Clinical efficacy of pembrolizumab is influenced by target occupancy in non-small cell lung cancer (NSCLC) patients: a joint target-mediated drug disposition (TMDD) – survival model

Clinical efficacy of pembrolizumab is influenced by target occupancy in non-small cell lung cancer (NSCLC) patients: a joint target-mediated drug disposition (TMDD) – survival model | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 25 February 2025 V1 Latest version Share on Clinical efficacy of pembrolizumab is influenced by target occupancy in non-small cell lung cancer (NSCLC) patients: a joint target-mediated drug disposition (TMDD) – survival model Authors : Marion Ferreira , Thomas Sécher , Sarah Lobet , Desvignes Céline , Lilou Guillot , Marion Pronost , Gilles Paintaud , Theodora Bejan-Angoulvant 0000-0002-0018-9996 , Nathalie Heuzé-Vourc'h , and David Ternant 0000-0003-4020-4545 [email protected] Authors Info & Affiliations https://doi.org/10.22541/au.174047957.71198545/v1 455 views 206 downloads Contents Abstract Introduction Methods Results Discussion Tables Figure legends References Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Introduction. Pembrolizumab is an anti-programmed-death 1 (PD-1) monoclonal antibody used in non-small cell lung cancer (NSCLC) for which dose-concentration-response relationship remains unclear. Our aim was to assess this relationship using joint population target-mediated drug disposition (TMDD) and parametric survival modeling. Methods. This prospective observational monocentric study (RICEPS, NCT04804137) included 19 NSCLC patients who received 200 mg pembrolizumab infusions every 3 weeks. Blood samples were collected at each visit before infusion. Pembrolizumab pharmacokinetics and hazard function for progression were described using one compartment TMDD Wagner model and a log-logistic model, respectively. The association of body surface area (BSA), baseline white blood cell count (WBC) and programmed-Death Ligand 1 tumor expression ratio (PDL1R) with pharmacokinetic parameters and progression-free survival PFS was assessed. Results. Pembrolizumab volume of distribution (V=5.3 L/m2) increased with BSA (p=0.025) while estimated baseline target level (R0=0.22 nM.G-1.L) increased with WBC (p=0.033). Other parameters of the model were clearance (CL=0.19 L/day), target elimination rate (kdeg=0.17 day-1) and pembrolizumab-target steady-state dissociation constant (KSS=4.9 nM). Hazard to progress was halved for concentration and target occupancy of C50=8 mg/L and R50=0.10 nM, respectively. Discussion. This study is the first that investigated the relationship between pembrolizumab pharmacokinetics, target occupancy and PFS. We observed target-mediated nonlinear pharmacokinetics and an association between increased pembrolizumab concentrations and improved clinical efficacy. These results do not support the use of flat dose, but rather suggest a benefit of individual dosing optimization. Introduction Pembrolizumab is an anti-programmed-death 1 (PD-1) humanized immunoglobulin G4 (IgG4) approved in several malignancies, including non-small cell lung cancer (NSCLC). The initial approved dosing regimen of pembrolizumab consisted in 2 mg/kg every 3 weeks (Q3W) in adults. However, this dosing regimen was changed in flat doses of 200 mg/Q2W or 400 mg/Q6W because weight-adjusted and flat doses were suggested to provide similar exposure distribution and clinical efficacy 1 . Up to date, the pharmacokinetics of pembrolizumab was investigated using compartmental modeling in only 5 publications, with contradictory results. Indeed, two studies reported different results although based on almost the same data (KEYNOTE-001, -002 and -006) 2–6 . Both used a two-compartment pharmacokinetic model, but one reported a clearance decreasing over time in almost all patients, linked to a decrease in disease severity over time 6 . However, previous works reported a possible nonlinear decay in pembrolizumab concentrations over time, which may not be attributed to variations in disease severity, but rather to target-mediated drug disposition (TMDD) 7–9 . Besides, most of previous publications reported no relationship between dose, exposure, tumor response and clinical response for tested dosing regimens, although an association between shortened time-to-event and higher pembrolizumab clearance was reported 10–13 . It was suggested that the apparently “flat” exposure-response relationship was attributable to sufficient target saturation even at low doses 7 . The approved pembrolizumab 200 mg/Q3W flat dosing regimen might lead to concentrations above those required for full target saturation, in addition to being economically ineffective. In addition, it may potentially expose patient to a higher risk of immune-related adverse events, even if the link between dose/exposure and adverse events is less documented than for anti-cytotoxic T lymphocyte–associated antigen 4 (CTLA-4) inhibitors 14,15 . In routine practice, dose-tapering based on pembrolizumab concentration measurements was suggested to be no less clinically effective and to be cost-effective 16 . Besides, some authors proposed strategies allowing dose-tapering while preserving sufficient target saturation 1,12 . Nevertheless, these evaluations used a linear two-compartment model, which accounted neither for time-varying clearance nor for TMDD 5 . Thus, available pharmacokinetic and PK-PD studies do not allow investigating the clinical benefit of alternative dosing regimens. The objectives of the present study were therefore to describe the relationship between pembrolizumab concentrations, target occupancy and clinical response using joint TMDD pharmacokinetic model and time-to-event modeling, and to investigate the interest of alternative dosing regimens. Methods Study design and data The present study is a post-hoc analysis of a prospective, observational, open, bicentric (Tours, France and Los Angeles, California, United States of America) study. The main objective of this study was to describe local and systemic anti-tumor immune system of patients treated with immune checkpoint inhibitors. It was conducted between May 2021 and August 2023 (NCT04804137, clinicaltrials.gov). This study was approved by local ethic committee and all participants gave written informed consent at the time of enrolment. The present study focused on 19 patients with NSCLC and treated in first-line setting with pembrolizumab (with or without chemotherapy) in Tours university hospital, Tours, France (table 1). Patients received 200 mg pembrolizumab infusions every three (Q3W) or 400 mg infusions every six weeks (Q6W) until unacceptable toxicity or progression. Chemotherapy regimen used was carboplatin and pemetrexed as patients had non-squamous NSCLC. Blood samples were collected to measure pembrolizumab concentrations before perfusions 1 to 4, then every three infusions and, in case of progression, every 8 weeks after progression. Pembrolizumab concentrations were measured using a validated enzyme-linked immunosorbent assay (ELISA) technique. The limit of detection (LOD), lower (LLOQ) and upper (ULOQ) limits of quantitation were 0.03 mg/L, 0.09 mg/L and 10 mg/L, respectively. Pembrolizumab concentrations that were < LLOQ were left-censored. Overall, 110 pembrolizumab concentrations, including 4 values < LLOQ, were available. Time to progression (TTP) was defined as the time from the first pembrolizumab administration to disease progression or death from any cause and was used to compute progression-free survival (PFS). Progression did not occur during follow-up in 4 patients and was right-censored at the time of the last known follow-up visit. The following demographic and biological characteristics were collected at baseline (table 1): age (years), sex, body weight, body surface area, leukocyte, neutrophile, lymphocyte, and monocyte counts (G/L), haemoglobin (g/dL), treatment (pembrolizumab alone or associated with chemotherapy), Eastern Cooperative Oncology Group Performance Status, smoking status (current/former/no smoking), metastases at diagnostic and programmed-cell death ligand 1 tumor proportion score. For the latter one, immunohistochemical staining are performed on slides from FFPE tissue, after dewaxing, using a Ventana BenchMark XT Platform (Roche Diagnostics, Basel, Switzerland) as instructed. PD-L1 antibody was used (clone 22C3, Dako) at dilution 1/50. Joint model development A joint pharmacokinetic-TTP model was built to assess the relationship between pembrolizumab concentrations, target occupancy and PFS. First, the pharmacokinetic model was designed and parameters were estimated by fitting the model to pembrolizumab concentration-time data. Then, estimations of pharmacokinetic parameter distributions using the final pharmacokinetic model were used as prior to the TTP model. Software Analysis of pharmacokinetic and TTP data was carried out using Monolix Suite 2023 (Lixoft®, Antony, France). A large number of iterations (1000 and 300 iteration kernels 1 and 2, respectively) and 3 Markov chains were used. Likelihood and Fisher Information Matrix were computed using importance sampling and stochastic approximation, respectively. Model-based simulations were computed using Simulx software (Lixoft®, Antony, France). Pharmacokinetic model To develop the structural pharmacokinetic model, linear one- and two-compartment models were first considered. Then, models accounting for potential non-linear pharmacokinetics of pembrolizumab were considered: - exponential increase or decay of clearance over time (time-varying CL), as CL = CL 0 + CL 1 ∙ e −kdes∙t where CL 0 is the initial value of the parameter CL, CL 1 is an additive parameter and k des is a first-order constant describing the rate of decrease of CL 1 over time. - quasi-steady-state (QSS) approximation of target-mediated drug disposition (TMDD) model, which assumes an immediate association of pembrolizumab with its target 17,18 - Wagner approximation, additionally assuming a constant value for total target (unbound + bound to pembrolizumab) 19,20 ; - Michaelis-Menten model approximation, additionally assuming irreversible binding of pembrolizumab to its target 18 Statistical interindividual models for