The Magnitude Matters: GSK Approvals Show How To Use PFS After Other Drugs Show Survival

Submitting an NDA based on progression-free survival when other drugs for the indication have been approved on the basis of overall survival benefits is a risky proposition, but GlaxoSmithKline was confident the magnitude of Mekinist and Tafinlar’s PFS effect in metastatic melanoma would make the findings clinically relevant.

Historically, FDA has declined to approve cancer drugs based on progression-free survival when earlier agents for the same disease had already been approved on the gold-standard endpoint of overall survival – but recent targeted therapy approvals suggest the agency is becoming more open to doing just that.

In those situations – as GlaxoSmithKline PLC’s Tafinlar (dabrafenib) and Mekinist (trametinib) and Boehringer Ingelheim Pharmaceuticals Inc.’sGilotrif (afatinib) showed – the magnitude of the PFS effect was the deciding factor.

The Gilotrif, Tafinlar and Mekinist reviews show a different paradigm emerging in terms of accepting PFS when competitors have shown OS.

FDA discouraged the use of PFS as a stand-alone primary endpoint in pivotal trials of Tafinlar, Mekinist and Gilotrif because competing drugs were already on the market with overall survival advantages, but the agency left the door open to approval based on what reviews and memos frequently refer to as “the magnitude of the difference between the arms and the clinical relevance of this difference” in PFS.

Review documents do not, however, show that FDA is overtly changing its position on the unsuitability of PFS-based NDAs after approvals based on OS. As Office of Hematology and Oncology Products Director Richard Pazdur stated in his dabrafenib approval memo, “an improvement in PFS of sufficient magnitude may be an appropriate endpoint provided that an improvement in OS is not demonstrated in a prior approval of another drug” in the same population.

Adapting To The New Era Of Targeted Therapy

The advent of targeted therapies in oncology, sometimes bringing much higher PFS and response rates than earlier drugs for the same indication, shows FDA’s flexibility in the service of bringing new options to patients with serious illnesses.

“Unmet medical need” is a crucial concept in earning regulatory flexibility from FDA. The Tafinlar and Mekinist approvals show how FDA is looking broadly at the concept, considering the GSK drugs to meet an unmet need in metastatic melanoma on the basis of differing toxicity profiles from the recently approved melanoma indications for Bristol-Myers Squibb Co.’s Yervoy (ipilimumab) and Roche’s Zelboraf (vemurafenib), both of which demonstrated a survival advantage (see related story, (Also see "FDA Considers Toxicity Profiles As Meeting Unmet Need In Melanoma" - Pink Sheet, 19 Nov, 2013.)). The approvals blocked GSK’s early plans to seek accelerated approval for the first-in-class MEK inhibitor Mekinist on the basis of response rate data.

The armamentarium for first-line metastatic non-small cell lung cancer is considerably more extensive than for melanoma. Gilotrif is the second approval for EGFR mutation-positive NSCLC patients, following the clearance of Roche’s Tarceva (erlotinib) on May 14, 2013. The Tarceva indication was also cleared on the basis of a large PFS benefit (a doubling of median PFS compared with chemotherapy), but Division of Oncology Products 2 Director Patricia Keegan’s summary review of Gilotrif also considers as available therapy numerous agents approved for first-line metastatic NSCLC without consideration of EGFR mutation status, including drugs like Pfizer Inc.’s Xalkori (crizotinib), Eli Lilly & Co.’s Alimta (pemetrexed), Roche’s Avastin (bevacizumab), and vinorelbine that showed an overall survival benefit.

Recent comments by Pazdur show that FDA is actively looking at how review standards should evolve as improvements in targeting molecular pathways have raised the bar for efficacy. Noting the “unprecedented” activity and response rates being posted by some targeted drugs, Pazdur said that traditional randomized trials might not be necessary. “We said, in the past, response rates do not correlate with overall survival, for example, in melanoma.” Now, he said, “one has a much different appreciation of a drug and its ability to relate to overall survival or progression-free survival if that response rate is” extremely high (Also see "The Next Phase In Oncology: FDA’s Pazdur Has New Vision For Drug Development" - Pink Sheet, 11 Nov, 2013.).

Consideration, Not Endorsement

The agency declined to endorse the PFS-based pivotal trial designs with special protocol assessments that would signify agreement on clinical endpoints. FDA rejected SPA requests for Tafinlar and Gilotrif and headed off an SPA for Mekinist, advising GSK that it was “unlikely that an SPA agreement can be reached, unless the primary endpoint is OS.”

Starting with end-of-Phase I meetings, FDA explicitly advised GSK and BI to use OS as the primary pivotal trial endpoint. “The agency continues to recommend that OS be the sole primary endpoint,” according to minutes of a July 30, 2010 trametinib meeting. However, “it is GSK’s decision whether to use PFS as a primary endpoint and to cross patients over at progression. … The agency will be willing to discuss the results.” Dabrafenib EOP1 minutes express an almost identical stance.

At an EOP1 meeting for afatinib on July 31, 2007, FDA “recommended overall survival as the primary endpoint” instead of BI’s proposed PFS measure. At an Oct. 16, 2008 meeting, FDA indicated growing acceptance of the PFS primary endpoint, telling the sponsor that “a substantial, robust improvement in PFS that was clinically meaningful and statistically persuasive, and has an acceptable risk/benefit profile, may be considered for regulatory decision making.”

However, as Division Director Keegan reported in her July 11, 2013 summary Gilotrif review, FDA continued to advise that “since all previous approvals in the first line setting of NSCLC were based on overall survival, an application based on PFS would likely be referred to the Oncologic Drugs Advisory Committee.” In the end, Gilotrif’s Phase III data won over the agency; an advisory committee was not necessary.

