Prediction of Neutropenia Risk Based on Exposure Information Without Plasma Concentration Measurement in Pemetrexed‒Platinum‒Based Chemotherapy: A Modeling Approach Using Real-World Clinical Data

Prediction of Neutropenia Risk Based on Exposure Information Without Plasma Concentration Measurement in Pemetrexed‒Platinum‒Based Chemotherapy: A Modeling Approach Using Real-World Clinical Data | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Prediction of Neutropenia Risk Based on Exposure Information Without Plasma Concentration Measurement in Pemetrexed‒Platinum‒Based Chemotherapy: A Modeling Approach Using Real-World Clinical Data Kazunori Morita, Keiichi Shigetome, Haruka Narise, Tetsuya Kaneko, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8989784/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 9 You are reading this latest preprint version Abstract Purpose Pemetrexed‒platinum chemotherapy is a key treatment option for non-squamous non-small cell lung cancer (NSCLC); however, its use is often limited by hematologic toxicity, particularly neutropenia. We aimed to model neutrophil dynamics following pemetrexed‒platinum chemotherapy using clinical data routinely available for patients and pharmacokinetic parameters of pemetrexed. Methods This real-world investigation included 86 patients with NSCLC who received pemetrexed‒platinum chemotherapy for model development, and an additional 83 patients who received the same chemotherapy plus pembrolizumab or bevacizumab as a validation cohort. We developed a nonlinear mixed-effects (NLME) model to predict neutrophil dynamics during the first cycle following pemetrexed‒platinum chemotherapy, using patient-specific clinical data collected before chemotherapy initiation and pemetrexed pharmacokinetic parameters derived from physiologically based pharmacokinetic (PBPK) modeling. Results The final model suggested that the area under the curve (AUC) 0–24 of pemetrexed, blood urea nitrogen value, and concomitant use of renin‒angiotensin system inhibitors influenced neutrophil suppression and delayed recovery. The goodness-of-fit plots, visual predictive check, nonparametric bootstrap analysis, and receiver operating characteristic (ROC) curves demonstrated good fit and high performance, with an area under the ROC curve (AUROC) of 0.92 for predicting severe neutropenia. The model maintained good performance in the validation cohort receiving antibody-containing regimens (AUROC 0.85). Conclusion This study demonstrated the importance of considering exposure to pemetrexed for an early prediction of neutropenia following pemetrexed‒platinum chemotherapy. Our model supports proactive pemetrexed dose optimization to prevent severe neutropenia, minimize treatment delays or interruptions, and preserve overall treatment intensity in patients undergoing pemetrexed‒platinum chemotherapy. Cancer chemotherapy Neutropenia Nonlinear mixed-effects model Pharmacometrics Physiologically based pharmacokinetic model Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Lung cancer was the most frequently diagnosed malignancy worldwide in 2022, with approximately 2.5 million new cases, representing 12.4% of all cancer diagnoses [ 1 ]. It also remains the leading cause of cancer-related death, with an estimated 1.8 million deaths, accounting for 18.7% of global cancer mortality [ 1 ]. Recent advances in diagnostic and therapeutic approaches have improved survival outcomes for patients with lung cancer [ 2 ]. For advanced non-small cell lung cancer (NSCLC) lacking driver gene mutations (approximately 50% of cases), the standard first-line treatment is either an immune checkpoint inhibitor (ICI) monotherapy or a combination of ICI and platinum-based chemotherapy [ 3 , 4 ]. In a phase III study, pemetrexed plus cisplatin demonstrated noninferiority in overall survival (OS) compared with gemcitabine plus cisplatin in untreated advanced NSCLC [ 5 ]. Subgroup analyses further showed that pemetrexed significantly improved survival among patients with adenocarcinoma and large-cell carcinoma, indicating its selective efficacy in non-squamous histology [ 5 ]. Furthermore, pemetrexed-containing regimens were associated with lower rates of hematologic and gastrointestinal toxicities than those containing gemcitabine [ 5 ]. These favorable efficacy and safety profiles establish pemetrexed as a key treatment option for non-squamous NSCLC [ 3 ]. Nevertheless, hematological toxicity, particularly neutropenia, is the most common adverse event associated with pemetrexed‒platinum chemotherapy [ 3 , 5 ]. Neutropenia is clinically significant because it markedly increases the risk of infection. These infections can reduce patients’ quality of life and, in severe cases, become life-threatening [ 6 ]. To prevent such infections, clinicians often delay or withhold subsequent chemotherapy cycles when neutropenia occurs. However, these treatment delays and prolonged intervals between chemotherapy cycles in first-line NSCLC treatment have been associated with poorer survival outcomes [ 7 ]. Therefore, early prediction of neutropenia is crucial for maintaining treatment continuity, which is essential for improving survival outcomes and preventing infections in patients with NSCLC. Platinum-based agents, such as cisplatin and carboplatin, require dose adjustments according to renal function [ 8 , 9 ]. Cisplatin dose adjustment based on renal function is well known to reduce the risk of myelosuppression and nephrotoxicity [ 10 ]. Similarly, carboplatin clearance is strongly correlated with glomerular filtration rate (GFR), and a linear relationship has been demonstrated between the area under the concentration-time curve (AUC) and carboplatin dose [ 8 , 9 , 11 ]. Based on these pharmacokinetic (PK) characteristics, Calvert et al. developed a formula to calculate the appropriate carboplatin dose from the patient’s measured GFR and target AUC [ 11 ]. Further, no standardized dose adjustment guidelines currently exist for pemetrexed, despite its primary renal excretion [ 8 , 9 ]. Whether renal function assessed by creatinine clearance (CLcr) or estimated GFR (eGFR) can predict pemetrexed-induced neutropenia remains controversial [ 12 – 14 ]. However, a recent study demonstrated that dose-adjusted pemetrexed can be safely used in patients with renal impairment, indicating that effective treatment may be feasible even in this traditionally excluded population [ 15 ]. Therefore, appropriate dose adjustment according to renal function is essential. Studies have reported a significant association between total systemic exposure to pemetrexed and an increased incidence of severe hematological adverse events [ 16 , 17 ]. Thus, considerable interindividual variability in the risk of pemetrexed-induced neutropenia may be influenced by patient-specific clinical and PK characteristics. However, blood pemetrexed levels are not routinely measured in clinical practice, which complicates the calculation of PK parameters. Several semi-mechanistic population PK/pharmacodynamic (PD) models have previously characterized pemetrexed-induced neutropenia in detail. For example, Latz et al. developed an integrated PK/PD model incorporating a physiological description of hematopoiesis within a nonlinear mixed-effects (NLME) framework [ 18 ]. More recently, Boosman et al. demonstrated that pemetrexed-related toxicity in patients with renal impairment is driven by a threshold exposure metric rather than cumulative AUC [ 19 ]. These semi-mechanistic approaches provide biologically grounded insights into exposure–toxicity relationships and allow simultaneous estimation of PK and PD parameters. Although these models were developed using well-controlled clinical trial datasets and provide valuable mechanistic insights, real-world clinical practice involves patients with heterogeneous backgrounds, comorbidities, and variable renal function, often without therapeutic drug monitoring. Such variability may result in substantial interindividual differences in neutrophil dynamics, underscoring the need for modeling strategies that remain