The challenge we have in considering this hypothesis [higher pathologic complete response rate will be associated with superior long-term outcomes] is simply that the available validation is limited and the potential for long-term error remains.
—Clifford A. Hudis, MD, FACP
Pathologic complete response as assessed surgically after neoadjuvant treatment is being touted by some researchers as a stand-alone endpoint justifying early drug approval for breast cancer. They argue that it provides a more efficient means of testing the value of agents that might be useful in the adjuvant setting and as a surrogate endpoint justifying accelerated drug approval. The FDA has since responded to a favorable vote by the Oncologic Drugs Advisory Committee (ODAC) in support of using pathologic complete response in labeling pertuzumab (Perjeta) for neoadjuvant use. The important and urgent overarching goal is to improve the sometimes cumbersome drug approval process. But ASCO President Clifford Hudis, MD, FACP, Chief of the Breast Cancer Medicine Service at Memorial Sloan-Kettering Cancer Center and Professor of Medicine at Weill Cornell Medical College in New York—assigned the “con” position during an education session at the 2013 Breast Cancer Symposium—noted that using pathologic complete response for drug approval carries potential risks that investigators should consider as more drugs are possibly brought forward on this pathway.
At the 2013 Breast Cancer Symposium, Dr. Hudis noted the increasing role that neoadjuvant trials play in the clinical trials portfolio. Results can be obtained less expensively and sooner than with later-stage trials. At the same time, he expressed concern that preoperative trials may sometimes overstate the benefits of novel therapies in terms of long-term endpoints, and early approval has the potential to complicate and compromise the conventional, off-study treatment of patients without delivering long-term benefits.
“Clinicians and clinical investigators need to appreciate some potential limits of using this approach,” he cautioned. “By aiming for efficiency and trying to change the clinical trials paradigm, some developers may think they see an open door for accelerated approval for drugs. But limited data sets may not be sufficient to support this. Pathologic complete response is, of course, a response endpoint and it has not always been an accurate surrogate of long-term benefit. In advanced disease, high-dose chemotherapy produced very high response rates but did not deliver long-term benefits. Small unconfirmed studies can also mislead. Remember that while the second randomized trial of bevacizumab was positive and led to accelerated approval in the metastatic stetting, subsequent studies and a meta-analysis failed to confirm an overall benefit. Hence, a drug's activity, as measured by surrogates, can be impressive but may not be consistent nor correlated with long-term benefit.
In the most current example of pertuzumab, not only was pCR increased, but this followed a demonstration of remarkably improved overall survival in the metastatic setting as well as the completion of a conventional adjuvant trial. The point being that pCR may enable the early availability of the drug, but in this setting it was accompanied by the same depth and extent of data as has been required in the past. An unsolved remaining problem is that forthcoming survival data will be based on a year of adjuvant treatment in the APHINITY study. Clinicians who chose to give "on-label" pertuzumab for 12 or 18 weeks may, or may not, be delivering a survival benefit but will not know for a long time, if ever.”
Validation of Pathologic Complete Response Limited
In the neoadjuvant setting, the hypothesis is that between two potential treatment options, the one yielding the higher pathologic complete response rate will be the one associated with superior long-term outcomes, specifically disease-free survival and overall survival. “The challenge we have in considering this hypothesis is simply that the available validation is limited and the potential for long-term error remains,” said Dr. Hudis.
While it has been established that among similarly treated patients, those achieving pathologic complete responses have a better prognosis than those who do not, pathologic complete response has not been fully validated as a surrogate for long-term outcomes, nor has such an association been observed in all studies and tumor subtypes. For example, in NSABP B27, the addition of docetaxel (Taxotere) to neoadjuvant doxorubicin (Adriamycin)/cyclophosphamide (Cytoxan) nearly doubled the pathologic complete response rate but did not improve disease-free and overall survival.1 For an approval based on pathologic complete response it is necessary to know what magnitude of improvement will be associated with long-term benefit, but a recent meta-analysis of the Collaborative Trials in Neoadjuvant Breast Cancer (CTNeoBC) failed to provide answers. The study, which included 12 randomized neoadjuvant traisl and nearly 13,000 patients, found that patients achieving a pathologic complete response had a 52% reduction in events (P < .001) and a 64% reduction in the probability of death (P < .001), but the magnitude of pathologic complete response improvement that predicted event-free and overall survival benefit could not be established.5
"Is it, for example, a 7% absolute increase or a 14% absolute increast? By the FDA's own analysis, we don't see a simple association," said Dr. Hudis.
