Perform an Analytical Method

Comparability Study13:15 – 14:45, Day 2, Wed. 22 06 16

• Steven S. Kuwahara, Ph.D.

• GXP BioTechnology

•6336 N Oracle Rd #326-313

• Tucson, AZ 85704-5480

• e-Mail: s.s.kuwahara@gmail.com

Day 2, 13:15 1

Regulatory Guidance, U.S. FDA

Draft Guidance for Industry: Comparability Protocols

Protein Drug Products and Biological Products -Chemistry,

Manufacturing, and Controls Information

U.S. DHHS, FDA, CBER, CDER, CVM: September 2003.

Guidance for Industry: Comparability Protocols for Human

Drugs and Biologics: Chemistry, Manufacturing, and

Controls Information.

U.S. DHHS, FDA, CDER, CBER: April 2016 , Revision 1

• Guidance for Industry

Guidance for Industry: Analytical Procedures and Methods

Validation for Drugs and Biologics

U.S. DHHS, FDA, CDER, CBER: July 2015

Day 2, 13:15 2

Comparing Two Analytical Methods: General

• From a regulatory perspective only a procedure that has

been submitted to a regulatory agency will require a

comparability protocol (CP).

– Follow the Guidance from April, 2016 (Rev. 1) for a

format of the study.

• There are two basic types of new analytical procedures

that require a comparability study.

– A new analytical method that will be an alternative to

the existing procedure that was submitted in the

original application.

– A modification to an existing analytical method.

– A third condition is the transfer of an existing method

to a new group or site.

Day 2, 13:15 3

Comparing Two Analytical Methods: General

Criteria

• A CP to establish a new regulatory analytical procedure or

a modification of an existing analytical procedure should

include justification for the change. If the procedure will

replace an existing regulatory procedure provided in an

approved application, then the new regulatory procedure

should be scientifically sound and equivalent to or better

than the currently approved one. The CP also should

include the specific plan and acceptance criteria for

validation of the modified or new procedure. Method

validation data should be submitted with the notification of

the implemented change. For alternative analytical

procedures, comparative data to the FDA-approved

procedure should also be submitted.

Day 2, 13:15 4

Preparing for the Study that will be Reported

in the CP. I.

• Obtain copies of the original method development

study for the existing procedure and its validation

study.

– The original development study will give you the

background and basic references for the existing method.

– The original validation study and any updates or

revalidations will provide the criteria that the existing

method was able to meet.

– These will provide the criteria that the new method must

either meet or improve upon.

• Remember that there are other criteria besides the validation

criteria. Cost effectiveness for example.

Day 2, 13:15 5

Preparing for the Study that will be Reported

in the CP. II.

• These documents should have been reviewed

before the development of the new method was

attempted.

• As a part of the design control/qualification for

the new method comparable information should

have been obtained prior to its development.

• In a method transfer, what information is

available to show that a successful transfer can be

accomplished?

Day 2, 13:15 6

Scenarios: 1. Alternative Analytical Procedures

A. • An alternative analytical procedure is an analytical

procedure that you use in place of the FDA- approved

analytical procedure. For an NDA or ANDA, you should

include any proposed alternate analytical procedures in the

application. You must include a description of the

procedure. After approval, for an NDA or ANDA, or for a

procedure approved in a BLA but not included in an FDA

regulation, the addition, revision, or deletion of an

alternative analytical procedure that provides the same or

increased assurance of the identity, strength, quality, purity,

or potency of the material being tested as the analytical

procedure described in the approved application, must be

documented in the next annual report.

Day 2, 13:15 7

Scenarios: 1. Alternative Analytical

Procedures B.

• For biological products, in rare cases an analytical procedure

may be included in an FDA regulation. If the analytical method

required is described by a regulation, however, and you want to

use an alternate method, you must submit the alternate method

for review and approval according to 21 CFR 610.9(a). You

must present evidence “…demonstrating that the modification

will provide assurances of the safety, purity, potency, and

effectiveness of the biological product equal to or greater than

the assurances provided by the method or process specified in

the general standards or additional standards for the biological

product.” Modification of such procedures requires FDA

approval during application review or in a postapproval

supplement.

Day 2, 13:15 8

Scenarios: 1. Alternative Analytical

Procedures C.

• You should identify the use of the alternative analytical

procedure (e.g., release, stability testing) and provide a

rationale for its inclusion, validation data, and

comparative data to the FDA- approved analytical

procedure. You should perform an analytical method

comparability study that demonstrates at a minimum that:

– The new method coupled with any additional control measures is

equivalent or superior to the original method for the intended

purpose.