applicable in routine clinical settings. In this study, we first investigated whether neutrophil dynamics could be predicted using routinely available clinical data from patients undergoing pemetrexed‒platinum chemotherapy, without relying on measured plasma concentration of pemetrexed. To achieve this, we employed physiologically based PK (PBPK) modeling, a mathematical approach that describes and predicts a drug’s behavior across various physiological tissues [ 20 ]. NLME modeling is a widely used pharmacometrics approach in PK and PK/PD analyses of longitudinal data [ 21 , 22 ]. It enables the evaluation of various structural models and covariate effects, uses all available observation points, and facilitates model validation through simulation. Using this framework, we aimed to identify patient-specific risk factors, including patients’ characteristics obtained from medical records as well as PK parameters of pemetrexed derived from a PBPK model, that influence changes in neutrophil counts during the first cycle of pemetrexed‒platinum chemotherapy in patients with lung cancer, by constructing a population model of neutrophil dynamics using an NLME modeling approach. The first cycle was selected for analysis because empirical dosing is generally applied at this stage, and patients have not yet been influenced by dose adjustments or supportive measures, allowing for an objective evaluation of baseline risk factors. Furthermore, we validated whether the developed model is generalizable to other patients receiving combination regimens that include ICIs or molecularly targeted agents. Materials and Methods Study population In this retrospective study, we analyzed data from two patient cohorts at Saiseikai Kumamoto Hospital from January 2015 to December 2023. The detailed inclusion and exclusion criteria for the modeling dataset were as follows: (1) received pemetrexed-platinum chemotherapy for at least one cycle; (2) had accurate records of treatment initiation and discontinuation; (3) underwent at least one hematological examination during the first treatment cycle; (4) had no prior exposure to pemetrexed or platinum agents; and (5) provided written informed consent. The development cohort comprised 86 patients with NSCLC (62 men and 24 women; mean age, 68.24 ± 7.85 years) who received pemetrexed‒platinum therapy, including 38 patients treated with cisplatin plus pemetrexed and 48 patients treated with carboplatin plus pemetrexed. To evaluate the applicability of the model to regimens combined with antibody drugs, such as pembrolizumab or bevacizumab, which are currently used as standard treatments, the validation cohort included 83 patients with NSCLC (51 men and 32 women, 65.52 ± 7.04 years), comprising 39 patients treated with carboplatin plus pemetrexed plus pembrolizumab, 20 with cisplatin plus pemetrexed plus bevacizumab, and 24 with carboplatin plus pemetrexed plus bevacizumab. In all regimens, chemotherapeutic agents and antibody drugs were administered on day 1 of the 21-day cycle. The standard doses of the chemotherapeutic agents were as follows: 500 mg/m 2 of pemetrexed, 75 mg/m 2 of cisplatin, and multiplying an AUC of 5 mg·min/mL by [25 + GFR (mL/min)] for carboplatin (mg); however, the actual doses of pemetrexed and platinum agents were determined empirically by physicians based on clinical judgment, taking into account patient-specific factors such as age, renal function, performance status, and comorbidities. For the validation cohort, pembrolizumab was administered at a fixed dose of 200 mg, and bevacizumab at 15 mg/kg, according to standard clinical protocols at the time. Neutrophil data from all measurement points in the first treatment cycle were used in the analysis, excluding any time points after the administration of granulocyte colony-stimulating factor. The final dataset comprised 316 and 375 data points from the development and validation cohorts, respectively, after excluding 6 data points (from 3 patients) and 5 data points (from 1 patient), respectively. The following information was collected from patient medical records: sex, age, height, weight, body mass index (BMI), creatinine, blood urea nitrogen (BUN), chemotherapy regimens, doses of anticancer agents, and other concomitant medications. Renal function was estimated using the Cockcroft‒Gault formula for CLcr and the Japanese equation for individualized eGFR [ 23 ]. The eGFR values were de-indexed from 1.73 m 2 to each patient’s body surface area (BSA), calculated using the Du Bois formula, to obtain absolute values (mL/min). The protocol was approved by the Ethics Committee of Saiseikai Kumamoto Hospital (Approval Nos. 1187 and 1319). The study was performed in accordance with the Declaration of Helsinki. Developing a PBPK model for pemetrexed A PBPK model for pemetrexed was developed using PK-Sim (version 11.3, Open Systems Pharmacology Suite). Physicochemical and PK parameters (including molecular weight, logP, pKa, plasma protein binding, and renal clearance) were obtained from the literature [ 24 ] and the PubChem database [ 25 ] (Supplementary Table 1). Model parameters not directly available were optimized using the Levenberg‒Marquardt algorithm implemented in PK-Sim. Pemetrexed is a hydrophilic antifolate primarily eliminated by renal clearance (~ 70–90%), with limited hepatic metabolism [ 24 ]. The renal excretion process was modeled to account for glomerular filtration and active secretion, reflecting the known involvement of transporters. Evaluation of the PBPK model for pemetrexed Model performance was evaluated by comparing predicted and observed values [ 26 – 28 ] for clearance (CL), maximum plasma concentration (C max ), and AUC from time zero to infinity (AUC 0–inf ). Predicted values for these PK parameters were obtained through simulations in which pemetrexed was administered to virtual populations generated in PK-Sim. In creating virtual populations, the number of individuals, race, proportion of females, and age (mean and range) were set to match those reported in the literature [ 26 – 28 ]. Model performance was assessed using the average fold error (AFE) between predicted and observed PK parameters. AFE was calculated using Eq. 1: AFE = Predicted value / Observed value (Eq. 1) A two-fold error range (AFE between 0.5 and 2) was considered acceptable for model evaluation [ 29 ]. First, the predictive performance of the PBPK model was evaluated without accounting for differences in varying renal function. Predicted PK parameters were obtained for eight pemetrexed doses (300, 500, 600, 700, 800, 900, 1,000, and 1,200 mg/m 2 ) for which observed values were reported in the literature [ 26 , 27 ]. These PK parameters were predicted based on simulations using healthy adult populations, as patients with impaired renal or hepatic function were excluded from the studies reporting the observed data [ 26 , 27 ]. To evaluate model performance, AFE was calculated for each PK parameter at each dose simulation, and the geometric mean of the AFEs was subsequently computed. Subsequently, model performance was assessed across populations with varying renal function by comparing predicted and observed [ 28 ] PK parameters for three groups: GFR ≥ 80, 60–79, and 40–59 mL/min. The mean GFR for each group was set to match previously reported values [ 28 ]. PK parameters for the GFR 40–59 mL/min group were predicted using simulations of a chronic kidney disease population, whereas those for the other groups were predicted using simulations of healthy adult populations. The AFE for the PK parameters was calculated for each group, and the predicted plasma concentration-time profiles (arithmetic mean and 5th–95th percentiles) were visually compared with the observed data [ 28 ] for each GFR group. Observed concentration‒time data [ 28 ] were digitized from the published figures using WebPlotDigitizer (version 4.8, Ankit Rohatgi, Pacifica, CA). Plasma concentration‒time profiles were illustrated using R (version 4.1.1, R Foundation for Statistical Computing, Vienna, Austria). Developing a population model for changes in neutrophil