Breast Cancer Subtypes Can Confound the Effects
Furthermore, Dr. Hudis pointed out that the molecular heterogeneity of breast cancer creates complexity in the neoadjuvant setting that is not always recognized. For example, different subtypes may have different sensitivities to new agents. Several neoadjuvant studies recently demonstrated that pathologic complete response rates after anti-HER2 therapies vary according to estrogen receptor/progesterone receptor status. In CALGB 40601, reported at 2013 ASCO Annual Meeting, pathologic complete response rates after dual HER2 blockade reached 77% in hormone receptor-negative patients, but were just 42% in hormone receptor-positive patients.2 And in GeparSixto, also reported at ASCO, the addition of carboplatin (Paraplatin) to paclitaxel (Taxol) and non-pegylated liposomal doxorubicin led to pathologic complete responses in almost 60% of patients with triple-negative breast cancer but failed to increase responses in HER2-positive patients.3 Other important modulators of response may not be recognized until later.
Even knowing this, trial results may not be completely confirmatory. In the Neo-ALLTO trial, which evaluated lapatinib (Tykerb), trastuzumab (Herceptin) and the combination, it was less clear that incremental improvements in pathologic complete response varied by hormone receptor status. But importantly, the neoadjuvant results failed to predict the eventual inferiority of single-agent lapatinib in the adjuvant setting, according to early analysis from the ongoing ALLTO study.4
While stratification of the population will help sort these things out, there may be additional factors that could influence pathologic complete response rates for specific targeted agents. “We need to acknowledge the potential for unrecognized imbalances in the typically smaller neoadjuvant studies testing new agents,” Dr. Hudis noted.
Where Do We Stand?
“Change in pathologic complete response rate is not yet an established stand-alone surrogate endpoint and we do not have sufficient experience to validate it. We still need a validated endpoint for the regular approval of drugs. The ODAC vote in favor of pertuzumab’s approval therefore relied heavily on the availability of an unprecedented overall survival improvement in the metastatic setting, as well as the existence of a completed, but unreported adjuvant trial,” Dr. Hudis said. “A similar triad of impressive overall survival data in metastatic disease and completed adjuvant studies seems unlikely to recur anytime soon.
Dr. Hudis made it clear that he strongly supports studies in the preoperative stetting as well as the development of a pathway for preoperative drug approval. “I run, conduct, design, and strongly support preoperative studies,” he said. “They are a good screen for a drug’s activity, may ultimately indicate improvements in long-term outcomes, and can provide informative imaging and tissue correlates,” he said. “When we have the rapid acquisition of data, as occurred with pertuzumab, accelerated approval may be justified. But this should not be misunderstood as evidence that [pathologic complete response] stands alone as a surrogate. Instead, the pCR data validated what was already known from the metastatic setting.” ■
Disclosure: Dr. Hudis reported no potential conflicts of interest.
1. Rastogi P, Anderson SJ, Bear HD, et al: Preoperative chemotherapy: Updates of National Surgical Adjuvant Breast and Bowel Project Protocols: B-18 and B-27. J Clin Oncol 26:778-785, 2008.
2. Carey LA, Berry DA, Ollila D, et al: Clinical and translational results of CALGB 40601: a neoadjuvant phase III trial of weekly paclitaxel and trastuzumab with or without lapatininb for HER2-positive breast cancer. 2013 ASCO Annual Meeting. Abstract 500. Presented June 2, 2013.
3. Von Minckwitz G, Schneeweiss A, Salat C, et al: A randomized phase II trial investigating the addition of carboplatin to neoadjuvant therapy for triple-negative and HER2-positive early breast cancer (GeparSixto). 2013 ASCO Annual Meeting. Abstract 1004. Presented June 3, 2013.
4. Baselga, J, Bradbury I, Eidtmann H, et al: Lapatinib with trastuzumab for HER2-positive early breast cancer (NeoALTTO): A randomized, open-label, multicentre, phase 3 trial. The Lancet, 379(9816):633-40.
5. Cortazar P, Zhang L, Untch M, et al: Meta-analysis results from the Collaborative Trials in Neoadjuvant Breast Cancer. 2012 San Antonio Breast Cancer Symposium. Abstract S1-11. Presented December 5, 2012.