PFS preferred regulatory endpoint in oncology trials, AstraZeneca exec says.(Endpoints)(progression free survival )

ARLINGTON, VA -- Progression-free survival (PFS) is the preferable regulatory endpoint in oncology clinical trials over time-to-progression (TTP) endpoints because PFS allows for smaller trial designs, and PFS includes deaths and better correlates with overall survival, a global product statistician with AstraZeneca stated here Sept. 17 at the CBI conference.

Renee Iacona, Ph.D., told the clinical trial summit that "PFS assumes deaths are randomly related to tumor progression."

She said progression-free survival is defined as a measurement in a clinical trial to help determine whether a new treatment is effective. It refers to the probability that a patient will remain alive, without the disease getting worse.

In cancer trials with a PFS endpoint, the goal is to stop the progression of the diseases, not necessarily to shrink the tumor.

Randomization can control for differences in population and such trial designs are becoming more widespread.

She said that PFS is needed in oncology studies because areas of high unmet need, such as, cancer studies require "creative and flexible approaches to trial design."

These designs, which include progression-free survival endpoints, must be robust, reliable and produces rapid trial results.

Also, "new therapies are targeted and efficacy may best be seen with delaying progression rather than shrinking tumors."

Iacona said that ways to look at PFS for randomized Phase II studies are:

* Comparison of HR based on full PFS assessment;

* Comparison of HR based on progression event count; and

* Comparison of PFS at a predetermined time point.

Also, PFS does not require frequent assessments. "If you assess patients at a frequency that is half of median, there is negligible loss in power," she added.

With event count, PFS causes only a "slight loss of sensitivity." The event count analysis requires little increase in sample size when less than 75% of patients have an event, she said.

The speaker said that considerations for using progression-free survival as an approval endpoint include:

* PFS only known to have occurred within an interval;

* How often should patients be assessed;

* Minimize the potential for time assessment bias;

* Censoring--think about the assumptions;

* Ascertainment bias--will an independent review get closer to the truth; and

* Use of PFS has resulted in extensive discussions with FDA regarding protocol.

In the use of PFS, she said that censoring of patients can occur as a "statistical process for handling of patients who don't progress." Censoring "reduces the power of analysis, which depends on the number of events."

Censoring also can cause bias, if rate differs between arms and is related to patient prognosis.

Patients who could be censored in a PFS study, Iacona noted, include:

* Patients who miss multiple visits; and

* Patients who stop therapy or start other therapy prior to progression.

For patients who stop therapy, she noted that rates are likely to depend on relative tolerability. "Decisions on patient management likely to be related to prognosis."

Iacona said that independent review is common for PFS studies, but does introduce some problems:

* Distance from the management of the patient;

* Analysis problems--handling of patients that the investigator declares as a progression but the independent reviewer disagrees;

* Helpful in open trials but not always necessary in a double blind trial; and

* If FDA agrees to PFS as the primary endpoint in a study, you will be required to use independent review.

By Joseph Pickett Managing Editor

Print Get the Mag Weekly Updates