Archived Content

The content on this page is provided for reference purposes only. This content has not been altered or updated since it was archived.

Vaccines, Blood & Biologics

Bioassay Statistical Review - December 6, 2009 - Prevnar 13


Statistical Review

Date December 6, 2009

To Julienne Vaillancourt
Michael Smith

From Lev A. Sirota
Vaccine Evaluation Branch

Through Tsai-Lien Lin
Acting Team Leader, VBT

A. Dale Horne
Branch Chief

File: BLA 125324

Product: Prevnar 12TM [Pneumococcal 13-valent Conjugate Vaccine (Diphtheria CRM197 Protein)

Subject: Bioassay statistical review

Reference: BLA 125324/0.1 and BLA 125324/0.13 (Wyeth Pharmaceuticals Inc.)

c.c. Chronological File
A.Dale Horne,
Henry Hsu
Christopher Egelebo

Executive Summary:

I did not find major bioassay-related statistical issues that may prevent this submission from being approved by the agency.


This review concentrates on statistical issues in validation of bioassay in section “Analytical and Bioanalytical Methods” of BLA 125324/0.1. It also covers the applicant’s response to reviewer’s question addressed by BLA 125324/0.13
Reviewer’s comments are in italics.

Results and Comments to CBER

Polio Neutralization Assay

Method validation was performed at -----------(b)(4)--------, Inc. The Validation Protocol contains clearly defined acceptance criteria for the following parameters:

• Accuracy
• Accuracy Recovery
• Specificity
• Sensitivity
• Intermediate Precision (Between-operator)
• Intermediate Precision (Between-day)
• Repeatability
• Linearity
• Range

Experimental design was customary and appropriate. Deviations from acceptance criteria were discussed in the biological bioassay review prepared by Dr. Dragunsky.

The applicant provided analysis of linearity based on consideration of higher order terms in the regression equation. They found statistically significant second order term. Nevertheless, the % difference between the predicted values using regression equation containing linear term only and the equation containing both linear and quadratic terms is very small (less than 5%).

CBER requested the applicant to provide a scatter plot with regression lines and residuals. The applicant presents the requested material for review in the amendment BLA 125324/0.13 for visual analysis of regression and residuals.

The presented data demonstrate satisfactory agreement with linear approximation. Only residuals at extreme ((b)(4) log) demonstrate small visible deviation from linearity and only for Poliovirus III.

Pneumococcal Enzyme-linked Immunosorbent Assays

The validation of pneumococcal enzyme-linked immunosorbent assays (PnELISAs) for pneumococcal serotypes 1, 3, 4, 5, 6A, 6B, 7F, 9V, 14, 18C, 19A, 19F, and 23F to support serological testing of pneumococcal polysaccharide or polysaccharide conjugate vaccines are presented. The Validation Protocol contains the following parameters:

• Accuracy
• Precision
• Linearity
• Lower Limit of Quantitation (LLOQ)
• Upper Limit of Quantitation (ULOQ) and Range
• Specificity

Accuracy acceptance criterion is not formally set because the reference standard is run only with “------(b)(4)-----” while test samples and controls are run using “-----(b)(4)-----.”

The applicant used a factorial design to address precision, repeatability, and linearity. A variance component analysis was performed to document the total precision of the assays and to determine the relative impact of the factors “operator,” “day,” and “repeatability” on the precision of the assay.

In terms of correlation coefficient the applicant used criterion for linearity of R>-(b)(4)

This criterion appears to be too liberal. All presented linear regression data in this part of the study demonstrate coefficient of correlation of 0.99 and above. I recommend that the applicant make criterion more stringent for the future validations of the method.

The applicant investigated deviations to the method validation protocol and concluded that they have no potential impact on validation study. Deviation summary is presented.

Haemophilus Influenzae Type b (Hib) ELISA

Extensive evaluation of this method in multiple laboratories resulted in many publications in peer-reviewed journals. The objective of current validation study was to support the use of this Hib ELISA to evaluate concomitant vaccine administration for the 13-valent pneumococcal conjugate (13vPnC) vaccine . This validation report is compiled from data from the Laboratory for Vaccine Development and Research on Immunity Mechanisms using the --(b)(4)-- ELISA that have been expanded and supplemented to include data from the Laboratory for Clinical Vaccines Research. The titers of the control serum are monitored over time to assure each ELISA continues to perform consistently and remains appropriate for its intended use.

