PD-1 axis inhibitors have become a standard treatment modality in the management of advanced lung cancer. Novel Natural Language Processing (NLP) and Artificial Intelligence (AI) technology enables automated extraction of real-world data at greater scale than current manual chart abstraction processes, which can be used to further explore the impact of these agents in the general population irrespective of PDL1 tumour expression.
Patients diagnosed with stage IIIB/IV lung cancer at the Princess Margaret Cancer Centre between 2015 and 2018 were reviewed using the DARWEN™ NLP and AI data abstraction platform developed by Pentavere. Data extracted include patient age, smoking status, ECOG performance status, tumour histology, biomarker status, PDL1 expression, sites of metastases, treatment information and survival.
Of 615 patients with accessible electronic pathology records, 540 (87.8%) had NSCLC and 280 (51.8%) of those received systemic therapy and were included in the analysis. 86 (30.7%) were EGFR sensitizing mutation positive, 18 (6.4%) ALK rearranged, PDL1>50%/1-49/<1/unknown in 21/8/10/61%. Almost one third (31.7%) of those that received treatment received immunotherapy for any line of therapy (12.1% first-line). Chemotherapy was used first-line in 56.1% and targeted therapy in 36.1% of those receiving systemic therapy Patients that were more likely to receive immunotherapy any line were smokers (OR: 2.7, 95% CI: 1.43-5.10, p=0.002) with a higher number of metastatic sites (OR: 1.23, 95% CI: 1.06-1.43, p=0.005). Those with EGFR sensitizing mutation and ALK rearrangement were less likely to be given immunotherapy (OR: 0.07, 95% CI: 0.03-0.19, p<0.001 and OR: 0.11, 95% CI: 0.01-0.84, p=0.03 respectively). There was no difference in the rates of immunotherapy being given in those with PDL1>50%/1-49/<1 (52/52/44%, p=0.8). Using Cox regression analyses after controlling for ALK, EGFR, PD-L1, age, sex, baseline ECOG, smoking status and number of metastatic sites, patients that received immunotherapy at any point had longer survival (HR: 0.28, 95%CI: 0.12-0.67, p=0.004) in a complete case analysis.
Novel NLP and AI technologies like DARWEN™ gives clinicians access to previously unavailable information on real world treatment strategies and outcomes. Increasing uptake of immunotherapy may further improve outcomes for patients with this challenging to treat cancer. This study demonstrates that the benefit of immunotherapy seen in clinical trials can be translated into the general advanced lung cancer population. Larger population studies will be needed to further analyze the impact of new treatments in the real world and will be facilitated by automated data abstraction to rapidly generate large datasets.