pharmacokinetic parameter distribution were all assumed exponential. Interindividual standard deviations (ω) were fixed to 0 when relative standard errors (RSE) and/or shrinkage values were high (>65%). Additive, proportional and mixed additive-proportional residual error models were tested. Time-to-progression model Description of time-to-progression data was attempted using parametric risk-proportional modeling using several models for the baseline hazard function (h 0 ): exponential, Weibull, Gompertz and log-logistic models (supplemental file). Joint pharmacokinetic and TTP model Joint pharmacokinetic-TTP modeling consisted in finding the best link between pembrolizumab pharmacokinetics and hazard function, i.e. h(t) = h 0 (t). LINK where h(t) is hazard function and LINK is the function that links pembrolizumab (or target) kinetics and hazard function. As in our previous publications, we assumed three possible relationships 20 : - Absence of influence of pembrolizumab pharmacokinetics: LINK = 1 - C-link: pembrolizumab acting as an inhibitor of hazard, described as a direct (Emax) relationship: \(LINK=1-\frac{C_{c}}{C_{50}+C_{c}}\)- R-link: target unbound to pembrolizumab acting as activator of hazard, described as a direct (Emax) relationship:\(LINK=\ \frac{R}{R_{50}+R}\)Where C c and R are unbound pembrolizumab concentrations in central compartment and unbound target levels, respectively, C 50 and R 50 are values of unbound pembrolizumab concentrations and target levels leading to a 50% decrease in baseline progression hazard, respectively. Model selection and evaluation Structural models were compared using Bayesian information criterion (BIC) which combines the –2 ln.likelihood (–2LL) and the number of parameters to be estimated. Model with the lowest BIC value was selected and the difference in BIC (ΔBIC) was reported. Statistical (interindividual, residual and covariate) models were compared and selected using the likelihood ratio test (LRT), assuming a difference in –2LL of two nested models following a χ 2 distribution. Pharmacokinetic models were evaluated using goodness-of-fit plots: observed vs. population (PRED) and individual fitted concentrations (IPRED); population (WRES) and individual (IWRES) weighted residuals vs. PRED and IPRED; visual predictive checks (VPC) and normalized prediction distribution errors (NPDE) were performed by simulating 1000 replicates using both fixed and random parameters of the final model. Covariates Due to the limited number of patients, only a few covariates (COV) could be tested: body surface area (BSA), programmed-cell death ligand 1 tumor proportion score (PD-L1 TPS) and leucocyte counts (LEU). Covariate analysis was performed on both PK and TTP models. Influence of covariates on pharmacokinetic structural parameters . To minimize the number of parameters to be estimated, the influence of continuous covariates was tested by estimating each structural parameter as unit of covariate, e.g. BSA on volume of distribution (V) consisted in estimating V in L/m 2 . PD-L1 TPS was tested as categorical covariate, with reference category being PD-L1 < 50% vs. other category being PD-L1 ≥ 50%. The effect of PD-L1 was coded as ln(θ TV,PD-L1≥50% ) = ln(θ PD-L1<50% ) + β PD-L1≥50% , where (θ TV,PD-L1≥50% is typical value of structural parameter θ for PD-L1 ≥ 50%, θ PD-L1<50% is θ value for reference category and β PD-L1≥50% is power parameter leading to the value for PD-L1 ≥ 50%. Structural parameters involved in covariate testing were those associated with estimated ω value. The association of covariates with parameter distribution was assessed in two steps. During the univariate step, the association of each covariate with structural parameters was tested separately. Covariates significantly associated with structural parameters (α 1 < 0.1) were retained for multivariate step. The limited number of covariates allowed a multivariate backward stepwise step (α 2 < 0.05), for which each covariate was removed from the full model. Covariates which exclusion led to p < α 2 were retained in the final model. Regarding time-to-progression model, covariates were coded as: h(t) = h 0 (t) . exp(β.COV). Simulations Simulations were conducted to assess the influence of (i) baseline target levels (R 0 ), (ii) covariates associated with R 0 and (iii) dosing regimen alterations on unbound pembrolizumab concentrations, unbound target levels and PFS. Parameter estimates from the final models were used. Continuous covariate values and categorical covariate proportions of virtual patients were drawn from Gaussian distributions of continuous covariates or according to the observed proportion of study patients, respectively. As for R 0 and continuous covariate levels, the 5 th , 50 th and 95 th percentiles were simulated. Basic simulated dosing regimen was 200 mg Q3W. Altered dosing regimens were 100 mg and 400 mg Q3W and 200 mg and 400 mg Q2W. Results Pharmacokinetic model A two-compartment model did not lead to better fitting of concentration-time data (ΔBIC=+0.07) than a one-compartment model. The TMDD model with Wagner’s approximation led