parametric tolerance interval (pti) test for improved control of delivered dose uniformity (ddu) in...
TRANSCRIPT
Parametric Tolerance Interval (PTI) Parametric Tolerance Interval (PTI) Test for Improved Control of Test for Improved Control of
Delivered Dose Uniformity (DDU) Delivered Dose Uniformity (DDU) in Orally Inhaled and Nasal Drug in Orally Inhaled and Nasal Drug
Products (OINDP)Products (OINDP)Presented by Bo Olsson (AstraZeneca)
on behalf of International Pharmaceutical Aerosol Consortium on Regulation and Science (IPAC-RS)
13 March 2003Rockville, MD
March 2003 IPAC-RS
2Purpose of Delivered Dose Purpose of Delivered Dose
Uniformity TestUniformity Test
DDU, which is one of several quality attributes for OINDP, combines performance of delivery device and formulation
DDU verifies delivered (emitted) dose in the batch– dose uniformity between containers– dose uniformity within multi-dose containers– closeness of mean dose to target - label claim
March 2003 IPAC-RS
3
Scope of ProductsScope of ProductsOral inhalation and nasal drug products (OINDP)
are either single-dose or multi-dose • pressurized metered dose inhalers (pMDIs)• dry powder inhalers (DPIs)• nasal sprays• inhalation solutions
deliver medication to the respiratory tract• lung• nasal passages
in the form of aerosol to treat diseases and conditions • respiratory (e.g., asthma, COPD, rhinitis)• systemic (e.g., diabetes, migraine)
March 2003 IPAC-RS
4
Inhalation Product HistoryInhalation Product History• First pressurized Metered Dose Inhaler (pMDI)
introduced in 1955
• Until early 1990 pMDI technology based on chlorofluorocarbons (CFC)
• CFCs linked to ozone depletion and banned by international environmental treaty (Montreal Protocol), phased out in medical use applications
• With phase-out of CFC for medical use, reformulation of pMDIs with hydrofluorocarbons (HFCs) and other new types of OINDP became necessary
March 2003 IPAC-RS
5
OINDP DDU Test HistoryOINDP DDU Test History• Regulatory requirements for DDU evolved based
on FDA experience with CFC products
• DDU testing requirements became more stringent over time – from multiple actuations per test to minimal clinical dose per test– from testing beginning of container only to testing beginning,
middle, end– from USP limits to tighter limits
• DDU requirements are more challenging for new technology from CMC perspective– Formulation flexibility significantly reduced with HFC’s due to their
physico-chemical characteristics and compatibility with excipients
– Challenging flow characteristics of powders
March 2003 IPAC-RS
6
Traditional DDU TestsTraditional DDU Tests
• USP: counting test, e.g.– 10 doses from one inhaler (3 beginning + 4 middle + 3 end)– no more than 1 of 10 doses outside 75-125% of label claim (LC)– none of 10 doses outside 65-135% LC (zero tolerance)– under certain conditions 2nd tier allowed with larger sample
• FDA: counting test with tighter limits, e.g. (through-container-life DDU test for multi-dose pMDIs/DPIs):
– 3 doses from each of 3 inhalers (3 beginning + 3 middle + 3 end)– no more than 1 of 9 doses outside 80-120% LC– none of 9 doses outside 75-125% LC (zero tolerance)– each of 3 means (B, M, E) within 85-115% LC– under certain conditions 2nd tier allowed with larger sample
March 2003 IPAC-RS
7
Reason for DDU ReplacementReason for DDU Replacement• Parametric Tolerance Interval (PTI) test is proposed as
replacement of current FDA DDU tests
BECAUSE• PTI test is more powerful and discriminating than current tests
– simultaneously uses mean and standard deviation to make quality assessment
• The FDA counting test is less efficient in use of data– unnecessarily rejects good batches
• The FDA counting test penalizes increased testing – e.g., stability testing increases chances to fail not due to product quality
change
• OINDP cannot routinely meet expectations in draft Guidances– e.g., many products have been approved with exceptions to DDU test and
acceptance criteria in published Guidances
March 2003 IPAC-RS
8
Origins of PTI Test Origins of PTI Test
