A comparative study of confidence intervals to assess biosimilarity from analytical data

Abstract

Assessment of analytical similarity of tier 1 quality attributes is based on a set of hypotheses that tests the mean difference of reference and test products against a margin adjusted for standard deviation of the reference product. Thus, proper assessment of the biosimilarity hypothesis requires statistical tests that account for the uncertainty associated with the estimations of the mean differences and the standard deviation of the reference product. Recently, a linear reformulation of the biosimilarity hypothesis has been proposed, which facilitates development and implementation of statistical tests. These statistical tests account for the uncertainty in the estimation process of all the unknown parameters. In this paper, we survey methods for constructing confidence intervals for testing the linearized reformulation of the biosimilarity hypothesis and also compare the performance of the methods. We discuss test procedures using confidence intervals to make possible comparison among recently developed methods as well as other previously developed methods that have not been applied for demonstrating analytical similarity. A computer simulation study was conducted to compare the performance of the methods based on the ability to maintain the test size and power, as well as computational complexity. We demonstrate the methods using two example applications. At the end, we make recommendations concerning the use of the methods. KEYWORDSbiosimilarity, confidence interval, exact-based intervals, generalized confidence intervals, Howe's method, Wald-type confidence intervals 316 317 have no clinical impact. If this information is unavailable, the Food and Drug Administration (FDA) recommends using = f × R , where the constant f is a fixed multiplier, and R is the variability of RP. The value of f = 1.5 is justified using power-based calculations as rationalized in Tsong et al 1 and Chow. 2 In this case, the set of hypotheses in Equation 1 becomes:( 2) Currently, analytical similarity is demonstrated using a two one-sided tests (TOST) of the hypotheses in Equation 2. The TOST is conducted by computing a 100(1 − 2 )% confidence interval on the difference T − R , and equivalence is demonstrated if the entire confidence interval falls within the range from −1.5 R to 1.5 R . In this setting, is the test size. However, in practice, R is unknown and is replaced in the EAC with a sample estimate. This approach does not account for the uncertainty in the estimation of R and is not recommended because it inflates the type I error and reduces the powers of the TOST. 3 Alternative approaches have been explored to address this issue.Two reformulations of the set of hypotheses in Equation 2 that facilitate development and implementation of statistical procedures to account for the uncertainty in the estimations of all the parameters, and in particular R , have been recently proposed in the literature. Burdick et al 4 reformulated the set of hypotheses in Equation 2 as effect size by dividing both sides of the inequalities by R...

Analytical Similarity Assessment of Biosimilars: Global Regulatory Landscape, Recent Studies and Major Advancements in Orthogonal Platforms

“…Along the same line of thought, statistical alternatives to similarity evaluation in cases where multiple reference products exist (EU-approved and US-licensed) have been proposed such as the use of simultaneous confidence approach as a possible means to reconcile the biosimilarity between the reference products as using it as a whole in comparison to the intended biosimilar ( Zheng et al, 2019 ) ( Supplementary Table S7 ). Studies comparing test procedures using confidence intervals for recently developed methods as well as other previously developed methods applied for demonstrating analytical biosimilarity have also been published ( Burdick et al, 2017 ; Wang and Chow, 2017 ; Quiroz et al, 2019 ; Chen and Hsiao, 2020 ) ( Supplementary Table S7 ). Comparative signature diagrams (CSDs) has been explored as an orthogonal visualization technique for analytical data sets in the form of colored contour plots ( Kim et al, 2016 ) ( Supplementary Table S7 ).…”

Contribution of Steel Fiber to the Mechanical Property Improvement of C80 Concrete Produced with a High Amount of Artificial Sand Powder

“…At present, the comparative method of arithmetic averages of fundamental mechanical characteristics for concrete prepared with various constituents is the main method [27,28]. From the method, in the experimental test results, as shown in Table 14, the axial compressive intensity of RP10-R0.4-83 is less than that of CP3, CP5 and CP7; its flexural intensity is greater than that of CP3, CP5 and CP7; its splitting tensile strength is greater than that of CP3, CP5 and CP7, and its elastic modulus is greater than that of CP7 and less than that of CP3 and CP5.…”

“…However, note that, at present, the influence of steel fiber content in comparison to RP content on the mechanical characteristics of concrete is mostly studied from an experimental perspective. The average experimental value method [27,28] is normally used to assess the experimental results, resulting in large divergences due to the obvious influences of different RP contents on different mechanical characteristics. In the literature, there is a lack of statistical theory to provide an alternative analysis of the influence of steel fiber content on RP content.…”

“…The possibility of the testing value being greater than the characteristic value is 91% in America [30]. Although a comparative approach of experimental arithmetic averages of mechanical characteristics for concrete prepared with various constituents has been considered [27,28], the comparison of characteristic values by means of possibility density distributions is close to real values and meets practical engineering applications [29,30].…”