to achieve the separation of the transesterification reac-
`tion products (positional isomers), two Hewlett Packard Zorbax
`SB-CN (3.5 µm) columns were coupled in series (4.6 × 75 mm
`and 4.6 × 150 mm). Column temperature was kept at 40°C. A
`mixture of 85:15 (v/v) THF–acetonitrile was used as mobile
`phase B and 0.1% (v/v) TFA in water as mobile phase A. The
`injection volume was 20 µL. Chromatograms were acquired at
`255 nm with a bandwidth of 4 nm. Three different gradient
`programs (Table I) were employed in this study. Gradient I
`was used for the separation of the transesterification reaction
`products of methylparaben with the twelve 3- to 6-carbon
`sugar alcohols and for the study on the isomerization of the six
`reaction products of sorbitol. Gradient II was used for the sep-
`aration of the reaction products between methylparaben and
`propylene glycol and also for all the LC–MS analyses. Gradient
`III was used for the study on the isomerization of the reaction
`products of xylitol and ribitol. The same flow rate (0.60
`mL/min) was used for all three gradients.
`
`Study of the isomerization of the
`transesterification products
`The isomerization of the transesterification reaction prod-
`ucts of methylparaben with sorbitol, xylitol, and ribitol were
`studied. The positional isomers that were generated according to
`the previously mentioned procedure were separated by HPLC
`using gradient I for the reaction products with sorbitol and gra-
`dient III for the reaction products with xylitol and ribitol. The
`fraction of each individual positional isomer was collected man-
`ually into an ice-chilled 2-mL HPLC vial from the outlet of the
`DAD. Among the six products generated from the reaction of
`methylparaben with sorbitol, peaks 1, 2, 3, and 4 were well-sep-
`arated, but peaks 5 and 6 overlapped significantly. Therefore,
`only the ascending part of peak 5 and the descending part of peak
`6 were collected to minimize cross contamination. Before the
`
`Table I. Gradients Used to Separate the
`Transesterification Reaction Products and for the
`Study on the Isomerization of the Reaction Products
`
`Gradient I
`Time (min) %B
`
`Gradient II
`Time (min)
`%B
`
`Gradient III
`Time (min) %B
`
`0
`5
`5.5
`6.5
`12
`16
`18
`25
`34
`
`0
`0
`5
`11
`11
`70
`70
`0
`0
`
`0
`14
`16
`36
`38
`40
`42
`51
`
`4
`4
`20
`20
`30
`30
`4
`4
`
`0
`17
`22
`27
`30
`39
`
`4
`4
`70
`70
`4
`4
`
`171
`
`Flat Line Capital Exhibit 1007
`Page 2
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 2
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`range of 50 amu. The actual mass ranges were chosen such that
`the molecular ions of the transesterification reaction products
`would fall approximately in the center of the 50-amu mass scan
`range.
`
`Results and Discussion
`
`Separation of transesterification reaction products
`The transesterification reaction of methylparaben with sugar
`alcohols results in the replacement of the methoxyl group of
`methylparaben by one of the alcohol groups of sugar alcohol,
`as shown in Figure 1 for sorbitol. In order to understand the
`transesterification reaction of sorbitol and parabens, other
`related sugar alcohols were also included in this study. The
`molecular structures of the twelve 3- to 6-carbon sugar alco-
`hols and propylene glycol with their plane of symmetry are
`shown in Figure 2. The number of transesterification reac-
`tion products that can be formed is in theory dependent on the
`number of unique hydroxyl groups present in the sugar alcohol
`
`Figure 1. Transesterification reaction of methylparaben with sugar alcohol
`showing one of the six sorbitol monoesters of p-hydroxybenzoic acid.
`
`fraction collection of peaks 1, 2, and 5, the tip of the capillary
`leading out of the detector was rinsed with water to avoid car-
`ryover. Because the same analytical columns were used for frac-
`tion collection, the amount of sample that could be loaded was
`limited. Approximately 10 collections were required for each
`peak in order to obtain a sufficient amount of sample for the
`study. In order to slow down the isomerization of the isolated
`positional isomers at the collection stage, the vials were kept on
`a bed of ice between collections. When the fraction collection was
`done, the mobile phase in the collected samples was evaporated
`overnight to dryness with a SpeedVac Plus Concentrator (Model
`SC210A) (Holbrook, NY). The dry samples were then kept tem-
`porarily in a –20°C freezer to minimize isomerization.
