1 Measuring and Managing Method Variability Lucinda (Cindy) Buhse, PhD Director, Division of Pharmaceutical Analysis Advisory Committee for Pharmaceutical.

1

Measuring and Managing Method Variability

Lucinda (Cindy) Buhse, PhDDirector, Division of Pharmaceutical Analysis

Advisory Committee for Pharmaceutical Science (ACPS)October 25, 2005

2

FDA Proposal to Advisory Committee in May

Alternative approach to dissolution calibration and validation:

• Perform stringent mechanical calibration to replace the need for a calibrator tablet

• ID and control all sources of variability– Apparatus type including sinkers– Set-up,calibration and operational parameters– Media including degassing

• Understand interaction between instrument and product during pharmaceutical development

• If necessary, establish an internal calibrator (bio-batch or clinical batch) for system suitability and stability

• Confirm suitability using Gauge R&R using pivotal clinical trial product or pivotal “bio-batch”

3

Advisory Committee Response May 3, 2005

DR. COONEY: So, the request is to embrace the content of slides…? We have 11 yes, zero no and zero abstentions for the vote. Wonderful!

4

Gauge R&R Design

• Chance to characterize variability on clinical or bio-batch – Sample from throughout manufacturing process to

demonstrate control– Benchmark variability prior to establishing

specifications– Establish internal reference

• For design include variables such as– Location (beginning, middle and end of lot)– Instrument– Operator

5

DPA gauge R&R study

• Product: NCDA#2 10mg Prednisone Tablet• Stable• History at DPA – known to be sensitive to degassing and

mechanical calibration tolerances

• Design of Study:• 2 operators • 2 mechanically calibrated USP Apparatus (Paddle)• Nested study design, N= 2x2x6• 6 replications for each operator on each apparatus

6

Operator 1 2

Apparatus A B A B

Vessel 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6

6 repetitions

DPA Gauge R&R Study

7

Comparison of % Dissolved at 30 Minutes for Two Apparatuses

Based on ANOVA results, no significant differences were observed between two USP apparatus 2 in testing NCDA#2 tablets.

#A-B

%D

iss@

30

27

2829

30

3132

33

3435

36

3738

39

#A #B

Apparatus

#A-#B

Assuming unequal variances

Difference

Std Err Dif

Upper CL Dif

Lower CL Dif

Confidence

0.41356

0.25750

0.92264

-0.09551

0.95

t Ratio

DF

Prob > |t|

Prob > t

Prob < t

1.606076

140.5884

0.1105

0.0553

0.9447 -1.0 -0.5 .0 .5 1.0

t Test

#A

#B

Level

72

72

Number

32.2868

31.8732

Mean

1.62056

1.46555

Std Dev

0.19098

0.17272

Std Err Mean

31.906

31.529

Lower 95%

32.668

32.218

Upper 95%

Means and Std Deviations

%D

iss0

@30

A B

8

Comparison of Variance Components

0

10

20

30

40

50

60

70

80

% o

f To

tal V

ari

an

ce

OperatorTablet Vessel

Variance Components

Results• Tablet is the main contribution to the variance.• Variability component from vessels for Apparatus A is larger than for Apparatus B.• Operator contributes minimally to variability for DPA.

Apparatus A

Apparatus B

9

30 Minute Dissolution for Individual Vessels for Apparatus A

#A-%

Dis

s@30

27

2829

30

3132

33

3435

36

3738

39

1 2 3 4 5 6

#A-vessels

%D

iss0

@30

Vessel

•Some vessels are above average and some below.

•Vessels were moved to different positions and trends above and below average were found to follow the vessel and not the shaft position.

•Although apparatus was level and shafts were vertical, vessels were found to be not vertical because of unevenness around their rims.

10

Vessel Centering and Vessel Verticality affect Hydrodynamics

Mechanical Calibration Status

% Dissolution at 30 min.

Avg. SD

Vessels Centered (Avg. of 30) 29.2 1.0

Vessels Offset 1 mm (Avg. of 12) 30.6 1.8

Vessels Offset 2 mm (Avg. of 12)* 35.6 3.9

Vessels Tilted 1.2º (Avg. of 12) 31.4 1.2

Vessels Tilted 2.3º (Avg. of 12) 36.3 2.6* Current USP Mechanical Calibration Tolerance

• 10 mg Prednisone Tablets, Lot PRED96-21, paddle, 50 RPM, degassed water• Data Range: 26.3 – 44.5%: so these effects would not be detected by a

calibrator tablet with assigned values similar to the current lot.