counts over time NONMEM (version 7.5.1, ICON Dev Soln, Ellicott City, MD) was used to model changes in neutrophil count during the first cycle of pemetrexed‒platinum therapy. The model was fitted to neutrophil data using the first-order conditional estimation with interaction method. Changes in neutrophil count during the first cycle were best described visually by a quadratic function. The base model comprised two quadratic functions: one for the decrease (Eq. 2) and the other for the recovery (Eq. 3) of the neutrophil count. y = A1 × (WEEK − B) 2 + Base − A1 × B 2 × exp(ε) if WEEK is lower than B (Eq. 2) y = A2 × (WEEK − B) 2 + Base − A1 × B 2 × exp(ε) if WEEK is greater than B (Eq. 3) WEEK is the elapsed time in weeks, calculated as the number of days since pemetrexed administration divided by seven. Base denotes neutrophil counts before pemetrexed administration (typically within 1 week). A1 and A2 represent the slopes of neutrophil decline and recovery, respectively. B is the WEEK when neutrophil recovery starts, and ε denotes intra-individual variabilities with mean zero and variance σ 2 . To identify factors influencing interindividual variability in neutrophil counts following pemetrexed and platinum administration, the effects of various covariates on parameters A1, A2, and B were assessed. Step 1 : A model was constructed using routinely available clinical variables in standard medical practice. The following covariates were evaluated: age, BMI, BSA, renal function parameters (CLcr, eGFR, and BUN), pemetrexed dose, type of platinum-based agent (cisplatin or carboplatin), and the use of concomitant medications, including non-steroidal anti-inflammatory drugs, renin‒angiotensin system (RAS) inhibitors, and diuretics. Covariate values were obtained at the initiation of chemotherapy during the first cycle. Step 2 In the second step, instead of the pemetrexed dose, PBPK model-derived PK parameters of pemetrexed (CL, C max , AUC 0–inf , and AUC from time zero to 24 h [AUC 0–24 ]) were also included as covariates to capture interindividual differences in drug exposure beyond dose. The covariate model was developed through univariate analysis in NONMEM, followed by a stepwise method using forward inclusion and backward elimination. The effect of covariates on parameters A1, A2, and B was assessed using the likelihood ratio test, wherein changes in the objective function value (OFV) computed by NONMEM were compared against critical values from the chi-squared (χ 2 ) distribution. In the forward step, covariates that exhibited significant effects in the univariate analysis were added to the model in descending order of their impact on OFV. The covariates were included in the model when the OFV decreased by > 3.84 (χ 2 , P < 0.05, one degree of freedom) and 5.99 (χ 2 , P < 0.05, two degrees of freedom), thereby developing the full model. In the backward step, models with only one covariate excluded from the full model were developed. The covariates were retained in the model when the OFV increased by > 3.84 (χ 2 , P < 0.05, one degree of freedom) and 5.99 (χ 2 , P < 0.05, two degrees of freedom), resulting in the final model. Model evaluation of the development cohort The model evaluation was performed using the same procedures previously reported [ 30 ]. The final model was internally validated using goodness-of-fit (GOF) plots and visual predictive checks (VPCs). A stratified nonparametric bootstrap analysis was also performed to investigate model robustness and parameter precision. The GOF plots used scatter plots of observed versus population-predicted values, observed versus individual-predicted values, and conditional weighted residuals versus population-predicted values and time after the pemetrexed dose. A random sampling of the original data generated 1,000 replicated datasets. In the VPC approach, the 5th, 50th, and 95th percentiles of simulations from the final model were compared with the observed values. The 95% confidence intervals (CIs) for the parameters after successful bootstrap convergence were compared with the final model estimates. In addition, we generated a receiver operating characteristic (ROC) curve for a decrease in neutrophil count to < 1,000/µL, and we confirmed prediction accuracy using the area under the ROC curve (AUROC). External validation of the final model To assess the predictive performance and generalizability of the final model, external validation was conducted using data from the validation cohort. PK parameters of pemetrexed—specifically CL, C max , AUC 0–inf , and AUC 0–24 —were estimated for each patient in the validation cohort using the PBPK model constructed in the development cohort. Subsequently, individual neutrophil counts during the first treatment cycle were predicted using Bayesian estimation based on the final model implemented in NONMEM. Model performance was assessed using GOF plots and the VPC approach, following the same procedure as in the model validation of the development cohort. Furthermore, an ROC curve for a decrease in neutrophil count to < 1,000/µL was generated, and its discriminative ability was assessed using the AUROC. Statistical analyses and graphics were generated using R. Results Patient demographics A comparison of the demographic and clinical characteristics of patients in the development (n = 86) and validation (n = 83) cohorts is summarized in Table 1 . Patients in the validation cohort received higher doses of pemetrexed, had a higher proportion of concomitant carboplatin use, were younger, and had higher BMI values than those in the development cohort. No significant differences were observed in other baseline characteristics. Table 1 Participant clinical characteristics Age (year) Development cohort (n = 86) Validation cohort (n = 83) P value 68.24 ± 7.85 65.52 ± 7.04 0.019 Sex (male/female) 62 (72.1) /24 (27.9) 51 (61.4) /32 (38.6) 0.191 Height (cm) 161.77 ± 9.10 161.88 ± 8.62 0.939 Weight (kg) 58.65 ± 11.01 62.10 ± 13.09 0.065 BMI (kg/ m 2 ) 22.34 ± 3.50 23.54 ± 3.79 0.034 BSA (m 2 ) 1.61 ± 0.18 1.65 ± 0.20 0.170 Creatinine (mg/dL) 0.76 ± 0.16 0.76 ± 0.18 0.985 CLcr (mL/min) 77.32 ± 24.38 82.93 ± 24.16 0.135 eGFR (mL/min) 71.09 ± 19.14 70.78 ± 16.16 0.908 BUN (mg/dL) 15.25 ± 5.16 14.44 ± 3.93 0.257 Pemetrexed dose (mg) 760.64 ± 127.05 802.64 ± 116.72 0.027 Pemetrexed dose per BSA (mg/m 2 ) 469.54 ± 51.35 484.27 ± 30.68 0.026 Co-administration Cisplatin/Carboplatin 38 (44.2) / 48 (55.8) 20 (24.1) / 63 (75.9) 0.009 Bevacizumab/Pembrolizumab – 44 (53.0) / 39 (47.0) – NSAIDs 23 (26.7) 12 (14.5) 0.058 RAS inhibitor 25 (29.1) 17 (20.5) 0.216 Diuretics 8 (9.3) 5 (6.0) 0.566 Data are presented as means ± standard deviations or proportions for categorical variables. BMI, body mass index; BSA, body surface area; BUN, blood urea nitrogen; CLcr, creatinine clearance; eGFR, estimated glomerular filtration rate; NSAIDs, non-steroidal anti-inflammatory drugs; RAS, renin‒angiotensin system. PK parameters estimation using the PBPK approach Supplementary Table 2 summarizes the predicted PK parameters, the corresponding observed values reported in the literature [ 26 , 27 ], and the AFE for each pemetrexed dose. The developed PBPK model predicted PK parameters within a two-fold error range across the wide range of pemetrexed dose levels. Moreover, the geometric mean (95% CI) of the AFE for CL, C max , and AUC 0–inf were 1.04 (0.86–1.26), 0.91 (0.80–1.05), and 1.00 (0.84–1.20), respectively, suggesting the favorable predictive performance. Model evaluation across three virtual populations with varying GFRs similarly showed PK predictions within a two-fold error range for each group (Supplementary Table 3). The developed PBPK model successfully captured the expected PK trend: as GFR declines, CL decreases, and AUC 0–inf increases, while C max generally remains relatively unchanged [ 28 ]. In addition, simulated 24-h plasma concentration‒time profiles of pemetrexed closely matched observed data [ 28 ] across groups (Supplementary Fig. 1), supporting the validity of the developed PBPK model. Subsequently, individual CL, C max , AUC 