`In order to study the isomerization (acyl migration) pathway
`for each isolated isomer (peak), the dry product was redissolved
`in an appropriate amount of water (150 to 450 µL) to obtain a
`suitable concentration for the study. Approximately 150 µL of
`the sample was then transferred to a 200-µL HPLC vial insert,
`and 20 µL of the sample solution was immediately injected
`onto the HPLC column to obtain an initial chromatogram
`before any significant isomerization. A small amount (0.5–
`2 µL) of 0.01–0.05M NaOH was then pipetted into the sample
`solution remaining in the HPLC vial insert as it was gently
`vortexed. The sample was again immediately injected onto the
`HPLC column. After the run was finished (approximately 45
`min after the NaOH addition), a second HPLC injection was
`made to determine the rate of isomerization and the time inter-
`vals for future HPLC injections. The rate of isomerization was
`visually estimated by the appearance of other isomer peaks and
`their relative peak sizes as compared with those at equilibrium.
`If the rate of isomerization was too slow to follow the reaction
`in a realistic time frame, more NaOH was added to
`the remaining sample solution. This process was
`repeated until a suitable rate of isomerization was
`obtained.
`
`LC–MS analyses
`The mass spectra of the transesterification reaction
`products of methylparaben with the twelve 3- to 6-
`carbon sugar alcohols and propylene glycol were
`acquired using a Finnigan (San Jose, CA) LCQ Deca
`ion-trap MS with an ESI source operated in negative
`ion mode. The HPLC conditions were mainly the
`same as those described in the LC section with some
`modifications. Because TFA was not suitable for
`LC–MS as a result of the significant suppression of
`ionization, 0.1% formic acid in water was used as
`mobile phase A. The mobile phase flow (0.60 mL/min)
`was split using a tee connected after the columns, and
`approximately 50 µL/min of the flow was introduced
`into the ESI source. The separation was carried out
`with gradient II for all transesterification reaction
`products. The operating parameters of the ESI source
`were as follows: the sheath gas flow rate was 60, the
`auxiliary gas flow rate was 5, the spray voltage 4.5 kV,
`the capillary temperature 350°C, the capillary voltage
`–41 V, and the tube lens offset –25 V. Mass spectra
`were acquired in full-scan MS mode with a mass scan
`
`172
`
`Figure 2. Molecular structures of the twelve 3- to 6-carbon sugar alcohols and propylene
`glycol used in this study shown in Fischer projection. The dot and dashed lines represent the
`point and plane of symmetry, respectively.
`
`Flat Line Capital Exhibit 1007
`Page 3
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 3
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`molecule and the symmetry of the molecule. Allitol, for
`example, possesses a plane of symmetry between the two
`middle carbon atoms, and therefore the expected number of
`transesterification reaction products is three. Sorbitol, talitol,
`and arabitol, however, have no symmetry element. The number
`of possible transesterification reaction products should be in
`theory equal to the number of hydroxyl groups present in the
`molecule (i.e., six for sorbitol and talitol and five for arabitol).
`In the previous studies on the transesterification reaction prod-
`ucts of methylparaben with sorbitol and other selected sugar
`alcohols (1–5), C18 columns were used to separate the reaction
`products. However, because the transesterification reaction
`products are hydrophilic compounds, their separation using a
`C18 column is not ideal. In the best case (1), only four out of
`six possible peaks of the reaction products of methylparaben
`with sorbitol were separated. In order to increase the retention
`and improve the separation of the reaction products, cyano
`columns were used in this study.
`During the method development, a Zorbax SB-CN (5 µm, 4.6
`× 250 mm) column was first employed and 0.1% TFA in water
`was used as mobile phase A and pure acetonitrile as mobile
`phase B. With this system, five peaks (including two majors and
`three minors) could be separated or partially separated. The
`two major peaks corresponded well with the two primary reac-
`tion products of methylparaben with the two primary hydroxyl
`groups present in sorbitol because they were expected to be
`much more reactive than the secondary hydroxyl groups (15).
