Treatment with pembrolizumab could elicit long-term survival rates of 21% to 25% for previously-treated patients with PD-L1–positive non–small cell lung cancer compared with 3% to 4% for docetaxel.
Matthew D. Hellmann, MD
Treatment with pembrolizumab (Keytruda) could elicit long-term survival (LTS) rates of 21% to 25% for previously-treated patients with PD-L1—positive non–small cell lung cancer (NSCLC) compared with 3% to 4% for docetaxel, according to a statistical analysis of findings from the KEYNOTE-010 and -001 trials presented at the 2017 ASCO-SITC Symposium.1
Findings from the analysis shed light on the number of patients with advanced NSCLC expected to benefit for up to 70 months from pembrolizumab. According to survival statistics from the SEER database for 2006 to 2012, the 5-year survival rate was 4.3% for those with lung or bronchus cancer with distant metastases.
"We estimate that among patients with previously-treated PD-L1—positive NSCLC treated with pembrolizumab the long-term survival is around 25%," said co-lead investigator Matthew D. Hellmann, MD, from the Memorial Sloan Kettering Cancer Center. "These aren't just numbers on a curve or dots on a plot, these are patients that we see who continue to do well years after starting pembrolizumab. The lives and expectations of these folks is remarkably changing."
With the field of immuno-oncology moving at a rapid pace, there is a need to understand the durability of response to PD-1 inhibition. To help accomplish this, Hellmann and colleagues turned to a LTS model, which is also known as a "cure" model. This type of model has been used by statisticians for over 20 years and involves the use of probability and logistic regression.
In most conventional models for predicting LTS in metastatic cancer, there is an assumption that the number of patients alive will eventually reach 0 given enough time, explained Hellmann. However, this immunotherapy model allows for the possibility of a tail on the curve. In this new cure model, the risk of progression is not constant and can stop at a certain point.
"We're caution but these outcomes really do start to approach the concept of cure for these patients with advanced cancer," said Hellmann. "Given that these outcomes are possible with immunotherapy, distinct statistical models are needed to estimate long-term outcomes."
The analysis of LTS was focused on patients with PD-L1—positive NSCLC (tumor proportion score ≥1%) who progressed on frontline therapy. Initial LTS estimates were based on 306 patients from the KEYNOTE-001 study, wherein pembrolizumab was administered at 10 mg/kg every 2 or 3 weeks. The median follow-up for this estimate was 21 months, with a maximum follow-up of 38.7 months. The LTS rate was 25.4% (95% CI, 14.8-33.6).
This initial estimate was followed by validation, using findings from the KEYNOTE-010 study. In this trial, pembrolizumab was given at 2 or 10 mg/kg every 3 weeks to 690 patients. The median follow-up for the validation group was 12 months with a maximum follow-up of 23 months. In this group, the LTS rate was 25.3% (95% CI, 9.0-36.6).
Findings from the 18-month analysis of the KEYNOTE-010 trial were compared with the LTS projection, with rates adjusted accordingly. After this confirmation step, the LTS rate was projected to be 21.5% (95% CI, 9.2-30.5).
A limited number of long-term results currently exist for PD-1 inhibitors. Long-term survival findings from patients with metastatic melanoma treated with the PD-1 inhibitor nivolumab (Opdivo) were presented at the 2016 AACR Annual Meeting.2 In these findings, which were from a phase I study, the 5-year overall survival rate was 34% for 107 heavily pretreated patients with metastatic melanoma who had not received prior ipilimumab (Yervoy). This rate was nearly double the SEER survival figure for melanoma of 17.9%.
"Melanoma, a disease that was previously characterized by progression, has now been profoundly changed by immunotherapies," noted Hellmann. "For ipilimumab, there's a plateau in the survival curve at about 20% that extends out more than 10 years."
Future studies will continue to assess characteristics that could help predict long-term outcomes, Hellmann noted. Additionally, biomarker analyses are also needed to identify the 25% of individuals with NSCLC who are most likely to experience a long-term response.
The same validated LTS model was used on the docetaxel arm of the KEYNOTE-010 study (n = 343) for comparison, Hellmann said. At the 18-month median follow-up, the LTS rate for docetaxel was 4.3% (95% CI, 0.0-14.4), which aligned with SEER survival expectations. "The long-term survival rates for docetaxel mirror the historical 5-year survival rates we would expect to see in lung cancer before the breakthrough of immunotherapy," he noted.