834 GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999
DRUGS, COSMETICS, FORENSIC SCIENCES
Validation of a Gas Chromatographic Method for Determining
Fatty Alcohols that Compose Policosanol in Five-Milligram
Film-Coated Tablets
GONZ LEZ CANAVACIOLO & MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999
VICTOR L. GONZÁLEZ CANAVACIOLO and JUAN MAGRANER HERNÁNDEZ
National Center for Scientific Research, Center of Natural Products, PO Box 6880, Cubanacán Playa, Havana City, Cuba
A gas chromatographic method using a packed
column and 1-eicosanol as an internal
standard was developed and validated for
determination of the aliphatic fatty alcohols
that compose policosanol in 5 mg film-coated
tablets. The alcohols were analyzed as
trimethylsilyl (TMS) derivatives, prepared with
N-methyl-N-trimethylsilyltrifluoroacetamide. The
method can detect degradation products with high
retention times without interfering with the peaks
of the active principle. Good linearity (correlation
coefficient = 0.9996) and accuracy (recovery =
100.44%) were proven over a range of 50–150% of
the nominal concentration. Within-day and
between-day precisions at the nominal 100% value
met the acceptance criteria (<2%). Ruggedness
was examined through an intralaboratory
experimental study in which 7 operational changes
were made and the observed results were
quantitation, repeatability, resolution, and relative
retention time. Among these results, only the
relative retention time (tC28,C20) was significantly
affected when the column used was 2.1 m instead
of 3.1 m. Repeatability and reproducibility (r =
0.1506 and R = 0.2450, respectively) were obtained
from a uniform-level interlaboratory test. The
method is suitable for quality control and stability
studies of these tablets.
olicosanol is a new cholesterol-lowering drug isolated
and purified from sugar cane (Saccharum officinarum
L.) wax. Its hypocholesterolemic effect has been demonstrated in experimental models (1, 2), healthy volunteers
(3), and patients with type II hypercholesterolemia (4–6). This
product is a mixture of 8 long-chain primary aliphatic alcohols
(C24–C34), in which 1-octacosanol is the major component (7).
In a previous work (8), we developed a gas chromatographic
(GC) method using a packed column for determination of the
fatty alcohols that compose policosanol in film-coated tablets.
P
Received August 14, 1998. Accepted by JM February 24, 1999.
In the present work, we validated an economical and less
time-consuming new version of this method.
Experimental
Apparatus
(a) Gas chromatograph.—Shimadzu GC 14B, interfaced
with a Shimadzu C-R4A computerized data processor, or
equivalent, equipped with a flame ionization detector
(Shimadzu, Kyoto, Japan). GC conditions: 3 mm (id) × 3.1 m
glass column, packed with 3% OV-101 on 80–100 mesh
Chromosorb W (HP) (Supelco, Bellefonte, PA). Injector and
detector temperatures, 320°C, column programmed from
250E to 320°C at 10°C/min and maintained isothermal for
20 min at 320°C; argon carrier gas flow, ca 35 mL/min.
(b) Gas chromatograph–mass spectrometer.—Fison Instruments (Manchester, England) GC 8000 Series MD-800,
coupled to a computerized data processor, or equivalent,
equipped with a capillary column SPB-5, 0.32 mm id × 25 m.
Operating conditions: column programmed from 100° to
200°C at 40°C/min, and from 200° to 320°C at 8°C/min and
then maintained isothermal for 90 min at 320°C; helium carrier gas flow, ca 1 mL/min. Injector, separator, and ion source
temperatures were fixed at 320°, 200°, and 250°C, respectively. Electron energy was 70 eV.
Chemicals
All chemicals were analytical reagent grade:
N-methyl-N-trimethylsilyltrifluoroacetamide
(MSTFA;
Fluka, Buchs, Switzerland, or equivalent), CHCl3, HCl
(0.1 mol/L), H2O2 (30%), and NaOH (0.1 mol/L; Merck,
Darmstad, Germany, or equivalent) were used.
To prepare the internal standard solution, 0.4 mg
1-eicosanol/mL, accurately weigh 100 mg 1-eicosanol (Janssen
Chimica, Geel, Belgium, or equivalent) into a 250 mL
volumetric flask. Dilute to volume with CHCl3 and mix.