to the best description of pembrolizumab concentration-time data (figure 1). This model was better than a linear one-compartment model (ΔBIC=−11.87), time-varying clearance model (ΔBIC=−7.75), QSS approximation (ΔBIC=−3.06) or Michaelis-Menten approximation (ΔBIC=−3.77). The interindividual standard deviations were estimable for volume of distribution (V), clearance (CL) and baseline target level (R 0 ); other standard deviations were poorly estimable and were set to zero. Best residual error model was proportional (table 2). The univariate step revealed an association of body surface area (BSA) and of PD-L1 TPS with volume of distribution and of leukocyte count (LEU) with R 0 . The multivariate step confirmed these associations; however, to optimize the number of parameters to be estimated, V was estimated by unit of BSA (m 2 ) and R 0 by unit of LEU (G/L): V was higher when PD-L1 tumor expression was high (ΔLL=4.8, p=0.028) and increased with BSA (ΔLL=5.04, p=0.025), while R 0 increased with LEU (ΔLL=4.52, p=0.033). All pharmacokinetic parameters were estimated with acceptable accuracy (table 2) and no obvious bias was observed (Supplemental file, figure S1). Two secondary parameters could be derived from Wagner model: - Saturable elimination rate of pembrolizumab (V max ) calculated as: V max = R 0 .k int .V = 0.22 x 9.5 x 0.17 x 5.3 = 1.9 mg/day; - Critical concentration (C crit ), i.e. pembrolizumab concentration associated with the onset of nonlinear elimination decay, calculated as C crit = (R 0 .k int .V)/CL = 9.5 mg/L 21 . Joint pharmacokinetic-TTP model Time-to-progression (TTP) data were better described using parametric log-logistic function for the baseline hazard function (h 0 ) than exponential (ΔBIC=−2.12), Weibull (ΔBIC=−2.96) or Gompertz (ΔBIC=−4.88) functions. The joint pharmacokinetic-TTP model with R-link led to better performance than no link (ΔBIC=−4.66) and C-link (ΔBIC=−2.21). None of tested covariates were significantly associated with TTP. All joint pharmacokinetic-TTP model parameters were accurately estimated (table 3) with no bias or model misspecification (Supplemental file, figure S1). Simulations Simulated pharmacokinetic, target occupancy and TTP profiles following repeated pembrolizumab 200 mg Q3W doses with increasing R 0 or LEU (figure 2) led to decreased pembrolizumab unbound concentrations, increased R levels and shortened PFS. High PD-L1 tumor expression (>50%) was associated with no significant alteration of pembrolizumab concentrations, R levels or of PFS (figure 2). In addition, simulations suggested that intensified dosing regimen led to increased pembrolizumab concentrations, resulting in higher target occupancy and longer PFS (figure 3): dosing regimens of 100 mg, 200 mg and 400 mg Q3W, and 200 mg and 400 mg Q2W were associated with PFS at 25 months of 0.12, 0.23, 0.37, 0.29 and 0.46, respectively. Discussion To our knowledge, this is the first study investigating the dose-concentration-response relationship of pembrolizumab in NSCLC using TMDD and joint pharmacokinetic-TTP modeling framework. These models allowed us to quantify the relationship between pembrolizumab concentrations, target engagement and PFS. The first part of this work focused on characterizing pembrolizumab pharmacokinetics and its target engagement over time. While previous studies analyzed pembrolizumab pharmacokinetics using compartmental models 7,8,22 , only one reported nonlinear elimination decay which was described using a Michaelis-Menten approximation of the TMDD model 7 . This has to be mentioned since, to our knowledge, nonlinear decay of concentrations over time was never reported for other anti-PD-1 or anti-PD-L1 monoclonal antibodies. Our findings support this nonlinearity, with TMDD providing a better description of the concentration-time data than time-varying clearance models. This suggests that pembrolizumab pharmacokinetics is primarily driven by target-mediated elimination rather than changes in clinical status over time. Our TMDD model allowed us to estimate a steady-state dissociation constant (Kss) of 4.9 nM, which is more than ten-fold higher than the value reported by Elssaiss-Schaap et al. (0.53 nM) 7 . Nevertheless, it is difficult to decipher which value is the most likely, since both estimates are inferior to the value of the dissociation constant measured in vitro (42 nM ) 22 . In addition, our V max estimate (1.9 mg/day) is superior to the one of Elssaiss-Schaap (0.113 mg/day) 7 . This difference might be explained either by differences in modeling techniques (Wagner vs. Michaelis-Menten models) or in included patients. Our k int (0.17 day -1 ) estimate is near the median [interquartile range] k int value from 55 previous publications of 0.15 day -1 [0.059 – 0.82] (updated from 23 ). Additionally, our calculated critical concentration (C crit ) of 9.5 mg/L suggests that pembrolizumab concentrations below this threshold may trigger nonlinear elimination decay, potentially associated with impaired therapeutic efficacy. One of the most relevant findings from our analysis is the influence of leukocyte counts on