Please Note: Blog posts are not selected, edited or screened by Seeking Alpha editors.

What Aveo Can Do To Save Company's Reputation – Prove The Efficacy Of Tivozanib

|Includes: AVEO Pharmaceuticals, Inc. (AVEO)

My first post about tivozanib was about why and how Aveo's tivozanib got rejected as treatment for kidney. The second post demonstrated the effect crossover had on the survival estimates. While doing research for those articles I learned about Memorial Sloan Kettering Cancer Center prognostic risk groups and the role they have in prediction of survival potential of kidney cancer patients. In this final post about tivozanib I will use MSKCC group approach to demonstrate how Aveo still can settle some troubling questions or maybe even prove the efficacy of tivozanib - without resorting to mathematics of my second post about tivozanib. No tricks - only generally accepted prognostics groups and a bit un-orthodox (but legitimate) use of them.

Breaking Overall Survival Curve to Survival Curves for MSKCC Risk Groups

All clinical trials I reviewed for my previous post reported the MSKCC risk group distribution in both trial arms. The information to do the breakdown is obviously available for all trials but very few have done it. Some trials reported median survival times for MSKCC risk group population. Only two trials published Kaplan-Meier survival curves for MSKCC risk groups - Axis trial [1] (axitinib vs. sorafenib) and everolimus trial [2]. I will use modified KM curves from the Axis trial to demonstrate how, under certain conditions, one can estimate the effect due to crossover.

Figure 1

Kaplan-Meier survival estimates from AXIS 1 for sorafenib are in Figure 1. Three curves for MSKCC risk groups are weighted according to the MSKCC risk group percentages and add up to overall survival estimate. If there is no censoring the result of addition is exact. Censoring makes Kaplan-Meier curves inaccurate on the probability axis - a death produces larger drop toward the end (right side of the curve) than it does at the beginning. In this case the addition result will be approximate.

Figure 2

Figure 2 has the same curves, except now the risk groups curves are not scaled by weighting factors - all curves have the same starting point - 1. Also, notice how overall survival settles somewhere in the middle of risk group curves, exactly where depends on weighting ratios.

Estimating Crossover Effect

Figure 3 shows a theoretical comparison of KM curves for two Intermediate risk groups - comparison is between clinical trial arms. I have assumed that 50 % of patients in the control arm (NYSE:A) crossed over to the study arm (NYSE:B) i.e. after cancer progressed with their initial drug they were given the drug under study. This was the situation, for instance, in Tivo - 1 trial (crossover about 60 %).

Figure 3

Usually Figure 3 is the end of the road - there are too many factors to explain (prior treatments, subsequent treatments). However, if there are no other significant prior and subsequent treatments but crossover, then one can estimate the crossover effect with good reliability. Figure 4 demonstrates the approach that can be used in this situation.

Figure 4

The new KM curve in Figure 4 is for a new sub-group of the Intermediate control arm group - only the patients in the Intermediate group who crossed over are included in this group - A followed By B 100 %. The new curve will outperform the 50 % group if the study drug has any positive effect when administered after the control drug. In other words, Figure 4 shows the difference in performance due to crossover being 100% instead of only 50 %. Adding a curve for those patients in the Intermediate control group not crossing over, 0 % group, will complete the estimate for the effect of crossover from 0 % to 100 %.

Aveo, thanks to 'bad study design' can utilize the above approach. In Tivo -1 trial only prior treatments having some significance to outcome were surgery and treatment with cytokines. All trial patients had surgery; hence it will not be a differentiating factor. About 20 % of trial population was treated with cytokines; too small percentage to make cytokines a differentiating factor. Plus, according to Axis-1 trial sorafenib is the likely beneficiary of cytokine treatment.

I am done with writing about tivozanib (for now, at least). It seems to me that tivozanib is in the class of sunitinib or better due lesser toxicity (oh, irony). What happens next is up to Aveo and people having interest in what kind of treatments are available for kidney cancer.

Sources:

  1. 2012 European Oncology Congress in Vienna, Clinical Spotlight in Renal Cell Carcinoma, Abstract LBA22, Abstract 793PD
  2. Phase 3 trial of everolimus for metastatic renal cell carcinoma

Disclosure: I am long AVEO.

Additional disclosure: No Change - Not Selling!