The Validation protocol includes the following parameters:
• Accuracy
• Specificity
• Linearity
• Precision
1. Intra-Assay Variation
2. Inter-Assay Variation
3. Inter-Laboratory Variation
• Range

No statistically based acceptance criteria were presented in the protocol summary therefore the extent to which this validation study can be reviewed from the statistical point of view is limited. The presented report corresponds more to an assay characterization study than an assay validation study. This may be acceptable given the method has been excessively studied previously, unless the biological reviewers express concerns regarding operational characteristics of the assay.

The accuracy was demonstrated by comparison to an accepted conventional measured value obtained by the traditional radioantigen binding assay. Linear Regression equation of ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------(b)(4)---------------- --------------------------------------------------------------------------------------------..

Repeatability was established using criterion of observed %CV(b)(4), where observed %CV was calculated based on (b)(4) repetitions within the same day and operator of (b)(4) control sera samples. The report does not contain any information indicating whether or not this and all other criteria in this section were pre-specified.

To assess intermediate precision, --(b)(4)-- %CVs were calculated for several datasets. Ability to reproduce data between laboratories was reported in the form of coefficient of correlation between measurements of the same samples obtained by -----(b)(4)-------- labs using (b)(4) serum panels with sample size --------(b)(4)------------. Coefficients of correlation between measurements produced by these labs were ------------(b)(4)-------------- correspondingly.

Bordetella Pertussis Antigens ELISAs

The intent of this validation report is to demonstrate that the methods described in SOP# 12C-ALG-37 “------(b)(4)----- ELISA for Determining Antibodies Against Bordetella pertussis PT, FHA, FIM2, FIM3, (b)(4), and PRN Subunits” can be used to evaluate concomitant vaccine administration for the 13vPnC vaccine via estimating serum antibody titers against the following Pertussis Antigens: Pertussis Toxin (PT), Filamentous Hemagglutinin (FHA), Pertactin (PRN), and Fimbriae protein (FIM). The ELISAs utilized in these studies follow methodology previously established and published by FDA.

The following parameters were included in the validation study:
• Accuracy
• Specificity
• Linearity
• Precision
• Range including Lower Limit of Quantitation
• Detection limit

The applicant did not summarize the pre-specified validation acceptance criteria in the single table, but discussed them by-sections.

The applicant tested ----(b)(4)-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------.

Equivalence test is currently considered by many statisticians as the more appropriate way to test ----(b)(4)------. Until consensus is achieved, CBER accepts both equivalence and significance tests in testing ------(b)(4)----.

The applicant’s acceptance criterion for accuracy is based on 95% confidence interval (CI) for the ratio of observed to expected titers, calculated from difference between observed titer and expected titer on a log scale. The criterion was: “The 95% CI for mean bias between (b)(4) and (b)(4).” The applicant presented data supporting the Accuracy according to ICH guidelines in Tables by antigen. Data for PT antigen demonstrates that ----------------------------------------------------------------------------------------------------(b)(4)--------------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------.

Precision is analyzed using data collected during (b)(4)-month period. Coefficient of variation for Control sera and infant sera at the lower end of each assay’s range were tabulated. The repeatability and intermediate precision were demonstrated with sufficient volume of data using an appropriate statistical method (variance component analysis per dilution).

Statistical methods used for accuracy and precision validation are adequate and customary.

The Range for PT, PRN, and FHA antigens were tabulated together with weighted %CV.

This section is lacking explanation of how each range was determined. The detailed explanation supporting range for Pertussis ELISA should be requested.

The lower limit of Quantitation (LLOQ) for each of FHA and PRN antigens was the lowest observed titer that met the acceptance criteria for Linearity, Accuracy and Precision. As PT ELISA consistently ----------------------------(b)(4)-------------------------------------, the LLOQ for PT ELISA was based solely on the acceptance criteria for Linearity and Precision. The LLOQ for FIM ELISA was derived from historic clinical data. An experimental limit of detection was calculated from ----(b)(4)---- on the standard reference curve that was (b)(4) times the assay’s background. From these data, a lower limit of quantitation was estimated as (b)(4) standard deviations above the experimental limit of detection.


1 page determined not to be releasable (b)(4)

Contact FDA

(800) 835-4709
(240) 402-8010
Consumer Affairs Branch (CBER)

Division of Communication and Consumer Affairs

Office of Communication, Outreach and Development

Food and Drug Administration

10903 New Hampshire Avenue

Building 71 Room 3103

Silver Spring, MD 20993-0002

Page Last Updated: 03/26/2010
Note: If you need help accessing information in different file formats, see Instructions for Downloading Viewers and Players.