Manual
In Situ Fiber Optic Dissolution Analysis in Quality Control
Schatz C.,(1
) Ulmschneider M.,(1 ) Altermatt R.,(1 ) Marrer S.,(1 ) and Altorfer
H.(2)
(1 ) Pharmaceutical
Quality Control and Quality Assurance, F. Hoffmann-La Roche Ltd.,
Basel, Switzerland
(2) Pharmaceutical Analysis,
Swiss Federal Institute of Technology, Zurich, Switzerland
email for correspondence: caspar.schatz@roche.com
Abstract
Manual dissolution testing entails time-consuming sample preparation.
Samples withdrawn from the dissolution vessels require filtration,
then the clear solutions have to be transferred into cells for
quantification by UV/VIS spectroscopy via an absorption reading.
This article describes the validation of a new method that enables
the extent of dissolution to be determined without any sample
preparation because the active substance concentration is measured
directly in the dissolution vessels with a fiber optic probe.
Using soft gelatin capsules and tablets as models, the method
was validated with regard to linearity, accuracy, precision, specificity
and robustness. The analytical results confirmed that the fiber
optic quantification method is simple, reliable, reproducible,
robust, time-saving, and easy to use in dissolution testing.
Introduction
As a compendial requirement the dissolution
methods for a given drug product are fixed in terms of medium,
temperature, stirring speed, and stirring devices. Any change
in a parameter that affects dissolution efficiency, such as paddle
speed or dissolution medium, requires extensive experiments to
get the approval of the regulatory authorities. Options for reducing
the analysis time for manual dissolution tests are therefore very
limited. However, the method employed to quantify the amount of
active substance dissolved is a key determinant of the analysis
time. Usually the quantification is done either by high performance
liquid chromatography (HPLC) or by direct UV/VIS spectroscopy.
Owing to the additional separation step, HPLC is more time-consuming
and more expensive than direct UV/VIS spectroscopy. Since the
dissolution testing of almost every solid drug product leads to
a turbid solution, sample withdrawal from dissolution vessels
must be followed by a filtration step to get rid of undissolved
components from the drug product and to eliminate potential interference
during quantification. When using direct UV/VIS spectroscopy for
quantification, the clear solutions are transferred to cells prior
to the absorption reading.
Using spectroscopy instead of chromatography
to determine the amount of active substance in solution is one
way of reducing the necessary analysis time. Another approach
minimizes the time spent on sample preparation prior to manual
dissolution testing. This can be achieved by using a fiber optic
probe to measure the absorption in the dissolution vessel directly.
Since the filtration step is omitted, the analytical method has
to compensate for the turbidity effects of undissolved excipients
by using two wavelengths. One wavelength - preferably at an absorption
maximum - is used for quantification and the other - at a point
where none of the excipients shows any absorption - is used for
the turbidity correction.
This approach was tested on two different
drug products. One was a soft gelatin capsule, containing 200
mg of active substance dissolved in a mono- and diglyceride filling.
While performing the dissolution test with these capsules the
medium becomes turbid owing to the metal oxide pigments from the
capsule shell. The other product examined was a benzodiazepine
tablet formulation containing 6 mg of active substance. In this
case the turbidity of the dissolution medium is mainly due to
microcrystalline cellulose.
Experimental
All dissolution tests were performed with apparatus 2 [1] (Sotax
AT 7, Sotax AG, Allschwil, Switzerland). A scanning spectrometer
(Varian Cary 50, Varian International AG, Zug, Switzerland) was
used with a fiber optic probe (Ultra Mini TS 10 mm + 2 LL UV Li/SMA
974725/18, Hellma GmbH & Co., Müllheim/Baden, Germany)
with a 10 mm path length. As shown in Figure 1, the absorption
readings with the fiber optic probe were taken via the sampling
hole of the dissolution tester about 3 cm below the surface of
the medium, with degassing by helium sparging or sonication under
vacuum [2].
Figure 1: Experimental arrangement for taking absorption readings using a fiber optic probe.
Soft
gelatin capsules
The soft gelatin capsules have a Q-value of 85% after 20 minutes
in 900 mL of dissolution medium (aqueous citrate buffer solution,
pH 3.0, with 12% Cremophor® EL) at 37.0°C ± 0.5°C
stirred at 100 rpm. Except for the robustness determination, all
dissolution tests were performed under these conditions.
Method development
Dissolution tests with the drug product, the active substance,
and the placebo formulation in dissolution medium were performed
in order to establish a suitable wavelength for the turbidity
correction.
Methods
In the case of the registered method (validated according to Ref.
