6/10/2011

1

Applying PAT Principlespp y g pto

Pharmaceutical Water Systems

By Bob LivingstonA i W t I

IVT ACE: Philly June 23 , 2011

Arion Water, Inc.bob@arionwater.com

Vocabulary • PAT – Process Analytical Technology• GAMP –Good Automated Manufacturing

Practice • CCP- Critical Control Points • CQA- Critical Quality Attributes • MDA - Multi-Variant Data Analysis• DOE – Design of Experiments • DCS -Distributive Control System • HACCP - Hazard Analysis and Critical

Control Points • FMEA – Failure Mode and Effects

Analysis • FMECA – Failure Mode Effects and

Criticality Analysis

6/10/2011

2

PAT - The Promise and the Problem

• PAT offers the promise of reducing or removingreducing or removing laboratory testing from the verification process and permitting real-time release of water for productionproduction

ASTM E-2500 -07

• New Regulatory expectationsp

• Dramatic reductions in the scope of validation now coming to a facility near you

6/10/2011

3

Verification replaces VALIDATION

• ASTM Standard E 2500 – 07• promotes a scientific

RISK BASED approach to Pharmaceutical FacilitiesPharmaceutical Facilities Processes, Utilities and Equipement

Verification is based upon RISK:-to the product

-and patient safety• Verification testing• Verification testing

assures the process or equipment is fit for the application

• The verification testing is gdetermined by the engineers, not just the QA/QC unit

6/10/2011

4

Verification is based upon RISK:-to the product

-and patient safety

• The level of verification testing and documentation is based

th RISKupon the RISK

The Problem

• Many USP Purified pharmaceutical water psystem designs may not be robust or reliable enough to qualify for PAT approval

Y t it– You cannot monitor a system into compliance

6/10/2011

5

PAT already exists-

• All continuous 24/7All continuous 24/7 utilities and processes in a Semiconductor production facility use PAT and PAI technology-

l b t t• no lab tests are administered 1

FDA PAT web site

• Timely measurement of raw materials and in

t i l dprocess materials and processes

• Analytical in PAT includes –chemical, physical, microbial,

th ti l d i kmathematical and risk analysis conducted in an integrated manner 2

6/10/2011

6

Industry: On-line analytical requirements 2

• Regulatory expectation• ReliabilityReliability• Analytical performance• Ease of use• Elimination of manual

errors

System Suitability test frequency

• For on-line instruments, it may be performed only y p yonce per year

• What if it failed?• Corroborating evidence

may be a necessity 2

6/10/2011

7

Real time release of data-• Computer, data acquisition,

process sensors, process analytical equipment (PAI) a a yt ca equ p e t ( )must be integrated in to a comprehensive management system.

• GAMP provides guidance in designing process control system.

• Not only monitoring- must have a control element involved-auto or manual

PAT (Process Understanding)

• CQA’s (Critical Quality Attributes))

– Identify– Monitor– Analyze– Control– Report

6/10/2011

8

The challenge in PAT

• Efficient management and effective analysis of ylarge amounts of streaming data, facilitating the correct response to changes

WFI Critical Quality Attributes (CQA’s)

• TOC• ConductivityConductivity• Temperature• Rapid Microbiological

Enumeration• On-line Endotoxin

6/10/2011

9

Multivariate data analysis

• Used to determine if any correlation exists between various combinations of the process inputs and critical quality attributes

Correlations-

• Mathematical relationships between in process measurement and critical quality attributes are critical to understanding the process, identifying data gaps, performing Design of Experiments and ultimatelyExperiments, and ultimately , developing feedback controls strategies to optimize the process. 3

6/10/2011

10

Design of Experiments (DOE)-

• Verifies correlations established via MVA (multivariate Analysis)

• Running experiments on water systems is not easy and may not be a good ideaidea 3

Demonstration-

• The analytical instrumentation

• the IT system• the statistical reporting

• With sufficient data to demonstrate reliable hardware/softwarehardware/software functionality, and establish a baseline of inter and intra batch variability 3

6/10/2011

11

Demonstration -

• Natural Variance Method• Useful for existing water

systems

• Imposed Variance Method– DOE

• Expanding Variance Method (The KernelMethod (The Kernel Approach) – Uses both of the above 3

Expanding Variance Method (The Kernel Approach) -

• Sufficient # of demo batches to correlate a range of variability smaller than the statistical model suggests could be employed (the kernel)

• If outside of kernel, in-process and final releaseprocess and final release testing must be added

• But then the kernel expands 3

6/10/2011

12

Monitoring and Control

• Monitor the analyzers– TOC, Temp and

Conductivity report to the Utilities DCS System

• The DCS system should be compatible with all the latest bus technologies (MODBUS fieldbus(MODBUS, fieldbus, profibus, 4-20 mA signals, etc.)

