SLIDE 1 Evalua uation f for Sta tability ty data a Q1E Q1E
Sum umie e Yo Yosh shio ioka, a, P Ph.
D. MHL HLW Nat ation
al I Ins nstit itut ute e of
Heal alth th S Scie ienc nces es
SLIDE 2 Q1E 1E pr prov
ides s re reco comme mend ndat ation
s on
:
How to to us use sta tabil ility d data ta gene nerat ated accord rding ng to Q Q1AR AR When a and d how a re retest t per eriod o
a shel elf l life c can n be ex exten ended b beyo yond the pe perio iod cov
red by y lon
erm d data
Q1E 1E co cont ntai ains
exampl ples s of st stati tistica cal a approa
hes to
ility ty data ta an analysi sis
SLIDE 3 Extrapolation Extrapolation
to toto
exten tend d re retes test t pe perio riod/ d/sh shelf elf lif ife
Statistical approaches Statistical approaches
re reco comme mmend nded ed in in t the he gu guid idel eline ine
SLIDE 4
Significant change
No Yes Accelerated condition
SLIDE 5
Where no significant change occurs at accelerated condition
No Yes
Little or no change Little or no variability
Accelerated data & Long-term data
SLIDE 6
Where accelerated data show significant change
No Yes
Significant change
Intermediate condition
SLIDE 7
Amenable? Performed?
No Yes
Statistical analysis
SLIDE 8
Available?
No Yes
Supporting data
SLIDE 9
12 month extension
Four outcomes passing through crossroads for Room Temperature Storage
No extension 6 month extension 3 month extension
SLIDE 10 Outcome 1 12 month extension
acceler erated d dat ata sho how no s signif ifica cant ch chang nge acc ccele lerated ed da data & & lon
litt ttle or
no chan ange litt ttle or
no vari riabi bility
Outcome 4 no extension
signif ifica cant ch chang nge at a accele lerat ated co condi dition n at i interm rmedi diate c cond ndition
SLIDE 11
Stat tatist istic ical al an analy alysi sis
longer retest period/shelf longer retest period/shelf life life (not necessarily required) (not necessarily required)
SLIDE 12
amenable? performed?
Yes No 12 month extension 6 month extension
Where Accelerated data show no significant change Changes and variations in accelerated data long-term data
with Supporting data
SLIDE 13
amenable? performed?
Yes No 6 month extension 3 month extension
Where Significant change at accelerated condition but not at intermediate condition
with Supporting data
SLIDE 14
Statistical analysis can be appropriate to verify retest period/shelf life
St Stat atis isti tical al a ana nalys ysis is
longer er re retest t per eriod/s /shel elf lif ife not al alway ays req equir ired Where significant change at accelerated & intermediate conditions variability in long-term data
SLIDE 15 Sta tatis isti tica cal a app ppro roach ches es re reco comm mmend nded ed i in t the he A Appe pend ndix ix
How to to an analyze ze lo long-term data for ap appro ropriat ate q quanti titat ative a attr tribute tes How to to us use reg egres ession n ana nalysis is for re retes est per eriod
elf l life e esti timatio ion Exampl ples s of st stati tistica cal p proced edure res to det eterm rmine pool
lity of da data a from m differ erent nt batc tches es or f fact ctor co combi binatio ions
SLIDE 16 Regression analysis
Est stabl blis ish h ret etes est t per erio iod/ d/she helf lf l life fe wit ith h a a hig igh h de degre ree e of
conf nfid idenc nce Quan uanti tita tativ tive e at attri tribu bute te wi will ll r rema emain in wit ithi hin n acc ccep epta tance ce c cri riter eria ia for
all ll fu futu ture re ba batc tche hes
SLIDE 17 Shelf-life Estimation with Upper and Lower Acceptance Criteria Based on Assay at 25C/60%RH
80 85 90 95 100 105 110 115 120 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 Time Point (Months) Assay (% of Label Claim) Raw Data Upper confidence limit Lower confidence limit Regression line Upper acceptance criterion: 105 Lower acceptance criterion: 95
SLIDE 18 Statistical approaches for determining whether data from different batches/factor combinations can be pooled (Appro roach ch #1) Wheth ther r data a fro rom al all batche hes/f /factor
combina natio ions su suppo port th the propos
d perio iod (Approach (Approach #2 #2 “Poolabili Poolability ty test test”) Wh Wheth ther da data a from m all ll batc tches es/fact ctor r combin inati tions c can n be co combi bined for ov
rall es estim imate o
a sing ngle e perio iod (Alter ernat ative a appr proache hes)
SLIDE 19 Ap Appr proac
hes s #1 #1 an and d #2 #2 ca can n als lso b be e ap appli lied ed t to d dat ata a ana naly lysi sis for
mult lti-fac acto tor r stu tudi dies es in incl cludi uding ng Bra racke keti ting ng & & Ma Matr trixi xing ng De Desi sign gns
SLIDE 20
Basic Principles
A shelf life is set based on long-term data The extent of extrapolation will depend on accelerated (and if applicable, intermediate) data, as well as long-term data Supporting data are useful in predicting long-term stability in primary batches
SLIDE 21
Basic Principles (cont’d)
Statistical analysis is not always necessary for setting a shelf life A shelf life beyond the period covered by available long-term data can be proposed with supporting data, with or without statistical analysis Where a statistical analysis is performed, longer extrapolation can be justified
SLIDE 22
MHLW Perspective - Q1E
Before Q1E EU---12 month extrapolation with or without statistical analysis; US--- max 6 month extrapolation with statistical analysis; Japan--- no practical extrapolation Q1E provides guidance on the extent of shelf life extrapolation in a variety of situations Q1E clearly describes the role of accelerated data and of supporting data in shelf life estimation
