In 2007 we started the Initiative for Molecular Profiling in Advanced Cancer Therapy. The hypothesis of the study was that human molecular profiling will help us optimise treatment selection based on patients’ tumour molecular abnormalities and therefore we would have superior outcomes compared to the standard approach.
And when it comes to the presentation from today, how much is that borne out to be accurate and true?
The hypothesis was proven to be accurate since from 2007 until 2013 we ordered tumour molecular profiling in 3,743 patients. Of those patients 1,307 patients had at least one molecular alteration, or 35%, and of these 1,307 patients 711 were treated with matched targeted therapy and 695 patients were treated with non-matched therapy. We treated these patients on clinical trials with investigational agents, first in human or with FDA approved drugs in new combinations. Patients were heavily pre-treated, they had a median of four prior therapies, ranging from zero to sixteen. 2.8% of patients were not previously treated, they had rare, incurable tumours.
Patients had various tumour types including gastrointestinal tumours, including colorectal, gynaecologic tumours, melanoma, breast and lung cancer. The overall response rate was 19% in patients treated with matched targeted therapy compared to 5% in those treated with non-matched therapy. When we took into consideration disease stabilisation for at least 6 months the overall tumour response control rate was 35% in the matched therapy group compared to 20% in the non-matched therapy group and this was statistically significant.
This improved disease control was also associated with longer progression free survival in the matched therapy group. The median progression free survival was 4 months in the matched therapy group compared to 2.8 months in the non-matched therapy group and the hazard ratio was 0.67. The median overall survival was 9.3 months in the matched therapy group compared to 7.3 months in the non-matched therapy group and the hazard ratio was 0.72.
Then we performed univariate and multivariate analysis, trying to determine the association between patients’ baseline characteristics and survival. We found that patients who had PI3 kinase alterations had inferior outcomes. In multivariate analysis factors predicting shorter overall survival were PI3 kinase, AKT mTOR pathway abnormalities and other characteristics indicating advanced metastatic cancer such as performance status greater than 1, increased liver LDH levels, liver metastases, abnormally low albumin levels, higher platelet counts as well as age 60 years or older. Then we used each of these parameters, pre-treatment characteristics, to develop a score. We gave one point to each of these parameters and we developed a score according to which we can predict patient survival taking into consideration their baseline characteristics. This survival ranges from if patients had zero points about 15 months to 2 months, or less than 2 months, if they have all 6-7 points.
Then in the multivariate analysis we added the type of therapy and we found that non-matched therapy was associated with shorter survival together with all the other pre-treatment characteristics we just described. Therefore matched targeted therapy was an independent factor associated with longer survival.
Therefore, in conclusion, we found that matched targeted therapy is associated with longer survival compared to non-matched therapy. It is an independent factor predicting survival. Overall the three year survival rate was 15% in patients treated with matched therapy compared to 7% in those patients treated with non-matched therapy. At ten years the ten year overall survival rate was 6% in the matched targeted therapy group compared to 1% in the non-matched therapy group.
Our results demonstrate the value of using personalised medicine in treating patients with advanced cancer. We would like to implement precision medicine which holds the promise of improving the clinical outcomes of patients with cancer.
