Download - Statcon

Is there a Sticky Sweet Spot?
Mixture Design of a Pressure Sensitive Adhesive Emulsion Formulation
1
M. Michaelis, C. S. Leopold
Department of Chemistry
Division of Pharmaceutical Technology
University of Hamburg, Germany

Aim of the study
Formulation of an adhesive mixture for transdermal patches that
comprises:
1. Improved adhesion properties
2. Improved solubility of an Active Pharmaceutical Ingredient (API)
2

Introduction
Transdermal Therapeutic Systems (TTS)
Patches for transdermal application of
APIs to achieve a systemic effect
About 50 TTS products on the market
14 Active Pharmaceutical Ingredients
Mostly Drug In Adhesive Design (DIA)
+ Controlled drug delivery
+ Good compliance
API level in blood
http://www.nipro-patch.co.jp
- Stability issues
- Lack of adhesion
3
http://www.pharmainfo.net
Crystalization of API
http://www.acino-pharma.com
Drug In Adhesive Design
http://www.novartis.com.ph
Backing
Drug in Adhesive
Release liner
http://www.quit.org.au
Lack of adhesion

Introduction
Pressure Sensitive Adhesives (PSAs)
Pressure Sensitive Adhesives:
Tacky at room temperature
Adhere to a variety of surfaces on light pressure
Adhere permanently
Ûsed in a lot of common products
Adhesive Performance
Tack: Ability to form a bond of measurable strength by
simple contact with a surface (stickiness)
Shear Adhesion: Ability to resist structural failure
(cohesiveness)
Peel Resistance: Force required to remove the tape without
leaving residues
4

Materials
Component 1: Polyacrylate
DuroTak® 387-2287
2-Ethylhexyl acrylate
Vinyl acetate
Hydroxyethyl acrylate
Component 2: Silicone Adhesive
BIO-PSA® 7-4302
Component 3: Oleyl Alcohol
Surfactant
Component 4: Ibuprofen
API / model drug
(high dose analgetic)
5
Oleyl Alcohol
67 %
28 %
5 %
Ibuprofen

Design
Mixture Design
3 (4) Components
Components
Design space with constraints IV Optimal Design
Suggested design: 16 runs, 5 replicates, 5 to estimate lack of fit
Evaluation: FDS Graph
Low
[%]
High
[%]
Polyacrylate Adhesive 20 70
Silicone Adhesive 10 60
Oleyl Alcohol 0 10
Ibuprofen 20
Total 100
6
= Design Space
= Replicate x 2
= Replicate x 3

Preparation of adhesive matrix/specimen
1. Components were dissolved in ethyl acetate (wet mix) and mixed in a shaker at
90 rpm for 15 min.
2. Wet mix was coated on a release liner
3. Films were dried at 80 °C for 30 min in an oven Adhesive matrix (a) - 100 µm thick
4. Films were laminated with a backing membrane Specimen (b)
a) Coating Knife
Release Liner
Drug In Adhesive
7
b)
Backing Membrane
Release Liner
Drug In Adhesive

Response 1 & 2:
1. Tack
Probe Tack Test at 21 °C
Adhesive matrix (a)
Stainless steel probe 3 mm
Contact time 1 s, contact force 0.4 N
Response: Stress maximum max [N/mm²]
2. Shear Adhesion
Test specimen (b) was attached to
a stainless steel plate
Area: 12 x 12 mm, weight: 250 g
Response: Time to failure t [min]
8
Probe Tack Test
Shear adhesion

Response 3 & 4:
3. Crystal growth
Area of 100 cm 2 of the adhesive matrix
after 24 h
Response: Area covered by crystals in %
of the whole area
4. Creaming behavior (phase separation)
Bottle with wet mix after 24 h
Separated phase of wet mix in % of the
wet mix
9
1:1 1:100
no creaming
20 % creaming

Response 5 & 6
5. Droplet size
Matrix was transferred to a glass slide
Microscope with 100-fold magnification
Response: Mean of droplet diameter of 30 droplets [µm]
6. Droplet distribution range
Response: Maximum range of 30 droplets [µm]
small & narrow large & narrow small & broad large & broad
10

Analysis
Model
Normal Plot
Transformation
Fit Summary and
Model
Lack of
Fit
Residuals
vs.
Predicted
ANOVA
Diagnostics
11
adj.
R-square
Externally
Studentized
Residuals
Model Graphs
pred.
R-square
Box-Cox
Plot

