Alicja Puszkiel (1,2), Pascaline Boudou-Rouquette (3), Jennifer Arrondeau (3), Marie Wislez (3), Elizabeth Fabre (4), Guillaume Bianconi (1), Xavier Declèves (1,2), Anne Jouinot (3,5), De Percin Sixtine (3), Jérôme Alexandre (3), François Goldwasser (3,5), Benoit Blanchet (1)
1) Biologie du Médicament - Toxicologie, Hôpital Cochin, AP-HP, Paris, France 2) Université de Paris, Inserm UMR-S1144, Paris, France 3) Medical Oncology Department, Hôpital Cochin, AP-HP; Cancer Research for PErsonalized Medicine (CARPEM), Paris, France 4) Thoracic Oncology Department, Hôpital Européen Georges Pompidou (HEGP), AP-HP; Cancer Research for PErsonalized Medicine (CARPEM); Paris University, France 5) Immunomodulatory Therapies Multidisciplinary Study group (CERTIM), Hôpital Cochin, AP-HP, Paris, France
Objectives: Nivolumab, an immune checkpoint inhibitor targeting PD1, is currently approved in multiple indications including non-small cell lung cancer (NSCLC) [1]. Although it has radically improved cancer treatment, the clinical benefit is observed in approximately 20% of patients [1]. Identification of patients who might benefit from the treatment is of high importance. Basal hypermetabolism (increased resting energy expenditure (REE) by more than 10%) was recently correlated with worse response to anti-PD1 immunotherapy in metastatic NSCLC [2]. Hypermetabolic patients are at risk of malnutrition and cachexia which are known to be deleterious prognostic factors. In addition, patients who present a decrease in nivolumab elimination clearance (CL) over time have a better best overall response (BOR) [3]. This phenomenon was associated with an improvement in disease state (cachexia, inflammation) and as a result, a decrease in catabolic elimination of proteins including nivolumab [4]. Since both REE and nivolumab CL are prognostic factors of treatment efficacy and both are hypothesized to be associated with cachexia, we aimed to evaluate the association between REE and nivolumab CL, and whether they are independent predictive factors of nivolumab efficacy in NSCLC patients.
Methods: Plasma nivolumab concentrations were collected in NSCLC patients included in the ELY study (NCT04879316). Concentration-time data were analyzed using population approach in Monolix (version 2023R1). Time-varying function on CL was tested in the base model [5]. The following covariates were tested for their impact on nivolumab parameters: sex, age, body weight (BW), baseline serum albumin, C-reactive protein (CRP), ECOG performance status (PS), Glasgow Prognostic Score (GPS) and basal REE. REE was measured (mREE) using ambulatory indirect calorimetry and compared with the theoretical (tREE) value prior to treatment initiation. Hyper- and hypometabolism were defined as mREE increased or decreased by more than 10% compared to tREE, respectively. The remaining patients were classified as normometabolic. Covariate analysis was performed according to COSSAC method [6]. The final model was validated using prediction-corrected visual predictive check (pcVPC) and used to obtain individual nivolumab CL. Survival analysis using Cox proportional Hazard models was performed in R software. The primary endpoint was progression-free survival (PFS) defined as time since treatment start until disease progression or death from any cause.
Results: A total of 98 patients and 324 nivolumab concentrations were included in the analysis. Concentration-time data were described using a two-compartment model with linear CL. Inclusion of a time-varying function on CL did not improve the model fit. The final estimates of CL, central (V1) and peripheral (V2) volumes of distribution and intercompartmental clearance (Q) were 0.228 L/day [RSE = 4.65%] (IIV = 34.0% [14.5%]), 4.23 L [7.84%] (IIV = 23.3% [40.6%]), 2.83 L [19.0%] (IIV = 50.8% [39.9%]) and 0.770 L/day (fixed), respectively. The residual additive and proportional error components were 1.75 mg/L [26.8%] and 13.3% [16.2%], respectively. Two significant covariates were identified in the multivariate analysis: baseline albumin was inversely correlated with CL (βalbumin = -0.914 [27.1%], p=0.0003) and BW was correlated with V1 (βBW = 0.867 [35.0%], p=0.002). Basal REE was not associated with nivolumab CL. Inclusion of these covariates decreased the IIV in CL and V1 from 37.7% to 34.0% and from 33.2% to 23.3%, respectively. Median PFS was 3.43 months (CI95=2.62-5.18). In the univariate Cox analysis, nivolumab CL ≥ median (p=0.016), baseline CRP (p=0.02), basal hypermetabolism (p<0.00001) and BW loss > 5% in the last 3 months (p=0.08) were significantly associated with PFS. In the multivariate step, nivolumab CL ≥ median (HR=1.77, CI95=1.11-2.82, p=0.017), basal hypermetabolism (HR=4.19, CI95=2.40-7.32 for hyper- vs hypo- or normometabolism, p<0.00001) and BW loss > 5% (HR=2.58, CI95= 1.49-4.47, p=0.0007) were independently associated with shorter PFS.
Conclusions: Basal REE did not influence nivolumab CL but both variables were independent predictors of PFS in NSCLC patients. These results suggest that basal hypermetabolism not only reflects cachexia but also a reduced ability of the patient to develop an active immune response to nivolumab.
References:
[1] Desnoyer et al. Eur J Cancer 2020. 128: 119–128.
[2] Boudou-Rouquette et al., EBioMedicine 2021. 73: 103630.
[3] Liu et al. Clin Pharmacol Therapeutics 2017. 101: 5.
[4] Chatelut et al. Pharmacol Res Perspect. 2021; 9:e00757.
[5] Bajaj et al. CPT Pharmacometrics Syst. Pharmacol. 2017. 6, 58–66.
[6] Ayral et al. CPT Pharmacomet Syst Pharmacol. 2021;10 (4): 318–29.
Reference: PAGE 32 (2024) Abstr 11224 [www.page-meeting.org/?abstract=11224]
Poster: Drug/Disease Modelling - Oncology