– The new analytical procedure is not more susceptible to matrix

effects than the original procedure.

Day 2, 13:15 9

Scenarios: 1. Alternative Analytical

Procedures D.

• If new process-related or product-related variants

or any new impurities are discovered with the new

procedure, testing on retention samples from

historical batches should be performed to

demonstrate that the variants/impurities detected

by the new method are a result of an increase in

the sensitivity or selectivity of the new procedure

and not a result of a change to process-related

impurities.

Day 2, 13:15 10

Scenarios: 1. Alternative Analytical

Procedures E.

• If the procedure has stability-indicating properties:

• Appropriate samples should be included that allow a comparison of

the ability of the new and original method to detect relevant product

variants and degradation species.

• The number of batches analyzed for comparison should provide

sufficient statistical power.

• Equivalence, non-inferiority, or superiority studies should be

performed with appropriate statistical methods to demonstrate that

the new or revised methods performance is comparable or better than

the original method.

• The statistical analyses performed to compare product testing should

be identified

• All bias or differences between analytical procedures seen with

comparative results should be discussed with an explanation, as

appropriate.

Day 2, 13:15 11

Scenarios: 2. Analytical Methods Transfer

Studies.

• Analytical method transfer is typically managed under a transfer

protocol that details the parameters to be evaluated in addition to the

predetermined acceptance criteria that will be applied to the results.

Transfer studies usually involve two or more laboratories or sites

(originating lab and receiving labs) executing the preapproved transfer

protocol. A sufficient number of representative test articles (e.g., same

lot(s) of drug substance or drug product) are used by the originating

and receiving laboratories.

• The comparative studies are performed to evaluate accuracy and

precision, especially with regard to assessment of interlaboratory

variability. In cases where the transferred analytical procedure is also

a stability-indicating method, forced degradation samples or samples

containing pertinent product-related impurities should be analyzed at

both sites. The USP General Chapter <1224> Transfer of Analytical

Procedures provides additional guidance on this topic.

Day 2, 13:15 12

Reporting Postmarketing Changes to an

Approved NDA, ANDA, or BLA

• Postmarketing changes to analytical procedures must be

reported to the FDA in compliance with 21 CFR 314.70 or

21 CFR 601.12.24 Additional information on the

appropriate reporting category for various kinds of

postapproval changes for NDAs and ANDAs is provided in

the 483 FDA guidance for industry on Changes to an

Approved NDA or ANDA and Changes to an Approved NDA

or ANDA; Specifications – Use of Enforcement Discretion

for Compendial Changes. Similar information on

postapproval changes to BLAs regulated by CDER and

CBER is provided in the FDA guidance: Changes to an

Approved Application for Specified Biotechnology and

Specified Synthetic Biological Products.

Day 2, 13:15 13

Statistical Procedures

• The formal statistical procedures may be found in:

– USP <1010> Analytical Data – Interpretation and

Treatment

– ASTM E1488 Standard Guide for Statistical Procedures

to Use in Developing and Applying Test Methods.

– ASTM E2782 Standard Guide for Measurement Systems

Analysis.

– ASTM E2935 Standard Practice for Conducting

Equivalence Testing in Laboratory Applications.

– There are equivalent methods to be found in the

European Pharmacopeia.

Day 2, 13:15 14

15

DEMONSTRATING COMPARABILITY

• The analyst is confronted with two situations.

– The first situation arises when a new analytical procedure must be compared with a previous method.

– The second situation is when analytical methods are transferred to a new laboratory or location, or both.

• A linear association study is useful in this case.

– It is based on the idea that if two test methods perform in identical ways, a plot of one against the other will produce a straight line with a slope of 1.0.

16

LINEAR ASSOCIATION

• If the straight line has an intercept other than zero (the origin), then one method is biased versus the other.

• This is a bit different from the usual linear regression analysis because of the fact that the independent variable must be assumed to be just as variable as the dependent variable.

• In regression analysis the dependent variable is the result of some action on the independent variable, and the independent variable was assumed to be absolute.

• In this situation, each variable is the independent result of different actions on common samples.

– The common samples bind the tests methods together.

17

LINEAR CORRELATION ANALYSIS 1.

• y = a + bx A = est. a B = est.b

• Take samples large enough to make two measurements by method X and two measurements by method Y. Each of the measurements will be considered to be different runs. Thus you will have Xa1 and Xa2; and Ya1 and Ya2. These measurements will have been made on sample a.