0–inf , and AUC 0–24 values were calculated for each participant using the developed PBPK model and patient-specific information, including race (Japanese), age, sex, height, weight, BMI, BSA, eGFR, and pemetrexed dose. Development of the final model for the change in neutrophil count Supplementary Table 4 summarizes the impact of tested covariates on the OFV in the neutrophil count variation model. In Step 1, the model was developed using routinely available information, without including PK parameters. The model parameters at this step were as follows: A1: 0.717 × (Dose/760) 1.27 × exp(η) (Eq. 4) A2: 0.0319 × exp(η) (Eq. 5) B: 1.32 × exp(η) (Eq. 6) where η represents interindividual variability with a mean of zero and variance ω 2 . Step 1 yielded a predictive model that may explain changes in neutrophil counts solely in terms of the pemetrexed dose. However, as we retrospectively obtained real-world clinical data, pemetrexed doses may have been adjusted based on individual patient characteristics, such as age and renal function. Furthermore, the model included only the pemetrexed dose, making it challenging to predict individualized dosing before starting chemotherapy. Therefore, we incorporated the PK parameters of pemetrexed derived from the PBPK model (Step 2) to capture interindividual variability in pharmacokinetics beyond empirically adjusted doses. This extended analysis identified BUN, concomitant use of RAS inhibitors, and pemetrexed AUC 0–24 as statistically significant covariates (Supplementary Table 5). The final model equations incorporating these covariates were as follows: A1: 0.334 × (BUN/15.2) −0.975 × exp(η) (Eq. 7) A2: 0.0443 × 0.0451 RAS inhibitor × exp(η) (Eq. 8) B: 1.39 × 1.44 AUC0–24>175 × exp(η) (Eq. 9) where RAS inhibitor = 1 if RAS inhibitor is administered concomitantly, otherwise 0; AUC 0–24 >175 = 1 if pemetrexed AUC 0–24 exceeds 175 µg·h/mL, otherwise 0; and η represents interindividual variability with a mean of zero and variance ω 2 . The effects of individual patient factors incorporated into the final model are illustrated in Fig. 1 The model suggested that patients with lower BUN levels experienced a more rapid decline in neutrophil counts (Fig. 1 A), those receiving concomitant RAS inhibitors showed delayed recovery (Fig. 1 B), and patients with a pemetrexed AUC 0–24 >175 µg·h/mL exhibited a delayed onset of neutrophil recovery (Fig. 1 C). Evaluation of the final model The GOF plots suggested a correlation between the neutrophil counts predicted by the final model and the actual measurements (Figs. 2 A and 2 B). The distribution of residuals showed no significant pattern across population-predicted values and time after pemetrexed administration (Figs. 2 C and 2 D). The VPCs showed that the final model described the observed neutrophil data without substantial misspecification (Fig. 3 ). Among 1,000 bootstrap runs, 622 exhibited successful minimization and were included in the bootstrap analysis. The 95% CIs for all parameters obtained using the bootstrap approach were generally consistent with the NONMEM estimates (Supplementary Table 6). The final model demonstrated high performance in predicting a decrease in neutrophil count to < 1,000/µL, with an AUROC of 0.92 (95% CI: 0.75–1.00). External validation Visual inspection of the GOF plots and VPCs confirmed that the final model accurately predicted changes in neutrophil counts among patients in the validation cohort, demonstrating performance comparable to that observed in the development cohort (Figs. 4 and 5 ). Notably, the model maintained a strong predictive ability for identifying patients with neutrophil counts < 1,000/µL, achieving an AUROC of 0.85 (95% CI: 0.57–1.00) in the validation cohort. Discussion In this study, we developed a predictive model for neutrophil dynamics following pemetrexed‒platinum chemotherapy using PBPK and NLME modeling approaches (Equations 7‒9). The model demonstrated high accuracy in predicting neutrophil counts < 1,000/µL (AUROC: 0.92, 95% CI: 0.75–1.00), enabling early identification of high-risk patients and supporting timely dose adjustment to reduce the risk of severe neutropenia. Furthermore, the model was externally validated in a separate cohort of patients receiving combination therapies that included ICIs or molecular-targeted agents, supporting the model’s broader applicability and potential clinical utility across diverse treatment settings beyond conventional chemotherapy alone. Traditionally, the risk factors associated with neutropenia following pemetrexed‒platinum therapy remain unclear, with dose adjustments often being made empirically. Our findings suggest the potential for data-driven, dose-individualized pemetrexed dosing based on quantitative predictions, even during the first treatment cycle, when neutrophil dynamics are not yet known, offering a path toward more personalized and safer chemotherapy regimens. In the model constructed using only routinely available clinical data, pemetrexed dose was the only significant covariate (Equations 4‒6). However, the pemetrexed dose showed a weak but significant correlation with age and renal function (Supplementary Fig. 2). For example, patients aged ≥ 75 years received significantly lower doses of pemetrexed than those aged < 75 years (477.44 ± 39.93 mg/m 2 vs. 441.69 ± 74.51 mg/m 2 , P = 0.007). These findings suggest that pemetrexed dosing was likely adjusted based on various patient characteristics, guided by clinical judgment and experience, as the data were collected retrospectively in a real-world setting. Therefore, the influence of different patient factors may have been indirectly reflected in the administered doses, which likely explains why only the dose was included in the initial model. In contrast to previous semi-mechanistic PK/PD models primarily developed using controlled clinical trial datasets [ 18 , 19 ], our approach was designed to accommodate the substantial variability encountered in real-world clinical practice. By directly modeling neutrophil dynamics using routinely collected clinical data and PBPK-derived PK parameters, our framework enhances applicability in settings where therapeutic drug monitoring is unavailable. Furthermore, we investigated the influence of individual PK differences, beyond empirical dose adjustments, on neutrophil dynamics following pemetrexed‒platinum chemotherapy. The final model demonstrated that a higher pemetrexed AUC 0–24 was associated with a delayed onset of neutrophil recovery (Eq. 9), consistent with the semi-mechanistic physiological PD model of pemetrexed proposed by Latz et al., which quantitatively attributed delayed neutrophil recovery to prolonged maturation time of precursor cells as they transition from the maturation compartment to peripheral blood [ 16 ]. In high-exposure settings, the degree of precursor cell depletion increases, leading to a marked delay in peripheral blood recovery [ 16 ]. It should be noted that previous work by Boosman et al. suggested that pemetrexed-related toxicity, particularly in patients with renal impairment, is more closely driven by a threshold exposure metric than by cumulative AUC [ 19 ]. Therefore, the association between AUC 0–24 and delayed neutrophil recovery observed in our study should be interpreted as a pragmatic exposure surrogate within our dataset rather than a definitive mechanistic driver of toxicity. We aimed to determine whether a parsimonious empirical exposure–response relationship could support clinically actionable risk prediction in settings where plasma concentration measurements are unavailable. Importantly, the exposure metrics in our model were estimated from patient characteristics without direct concentration measurements, supporting its applicability in real-world clinical settings. Moreover, prolonged dosing intervals, potentially resulting from delayed neutrophil recovery, can increase mortality in patients undergoing first-line treatment for NSCLC [ 7 ]. By leveraging PBPK and NLME modeling approaches, this study provides a novel framework for detailed evaluation of exposure‒toxicity