3763s1_15_raju

7/27/2019 3763s1_15_raju

1/56

MIT PHARMACEUTICAL MANUFACTURING INITIATIVE (PHARMI)

CONTINUOUS QUALITYVERIFICATION(CQV)

G.K.Raju, Ph.D.Pharmaceutical Manufacturing Initiative (PHARMI),

MIT Program on the Pharmaceutical Industry,Massachusetts Institute of Technology

July 2001

7/27/2019 3763s1_15_raju

2/56


Objective:To Describe the Opportunity to Improve

Pharmaceutical Manufacturing Performance

MIT Pharmaceutical Manufacturing Initiative

Research Development Manufacturing Marketing

7/27/2019 3763s1_15_raju

3/56


Pharmaceutical Manufacturing

Research Development Manufacturing Marketing

Bulk

Active

Bulk

Formulation

Filling &

FinishPackaging

Inbound

Logistics

Outbound

Logistics

7/27/2019 3763s1_15_raju

4/56


HORIZONTAL

APPROACH

STUDYING PHARMACEUTICAL MFG:

VERTICAL VS. HORIZONTAL APPROACH

Bulk

Active

Bulk

Formulation

Filling/

Tableting

Packaging/

Finishing

Bulk

Active

Bulk

Formulation

Filling/

Tableting

Packaging/

Finishing

Plant A

Bulk

Active

Bulk

Formulation

Filling/

Tableting

Packaging/

Finishing

Bulk

Active

Bulk

Formulation

Filling/

Tableting

Packaging/

Finishing

Plant B

Plant A

Plant B

VERTICAL

APPROACH

7/27/2019 3763s1_15_raju

5/56


Bulk

Active

Bulk

Formulation

Filling/

Tableting/

etc.

Packaging/

Finishing

Bulk

Active

Bulk

Formulation

Filling/

Tableting/

etc.

Packaging/

Finishing

Company A

Company B

Company CBulk

Active

Bulk

Formulation

Filling/

Tableting/

etc.

Packaging/

Finishing

THE HORIZONTAL APPROACH

PHARMACEUTICAL MANUFACTURING

7/27/2019 3763s1_15_raju

6/56


DESCRIBING THE OPPORTUNITYIN

ROUTINE MANUFACTURING

CONTINUOUS QUALITY VERIFICATION (CQV)

7/27/2019 3763s1_15_raju

7/56MIT PHARMACEUTICAL MANUFACTURING INITIATIVE (PHARMI)

WHICH PROCESSES?

PROCESS CYCLE TIMES

In ProcessDevelopment

In RoutineManufacturing

Biggest Impact Here?Time-to-marketEnablingPotent Products

Place forValidation?Potent ProductsDifficult ProcessesHigh Volume ProductsProducts with Tough QC TestsGeneric Competition