But Gilotrif underscores that having a dramatic benefit on PFS is crucial because the endpoint is more subjective than OS. Keegan pointed to the “approximate doubling of the median progression-free survival from 6.9 months in the chemotherapy arm to 11.1 months in the afatinib arm.” She noted that “because documentation of PFS assessments is often based on both subjective and objective criteria and these assessments depend on the frequency, accuracy, reproducibility, and completeness of tumor assessments, it is important that the magnitude of effect is robust.”

“Treatment effects of this magnitude are also considered to be evidence of clinical benefit provided that the risks are acceptable,” Keegan said in her Gilotrif review.

Improved PFS As Clinical Benefit, Not A Predictor Of Benefit

Historically, PFS has been considered an endpoint that signals the possibility of clinical benefit – an improvement in survival – less than a clinical benefit in and of itself. Thus, FDA has traditionally held that once a survival benefit has been demonstrated, all future approvals in that space should also show a survival benefit. The Gilotrif, Tafinlar and Mekinist reviews show a different paradigm emerging.

Only two of seven novel oncologics approved this year have had OS data.

In melanoma, FDA questioned the association of PFS and OS at all. During the July 2010 trametinib EOP1/Pre-Phase III meeting, FDA explained that “the relationship between progression-free survival and clinical benefit has not been established in melanoma.” PFS, a qualitative endpoint, also suffers from greater variability in interpretation.

“Whether the prolongation in PFS will also result in improvement in OS, similar to the treatment effects observed with vemurafenib, is uncertain,” Mekinist and Tafinlar clinical reviewer Marc Theoret stated.

Nonetheless, he described dabrafenib’s effect as “a prolongation of PFS of sufficient magnitude, when compared to treatment effects of available therapy, to be considered a clinical benefit.” The dabrafenib trial compared the drug to dacarbazine, “a clinically relevant comparator at the time of initiation of the BRF113683 trial.”

Tafinlar posted a 5.4 month median PFS, compared with 2.7 months for the DTIC arm. Mekinist posted a 4.8 month median PFS, compared with 1.5 months for chemotherapy (DTIC or paclitaxel). Early data suggest that combination use may be even more effective; priority review sNDAs for combination use are pending, with user fee goals falling on Jan. 8, 2014 (Mekinist) and Jan. 9, 2014 (Tafinlar).

The idea of PFS as a stand-alone indicator of clinical benefit gained prominence during the debate over Avastin’s accelerated approval for first-line treatment of metastatic breast cancer (Also see "Avastin Debate Could Provide Clarity On Use Of Progression-Free Survival Endpoints" - Pink Sheet, 11 Jul, 2011.). CDER called Avastin the first non-hormonal agent approved in MBC “in which evidence of a treatment effect on PFS alone was viewed not as a surrogate endpoint, but rather as a clinical benefit because of the magnitude of improvement in PFS” (Also see "Avastin Breast Cancer Dispute Centers On “Magnitude Of Benefit” Questions" - Pink Sheet, 30 May, 2011.).

When confirmatory trials failed to confirm the 5.5 month median PFS benefit seen in the study supporting accelerated approval, FDA moved to withdraw the Avastin indication. Genentech was unsuccessful in opposing the action, but the high-profile debate highlighted issues in using PFS as a primary endpoint for approval that may have guided future applicants, like GSK and BI, who provided NDAs with clear benefits on median PFS without resorting to arguments centered on hazard ratios, as Genentech attempted.

And, importantly, FDA Commissioner Margaret Hamburg’s decision memo on the Avastin MBC withdrawal made clear that she agreed with CDER’s position that improvement in PFS may constitute clinical benefit in appropriate circumstances (Also see "Avastin Decision Leaves Lingering Uncertainty As To Threshold Level Of Progression-Free Survival Benefit" - Pink Sheet, 28 Nov, 2011.).

FDA has been working to clarify standards for PFS in light of its growing popularity as a primary endpoint in drug development (Also see "FDA Will Revisit Appropriate Use Of PFS Endpoints At Advisory Committee" - Pink Sheet, 1 Sep, 2009.). A July 2012 advisory committee meeting addressed radiographic review of PFS; the panel supported easing the requirement that 100% of scans in studies with PFS as the primary endpoint be reviewed in a blinded, independent central review (Also see "ODAC Backs Random Sample Audits Of Progression Events" - Pink Sheet, 30 Jul, 2012.). For Mekinist and Tafinlar, assessment of scans was complicated by extensive technical deficiencies in the NDAs, which required the review team to analyze raw data themselves (Also see "Technical Difficulties Prevented Priority Review Of GSK Melanoma Drugs" - Pink Sheet, 15 Oct, 2013.).

FDA’s oncology NME approvals in 2013 to date highlight the widespread use of endpoints other than OS to support approval. The agency has approved seven NMEs for cancer this year. In addition to the three full approvals based on PFS – Gilotrif, Tafinlar and Mekinist – the agency has granted two accelerated approvals, for Celgene Corp.’s Pomalyst (pomalidomide) and Johnson & Johnson/Pharmacyclics Inc.’s Imbruvica (ibrutinib), based on response rate data. Only Genentech Inc.’s Kadcyla (ado-trastuzumab) and Bayer HealthCare LLC/Algeta ASA’s Xofigo (radium 223 dichloride) provided OS data. Kadcyla met both co-primary endpoints of OS and PFS, and Xofigo is approved for prostate cancer, a setting where symptoms patterns complicate assessment of clinical endpoints other than OS (Also see "Why Survival Is The Fittest Endpoint For HRPC: The Problem With Pain" - Pink Sheet, 1 Mar, 2011.).