relationships, suggesting the possibility of designing individualized dosing regimens that maximize efficacy and safety, even without the need for plasma pemetrexed concentration measurements. Collectively, these findings emphasize the importance of optimizing pemetrexed dosing from the first treatment cycle onward to minimize the risk of severe toxicities, maintain the intended treatment schedule, and ultimately improve therapeutic outcomes. In addition to drug exposure metrics, patient-specific factors such as BUN levels also influenced neutrophil dynamics (Eq. 7). This study demonstrated that the rate of neutrophil decline after pemetrexed administration was greater in patients with lower BUN levels. When participants were stratified by median BUN, the low-BUN group was younger, had better renal function, and had a higher rate of concomitant cisplatin use than the high-BUN group (Supplementary Table 7). Notably, no significant between-group differences in pemetrexed or platinum doses were observed, indicating that the observed decline in neutrophils was not attributable to differences in chemotherapy intensity. Given that cisplatin is generally associated with a lower frequency of severe neutropenia than carboplatin [ 31 ], the greater neutrophil decline observed in the low BUN group is unlikely to be explained solely by differences in chemotherapy regimens. Notably, younger age and preserved renal function are typically considered protective against severe myelosuppression [ 13 , 14 , 32 ]. Therefore, this finding appears counterintuitive and suggests the involvement of other underlying factors. BUN may have been incorporated into the model not solely as a marker of renal clearance, but also as a surrogate for broader physiological conditions—such as nutritional status or metabolic reserve—that influence hematologic sensitivity [ 32 , 33 ]. Thus, the observed association between low BUN and greater neutrophil decline may reflect a complex interplay of the aforementioned patient-specific factors. Nevertheless, further studies are warranted to confirm these findings and elucidate the underlying mechanisms. Further, this study suggests that neutrophil recovery after pemetrexed administration is slower in patients taking RAS inhibitors (Eq. 8). Angiotensin II reportedly regulates hematopoiesis in vivo at the level of stem cells [ 34 ]. Moreover, angiotensin II analogs reportedly promote hematopoiesis and facilitate the recovery of circulating cells [ 35 ]. Therefore, RAS inhibitors may disrupt the hematopoietic regulatory function of angiotensin II, leading to delayed neutrophil recovery following pemetrexed administration. In addition, some RAS inhibitors, such as olmesartan and valsartan, have been reported to inhibit the organic anion transporter 3 (OAT3) [ 36 , 37 ], which is crucial for pemetrexed excretion. Concomitant use of OAT3 inhibitors increases pemetrexed AUC, suggesting that RAS inhibitors could impair pemetrexed elimination and potentially increase drug exposure and toxicity risk [ 38 , 39 ]. However, the detailed mechanisms underlying this interaction remain unclear and warrant further investigation. Recent advances in NSCLC treatment have demonstrated that the addition of antibody drugs, such as bevacizumab or pembrolizumab, to pemetrexed‒platinum regimens can enhance therapeutic efficacy [ 3 , 40 , 41 ]. Although the addition of antibody drugs can cause a range of side effects, their impact on hematologic toxicity is generally minimal [ 3 , 41 ]. Given the increasing clinical use of these combination therapies, we performed external validation of the developed model using a separate cohort of patients receiving pemetrexed‒platinum chemotherapy in combination with these antibody drugs. The final model demonstrated the ability to predict neutrophil count fluctuations after pemetrexed‒platinum administration, even in the presence of concomitant antibody drug use. Therefore, the results of this study may be applicable across a wide range of pemetrexed‒platinum regimens. This study has some limitations. First, as blood samples for the PK analysis were not collected from our study participants, the predictive performance of the PBPK model could not be evaluated. Therefore, future studies incorporating clinical PK data will be necessary to verify its accuracy. In addition, the developed PBPK model focused solely on pemetrexed and did not consider the effects of concomitant medications on its pharmacokinetics. In particular, the potential impact of drugs that interact with OAT3, which plays a significant role in pemetrexed renal elimination [ 38 ], was not accounted for in the pemetrexed PBPK model. Accordingly, future enhancements to the PBPK model should incorporate the effects of concomitant medications, particularly OAT3 inhibitors, to better predict pemetrexed PK. Moreover, the model did not explicitly include the type of platinum agent used (cisplatin or carboplatin). As previously mentioned, the choice and dosage of platinum agents may have been adjusted based on patient-specific factors, such as renal function. In addition, this study did not consider the pharmacokinetics of platinum agents. To enhance the model's accuracy and applicability, future studies should consider these factors. Furthermore, this study did not assess the relationship between pemetrexed exposure and antitumor efficacy. Although the model focused on hematologic toxicity, evaluating exposure‒efficacy relationships is also crucial for dosage optimization. Future studies should investigate whether pemetrexed exposure is correlated with clinical outcomes, such as response or survival. Furthermore, although this observational study was designed to reflect real-world clinical conditions, its findings may have limited generalizability because of the relatively small sample size, the single-institution setting, and the exclusive inclusion of Japanese patients. These factors may restrict the applicability of the results to broader, more diverse populations. Many patient-specific factors may influence the pharmacokinetics and pharmacodynamics of pemetrexed. Therefore, it is essential to confirm that the study findings are not limited to the study population through retrospective analyses in other patient groups with comparable backgrounds. Subsequently, prospective validation in larger, more diverse populations will be necessary to assess the generalizability of our findings. Conclusion This study demonstrated the potential to predict neutrophil dynamics after pemetrexed‒platinum chemotherapy by integrating exposure metrics estimated from patient characteristics, without requiring measured plasma concentrations of pemetrexed or patient-specific factors such as BUN levels or concomitant RAS inhibitor use. Our findings suggest that data-driven, individualized pemetrexed dosing may be feasible, starting with the first treatment cycle, by accounting for individual patient characteristics. Although further validation with actual PK data and consideration of concomitant drug effects are needed, these results highlight the promise of model-informed strategies to optimize pemetrexed dosing and improve therapeutic outcomes. Declarations Acknowledgments The authors thank all of the study participants. We would like to thank Editage (www.editage.jp) for English language editing. Author contributions KM, KS, HN, TK, AKM, KO, and JS contributed to the conceptualization and study design. KM, KT, AS, HT, and KI contributed to data curation and investigation. KM, KS, HN, TK, NS, and JS performed formal analysis and interpreted the data. KM, KS, HN, AKM, KO, and JS drafted the manuscript. TK, RT, HJ, HI, DK, KT, AS, HT, KI, AKM, KO, and JS critically reviewed and revised the manuscript for important intellectual content. All authors approved the final version of the manuscript. Funding This work was supported by JST SPRING (No. JPMJSP2127). None of the funders played a role in the design, implementation, analysis, and interpretation of the data. Conflict of Interest Authors do not report any conflict of interest for this work. Ethics approval The protocol was approved by the Ethics Committee of Saiseikai Kumamoto Hospital (Approval Nos. 1187 and 1319). The study was performed in accordance with the Declaration of Helsinki. References Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-263.10.3322/caac.21834 Ariyasu R, Kakuto S, Miyadera K, Akita T, Kiritani A, Tsugitomi R, et al. Real-World outcome analysis of patients with stage IV NSCLC treated with tyrosine kinase and immune checkpoint inhibitors. JTO Clin Res Rep. 2023;4(6):100524.10.1016/j.jtocrr.2023.100524 Gandhi L, Rodriguez-Abreu D, Gadgeel S, Esteban E, Felip E, De Angelis F, et al. Pembrolizumab plus chemotherapy in metastatic non-small-cell lung cancer. N Engl J Med. 2018;378(22):2078-2092.10.1056/NEJMoa1801005 Meyer ML, Peters S, Mok TS, Lam S, Yang PC, Aggarwal C, et al. Lung cancer research and treatment: global perspectives and strategic calls to action. Ann Oncol. 