7/27/2019 3763s1_15_raju


PROCESS A WITH QC TESTS

WEIGHING WET

GRANULATION

STEP FB DRY

STEP

BLEND

ENCAPSULATESIEVE

API

MICRO

Particle Size Description

IDAssay

CU

Impurity

Dissolution

MICRO

QC1 QC3 QC4

DRY MIX

QC2

LOD

7/27/2019 3763s1_15_raju


PROCESS A WITH CYCLE TIMES

WEIGH WETGRANULN

ProcessingFB DRY

STEP

BLEND

ENCAPSULATE

SIEVE

API

MICRO

Particle

Size

Description

ID

AssayCU

Impurity

Dissolution

MICRO

QC1QC3

QC4

DRY

MIX

QC2

LOD

7 DAYS 13 DAYS

< 3 DAYS

7/27/2019 3763s1_15_raju


PROCESS B WITH QC TESTS

CHEMICAL

WEIGHING

BLEND FILL CAPSULES BOTTLE

PACKAGING

API

OVI

Description

IDAssay

CU

Impurity

Dissolution

QC1 QC2

7/27/2019 3763s1_15_raju


PROCESS B WITH CYCLE TIMES

CHEMICAL

WEIGHING

BLEND FILL

CAPSULES

BOTTLE

PACKAGING

APIOVI DescriptionID

Assay

CU

Impurity

Dissolution

QC1 QC27 DAYS14 DAYS

17 DAYS

7/27/2019 3763s1_15_raju


PROCESS C WITH QC TESTS

WEIGHING

GRANULATION

FB DRY

STEP

BLEND

FILM

COATINGCOMPRESS

BOTTLE

PACKAGING

API Particle Size

LOD

Description

IDAssay

CU

Impurity

Dissolution

QC1 QC2 QC2

7/27/2019 3763s1_15_raju


PROCESS C WITH CYCLE TIMES

WEIGHING

GRANULATION

FB DRY

STEP

BLEND FILMCOATING

COMPRESS

BOTTLE

PACKAGING

API Description

IDAssay

CU

Impurity

Dissolution

QC1 QC2

21 DAYS

6 DAYS 14 DAYS

7/27/2019 3763s1_15_raju


PROCESS D WITH QC TESTS

CHEMICAL

WEIGHING

GRANULATION

Processing

STEP

BLEND 1: BLEND 2:

PRE-BLEND

FINAL

BLEND

FILM

COATING

COMPRESS BOTTLE

PACKAGING

API Particle Size

LOD

Description

IDAssay

CU

Impurity

Dissolution

QC1 QC2 QC3

7/27/2019 3763s1_15_raju


PROCESS D WITH CYCLE TIMES

QC1

BLEND 2:

PRE-BLEND

CHEMICAL

WEIGHING

GRANULATION

PROCESSING

STEP

BLEND 1: FINAL

BLEND

COMPRESS

FILM

COATING

BOTTLE

PACKAGING

QC2 QC3

15 DAYS10 DAYS

20 DAYS 15 DAYS

60 DAYS

Description

ID

Assay

CU

Impurity

Dissolution

Particle Size

LOD

API

7/27/2019 3763s1_15_raju


PROCESS D WITH QC TESTS:

Cycle Times includingBULK ACTIVE

QC1

BLEND 2:

PRE-BLEND

CHEMICAL

WEIGHING

GRANULATIONSTEP

BLEND 1:FINAL

BLEND

COMPRESS

FILMCOATING

BOTTLE

PACKAGING

QC2QC3

15 DAYS10 DAYS

20 DAYS 15 DAYS

60 DAYS21-90 DAYS

PROCESSING

7/27/2019 3763s1_15_raju


PROCESS D WITH QC TESTS

Cycle Times

QC1

BLEND 2:

PRE-BLEND

CHEMICAL

WEIGHING

GRANULATION

PROCESSING

STEP

BLEND 1: FINAL

BLEND

COMPRESS

FILM

COATING

BOTTLE

PACKAGING

QC2QC3

15 DAYS10 DAYS

20 DAYS 15 DAYS

60 DAYS

0

5

10

15

20

QC1 PFD QC3 Release

Actual

Target

Potential

7/27/2019 3763s1_15_raju


WHAT DRIVES THE

QC TESTING TIMES?

0

5

10

15

20

QC1 PFD QC3 Release

Actual

Target

Potential

SamplingBatching

Other Products

Waiting

Coordinating

Other ProductsOther Paperwork

Waiting

Coordinating

2%

TEST

7/27/2019 3763s1_15_raju


WHICH PROCESSES?