• Statistical design built on previous work– 1999 AAPS/FDA/USP Workshop presentation by Dr. Walter Hauck– Williams, Adams, Poochikian and Hauck, Pharm. Res. (2002) – JP/EP/USP parametric test for dose uniformity in tablets – Features of FDA Draft Guidance test (e.g., target interval and mean
criteria)
• Acceptance criteria designed to match or exceed statistical consumer protection implied by published Draft Guidances – Metered Dose Inhaler (MDI) and Dry Powder Inhaler (DPI) Drug
Products 1998 – Nasal Spray and Inhalation Solution, Suspension, and Spray Drug
Products Draft: 1999, Final: 2002
March 2003 IPAC-RS
9
Define Batch Quality With CoverageDefine Batch Quality With CoverageCoverage = Proportion of doses in the batch that are within
target interval
Batches having same coverage of given target interval are considered to be of equal quality
Emitted DoseEmitted Dose
Target IntervalTarget Interval
March 2003 IPAC-RS
10
Hypothesis Testing FrameworkHypothesis Testing Framework
• To assure with high confidence that sub-standard batches are rejected, set statistical hypothesis as:
– H0: batch quality is out of specification
– Ha: batch quality is in specification
• Type I error: batch released but is outside specification
• Type II error: batch rejected but is within specification
March 2003 IPAC-RS
11
Hypothesis Testing Usefulness Hypothesis Testing Usefulness
• Quality of batches released to consumer is of greatest importance– Need to control risk of releasing substandard batches to
consumer (i.e., Type I error)– Type I error is controlled independently of sample size
• Therefore UNACCEPTABLE (limiting) quality is defined as baseline standard
March 2003 IPAC-RS
12
Proposed Standard of Batch QualityProposed Standard of Batch Quality
The proposed limiting quality is set at 85% batch coverage of the 75-125% LC target interval
– corresponds to 5% acceptance for FDA multi-dose test
This means that commercial batches must far exceed 85% coverage with high confidence
March 2003 IPAC-RS
13Comparison of Coverage at Comparison of Coverage at Limiting Quality Limiting Quality
between FDA and PTI testsbetween FDA and PTI tests
DDU Test Approach Minimum BATCH Coverage (95% confidence)
85% coverage of 75-125%LC
FDA for multi-dose products (Counting Test)
IPAC-RS for all OINDP(Parametric Tolerance Interval Test)
85% coverage of 75-125%LC
78% coverage of 75-125%LC
FDA for single-dose products (Counting Test)
March 2003 IPAC-RS
14
SAMPLE: pre-defined number of units (n), from different portions in container (Life Stages - LS), one dose from each unit
CALCULATE: mean (m), standard deviation (s), Acceptance Value (AV) =100-mLS + ks
COMPARE: AV 25,
100-mLS 15, and
s 25f/k (Maximum Sample Standard Deviation)
IF NOT ACCEPTED, GO TO SECOND TIER
PTI Test MechanicsPTI Test Mechanics
Emitted Dose
Target (100%LC)Mean (m)
Standard Deviation (s)
• k and f pre-defined based on n• n determined as appropriate for each product
March 2003 IPAC-RS
15
n f k1 k2
10/30 0.8366 2.152 1.60212/36 0.8254 2.001 1.55114/42 0.8189 1.900 1.51515/45 0.8156 1.860 1.49618/54 0.8067 1.765 1.45924/72 0.7967 1.649 1.41030/90 0.7887 1.577 1.375
Revised PTI Test CoefficientsRevised PTI Test Coefficients
• Test plan (sample size) selected as appropriate for each product• Consumer protection the same for all sample sizes by design• Producer risk decreases with increasing sample size
n - sample size in tier 1/ total sample size for tiers 1+ 2k1 and k2 - acceptability coefficients in 1st and 2nd tier f - factor for maximum sample standard deviation
Ensure less than 5.1% Type I error for all batch means
March 2003 IPAC-RS
16
Maximum Type I Error is 5.1% Maximum Type I Error is 5.1% (occurs for smallest sample size at mean deviation of ±9 %LC)(occurs for smallest sample size at mean deviation of ±9 %LC)
0
1
2
3
4
5
6
85 90 95 100 105 110 115
Mean, μ (% LC)
Acc
epta
nce
pro
bab
ility
(%
)
n=10/30
n=30/90
n=100/300
Normal distribution at limiting qualityNormal distribution at limiting quality(85% coverage of 100±25% LC)(85% coverage of 100±25% LC)
Sample size
March 2003 IPAC-RS
17
Issues Discussed in Previous Issues Discussed in Previous Meetings with FDA Meetings with FDA
