A Science and RiskBased Approach for Establishing Critical Process Parameters and Critical Intermedi

1
A Science and Risk-Based Approach for
Establishing Critical Process Parameters and
Critical Intermediate Quality Attributes

• Presented 28 30 September 2005
• Thomas Parks, PhD
• Principal Research Scientist Statistics
• Eli Lilly Company

2
Considerations

• When developing a methodology for determining a
  critical process parameter (CPP) or critical
  intermediate quality attribute (CIQA), it is
  appropriate to consider the following question

What defines critical?
3
Principles

• A parameter (process parameter or intermediate
  quality attribute) is critical when it directly
  or indirectly causes a critical quality attribute
  (CQA) of the API to be outside of its acceptable
  range/specification.
• No parameter is critical within its proven
  acceptable range (PAR).
• A parameter is potentially critical when it is
  outside its PAR.

4
Therefore

• To assess criticality of a parameter requires an
  assessment of
• The strength of relationship of the parameter to
  one or more API CQAs.
• The probability that a parameter will be outside
  its PAR (or specification).

5
The Science/Risk-Based (SciRi) Approach
Select a Critical Quality Attribute (CQA) of the
API, e.g. an impurity
Select a parameter (?), e.g. temperature
Does CQA f(?) over PAR?
How strong is the relationship?
What is Prob(? outside PAR)?
Is PAR a limit of failure (LOF) or limit of
success (LOS)?
6
Assessing Criticality The SciRi Matrix

• The SciRi Matrix uses a table where the rows
  represent the probability of being outside the
  PAR and the columns represent the strength of
  relationship of the process parameter (or
  intermediate quality attribute) with the API CQA.
• This table provides a qualitative framework for
  assessing criticality.
• The table can be useful when limited data are
  available for the specific process and scientific
  judgment must be applied.

7
Assessing CriticalityThe Qualitative SciRi
Matrix
Note When a PAR limit is a limit of failure, the
strength of relationship to the API CQA is by
definition large. Note A process parameter or
intermediate quality attribute that has a
relationship to an API CQA is defined as a Key
Parm/Attribute
8
Establishing Consistent Assessments

• The Qualitative SciRi Matrix provides an
  overarching framework for criticality assessment.
• However, to provide a consistent criticality
  assessment it is desirable to establish a
  quantitative approach to the SciRi Matix.
• One quantitative approach is
• Assess Strength of Relationship to the CQA by
  determining the number of standard deviations
  (sigma) departure from the PAR required to
  predict a failure of the API CQA
• Assess Probability of Exceeding the PAR by the
  process capability index Cpk

9
Strength of Relationship - Linear
Over 3 Sig away
Max acceptable value for CQA
3.0 Sig
1.5 Sig
L
M
S
CQA
PAR
Parameter being assessed
10
Strength of Relationship Non-Linear
Over 3 Sig away
Max acceptable value for CQA
3.0 Sig
1.5 Sig
L
M
S
CQA
PAR
Parameter being assessed
11
Assessing CriticalityA Quantitative SciRi Matrix
12
An Alternate Probability Assessment

• An alternate approach to using process capability
  to assess the probability of exceeding the PAR is
  to make the assessment based on measurement
  uncertainty. In this approach, an alarm limit is
  set inside the PAR a certain number of
  measurement standard deviations (?). This alarm
  limit is often referred to as a guardband (GB).
• Alternate quantitative approach
• Assess Strength of Relationship to the CQA by
  determining the number of standard deviations
  (sigma) departure from the PAR required to
  predict a failure of the API CQA
• Assess Probability of Exceeding the PAR by the
  number of measurement ? the GB is set inside the
  PAR

13
Probability Assessment Using Meas Uncert
Lower PAR
GB
S
M
L
14
Assessing CriticalityAn Alternate Quantitative
SciRi Matrix
15
Benefits of the SciRi Matrix Approach

• Science Based
• Strength of relationship to CQA
• Risk Based
• Probability of exceeding PAR
• Flexibility with respect to amount of available
  data
• Consistency in CPP assessment and justification
• Covers entire space, i.e. all parameters
  explicitly defined as critical or non-critical
• Consistent with PhRMA API Working Group approach
  proposed in July issue of Pharmaceutical
  Technology