PROCESS CYCLE TIMES



Biggest Impact Here?Time-to-market

EnablingPotent Products


7/27/2019 3763s1_15_raju


PROCESS E WITH QC TEST POINTS

WEIGH DRY MIX WET

GRANULN

WET GRANULN FL BED DRY MILL

MILL

STOREDRY SIFTWET GRAN MIX MIX

COAT

STORE

MILL

SIFT

STORE

BLEND BLEND BLEND BLEND STORE BOTTLE FILL

DRY

MIX MIX GRANUL DRY MILL

MILL

ACTIVE INITIAL GRANULATION STAGE

SECOND GRANULATION STAGE COATING STAGE

EXCEPIENT PREPARATION STAGE

SIFT&BLEND STAGE BLEND&FILL STAGE

QC4QC3QC2

QC1

LOD

LOD LOD

LOD

7/27/2019 3763s1_15_raju


PROCESS E WITH QC TEST TIMES

ACTIVE

FIRST

GRAN

SECOND

GRAN

QC1

7 days

COATSIFT&

BLEND

BLEND

QC4QC3QC2

FILL

7 days7 days3 days

< 1

day

< 1

day

1-2

days

< 1

day

< 1

day

< 1

day

7/27/2019 3763s1_15_raju


WHICH PROCESSES?

TOWARDS PARAMETRIC RELEASE





7/27/2019 3763s1_15_raju


PROCESS F: LIQUID LINE

WEIGH

pH ADJ COMPOUND

FILTER

FILL STOPPER CAP

LABEL/PKG

TERMINAL

STERILIZATION

WEIGH

WEIGH

VIALS

STOPPERSWFI

BUFFER

WASH

WASH AUTOCLAVE

DEPYROGEN

SEALS

Appearance

ID

Assay

Impurity

Fill Vol, Osmolarity, Partic.

Endotoxin

STERILITY TESTING

ENVIRO. MONITORING

BIOBURDEN

WFI TESTINGEndotoxin

TOC

ID

ID

pH Visual CheckWt Check

QC Check

QC CheckQC Check

7/27/2019 3763s1_15_raju

24/56


PROCESS F:

LIQUID LINE WITH CYCLE TIMES

10 days

7 daysENVIRONMENTAL MONITORING

BIOBURDEN TESTING

17-20 days

7 days

3-4 days

STERILITY TESTING

WFI TESTING

3-4 months

7/27/2019 3763s1_15_raju

25/56


PERCEIVED PROCESS CYCLE TIMES:SUMMARY

Process Process MICRO Optimized Process QC QC:ProcessFlow Type Test? Process? Cycle Time Cycle Time Time

A High Vol. Y Y 3 20 6.7

B High Vol. N N 17 21 1.2

C High Vol. N N 21 20 1.0

D High Vol. N N 20 25 1.3E Variable N Y 7 24 3.4

F Parentals Y N 3 20 6.7

7/27/2019 3763s1_15_raju

26/56


CYCLE TIME COMPONENTS

STEPS IN THE PROCESS/PLANT IN QC/QAProcess/Unit operation

Interruption of the process

Securing of sample from process

Holding of sample in plant

Documentation and verification of sampling

Transferring of samples to QC lab

Batching of samples in QC

Preparation of test samples

Actual test - separation

Actual test - measurement

Test data collection and processing

Documentation and verification of testing

Transferring of results for review

Decision regarding impact on process

Process/Process Step Primarily Manual Operation

Inventory Hold Actual test

7/27/2019 3763s1_15_raju

27/56


ON-LINE TECHNOLOGY IMPACTS

DOMINANT CYCLE TIMES

STEPS IN THE PROCESS/PLANT IN QC/QA

Process/Unit operation

Interruption of the process

Securing of sample from process

Holding of sample in plant

Documentation and verification of samplingTransferring of samples to QC lab

Batching of samples in QC

Preparation of test samples

Actual test - separation

Actual test - measurement

Test data collection and processing

Documentation and verification of testing

Transferring of results for review

Decision regarding impact on process

Process/Process Step Primarily Manual Operation

Inventory Hold Actual test

On-line LIF, NIR, Pattern Recognition, etc.