• “Gap” in OC curves
• Quality Standard
• “Zero tolerance”
• Performance for normal and non-normal distributions
• Representative sampling from batch
• Differences between product types
March 2003 IPAC-RS
18
Progress to Date for DDU Progress to Date for DDU ReplacementReplacement
• FDA has stated that conceptually, PTI approach is acceptable
• Need to resolve: acceptance criteria to be used by PTI test
““Gap”Gap”
March 2003 IPAC-RS
20
0
10
20
30
40
50
60
70
80
90
100
Pro
bab
ility
to
Acc
ept
(%)
5%
Area of uncertainty
Consumer protection region
Producer protection region
Variability (either standard deviation at a given batch mean or coverage)
Operating Characteristic CurveOperating Characteristic Curve
0
10
20
30
40
50
60
70
80
90
100
6 8 10 12 14 16 18 20
Standard Deviation, (% of LC)
Pro
bab
ility
to
Acc
ept,
%
FDA DCU&TCL
PTI 12/36
USP 10/30
True Mean (μ)at 100% LC
• Consumer protection (Limiting Quality, LQ) same• PTI test’s curve is sharper (narrowed area of uncertainty)• Fewer acceptable batches rejected (lower producer risk)• ““Gap”: Fewer rejections does not mean lower quality of Gap”: Fewer rejections does not mean lower quality of
accepted batches accepted batches (see simulated production illustration below)(see simulated production illustration below)
Comparison of Operating Comparison of Operating Characteristic CurvesCharacteristic Curves
21
“Gap”
LQ FDA test: as in Draft MDI/DPI Guidance
0
10
20
30
40
50
60
70
80
90
100
6 8 10 12 14 16 18 20
Standard Deviation, (% of LC)
Pro
bab
ility
to
Acc
ept,
%
FDA DCU&TCL
Outlier retesting
BE Wider limits
Double doses
PTI 12/36
Examples of Effect of Deviations Examples of Effect of Deviations on current FDA OC Curveon current FDA OC Curve
• Majority of OINDP products approved by FDA in 1990-2001 have DDU test or acceptance criteria or both that deviate from draft Guidances (IPAC-RS 2001 survey)
• The ”gap” between FDA and PTI OC curves decreases with such deviations, and consumer protection is eroded
• PTIT provides reduction of producer risk without compromising consumer protectionPTIT provides reduction of producer risk without compromising consumer protection
FDA DCU & TCL test as in Draft MDI/DPI Guidance
22
March 2003 IPAC-RS
23
Simulated Illustration* Simulated Illustration* (see figures in next two slides)(see figures in next two slides)
• Unacceptable quality: FDA and PTI tests have comparable performance
• Acceptable quality: PTI test rejects fewer acceptable batches than FDA test
* Used 5000 simulations, normal distributions
March 2003 IPAC-RS
24Unacceptable Quality Batch Mean ~10014, Batch SD ~ 203
0
5
10
15
20
25
75 85 95 105 115 125
Batch mean, µ (% LC)
Ba
tch
SD
, (
% L
C)
Rejected by FDA DCU&TCL (98.8%)
0
5
10
15
20
25
75 85 95 105 115 125
Batch mean, µ (% LC)
Ba
tch
SD
, (
% L
C)
Accepted by FDA DCU&TCL (1.2%)
0
5
10
15
20
25
75 85 95 105 115 125
Batch mean, µ (% LC)
Bat
ch
SD
, (
% L
C)
Accepted by PTI test 24/72 (0.3%)0
5
10
15
20
25
75 85 95 105 115 125
Batch mean, µ (% LC)
Ba
tch
SD
, (
% L
C)
Rejected by PTI test 24/72 (99.7%)
Bat
ch S
D,
(%
LC
)
Bat
ch S
D,
(%
LC
)B
atch
SD
, (
%L
C)
Bat
ch S
D,
(%
LC
)
1.2%Accepted by FDA test(median cov= 80.3%)
0.3% Accepted by PTI test(median cov= 81.2%)
98.8%Rejected by FDA test(median cov= 76.5%)
99.7%Rejected by PTI test(median cov= 76.6%)
March 2003 IPAC-RS
25
0
5
10
15
20
75 85 95 105 115 125
Batch mean, µ (% LC)
Ba
tch
SD
, (
% L
C)
Rejected by FDA DCU&TCL (35%)
0
5
10
15
20
75 85 95 105 115 125
Batch mean, µ (% LC)
Bat
ch
SD
, (
% L
C)
Accepted by FDA DCU&TCL (65%)
0
5
10
15
20
75 85 95 105 115 125
Batch mean, µ (% LC)
Bat
ch
SD
, (
% L
C)
Accepted by PTI test 24/72 (95%)0
5
10
15
20
75 85 95 105 115 125
Batch mean, µ (% LC)
Ba
tch
SD
, (
% L
C)
Rejected by PTI test 24/72 (5%)
Bat
ch S
D,
(%
LC
)
Bat
ch S
D,
(%
LC
)B
atch
SD
, (
%L
C)
Bat
ch S
D,
(%
LC
)
65%Accepted by FDA test(median cov= 98.6%)
95%Accepted by PTI test(median cov= 98.1%)
35%Rejected by FDA test(median cov= 96.5%)
5%Rejected by PTI test(median cov= 91.5%)
Acceptable Quality Batch Mean ~1009, Batch SD ~103