Monitoring and Control

• Data from all skids should be available on

l tfone platform• Phase in new sensors as

technology becomes proven– New analyzers may be

evaluated and added to the platform after successful validation

6/10/2011

13

PAT Dossier- consensus on-• Extent of experimentation• Interpretation of results• Accuracy of mathematical models• Accuracy of mathematical models• Capability and reliability of new

technology• Assessment of risks• Selection of CCP and CQA• Selection and placement of

analyzersanalyzers• Determination of endpoints and

process ranges• Design decisions

Change Procedures

• Risk assessments for Failures– Instrument failure– Hard/software failure– Adverse trends– Variation outside up/low

control limits• For likelihood, ability to

detect, and severity of failures 3

6/10/2011

14

Tools for Risk Assessment

• HACCP (say ”hassup”)

– Critical control pointp– Process input– Process Parameters– Process Output

– Critical Quality attribute– Use a Pareto Chart

• FMEA• FMECA

Analytical Tools-

• Spectroscopy-• (Vis, UV, NIR, X-ray, Turbidity,

Refractivity)Refractivity)

• Chromatography • (HPLC, GC, IC)

• Electrochemical• (pH, DO2, CO2, Conductivity, TOC)

• PhysicalPhysical• Particle count, temp, pressure,

flow)

• Wet Chemistry- in lab

6/10/2011

15

Create a PAT Project Plan 3

• Opportunity description• Scope- process and equipment• Scientific Rational

P D i ti• Process Description• Roles and responsibilities• Training• Benchmarking, Data Collection, Metrics,• IT requirements• Financial analysis • Validation strategy• Regulatory submission strategyRegulatory submission strategy• Implementation strategy• Project execution plan• Project Change Management

Critical Evaluation of USP Purified Water S t D iSystem Designs

6/10/2011

16

City Feed

• Potability required• Risk – Low to moderateRisk Low to moderate

– Historical data trending– Quality of potable feed is

of sufficient stability and quality to prevent risk to final product qualityfinal product quality

City feed on-line Monitoring

• Conductivity• Turbidityy• Temperature• Pressure• Flow

• Lab – Bioburden– Potability

6/10/2011

17

Pretreatment Functions

• Multi-media • Filtration

• Softener

• Carbon

• Hardness

• Oxidants• TOC

• Filters • Pump protection

6/10/2011

18

Multi media vessel

• Function- Filtration• Risk- Low to moderate• Instrument -turbidity(no good on-line SDI test) • Operation-

– Pressure dropFl t– Flow rates

• (Service and backwash)• Lab- Bioburden

Water Softener

• Function- Hardness Removal

• Risk- Low to moderate• Often non-critical

• Instrument- Hardness meter

• Operation– Brine level, flow rates,

pressure profile

6/10/2011

19

Carbon Bed• Function:

– Oxidant and TOC removal• Risk:• Risk:

– Moderate to High (for WFI only)• (Amines in Still or clean steam)

• Instrument : – Chlorine analyzer

• Operation:• Pressure dropPressure drop• Flow rates

– (Service and backwash)

• Lab- Bioburden

Pretreatment Performance • Low Risk for most

pretreatment functions not directly feeding WFIdirectly feeding WFI

• Consolidate allinstrumentation to monitor RO feed water only

• Pressure drop• Flow (service and backwash)• Conductivity• Conductivity• Turbidity• Hardness• Temperature • Oxidants (Total Chlorine, not ORP)

6/10/2011

20

Purification Technologies and Functions

• Risk : High(critical process step)(critical process step)

• Reverse Osmosis• Double Pass RO• EDI/CDI

Mi d B d I E h• Mixed Bed Ion Exchange• Or combination of above

Purification Functions

• Reverse Osmosis is a universal critical process pstep for Purified Water production

• > 95% ion reduction• >95% bacteria reduction• >95 % TOC reduction• >99 % Suspended Solids

6/10/2011

21

Primary RO Functions

• Risk : High to moderate• InstrumentationInstrumentation

– % Ion Rejection– TOC reduction

• Operation– Flow profiles

– Rates and totals

– Pressure profiles• Lab- bioburden

6/10/2011

22

Secondary Purification

• Risk : High to moderate• InstrumentationInstrumentation

– % Ion Rejection– TOC reduction

• Operation– Flow profiles

– Rates and totals

– Pressure profiles• Lab- bioburden

Design Matters

• Are we finished yet?• Often no additional

processing after storage is utilized

• Water System Design must be robust to utilize PAT approachespp

• Pharmaceutical water systems designs are not often robust

6/10/2011

23

Conventional Purified Design

• RO and CDI run continuously• TOC Control is insufficient• Water is never 18 MegΩ