2024;35(12):1088-1104.10.1016/j.annonc.2024.10.006 Scagliotti GV, Parikh P, von Pawel J, Biesma B, Vansteenkiste J, Manegold C, et al. Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advanced-stage non-small-cell lung cancer. J Clin Oncol. 2023;41(14):2458-2466.10.1200/JCO.22.02544 Wang XJ, Wong M, Hsu LY, Chan A. Costs associated with febrile neutropenia in solid tumor and lymphoma patients - an observational study in Singapore. BMC Health Serv Res. 2014;14:434.10.1186/1472-6963-14-434 Crawford J, Denduluri N, Patt D, Jiao X, Morrow PK, Garcia J, et al. Relative dose intensity of first-line chemotherapy and overall survival in patients with advanced non-small-cell lung cancer. Support Care Cancer. 2020;28(2):925-932.10.1007/s00520-019-04875-1 Kidney Disease: Improving Global Outcomes CKDWG. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int. 2024;105(4S):S117-S314.10.1016/j.kint.2023.10.018 Yanagita M, Muto S, Nishiyama H, Ando Y, Hirata S, Doi K, et al. Clinical questions and good practice statements of clinical practice guidelines for management of kidney injury during anticancer drug therapy 2022. Clin Exp Nephrol. 2024;28(2):85-122.10.1007/s10157-023-02415-0 Zhang J, Ye ZW, Tew KD, Townsend DM. Cisplatin chemotherapy and renal function. Adv Cancer Res. 2021;152:305-327.10.1016/bs.acr.2021.03.008 Calvert AH, Newell DR, Gumbrell LA, O'Reilly S, Burnell M, Boxall FE, et al. Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J Clin Oncol. 1989;7(11):1748-1756.10.1200/JCO.1989.7.11.1748 de Rouw N, Beunders R, Hartmann O, Schulte J, Boosman RJ, Derijks HJ, et al. A comparison of the renal function biomarkers serum creatinine, pro-enkephalin and cystatin C to predict clearance of pemetrexed. Cancer Chemother Pharmacol. 2024;94(6):799-806.10.1007/s00280-024-04717-w Saito Y, Taniguchi O, Takekuma Y, Sakakibara-Konishi J, Shimizu Y, Kinoshita I, et al. Impact of baseline renal impairment on severe neutropenia development in pemetrexed and carboplatin thoracic cancer treatment. Support Care Cancer. 2024;32(12):829.10.1007/s00520-024-09006-z Shima Y, Yoshida H, Suminaga K, Yoshida H, Hashimoto K, Ogimoto T, et al. Safety and efficacy of pemetrexed for patients with non-small cell lung cancer in the setting of renal impairment: a retrospective study. BMC Cancer. 2025;25(1):388.10.1186/s12885-025-13785-x de Rouw N, Otten LS, Kicken MP, Piet B, Biesma B, van Veggel B, et al. Folinic acid prophylaxis and dose adjustments enable safe treatment with pemetrexed in patients with renal impairment. Clin Pharmacol Ther. 2025;118(3):715-722.10.1002/cpt.3735 Latz JE, Schneck KL, Nakagawa K, Miller MA, Takimoto CH. Population pharmacokinetic/pharmacodynamic analyses of pemetrexed and neutropenia: effect of vitamin supplementation and differences between Japanese and Western patients. Clin Cancer Res. 2009;15(1):346-354.10.1158/1078-0432.CCR-08-0791 Visser S, Koolen SLW, de Bruijn P, Belderbos HNA, Cornelissen R, Mathijssen RHJ, et al. Pemetrexed exposure predicts toxicity in advanced non-small-cell lung cancer: A prospective cohort study. Eur J Cancer. 2019;121:64-73.10.1016/j.ejca.2019.08.012 Latz JE, Karlsson MO, Rusthoven JJ, Ghosh A, Johnson RD. A semimechanistic-physiologic population pharmacokinetic/pharmacodynamic model for neutropenia following pemetrexed therapy. Cancer Chemother Pharmacol. 2006;57(4):412-426.10.1007/s00280-005-0077-5 Boosman RJ, Dorlo TPC, de Rouw N, Burgers JA, Dingemans AC, van den Heuvel MM, et al. Toxicity of pemetrexed during renal impairment explained-Implications for safe treatment. Int J Cancer. 2021;149(8):1576-1584.10.1002/ijc.33721 Kuepfer L, Niederalt C, Wendl T, Schlender JF, Willmann S, Lippert J, et al. Applied concepts in PBPK modeling: How to build a PBPK/PD model. CPT Pharmacometrics Syst Pharmacol. 2016;5(10):516-531.10.1002/psp4.12134 Jian W, Xue J, Yao Q, Chen R, Yao Y, Wang M, et al. Starting dose selection of palbociclib in Chinese patients with breast cancer based on population kinetic-pharmacodynamic model of neutropenia. Cancer Chemother Pharmacol. 2022;90(6):489-497.10.1007/s00280-022-04484-6 Tang F, Ng CM, Horn J, Bada HS, Leggas M. Pharmacokinetic modeling and model-based hypothesis generation for dose optimization of clonidine in neonates with neonatal opioid withdrawal syndrome. Clin Pharmacol Ther. 2025;117(5):1254-1263.10.1002/cpt.3507 Matsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009;53(6):982-992.10.1053/j.ajkd.2008.12.034 Posada MM, Bacon JA, Schneck KB, Tirona RG, Kim RB, Higgins JW, et al. Prediction of renal transporter mediated drug-drug interactions for pemetrexed using physiologically based pharmacokinetic modeling. Drug Metab Dispos. 2015;43(3):325-334.10.1124/dmd.114.059618 National Center for Biotechnology Information. PubChem database. 2025 18 September. Available from: https://pubchem.ncbi.nlm.nih.gov/compound/Pemetrexed. Nakagawa K, Kudoh S, Matsui K, Negoro S, Yamamoto N, Latz JE, et al. A phase I study of pemetrexed (LY231514) supplemented with folate and vitamin B12 in Japanese patients with solid tumours. Br J Cancer. 2006;95(6):677-682.10.1038/sj.bjc.6603321 Kavathiya K, Gurjar M, Patil A, Naik M, Noronha V, Joshi A, et al. A comparative pharmacokinetic study of 2 pemetrexed formulations in Indian adult chemonaive patients with adenocarcinoma stage III/IV non-small cell lung cancer. Clin Pharmacol Drug Dev. 2017;6(3):234-239.10.1002/cpdd.347 Mita AC, Sweeney CJ, Baker SD, Goetz A, Hammond LA, Patnaik A, et al. Phase I and pharmacokinetic study of pemetrexed administered every 3 weeks to advanced cancer patients with normal and impaired renal function. J Clin Oncol. 2006;24(4):552-562.10.1200/JCO.2004.00.9720 Rostami-Hodjegan A. Reverse Translation in PBPK and QSP: Going Backwards in Order to Go Forward With Confidence. Clin Pharmacol Ther. 2018;103(2):224-232.10.1002/cpt.904 Shigetome K, Egashira T, Tomita T, Higa N, Iwashita K, Morita K, et al. Effect of cumulative exposure on the efficacy of paroxetine: A population pharmacokinetic-pharmacodynamic and machine learning analyses. CPT Pharmacometrics Syst Pharmacol. 2025;14(6):1119-1127.10.1002/psp4.70032 Griesinger F, Korol EE, Kayaniyil S, Varol N, Ebner T, Goring SM. Efficacy and safety of first-line carboplatin-versus cisplatin-based chemotherapy for non-small cell lung cancer: A meta-analysis. Lung Cancer. 2019;135:196-204.10.1016/j.lungcan.2019.07.010 Kulkarni PM, Chen R, Anand T, Monberg MJ, Obasaju CK. Efficacy and safety of pemetrexed in elderly cancer patients: results of an integrated analysis. Crit Rev Oncol Hematol. 2008;67(1):64-70.10.1016/j.critrevonc.2008.01.011 Okada S, Yamazaki S, Kaiga T, Funada T, Kochi M, Takayama T. Impact of nutritional status in the era of FOLFOX/FIRI-based chemotherapy. World J Surg Oncol. 2017;15(1):162.10.1186/s12957-017-1226-0 Kim S, Zingler M, Harrison JK, Scott EW, Cogle CR, Luo D, et al. Angiotensin II regulation of proliferation, differentiation, and engraftment of hematopoietic stem cells. Hypertension. 2016;67(3):574-584.10.1161/HYPERTENSIONAHA.115.06474 Gaffney K, Weinberg M, Soto M, Louie S, Rodgers K. Development of angiotensin II (1-7) analog as an oral therapeutic for the treatment of chemotherapy-induced myelosuppression. Haematologica. 