`In order to find out if all six hydroxyl groups in sorbitol reacted
`with methylparaben, a further increase in resolution was
`
`required. This was done by coupling in series two Zorbax SB-
`CN columns (4.6 × 75 mm and 4.6 × 150 mm) with a 3.5-µm
`particle size. With this setup, one of the three minor peaks
`broadened significantly, which indicated the possibility of two
`overlapping peaks. By adjusting the gradient program, that
`minor peak was resolved into two peaks while the resolution of
`the two major peaks was completely lost. This experiment
`showed that all six hydroxyl groups of sorbitol did react with
`methylparaben to form six transesterification products.
`In order to improve the selectivity, mobile phase B (pure ace-
`tonitrile) was modified by a few different solvents. The best
`resolution was obtained with THF as the modifier. Figures 3
`and 4 show the separation or partial separation of all possible
`transesterification reaction products of methylparaben with
`the twelve 3- to 6-carbon sugar alcohols using 85:15
`THF–acetonitrile as mobile phase B and gradient I. It can be
`seen from these figures that the number of peaks recognizable
`from the chromatograms correlated well with the expected
`number of transesterification reaction products. Because the
`HPLC conditions were optimized for the separation of the
`reaction products with sorbitol (Figure 3A), they were not
`ideal for the separation of arabitol and talitol reaction products.
`In the case of arabitol (Figure 4C), five peaks were indeed
`found in the chromatogram, although two minor peaks were
`severely overlapped. For talitol (Figure 3E), two broad minor
`peaks instead of four were seen in the chromatogram, indi-
`cating a possible coelution of the two pairs of minor peaks. It
`is interesting to note that all the primary reaction products
`eluted later than the secondary products. This could be related
`
`A
`
`B
`
`C
`
`D
`
`E
`
`F
`
`Figure 3. Chromatograms of the reaction products of methylparaben with 6-carbon sugar alcohols: (A) sorbitol, (B) mannitol, (C) iditol, (D) dulcitol, (E) talitol,
`and (F) allitol.
`
`173
`
`Flat Line Capital Exhibit 1007
`Page 4
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 4
`
`

`

`to the molecular shape rather than the hydrophobicity of the
`reaction products. When methylparaben reacts with a primary
`hydroxyl group, the reaction product has a more linear struc-
`ture, which may allow for a stronger interaction with the sta-
`tionary phase. It was also observed from the chromatograms
`that the smaller the sugar alcohol molecule then the longer the
`retention time of the reaction product because the products
`become less hydrophilic.
`It has been reported that there is no significant reaction
`between methylparaben and propylene glycol at 90°C and pH
`7.3 (2). We found that propylene glycol was indeed much less
`reactive than sugar alcohols under the same reaction condi-
`tions as the twelve 3- to 6-carbon sugar alcohols (i.e., pH ~11
`and 90°C for 2 h). In order to clearly show the reaction between
`propylene glycol and methylparaben, the concentration of
`propylene glycol was increased to 25 times higher than that of
`glycerol and the reaction time at 90°C was extended to 48 h.
`The chromatogram of the two expected reaction products is
`shown in Figure 5. Because propylene glycol is less hydrophilic
`than glycerol, its reaction products were found to elute after
`p-hydroxybenzoic acid. In order to obtain a good separation of
`the reaction products from p-hydroxybenzoic acid and methyl-
`paraben, gradient II was used.
`Theoretically, sugar alcohol monoesters of p-hydroxyben-
`zoic acid may react further with methylparaben to form sugar
`alcohol diesters of p-hydroxybenzoic acid because sugar alco-
`hols have more than one hydroxyl group. The mole ratios of
`methylparaben to the sugar alcohols used in this study ranged
`
`Journal of Chromatographic Science, Vol. 40, March 2002
`
`from 3:1 to 1:99. If diesters were formed, they would have
`longer retention times than monoesters because of the pres-
`ence of two acyl groups. Under the current experimental con-
`ditions, however, no such peaks were found in the
`chromatograms of the transesterification reaction products.