Quality StandardQuality Standard
March 2003 IPAC-RS
27
Quality StandardQuality StandardQuality of a batch should be judged against a specific standard. Within presented hypothesis framework that standard is– Limiting (Unacceptable) Quality
= coverage corresponding to 5% acceptance probability= consumer protection = typical batch quality has to be far above the limiting quality to achieve
reasonable batch acceptance probability
not– Typical Batch Quality
= coverage corresponding to e.g., greater than 95% acceptance = producer risk = is different for different products
March 2003 IPAC-RS
28
Proposed Quality StandardProposed Quality Standard• IPAC-RS 2001 Proposal: Limiting quality set to
85% coverage of 75 -125% LC interval – same limiting quality as implied by Draft Guidances– demonstrated for each batch with high confidence
• FDA comment: tighter standard may be needed
• Significantly tighter standard will be problematic – in setting the standard, both producer risk and consumer protection
should be considered
– standard must be compatible with capability of current and pipeline products and analytical methodology
– if standard exceeds capability, it will create difficulties for manufacturing of current products, and development and approval of new products and generic versions
Non-Normal Distributions and Non-Normal Distributions and Zero Tolerance Criterion Zero Tolerance Criterion
March 2003 IPAC-RS
30
Non-Normal DistributionsNon-Normal Distributions• IPAC-RS database demonstrates that assumption of
normality is appropriate
• Studied non-normal distributions: symmetric short-tailed (beta) or bimodal, asymmetric short-tailed (beta and gamma) or long-tailed (outliers)
• Revised PTI test assures < 5.1 % Type I error for all normal and most non-normal distributions
• For a few extreme distributions, 5% is exceeded at the limiting quality:– Notably off-target, relatively symmetric distributions with extremely short
tails
– Notably off-target, notably asymmetric distributions with longer tail in the off-target direction
• PTI test is appropriate for real products
March 2003 IPAC-RS
31
Effect of Zero Tolerance (ZT) CriteriaEffect of Zero Tolerance (ZT) Criteria• A ZT criterion does not protect from having values
outside ZT limits in the batch
• A fixed ZT criterion degrades parametric tests;this effect escalates with sample size
• A ZT must scale with sample size in order to avoid degrading parametric test and to have no effect on producer risk
• A scaled ZT has little or no effect on consumer protection even for extreme non-normal distributions
• Any ZT penalizes thorough testing (stability, validations)
• Conclusion: ZT does not help control product quality
March 2003 IPAC-RS
32ZT Criterion Has No or Little Effect ZT Criterion Has No or Little Effect on Acceptance rate at 85% coverageon Acceptance rate at 85% coverage
Distribution type mean PTIT 10/30 PTIT 30/90no ZT w ZT=35 no ZT ZT=45
Normal 100 4.3 4.1 2.1 2.4108 5.0 5.0 4.8 4.8
Bimodal d=12 100 2.3 2.2 1.0 0.9108 1.0 1.0 0.5 0.4
Spiked p=15, l=35 100 5.7 5.6 2.7 2.5108 6.5 6.2 3.0 3.0
Sym short-tailed a=2, b=2 100 2.1 2.1 1.2 1.2108 4.0 4.0 5.1 4.9
Asymm short-tailed a=2, b=100 100 1.6 1.6 0.5 0.7108 11.3 10.4 7.9 6.5
The addition of ZT criterion does not materially improve consumer protection
March 2003 IPAC-RS
33
Desired Outcome of DDU Effort Desired Outcome of DDU Effort for IPAC-RSfor IPAC-RS
• Agree that PTI test is conceptually acceptable as a replacement– Parametric (no Zero Tolerance)– Coverage as quality definition
• Allow product-by-product justification of sample size n– multiple sampling plans, e.g., 12/36 to 30/90
• Maintain limiting quality standard implied by FDA Guidances– 85% coverage of 75-125% LC target interval– 5% acceptance (95% rejection) at limiting quality
March 2003 IPAC-RS
34
AcknowledgementsAcknowledgements• FDA / CDER / OPS
• IPAC-RS Members– Aradigm– AstraZeneca– Aventis– Boehringer Ingelheim– Eli Lilly – GlaxoSmithKline
• Members of IPAC-RS DDU Working Group
• IPAC-RS Secretariat
– IVAX– Kos Pharmaceuticals– Nektar Therapeutics– Novartis– Pfizer– Schering-Plough