Virtual Library

Start Your Search

Andrea Fülöp



Author of

  • +

    OA14 - Update of Phase 3 Trials and the Role of HPD (ID 148)

    • Event: WCLC 2019
    • Type: Oral Session
    • Track: Immuno-oncology
    • Presentations: 1
    • +

      OA14.01 - KEYNOTE-024 3-Year Survival Update: Pembrolizumab vs Platinum-Based Chemotherapy for Advanced Non–Small-Cell Lung Cancer (ID 1465)

      11:30 - 13:00  |  Author(s): Andrea Fülöp

      • Abstract
      • Slides

      Background

      In the phase 3 KEYNOTE-024 trial (NCT02142738), first-line pembrolizumab significantly improved PFS (hazard ratio [HR] 0.50, P<0.001) and OS (HR 0.60, P=0.005) vs platinum-based chemotherapy in patients with advanced NSCLC, PD-L1 tumor proportion score (TPS) ≥50%, and no targetable EGFR/ALK alterations (median follow-up, 11.2 months). We present data with 3-years minimum follow-up.

      Method

      Patients were randomized to pembrolizumab 200 mg Q3W for 2 years or platinum doublet (investigator’s choice) for 4‒6 cycles plus optional maintenance (nonsquamous), with stratification by ECOG PS (0/1), tumor histology (squamous/nonsquamous), and region (East Asia/non‒East Asia). Patients in the chemotherapy arm could cross over to pembrolizumab upon disease progression if they met eligibility criteria. The primary endpoint was PFS; OS was a key secondary endpoint. Response per investigator by RECIST version 1.1 is reported.

      Result

      305 patients were randomized (pembrolizumab, n=154; chemotherapy, n=151). At data cutoff (February 15, 2019), median (range) follow-up was 44.4 (39.6‒52.9) months. 210 patients had died (pembrolizumab, n=97; chemotherapy, n=113). 98 (64.9%) patients crossed over from chemotherapy to anti‒PD-(L)1 therapy during/outside of the study. Median (95% CI) OS in the pembrolizumab arm was 26.3 (18.3‒40.4) months vs 14.2 (9.8‒18.3) months in the chemotherapy arm (HR, 0.65; 95% CI, 0.50‒0.86). 36-month OS rate was 43.7% in the pembrolizumab arm vs 24.9% in the chemotherapy arm. Despite longer mean treatment duration in the pembrolizumab arm (11.1 vs 4.4 months), grade 3‒5 treatment-related adverse events (AEs) were less frequent with pembrolizumab vs chemotherapy: 31.2% vs 53.3%. 38 patients in the pembrolizumab arm completed 2 years (35 cycles) of therapy. Among these, 34 were alive, 31 (81.6%) had an objective response (including 3 with complete response), and median duration of response was not reached (range, 4.2‒46.7+ months). OS rate 12 months after completing pembrolizumab treatment (ie, ~36 months after initiating treatment) was 97.4% (95% CI, 82.8‒99.6). Among the 38 patients who completed 2 years, 5 (13.2%) had treatment-related grade 3-4 AEs; no fatal treatment-related AEs occurred. 10 patients who completed 2 years (1 completed 34 cycles) and subsequently progressed received second-course pembrolizumab; 7 had an objective response, 8 remain alive.

      Conclusion

      With >3 years’ follow-up, first-line pembrolizumab monotherapy continued to provide durable long-term OS benefit vs chemotherapy despite a majority of patients assigned to chemotherapy crossing over to pembrolizumab. Pembrolizumab was associated with less toxicity than chemotherapy. Patients who completed 35 cycles of pembrolizumab had durable clinical benefit and most were alive at data cutoff.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.

  • +

    P1.04 - Immuno-oncology (ID 164)

    • Event: WCLC 2019
    • Type: Poster Viewing in the Exhibit Hall
    • Track: Immuno-oncology
    • Presentations: 1
    • Moderators:
    • Coordinates: 9/08/2019, 09:45 - 18:00, Exhibit Hall
    • +

      P1.04-49 - Quantitative Computed Tomography (CT) Based Texture Analysis (QTA) Might Identify Responders to Immunotherapy in Non-Small Cell Lung Cancer (ID 2762)

      09:45 - 18:00  |  Author(s): Andrea Fülöp

      • Abstract
      • Slides

      Background

      Quantitative computed tomography (CT) based texture analysis (QTA) can characterize tumor heterogeneity, that might be associated with tumor infiltrating immune cells, including T-cells, a hallmark of ongoing immune surveillance with potential therapeutic importance. Therefore, QTA might represent prognostic and predictive biomarker in terms of immunotherapy administration. In this study, we investigate the potential of tumoral QTA and a novel machine learning approach to differentiate responders from non-responders using baseline pre-treatment CT imaging in advanced-stage NSCLC patients treated with immune checkpoint inhibitors.

      Method

      The QTA was applied separately on 50 contrast and 50 non-contrast CT images of histologically confirmed NSCLC patients. All patients included were treated with second-line immunotherapy (nivolumab, pembrolizumab or atezolizumab). Three-dimensional tumor segmentation was performed using the 4.10 version of 3D Slicer, and a total of 104 CT parameters from each CT image was obtained. For data pre-processing and standardization, we used the Sklearn machine learning library in Python programming language, reducing the number of CT parameters by Principal Component Analysis (PCA). The components thus obtained were further analyzed with hierarchical cluster analysis. According to QTA, responders were differentiated from non-responders based on naïve Bayes and k-means clustering. Response was defined as complete response, partial response, or stable disease at the first follow-up CT scan after immunotherapy initiation while non-response was defined as either visible signs of progression at time of or death prior to first follow-up CT scan. To verify the accurateness of the machine learning algorithms, leave-one-out cross-validation was performed.

      Result

      Overall, we analyzed the CT scans of 88 advanced-stage NSCLC patients including 40 woman and 48 men. PCA identified eight major principal components which characterize the 104 CT features with an accuracy of 90%, suggesting no loss of essential CT-related information. We further evaluated these eight principal components and hierarchical cluster analysis clearly identified two major subgroups both in the contrast and non-contrast CT image group. According to the QTA, based on naïve Bayes clustering, the machine learning algorithm was able to differentiate responders from non-responders with an accuracy of 66.6% in the contrast CT image group but was not predictive in the non-contrast group. K-means clustering also showed an accuracy of 66.6%, thus confirming the results of the naïve Bayes clustering.

      Conclusion

      Advanced-stage NSCLC patients can be classified into two major subgroups according to their CT features principal components with hierarchical cluster analysis using both contrast and non-contrast CT images. The clinical relevance of these principal component related subgroups should be further investigated in future prospective studies. By evaluating the contrast CT images, machine learning algorithms based on naïve Bayes and k-means clustering may predict the response to immunotherapy, however, further studies on “Big Data” are needed to define the exact prognostic value of QTA in NSCLC patients regarding immunotherapy administration. Application of QTA to prognostication for progression-free survival and overall survival is in progress.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.