`
`LC–MS study on the transesterification reaction products
`In order to ascertain that the peaks separated or partially sep-
`arated in the chromatograms shown in Figures 3–5 are indeed
`the transesterification reaction products, the UV and mass
`spectra of these peaks were acquired by LC–DAD and LC–MS
`using mobile phase gradient II. It was found that the UV spectra
`of all the peaks were almost identical to those of methylparaben
`and p-hydroxybenzoic acid with a maximum absorption at 255
`nm. This indicated the structural similarity of these reaction
`products to methylparaben and p-hydroxybenzoic acid. The
`more definitive information was obtained from the LC–MS
`study.
`The initial investigation using LC–MS with electrospray was
`done in the positive ion mode. However, the mass detector
`sensitivity could be significantly increased by using electro-
`spray in the negative ion mode even though the mobile phase
`(0.1% TFA) was acidic. It indicated that the p-hydroxyl group
`of the paraben moiety was easily deprotonated at the electro-
`spray ion source under the current experimental conditions to
`form the negative molecular ions, (M–H)–. Therefore, electro-
`spray in the negative ion mode was employed in this study.
`Figure 6 shows some typical total ion chromatograms (Fig-
`
`A
`
`B
`
`C
`
`D
`
`E
`
`F
`
`Figure 4. Chromatograms of the reaction products of methylparaben with 5-, 4-, and 3-carbon sugar alcohols: (A) ribitol, (B) xylitol, (C) arabitol, (D) erythritol,
`(E) threitol, and (F) glycerol.
`
`174
`
`Flat Line Capital Exhibit 1007
`Page 5
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 5
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`ures 6A–6C) and mass spectra (Figures 6D–6F) of the transes-
`terification reaction products. Because the mass scan range was
`limited to 50 amu around the expected molecular ions of the
`reaction products, p-hydroxybenzoic acid and methylparaben
`were not detected and therefore excluded from the mass chro-
`matograms. All of the peaks shown on the total ion chro-
`matograms were expected to be the reaction products. Because
`they were acquired using gradient II, the total ion chro-
`matograms were slightly different from those shown in Figures
`
`Figure 5. Chromatogram of the reaction products of methylparaben with
`propylene glycol.
`
`3–5 in terms of peak retention and resolution. For example, the
`minor peaks 3 and 4 shown in Figure 3A for the reaction prod-
`ucts of sorbitol were overlapped on the mass chromatogram in
`Figure 6C. It was found that all the peaks on the same mass
`chromatogram had identical mass spectra with the same m/z
`ratio for their molecular ions. The m/z ratio matched exactly
`what was expected based on the molecular weight of sugar
`alcohol monoesters of p-hydroxybenzoic acid. For example,
`the mass spectra of the five peaks shown in Figure 6C gave the
`same m/z ratio of 301, (M–H)–, as shown in Figure 6F, and the
`molecular weight of the sorbitol monoesters of p-hydroxyben-
`zoic acid was 302. The LC–MS data confirmed that all the
`“unknown” peaks that were separated or partially separated in
`Figures 3–5 were indeed the transesterification reaction prod-
`ucts of methylparaben with sugar alcohols and propylene
`glycol.
`
`Isomerization of the transesterification reaction products
`During this study, effort was made to isolate individually the
`six positional isomers generated from the reaction of methyl-
`paraben with sorbitol by fraction collection for structural iden-
`tification using other spectroscopic techniques such as NMR.
`However, it was found that the isolated positional isomers could
`
`D
`
`E
`
`F
`
`Relativeabundance
`
`Relativeabundance
`
`Relativeabundance
`
`A
`
`B
`
`C
`
`Relativeabundance
`
`Relativeabundance
`
`Relativeabundance
`
`Figure 6. Total ion chromatograms of the transesterification reaction products of methylparaben with (A) propylene glycol, (B) xylitol, and (C) sorbitol. Their typical
`mass spectra of the isomer peaks are shown in D, E, and F, respectively.