pembrolizumab pharmacokinetics. Leucocyte count, which includes various PD-1-expressing immune cell types such as T cells, B cells, NK cells and monocytes, was positively correlated with estimated baseline target levels (R 0 ). This finding reinforces the hypothesis that pembrolizumab undergoes target-mediated elimination, as high leukocyte count appears to be associated with high drug-target interactions, further supporting the role of immune cells in influencing pembrolizumab pharmacokinetics. To better understand the complex relationship between pembrolizumab concentrations, target engagement, and clinical outcomes, we employed a joint PK-TTP model. This integrated approach allowed us to assess the impact of drug exposure on time-to-progression (TTP) while accounting for potential biases in the exposure-response relationship 20,24,25 . Our model revealed that concentrations of pembrolizumab ≥8 mg/L and target occupancy ≥90% were associated with a 50% reduction in the hazard of progression. This finding highlights the importance of maintaining sufficient drug concentrations to sustain high target occupancy, as lower exposure may lead to reduced target engagement and accelerated disease progression. Our simulations further confirmed that higher baseline target levels and leukocyte counts, combined with decreased drug exposure, led to shorter time to progression, emphasizing the need to optimize pembrolizumab dosing to ensure sustained efficacy. Interestingly, contrary to previous studies 7,10–12 , our results support a relationship between target engagement and clinical efficacy. Indeed, patients with high baseline target levels may be at risk of underexposure to pembrolizumab, and thus of treatment failure. Our results therefore challenge both strategies of ”one-size-fits-all” dosing and dose-tapering. Indeed, our study suggest a benefit of model-informed precision dosing (MIPD) 26 . This approach, based on limited sampling strategy and Bayesian analysis, allows to estimate individual pharmacokinetic parameters and to define optimal dosing scheme given a target in drug exposure. While patients with high target engagement levels would benefit of dose-tapering to reduce financial toxicity, patients with insufficient target engagement would benefit of intensified dosing to reduce the risk of relapse. Of note, in the context of TMDD, relying on pembrolizumab serum trough concentrations or in linear pharmacokinetic models may lead to an over-estimation of drug exposure, potentially resulting in under-dosing and compromising treatment effectiveness 16 . Model-informed prediction dosing necessitates a TMDD model, as previously demonstrated for eculizumab, an antibody targeted against the C5 fraction of complement 27 . These trends reinforce the need to personalize and optimize dosing-regimens of anti-PD-1 or anti-PD-L1 monoclonal antibodies. First, finding the optimal dose that maximizes efficacy while minimizing toxicity is key 28 . In clinical practice, the goal is to balance the benefits of immune activation against cancer cells with the risk of severe immune related adverse events, that need sometimes long-term management and inevitably health costs 29 . This is why fixed dosing regimens are commonly used, as they have been shown to offer a manageable safety profile while providing effective tumor control, occasionally with extended interval dosing and double dose 2,30,31 . Second, the cost-effectiveness of these drugs remains unclear. While ICIs have demonstrated substantial improvements in overall survival (OS) and PFS compared to traditional therapies, the high upfront cost leads to a higher cost per quality-adjusted life year (QALY) 32,33 . The threshold for cost-effectiveness varies by healthcare system, but ICIs often exceed these thresholds, particularly when the cost of long-term therapy and management of side effects are factored in. There is ongoing debate on the long-term sustainability of funding for high-cost cancer therapies like ICIs. Healthcare systems must weigh the potential interest of improved survival against the financial burden and explore models that balance innovation with cost containment. In the future, the role of interventional pharmaco-economics may be crucial 34 . Our study presents some limitations. First, data were obtained in a limited cohort of patients treated with approved dosages, which (i) hampered us to use a two-compartment structural model and (ii) may have limited a thorough description of interindividual pharmacokinetic variability. Even if no structural parameter value had to be fixed, our data did not allow to estimate interindividual standard deviations of some parameters (K SS , k int ) and imposed linear relationships between structural parameters and covariates, therefore decreasing the power of covariate testing. Second, target level measurements were not available, as in all previous anti-PD-1 monoclonal antibody pharmacokinetic studies. It may have hampered the quantification of the actual target-mediated elimination of pembrolizumab, even if estimating TMDD parameters using only concentration-time data