ESMO 2016
Atezolizumab v docetaxel for NSCLC: results from the OAK trial
Dr Fabrice Barlesi - Hôpital Nord, Marseille, France
I will present the primary analysis from the OAK study which is a randomised phase III study comparing atezolizumab with docetaxel in advanced non-small cell lung cancer. The phase III OAK study is the first study assessing a PD-L1 antibody in non-small cell lung cancer and this is also the largest phase III trial in the second line looking at the assessment of immune checkpoint inhibitors.
The OAK study recruited locally advanced or metastatic non-small cell lung cancer that had received one or two prior lines of therapy, including one platinum-based line of therapy, whatever their PD-L1 status. They were stratified by PD-L1 expression, histology and number of prior chemotherapy regimens. They were randomised one to one to receive either atezolizumab 1200mg IV every three weeks until disease progression or loss of clinical benefit or they received docetaxel at standard dosage until disease progression.
The OAK study has two co-primary endpoints: the first endpoint was the overall survival in the intention to treat population and the co-primary endpoint was the overall survival in patients with PD-L1 expression, meaning at least 1% of PD-L1 expression in tumour cells, that’s TC, or tumour infiltrating immune cells, referred to as IAC. There was a pre-specified analysis after the first 850 enrolled patients, this analysis was pre-specified and provided enough power to assess the overall survival. The secondary endpoints were overall response rate, progression free survival, duration of response and safety. There was no crossover allowed between the arms.
The OAK study met its primary endpoint. As you can see, the curves separate very early and remain separated along over the time. It translates in a hazard ratio of 0.73 which is highly statistically significant. There was also a clinically meaningful benefit as the median overall survival increased from 9.6 months for the patients randomised in the docetaxel arm to 13.8 months for the patients randomised in the atezolizumab arm, meaning an increase of 4 months for the patients receiving the PD-L1 inhibitor.
The OAK study also met its co-primary endpoint, the assessment of the overall survival in the PD-L1 expression subgroups which represent 50% of our patients. In these subgroups the benefit was similar with a hazard ratio which was 0.74 which was also highly statistically significant. As you can see, there is also a clear separation of the curves and the increase in the median overall survival goes from 10.3 months to 15.7 months in this group, meaning an improvement of 5 months which is obviously clinically significant.
When we look at the efficacy and the overall survival across all PD-L1 subgroups you can see on the right the median overall survival for each group. You can see in the middle the hazard ratio and on the left the PD-L1 subgroups. What you can see is the efficacy is consistent across all the subgroups with a hazard ratio of 0.73 in the intention to treat population, 0.75 in the IC0 TC0 patient population, meaning patients without any PD-L1 expression, and the patients with the highest PD-L1 expression, patients TC3 IC3, had the greatest benefit.
Regarding the safety summary and the immune mediated adverse events, you can see that the atezolizumab was well tolerated, despite a longer median treatment duration. The fact that 20% of the patients in the atezolizumab arm were treated for more than 12 months you can see that there was an advantage for the atezolizumab arm with less treatment related grade 3 or 4 events, 15% compared with 43%. There were no treatment related deaths for the atezolizumab arm and there were less adverse events leading to treatment withdrawal, dose modification, delay or interruption of the treatment. Looking at selected immune mediated adverse events, you can see that the number of these events were very low with for the grade 3 and 4 events less than 1% of pneumonitis, hepatitis and colitis.
In conclusion, the OAK data represents the first phase III results for a PD-L1 directed antibody. The atezolizumab improved the overall survival in all patients with in the intention to treat population a median overall survival going from 9.6 for the docetaxel arm to 13.8 months for the atezolizumab arm and a hazard ratio of 0.73, meaning a decrease of 27% in the risk of death. This benefit was seen regardless of the PD-L1 expression with a hazard ratio of 0.75 in the patient population with no or minor PD-L1 expression, less than 1%, to 0.41 for the patients with the highest PD-L1 expression. The overall survival benefit was consistent across all subgroups including squamous or non-squamous histology and never smokers. Atezolizumab was well tolerated with a favourable safety profile compared to docetaxel. No new safety signals were identified and the rate of immune mediated adverse events was very low, below 1%. Thank you.
