PREDICTABILITY OF ENANTIOMERIC CHROMATOGRAPHIC BEHAVIOR ON A MODELING SOFTWARE
507
two parameters as modifiers of the chiral recognition. The % organic
phase and the pH were the investigated parameters.
Polysaccaride CSPs (Chiralpak AD-RH) are limited with respect to
operating pressure; accordingly the flow rate was not a studied var-
iable. The % organic phase and the temperature were the parame-
ters under investigation.
Protein CSPs (Ultron ES-OVM) are limited regarding pressure
and temperature (40 ꢀC), as well as the % organic phase, which
was only studied in the limited range allowed.
industry, environmental and agricultural fields as it will reduce
the time required for the chiral chromatographic method deve-
lopment. Three CSPs are investigated in the present study
namely, macrocyclic glycopeptides (Chirobiotic V and T), poly-
saccharides (Chiralpak® AD-RH) and proteins (Ultron ES-OVM).
Few reported trials have aimed to predict the enantioselectivity
of chiral drugs. One of the earliest studies predicted the
enantioselectivity of a number of chiral drugs on derivatized
ß-cyclodextrin stationary phase based on free energy calcula-
tions of substituents present on the stereogenic center. This
study was able to predict the possibility of enantiomeric
separation on a specific stationary phase rather than the
elution order.22 Previous works relating chiral separation to
two types of HPLC method development software, ACD Lab
and Chromsword, were reported. ACD Lab software was able
to predict the optimum %B required for the separation of
enantiomers of eszopiclone, using tandem mass spectromet-
ric detection on chiral AGP, which is an a1-acid glycoprotein
stationary phase23; for Chromsword, chiral drugs were
separated on a polysaccharide-based stationary phase using
normal phase mode. DryLab was the tool for prediction in only
two studies, which were performed on a quinine carbamate-
based chiral anion-exchanger24 and a tert-butyl carbamoylated
quinine25 stationary phase, where N-derivatized amino acids
were successfully resolved.
•
•
Experimental preliminary runs. The preliminary-run conditions
required to start predictions via computer simulations were selected
according to the column restrictions previously mentioned and the corre-
sponding DryLab operational mode. These conditions are presented in
Table 1 as follows: for Chirobiotic V and T, two isocratic runs at 75%
and 95%B; for Chiralpak AD-RH, four isocratic runs at 40% and 60%B each
at two different temperatures (30 ꢀC and 60 ꢀC); and for Ultron ES-OVM,
two isocratic runs at 5% and 25%B.
The preliminary runs were exported to PeakMatch software in artificial
intelligence assemblage (AIA) format for peak tracking where the peak
areas were taken as identifiers for the peaks. These were subsequently
exported to DryLab to build the resolution map of the method.
The optimum chromatographic separation conditions were selected via
the resolution map to achieve the best resolution for the enantiomers in
the least possible run time. Then the computed model was experimen-
tally checked for accuracy of the retention time via comparison between
predicted and experimental chromatograms.
Determination of the Effect of pH
MATERIALS AND METHODS
Three preliminary runs were performed each at a different
pH value (differing by 0.5–0.6 pH units).17 After peak tracking
and obtaining the resolution map, retention time was tested
for accuracy between virtual and experimental chromatograms.
Chemicals and Reagents
Methanol, acetonitrile (HPLC grade), trifluoroacetic acid (TFA), glacial
acetic acid, monobasic sodium monophosphate (NaH2PO4), dibasic
sodium monophosphate (Na2HPO4), ammonium hydroxide (NH4OH),
phosphoric acid (H3PO4), and triethylamine (TEA) were purchased from
Sigma-Aldrich, Germany. Rac-Propranolol, rac-salbutamol sulfate, rac-
terbutaline sulfate, rac-tetoprolol, and rac-ibuprofen were provided by
SEDICO Pharmaceutical Company, Egypt. Rac-Warfarin was provided
by GlaxoSmithKline (GSK, Cairo, Egypt).
RESULTS AND DISCUSSION
Chirobiotic V
Results of retention time at various conditions are shown in
Table 2. The resolution map (plot of minimum resolution for
the poorest-resolved band pair in the sample versus % organic),
presented in Figure 1, showed that the best resolution for pro-
pranolol enantiomers was obtained at 100%B, proven by com-
parison of the results of the experimental run and the
predicted. A difference in retention times between predicted
and experimental of only 1.50 min and 0.51 min, for the first
and the second propranolol enantiomers, respectively, was
found, resulting in 97.80% average accuracy between experi-
mental and predicted as demonstrated in Table 3.
The accuracy of the predictions was tested for warfarin as
well—both the effect of the % organic phase and the pH were
studied. For the effect of % organic phase, the retention time
of the two enantiomers for each of the two preliminary runs,
5% and 25%B, are shown in Table 2, and indicate that when
%B (organic phase) increased, both the retention time and
the Rs decreased. Optimum separation conditions, according
to the computed resolution map, were obtained when the %B
was at 10%, as shown in Figure 2. A comparison of the in-silico
model chromatogram results with the experimental results is
presented in Table 3, indicating a difference of 1.188 min and
0.585 min for the first and second eluted warfarin enantio-
mers, which proved an average accuracy on prediction of
the retention time by 93.45%. Some peak tailing was observed
in the later enantiomer of warfarin.
Equipment and Software
HPLC (Thermo Finnigan SpectraSYSTEM, Thermo Electron Corpora-
tion, UK) consisted of: pump model P2000, detector model UV3000,
autosampler model AS3000, and data acquisition for the HPLC Chromquest
4.2 data system (UK).
The columns used are Chirobiotic V bonded vancomycin-based phase,
5-mm particle size (150 Â 4.6 mm) and the Chirobiotic T bonded
teicoplanin-based column, 5-mm particle size (150 Â 4.6 mm), both
purchased from Advanced Separations Technologies, Inc. (Astec)
and Supelco (Bellefonte, PA, USA); Chiralpak AD-RH amylose tris
(3,5-dimethylphenylcarbamate) coated on 5-mm silica gel (150 Â 4.6
mm) purchased from Daicel Chemical Industries Ltd (Tokyo, Japan);
and Ultron ES-OVM: protein, ovomucoid, chemically bonded to the silica
support (150 Â 4.6 mm), purchased from Agilent Technologies, USA.
The computer programs used were Chromquest 4.2 data system (UK) for
data acquisition, DryLab2000Plus, and PeakMatch, v. 3.60 (Molnár Institute
for Applied Chromatography, Germany).
METHODS
Determination of Predictability of the CSPs
Experimental limitations. The parameters that can be optimized by
DryLab software are: % organic phase, temperature, pH, and flow rate.
Three different CSPs were investigated:
•
Macrocyclic glycopeptides (Chirobiotic V and T) have limitations
regarding operating pressure and temperature (max. 45 ꢀC) and
considerable variations of temperatures and flow rate from low to
high values were not possible, limiting the ability to study these
The effect of pH on separation of warfarin enantiomers on
Chirobiotic V was studied. The three isocratic runs at 10%B
at three different pH values (3.30, 3.90, and 4.50) showed that
Chirality DOI 10.1002/chir