7/27/2019 3763s1_15_raju

28/56


CQV OPPORTUNITY IN ROUTINE MANUFACTURING

Quality Monitoring is Discontinuous

QC testing times are approximately = 1 month

Factor of 20-25 opportunity in cycle time: Process

Factor of 20-25 opportunity in cycle time: QC

QC Cycle Times >= Process Cycle Times

Time is driven by off-line natureof test Exception is MICRO test

SUMMARY

7/27/2019 3763s1_15_raju

29/56


DESCRIBING THE OPPORTUNITY

IN PROCESS DEVELOPMENT


7/27/2019 3763s1_15_raju

30/56


Bulk

Active

Bulk

FormulationFilling &

Finish Packaging

BulkActive

BulkFormulation

Filling &

Finish Packaging

Bulk

Active

Bulk


Finish Packaging

Company A

Company B

Company C

THE VERTICAL APPROACH

Blending

Drying

Granulation

Fermentation Rapid

Microbial

Detection

Flow

Tableting

Transport

7/27/2019 3763s1_15_raju

31/56


Bulk

Active

Bulk


Finish Packaging

Bulk

Active

Bulk


Finish Packaging

Bulk

Active

Bulk

Formulation Filling &Finish Packaging

Company A

Company B

Company C

VERTICAL ANALYSIS I:

BLENDING UNIT OPERATION

Eg. Blending

7/27/2019 3763s1_15_raju

32/56


WHICH PROCESSES?

PROCESS CYCLE TIMES





7/27/2019 3763s1_15_raju

33/56


FOCUS

Explore the Potential Impact of On-lineMonitoring Technology on Blending

Process Development

MIT Pharmaceutical Manufacturing Initiative

7/27/2019 3763s1_15_raju

34/56


Blending Operation Model

Blender

cleaning

Mixing

Active

ingredientExcipients

Raw material load Sampling

Transporting

Homogeneity test

OK?

Results & Decision

Making

Undermixed

mix-longer

Homogeneous

Next batch

Discarded

Next batch

Analysis

7/27/2019 3763s1_15_raju

35/56


LIGHT INDUCED FLUORESCENCE SYSTEM

FOR THE DETERMINATION OF THE

HOMOGENEITY OF DRY POWDER BLENDING

0

500

1000

1500

2000

2500

3000

3500

4000

4500

0 10 20 30 40 50

Number of Rotations

LIFSign

alUnits

A: 5% T/L

B: 5% T/L

C: 5% T/L

PHARMACEUTICAL MANUFACTURING:

LIF FOR ON-LINE MONITORING OF BLENDING

7/27/2019 3763s1_15_raju

36/56


LIF VERIFICATION STUDIES

Established a correlation between LIF assessment ofhomogeneity and thief-sampling with off-line analysis

0

1

2

3

4

5

6

4.75% 3.22% 1.64%

Run Batch

En

dPointDetermination

LIF % Triamterene A

LIF % Triamterene BThief Assay A

Thief Assay B

7/27/2019 3763s1_15_raju

37/56


PROCESS D:

BLENDING PROCESS DEVELOPMENT

CHEMICAL

WEIGHING

GRANULATION

Processing

STEP

BLEND 1: BLEND 2:

PRE-BLEND

FINAL

BLEND

FILM

COATING

COMPRESS BOTTLE

PACKAGING

Mixing Of 1:10 Triamterene-Lactose

@ 70% Fill & 27.4 RPM

0

5

10

15

20

25

30

35

40

0 50 100 150 200 250 300

Number of Rotations

PMTSignals,

Volts

OFF LINE QC TEST

ON LINE SENSOR

7/27/2019 3763s1_15_raju

38/56


Raw Materials Sampling Transport Analysis

Results &Decision making

Reprocessed Discarded Well Blended

Information FlowMaterials FlowR/D/W

Process

knowledgeWaiting Stock

Blending

a- Process Development

b- manufacturing

Blending Operation: Two Technologies, Two ApproachesProcess Development, Val idat ion and Manufactur ing