has the advantage of dealing with total level of targets. Third, as almost all TMDD models, we assumed that turnover of targets was in equilibrium which, in the context of oncology, might result in an over-simplistic TMDD model. Taken together, these limitations impose to consider our results as indicative trends rather than accurate predictions. In conclusion, our study provides valuable insights into the dose-exposure-response of pembrolizumab in NSCLC. By using TMDD and joint pharmacokinetic-TTP modeling, we demonstrated for the first time the strong interplay between exposure to pembrolizumab, target occupancy, and clinical efficacy. Our findings could provide the basis for a valuable MIPD tool, maximizing both treatment efficacy and cost-effectiveness. Developing optimized immune checkpoint inhibitor protocols will be challenging, but it is a crucial step toward ensuring global access. Funding This study was made on data from the NCT00489697 trial (clinicaltrials.gov) and was partly supported by the Higher Education and Research Ministry under the program “Investissements d’avenir” Grant Agreement: LabEx MAbImprove ANR-10-LABX-53-01. Acknowledgements The authors thank the investigators who included the patients in this study, Jean-Philippe Metges, Antoine Adenis, Jean-Luc Raoul, You Heng Lam, Roger Faroux, Claude Masliah, Virginie Berger and Erick Gamelin; Anne-Claire Duveau and Caroline Guerineau-Brochon for technical assistance Conflict of interest statement Marion Ferreira acted as a consultant for and received fees by AstraZeneca and Janssen. Sarah Lobet is currently employed by Calvagone SAS; however, this work was part of her PhD work, conducted at Tours University, and is independent of her professional duties at Calvagone SAS. Gilles Paintaud Gilles Paintaud has received grants for his research team from Roche Pharma, Chugai, Pfizer, Novartis, Lundbeck and Sanofi-Genzyme. Nathalie Heuzé-Vourc’h acted as as a consultant for Novartis, European Commission, Immune Biosolution, ANR, Alvea/Telis Bioscience and received grants for his research team from Aerogen, Nemera, Aptar Pharma, Resyca, Affilogic and 4P-Pharma. David Ternant acted as a consultant and has given lectures on behalf of his institution for Astra-Zeneca, Amgen, Novartis, Lundbeck, Roche and Sanofi. Thomas Sécher, Céline Desvignes, Lilou Guillot, Marion Pronost and Theodora Bejan-Angoulvant have no conflict of interest to declare regarding the study. Author contributions Marion Ferreira designed the study, managed and analyzed the data interpreted the results and wrote the manuscript. Thomas Sécher participated to data acquisition, interpreted the results and reviewed the manuscript. Sarah Lobet analyzed the data, interpreted the results and reviewed the manuscript. Céline Desvignes managed the data, participated to data acquisition and reviewed the manuscript. Lilou Guillot and Marion Pronost participated to data acquisition and reviewed the manuscript. Gilles Paintaud and Theodora Bejan-Angoulvant interpreted the results and reviewed the manuscript. Nathalie Heuzé-Vourc’h designed the study, interpreted the results and reviewed the manuscript. David Ternant designed the study, analyzed the data, interpreted the results and wrote the manuscript. Ethics approval and consents This study was registered in ClinicalTrials.gov database (NCT04804137) and was approved by the local ethics committee. Written informed consent was obtained from all patients. Data availability statement Data, material and codes are available on request to the corresponding author. Tables Table 1. Summary of patient characteristics nb. pembrolizumab infusions 8 [3-30] Treatment Pembrolizumab alone 6 (32 %) Pembrolizumab + chemotherapy 13 (68 %) Sex Male 11 (58 %) Female 8 (42 %) Age (years) 62 [42 − 83] Body weight (kg) 65 [39 − 87] BMI (kg/m 2 ) 21.7 [16.9 − 29.1] Body surface area (m 2 ) 1.78 [1.28 − 2.08] Leukocyte count (G/L) 9.5 [4.7 − 14.2] Neutrophile count (G/L) 6.5 [2.6 − 10.4] Lymphocyte count (G/L) 1.7 [0.6 − 2.6] Monocyte count (G/L) 0.8 [0.4 − 1.4] Hemoglobin (g/dL) 13.5 [9.7 − 15.9] Smoking status Current smoker 8 (42 %) Former smoker 11 (58 %) ECOG PS 0 8 (42 %) 1 11 (58 %) PD-L1 Tumor Proportion Score Highly positive (> 50%) 9 (47 %) Positive (1-49%) 7 (37 %) Negative (< 1%) 3 (16 %) Metastases at diagnostic Lung and/or pleura 5 (26 %) Brain 2 (10 %) Extra-thoracic 14 (74 %) Data are given in median [range] or in number (%). Table 2. Population pharmacokinetic parameter estimates Unit Estimate RSE Unit Estimate RSE V 1 L 10.2 7.2 L.m -2 5.3 5.9 PD-L1-bio_V1 ─ 0.18 13 ─ 0.23 46 CL L.day -1 0.18 13 L.day -1 0.20 11 R 0 nM 2.2 34 nM.G -1 .L 0.22 38 k int day -1 0.11 12 day -1 0.17 14 K SS nM 2.8 26 nM 4.9 40 ω V1 ─ 0.21 36 ─ 0.13 39 ω CL ─ 0.50 21 ─ 0.46 18 ωR R0 ─ 1.3 29 ─ 1.0 26 σ prop ─ 0.19 9.5 ─ 0.19 9.6 Parameter estimates are given for base and final models. Legends: RSE relative standard error, V volume of distribution, CL clearance, R 0 baseline target amount, k int first-order free target elimination and complex elimination rate constant, K SS steady-state dissociation constant, ω interindividual