[3, 4]) an aliquot of about 20 mL was withdrawn from each vessel
after 20 minutes. The samples were filtered (pore size 0.45 µm)
and the absorption of the clear solutions was measured at room
temperature against a dissolution medium blank at 315 nm (A315)
with a cell width of 10 mm. The amount of active substance dissolved
was calculated as a percentage (%dissolved) using Formula 1, where
4,500 is a factor accounting for the dilution and the conversion
to percent and 53.0 is the A (1%, 1 cm) value of the active substance
in clear dissolution medium at room temperature.
Formula 1:
%dissolved=A315*4,500/53.0
Using the fiber optic probe, the
absorptions at 315 nm (A315) and 380 nm (A380) were measured directly
in the vessel. The amount of dissolved active substance in percent
(%dissolved) was calculated using Formula 2, where 53.0 is the
A (1%, 1 cm) value of the active substance in placebo-spiked dissolution
medium at 37.0° ± 0.5°C.
Formula 2:
%dissolved=(A315-A380)*4,500/53.0
Validation of the fiber optic
method
The linearity of active substance in placebo-spiked dissolution
medium was examined in replicates of six at six concentrations
ranging from 5 to 120% dissolved active substance (0.011
0.267 mg/mL).
A (1%, 1 cm) was calculated with
these data, as well as the accuracy and the precision as an internal
validation. A method comparison with the registered method for
dissolution testing of the soft gelatin capsules was carried out
for the additional determination of accuracy and precision. For
these experiments six capsules from seven lots were used to determine
the amount of active substance dissolved.
The placebo response was determined
with six placebo capsules as well as six soft gelatin capsule
shells of the drug product. The robustness of the analytical method
for determining the amount of active substance dissolved was examined
in terms of temperature, paddle speed, and the immersion depth
of the fiber optic probe window. In addition, dissolution profiles
over 20 minutes were measured with six soft gelatin capsules.
Benzodiazepine
tablets
The registered Q-value of the benzodiazepine tablets in 900 mL
of simulated gastric fluid without pepsin [1] is 75% after 20
minutes, the vessels being stirred at 50 rpm at 37.0° ±
0.5°C. Accordingly, all dissolution tests were performed with
these parameters.
Method development
To establish a turbidity correction wavelength, dissolution tests
with the drug product, the active substance, and placebo powder
were performed.
Methods
For the registered method (validated according to Ref. [3, 4])
and the fiber optic method, a standard of about 0.0065 mg/mL (cstandard)
was prepared and its absorption at 239 nm (Astandard) was measured
at room temperature 20 minutes later.
In the case of the registered method,
a sample volume of about 20 mL was withdrawn from each vessel
and filtered (pore size 0.45 µm). The absorption of the
clear solutions at 239 nm (A239) was measured against simulated
gastric fluid without pepsin [1] at room temperature, with a cell
width of 10 mm. The amount of dissolved active substance in percent
(%dissolved) was calculated using Formula 3, where 15,000 is a
factor accounting for conversion to percent and dilution.
Formula 3:
%dissolved=A239*cstandard * 15,000/Astandard
For the fiber optic method the absorption
was measured directly in the vessels at 239 (A239) and 450 (A450)
nm with a path length of 10 mm. Formula 4 was used to determine
the amount of active substance dissolved.
Formula 4:
%dissolved=(A239 -A450)*cstandard * 15,000/Astandard
Validation of the fiber optic
method
The linearity was investigated in triplicate at five concentrations
ranging from 25 to 125% dissolved (0.0017 0.0083 mg/mL).
To spike the solutions, placebo powder was added before taking
the absorption readings in the dissolution vessels under regular
test conditions. The same data set was used to determine the accuracy
as well as the precision. The fiber optic method was then compared
with the validated and registered method. Six tablets from three
lots were used. For reference, six vessels were filled with placebo-spiked
model solutions corresponding to a dissolution of about 90% (0.0060
mg/mL). This experiment was carried out in replicates of six (n=6),
filtering with a pore width of 0.2 µm. The resulting data
sets enabled the accuracy as well as the precision to be calculated.
The placebo response (six assays) and the dissolution profiles
of six tablets were determined.
Results
Soft gelatin capsules
Method development
From Figure 2 it can be seen that the active substance does not
show any absorption above 365 nm. The absorption of the placebo
capsule is 0.050 over the entire wavelength range, owing to the
turbidity. Above 365 nm the soft gelatin capsule absorption curve
matches the turbidity line.
Figure 2: Absorption spectra of
0.26 mg/ml active substance (1), a soft gelatin capsule (2), and
a placebo capsule (3) in 900 ml dissolution medium. .
For the new fiber optic method the shoulder of the active substance at 315 nm was chosen for quantification, just as in the registered model. The wavelength of 380 nm was chosen for the turbidity correction.
Validation of the fiber optic
method
For the A (1%, 1 cm) calculation the six linearity assays were
evaluated separately. The mean of the six A (1%, 1 cm) values
was 52.96 cm-1 %-1 with a relative standard deviation of 0.51%.