To prepare the stock solution, accurately weigh 10 mg
1-tetracosanol (0.04 mg/mL), 18 mg 1-hexacosanol
(0.07 mg/mL), 10 mg 1-heptacosanol (0.04 mg/mL), 182 mg
1-octacosanol (0.73 mg/mL), and 35 mg 1-triacontanol
(0.14 mg/mL; Janssen Chimica, or equivalent) into a 250 mL
volumetric flask. Dilute to volume with CHCl3 and mix.
GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999 835
Table 1. Factors and conditions investigated in
ruggedness study
Conditions examined
Factors
(A) Temperature of injector
(B) Temperature of detector
Upper level
(+)
Lower level
(–)
320°C
310°C
320°C
310°C
(C) Flow rate
30 mL/min
25 mL/min
(D) Temperature program
12°C/min
10°C/min
(E) Initial oven temperature
250°C
240°C
(F) Column length
3.1 m
2.1 m
(G) Injection volume
3 µL
2 µL
To prepare the working standard solution of policosanol
(1.25 mg/mL), accurately weigh 125 mg policosanol
(DALMER S.A. Laboratories, Havana, Cuba; batch No. 294)
into a 100 mL volumetric flask. Dilute to volume with CHCl3
and mix.
Tablet Formulation
The test procedure was applied to a commercial
formulation of policosanol in film-coated tablets, consisting
of 5 mg policosanol, lactose, sucrose, gelatin, cornstarch,
special talc for tablets, microcrystalline cellulose,
magnesium stearate, and sodium croscarmelose as
excipients. Tablets were coated with a mixture of cellulose
acetophthalate, polyethylene glycol 20000, special talc for
tablets, titanium dioxide, and FD&C blue No. 2 lake
(indigotine). The tablet mass was ca 120 mg.
Test Procedure
Twenty tablets were randomly selected and weighed. The
average mass was calculated, and then the tablets were
crushed to a fine powder. A mass equivalent to that of one tablet was transferred quantitatively to a 10 mL test tube with
screw cap. Internal standard solution (3 mL) was added, and
the tube was heated at 60°C for 10 min with occasional shaking. The hot solution was filtered (Whatman No. 1 filter paper;
Maidstone, UK), and 0.5 mL of the filtrate was transferred to
another tube. MSTFA (0.05 mL) was added, the tube was
heated for 15 min at 60°C, and 2 µL portions were examined
by GC.
The mass of each alcohol was obtained by the internal standard method (9) according to the following equation:
Mass of compound =
area of compound i × mass of internal standard × f iw
area of internal standard
where fwi is the relative mass response factor for compound i.
To determine fwi, a portion of the stock solution (0.5 mL)
was evaporated to dryness at 60°C with an air stream. Internal
standard solution (0.15 mL) and MSTFA (0.05 mL) were
added, and the mixture was heated for 10 min at 60°C. Portions (2 µL) of this solution were injected into the gas
chromatograph. This procedure was performed in triplicate,
and fwi was calculated as follows:
Figure 1. GC profiles of (a) 1-eicosanol, (b) policosanol (working standard solution), and (c) placebo tablets after
TMS derivatization. Peaks correspond to TMS derivatives of (IS) 1-eicosanol, (1) 1-tetracosanol, (2) 1-hexacosanol,
(3) 1-heptacosanol, (4) 1-octacosanol, (5) 1-nonacosanol, (6) 1-triacontanol, (7) 1-dotriacontanol, and (8)
1-tetratriacontanol.
836 GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999
Figure 2. GC profiles of (a) tablets containing 5 mg policosanol and (b) thermally degraded tablets after TMS
derivatization.
fwi =
area of internal standard × mass of compound i
area of compound i × mass of internal standard
Because commercial standards of 1-nonacosanol,
1-dotriacontanol, and 1-tetratriacontanol were not available,
the relative response factor of 1-octacosanol was used for
1-nonacosanol, and for the other alcohols, the relative mass response factor of 1-triacontanol was used. The total mass of
policosanol was determined by adding the masses of all the alcohols. This value was corrected by taking into account sample mass and tablet average mass.