• Will support microbiology

• No equipment after storage– Lose control of product water quality

• Routine Sanitization is Required

6/10/2011

24

Purified Plastic Design

• RO operates On /OFF as requiredq

• Process equipment in distribution loop

• 18 MegΩ water made and delivered

• But bacteria are often in product water– Must be sanitized

Storage

• Storage tanks cannot control microbiologygy– Attempts to do so will

prove frustrating• Water in storage is

essentially stagnantC t h ld• Counts should range from 1-20 cfu/mL

• Walls of tank should not feel slimy

6/10/2011

25

Water Storage

• PAT is not only monitoring- must strive to ghave a control element involved– auto or manual

• Sufficient control of prod ct ater q alit isproduct water quality is difficult to maintain in storage

Quality Process Design

• PAT benefits and results are best applied to high

lit b t d llquality, robust and well understood Process Designs

• Insufficient process expertise may be lacking i th d i h fin the design phase of some Pharmaceutical Processes

6/10/2011

26

FDA web site -PAT Guidance for Industry (Principles and Tools) 2

• These PAT tools used in a system can provide effective and efficient means ofand efficient means of acquiring info to facilitate process understanding, develop risk mitigation strategies, achieve continuous improvement,p ,and share knowledge.

• Must include two or more tools to be PAT

Postreatment equipment after storage is highly recommended

• Do not “lose control” of the high purity water g p yonce produced

• Add TOC control to distribution loop– TOC control=

microbiological controlmicrobiological control• Do not use regenerated

ion-X resin

6/10/2011

27

Bio-adhesion Control in Ambient Distribution Loops

• Most attention has been focused on the distribution piping, materials and Loop design to control product water microbiology

Many a piping system has– Many a piping system has been blamed for bacteria problems

6/10/2011

28

Product Water Quality controls microbiology

A properly designed water system will control bacteria in ambient distribution loops via :distribution loops via :

• Inoculation Prevention• Nutrient deprivation

– Both of which are functions of the water system design, not the distribution piping design

Robust Design facilitates “Real Time Release”

• Water quality exceeds specifications by orders p yof magnitude

• Reduced maintenance requirements ensure reliable operation

6/10/2011

29

Post Treatment Design

• Ambient temperature water systems must have yhigh product water quality with no fluctuations

• Post Treatment Equipment permitsEquipment permits precise product water quality control

TOC UV Irradiation

• Function: Microbial and TOC control

• Risk: Low • (high reliability)

• Instrumentation:– UV Monitor– (On-line TOC not required

until final product)

6/10/2011

30

Mixed Bed Ion Exchange

• Function: Trace impurity removal– (ionic, CO2, microbial,

endotoxin, particles, silica)• Risk: Moderate to low

• Due to redundant operation

Instrumentation:• Instrumentation:• Resistivity probe

• Lab: Bioburden

Final Filtration

• Function: Backup particulate removalp

• Risk: Low • (high reliability at this stage)

• Instrumentation:– Pressure drop and

di t ib ti fl tdistribution flow rate

6/10/2011

31

Real-time Release of Product Water for Production e

6/10/2011

32

Product Water

• USP Purified Specs – Conductivityy– TOC– Microbio – Endotoxin

• A common risk is permitting Carbon Dioxide in product water

Product Water Critical Quality Attributes (CQA)

• Microbiology• ConductivityConductivity• Endotoxin• TOC• Flow • PressurePressure• Temperature • Particle Counts

6/10/2011

33

Validate after Changes ?

• If the changes will result in little or no risk to the

d t t CQA’product water CQA’s, or even improvement or increase in reliability,

• NO re-validation of the system is required.

• A change control with continued monitoring will suffice!