11

Solution: Two-Point Method for Vessel Centering and Vessel Verticality

12

Comparison of Variance Components after “2 Points Centering Check” for Apparatus A

Y

Apparatus #A

Apparatus #B

1pt centering

2pts centering

0

10

20

30

40

50

60

70

80

90

100

% o

f To

tal V

ari

an

ce

Variance Components

OperatorVesselTablet

2 point centering to ensure verticality of the vessels eliminated the vessel component of the variability.

13

Dissolution Testing Good Practices

• Apparatus Set Up• Vessel Dimensions• Basket Dimensions (Basket Clips)• Paddle Dimensions• Belts and Ball Bearings

• Mechanical Calibration• Shaft Wobble • Vessel Centering / Vessel Verticality• Basket and Paddle Depth• Paddle and Basket Shaft Verticality• Rotational Speed• Basket Wobble

• Operation• Basket Examination• Paddle Examination• Vessel Temperature• Vibration• Sinkers

14

Apparatus Set Up

Vessel Dimensions

Per USP illustrations and dimensionsBasket Dimensions

(including use of Basket Clips)

Paddle Dimensions

Belts and Ball Bearings None Belts are tight, clean, free of cracks,

properly aligned, shafts turn freely

USP FDA

15

Mechanical Calibration Tolerances

Shaft Wobble No significant wobble

≤ 0.5 mm total run out

Vessel/Shaft Centering and Vessel Verticality

2 mm from centerline

1 mm from centerline at 2 points

Height check/Basket or Paddle Depth as measured at Basket or Paddle bottom

25 + 2 mm 25 + 2 mm

Shaft Verticality None Centered in bubble level at 2 pts 90 apart

Rotational speed + 4% + 2 rpm

Basket Wobble (Bottom Rim) + 1mm + 0.5mm

(≤ 1.0mm total)

USP FDA

FDA Tolerances consistent with specs or performance of Varian, Distek, Erweka and Hanson instruments. PhRMA Dissolution Expert Team participated in setting tolerances.

16

Operation

Basket and Paddle Examination

None No gross defects as determined by visual

inspection

Vessel Temperature 37 0.5 °C 37 0.5 °C

Vibration No significant vibration

No significant vibration (pending further studies)

Sinkers A few turns of wire helix may be attached to dosage unit that would otherwise float. Other validated sinker devices may be used.

Recommend USP sinker or ensure commercial sinker does not affect results

USP FDA

17

Vibration

1. 1999 Collaborative Study: Displacement

2. 1998 Japanese Study: Acceleration

Apparatus 1, 50 rpm, Enteric Coated product Average of 5 – 8 apparatus

Low Vibration Acceleration < 0.01 m/s2 32 1

High Vibration: Acceleration > 0.09 m/s2 39 4

3. 2005 Study by Bryan Crist and Dan Spisak, Varian Inc. : Frequency

Apparatus 2, 50 rpm 10 mg Prednisone USP Lot #O0C056 Average 6

Benchmark 34 4

Vibration Displacement < 0.20 mils Frequency at 20 Hz 34 2

Vibration Displacement < 0.02 mils Frequency at 130 Hz 42 9

18

Hydrodynamics

Challenges• Paddle method is operated at flow conditions between

laminar and turbulent which makes modeling difficult and shear stress distribution is non-uniform at base of vessel.

• Degree of mixing with basket method is limited leading to solute stratification in the vessels, and the dosage form remaining in the basket is subjected to different shear stress than the fragments that settle at bottom of vessel.

• Hydrodynamic variables that are important to a calibrator tablet may not be important to tested drug product.

Statements based on information from Dr. Armenante, Dr. Muzzio and Dr.Kakhi

19

Benefits of Mechanical Calibration and Gauge R&R

• The sources of variability in the dissolution measurement system can be identified and minimized.

• If done during clinical or bio-batch lot, knowledge of variability can assure development of meaningful specifications.

• An internal calibrator can be developed from clinical or bio-batch which is more applicable to specific product than the USP calibrator tablets.

• This approach provides a higher assurance of quality than the current system where failure results can be caused by product failure OR measurement system variability.

20

Next Steps

• Additional Collaborative Research– Hydrodynamics– Vibration– New approaches to assess drug release (PAT,

spectroscopy, first principles and modeling)• Training of FDA Labs• Compliance Policy Guide• Future Guidance