On-line

Information Feedback

Well Blended

Discarded

Analysis &

Decision making

Blending

Raw Materials

O

FF

LINE

ON

LINE

7/27/2019 3763s1_15_raju

39/56


Blending Data Collected from CAMP Companies

Operation Characteristics

Low Medium High

Cleaning time (min) 20

10

6

250

35

18

480

60

30

60 90 120

6

20

2

30

250

25

60

480

48

Loading time (min)

Discharge time (min)

Sampling time (min)

Transport time (min)

QC Testing time (min)

QC Holding time (min)

7/27/2019 3763s1_15_raju

40/56


Results

Blending PerformanceProcess Development and Validat ion

6% no wait between blends

1 Blend 2 Blends 3 Blends

Best

Med.

Worst

2.32

13.19

25.82

0.36

1.31

2.57

4.96

23.45

43.40

0.68

1.93

3.56

8.45

30.65

56.17

1.07

2.41

4.46Time(days)

7/27/2019 3763s1_15_raju

41/56


Impact of the number of blends on totalprocess time

1.31 1.93

13.19

30.65

23.45

2.410

10

20

30

40

50

60

1 2 3

Number of blends

Time(days)

Avg Off line Avg On line

Blending PerformanceProcess Development and Validation

APPROACH TO LEARNING

7/27/2019 3763s1_15_raju

42/56


APPROACH TO LEARNING:

Current Approach Proposed Approach

xx x

x

x

x

x

x

x

x

x

x

xx

xxxx

x

x

xxx

x

xxx

xxx

xxxxxxxxx

xxx

x

xxxxxxxxxxxxx

xxx

xxx

xxxxxxxxx

xxx

xx

x

Process Development

Commercial Production

1a

1b

2a

2b

Consequences on Process Development & Commercial Production

7/27/2019 3763s1_15_raju

43/56


CQV Opportunity in Process Development

Process development can be on the critical path

Factor of 10-15 reduction in cycle time in blendprocess development (maybe more..)

Variability reduction in blend process dev. time

independence of organization/product -> predictability Benefits not restricted to use of new on-line sensors

improvement data analysis of existing sensor data

use of this for experimental design

SUMMARY

7/27/2019 3763s1_15_raju

44/56


WHAT ARE THEIMPLICATIONS?


C ti f th Ad t f

7/27/2019 3763s1_15_raju

45/56


FDA

Pharmaceutical Companies

CAMP

Purdue MIT

Abbott

Aventis

Bristol-Myers Squibb

Johnson & Johnson

Hoffmann-La Roche

Glaxo SmithKline

Wyeth-Ayerst

Consortium for the Advancement of

Manufacturing in Pharmaceuticals (CAMP)

Vendors

PROCESS A WITH CURRENT QC TESTS

7/27/2019 3763s1_15_raju

46/56


PROCESS A WITH CURRENT QC TESTS

AND NEW POSSIBILITIES

WEIGHING WET

GRANULATION

STEP FB DRY

STEP

BLEND

ENCAPSULATESIEVE

API

MICRO

Particle Size Description

ID

Assay

CU

Impurity

Dissolution

MICRO

QC1 QC3 QC4

DRY MIX

QC2

LOD

N d t F B th

7/27/2019 3763s1_15_raju

47/56


Need to Focus on Both

Material Flowand Information Flow

Bulk

Active

Bulk

Formulation

Filling/

Tableting/

etc.

Packaging/

Finishing

Sensor1(or QC test1)

YXY

1Y

2Y3

t 1 t 2 t 3 t 4

Y4

t

Sensor2(or QC test2) Sensor3(or QC test3) Sensor4(or QC test4)

INFORMATION FLOW

MATERIAL FLOW

M f t i I f ti M t

7/27/2019 3763s1_15_raju

48/56


Manufacturing Information Management:

Has Hardware and Software Components

Bulk

Active

Bulk

Formulation

Filling/

Tableting/

etc.