To determine whether common acceptance
criteria for linearity [3, 4] are met, the data sets were evaluated
as a batch. The coefficient of correlation was 0.99968, exceeding
the acceptance limit of > 0.99. The y-intercept met the acceptance
limits as well, lying within the 95% confidence interval of 2%
of the reference x-value (100% of active compound dissolved) around
the origin. All group standard deviations as well as the relative
repeatability standard deviation (0.44%) were within the acceptance
criterion of <1.00%. The internal validation gave a mean recovery
of 98.87%, satisfying the acceptance limits of 98.00 to 102.00%.
The method comparison data showed
no significant difference (p = 95%). Thus the 95% confidence interval
of the mean of the fiber optic method was entirely within the
acceptance interval of 2.00% around the mean of the validated
and registered method with filtration. The relative within-sample
standard deviation with this data set was found to be 0.49%, satisfying
the acceptance limit of <1.00%.
The placebo response was -0.17% for
the placebo capsules and -0.35% for the soft gelatin capsule shells.
Robustness
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Table 2: Dissolution values obtained at various probe immersion depths within the vessels. The depth of 3.0 cm was used for all other measurements.
|
|
|||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Figure 3 shows the influence of paddle
speed on the detected dissolution.
With the equipment used here the radial distance between the probe
and the paddle could not be varied and was within the requirements
of Ref. 1. Findings for probe window immersion depths of 0.5,
2.0, 3.0 and 4.5 cm are presented in Table 2.
A statistical comparison of the depths 0.5 and 4.5 cm did not
show any significant differences (p = 95%).
Figure 3: Influence of paddle
speed on the fiber optic quantification method. The dots correspond
to the mean of twelve soft gelatin capsules and the vertical lines
represent ± 1 standard deviation.
Furthermore, the absorption was measured
(sixfold determination) in one vessel at six different immersion
depths at 315 nm and 380 nm. The mean absorption readings at both
wavelengths as well as their differences are presented with the
corresponding standard deviations in Table 3 .
Table 3: The means of six absorption readings at 315 (A315) and 380 nm (A380) as well as the difference between the two (D A) are shown for different immersion depths in the same vessel. All obtained values are given with the corresponding standard deviations.
| Immersion depth |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
These results show that the turbidity
as well as the concentration of the active substance is homogeneous
in the examined region of the vessel because the different immersion
depth do not lead to significantly different results (p= 95%).
Although the turbidity is measured during each quantification,
further evidence about the robustness of the method is given owing
to the small turbidity differences.
As can be seen in Figure 4, the mean
dissolution profile has a sigmoid shape, with high variability
between 6 and 15 minutes. The acceleration at around 6 minutes
corresponds to the rupture of the capsule shell. After 15 minutes
the entire liquid content of the soft gelatin capsule has escaped
and the 100% plateau is reached.
Figure 4: Mean dissolution profile of six soft gelatin capsules. The vertical lines represent ± 1 standard deviation.
Benzodiazepine
tablets
Method development
For the fiber optic method the maximum at 239 nm in Figure 5 was
used for the determination of the active substance, just as in
the registered method. Since the spectrum of the active substance
meets the x-axis just before 450 nm, this wavelength was chosen
for the turbidity correction. At the upper end of the spectrum
the turbidity line is higher than the benzodiazepine tablet curve
owing to the difference in particle size between the placebo powder
turbidity and the turbidity resulting from tablet dissolution.
Validation of the fiber optic
method
The linearity assays met the above-mentioned acceptance criteria
concerning the coefficient of correlation (0.99993), the y-intercept,
all the group standard deviations and the relative sample standard
deviation (0.50%). The internal validation led to a mean recovery
of 100.38%.
The results of the method comparison with tablets are shown in Table 4.
Table 4: Mean dissolution test results after quantification by the two methods. The standard deviation of the six readings is given in brackets.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The dissolution values differ by less than 2.00% and so meet the
acceptance limits.
The method comparison with model solutions proved that the methods
are equivalent: the 95% confidence interval of the mean of the
fiber optic method lay fully within the acceptance interval of
2.00% around the mean of the registered method. The relative within-sample
standard deviation determined with this data set was 0.60%, similarly
meeting the acceptance criteria.
For the mean placebo response a value
of -0.20% was found.
The benzodiazepine tablets show the typical dissolution profile
(Figure 6) of an immediate release tablet formulation, characterized
by a steep initial slope which gradually levels off, resulting
in a final relative standard deviation of about 2%.
Figure 6. Mean dissolution profile of six benzodiazepine tablets. The vertical lines indicate the interval of ±1 standard deviation.
