The Role of Neoadjuvant Trials in Drug Development for Solid Tumors

The primary objective of phase II trials in oncology has been to determine whether a novel drug has sufficient antitumor activity to warrant further investigation in phase III trials. Because of the cost and number of patients required, the failure of a phase III trial has detrimental financial and human ramifications. Most observers have concluded that phase II studies in oncology have had a poor track record in predicting success in phase III trials as well as eventual regulatory approval (1). By one estimate, only 57% of oncology drugs taken from phases II to III obtain FDA approval, a proportion notably lower than for non-oncology drugs (2). The likelihood that a positive phase II trial of combination therapy will result in a subsequent trial that improves the standard of care within 5 years has been reported to be only 0.038 (3).

This low predictive value of phase II trials may be due to reliance on the endpoint of response rate (RR) as defined by the RECIST (2). RECIST was based on work done decades ago to standardize tumor responses in the absence of modern cross-sectional imaging techniques (4, 5). Criteria for partial responses (PR) were based on the precision with which oncologists could differentiate solid spheres of different sizes under a layer of foam rubber (4), not based on clinical outcomes. There was never any claim that the PR criteria were correlated with clinically meaningful outcomes. In fact, the original WHO response criteria (5) and later RECIST (6) stressed that “it is not intended that these RECIST guidelines play a role in…decision making, except if determined appropriate by the treating oncologist” (6). Therefore, it is not surprising that RR has not been a reliable endpoint for phase II trials (7–9).

It seems reasonable to speculate that pathologic complete response (pCR) might correlate more strongly with overall survival (OS) than RRs defined by RECIST. pCR is a biologically rational reflection of a therapy's ability to eradicate metastatic disease, and may therefore serve as a surrogate for OS. Because the number of tumor cells in undetectable micrometastases is many logs lower than the cell number in clinically evident tumors, even pathologic responses less than complete eradication of the detectable tumor might correlate with improved OS. The strategy of neoadjuvant therapy (treatment before complete surgical resection) permits investigators to assess tumor response on a pathologic basis. In this setting, it may be possible to assess treatment benefit in a more meaningful way than in traditional phase II trials using RECIST. In this review, we will consider the use of neoadjuvant trials as a way to identify treatments perhaps more likely to improve OS and as a strategy for drug development.

Different measurements of pathologic response after neoadjuvant therapy have been demonstrated to correlate with OS across various solid tumors (Table 1). The most extensive experience using neoadjuvant therapy is in breast cancer. A strong association between pCR and survival has been demonstrated in many multi-institutional, randomized neoadjuvant trials of chemotherapy in early-stage breast cancer (10–15). However, the definition of pCR varied across these neoadjuvant breast cancer studies and the relationship between pCR and long-term benefit was not always clear. To address these challenges, the FDA performed a pooled analysis of nearly 13,000 patients enrolled in neoadjuvant breast cancer trials (16). The eradication of invasive cancer in both the resected breast tissue and regional lymph nodes was found to correlate more strongly with improved long-term outcomes than was tumor eradication in the breast alone. In addition, the pooled analysis found individual patients who attain a pCR had a 64% reduction in the risk of death compared with patients who did not. At the individual trial level, only a weak association was evident between increases in the proportion of patients achieving a pCR and the ability of treatment to improve OS. However, the heterogeneous patient populations, the low overall rates of pCR, and the lack of targeted therapy in the trials included in the analysis can explain this finding (17). Although the FDA ultimately concluded that pCR meets the surrogate endpoint criterion of being “reasonably likely to predict clinical benefit” (18), future pooled analyses of targeted therapy trials in biomarker-defined breast cancer subtypes could help solidify pCR as a surrogate endpoint for long-term outcomes (19).

In neoadjuvant bladder cancer trials, pCR, less than CR, has been associated with improved OS (20). In an analysis of 147 patients by Splinter and colleagues (21), patients with muscle-invasive bladder cancer (MIBC) whose disease was downstaged to no muscle invasion (<pT2) after neoadjuvant chemotherapy experienced a 75% survival rate at 5 years compared with 20% survival for those whose tumors still showed muscle invasion (≥pT2 residual disease). The equivalency of pCR and <pT2 in predicting OS after neoadjuvant chemotherapy for bladder cancer was confirmed in a prospective trial (22) and a retrospective analysis (23). The development of the pCR for bladder cancer highlights the importance of considering the disease biology when defining the endpoint. Neoadjuvant phase II trials in MIBC are a model for rational drug development and can use <pT2 as a surrogate endpoint (24).