Response Analysis 1: Tack
Fit Summary and Model
ANOVA
Highest order polynomial: Reduced Special Quartic Model
Model: 0.0001 significant
Lack of fit: 0.3626 not significant
Adj. R-square: 0.9226
Pred. R-square: 0.8641
Diagnostics
Normal Plot
Residuals vs. Predicted
Externally Studentized Residuals
Box-Cox plot
Model Graphs
12

Normal % Probability
Normal % Probability
Response 1: Tack - Diagnostics
99
95
90
80
70
50
30
20
10
5
1
Normal
Normal
Plot
Plot
of
of Residuals
Residuals
-2.00 -1.00 0.00 1.00 2.00
-2.00 -1.00 0.00 1.00 2.00
Internally Studentized Residuals
Internally Studentized Residuals
13
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
Residuals
Residuals
vs.
vs. Predicted
Predicted
-2.00 1.00 -1.00 1.50 0.00 2.00 1.00 2.50 2.00 3.00
Predicted
Predicted

Externally Studentized Residuals
Externally Studentized Residuals
Response 1: Tack - Diagnostics
6.00
4.00
2.00
0.00
-2.00
-4.00
-6.00
Externally Externally Studentized Residuals Residuals
1 4 7 10 13 16
1 4 7 10 13 16
Run Number
Run Number
14
Ln(ResidualSS)
Ln (ResidualSS)
-1.00
-1.20
-1.40
-1.60
-1.80
Box-Cox-Plot Box-Cox Plot for for Power Power Transforms Transforms
-3 -2 -1 0 1 2 3
Lambda
Lambda

2
Response 1: Tack - Model Graphs
100 % Polyacrylate = 0.50 N/mm²
100 % Silicone = 0.55 N/mm²
0.4
2
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
0.3
3
0.2
2
0.2
20.000
Tack
Design points below predicted value
50.000
C: Oleyl Alcohol
15
Tack
0.5
0.4
0.3
0.2
0.1
C (0.000)
A (70.000)
B (10.000)
B (60.000) A (20.000)
C (50.000)

Normal % Probability
Normal % Probability
99
95
90
80
70
50
30
20
10
5
1
Response 2: Shear Adhesion - Diagnostics
Fit Summary and Model
ANOVA
Highest order polynomial: Quadratic Model
Model: 0.0001 significant
Lack of fit: 0.6446 not significant
Adj. R-square: 0.99875
Pred. R-square: 0.9734
Diagnostics
Normal Plot of Residuals
-2.00 -1.00 0.00 1.00 2.00
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
Residuals vs. Predicted
0.00 0.20 0.40 0.60 0.80 1.00 1.20 1.40 1.60
Predicted
16
Externally Studentized Residuals
Externally Studentized Residuals
4.00
2.00
0.00
-2.00
-4.00
Externally Studentized Residuals
1 4 7 10 13 16
Run Number
Ln(ResidualSS)
Ln (ResidualSS)
10.00
8.00
6.00
4.00
2.00
0.00
Recommended
transformation:
Log
Normal Plot of Residuals Residuals vs. Predicted Externally Studentized Residuals Box-Cox Plot for Power Transforms
Box-Cox Plot for Power Transforms
-3 -2 -1 0 1 2 3
Predicted Run Number Lambda
Lambda

2
Response 2: Shear Adhesion - Model Graphs
100 % Polyacrylate = 17.2min
100 % Silicone = 56.9 min
20
2
15
60.000
B: Silicone
5
3
A: Acrylat
70.000
2
Design-Expert® Software
Component Coding: Actual
Original Scale
(median estimates)
Shear
Design points above predicted value
Design points below predicted value
30.2
1.4
0.000 10
10.000
20.000
Shear Shear Adhesion
X1 = A: Acrylat
X2 = B: Silicone
X3 = C: Oleyl Alcohol
50.000
C: Oleyl Alcohol
17
Shear Shear Adhesion
30.2
23
15.8
8.6
1.4
C (0.000)
A (70.000)
B (10.000)
B (60.000) A (20.000)
C (50.000)

