Evidence for pCR

FDA's draft guidance on the use of pCR as a surrogate endpoint cites several studies that suggest a correlation between the laboratory measure and clinical endpoints like disease-free survival (DFS) and overall survival (OS).

For example, in the NSABP B-18 trial comparing preoperative and postoperative delivery of four cycles of doxorubicin plus cyclophosphamide, patients in the preoperative arm who attained a pathologic complete response had about a 68% reduced risk of death (HR: 0.32, p<0.0001).

The NOAH trial of Genentech Inc.'s Herceptin trastuzumab in neoadjuvant breast cancer showed a pCR rate in the Herceptin plus chemotherapy arm of 38% vs. 19% for chemotherapy alone (p=0.001). With 3.2 years of median follow-up, the three-year DFS was 71% for Herceptin plus chemo vs. 56% for chemotherapy alone (p=0.013).

There was no statistically significant difference in OS, but fewer deaths occurred in the Herceptin arm (18 vs. 26; HR=0.62).

According to Philippe Bishop, VP of clinical development for Genentech's Avastin bevacizumab, the Roche unit uses pCR to screen new agents because of this correlation between pCR and survival.

Bishop would not disclose the pCR rate a compound must achieve for Genentech to advance it into Phase III. "Internally, we are looking for big differences that our own modeling predicts will have an impact on disease-free and/or overall survival," he said.

Mark Benyunes, senior group medical director for clinical hematology/oncology, cited the company's Perjeta pertuzumab as an example.

"NEOSPHERE told us that pertuzumab was likely to be very active in early breast cancer, and we then designed the large adjuvant study APHINITY," he said.

In the Phase II NEOSPHERE trial in neoadjuvant breast cancer, patients on Perjeta plus Herceptin and docetaxel had a pCR rate of 45.8% vs. 29% for patients on Herceptin plus docetaxel (p=0.014).

In June, FDA approved Perjeta in combination with Herceptin and docetaxel in first-line metastatic, HER2-positive breast cancer.

According to Debasish Tripathy, professor of medicine at Keck School of Medicine, there is no specific pCR threshold for I-SPY 2, an adaptive trial that is evaluating several agents in neoadjuvant breast cancer.

The investigators will use Bayesian statistics to analyze the pCR rates and the patients' genomic signatures to determine a drug's likelihood of success in the Phase III trials. Drugs predicted to have at least an 85% likelihood of success will graduate.