cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
The Discovery Summit 2025 Call for Content is open! Submit an abstract today to present at our premier analytics conference.
Choose Language Hide Translation Bar
View Original Published Thread

JMP 17 Adds Key Options for Drug Stability Analysis

The current latest version, JMP 17, adds new capabilities for pharmaceutical stability testing analysis. Specifically, it introduces the option to pool and compute error variances for models with different intercepts and slopes.

 

With this option, if you select a model with different intercepts and slopes, you will see the message "Error variances calculated pooled", as shown below. Then, the validity period is calculated based on this calculation method.

 

undefined

 

In particular, JMP users in Japan who are analyzing stability studies will find this feature very useful. However, it is possible that this option is not well known, so this blog will explain the meaning and usage of this option in detail with concrete examples.

 

Compliant with description of ICH Q1E guideline

Ultimately, This option adapts JMP to the description of ICH Q1E (Guidelines for the Evaluation of Stability Data).

 

In the stability test, analysis of covariance is performed on the data of multiple lots (three lots or more), and it is classified into Model 1, Model 2, and Model 3 below. (Models 1, 2, and 3 are aligned with the terminology output in JMP reports.)

 

undefined

 

The focus here is on the case when the slope and intercept fall into model 1, which is a separate model for each lot.

 

Regarding this model, the ICH Q1E guideline states:

 

If the test rejects the hypothesis of equal slopes (ie, there is a significant difference in slopes between lots), then it is considered inappropriate to pool the data for all lots. The retest period or shelf life for each lot undergoing stability testing shall be determined by the individual ordinate intercept and individual slope, and using the mean squared error calculated from all lots , can be estimated by applying the method described in Section B.1. Choose the shortest individual lot estimate for the retest period or shelf life for all lots.

 

The part shown in red above corresponds to pooling the error variance.

 

If you do not check the option shown above, the error sum of squares obtained for each lot is used according to the specifications of the STAB macro. Until JMP 16, we could only calculate with this specification, With JMP 17, you can now pool error variances for analysis using the options shown above. In other words, it is now possible to perform analyzes in line with the descriptions of the ICH Q1E guidelines.

 

Take advantage of new options

To run a drug stability study analysis, from the JMP menu bar Analyze > Reliability/Survival Analysis > Degradation Analysis from the top left tab in the Select Columns dialog "Stability test" Choose.

 

in the lower left "Pool error variances for models with separate intercepts and slopes" is the new option. This option is unchecked by default, but if you plan to apply for drug approval using JMP reports, you should check this option.

 

undefined

 

A report of the degradation data analysis is displayed. In this example, model 1 (separate intercept, separate slope) is selected based on the above flow, It shows how the error variance was calculated by pooling it as shown in red on the right.

 

 

undefined

 

This example specifies a lower specification limit for concentration of 96. Therefore, the shortest time (17.986) to intersect concentration=96 on the Y-axis is shown as the validity period.

 

Try calculating the validity period in "Fit Model"

Degradation analysis is characterized by the ability to automatically select a flow-based model and indicate the effective period for that model. I'll try using Fit Model.

 

1.Select Analyze > Fit Model and specify 'Y' and 'Construct Model Effects' as follows:

 

Lot Number*Time (Months) shows the interaction effect of Lot Number and Time (Months). By including this effect, we are fitting a regression model that estimates the slope separately and the intercept separately.

 

undefined

 

A report of the model fit is displayed. Looking at the Regression Plot, we can see that a straight line with a separate slope and intercept has been fitted to each lot.

 

2.Select Estimates > Inverse Estimates from the red triangle at the top left of the report. Here, change the check to "All" for (Prediction target), change it to "Lower one-sided", and enter "96" as the lower specification limit for concentration.

 

undefined

 

In the inverse estimation report, the estimated value of time (month) and the lower 95% of the inverse estimation for each lot number are displayed.

 

undefined

 

In this example, the lower 95% of lot number (4_14) is 17.986, which agrees with the value calculated in the degradation analysis stability test.

 

Technical documentation for analysis of stability test data (Japanese)

The JMP Japan Division provides a summary of technical materials on JMP's stability test functions.

If you are interested, please refer to the commentary on model selection when there are multiple packaging forms, accelerated tests using the Arrhenius model, etc.

 

Analysis of Stability Test Data Using JMP -For Determining Shelf Life of Drugs- (PDF) Application | JMP

 

by Naohiro Masukawa (JMP Japan)

This post originally written in Japanese and has been translated for your convenience. When you reply, it will also be translated back to Japanese.