Normal % Probability
Normal % Probability
99
95
90
80
70
50
30
20
10
5
1
Response 3: Crystal Growth - Diagnostics
Fit Summary and Model
ANOVA
Highest order polynomial: Reduced Cubic Model
Model: 0.0001 significant
Lack of fit: 0.7984 not significant
Adj. R-square: 0.9819
Pred. R-square: 0.9702
Diagnostics
Normal Plot of Residuals Residuals vs. Predicted Externally Studentized Residuals Box-Cox Plot for Power Transforms
Normal Plot of Residuals
-3.00 -2.00 -1.00 0.00 1.00 2.00
Internally studentized residuals
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
Residuals vs. Predicted
2 2
0.00 20.00 40.00 60.00 80.00 100.00
Predicted
18
Externally Studentized Residuals
Externally Studentized Residuals
6.00
4.00
2.00
0.00
-2.00
-4.00
-6.00
Externally Studentized Residuals
1 4 7 10 13 16
Run Number
Ln(ResidualSS)
Ln (ResidualSS)
30.00
25.00
20.00
15.00
10.00
5.00
0.00
Box-Cox Plot for Power Transforms
-3 -2 -1 0 1 2 3
Predicted Run Number Lambda
Lambda

Response 3: Crystal Growth - Model Graphs
Crystal Growth
100 % Polyacrylate= 70 %
100 % Silicone= 100 %
2
2
60.000
B: Silicone
A: Acrylat
70.000
0.000 20
10.000
100
60
80
40
3
0
2
0
0
20.000
20
Crystal Growth
Design points below predicted value
50.000
C: Oleyl Alcohol
19
Crystal Growth
120
100
80
60
40
20
0
-20
C (0.000)
A (70.000)
B (10.000)
B (60.000) A (20.000)
C (50.000)

Normal % Probability
Normal % Probability
99
95
90
80
70
50
30
20
10
5
1
Response 4: Creaming behavior - Diagnostics
Fit Summary and Model
ANOVA
Highest order polynomial: Reduced Cubic Model
Model: 0.0001 significant
Lack of fit: 0.6410 not significant
Adj. R-square: 0.9460
Pred. R-square: 0.7104
Diagnostics
Normal Plot of Residuals Residuals vs. Predicted Externally Studentized Residuals Box-Cox Plot for Power Transforms
Normal Plot of Residuals
-2.00 -1.00 0.00 1.00 2.00
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
Residuals vs. Predicted
5.00 10.00 15.00 20.00 25.00 30.00
Predicted
20
Externally Studentized Residuals
Externally Studentized Residuals
6.00
4.00
2.00
0.00
-2.00
-4.00
-6.00
Externally Studentized Residuals
1 4 7 10 13 16
Run Number
Ln(ResidualSS)
Ln (ResidualSS)
9.00
8.00
7.00
6.00
5.00
4.00
3.00
Box-Cox Plot for Power Transforms
-3 -2 -1 0 1 2 3
Predicted Run Number Lambda
Lambda

2
Response 4: Creaming Behavior -
Model Graphs
10
2
15
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
20
25
3
25
20
15
2
10
20.000
Creaming Creaming
Design-Expert® Software
Component Coding: Actual
Creaming
Design points above predicted value
Design points below predicted value
27.3743
4.44444
X1 = A: Acrylat
X2 = B: Silicone
X3 = C: Oleyl Alcohol
50.000
C: Oleyl Alcohol
21
Creaming
30
25
20
15
B (60.000)
10
5
0
C (0.000)
A (20.000)
A (70.000)
B (10.000)
C (50.000)

Normal Normal % % Probability
Response 5: Droplet Size - Diagnostics
99
95
90
80
70
50
30
20
10
5
1
Fit Summary and Model
ANOVA
Highest order polynomial: Reduced Cubic Model
Model: 0.0001 significant
Lack of fit: 0.1123 not significant
Adj. R-square: 0.9733
Pred. R-square: 0.7747
Diagnostics
Normal Plot of Residuals Residuals vs. Predicted Externally Studentized Residuals Box-Cox Plot for Power Transforms
Normal Plot of Residuals
-3.00 -2.00 -1.00 0.00 1.00 2.00 3.00
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
2
Residuals vs. Predicted
0.00 10.00 20.00 30.00 40.00
Predicted
2
22
Externally Studentized Residuals
Externally Studentized Residuals
6.00
4.00
2.00
0.00
-2.00
-4.00
-6.00
Externally Studentized Residuals
1 4 7 10 13 16
Run Number
Ln(ResidualSS)
Ln (ResidualSS)
8.00
7.00
6.00
5.00
4.00
3.00
2.00
Box-Cox Plot for Power Transforms
-3 -2 -1 0 1 2 3
Predicted Run Number Lambda
Lambda

2
Response 5: Droplet Size - Model Graphs
10
2
40
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
30
20
3
10
2
20.000
Droplet Size
Design-Expert® Software
Component Coding: Actual
Droplet Size
Design points above predicted value
Design points below predicted value
40
3.75
X1 = A: Acrylat
X2 = B: Silicone
X3 = C: Oleyl Alcohol
50.000
C: Oleyl Alcohol
23
Droplet Size
Droplet Size
50
B (60.000)
40
30
20
10
0
C (0.000)
A (20.000)
A (70.000)
B (10.000)
C (50.000)