Method Validation
To enable formation of degradation products, the tablets
were crushed to a fine powder and subjected to thermolysis
(55°C, 9 months), base and acid hydrolysis (sample sus-
pended in 0.1 mol NaOH/L and 0.1 mol HCl/L, respectively,
at ca 1 g in 10 mL, at 108°C, 1 day), oxidation (sample suspended in 30% H2O2, at ca 1 g in 10 mL, at 25°C, 1 week), and
photolysis (254 nm UV light, at 25°C, 1 week). These tests
were performed in neutral glass ampoules, which were
flushed with nitrogen and sealed (n = 3). Specificity of the
chromatographic system was proven by comparing
chromatograms of the placebo, the internal standard solution,
the working standard solution of policosanol, the original tablets, and the tablets stressed under degradation conditions.
Peak purity was checked by gas chromatography/mass spectrometry (GC/MS) analysis.
Linearity and accuracy were assessed over the range
50–150% of the nominal concentration. The matrix (120 mg
placebo) was spiked with 2, 3, 4, 5, and 6 mL of the working
standard solution. Samples were evaporated to dryness, and
Table 2. Linearity and accuracy of GC determination of policosanol in film-coated tablets
Amount found, mg
Amount added, mg
Recovery, %
1
2
3
1
2
3
Average, %
2.50
2.51
2.50
2.52
100.40
100.00
100.80
100.40 ± 0.40
3.75
3.80
3.73
3.83
101.33
99.47
102.13
100.98 ± 1.36
5.00
5.07
4.99
5.01
101.40
99.80
100.20
100.47 ± 0.83
6.25
6.17
6.22
6.20
98.72
99.52
99.20
99.15 ± 0.40
7.50
7.62
7.57
7.51
101.60
100.93
100.13
100.89 ± 0.74
Average, %
RSD, %
100.38 ± 0.97
0.97
GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999 837
Table 3. Results of repeatability assay for GC determination of policosanol in film-coated tablets
Amount found, mg, by laboratory 1 in indicated samples
Tabletsa
Spiked placeboa
Tabletsb
Amount found, mg,
by laboratory 2 in tabletsa
1
4.94
4.99
4.86
4.89
2
5.01
4.91
4.82
4.81
3
4.97
5.07
4.86
4.87
4
4.89
5.01
5.05
4.91
5
5.02
5.12
4.84
4.87
6
5.03
4. 91
4.80
4.90
7
4.91
4.90
4.81
5.05
8
4.94
4.88
4.84
4.95
Average, mg
4.94
4.98
4.86
4.92
SD
0.07
0.086
0.08
0.06
RSD, %
1.45
1.70
1.64
1.20
Determination
a
b
Current method.
Previous method (8).
then the mixture was analyzed. Three independent determinations were performed in each case. The regression line (y = a +
bx) was calculated by the method of least squares based on the
amount found (y) versus the amount added (x). Evaluation was
made by linearity and proportionality tests for p = 0.05.
Acceptance criteria were as follows: (1) Correlation
coefficient > 0.999. (2) Relative standard deviation of response
factors (RSDf) < 5%, where response factor is defined as y/x. (3)
Relative standard deviation of slope (RSDb) < 2%, with
RSDb, % =
where SDa = standard deviation of the intercept.
The accuracy was assessed by a recovery study. Recoveries were calculated according to the following equation:
Recovery, % =
Average recovery was checked to 100% with the Student t
test. The experimental value of t was calculated as follows:
SDb
× 100
b
t=
where SDb = standard deviation of the slope. (4) The zero
value should be included in the confidence interval (CI):
CI = a ± t × SDa
amount found
× 100
amount added
100− recovery n
RSD
The null hypothesis (the recovery is close to 100% and the
method is accurate) was accepted for a significance level
greater than 5%.
Within-day and between-day precisions were evaluated by
assaying commercial tablets in 2 different laboratories by the
Table 5. Effects of operational changes on results of
ruggedness study
Table 4. Means of results in ruggedness study (n = 6)
Factora
Policosanol, mg
SD
tC28,C20
RsC28-C30
A
0.02
0.01
0.07
0.10
B
0.04
0.01
0.06
0.26
3.96
C
0.06
0.00
0.14
0.06
2.25
4.69
D
0.02
0.02
0.09
0.24
0.02
2.02
4.54
E
0.06
0.00
0.13
0.16
4.88
0.02
1.71
4.11
F
0.05
0.00
0.28
0.06
6
4.80
0.01
2.31
4.41
G
0.03
0.00
0.04
0.14
7
4.86
0.02
2.18
4.33
0.09
0.20
0.28
0.32
8
4.94
0.02
2.13
4.24
Experiment Policosanol, mg
SD
tC28,C20
RsC28-C30
1
4.91
0.04
2.04
4.27
2
4.77
0.04
2.30
3
4.90
0.02
4
4.80
5
%2 SD
a
As defined in Table 1.