`
`175
`
`Flat Line Capital Exhibit 1007
`Page 6
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 6
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`isomerize (convert) reversibly from one to another via intramol-
`ecular acyl migration, which under suitable conditions could be
`followed using the developed HPLC–UV method. For simplicity,
`a peak and positional isomer were used interchangeably in the
`following discussion and figures.
`After the eluent in the collected fractions was evaporated to
`dryness using a vacuum concentrator, the dry isomer samples
`were redissolved in a small amount of water and reinjected
`immediately onto the HPLC column. In water, the rate of iso-
`
`Figure 7. Summary of the acyl migration pathways of the three positional
`isomers generated from the reaction between methylparaben, ribitol, and
`xylitol.
`
`merization of the reaction products was slow, but it could be
`
`A
`
`B
`
`C
`
`D
`
`Figure 8. Typical chromatograms of the isomerization process of the isolated transesterification reac-
`tion products of methylparaben with sorbitol: (A) isolated peak 1 initially, (B) 1.2 h after the first addi-
`tion of NaOH (approximately 1.5
`10–4M in sample), (C) 17 h after the first addition and 3 min after
`×
`the second addition of NaOH (totaling approximately 4
`10–4M in sample), (D) 1.2 h after the second
`×
`addition of NaOH, (E) isolated peak 4 initially, (F) 2 h after the addition of NaOH (approximately 4
`10–4M in sample), (G) 13 h after NaOH addition, and (H) 66 h after NaOH addition.
`
`×
`
`176
`
`drastically increased by the increase in solution pH because the
`reaction was catalyzed by base. When the pH of the solution
`containing one isolated isomer was adjusted to approximately
`12 using diluted NaOH, the rate of isomerization was so fast
`that the chromatogram obtained at 15 min after the NaOH
`addition showed the presence of all six isomer peaks with the
`distinct relative peak sizes (shown in Figure 3A). No further
`change in relative peak sizes was found at a later time point.
`This means that isomerization from one isomer to the other
`five isomers reached equilibrium.
`In order to understand the acyl migration pathway, the
`isomerization (interconversion) process of the isolated posi-
`tional isomers from the reaction of methylparaben with ribitol
`and xylitol was first followed by HPLC because the isomer
`peaks could be readily assigned based on their relative peak
`size. For example, there were three isomer peaks as shown in
`Figures 4A and 4B that correspond with the three distinct
`hydroxyl groups present in ribitol and xylitol, respectively.
`Based on the understanding of the relative reactivity of the
`hydroxyl groups, the largest peak of the chromatogram could
`be easily assigned to the isomer with the acyl group linked to
`the primary hydroxyl group, the second largest peak could be
`assigned to the isomer with the acyl group linked to one of the
`two equivalent secondary hydroxyl groups, and the smallest
`peak to the reaction product of the central
`hydroxyl group. The two secondary isomer
`peaks should have a peak-area ratio of
`approximately 2:1 because all of the sec-
`ondary hydroxyl groups were expected to
`have similar reactivity. Indeed, the peak-
`area ratios were found to be 2.3:1 and 1.9:1
`for ribitol and xylitol, respectively. The iso-
`merization results show that at each step
`the acyl group migrated to an adjacent
`hydroxyl group via a less sterically hindered
`five-member intermediate as shown in
`Figure 7, which is similar to acyl migration
`on a glucuronic acid ring (6–14).
`Figure 8 shows typical chromatograms
`illustrating the isomerization (acyl migra-
`tion) process of isolated peak 1 and peak 4. In
`Figures 8A–8D, isolated peak 1 first converted
`to peaks 2 and 3, which in turn converted to
`peaks 4, 5, or 6. In Figures 8E–8H, isolated
`peak 4 initially converted primarily to peak 6
`and then to other peaks. This suggests that
`isomer 1 is on the adjacent hydroxyl group
`of isomers 2 and 3 and that isomer 4 is on
`the adjacent hydroxyl group of isomer 6. By
`following the isomerization process of the
`other isolated peaks, the complete acyl
`migration pathway can be obtained. Figure
`9 shows two tentative assignments of the
`peaks based on these observations. Two
`assignments were given because the two
`primary monoester isomers cannot be
`specifically assigned based on this study.
