Last Friday, Peregrine announced their initial phase IIb results with bavituximab in second line non squamous non-small cell lung cancer (NSCLC). The topline results were most interesting to say the least, since the company suggested that adding bavituximab to docetaxel in second line doubled the rate of survival in these patients. The data was presented by Dr David Gerber (University of Texas, Southwestern in Dallas) at the Chicago meeting of the Multidisciplinary Symposium in Thoracic Oncology in the late breaking session.

What is bavituximab?

Bavituximab is a monoclonal antibody that targets phospholipid phosphatidylsterine (PS), which exists in the membrane of vascular endothelial cells.

In normal conditions, PS is hidden in the inner membrane, but in cancer conditions, the tumour microenvironment can cause the PS cells to also be expressed on the outer membrane. It is thought that chemotherapy can also heighten the PS exposure in the tumour microenvironment, leading to reduced anti-tumour responses.

Bavituximab’s role is to bind to the activated PS on the outer membrane, essentially blocking PD mediated tumour suppression. This is an immune effect, similar to PD–1, whereby it allows the body to recognise the presence of cancer and stimulates it to fight the tumour.

What was the trial design?

The phase II trial randomised patients with non-squamous NSCLC (n=121) to receive one of the following:

1) Docetaxel 75mg/m2 3qw + placebo for up to 6 cycles (n=40)
2) Docetaxel 75mg/m2 3qw + bavituximab 1mg/Kg qw for up to 6 cycles (n=40)
3) Docetaxel 75mg/m2 3qw + bavituximab 3mg/Kg qw for up to 6 cycles (n=41)

Each of the groups then received maintenance therapy ie placebo or bavituximab (1 or 3mg/Kg) until progression.

The primary endpoint of the trial was response rate and the secondary endpoints included PFS, OS, safety and duration of response.

Where were the centres?

Aside from University of Texas Southwestern in Dallas, other US sites included Hershey, several regional cancer centres and a number of community oncology sites, presumably US Oncology. Internationally, the centers included sites in Russia, India and Ukraine.

The trial details can be found in the clinical trials database.

What did the results show?

Based on the company’s press release, the interim data analysis appeared to suggest that adding bavituximab to docetaxel doubled the overall survival from 5.6 months with docetaxel/placebo to 12.1 months in the pooled bavituximab arm.  Each active arm was 11.1 and 13.1 months for 1 and 3 mg/Kg respectively.  I’m not sure that the individual arms were significantly different from docetaxel alone or just the pooled data, which naturally doubles the N number and hence power in determining the difference.  That’s quite a different proposition to start with.

While these results are not unheard of, they require caution for a number of reasons.  We need to remember that they are reminsicent of other, similar small company phase II trials in lung cancer such as Antisoma’s ASA–404, where an increase of 8 months or so in the phase II study was not repeated in a large scale randomised phase III trial. I do think there is enough here from the current bavituximab study to warrant testing the immunotherapy in phase III RCT, but I would urge caution in extrapolating the results at this time.

What are the potential issues with this study?

a) Small phase 2 trials in oncology are fraught with bias, which often disappears when tested on a larger scale in reputable US and EU centers.

b) Not all patients in Eastern Europe are aggressively treated to progression upfront in the same way they are in the US, in other words they may be relatively undertreated, which can influence the subsequent outcome.

c) We don’t have a breakdown by country of the results across groups – did the Eastern EU and Indian patients do better than the US ones? How many patients came from each country that participated and in which arm?

d) It is always tricky to extrapolate from small N numbers of this size, especially when no data for how well the groups were matched in terms of prognostic factors is provided beyond age, performance status etc. There was, howver, a noticeable lower number of caucasians and a higher number of asians in the 3mg bavituximab arm, for example. Would this affect the results? Who knows in a small sample!

e) No mutational analysis was made available – this means more patients with a poorer prognosis could confound the data in the 3 groups and we have no idea why some patients did better than others.

f) The amount of prior Avastin varied across the groups – from 20% (placebo) to 18% and 15% in the bavituximab arms and although not significant in a small sample size, the groups should have been better balanced for this. I would be very wary if these numbers were repeated in a much larger scale trial, especially when considering the bleeding events were higher in the bavituximab 1mg group, which also received more prior Avastin than the 3mg group but a lower number overall any adverse events or grade 3+ events.

g) It’s also difficult to interpret these data in a small sample size, especially when you consider that the toxicities in the docetaxel plus bavituximab 1mg group appear to be better than docetaxel alone. This make little sense to me.

When considering the efficacy data, there are also some other noticeable anomalies:

1) The differences in PFS are small (3.0, 4.2 and 4.5 months) between the three groups but the OS curves show a more dramatic difference. This is normally an obvious red flag – you wouldn’t predict this to happen from the marginal PFS data based on central review.

You can see the OS data as shown on Yahoo Finance:

bavituximab
bavituximab

2) The other odd thing about the PFS data is that at around month 7, the curve for the 3mg bavituximab arm suddenly drops below the 1mg arm and parallels the placebo arm, yet the OS is still better in the bavituximab arm compared with docetaxel alone? Crossover does occur in trials, but it does seem an odd drop and leads me to wonder why and what happened at that point? In addition, with immunotherapy there is usually a delayed effect rather than an upfront effect so one would expect the results in the bavituximab arm to prolong over time if the agent is working effectively.

3) While PFS of 3–4 months for a docetaxel combination is normally acceptable in second line, I would have expected the OS to be longer than 5–6 months for docetaxel alone. It is possible, however, that MOS has not yet reached and these data will improve over time since 10–11 months is a realistic target. If median OS has not yet been reached since this is interim data, then we need to wait and see what the final data will be – with and without bavituximab – since any initial benefit seen before MOS can potentially disappear over time.

Overall:

These data are very early and offer both hints of promise and portends to future doom (ie a negative phase III trial).  Three things I would like to see:

1) Full data analysis with median rather than interim OS for the phase II trial
2) Subset analysis by country and prognostic factors such as mutational analysis
3) Phase III results from a randomised controlled trial involving US and EU academic centres that eliminates inevitable phase II bias

I would also caution readers to avoid extrapolating too far from interim data from small phase II trials based on the Antisoma experience in NSCLC!