Case Study – Evaluation of Health-Based Exposure Limits and

Potential Impact on Manufacturing Equipment Cleaning Limits

EMA Workshop on generation and use of Health Based Exposure Limits (HBEL)
Date: 21 June 2017 * Version: 2.0

Presentation

Gretchen Allison, Pfizer Global Quality Operations – Validation Team Leader

www.efpia.eu
 Outline
• Background/Scope

• Step By Step Example

• Industry Experience

• Considerations for Inspections/Key

Messages

www.efpia.eu 2
 Scope
Primary Scope of this Case Study: Drug Product
• Commercial Drug Product
• Small Molecule
• New vs Legacy
• Product residue removal (vs micro, cleaning agent removal)
• Non-dedicated Manufacturing Equipment

www.efpia.eu 3
 Background
• Historically used drug product manufacturing equipment cleaning
limits based on 1/1000 minimum therapeutic dose
• Some markets still expect 1/1000 dose limit, or lower of that and
NMT 10 ppm limit
• Any cleaning must pass visual inspection
• EMA Health Based Exposure Limit (HBEL) guide published, effective
2015
• Some other markets (e.g. PICS, China) also expect HBEL assessment
in establishing cleaning limits
• New active ingredients and associated drug product have
documented HBEL at the time of commercialization
• ADE = PDE
• For older (legacy) products already on the market, it’s a bit more
complicated….
www.efpia.eu 4
For existing/older (legacy) products already on the market, it’s a bit
more complicated….
~74
63 Billion Doses
in one year ~30,000
Manufacturing
Sites Colleagues

850+ 25,000+
Products SKUs

~450 130
Contract Logistic
Manufacturers Centers

~5,000 30+
Patients in Legacy
Suppliers
129+ Companies
Countries

5
 Overview of Workprocess
For Legacy Products: Assess ADE based cleaning limits

Step 1: Identify and prioritize existing Active Ingredients and

associated Drug Products for HBEL evaluation
Step 2: Establish ADE value: see scenarios on next slide
Step 3: Compare health based cleaning limit to existing cleaning limit
• Where existing cleaning limit (e.g. based on 1/1000 minimum therapeutic
dose) is lower than health-based (ADE) derived cleaning limit, the existing
cleaning limit maybe retained as an acceptable approach
• Where health based derived cleaning limit is lower than existing cleaning
limit, a new health based cleaning limit is implemented for use. This may
require:
• Cleaning, sampling, and analytical test method detection level
evaluation
• Potentially lower detection limits to be established and validated
Must be done on product by product basis. For sites with many legacy
products = significant time/resource
Reference flowchart slide for workprocess
www.efpia.eu 6
Step 2 Scenarios: Establishing ADE value
Scenario A: For new active ingredients/drug product documented HBEL
in place at the time of commercialization i.e. ADE values available
Scenario B: If no ADE available:
• Screen out highly sensitizing beta lactams and review Segregation
product assessment
• For all Other products with no ADE value: Check OEL/OEB monograph:
is it based on clinical data?
• If the OEL/OEB is based on the low clinical dose, then ADE
development can be assigned a lower priority as 1/1000 the same
minimum therapeutic dose (MTD) will be sufficiently conservative
– regardless of how high or low (i.e., “potent”) the MTD may be.
• If the OEL or OEB is NOT based on the low clinical dose then the
product is prioritized for Tox review and estimation of the ADE
using the OEL/OEB monograph. Compare against current
cleaning limit and prioritize full ADE development, as needed
• For lower priorities, ADE to be developed per prioritized plan
• Once ADE developed, calculate cleaning limit based on ADE and
compare to existing cleaning limit www.efpia.eu 7
Flow chart (derived from PharmTech Europe Dec2015 publication)

www.efpia.eu 8
Example of small molecule cleaning limits calculation
using the ADE value and minimum therapeutic dose

Common Industry Cleaning Limit Calculations:

Step 1: Calculate Maximum Allowable Residue (MAR) as mg of Product
A per kg of Product B for either:
• A specific product changeover of Product A to Product B, or
• Calculate a worst case cleaning limit considering all products made in same
equipment

Minimum Therapeutic Dose MAR : For 1/1000 min. therapeutic dose, SF=0.001
Dose MAR = TA (mg of A) · conversion (106 mg of B/kg of B) · (SF)
BB (units of product B) · CB (mg of B/unit)

Health Based Exposure Limit (Tox) MARADE/PDE

TOX MARADE/PDE = ADE or PDE (mg of A/day) • conversion (106mg of B/kg of B)

BB (units) • CB (mg of B/unit)
www.efpia.eu 9
Example of cleaning limits calculation using the ADE value
and minimum therapeutic dose
Common Industry Cleaning Limit Calculations (Continued):
Step 2: Using Calculated MAR from Step 1, convert MAR to the allowed
amount per cleaning sample:

Residue Acceptance Limit (RAL) for Swabs as (mcg of A) per Swab:

Swab RAL = MAR (mg of A/kg of B) · LB (kg of B) · AS (cm2/swab) · conversion (103 mcg A)/(mg A)
EW (Equipment Surface Area in cm2)

RAL for Rinsate as (mg of A) per kg of Rinse:

Rinsate RAL = MAR (mg of A)/(kg of B) · LB (kg of B)

WR (kg of Rinse Used)

www.efpia.eu 10
 Example 1: Resulting Swab Cleaning Limit Grid
Product A = Drug Product being cleaned out of equipment
Product B = Next Drug Product to be made in that equipment
For a given manufacturing equipment producing 4 different Drug
Products, one would have the following permutations of cleaning
limits for each change-over cleaning:
Changeover from Product 1 to product 3 results in the worst case
lowest cleaning limit

www.efpia.eu 11
 Example 2: Establishing Worst Case Cleaning
Limits for Therapeutic Compounds
Input Input Input Result Input Result Input Input
Product TA BB CB Max Daily LB Dose AS EW
(mg (kg) MAR (cm2) (cm2)
Dosage (mg
dosage dosage (mg A/Kg
unit) unit/day)
next drug
Product1
to 3)

1 2.5 1 150 150 650 1.89 100 100,000

2 5 2 250 500 1000 -- 100
3 10 3 440 1,320 200 -- 100
4 50 5 250 1,250 750 -- 100
5 100 4 800 3,200 700 -- 100
Worst case Prod 1 is Highest Next Prod 3 is Prod 1 100,000
comment lowest weight product 5 is smallest to 3 is
highest batch lowest

Note when establishing worst case limits, the worst case Product A to
Product B changeover scenario is typically used in the calculation.
www.efpia.eu 12
Resulting in overly conservative limit for most changeover scenarios.
Example 3: Spreadsheet Reflecting the Inputs for
Provisional ADE Values
After screening of products for potential segregation, assess
Other products made in that equipment
Drug Endpoint Dose PK/MF UFc ADE Value ADE Value
Substance Used (mg) (Oral) (Parenteral)
(mg/day) (mg/day)

Product 1 MTD 250 0.4 90 2 1

Product 2 MTD 100 0.7 30 3 2
Product 3 MTD 1 0.6 30 0.03 0.02
Product 4 MTD 20 0.5 30 0.7 0.3
Product 5 Long- 0.3 1 180 0.08 0.08
term rat mg/kg
LOAEL /day
Product 6 MTD 0.5 1 30 0.02 0.02

PK/MF = Pharmacokinetic Adjustment Factor or Modifying Factor

UFc = Composite Uncertainty Factor
MTD = Minimum Therapeutic Dose
“Other” = Segregation not required per Segregation Quality Standard www.efpia.eu 13
 Step by Step Example
Prioritizing Development of Health-Based Exposure Values and
Evaluating Effectiveness of Current Cleaning Limits

Follow the decision tree tool in Flowchart

1. Identify the product: ‘Capzone’
2. Check if the ADE value for Capzone is available.

www.efpia.eu 14
 Step 3: If an ADE value is available

• If an ADE value is available, perform the cleaning limit (CL)

calculation. Or, compare the ADE value directly to the 1/1000
minimum therapeutic dose value used in the cleaning limit
calculation, since the rest of the variables of the cleaning calculation
are identical whether performing a dose based limit or an ADE based
limit calculation.
• Compare if the CL using ADE is greater than, or less than the current
limit. Document the assessment and conclusions, including effect on
the Cleaning Validation (CV) Program:
— If the new health-based CL > or equal to Current CL, can
maintain current CL.
— If the new health-based CL is <current CL, evaluate the
difference between the two CLs with site Quality Assurance,
Production, EHS, Quality Control (if applicable) for further
discussion and resolution.

www.efpia.eu 15
 Step 3: If an ADE value is not available for
Capzone
• Determine if Capzone has been assessed for segregation (e.g.
highly sensitizing beta lactam). If it does not require segregation:
• Verify if the OEL or OEB monograph is available
• Verify if the monograph is based on the minimum therapeutic
dose:
— Refer to the abstract on the first page of the monograph i.e. the
monograph is based on the minimum therapeutic dose when
the monograph states that it is based on clinical data, or both
clinical and nonclinical data
— If it is not clear in the monograph that this was based on the
minimum therapeutic dose then further toxicologist analysis is
required (step 6)
• Verify if current cleaning limit (CL) was calculated using 1/1000th
of the same minimum therapeutic dose used to calculate the OEL
or OEB.
• If so, then current CL is considered appropriate. Document the
assessment and conclusions.
www.efpia.eu 16
 Further toxicological analysis

• Further toxicological analysis is required when:

— An ADE is not available and the product requires segregation assessment
— The OEL or OEB is not based on the minimum therapeutic dose or is not
available
— If current CL was not calculated using 1/1000 of the same minimum
therapeutic dose used to calculate the OEL/OEB (e.g., it was calculated
using NMT 10ppm, or used a SF of 1/100, etc.)
• The toxicologist determines if a provisional ADE value can be defined. If so, the
site may utilize it in the comparison to existing cleaning limits.
• The site will calculate a health based CL
• If the new health-based CL is lower than current CL, notify the toxicologist to
prioritize the development of ADE value for re-analysis. Communicate to site
Quality Assurance, Production, EHS, Quality Control (if applicable) the potential
for change to the current CL for further discussion and resolution.

www.efpia.eu 17
 Finalize the assessment
• Once the ADE has been determined and approved, the
site will verify that the provisional and approved ADE
values are the same. If not the same perform step 4 –
recalculate the health based CL and compare to existing
CL.