2018;103(12):e567-e570.10.3324/haematol.2018.193771 Sato M, Iwanaga T, Mamada H, Ogihara T, Yabuuchi H, Maeda T, et al. Involvement of uric acid transporters in alteration of serum uric acid level by angiotensin II receptor blockers. Pharm Res. 2008;25(3):639-646.10.1007/s11095-007-9401-6 Watanabe T, Kusuhara H, Watanabe T, Debori Y, Maeda K, Kondo T, et al. Prediction of the overall renal tubular secretion and hepatic clearance of anionic drugs and a renal drug-drug interaction involving organic anion transporter 3 in humans by in vitro uptake experiments. Drug Metab Dispos. 2011;39(6):1031-1038.10.1124/dmd.110.036129 Kurata T, Iwamoto T, Kawahara Y, Okuda M. Characteristics of pemetrexed transport by renal basolateral organic anion transporter hOAT3. Drug Metab Pharmacokinet. 2014;29(2):148-153.10.2133/dmpk.dmpk-13-rg-042 Sweeney CJ, Takimoto CH, Latz JE, Baker SD, Murry DJ, Krull JH, et al. Two drug interaction studies evaluating the pharmacokinetics and toxicity of pemetrexed when coadministered with aspirin or Ibuprofen in patients with advanced cancer. Clin Cancer Res. 2006;12(2):536-542.10.1158/1078-0432.CCR-05-1834 Garassino MC, Gadgeel S, Speranza G, Felip E, Esteban E, Domine M, et al. Pembrolizumab plus pemetrexed and platinum in nonsquamous non-small-cell lung cancer: 5-year outcomes from the phase 3 KEYNOTE-189 study. J Clin Oncol. 2023;41(11):1992-1998.10.1200/JCO.22.01989 Patel JD, Socinski MA, Garon EB, Reynolds CH, Spigel DR, Olsen MR, et al. PointBreak: a randomized phase III study of pemetrexed plus carboplatin and bevacizumab followed by maintenance pemetrexed and bevacizumab versus paclitaxel plus carboplatin and bevacizumab followed by maintenance bevacizumab in patients with stage IIIB or IV nonsquamous non-small-cell lung cancer. J Clin Oncol. 2013;31(34):4349-4357.10.1200/JCO.2012.47.9626 Additional Declarations No competing interests reported. Supplementary Files Moritaetal.SupplementalMaterial.docx Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 20 Apr, 2026 Reviews received at journal 19 Apr, 2026 Reviews received at journal 30 Mar, 2026 Reviewers agreed at journal 18 Mar, 2026 Reviewers agreed at journal 03 Mar, 2026 Reviewers invited by journal 03 Mar, 2026 Editor assigned by journal 28 Feb, 2026 Submission checks completed at journal 28 Feb, 2026 First submitted to journal 27 Feb, 2026 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8989784","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":599807035,"identity":"f3c08ea0-cf9d-42b8-8b49-31697e56e3fd","order_by":0,"name":"Kazunori Morita","email":"","orcid":"","institution":"Saiseikai Kumamoto Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kazunori","middleName":"","lastName":"Morita","suffix":""},{"id":599807036,"identity":"b73a26d5-7d0f-4845-80a0-195ce2f22620","order_by":1,"name":"Keiichi Shigetome","email":"","orcid":"","institution":"Kumamoto University","correspondingAuthor":false,"prefix":"","firstName":"Keiichi","middleName":"","lastName":"Shigetome","suffix":""},{"id":599807038,"identity":"8cc490ac-6567-43aa-8566-be8232b806f5","order_by":2,"name":"Haruka Narise","email":"","orcid":"","institution":"Kumamoto University","correspondingAuthor":false,"prefix":"","firstName":"Haruka","middleName":"","lastName":"Narise","suffix":""},{"id":599807040,"identity":"d400a828-1ae4-4ee8-bc90-2af4742ad8b6","order_by":3,"name":"Tetsuya Kaneko","email":"","orcid":"","institution":"Kumamoto University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Tetsuya","middleName":"","lastName":"Kaneko","suffix":""},{"id":599807041,"identity":"ffcb52e1-306a-4454-aad2-1a18a71b01ee","order_by":4,"name":"Naoto Soejima","email":"","orcid":"","institution":"Kumamoto University","correspondingAuthor":false,"prefix":"","firstName":"Naoto","middleName":"","lastName":"Soejima","suffix":""},{"id":599807042,"identity":"0f67c6d7-462f-42ac-8708-a10649271c11","order_by":5,"name":"Ryota Tanaka","email":"","orcid":"","institution":"Oita University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Ryota","middleName":"","lastName":"Tanaka","suffix":""},{"id":599807043,"identity":"788b7b04-98cf-40f0-a693-64fdc5d53bff","order_by":6,"name":"Hirofumi Jono","email":"","orcid":"","institution":"Kumamoto University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Hirofumi","middleName":"","lastName":"Jono","suffix":""},{"id":599807047,"identity":"608d612d-bcc1-449e-a2f8-a18b47ad3880","order_by":7,"name":"Hiroki Itoh","email":"","orcid":"","institution":"Oita University Hospital","correspondingAuthor":false,"prefix":"","firstName":"Hiroki","middleName":"","lastName":"Itoh","suffix":""},{"id":599807048,"identity":"2433cb89-277a-41ab-b706-b1a210a98e80","order_by":8,"name":"Daisuke Kadowaki","email":"","orcid":"","institution":"Sojo University","correspondingAuthor":false,"prefix":"","firstName":"Daisuke","middleName":"","lastName":"Kadowaki","suffix":""},{"id":599807053,"identity":"49a6355e-e238-4f32-89a9-d69274411708","order_by":9,"name":"Koki Tokunaga","email":"","orcid":"","institution":"Saiseikai Kumamoto Hospital","correspondingAuthor":false,"prefix":"","firstName":"Koki","middleName":"","lastName":"Tokunaga","suffix":""},{"id":599807054,"identity":"1b2e217b-4678-4f2c-a58e-822b58dfd9cd","order_by":10,"name":"Akitomo Shibata","email":"","orcid":"","institution":"Saiseikai Kumamoto Hospital","correspondingAuthor":false,"prefix":"","firstName":"Akitomo","middleName":"","lastName":"Shibata","suffix":""},{"id":599807055,"identity":"4c87db41-2b43-4a3a-adb5-c13afd089e5d","order_by":11,"name":"Harumi Tanoue","email":"","orcid":"","institution":"Saiseikai Kumamoto Hospital","correspondingAuthor":false,"prefix":"","firstName":"Harumi","middleName":"","lastName":"Tanoue","suffix":""},{"id":599807056,"identity":"675e6f5e-e8e6-45e7-9dc4-5bd2cbf49028","order_by":12,"name":"Kazuya Ichikado","email":"","orcid":"","institution":"Saiseikai Kumamoto Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kazuya","middleName":"","lastName":"Ichikado","suffix":""},{"id":599807057,"identity":"f54c9949-0667-4414-94ac-fa60481f1009","order_by":13,"name":"Ayami Kajiwara-Morita","email":"","orcid":"","institution":"Kumamoto University","correspondingAuthor":false,"prefix":"","firstName":"Ayami","middleName":"","lastName":"Kajiwara-Morita","suffix":""},{"id":599807058,"identity":"76ea6cdb-0c2b-47bb-b7d6-60a93e1d7403","order_by":14,"name":"Kentaro Oniki","email":"","orcid":"","institution":"Kumamoto University","correspondingAuthor":false,"prefix":"","firstName":"Kentaro","middleName":"","lastName":"Oniki","suffix":""},{"id":599807059,"identity":"02c742f5-31ef-4f73-a34d-3816134f0dc0","order_by":15,"name":"Junji Saruwatari","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABG0lEQVRIiWNgGAWjYBACxgYILYfMgYMEfFqMkTjM+LXAQGIDNi1YAfOM3MMffu45nD4/uvkA48y2bfK67f0HHzC2McjzNzA8e4DNYTPy0iR7nh3O3XjnWALjxrbbhtvOHGY2AGoxnHGAId0Aq5YcMwaeA0AtM3IMGB+23WbcdiOZTfpvGwPjBgaGNAnsWow//jlwON0QqsUepEUCaIs9Hi0G0kBbEuQlgFqADkuEaUnEqaXnjZm0zIF0ww0yxxIOzjh3OxnoF2MDhnMSyTMOY/eLYTvQYW8OWMvLz24++LCn7LbttuONDx8wlNnY9rf3pGELMcMGKMPgBgPDASQJoJOYedKw6GCQhzNmYEqyH8OmZRSMglEwCkYcAAD7dmotUiPXPgAAAABJRU5ErkJggg==","orcid":"","institution":"Kumamoto University","correspondingAuthor":true,"prefix":"","firstName":"Junji","middleName":"","lastName":"Saruwatari","suffix":""}],"badges":[],"createdAt":"2026-02-27 15:39:49","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8989784/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8989784/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104179490,"identity":"5932ee27-4dce-4cea-95bb-09995099d5c9","added_by":"auto","created_at":"2026-03-08 17:05:34","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":150808,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRelationship between the time after administration and neutrophil count based on the prediction model for neutropenia\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a)\u003cstrong\u003e \u003c/strong\u003eRelationships between the time after administration and neutrophil count are represented as a solid curve for patients with a low BUN. They are presented as a dotted curve for average patients. (b) Relationships between the time after administration and neutrophil count are represented by a solid curve for patients taking a RAS inhibitor and are presented as a dotted curve for the average patient. (c) Relationships between the time after administration and neutrophil count are represented by a solid curve for patients with a high value of AUC\u003csub\u003e0–24\u003c/sub\u003e. They are presented as a dotted curve for average patients.