Here is the presentation from the FDA at AASLD with respect to this issue.
"Consideration of New Endpoints for Regulatory Approval and Dose Selection of Hepatitis C Therapies
J. Florian1; J. Chen1; P. R. Jadhav1; J. Murray2; D. Birnkrant2
1. Division of Pharmacometrics, US Food and Drug Administration, Silver Spring, MD, United States.
2. Division of Antiviral Drug Products, US Food and Drug Administration , Silver Spring, MD, United States.
Aims:
This report summarizes the basis for new endpoints, namely SVR12 and SVR4 that can be used to support key regulatory decisions including dose selection of new hepatitis C (HCV) therapies. The current primary endpoint for approval is SVR24, which is assessed 24 weeks after stopping treatment. However, there is increasing evidence that SVR at earlier time points may be correlated with SVR24. The use of earlier time points for these key decisions could improve efficiency in a the large number of HCV therapies under development.
Methods:
The data from 10,062 subjects studied in fifteen Phase II and III trials of six drug development programs were combined to assess the concordance between SVR24 and SVR12 or SVR4. These trials were submitted to FDA through IND and NDA submissions under the Antiviral Information Management System (AIMS) initiative. The data included treatment arms of various durations with interferon, pegylated interferon with and without ribavirin (P/R), and direct-acting antivirals with and without P/R.
Results:
The positive predictive value (PPV) between SVR12 and SVR24 measurements was 98% and the negative predictive value (NPV) was 99%. In other words, two-percent of subjects with SVR12 (114/5665) had detectable HCV virus at week 24 of follow-up due to relapse, reinfection, or other reasons. Likewise, 1.2% of subjects (55/4397) achieved SVR24 but had a detectable viral load at week 12 of follow-up and would be mistakenly listed as treatment failures. The PPV between SVR4 and SVR24 measurement was 91% and the NPV was 98%. However, the effect size (difference between treatment and active control arms) was similar for both SVR4 and SVR24.
Conclusion:
SVR12 and SVR24 measurements agreed across a large population database involving multiple trials, regimens, and durations. Consideration should be given to using SVR12 instead of SVR24 as a primary endpoint for regulatory approval, noting that this approach may not be suitable in clinical practice for confirming a patient’s clinical response. The SVR4 measurement is less amenable as a primary endpoint due to overestimation of true SVR but may be useful in selecting treatment arms and designing registration trials as the estimate of drug effect is similar. These innovative analyses aid the agency in proactively addressing the public health need for additional treatments while ensuring that strict regulatory standards are met.
Concordance between SVR12 and SVR24 from Pooled Data
SVR 24
_______________ Undetectable --- Detectable
SVR12 Undetectable ---- 5551 ---- 114
SVR12 Detectable ---- 55 ---- 4342
(sorry for the ugly format of the contingency table)
Market Data 
Markets 
AI