• If an ADE value cannot be determined (e.g. highly

sensitizer Beta lactam products) then the site will review
the manufacturing segregation assessment

Note Steps need to be repeated for each product

www.efpia.eu 18
 Large Molecule Considerations

• Biological products often degrade and

denature during cleaning, rendering them
inactive

• EMA HBEL guide Sect 5.3: using ADE value of

the active and intact product may not be
required for macromolecules

www.efpia.eu 19
 Industry Experience
At Pfizer: based on internal survey
• Assessment outcomes vary by manufacturing site: e.g. dependant
on their equipment use (e.g. dedicated or multi-product), type of
products (e.g. small or large molecules)

• Generally in > 85% of cases, dose cleaning limit is lower than health
based cleaning limit using ADE, often by an order of magnitude or
more

• Performing the assessment for legacy products consumes significant

resources (e.g. time)

In industry: based on prior EfPIA survey: Generally in > 85% of cases,

dose cleaning limit is lower than health based cleaning limit using ADE

www.efpia.eu 20
 Industry Experience
At Pfizer: Why are sites reluctant to use/change to less stringent
health based cleaning limits?
• Most sites have cleaning validation complete (or ongoing for
new products, other changes)
• Changing the cleaning limits requires re-assessment of:
— The analytical test and sampling methods used to detect the product
residue (e.g. have they been validated in a range to include the new
higher limit?)
— The equipment cleaning validated status: does validation need to be
re-executed?
• There is risk of failing visual inspection after cleaning when using
health based cleaning limits, which are often significantly higher
than dose based cleaning limits
• If site experience is that they can consistently clean to the lower
dose limits, there is often little motivation to consume resources
to assess using less conservative cleaning limits
• Some markets still expect dose based cleaning limits
www.efpia.eu 21
 Considerations for Inspections
• Developing HBELs for legacy products is time consuming, especially for
sites with hundreds of products
• For legacy processes, assessing current cleaning limits vs HBEL derived
cleaning limits can take significant time and effort, depending on the
number of products at a manufacturing site
• In majority of cases (>85%), dose based cleaning limits used historically
in industry are more conservative than use of HBEL cleaning limits
• Risked based approach to prioritization and assessment is wise use of
resources
• Do manufacturing sites have a documented risked based
plan/approach, especially for legacy products, with established
timelines?
• Are qualified toxicologists intimately involved with the process per that
documented plan/approach?

www.efpia.eu 22
 Concluding remarks from the Associations
• All express their thanks for the opportunity to
engage in dialogue at this workshop, and offer
their resources for future discussion
• For us, a successful outcome will ensure:
• The difference between risk and hazard is well
understood
— Regulators will have confidence in the processes used
by industry to derive HBELs
• It’s not just about the numbers
• There is a lot of scholarship associated with deriving
HBELs
— It is understood that many, especially smaller,
companies may not have site-based or in-house
expertise and resources and may have to outsource
the work
— Extensive discussion about the derivation of HBELs
during an inspection may not be possible
www.efpia.eu 23
 Next Steps
• We recommend a substantial revision of the Q&A document
— Or revision to the SWP guideline
o We would appreciate an opportunity to comment on any future Q&A,
reflection paper or revised guideline prior to release
• There are also topics of importance that remain to be resolved e.g.,
— Relating to animal health products
— Responsibilities of the manufacturer vs MAH
• Given the complexity of the subject matter, future timelines should be
carefully considered
— It may be that the recent implementation timelines were too short and
resulted in some companies being unable to make sufficient investment
in HBEL determinations
• Similar to the increased collaboration between assessors and inspectors, we
support the greater involvement of safety experts with the inspectors

www.efpia.eu 24
Questions?

www.efpia.eu
Backup

www.efpia.eu
Approach 2: How can I use the existing OEB to estimate a default
ADE band for comparison against the current cleaning limit for
investigational product?

• OEB paradigm closely aligns with ADE value methodology

so bottom of band can be read to a default ADE value
• Same value as derived using Threshold of Toxicological
Concern (TTC) outlined by Dolan et al 2005
• Where existing cleaning limit is lower than the calculated
HBEL cleaning limit, development of a full ADE maybe
lower priority
Pfizer OEB Default ADE Value
(default TTC per Dolan et al)
OEB 1 10,000 ug/day
OEB 2 1,000 ug/day
OEB 3 100 ug/day
OEB 4 10 ug/day
www.efpia.eu 27
OEB 5 1 ug/day