Celgene: An Upside Surprise for Abraxane and a Clear Path to Continued Growth

On Friday November 9th, Celgene (NASDAQ:CELG) released top-line data from its phase III 1st-line metastatic pancreatic cancer trial and noted that the study met its primary endpoint, where abraxane plus gemcitabine generated a statistically significant improvement in overall survival (OS) as compared to gemcitabine alone. While expectations for abraxane in pancreatic cancer had recently risen, this is still an important opportunity for Celgene as few drugs work in pancreatic cancer (note the recent failure of Clovis’ [NASDAQ:CLVS] compound). Celgene has not specified the exact size of the effect as it will be presenting more details at the American Society of Clinical Oncology’s (ASCO) GI Symposium in San Francisco on January 24-26, 2013. Without knowing the precise size of the effect, the market opportunity is difficult to gauge. That being said even a marginally significant improvement in OS would be a considerable market opportunity.

We can, however, estimate the possible size of the benefit. The phase III trial was 90% powered to show an OS hazard ratio (HR) of 0.769 and if we assume that the gemcitabine arm has a median overall survival of 6 months (consistent with previous studies), then the combination arm with abraxane would at least have a median survival of about 7.3 months (1.3 month benefit). Celgene management, however, noted that the effect was clearly significant and so that 1.3 month is likely conservative and if we assume benefit is similar to phase II data then a benefit close to 2 months is quite possible. The question then becomes what is the commercial opportunity given a 2 month benefit?

For the past decade or so, gemcitabine had been the primary regimen in 1st-line metastatic pancreatic cancer. This started in the late 1990s after the publication of a study that showed gemcitabine significantly improved the median OS as compared to fluorouracil (5.6 vs. 4.4 months, P=0.002). Since then, gemcitabine has been tested in combination with other chemotherapeutic compounds, but these combinations have seen limited success. In the table below, Ying et al (2012) looked at 13 studies that have a gemcitabine combination treatment and a couple of interesting points should be noted. First, none of the combinations significantly improve OS (p-value threshold of 0.05). Second, the effect of gemcitabine is fairly consistent in generating a median OS of around 6 months with a range that has a low of 4.9 months and a high of 8.8 months. So even though the efficacy of gemcitabine is relatively small (6 months survival), it has been the best option.

Source:Ying et al (2012)

A new option in pancreatic cancer appeared in 2011 with the FOLFIRINOX (oxaliplatin, irinotecan, fluorouracil, and leucovorin) regime. In the study, the median OS for FOLFIRINOX treated patients was 11.1 months as compared to 6.8 months in the gemcitabine group (4.2 months advantage and an overall survival hazard ratio of 0.57). While it is clear that FOLFIRINOX was an efficacious combination, it comes at a cost in terms of adverse events. In the same study, they noted that “incidences of grade 3 or 4 neutropenia, febrile neutropenia, thrombocytopenia, diarrhea, and sensory neuropathy were significantly higher in the FOLFIRINOX group.” Ying et al (2012) draws what is a fairly common conclusion about this regime in that “FOLFIRINOX could be a viable option in selected patients (good PS, normal bilirubin, and a good supportive care system).” In other words, FOLFIRINOX can (should) only be used in the healthier subset of patients as those are the ones that can best handle the toxicity. In addition to concerns over toxicity, there is also the difficulty in administering FOLFIRINOX because it requires a central catheter directly into a vein and hospitalization. In general, then, even with the impressive efficacy of FOLFIRINOX, there is more than enough room for a drug with a similar efficacy but a cleaner side effect profile.

How does the expected benefit of abraxane stack up to FOLFIRINOX? This is a difficult question as the base characteristics of the patients in the study can dramatically affect the difference. Note, for instance, that the FOLFIRINOX study showed a median OS for the gemcitabine arm of 6.8 months, which is on the high side. This is likely because the study needed to recruit healthier patients as the sicker ones would not be able to tolerate FOLFIRINOX. In addition, in the phase I/II study of abraxane, it had a median overall survival of 12.2 months, which is a full month higher than FOLFIRINOX but we are unable to determine the net benefit as the trial was single armed. Given what we know of the trial design and indications from the management, it is quite likely that the benefit will be between 1.5 to 2 months with a possible upside surprise. While the likely efficacy benefit of abraxane will be less than FOLFIRINOX, the side effect profile was much cleaner, where in the phase I/II study “most of these treatment-related AEs were grade 1 and 2 … the most common grade3 nab-paclitaxel related non-hematologic AEs were fatigue (21%) and sensory neuropathy (15%) [and] Of the grade 3 treatment-related hematologic AEs, neutropenia (67%), leucopenia (44%), and thrombocytopenia (23%) were the most common” (Von Hoff et al 2011). In general, then we can expect an efficacy slightly less than FOLFIRINOX but a much more tolerable side effect profile.

In 2005 the FDA approved Tarceva (marketed by Roche) for the treatment of advanced pancreatic cancer. The addition of Tarceva to gemcitabine generated an OS HR of 0.82 and a median OS of 6.24 months compared to 5.91 months (0.33 advantage). The company has said that the effect of abraxane is clearly better than Tarceva, which could already be implied by what we know about the powering of the study. Regardless, if the hurdle for approval is at least a 0.33 month advantage, then clearly approval risk is not very high if even existent. The key is commercial potential against gemcitabine and FOLFIRINOX. One would imagine that gemcitabine would not be a competitive threat given that we know abraxane is clearly more efficacious without significantly adding to the side effect profile.

The real question is still how an abraxane regime would stack up to FOLFIRINOX. In a note after the announcement, a Citi analyst modeled a scenario where abraxane is only used in patients who cannot tolerate FOLFIRINOX and this led to an $850 million market in the US alone and one that grows to $1 billion in 2017. A Bernstein analyst had a slightly lower US market estimate of $600 million (and $300 million outside US) but that was driven by a lower estimated price and assuming 4 cycles as opposed to the 7 cycles used by Citi (keep in mind that the average number of cycles in the phase II trial was 8 so both might be low). Cowen comes to a similar conclusion with estimates of $750 million to $1.1 billion in the US and $2 billion worldwide.

Of course, these estimates seem to assume that abraxane will not be able to meaningfully penetrate the FOLFIRINOX market share. Even with that conservative assumption, it looks like the pancreatic cancer indication will add at least $1 billion (between US and rest of world) in incremental revenues. The revenues are incremental as abraxane is already approved for breast cancer and more recently non-small cell lung cancer, which essentially means that Celgene does not need to hire a sales force, and outside of costs of goods, these new revenues will go straight to the bottom line. In addition, there is still upside if the benefit is large enough to steal some of the FOLFIRINOX market share. As such, for those interested in the relative safety of a large capitalization company, the pancreatic cancer data provides additional earnings power and solidifies the growth profile.