838 GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999
Table 6. Collaborative results for GC determination of policosanol in film-coated tablets
Found, mg/tablet
Laboratory
1
2
3
Mean, mg/tablet
SD
RSD, %
1
4.63
4.64
4.69
4.66
0.031
0.66
2
4.63
4.68
4.70
4.67
0.036
0.77
3
4.77
4.77
4.78
4.77
0.006
0.13
4
4.62
4.70
4.70
4.67
0.046
0.99
5
4.64
4.77
4.80
4.74
0.085
1.79
6
4.63
4.66
4.68
4.66
0.025
0.54
7
4.72
4.72
4.87
4.77
0.087
1.82
8
4.70
4.73
4.83
4.75
0.068
1.43
0.036
0.79
9
a
4.50
4.55
4.57
a
4.54
Value outlier according to Grubb’s test.
current method. One laboratory also assayed a sample of
spiked placebos and further assayed tablets by a previously reported method (8): A mass of powder equivalent to that of a
tablet was transferred quantitatively to a round-bottom flask;
1,2-dichloroethane (20 mL) and a solution of 1-eicosanol
(1 mg/mL) in 1,2-dichloroethane (0.5 mL) were added; the
mixture was heated to reflux for 20 min; the hot solution was
filtered (Whatman No. 1 filter paper); the filtrate, collected in
another round-bottom flask, was evaporated to dryness;
MSTFA (0.1 mL) and CHCl3 (1 mL) were added; and this
mixture was heated at 60°C for 15 min. Portions (2 µL) were
analyzed by GC. Each operator followed the procedure under
conditions of repeatability (n = 8). Fisher (F) and Student (t)
tests for p = 0.05 were performed to determine significant differences between results.
Ruggedness was evaluated by an intralaboratory experimental study in which the influence of small changes in the
operating conditions on results was evaluated. Seven variables were studied by performing 8 separate experiments (n =
6) as described by Youden (10). The selected factors and levels are shown in Table 1. The observed results were amount of
policosanol (P), precision (SD), resolution between alcohols
C28 and C30 (RsC28-C30), and relative retention for alcohol C28
(tC28,C20).
Precision was also determined from a uniform-level
interlaboratory test involving 9 laboratories. Participants were
requested to perform 3 replicates under repeatability
conditions. Statistical analyses were performed according to
the guidelines in ISO 5725.
Results and Discussion
The analytical procedure allows rapid, reliable, and reproducible determination of the mixture of long-chain primary
aliphatic alcohols identified as policosanol. The specificity assay showed that degradation occurred only when tablets are
subjected to thermolysis. The method detects degradation
products (octacosanyl stearate and palmitate; 11) at high re-
tention times (55 and 75 min), without interfering with the determination of policosanol.
Figure 1 shows that the chromatographic conditions gives
well-resolved peaks corresponding to samples of placebo,
1-eicosanol, and policosanol. No difference between the partial chromatogram of degraded tablets and the chromatogram
of tablets containing 5 mg policosanol can be seen in Figure 2.
The peaks corresponding to degradation products are not
shown because they elute at high retention times.
Linearity and Accuracy
The calculated regression line was y = (1.0035 ± 0.0075)x
+ (0.0033 ± 0.0400), with CI calculated at p = 0.05. The
correlation coefficient was 0.9996. Because use of the
correlation coefficient is not always sufficient to assess
linearity, RSDf (0.92%) and RSDb (0.75%) were calculated.
The CI (0.003 ± 0.086) includes the zero value; thus it can be
concluded that the line passes through the origin. The
procedure could be considered linear and proportional in the
concentration range studied.
Recoveries from spiked placebos were 99.15–100.98% of
the expected amounts, with RSD values <2% (Table 2). These
high recoveries were made possible by the addition of an internal standard at the beginning of extraction. The average recovery and 100% of recovery were not significantly different
according to the calculated t (1.850), which was lower than the
critical value for p = 0.05.
