AACR 2014

I-SPY 2 trial shows promising development for some breast cancer patients using neratinib

Dr John W. Park - University of California, San Francisco Hellen Diller Family Comprehensive Cancer Center, San Francisco, USA

It is my pleasure on behalf of the I-SPY 2 investigator team to present neratinib plus standard neoadjuvant therapy for high risk breast cancer efficacy results from the I-SPY 2 trial. I’m going to advance to the adaptive trial design that was used in I-SPY 2 which is highly innovative. The trial begins with equal randomisation probabilities to different breast cancer subtypes, the primary endpoint is pathCR rate which is estimated for each subtype and for each associated molecular signature. Based on the success probability in a Bayesian analysis an agent will be found to either graduate, which is meet a predefined threshold, for successful completion of the endpoint or futility, meaning the endpoint was not reached. If at any point the trial this endpoint has not been met, the trial continues to learn from the responses obtained to date and the randomisation is revised to reflect the results that have been obtained thus far. The trial continues to accrue patients but in this weighted randomisation fashion and so on iteratively until again graduation or futility is the result. An experimental agent is grafted onto a standard chemotherapy backbone and a series of investigational agents in fact are included and are simultaneously evaluated. Once an agent is found to have graduated or to have completed the cohort without graduating it then stops further accrual and that agent’s evaluation is done.

I-SPY 2 was open to patients with primary breast tumours of at least 2.5cm who had a molecular profile indicating any subtype that was not so-called luminal A or hormone receptor positive HER2 negative MammaPrint low. Otherwise patients had the ability to have core biopsies obtained, serial MRIs and then treatment with the investigational agent with standard of care or standard of care alone in the control group.

Today we’re reporting the experimental arm containing neratinib. Neratinib is an irreversible inhibitor of the EGFR, HER2 and HER4 receptor tyrosine kinases. Patients were adaptively randomised to receive the investigational agent in combination with paclitaxel given weekly followed by AC chemotherapy followed then by surgery versus, in the control group, paclitaxel alone followed by AC for the non-HER2 positive patients or paclitaxel with trastuzumab followed by AC for the HER2 positive patients, representing standard of care.

Therefore, for the HER2 negative patients neratinib was able to be open for enrolment, this represented recognition that neratinib has preclinical activity against both HER2 negative and HER2 positive tumour cells, potentially by virtue of its pan-HER spectrum. Therefore neratinib was evaluable for graduation and was eligible to graduate in all ten of the pre-specified signatures.

Here’s a schema of the neratinib arm. Again, HER2 positive patients received Taxol with trastuzumab followed by AC versus the investigational version, Taxol plus neratinib in place of trastuzumab followed by AC. HER2 negative patients received the same chemotherapy without the trastuzumab.

116 patients were randomised to receive the investigational therapy, one patient withdrew consent therefore 115 patients were evaluable. The primary endpoint, as mentioned, is pathologic complete response rate. 78 patients were concurrently randomised to the control group, the control group was shared among the investigational agents in the trial. As the trial proceeded the adaptive design, as planned, continued enrolment and enriched enrolment for the responding subtypes and decreased and eventually stopped assignment to non-responding subtypes. So, for example, the hormone receptor positive HER2 negative group and the hormone receptor negative HER2 negative group ceased accruing patients late in the trial.

The profile of the patients in the trial is consistent with a predominantly locally advanced breast cancer population as you see here. Once again the adaptive randomisation enriched the neratinib treatment arm for patients who had the appropriate subtype consistent with good response and therefore the HER2 positive subset, as you see, was enriched relative to the control group, once again which was shared among all the investigational agents.

The format for I-SPY 2 results reporting is a Bayesian estimate of the probability of pathCR by signature and there are in fact three analyses that are performed generating an estimate for pCR rate by signature, a probability that the drug is better than the control for a given signature, and the predicted probability of success in a subsequent 300 patient phase III trial based on pathCR as an endpoint.