Normal % Probability
Normal % Probability
99
95
90
80
70
50
30
20
10
5
1
Response 6: Droplet Distribution Range -
Diagnostics
Fit Summary and Model
ANOVA
Highest order polynomial: Reduced Cubic Model
Model: 0.0001 significant
Lack of fit: 0.2831 not significant
Adj. R-square: 0.9230
Pred. R-square: 0.7410
Diagnostics
Normal Plot of Residuals
-2.00 -1.00 0.00 1.00 2.00
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
Internally Studentized Residuals
3.00
2.00
1.00
0.00
-1.00
-2.00
-3.00
Residuals vs. Predicted
0.00 2.00 4.00 6.00 8.00 10.00
Predicted
2
24
Externally Studentized Residuals
Externally Studentized Residuals
6.00
4.00
2.00
0.00
-2.00
-4.00
-6.00
Externally Studentized Residuals
1 4 7 10 13 16
Run Number
Ln(ResidualSS)
Ln (ResidualSS)
12.00
10.00
8.00
6.00
4.00
Recommended
Transformation:
Square Root
Normal Plot of Residuals Residuals vs. Predicted Externally Studentized Residuals Box-Cox Plot for Power Transforms
Box-Cox Plot for Power Transforms
-3 -2 -1 0 1 2 3
Predicted Run Number Lambda
Lambda

2
Response 6: Droplet Distribution Range -
Model Graphs
2
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
60
40
20
3
2
20.000
Droplet Droplet Distribution Distribution Range
Design points above predicted value
Design points below predicted value
50.000
C: Oleyl Alcohol
25
Droplet Distribution
Droplet Distribution Range
80
60
B (60.000)
40
20
0
C (0.000)
A (20.000)
A (70.000)
B (10.000)
C (50.000)

2
2
60.000
B: Silicone
Discussion: Response 3 & 5 & 6
A: Acrylat
70.000
0.000 20
10.000
100
40
60
80
3
0
2
0
0
20.000
20
Crystal Growth
50.000
C: Oleyl Alcohol
2
10
2
40
60.000
B: Silicone
A: Acrylat
70.000
10
0.000 10.000
30
20
3
2
20.000
Droplet Size
26
50.000
C: Oleyl Alcohol
2
2
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
60
40
20
3
2
20.000
Droplet Distribution
50.000
C: Oleyl Alcohol
Crystal Growth Droplet Size Droplet Distribution Range

Optimization
Constrains
Solutions
Lower Upper Lower Upper
Name Goal Limit Limit Weight Weight Importance
A:Acrylate is in range 20 70 1 1 3
B:Silicone is in range 10 60 1 1 3
C:Oleyl Alcohol is in range 0 10 1 1 3
Tack maximize 0.3 0.41 1 1 3
Shear Adhesion maximize 5.0 30.2 1 1 3
Crystal Growth is target = 0 0 10.0 1 1 5
Droplet Size minimize 3.75 40 1 1 3
Droplet Distribution minimize 2.5 70 1 1 3
Number Acrylate Silicone Oleyl Alcohol Tack
27
Shear
Adhesion
Crystal
Growth
Droplet
Size
Droplet
Distribution Desirability
1 20.2 59.8 0.0 0.41 26.19 4.81 7.3 5.0 0.750
2 45.0 35.0 0.0 0.31 12.89 0.53 10.9 4.4 0.540

Optimization
2
2
0.600
60.000
B: Silicone
A: Acrylat
70.000
0.000 10.000
0.400
0.600
0.200
Prediction 0.750
0.400
3
2
Prediction 0.540
20.000
Desirability
28
50.000
C: Oleyl Alcohol

Conclusion
Is there a sticky sweet spot?
No, there is not.
But…
29

Conclusion
We know that:
Oleyl alcohol neither stabilizes the polymer-polymer interaction nor the
polymer-API interaction.
Oleyl alcohol decreases the adhesion properties in most cases.
Wet mixes with equal amounts of polymer tend to be less stable.
The best results were found at the periphery of the design space.
Crystal growth correlates with droplet size and droplet distribution.
No profit of oleyl alcohol, the addition is unnecessary.
Limited processing time must be taken into account for scale up.
Design space needs to be augmented.
Do process variables such as mixing speed or viscosity of the wet mix have
an influence on crystal growth? (combined – design)
30

Thank you!
31