`
`E
`
`F
`
`G
`
`H
`
`Flat Line Capital Exhibit 1007
`Page 7
`
`KVK-Tech, Flat Line Capital Exhibit 1007
`Page 7
`
`

`

`Journal of Chromatographic Science, Vol. 40, March 2002
`
`Conclusion
`
`This study has demonstrated that all the hydroxyl groups in
`a sugar alcohol molecule can react with methylparaben to
`form transesterification reaction products. The number of pos-
`sible reaction products is dependent on the number of hydroxyl
`groups present in the sugar alcohol molecule and the sym-
`metry characteristics of the molecule. These reaction products
`are positional isomers and if isolated individually they can iso-
`merize (convert) reversibly to other isomers. Based on the
`information obtained from acyl migration reaction, two pos-
`sible assignments could be made for the six isomer peaks from
`the transesterification reaction of methylparaben with sor-
`bitol. A definitive assignment of the six isomer peaks may be
`achieved by an LC–NMR study in the future. Derivatization of
`an isolated individual positional isomer may also be useful for
`this purpose.
`
`Figure 9. The tentative assignments of the six sorbitol monoesters of
`p-hydroxybenzoic acid to the six isomer peaks.
`
`Acknowledgments
`
`We thank Dr. Laird Trimble for helpful discussions and Ms.
`S. Spagnoli and Dr. J. Visentini for their initial LC–MS study.
`We also would like to thank Dr. J. Grove and Ms. M.-P. Quint
`of Chibret, Merck Sharp and Dohme, for leading us to this
`project.
`
`References
`
`1. M.J. Thompson, A.F. Fell, B.J. Clark, and M.L. Robinson. LC studies
`on the potential interaction of paraben preservatives with sorbitol
`and glycerol. J. Pharm. Biomed. Anal. 11: 233–40 (1993).
`2. B. Bunesson and K. Gustavii. Stability of parabens in the presence
`
`polyols. Acta Pharm. Suec. 23: 151–62 (1986).
`3. A. Hensel, S. Leisenheimer, A. Muller, E. Busker, E. Wolf-Heuss,
`and J. Engel. Transesterification reactions of parabens (alkyl 4-
`hydroxybenzoates) with polyols in aqueous solutions. J. Pharm. Sci.
`84: 115–18 (1995).
`4. M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, and M.A. Romero.
`Analytical control of a pharmaceutical formulation of sodium
`picosulfate by capillary zone electrophoresis. J. Chromatogr B
`751: 29–36 (2001).
`5. M. Blanco, J. Coello, H. Iturriaga, S. Maspoch, and M.A. Romero.
`Resolution of isomers of sorbitolparaben esters by chromatographic
`and electrophoretic techniques. J. Chromatogr B 752: 99–105
`(2001).
`6. U. Sidelmann, S.H. Hansen, C. Gavaghan, A.W. Nicholls, H.A.J.
`Carless, J.C. Lindon, I.D. Wilson, and J.K. Nicholson. Development
`of a simple liquid chromatographic method for the separation of
`mixtures of positional isomers and anomers of synthetic 2-, 3-
`and 4-fluorobenzoic acid glucuronides formed via acyl migration
`reactions. J. Chromatogr. B 685: 113–22 (1996).
`7. U. Sidelmann, S.H. Hansen, C. Gavaghan, H.A.J. Carless, J.C.
`Lindon, R.D. Farrant, I.D. Wilson, and J.K. Nicholson. Measure-
`ment of internal acyl migration reaction kinetics using directly
`coupled HPLC-NMR: application of the positional isomers of syn-
`thetic (2-fluorobenzoyl)-D-glucopyranuronic acid. Anal. Chem.
`68: 2564–72 (1996).
`8. U. Sidelmann, A.W. Nicholls, P.E. Meadows, J.W. Gilbert, J.C.
`Lindon, I.D. Wilson, and J.K. Nicholson. High-performance liquid
`chromatography directly cou