standard deviation, σ prop residual proportional error, PD-L1 tumor expression of PD-L1. Table 3. Estimates of joint pharmacokinetic-time-to-progression model est. RSE est. RSE est. RSE est. RSE Exponential 271 18 ─ ─ ─ ─ ─ ─ Weibull 275 19 1.4 3.3 ─ ─ ─ ─ Gompertz 207 18 2.1 14 ─ ─ ─ ─ Log-logistic 186 19 2.1 24 ─ ─ ─ ─ Log-logistic 83 14 5.5 18 8.0 30 ─ ─ Log-logistic 88 17 5.1 8.2 ─ ─ 0.10 36 Parameter estimates are given for base exponential, Weibull, Gompertz and log-logistic models, and then for join C-link and R-link log-logistic models. Legends: RSE relative standard error, Te scale parameter of survival function, s shape parameter of survival function, C 50 pembrolizumab concentration leading to a decrease in hazard function of 50%, R 50 target level leading to a decrease in hazard function of 50%. Figure legends Figure 1. Pharmacokinetic target-mediated drug disposition (TMDD) model and joint pharmacokinetic-time-to-progression model for pembrolizumab in NSCLC patients. Legends: C pembrolizumab serum concentration, R target levels, RC pembrolizumab-target complex levels, PFS progression-free survival, CL clearance, k syn zero-order target input rate constant, k int first-order target and complex elimination rate constant, K SS steady-state dissociation constant, h(t) hazard function, h 0 (t) baseline hazard function, R 50 target level associated with decrease in hazard of 50%. Figure 2. Pembrolizumab pharmacokinetics (A), target occupancy (B) and progression-free survival (C) profiles for (1, top) 5 th (pink) median (green) and 95 th (blue) percentile values of latent baseline target amount (R 0 ), (2, middle) 5 th (pink) median (green) and 95 th (blue) percentile values of leukocyte counts, and (3, bottom) PD-L1 tumor expression of less than (pink) or superior to (blue) 50%. Lines and areas are median and 90% prediction intervals, respectively. Figure 3 . Pembrolizumab pharmacokinetics (A), target occupancy (B) and progression-free survival (C) profiles for (1, top) dosing regimen of 100 mg every 3 weeks (pink), 200 mg every 3 weeks (green) and 400 mg every 3 weeks (blue), and (2, bottom) for dosing regimen of 200 mg every 2 weeks (pink) and 400 mg every 2 weeks (blue). Lines and areas are median and 90% prediction intervals, respectively. References 1. 1. Freshwater T, Kondic A, Ahamadi M, et al. Evaluation of dosing strategy for pembrolizumab for oncology indications. J Immunother Cancer . 2017;5(1):43. Garon EB, Rizvi NA, Hui R, et al. Pembrolizumab for the Treatment of Non–Small-Cell Lung Cancer. N Engl J Med . 2015;372(21):2018-2028. Ribas A, Puzanov I, Dummer R, et al. Pembrolizumab versus investigator-choice chemotherapy for ipilimumab-refractory melanoma (KEYNOTE-002): a randomised, controlled, phase 2 trial. Lancet Oncol . 2015;16(8):908-918. Robert C, Ribas A, Schachter J, et al. Pembrolizumab versus ipilimumab in advanced melanoma (KEYNOTE-006): post-hoc 5-year results from an open-label, multicentre, randomised, controlled, phase 3 study. Lancet Oncol . 2019;20(9):1239-1251. Ahamadi M, Freshwater T, Prohn M, et al. Model-Based Characterization of the Pharmacokinetics of Pembrolizumab: A Humanized Anti-PD-1 Monoclonal Antibody in Advanced Solid Tumors. CPT Pharmacomet Syst Pharmacol . 2017;6(1):49-57. Li H, Yu J, Liu C, et al. Time dependent pharmacokinetics of pembrolizumab in patients with solid tumor and its correlation with best overall response. J Pharmacokinet Pharmacodyn . 2017;44(5):403-414. Elassaiss-Schaap J, Rossenu S, Lindauer A, et al. Using Model-Based ‘Learn and Confirm’ to Reveal the Pharmacokinetics-Pharmacodynamics Relationship of Pembrolizumab in the KEYNOTE-001 Trial. CPT Pharmacomet Syst Pharmacol . 2017;6(1):21-28. de Vries F, Smit AAJ, Wolbink G, de Vries A, Loeff FC, Franssen EJF. Case report: Pharmacokinetics of pembrolizumab in a patient with stage IV non-small cell lung cancer after a single 200 mg administration. Front Oncol . 2022;12:960116. Ternant D, Azzopardi N, Raoul W, Bejan-Angoulvant T, Paintaud G. Influence of Antigen Mass on the Pharmacokinetics of Therapeutic Antibodies in Humans. Clin Pharmacokinet . 2019;58(2):169-187. Chatterjee MS, Elassaiss-Schaap J, Lindauer A, et al. Population Pharmacokinetic/Pharmacodynamic Modeling of Tumor Size Dynamics in Pembrolizumab-Treated Advanced Melanoma. CPT Pharmacomet Syst Pharmacol . 2017;6(1):29-39. Centanni M, Moes DJAR, Trocóniz IF, Ciccolini J, van Hasselt JGC. Clinical Pharmacokinetics and Pharmacodynamics of Immune Checkpoint Inhibitors. Clin Pharmacokinet . 2019;58(7):835-857. Renner A, Burotto M, Rojas C. Immune Checkpoint Inhibitor Dosing: Can We Go Lower Without Compromising Clinical Efficacy? J Glob Oncol . 2019;5:JGO.19.00142. Turner DC, Kondic AG, Anderson KM, et al. Pembrolizumab Exposure–Response Assessments Challenged by Association of Cancer Cachexia and Catabolic Clearance. Clin Cancer Res . 2018;24(23):5841-5849.14. Yin Q, Wu L, Han L, et al. Immune-related adverse events of immune checkpoint inhibitors: a review. Front Immunol . 