In other solid tumors, where eradication or significant downstaging of disease after neoadjuvant therapy is a rare event, investigators have used histopathologic methods to categorize response and correlate them with outcome. In stage IB-IIIA non–small cell lung cancer (NSCLC), the median rate of pCR from 15 trials of neoadjuvant chemotherapy was only 4% (range, 0–16%; ref. 25). Pataer and colleagues (26) developed a technique for assessing response to neoadjuvant chemotherapy based on mean a percentage of residual viable tumor cells taken from sampled tissue. In a comprehensive tissue analysis of 192 patients with resected stage I–IV NSCLC given neoadjuvant chemotherapy, a cutoff value of ≤ 10% viable tumor was associated with improvement in OS (HR, 2.39; P = 0.05 if >10% viable tumor) on multivariate analysis (26, 27). These methods of evaluating pCR (26) were applied prospectively in a trial of 50 patients with stage IB–IIIA NSCLC given neoadjuvant chemotherapy and bevacizumab (28). Of the 22% of patients with ≤10% viable tumor, 100% were alive at 3 years compared with only 49% of those who had >10% residual tumor (P = 0.01); this remained statistically significant after adjustment for stage (P = 0.02). Although validation in larger studies across NSCLC histologies is needed, Hellman and colleagues (25) propose that ≤10% residual tumor in resected lung and lymph node tissue should be regarded as a surrogate of OS in patients with resectable NSCLC given neoadjuvant chemotherapy.

These experiences, and others not discussed here (29–31), indicate that neoadjuvant treatment can lead to improved OS depending on how response is defined. Although some trials used pCR as the endpoint, pathologic responses, less than complete responses, may be associated with improved survival in certain tumor types.

Neoadjuvant administration of systemic therapy has the following potential benefits to patients: (i) tumor cytoreduction leading to improved surgical outcome or, in some cases, less radical surgery; (ii) sooner treatment of systemic metastases without the possible barrier of postoperative complications; (iii) ability to determine, in a short period of time, if the tumor is sensitive to the systemic therapy. Indeed, neoadjuvant therapy is used as a standard of care for some tumors in carefully selected patients (29, 32–34).

In early-stage breast cancer, multiple studies have shown that neoadjuvant therapy improves OS and may also reduce the extent of local surgery required (10, 35–37). The multidisciplinary approach required and the current controversies have been reviewed recently (32). Neoadjuvant therapy appears to be more beneficial in aggressive subtypes such as triple-negative and HER2-positive tumors that are more chemosensitive and have higher pCR rates (15, 38).

Neoadjuvant cisplatin–based chemotherapy followed by radical cystectomy is a standard of care for MIBC. Randomized neoadjuvant trials using MVAC (methotrexate, vinblastine, doxorubicin, and cisplatin; ref. 20) or CMV (cisplatin, methotrexate, and vinblastine; ref. 39) before radical cystectomy have both been shown to improve survival compared with cystectomy alone. Patients receiving neoadjuvant MVAC had superior disease specific survival [HR, 1.66; 95% confidence interval (CI), 1.22–2.45; P = 0.002] and a trend toward superior OS (HR, 1.33; 95% CI, 1.00–1.76) compared with patients who were managed with surgery alone, with OS rates of 57% and 43% at 5 years, respectively (P = 0.06). The neoadjuvant CMV trial reported that neoadjuvant CMV was associated with a 16% relative improvement in survival (P = 0.037) and a 23% relative improvement in metastasis-free survival (P = 0.0001) at 10 years. Given that gemcitabine and cisplatin (GC) is better tolerated and achieves similar survival rates to those from MVAC in the metastatic disease setting (40), it is frequently used as a substitute for MVAC in the neoadjuvant setting and has similar pathologic RRs (41). In a meta-analysis of over 3,000 patients with MIBC, there was a 5-year OS benefit seen with neoadjuvant cisplatin–based chemotherapy followed by radical cystectomy versus radical cystectomy alone (HR, 0.86; 95% CI, 0.77–0.95; P = 0.003; ref. 33).

In patients with resectable NSCLC, neoadjuvant chemotherapy can improve OS. In the 1990s, two small, randomized trials were terminated early on the basis of an interim analysis showing significant improvement in survival for neoadjuvant chemotherapy followed by surgery versus surgery alone (42, 43). In a meta-analysis performed by Song and colleagues (34) of 13 randomized controlled trials patients with stage IB–IIIA NSCLC who received neoadjuvant chemotherapy had improved OS compared with patients who were managed with surgery alone (combined HR, 0.84; 95% CI, 0.77–0.92; P = 0.0001). These results were similar to those from a prior meta-analysis, in which a combined HR of 0.82 (95% CI, 0.69–0.97) was obtained. A more recent phase III trial in patients with stage IB–IIIA NSLCC randomly assigned patients to surgery alone or surgery plus neoadjuvant CG (44). The HR for OS was 0.63 (95% CI, 0.43–0.92; P = 0.02), favoring the neoadjuvant chemotherapy arm. In addition, patients with stage IIB/IIIA disease had a survival benefit of 23.4% at 3 years.

