Minor delay puts neratinib on collision course with Aphinity

Yesterday’s tweak to the proposed statistical analysis of the Extenet trial of Puma’s neratinib comes as another pain for investors, whose heads must already be spinning after trying to understand the various data iterations from this complicated study.

The good news is that the FDA-mandated change basically does not alter the robustness of Extenet’s final readout. But there is a serious sting in the tail: with a three-month filing delay neratinib now looks like it will be reviewed with the US agency already in possession of crucial competing data from Roche’s Aphinity trial.

In Aphinity Roche’s Perjeta is added on top of Herceptin for one year in the adjuvant breast cancer treatment setting. And if the data favour the Perjeta addition this could wipe out the case for neratinib, for which Puma is seeking an “extended adjuvant” label on the basis of Extenet – an extra year of neratinib after a year’s Herceptin alone.

In January Roche’s pharma chief, Daniel O’Day, said Aphinity was on track to read out towards the end of 2016. With the added statistical tweak to Extenet neratinib will now not be filed with the FDA until mid-2016, rather than in the first quarter, Puma said yesterday.

Patient censoring

There have always been two main doubts about Extenet: a high degree of patient censoring that is spread unevenly between the study’s active and placebo arms, and the fact that up to 40% of neratinib recipients experienced grade 3 diarrhoea.

Patients are censored either owing to loss to follow-up, or because they have not reported for their scheduled doctor visit. As such the 24-month data, which will be used in the filing, differed with longer-term follow-up that included previously censored patients added back if they made a visit after month 24 (Analysts see a new hope in three-year ExteNET data, December 15, 2015).

The FDA has now asked Puma to censor progressions in patients missing two or more visits at the last available assessment time before progression. The upshot: six fewer progressions, the same level of benefit over placebo in invasive disease-free survival (2.3 points), and a still highly significant p value.

Iterations of 2-year Extenet data...*
Presented at Neratinib Placebo Hazard ratio p value Note
ASCO 2015 93.9% 91.6% 0.67 0.005 Original readout
SABCS 2015 94.1% 91.6% 0.68 *** Longer follow-up, reduces censoring
28 Mar 2016 94.2% 91.9% 0.66 0.004 Amended statistical analysis
NB: *1 yr Herceptin, then 1 yr neratinib; **invasive disease-free survival; ***p=0.004, but not relevant owing to multiple analyses.

As for the diarrhoea problem, Puma has long stressed that the original Extenet design did not allow prophylaxis with Imodium; using this drug in a subsequent neratinib study cut grade 3 diarrhoea to 17% – still high, critics contend – and the company said the FDA would consider this also in its deliberations.

There are other problems, however. One is the extent to which the agency will look at neratinib’s extended adjuvant benefit over placebo at years three to five. Already initial data have shown neratinib’s 24-month benefit getting even smaller at three and four years.

…and neratinib's vanishing Extenet benefit
Time point Neratinib Placebo Relative benefit (pct points)
2 years 93.9% 91.6% 2.3
3 years* 92.0% 89.9% 2.1
4 years* 90.5% 88.6% 1.9
5 years Not yet available
NB: *as presented at SABCS 2015.

A separate issue is whether the small – though statistically significant – 24-month benefit is robust when so many patients are censored, and when just a handful of events could swing the result.

Add to that the Aphinity data and the threat to Puma becomes even greater. Shares in the group were down 26% to $26.00 in early trading, putting its market cap below $1bn; at the height of the euphoria following the first Extenet readout its valuation had touched $9bn.

Puma continues to cling to neratinib’s improved efficacy in hormone receptor-positive patients as a potential benefit over Perjeta, and indeed some analysts have removed hormone receptor-negatives from their forecasts.

Still, whether the FDA will pay this subgroup any attention is, like the myriad other Extenet analyses, too risky to bet on right now.

Study Trial ID
Extenet NCT00878709
Aphinity NCT01358877

To contact the writer of this story email Jacob Plieth in London at jacobp@epvantage.com or follow @JacobPlieth on Twitter

Share This Article