So the results that I-SPY 2 obtained are shown here. Neratinib graduated, that is to say met the predefined threshold for positive results, specifically in the HER2 positive hormone receptor negative subset. The estimated pCR rate for neratinib was 56% for the neratinib containing treatment versus 33% for the concurrent control treatment, again trastuzumab containing. The second analysis compared the probability of neratinib to obtaining pathCR versus the trastuzumab therapy in this signature and there was a 95% probability that neratinib was superior to standard of care. The third analysis indicated a predicted probability of success for a similarly designed phase III trial of 300 patients in the population, a probability of success of 79%. The remaining nine signatures are shown below and they contain also some evidence of benefit but not reaching the threshold mark for graduation.

This is actually the data as it occurs from the Bayesian analysis so, for example, signature one which represents all patients, the distribution, the Bayesian distribution of pathCR results, is shown in red, the predicted probability of superiority is 48% over control. The median point of this distribution shows the point estimate for pathCR for neratinib, that is 33%, for the control treated patients it is 23%. For all HER2 patients indicating signature 4 the pathCR estimate is 39% for neratinib, it is 23% for the control. For the HER2 negative group there was minimal evidence of any benefit, pathCR was 28% for neratinib estimated versus 24% for control.

Turning now specifically to the HER2 positive signature and its additional sub-signatures or more specified signatures, the HER2 positive hormone receptor negative signature, which is the graduating signature, again showed a difference in estimated pCR of 56% versus 33% respectively for neratinib over control. Again, consistent with the predicted probability of phase III success of 79%. The HER2 positive hormone receptor positive complement to that signature, which is itself signature 10, showed predicted probability of success in phase III of 65% with 30% estimated pCR rate for neratinib as compared to 17% for control. The remaining signatures include the HER2 negative signatures, all HER2 negative patients at the top with minimal evidence of different pathCR rate. The triple negative signature with a difference between 38% for neratinib, 31% for control, and the HER2 negative hormone receptor positive signature which showed minimal difference.

Finally, the MammaPrint 2, this represents the median cut-point or above for MammaPrint score. The MammaPrint high 2 or ultra-high showed a numerical advantage for neratinib with an estimated pathCR rate of 47% for neratinib as compared to 29% for control.

We have some preliminary and selective safety information at this time which I will share. Neratinib was largely well tolerated however there were safety signals, particularly involving diarrhoea, as you see here, including grade 3 or above as well as frequent grade 1 or 2 events that were greater than the control. There was not a particular cardiac safety event that was frequent in either arm and there was, in fact, no congestive heart failure events noted to date. Focussing further on the diarrhoea there was an initial diarrhoea supportive care guideline, it was then modified to include vigorous patient monitoring and early institution of supportive care which appeared to make a difference in terms of the frequency of diarrhoea and amelioration of that side effect. Further to that there was also the institution of a prophylactic loperamide regimen which was also instituted later in the trial.

In terms of early discontinuation and delay of surgery, there was no apparent difference in time to surgery from initiation of treatment in the control versus the neratinib treated groups. Early discontinuation was somewhat more frequently observed with neratinib primarily due to toxicity whereas early discontinuation due to progression was observed more frequently in the control group.

So, in conclusion, this I-SPY 2 adaptive trial has identified a biomarker signature for neratinib. It is the HER2 positive hormone receptor negative signature. It has graduated in this signature with a 79% predicted probability of success in a subsequent phase III trial with similar design, such as neratinib paclitaxel versus trastuzumab paclitaxel both followed by AC. In the HER2 positive hormone receptor negative signature in which it graduated neratinib was calculated to be superior to trastuzumab with 95% probability. The toxicity was most notable for diarrhoea, however, this was considered manageable with a number of measures instituted for supportive care and we in fact recommend aggressive supportive care based on this experience as well as other studies, including studies now using higher doses of supportive care prophylaxis.

I-SPY 2 is a biomarker rich trial, additional response predictors and biomarker development are under investigation and we anticipate will be reported in a subsequent forum. Based on these results neratinib is under consideration for phase III testing in the neoadjuvant population, for example in I-SPY 3 in the HER2 positive hormone receptor negative signature or in the all HER2 positive group.

We would like to thank all the investigators participating in I-SPY 2 and most especially the patients. Thank you very much.