Discussion
With the benzodiazepine tablet model the spectra for the method
development were not as smooth as in the case with the soft gelatin
capsules since the particles causing turbidity were larger. While
quantifying the amount of dissolved active substance, the influence
of the turbidity and the turbidity particle size on the accuracy
of the absorption reading was eliminated by accumulating 80 xenon
lamp flashes in order to obtain one absorption reading. (The spectrometer
used was equipped with a xenon flash lamp as light source.) This
procedure was applied for the examination of both drug products.
The validation experiments for the
two drug products were performed differently owing to the differences
in quantification. In the case of the soft gelatin capsules the
linearity assays involved more samples because they served to
estimate A (1 %, 1 cm). The data range (5 to 120%) was chosen
so as to optimally trace the sigmoid dissolution curve.
To compensate for the disintegration of the active substance provoked
by UV light [5, 6], a one-point quantification with a reference
solution was performed for each experiment. The standard solution
for the one-point calibration is measured at room temperature,
leading to a maximal theoretical deviation of 0.002 AU corresponding
to the density decrease with increasing temperature [7, 8].
When examining the benzodiazepine
tablets, the placebo powder turbidity differed from the tablet
turbidity owing to the difference in particle size, as can be
seen in the results of the method development experiments. For
this reason the pore size of the filter had to be changed to 0.2
µm for the filtration of solutions containing placebo powder
[9]. Because of the use of two different filter sizes, the method
comparison was performed with tablets as well as model mixtures.
For the model mixtures a dissolution of 90% was chosen, which
is the regular value to be expected [9]. However, the method comparison
with tablets required different acceptance criteria to deal with
the scattering of the dissolution profiles, as can be deduced
from the relative standard deviation of about 2% at the end of
the mean benzodiazepine tablet dissolution curve [9]. In the case
of the soft gelatin capsules, scattering was virtually undetectable
at the end of the dissolution run owing to the design of this
dosage form.
Although the robustness was examined
with the soft gelatin capsules only, there is strong evidence
that the robustness is the same in the case of the benzodiazepine
tablets [10].
Conclusions
With the new method presented here the analysis time for manual
dissolution testing can be drastically reduced. Thanks to the
lack of sample preparation, the new fiber optic method is 6 to
8 times faster and avoids the use of disposable materials such
as filters and syringes. The new method rapidly repays the necessary
investments in the fiber optic probe and fiber optic coupler.
For automated dissolution testing,
the filter and pump station can be replaced by fiber optic probes,
thus eliminating the laborious qualification and validation of
this equipment. Another advantage is the elimination of typical
pumping and filtration problems such as air bubbles in the flow-through
cells and pumping system, absorption by tubing, compatibility
issues, volume deviations, filter clogging and drug hold-up.
Thus the use of fiber optics offers considerable advantages in
manual dissolution testing and may be the ideal tool for automation
of dissolution testing [11].
References
[1] USP 23 NF 18, United States Pharmacopeial Convention, Inc.,
Rockville, MD, (1995)
[2] Rohrs B.R., Stelzer D.J., Deaeration Techniques for Dissolution
Media, Dissolution Technologies, 2 (2) 1-9 (1995)
[3] International Conference on Harmonization, Validation of Analytical
Procedures: Methodology, ICH Harmonised Tripartite Guideline Q2B
(1995)
[4] Pharma Switzerland, Quality Assurance and Quality Control,
Validation of Analytical Methods, F. Hoffmann-La Roche Ltd., Basel
(1998)
[5] Roth H.J., Eger K., Troschütz R., Arzneistoffanalyse,
Gustav Fischer, Stuttgart, Jena, Lübeck, Ulm, 3 (1997)
[6] Debesis E., Revised Analytical Data Sheet for the Active Substance*,
Red Corner Research Report No. N-33671, F. Hoffmann-La Roche Ltd.,
Nutley (1979)
Fiber Optic References...continued
[7] Stricker H. (ed.), Physikalische Pharmazie,
Wissenschaftliche Verlagsgesellschaft mbH, Stuttgart, 3 (1987)
[8] Lide D.R. (ed.), Handbook of Chemistry and Physics, CRC Press,
Boca Raton, New York, London, Tokyo, 76 (1996)
[9] Schatz C., Internal Communication about the Benzodiazepine
Tablets, F. Hoffmann-La Roche Ltd., Basel (1999)
[10] Schatz C., Internal Communication about the Soft Gelatin
Capsules, F. Hoffmann-La Roche Ltd., Basel (1999)
[11] Kostek L.J. et al., Automated Dissolution Testing Utilizing
On-line Fiber Optic Probe UV Analysis, ISLAR '95 Proceedings,
355-367 (1995)
Correspondence:
Caspar Schatz
F. Hoffman-LaRoche, Ltd.
POBQ, Building 61/133
CH-4070 Basel Switzerland
Phone +41-61-687 14 33 Fax +41-61-688 8020
caspar.schatz@roche.com