• Take samples b, c, d, and e or more. These samples (a - e) should have concentrations covering the expected range of test results. There will be Xb1, Xb2; and Yb1, Yb2, etc.

• Actually, more than two measurements are often taken and will give more information on method variability.

– However the measurements should be in pairs.

18

LINEAR CORRELATION 3.

• Let d = the difference between the odd and corresponding even results. (i.e. Xa1 - Xa2 and Ye1 -Ye2) . Square the differences (d2

xa, d2xb….d2

ya, d2

yb, etc. ).

• Calculate the variances as:

pairs ofnumber the

2

2

22

22

n

n

nnn

Y

YY

YXX

XX

ds

ds

19

SLOPE ESTIMATION

• The equation is: Y = a + bX

• The estimate (b) of the slope is:

2

4

2

2

22

X

Y

XY

XYXYXY

S

S

SS

SSSSSSSSSSb

20

TERMS

• The sign of b is the sign of SSxy.

n

YXXYSS

n

XXSS

n

YYSS

XY

X

Y

22

22

21

SY•X

• SY•X is the standard deviation of y on x.

222

2

1XYXY SbS

n

nS

22

REGRESSION DATA

• METHOD X METHOD Y• #Run 1 Run 2 ave. dX Run 1 Run 2 ave. dY

• 1 22 24 23 -2 19 19 19 0• 2. 24 20 22 4 24 28 26 -4• 3. 15 17 16 -2 20 16 18 4• 4. 25 27 26 -2 28 20 24 8• 5. 38 42 40 -4 35 41 38 -6• 6. 26 28 27 -2 28 24 26 4• 7. 20 22 21 -2 29 31 30 -2• 8. 39 35 37 4 27 31 29 -4• 9. 18 18 18 0 31 25 28 6• 10. 33 37 35 -4 26 30 28 -4• 11. 31 32 32 -1 37 33 35 4• 12. 29 31 30 -2 39 35 37 4• 13. 24 26 25 -2 17 19 18 -2• 14. 25 23 24 2 31 25 28 6• 15. 31 33 32 -2 40 36 38 4

23

S2Y and S2

X

• These are calculated using the averages of the actual

data points, not the differences. So n = 15.

22

22

1

1

YYXX

Y

YY

X

XX

SSSS

n

SSS

n

SSS

24

CORRELATION PARAMETERS

1333.28 0.422 2.27 0.408

0495.33667.3

2667.10

2667.1030

308

3667.330

101 2

2

2

2

2

2

YYXX

S

S

S

Sn

dS

X

Y

Y

X

25

SUMS OF SQUARES

16.064,176 60.419

40.478,11 898,11

7340.639SS 40.684

512,12 782,11

2

Y

22

XYXY

X

SSSS

n

YXXY

SS

YX

26

Calculation of Run to Run Variation

• Note: This is not the same as the d calculation.

7598.6 6952.4514

7333.639

9918.6 8857.4814

4.684

2

2

YY

XX

SS

SS

27

SLOPE CALCULATION

7297.02.839

37.612

2.839

72.20593467.447,1

2.839

615.446,242,43467.447,1

2.839

14.214763447.209481208.208773.639

b

b

b

28

INTERCEPT AND TEST

• a = 28.1333 - (0.7297x27.20) = 8.2852

602.41781.21)6654.19(077.1

8857.485325.06952.4513

14

0 13.2 4331.0131.19

2852.8

282.17602.4

2852.8

40.684

0.782,11

15

1602.4

2852.80

15,2/05.0

XY

XY

S

XS

Att

t

29

Correlation Coefficient Calculation

X.in variations

by explained is Yin variation theof 40.2% So

4021.06341.0r

exists.n correlatiot significanA 514.0

13df 6341.0

267.872,110.512,126.097,110.781,11

40.478,11898,11

22

13,05.0

r

r

r

30

REFERENCES

• STATISTICAL INFORMATION CAME FROM

SEVERAL SOURCES, BUT THE PRIMARY SOURCE

WAS:

• “PRACTICAL STATISTICS FOR ANALYTICAL

CHEMISTS” by ROBERT L. ANDERSON

• Van Nostrand Reinhold CO., New York, 1987

• The table for the significance of r values is in:

• E.S. Pearson & H.O. Hartley, Biometrika Tables for

Statisticians, Vol. 1, pp. 138 (1962).