Packaging/Finishing

Sensor1(or QC test1)

YXY

1Y2

Y3

t1

t2

t3

t4

Y4

t

Sensor2(or QC test2) Sensor3(or QC test3) Sensor4(or QC test4)

0

2

4

6

8

10

12

14

16

18

20

0 50 100 150 200 250

Readings @ 0.375 seconds

1:10 powder flow

1:20

y = 1.9757x + 0.4886

R2= 0.9983

0.00

5.00

10.00

15.00

20.00

25.00

0.0 2.0 4.0 6.0 8.0 10.0 12.0

% of Triamterene

Mixing Of 1:10 Triamterene-Lactone @ 85%

Fill (10g) And 27.4 RPM

0

5

10

15

20

25

30

35

40

0 50 100 150 200 250 300

Number of Rotations

i

l

l

0

5

10

15

20

25

30

35

0 50 100 150 200

Number of Readings @ 0.375 s/reading

10%

5%

1%

0.5%0.1%

Fast ResponseOn-Line

Real-Time

Accurate

Robust

Rapid Rate of LearningShort Cycle Times

Benchmarking

Modeling

Continuous Problem Solving

7/27/2019 3763s1_15_raju

49/56


Bulk

Active

Bulk


Finish Packaging

BulkActive

BulkFormulation

Filling &Finish Packaging

Bulk

Active

Bulk


Finish Packaging

Company A

Company B

Company C

HORIZONTAL AND VERTICAL APPROACHES

Blending

Drying

Granulation

Fermentation Rapid

Microbial

Detection

Flow

Tableting

Transport

High Vol

Variable

Liquids

7/27/2019 3763s1_15_raju

50/56


CQV: BENEFITS

Data Mining of Process Data

Data -> Information-> Knowledge

Rationale for New Sensors

Variable Categorization: PCCPs, etc.

Basis for Specifications, Batch Record Design

Basis for Experimental Design, Etc.

7/27/2019 3763s1_15_raju

51/56


Learning Curve: Cycle Times

Total cycle time

10

100

1000

0 50 100 150 200 250 300 350 400 450 500 550 600 650 700

batch #

days

Accelerated Learning Curve Facilitated By

Continuous Quality Verification

7/27/2019 3763s1_15_raju

52/56


CQV: SO WHAT?

On-line sensorsdoing the same thing will have

only incremental impact

This impact will still be only incremental

even if there is an MES/EBR system

Data Warehousingfocused on exceptions can

have a large impact

On-line sensors + EBR + Data Warehousingcan fundamentally changepharma. mfg.

7/27/2019 3763s1_15_raju

53/56


PRODUCT LIFE CYCLE: OPPORTUNITIES

7:15 AM T ue, Mar 02, 1999

1.00 8.25 15 .5 0 22 .7 5 30 .00

Years

1:

1:

1:

-141.04

114.92

370.88

1: NET INCOME 2: NET INCOME 3: NET INCOME

1

1

1

2

2

2

2

3

3

3

3

Net Inco me (Un ti tle d Graph)

Reduction of

time-to-market

Reduction of

manufacturing cost

Pharmaceut ical Manu factu r ing :

7/27/2019 3763s1_15_raju

54/56


Pharmaceut ical Manu factu r ing :

Oppo rtuni ty Areas

Manufacturing: Cost --> Profit

Organizational focus: Functional --> Process

Optimization: Local --> Supply chain

Inventory Management:JIC --> JIT

Cost of Quality: Inspection --> Prevention

KEY TECHNOLOGY OPPORTUNITY:

On-line Sensors+EBR+Data Warehousing!

7/27/2019 3763s1_15_raju

55/56


ACKNOWLEDGEMENTS

Professor Charles Cooney (MIT)

Professor Stephen Byrn (Purdue)

CAMP

7/27/2019 3763s1_15_raju

56/56

NOTE ON CONTEXT

This presentation does not necessarily represent

the views of MIT, Purdue or CAMP

Some data have been disguised for reasons of

sensitivity and confidentiality

3763s1_15_raju

Documents

Transcript of 3763s1_15_raju