Materials and methods
Materials. Propranolol hydrochloride, US Pharmacopeia tablets (blue color), 20 mg, and rofecoxib (Vioxx) tablets (yellow color), 25 mg, were purchased from Pliva (East Hanover,
NJ), and Merck & Co. (Whitehouse Station, NJ), respectively. DBcaps capsules (DBcaps Size B)
and standard gelatin capsules (HGC size 00) were supplied by Capsugel (Peapack, NJ). Microcrystalline cellulose (Avicel PH102)
was donated by FMC Biopolymer (Philadelphia), and Foremost Farms (Baraboo, WI) supplied anhydrous lactose. A Vankel dissolution
apparatus and a Delphian fiber-optic dissolution-monitoring system from Delphian (Woburn, MA) were used.
Experimental design. Two blocks of 2 × 2 × 2 randomized full-factorial design were used to determine the effect of three independent parameters.
Two levels (i.e., two drug types, propranolol and rofecoxib), two capsule types (i.e., DBcaps and HGC), and two filler levels
(i.e., microcrystalline cellulose–lactose 1:1 mixture and no filler) were used in the design.
A predetermined weight of a 1:1 physical mixture of anhydrous lactose:microcrystalline cellulose was filled into the capsule
body. The tablet was placed on the backfill, and the capsule cap was snapped on to lock in place. When backfill was not required,
only the tablet was filled into the capsule.
Twenty capsules from each batch were weighed, and the average weight, standard deviation, and percent relative standard deviation
(RSD) were computed. To remove bias, the hand filling of the capsules was blinded from the analyst who conducted the weight-variation
experiment. The samples were also coded.
A Vankel disintegration apparatus, running at 30 dips per minute (dpm), and simulated gastric fluid (pH 1.2) without pepsin
were used for the disintegration test. Two randomly selected capsules from a batch were tested at a time. This test was repeated
three times, making a total of six capsules from each batch. The average disintegration (D-time) was computed. Two capsules
were run at a time to allow a video camera (MDS 100, Kodak, Rochester, NY) to capture the process of capsule disintegration.
The Vankel dissolution apparatus and Delphian RAINBOW dynamic dissolution-monitor system, consisting of six photodiode-array
probes with a cell length of 10 mm, were used. Simulated gastric fluid (pH of 1.2) without pepsin, degassed at 40 °C, was
used as the medium at 37 °C.
A spectrophotometric linearity test was performed to test each individual probe for linearity and reproducibility. Using the
blank dissolution medium, the fiber-optic system first acquired the 100% transmittance. This step was followed by calibration,
which involved collecting the transmittance from three replicates of six standard solutions. The percent difference and percent
RSD were calculated, and the linearity was determined.
The dissolution of propranolol was determined using the USP Apparatus I (or basket method) at 50 rpm in 900 mL of simulated gastric fluid. A filtered solution of 20 mg of propranolol
in 900 mL of simulated gastric fluid was used as the standard. Peak absorbance wavelength and baseline correction wavelength
were 288 nm and 370 nm, respectively.
For rofecoxib, the USP Apparatus II (or paddle method) was used at 50 rpm in 900 mL of simulated gastric fluid. A filtered solution of 25 mg of
rofecoxib in 900 mL of simulated gastric fluid was used as the standard. The peak absorbance wavelength and baseline correction
wavelength were 268 nm and 370 nm, respectively. Replicates of eight batches were tested, making a total of 16 runs.
2 similarity factor (see Eq. 1) was used to compare the dissolution profiles. Six and 16 time points were used for propranolol
and rofecoxib, respectively, because of differences between the two drugs' dissolution rates. When comparing the dissolution
profiles of encapsulated and unencapsulated tablets, the data were normalized for a lag time of 2 min (13). The significance
of individual parameters' effects and their influence on disintegration time and dissolution were determined using a least-squares
regression model (p-value < 0.05).