Final Product Water Quality

• Meet final product water quality specification with q y proom to spare with no evidence of trending at all towards regulatory specifications

6/10/2011

34

TEST ATTAINABLE ACCEPTABLE

ALERT

CRITICAL PURIFIED WFI

Conductivity

Resistivity (MegOhm) @ 25 C

0.055

(18.2)

0.10(17.9)

>0.5

(<17.5)

>1.0

(<1.0)

<1.0 <1.0

TOC (ppb) Off Line (grab sample)

<10 < 50 >50 >300 <500 <500

TOC (ppb)On Line - Instrument

<2 <10 >100 >300 <500 <500

Bacteria (CFU)Count / 10 L sample

Count / 1000 mL sampleCount / 100 mL sample

Count / 1 -ml

<1<1<1<1

<100<10<1

>100>10>1>1

>500>50>5>1

1 Million100K

10,000**100**

1,00010010<1

<1

Endotoxin (by LAL)Units: EU (Endotoxin Units)

<0.001 <0.001 0.005 >0.10 N/A 0.25

Silica (ppb) dissolved(For information purposes only)

<3 <5 >25 N/A N/A

Required future on-line testing-1

• Endotoxin • Microbial enumerationMicrobial enumeration• Microbial Identification• Ion chromatography

– Will be on-line or at-line

6/10/2011

35

CQA- WFI Microbiology

• Historical Data Trending• >65 C water has no recorded

failuresfailures• Ambient temp WFI systems have

no recorded failures

• Rapid Enumeration• Emerging technologies

• Particle counts• WFI is contaminated with particles

• TOC and Endotoxin• Not correlated to bacteria #’s

CQA- “HP” (Purified) Microbiology

• Historical Data Trending• New water quality

designation

• Rapid Enumeration• Emerging technologies

• Particle counts• Very low counts possibleVery low counts possible

• TOC and Endotoxin• Not correlated to bacteria #’s

6/10/2011

36

CQA – Conductivity, TOC, Endotoxin

• Standard Pharmaceutical design not robust enough

– No post treatment

• Conductivity fluctuates from 0.9 – 1.3uS with occasional excursions

• TOC fluctuates and often t l h tnot low enough to

provide microbial inhibition

Robust HP Purified Design

• Post treatment equipment in distribution loopp

• On-line Conductivity does not fluctuate (0.05 uS)

• On-line TOC does not fluctuate

• 5- 1 ppb without excursions

• Endotoxin does not fluctuate <0.001EU

6/10/2011

37

Product Water Operational Parameters

• Product Flow and Pressure

• (outgoing and return)

• Product Particle Counts• (return water for worst case)• Correlate with lab bioburden

Multivariate Data Analysis-3

• Process understanding and historical data may ybe used to build models to predict in process material attributes from the engineering measurements 3measurements. -3

6/10/2011

38

Real-Time ReleaseProcess understanding can

provide high assurance f lit b t hof quality on every batch

and provide alternativeeffective mechanisms to achieve validation.

• “Process understanding j tif l tican justify real time

release”• HOOOORRAAYY !!

The Problem with PAT

6/10/2011

39

Purified Water is not good enough

• Often Purified Water is not of sufficient microbial quality for most BioPharm applications

• 10 - 50 cfu/ml cannot be reliably controlled

B ik OOS f– Bugs can spike OOS for a variety of reasons

HP Purified Water is secretly desired

• HP (Highly Purified) Water requires q– <10 cfu/ 100 ml bioburden– < 0.25 EU Endotoxindue to popular demand

• Low microbial HP Purified W t i i d fWater is required for diverse application,

• Lab, media and buffer prep

6/10/2011

40

Distribution Loop and Point of Use Configuration

It’s is better than you think

• The microbial quality of the water is distribution is typically better than the microbio test results indicate

• The difficulty is getting the water out of thethe water out of the distribution system without contamination

6/10/2011

41

Figure 7Standard Plate Count for Ambient Temperature Purified Water System

No Sanitizations Performed Since Start-up (May 2000)Average of 6 samples collected weekly8 out of 300 samples were discarded

90

100

30

40

50

60

70

80

cfu/

100m

L

Arion System DesignNon Arion System Design

0

10

20

30

1/8/

2002

1/22

/200

2

2/5/

2002

2/19

/200

2

3/5/

2002

3/19

/200

2

4/2/

2002

4/16

/200

2

4/30

/200

2

5/14

/200

2

5/28

/200

2

6/11

/200

2

6/25

/200

2

7/9/

2002

7/23

/200

2

8/6/

2002

8/20

/200

2

9/3/

2002

9/17

/200

2

10/1

/200

2

10/1

5/20

02

10/2

9/20

02

11/1

2/20

02

11/2

6/20

02

12/1

0/20

02

Sample Date

Problem Water Systems

• Allow stabilization or equilibration of microbial

ltresults• If numbers remain

excessive, look for ways to reduce nutrients (TOC)

• If quality is not q yacceptable, the water system design is likely at fault

6/10/2011

42

Sanitization may not help

• Hot Water and Chemical Sanitization of problem pwater systems may exacerbate problems

