Plasma Proteome-Based Test for First-Line Treatment Selection in Metastatic Non-Small Cell Lung Cancer

Abstract

Purpose: Current guidelines for the management of metastatic non-small cell lung cancer (NSCLC) without driver mutations recommend checkpoint immunotherapy with PD-1/PD-L1 inhibitors, either alone or in combination with chemotherapy. This approach fails to account for individual patient variability and host immune factors and often results in less-than-ideal outcomes. To address the limitations of the current guidelines, we developed and subsequently blindly validated a machine learning algorithm using pretreatment plasma proteomic profiles for personalized treatment decisions.

Patients and methods: We conducted a multicenter observational trial (ClinicalTrials.gov identifier: NCT04056247) of patients undergoing PD-1/PD-L1 inhibitor-based therapy (n = 540) and an additional patient cohort receiving chemotherapy (n = 85) who consented to pretreatment plasma and clinical data collection. Plasma proteome profiling was performed using SomaScan Assay v4.1.

Results: Our test demonstrates a strong association between model output and clinical benefit (CB) from PD-1/PD-L1 inhibitor-based treatments, evidenced by high concordance between predicted and observed CB (R² = 0.98, P < .001). The test categorizes patients as either PROphet-positive or PROphet-negative and further stratifies patient outcomes beyond PD-L1 expression levels. The test successfully differentiates between PROphet-negative patients exhibiting high tumor PD-L1 levels (≥50%) who have enhanced overall survival when treated with a combination of immunotherapy and chemotherapy compared with immunotherapy alone (hazard ratio [HR], 0.23 [95% CI, 0.1 to 0.51], P = .0003). By contrast, PROphet-positive patients show comparable outcomes when treated with immunotherapy alone or in combination with chemotherapy (HR, 0.78 [95% CI, 0.42 to 1.44], P = .424).

Conclusion: Plasma proteome-based testing of individual patients, in combination with standard PD-L1 testing, distinguishes patient subsets with distinct differences in outcomes from PD-1/PD-L1 inhibitor-based therapies. These data suggest that this approach can improve the precision of first-line treatment for metastatic NSCLC.

FIG 1.

Study cohort flowchart: the PROPHETIC study assessed patients with NSCLC who were treated with a PD-1/PD-L1 inhibitor (n = 616; n = 540 after exclusions), represented in gray boxes. In addition, a comparative control group of chemotherapy-treated patients was included (n = 143; n = 85 after exclusions), represented in red boxes. PROphet model development used patients treated with PD-1/PD-L1 inhibitors (ICI-treated) and containing CB evaluations. This cohort was divided into a development set (n = 228 patients) and a blinded validation set (n = 272). Within the development set, 68 advanced-line patients were used as well. The second arm of the study aimed at assessing clinical utility, contrasting a chemotherapy cohort with a set of patients treated with PD-1/PD-L1 inhibitors (total 529 patients). This arm included individuals with measurable PD-L1 stain levels (n = 444) and incorporated the chemotherapy cohort (n = 85). CB, clinical benefit; ICI, immune checkpoint inhibitor; NSCLC, non–small cell lung cancer.

FIG 2.

PROphet model performance evaluation for clinical validity. The predictive accuracy of the PROphet model was assessed using an independent validation set of 272 patients. (A) The ROC curve yielded an AUC of 0.68, with a P value <.0001, indicating a statistically significant predictive power. (B) Observed CB rate versus predicted CB probability, a nearest neighborhood smoother analysis. Each data point represents a patient in the validation set. The x-axis shows the predicted CB probability (CB score), and the y-axis displays the observed CB rate. The relationship between these values reached a goodness of fit with an R² value of 0.98 and a slope of 1.06, highlighting the accuracy of the predictions. CB, clinical benefit; ROC, receiver operating characteristic.

FIG 3.

Clinical utility of PROphet in predicting differential OS outcomes within PD-L1–high expression level subgroups. (A and B) Kaplan-Meier plots for PD-L1–high (≥50%) patients: the survival outcomes for patients who are (A) PROphet-positive and (B) PROphet-negative, treated with either an ICI-chemotherapy combination or ICI monotherapy. (C) Forest plot for multivariate analysis focuses on PROphet predictors in PD-L1–high (≥50%) patients. n = 194 all patients who had all available clinical data. (D) Boxplot and swarm plot of PROphet score separated by PD-L1 staining groups (high [≥50%], low [1%-49%], negative [<1%]): each point represents a patient, demonstrating the relationship between the PROphet score and PD-L1 staining independence at the level. CB, clinical benefit; chemo, chemotherapy; ICI, immune checkpoint inhibitor; OS, overall survival.

FIG 4.

PROphet decision tree for 1L metastatic NSCLC without targetable driver mutations. The chart delineates distinct paths for patient management depending on the combination of tumor PD-L1 expression and PROphet score, highlighting the recommended treatment regimens (either PD-1/PD-L1 inhibitor therapy alone or in combination with chemotherapy) and their corresponding expected CBs. Additionally, it presents the anticipated median rwOS outcomes, offering a prognostic overview correlating with each treatment pathway. 1L, first-line; CB, clinical benefit; NR, not reached; NSCLC, non–small cell lung cancer; rwOS, real-world overall survival.