Randomized trials of neoadjuvant chemotherapy in other tumor types have demonstrated an improvement in OS when compared with surgery alone (29, 30, 45, 46). Chemoradiotherapy is also routinely used in the neoadjuvant treatment of some solid tumors (47–49). These experiences further support the notion that pCR in the neoadjuvant setting can serve as a surrogate for OS.

In May 2012, the FDA published a draft guidance outlining the reasons for its acceptance of pCR as a surrogate in early-stage breast cancer and proposing a framework for its use in clinical trials (50). In September 2013, the FDA granted accelerated approval to pertuzumab for use in combination with trastuzumab and docetaxel as neoadjuvant treatment of patients with HER2-positive, early-stage breast cancer. The approval was based on the NeoSphere trial, which demonstrated an improvement in the pCR rate seen with pertuzumab and trastuzumab plus docetaxel compared with trastuzumab plus docetaxel (45.8% vs. 29.0%; P = 0.0141; ref. 51).

Although an improvement in pCR ultimately led to the accelerated approval of pertuzumab, the other factors that were crucial to the FDA decision reveal important, generalizable considerations for using neoadjuvant therapy as a platform for drug development (Fig. 1). First, patients with HER2-positive breast cancer are at high risk for relapse with standard therapy (15). Second, the safety profile of pertuzumab alone and in combination with standard-of-care chemotherapy and trastuzumab had been reported in nearly 10,000 patients (17). Third, the FDA had conducted the aforementioned pooled analysis of neoadjuvant trials to establish the definition of pCR and to support the relationship between pCR and OS at the patient level (16). Finally, the adjuvant APHINITY confirmatory trial was fully accrued and well under way at the time of accelerated approval, complying with the FDA requirement for a postmarketing clinical trial to verify meaningful clinical benefit (50).

The bar for the first approval of a drug based on pCR was set high, but has provided an impetus to pursue this pathway of expedited drug development in breast cancer and other solid tumors. Indeed, once a pCR assessment has been validated as a surrogate in randomized clinical trials, it can be used as an endpoint in nonrandomized trials as well to screen drugs for antitumor activity based on results from historical controls. For example, two recent, single-arm, phase II trials in MIBC have evaluated dose-dense MVAC (ddMVAC) in the neoadjuvant setting (52, 53). On the basis of the pathologic RRs seen in these trials, ddMVAC has been carried forward to randomized clinical trials in the neoadjuvant setting (NCT02177695 and NCT01812369).

Advances in molecularly targeted therapy and immunotherapy have led to an explosion of new drugs for cancer patients. Despite this progress, the field remains reliant on RECIST, an antiquated method of measuring responses never designed to correlate with clinically meaningful outcomes. We speculate that pCR, as assessed in neoadjuvant trials, might be expected to correlate more strongly with OS. Indeed, pCR has already been endorsed by the FDA as a surrogate for OS in early-stage breast cancer (50). However, patient selection will remain important, and the few patients who are thought to have a brief window of opportunity for curative surgery may not be appropriate for neoadjuvant treatment with experimental therapies. Conversely, if a neoadjuvant therapy has already been shown to improve OS, new drugs may need to be tested as add-ons to the standard therapy.

Another advantage of neoadjuvant trials is the availability of pre- and posttreatment tumor tissue for study. Investigation facilitated by the neoadjuvant paradigm has already contributed to our understanding of how tumor genomics and the immune microenvironment can serve as predictive biomarkers of response in solid tumors (Table 2). In patients who do not respond to treatment, the posttreatment tumor can provide critical information about mechanisms of treatment failure. Tumors can be studied for intrinsic resistance or for the failure of the drug treatment to hit the intended molecular target. For immunotherapy treatments, studies of the tumor environment are likely to be very important. The ability to assess the activity of a drug in vivo and to interrogate posttreatment tissue will aid in the discovery of response and resistance mechanisms that can guide the next generation of clinical trials.

S.A. Funt has ownership interest in Kite Pharma. P.B. Chapman reports receiving commercial research support from Bristol-Myers Squibb, Genentech, GlaxoSmithKline, and Novartis and is a consultant/advisory board member for Bristol-Myers Squibb, Genentech, and GlaxoSmithKline. No other potential conflicts of interest were disclosed.

Conception and design: S.A. Funt, P.B. Chapman

Writing, review, and/or revision of the manuscript: S.A. Funt, P.B. Chapman

P.B. Chapman was supported in part by the John K. Figge Research Fund.