• TNTC spikes are common after sanitization

Thi i lik l th– This is likely the recolonization of biofilms

The Microbiological Quality of Water in Distribution

• If higher quality Purified Water is required, 100 ml q ,sample size is statistically too small

• To demonstrate < 1 cfu/ 100 ml quality,

lit l ione liter sample sizes are required

6/10/2011

43

If Better Microbio Results are Required

• Reliable HP Purified water system design is requiredy g q

• Point of Use and Sample valve configuration is critical

• Properly designed and t d di t ib ti loperated distribution loop

is required

6/10/2011

44

Microbio Quality of Product Water

• Point of Distribution (PoD)( )– the quality of water the

system can produce. If it is not in spec, the water system design is at fault.

• PoD must be sampled atPoD must be sampled at high volume – (1L minimum) for

statistical relevance.

Micro Bio Quality of Return

• Point of Return (PoR) – The quality of water the q y

distribution system can deliver. If not in spec; the use points or distribution piping design is suspect.

• PoR must also be sampled in large volume– 1 Ltr sample size

6/10/2011

45

Point of Use Configuration

• Point of Use (PoU) – The quality of water procured q y pfrom the use point configurations and the sample points

• The PoU bacteria results are often significantlyare often significantly poorer than the water in distribution

Point of Use Valve Multiplier (estimated)

(distribution bacteria = 1X)

• Zero dead leg diaphragm (1 5x)• Zero dead-leg diaphragm (1.5x) • Diaphragm (3 x) • Ball Valve (5 X)• Needle Valve (10X)• Goose Neck (20 X)• Goose Neck (20 X)• Hoses or Tubing (30 X)

6/10/2011

46

Use Point Configuration

• The number and size of the use points must bethe use points must be in realistic proportion to the size of the pipe

• PoU valves must not exceed ½ the size of the distribution pipe sizedistribution pipe size

Size and # of Use Points

Pipe Size Use Point Max #

1.0” ¼ - 1/2” 15- 101.5” 3/8” – ¾” 30- 202.0” ½” – 1.0” 40 - 203.0” ¾” – 1.5” 50 - 204 0” 1 ” 2 ” 60 204.0” 1 ” – 2 ” 60 - 205.0” 1 ” – 3 ” 70 - 20

6/10/2011

47

Distribution Loop Design

• Instantaneous water demand should not exceed 75% of the flow in distribution

• In critical applications, max water demand should not exceed 50% ofshould not exceed 50% of the flow in distribution

Distribution Piping

Pipe Size Flow rate Length

1.0” 12 gpm 400 ft1.5” 30 gpm ~1,500 ft 2.0” 60 gpm ~3,000 ft 3.0” 140 gpm ~4,000 ft 4 0” 200 gpm 6 000 ft4.0” 200 gpm ~6,000 ft 5.0” 350 gpm ~8,000 ft

6/10/2011

48

Distribution Loop Design

• Do not disrupt flow• Flow surges and g

pressure fluctuations in the distribution loop scour pipe and dislodge biofilm

Do not disrupt flow

• Only distribution pumpsneed be on back up ppower supply

• If distribution is disrupted, await one or two loop recirculation volumes to be assured ofvolumes to be assured of a return to equilibrium quality

6/10/2011

49

Water Sampling

Water Sampling

• Sampling still remains problematicproblematic

• Sampling should reflect the way water is used

• Do not use alcohol or long rinse regimes

• Require Max velocity rinse for 3-5 seconds prior to use or sampling

6/10/2011

50

Rinse at Maximum Velocity

• Use points must be capable of max velocity irinse

• Do not place in ceiling or pointed at electrical panels

• Use points over 1.5” cannot be rinsed at max velocityvelocity– Must add a hose or a

sample valve

The FDA PAT Team wants to help

• Should be included from the beginning g g

• The FDA PAT team performs inspections

6/10/2011

51

Thank You Questions?

Contact Us:

Bob@arionwater.com

References• 1-”An Introduction to Process Analytical

Technologies” , by Nissan Cohen, Pharmaceutical Engineering, Jan/Feb 2004, Vol. 24 no.1

• 2-”Auto Release of Water Using TOC Analysis and Risk based cGMP, Inspection, and PAT Principles,” Rick Godec and Nissan Cohen, Pharmaceutical Engineering, Jan/Feb 2005, Vol 25 No.1

• 3-”The Four Steps of PAT Implementation-by John Davis, PE and John Wasynczuk, PhD. Pharmaceutical Engineering, Jan/Feb 2005.