4 March 2014
The emerging view of cancer is an increasingly diverse set of rare diseases with overlapping etiology and molecular drivers (1). The evolving definition of personalized oncology centers on the development and vetting of many specific, targeted, low-toxicity molecular scalpels that can be used individually or in combinations to tailor a patient’s cancer treatment. The bottleneck of drug development is the testing phase, and there are numerous barriers to novel chemotherapeutic introduction to the clinic.
The I-SPY2 trial has received a lot of press lately, and rightfully so: it’s a bayesian study designed to adaptively match patient genotypes of appropriate therapies in a rotating 5-arm clinical trial that can quickly take on a new agent once one of the test arms graduates, ready for Phase III trials that (theoretically) have a might higher chance for success, with fewer patients (2, 3). The trial has been engineered to make cancer drug trials quicker and less expensive, and more information can be found here.
The I-SPY2 trial is co-directed by Laura Esserman, who gave an illuminating talk at the UCSD Moores Cancer Center last week. Most targeted therapies are tested in clinical trials in heavily pre-treated patients in the metastatic setting, and often without molecular biomarkers. Agents tested in the metastatic setting often see a 2-4 year knowledge turn, while those in the adjuvant setting often come with a 6-9 year knowledge turn. This is simply too slow (and uses too many patients and is too expensive) to realize the vision of personalized oncology.
In the neoadjuvant setting (pre-surgery, chemo-naive patients) the turnaround can be much faster. Instead of disease recurrence, the trial readout is volumetric change in tumors. Tumor samples are subject to panomic (genomic, proteomic, methylomic, etc) analysis before and after neoadjuvant therapy, enabling post-therapy clues to efficacy or failure of tested agents. While this sounds straightforward, this is not the norm for current oncology clinical trials, especially ones for targeted therapies. The reasons for this are beyond the scope of this post. There has been an extreme paucity of information for why certain drugs do not work, and very limited information on what –omics background produce responders.
While pre-operative administration could permit greater organ conservation in the patient, I am also interested in this approach for two additional reasons:
1) Chemotherapy-naive patients might be less susceptible to drug resistance out of the gate. Many drug resistance mechanisms are shared, and patients will not have been weakened from systemic chemotherapy in the neoadjuvant setting (4, 5).
2) Surgery is a highly invasive procedure that damages tissue, releasing cytokines and growth factors that can promote inflammation and tumor growth (6).
The neoadjuvant is a better stage to test for agent efficacy, and opens the door to glimpses of tumor biology that might enable more curative approaches. The neoadjuvant drug administration combined with panomics approaches could be a boon for the emerging “Rapid Learning Precision Oncology” paradigm proposed as part of “Personalized Oncology 3.0″ by Shrager and Tenenbaum (7). Eventually, it may be possible to consider each patient encounter as an experiment, with each additional “experiment” better informed than the last.
I-SPY2 has already graduated two agents to Phase III trials: the small molecule dual HER2 and EGFR inhibitor Neratinib, and the PARP inhibitor Veliparib. The speed at which they passed through Phase II trials is encouraging. Because the trial actively adapts with genotype efficacy, it’s anticipated that Phase III trials will have a greater chance of success. I will be watching this with great anticipation, and at this point in time I am skeptically optimistic about I-SPY2. We need more bullets, big or small, and every new target and every new agent adds another small step toward realizing the vision of personalized oncology.
1. Kandoth C, McLellan MD, Vandin F, Ye K, Niu B, Lu C, et al. Mutational landscape and significance across 12 major cancer types. Nature. 2013 Oct 17;502(7471):333-9. PubMed PMID: 24132290. Epub 2013/10/18. eng.
2. Berry DA. Adaptive clinical trials in oncology. Nature reviews Clinical oncology. 2012 Apr;9(4):199-207. PubMed PMID: 22064459. Epub 2011/11/09. eng.
3. DeMichele A, Berry DA, Zujewski J, Hunsberger S, Rubinstein L, Tomaszewski JE, et al. Developing safety criteria for introducing new agents into neoadjuvant trials. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013 Jun 1;19(11):2817-23. PubMed PMID: 23470967. Epub 2013/03/09. eng.
4. Junttila MR, de Sauvage FJ. Influence of tumour micro-environment heterogeneity on therapeutic response. Nature. 2013 09/19/print;501(7467):346-54.
5. Lord CJ, Ashworth A. Mechanisms of resistance to therapies targeting BRCA-mutant cancers. Nature medicine. 2013 Nov;19(11):1381-8. PubMed PMID: 24202391. Epub 2013/11/10. eng.
6. Costanzo ES, Sood AK, Lutgendorf SK. Biobehavioral influences on cancer progression. Immunol Allergy Clin North Am. 2011 Feb;31(1):109-32. PubMed PMID: 21094927. Pubmed Central PMCID: PMC3011980. Epub 2010/11/26. eng.
7. Shrager J, Tenenbaum JM. Rapid learning for precision oncology. Nature reviews Clinical oncology. 2014 Feb;11(2):109-18. PubMed PMID: 24445514. Epub 2014/01/22. eng.