Design Space Integration

1

NIPTE-FDA Collaborative Case StudyOn Model-based Design Space Development

Across Scales & with Stability Considerations

Design Space Integration

2

Design Space Integration

• Considered– Degradation

• NMT 0.4 mole % lactam– Hardness

• NLT 3 kP– Weight variation

• 90 % - 110%• % RSD NMT 6%

– Dissolution• NLT 40% @ 5 min

– Disintegration• NMT 10 min

3

Shelf Life Design Space

Probability of lactam < 0.4 mole % after 2 years @

22°C

Low Moisture max p=0.7Medium Moisture max p=0.71

0

~

a

obs

o

ERT

obs

o

dLk

dtL L

k Ae

A D

4

Tabletting Design Space

20

30

40

50

60

70

Mea

n D

isso

lutio

n 1

5

Leve

rage

Re

sid

uals

0 50 100 150 200 250

Hydrophobicity

Leverage, P=0.0271

20

30

40

50

60

70

Mea

n D

isso

lutio

n 1

5

Leve

rage

Re

sid

uals

0 5 10 15

Mean Tablet Hardness

Leverage, P<.0001

0

5

10

15

20

25

Ave

rag

e di

sin

tegr

atio

n

time

Lev

era

ge

Res

idu

als

50 100 150 200 250

Hydrophobicity

Leverage, P<.0001

0

5

10

15

20

25

Ave

rag

e di

sin

tegr

atio

n

time

Lev

era

ge

Res

idu

als

0 2 4 6 8 10 12 14 16 18

Mean Tablet Hardness

Leverage, P<.0001

disintegration time =-4.54 + 0.86 * hydrophobicity+ 0.06 * hardness

Hydrophobicity Hardness

% dissolved @5 min =78.1 – 0.1 * hydrophobicity– 3.0 * hardness

5

Tabletting Design Space

Disintegration

Dissolution

Hydrophobicity = -26 + 36 * Blend Time

=># of rotations are limited

50

100

150

200

250

Hyd

rop

hobi

city

Act

ual

50 100 150 200 250

Hydrophobicity Predicted

P<.0001 RSq=0.64 RMSE=37.617

6

Dissolution Data (provided by FDA)

0 5 10 15 20 25 30 35 40 45 500

20

40

60

80

100

120

96251-2 #1

96251-3 #1

96251-4

96251-5

96251-6

96251-7

96251-8

96251-9

96251-10

96251-11

96251-12 #1

96251-13

96251-14

96251-15

Time (h)

% d

isso

lved

7

Dissolution

20

30

40

50

60

70M

ean

Diss

olut

ion

15

0 5 10 15 20 25

Average disintegration time

% dissolved @5 min = 63.3 – 2.4 * disintegration time

Disintegration is often limiting step for highly soluble drugs

8

Single Unit Op

9

Unit Op Interactions

10

Design SpaceSingle Unit Op

11

Unit Op Interactions

12

Design Space Integration

• Tablet weight variation = f (spray rate, impeller speed)– within USP limit

• In-process lactam is well below 0.4 mole %– a meaningful limit may be established only after

accelerated stability studies

13

Model Building

20 25 30 35 40 45 500.01

0.015

0.02

0.025

0.0300000000000001

f(x) = 0.000330905350366283 x + 0.00668066973408026R² = 0.975504874980725

Temperature (C)

In-p

roce

ss la

ctam

(mo

le %

)

In-process lactam ~ max. product temperatureLab scale

14

Model Validation (Lab Scale)

0 242.7 485.4 728.1 970.8 1213.5 1456.2 1698.9 1941.6 2184.3 2427

16.4728

24.7093

32.9457

41.1821

49.4185

57.6549

65.8914

74.1278

82.3642

Fluidized bed drying

Time (s)

Tem

pera

ture

(C

)

Inlet air temperature

Inlet air temperature, smoothed

Granule temperature

Predicted granule temperature

15

Model Prediction

EEF EMT0.175 0.50.175 10.45 1

0.45 0.5

No need to run full factorial!!!

Single experiment to calibrate model parametersSecond experiment to verify model predictions

16

Model Prediction (Intermediate Scale)

0 250 500 750 1000 1250 1500 1750 2000 2250 2500

10.1

20.2

30.3

40.4

50.5

60.6

70.7

80.8

90.9

101

Fluidized bed drying

Time (s)

Tem

pera

ture

(C

)

Inlet air temperature

Predicted granule temperature

17

Summary

• Design space for shelf life stability determined• Unit op interactions determined• Design space for each unit op can be determined by

backward propagation• Models help to predict process behavior at larger

scale => save on experimental effort