2023;14:1167975. Shulgin B, Kosinsky Y, Omelchenko A, et al. Dose dependence of treatment-related adverse events for immune checkpoint inhibitor therapies: a model-based meta-analysis. Oncoimmunology . 2020;9(1):1748982. Wang S, Li Y, Liu Z, et al. Efficacy and safety of first-line immune checkpoint inhibitors combined with chemotherapy for extensive-stage small cell lung cancer: A network meta-analysis. Lung Cancer Amst Neth . 2023;178:47-56. Mager DE, Jusko WJ. General pharmacokinetic model for drugs exhibiting target-mediated drug disposition. J Pharmacokinet Pharmacodyn . 2001;28(6):507-532. Gibiansky L, Gibiansky E, Kakkar T, Ma P. Approximations of the target-mediated drug disposition model and identifiability of model parameters. J Pharmacokinet Pharmacodyn . 2008;35(5):573-591. Dua P, Hawkins E, van der Graaf PH. A Tutorial on Target-Mediated Drug Disposition (TMDD) Models. CPT Pharmacomet Syst Pharmacol . 2015;4(6):324-337. Lobet S, Paintaud G, Azzopardi N, et al. Relationship Between Cetuximab Target-Mediated Pharmacokinetics and Progression-Free Survival in Metastatic Colorectal Cancer Patients. Clin Pharmacokinet . 2023;62(9):1263-1274. Stein AM, Peletier LA. Predicting the Onset of Nonlinear Pharmacokinetics. CPT Pharmacomet Syst Pharmacol . 2018;7(10):670-677. Patnaik A, Kang SP, Rasco D, et al. Phase I Study of Pembrolizumab (MK-3475; Anti-PD-1 Monoclonal Antibody) in Patients with Advanced Solid Tumors. Clin Cancer Res Off J Am Assoc Cancer Res . 2015;21(19):4286-4293. Bensalem A, Ternant D. Pharmacokinetic Variability of Therapeutic Antibodies in Humans: A Comprehensive Review of Population Pharmacokinetic Modeling Publications. Clin Pharmacokinet . 2020;59(7):857-874. Kawakatsu S, Bruno R, Kågedal M, et al. Confounding factors in exposure–response analyses and mitigation strategies for monoclonal antibodies in oncology. Br J Clin Pharmacol . 2021;87(6):2493-2501. Dai HI, Vugmeyster Y, Mangal N. Characterizing Exposure–Response Relationship for Therapeutic Monoclonal Antibodies in Immuno-Oncology and Beyond: Challenges, Perspectives, and Prospects. Clin Pharmacol Ther . 2020;108(6):1156-1170. Pérez-Blanco JS, Lanao JM. Model-Informed Precision Dosing (MIPD). Pharmaceutics . 2022;14(12):2731. Le Tilly O, Gatault P, Semlali S, et al. Eculizumab dose tapering should take into account the nonlinearity of its pharmacokinetics. Br J Clin Pharmacol . 2024;90(5):1312-1321. Johnson DB, Nebhan CA, Moslehi JJ, Balko JM. Immune-checkpoint inhibitors: long-term implications of toxicity. Nat Rev Clin Oncol . 2022;19(4):254-267. Cantini L, Paoloni F, Pecci F, et al. Safety of extended interval dosing immune checkpoint inhibitors: a multicenter cohort study. J Natl Cancer Inst . 2023;115(7):796-804. Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial. Lancet Lond Engl . 2016;387(10027):1540-1550. Paz-Ares L, Luft A, Vicente D, et al. Pembrolizumab plus Chemotherapy for Squamous Non-Small-Cell Lung Cancer. N Engl J Med . 2018;379(21):2040-2051. Verma V, Sprave T, Haque W, et al. A systematic review of the cost and cost-effectiveness studies of immune checkpoint inhibitors. J Immunother Cancer . 2018;6:128. Ding H, Xin W, Tong Y, et al. Cost effectiveness of immune checkpoint inhibitors for treatment of non-small cell lung cancer: A systematic review. PloS One . 2020;15(9):e0238536. Wesevich A, Goldstein DA, Paydary K, Peer CJ, Figg WD, Ratain MJ. Interventional pharmacoeconomics for immune checkpoint inhibitors through alternative dosing strategies. Br J Cancer . 2023;129(9):1389. Crossref Google Scholar Information & Authors Information Version history V1 Version 1 25 February 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords immunology immunotherapy monoclonal antibodies pharmacodynamics pharmacokinetic-pharmacodynamic pharmacokinetics population analysis Authors Affiliations Marion Ferreira View all articles by this author Thomas Sécher View all articles by this author Sarah Lobet 1Inserm UMR 1069, Nutrition Croissance et Cancer (N2C), Tours University, Tours, France View all articles by this author Desvignes Céline Université de Tours View all articles by this author Lilou Guillot View all articles by this author Marion Pronost View all articles by this author Gilles Paintaud Universite Francois Rabelais de tours View all articles by this author Theodora Bejan-Angoulvant 0000-0002-0018-9996 University Hospital Tours View all articles by this author Nathalie Heuzé-Vourc'h View all articles by this author David Ternant 0000-0003-4020-4545 [email protected] Université de Tours View all articles by this author Metrics & Citations Metrics Article Usage 455 views 206 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Marion Ferreira, Thomas Sécher, Sarah Lobet, et al. Clinical efficacy of pembrolizumab is influenced by target occupancy in non-small cell lung cancer (NSCLC) patients: a joint target-mediated drug disposition (TMDD) – survival model. Authorea . 25 February 2025. DOI: https://doi.org/10.22541/au.174047957.71198545/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174047957.71198545/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a007982c19a641e2',t:'MTc3OTU3NjM2Mg=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();