Repeatability
The method performed well in the initial precision study
with real tablets by 2 laboratories (Table 3). No significant
variations between the 2 series of analysis (Fcalc = 1.45 and
tcalc = 0.613) were found through the F and t tests. Within-day
RSD values were <2% in both cases and between-day RSD (n
= 16) was 1.32%. Confidence limits were 4.93 ± 0.04 mg;
therefore in 95 of 100 cases, the real value will lie within the
range 4.89–4.97 mg.
GONZ
LEZ CANAVACIOLO
& MAGRANER HERNÁNDEZ: JOURNAL OF AOAC INTERNATIONAL VOL. 82, NO. 4, 1999 839
Real tablets and spiked placebos were analyzed to evaluate
matrix interference effects. The amount of policosanol per
tablet was 98.80% (4.94 mg) of the declared dose, and the
recovery from spiked placebos was 99.60% (4.98 mg) of the
added amount (Table 3). These results are very similar, and
the calculated values for F and t (1.467 and 1.192,
respectively) are less than the critical values for p = 0.05.
Therefore, the mixing of active ingredients and excipients
does not affect recovery.
Finally, statistical analyses (experimental F and t values
1.306 and 2.128, respectively) show no significant difference
in either the precision or the means obtained from the 2 sample
preparation methods.
Ruggedness
The means of results for each experiment (Table 4), the effects on the observed results, and the limit values of these effects as expressed by %2 SD (Table 5) were calculated. Only
the tC28,C20 was significantly affected when the column used
was 2.1 m. No significant effect was observed on other results.
Interlaboratory Assay
Results of the interlaboratory assay are presented in Table 6. According to the Cochran test, there are no outliers in
this table, but according to Grubb’s test, one collaborator deviated from the procedure. For subsequent analyses, all values
were considered. Method precision was good, as demonstrated by the following statistical data for repeatability: r =
0.1506, Sr = 0.053, and RSDr = 1.13%; and for
reproducibility: R = 0.2450, SR = 0.087, and RSDR = 1.85%.
Acknowledgments
We thank R. Sierra and C. Velázquez for their help during
analysis of samples. We are also grateful to DALMER S.A.
Laboratories for supplying policosanol tablets.
References
(1) Arruzazabala, M.L., Carbajal, D., Más, R., Castaño, G.,
Sotolongo, R., & Mesa, R. (1991) Revista CENIC Ciencias
Biológicas 22, 60–61
(2) Arruzazabala, M.L., Carbajal, D., Más, R., Molina, V.,
Valdés, S., & Laguna, A. (1994) Biol. Res. 27, 205–208
(3) Hernandez, F., Illnait, J., Más, R., Castaño, G., Fernández, L.,
Gonzalez, M., Cordoví, N., & Fernández, J.C. (1992) Curr.
Ther. Res. 51, 568–575
(4) Aneiros, E., Más, R., Calderón, B., Illnait, J., Fernández, L.,
Castaño, G., & Fernández, J.C. (1995) Curr. Ther. Res. 56,
176–182
(5) Canetti, M., Morera, M., Illnait, J., Más, R., Castaño, G.,
Fernández, L., & Fernández, J.C. (1995) Int. J. Clin. Pharm.
Res. XV, 159–165
(6) Castaño, G., Más, R., Fernandez, J.C., & Illnait, J. (1997)
Curr. Ther. Res. 58, 154–162
(7) Laguna, A., Magraner, J., Carvajal, D., Arruzazabala, M.L.,
Más, R., & García, M. (1993) Patent Cooperation Treaty EP
937/00007
(8) Magraner, J., Gonzalez, V., Laguna, A., Velazquez, C.,
Lorenzo, M., & Uribarri, E. (1998) Revista CENIC Ciencias
Químicas 29, 127–130
(9) Gasco, S. (1969) Teoria y Práctica de la Cromatografía en
Fase Gaseosa, JEN Editions, Madrid, Spain, p. 340
(10) Youden, W.J., & Steiner, E.H. (1975) Statistical Manual of
the AOAC, AOAC, Arlington, VA
(11) González, V.L., Magraner, J., Cabrera, L., & Rivero, B.
(1998) Revista CENIC Ciencias Químicas 29, 27–30