\u003c/p\u003e\n\u003cp\u003eAUC\u003csub\u003e0–24\u003c/sub\u003e, area under the plasma concentration‒time curve from time zero to 24 h; BUN, blood urea nitrogen; and RAS, renin‒angiotensin system.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/276f95e019f3ce6c535825f3.png"},{"id":104179495,"identity":"3c69f42b-cdb9-48d8-82cd-9ccb0bd0acf5","added_by":"auto","created_at":"2026-03-08 17:05:34","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":85618,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGoodness-of-fit plots for the final model in the development cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Observed vs. population-predicted neutrophil count. (b) Observed vs. individual-predicted neutrophil count. (c) Conditional weighted residualvs. population-predicted neutrophil count. (d) Conditional weighted residual vs. time after administration. The solid lines in (a) and (b) represent the identity lines. The solid lines in (c) and (d) represent the zero line. The dashed lines represent smoothing lines.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/b506d81dcb3c5a549d9a33e0.png"},{"id":104179502,"identity":"d9c634b2-9f6f-4d1e-a369-22906ba3f844","added_by":"auto","created_at":"2026-03-08 17:05:34","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":192074,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVisual predictive check for the final model in the development cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBlue circles indicate observed neutrophil counts. The solid red line represents the median of the observed neutrophil count. The solid black line represents the median of the predicted neutrophil count in the simulated data. The dashed red lines are the 5th and 95th percentiles of the observed neutrophil count. The dashed black lines are the 5th and 95th percentiles of the predicted neutrophil count in the simulated data. The red area represents the 95% confidence interval for the median, while the blue area represents the 95% confidence interval for the 5th and 95th percentiles in the simulated data.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/53912e488fa9d4d85b0b1ef8.png"},{"id":104179489,"identity":"cfb8b56b-2e96-4bbb-9cc3-ea99d38e3cc8","added_by":"auto","created_at":"2026-03-08 17:05:33","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":229074,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGoodness-of-fit plots for the final model in the validation cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(a) Observed vs. population-predicted neutrophil count. (b) Observed vs. individual-predicted neutrophil count. (c) Conditional weighted residualvs. population-predicted neutrophil count. (d) Conditional weighted residual vs. time after administration. The solid lines in (a) and (b) represent the identity lines. The solid lines in (c) and (d) represent the zero line. The dashed lines represent smoothing lines.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/ff2cc53c4be2afdd7f2d3c69.png"},{"id":104179503,"identity":"ac598bd7-b5fd-4ab6-8f2f-565afd2487d3","added_by":"auto","created_at":"2026-03-08 17:05:34","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":184188,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eVisual predictive check for the final model in the validation cohort\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBlue circles plot the observed neutrophil counts. The solid red line represents the median of the observed neutrophil count. The solid black line represents the median of the predicted neutrophil count in the simulated data. The dashed red lines are the 5th and 95th percentiles of the observed neutrophil count. The dashed black lines are the 5th and 95th percentiles of the predicted neutrophil count in the simulated data. The red area represents the 95% confidence interval for the median, while the blue area represents the 95% confidence interval for the 5th and 95th percentiles in the simulated data.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/5b49f1dbaff3318bea1c8e71.png"},{"id":104409321,"identity":"eb5fd538-36e7-441f-aa28-121164c71b15","added_by":"auto","created_at":"2026-03-11 12:44:44","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1787443,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/09756629-efec-44f2-8fbf-412c27020c07.pdf"},{"id":104404190,"identity":"5d403353-9148-44bf-84cf-ef57400ea275","added_by":"auto","created_at":"2026-03-11 12:19:47","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":970919,"visible":true,"origin":"","legend":"","description":"","filename":"Moritaetal.SupplementalMaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-8989784/v1/cd58d07d949d4493d1ac64cd.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Prediction of Neutropenia Risk Based on Exposure Information Without Plasma Concentration Measurement in Pemetrexed‒Platinum‒Based Chemotherapy: A Modeling Approach Using Real-World Clinical Data","fulltext":[{"header":"Introduction","content":"\u003cp\u003eLung cancer was the most frequently diagnosed malignancy worldwide in 2022, with approximately 2.5\u0026nbsp;million new cases, representing 12.4% of all cancer diagnoses [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. It also remains the leading cause of cancer-related death, with an estimated 1.8\u0026nbsp;million deaths, accounting for 18.7% of global cancer mortality [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Recent advances in diagnostic and therapeutic approaches have improved survival outcomes for patients with lung cancer [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. For advanced non-small cell lung cancer (NSCLC) lacking driver gene mutations (approximately 50% of cases), the standard first-line treatment is either an immune checkpoint inhibitor (ICI) monotherapy or a combination of ICI and platinum-based chemotherapy [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. In a phase III study, pemetrexed plus cisplatin demonstrated noninferiority in overall survival (OS) compared with gemcitabine plus cisplatin in untreated advanced NSCLC [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Subgroup analyses further showed that pemetrexed significantly improved survival among patients with adenocarcinoma and large-cell carcinoma, indicating its selective efficacy in non-squamous histology [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Furthermore, pemetrexed-containing regimens were associated with lower rates of hematologic and gastrointestinal toxicities than those containing gemcitabine [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. These favorable efficacy and safety profiles establish pemetrexed as a key treatment option for non-squamous NSCLC [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eNevertheless, hematological toxicity, particularly neutropenia, is the most common adverse event associated with pemetrexed‒platinum chemotherapy [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. Neutropenia is clinically significant because it markedly increases the risk of infection. These infections can reduce patients\u0026rsquo; quality of life and, in severe cases, become life-threatening [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. To prevent such infections, clinicians often delay or withhold subsequent chemotherapy cycles when neutropenia occurs. However, these treatment delays and prolonged intervals between chemotherapy cycles in first-line NSCLC treatment have been associated with poorer survival outcomes [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Therefore, early prediction of neutropenia is crucial for maintaining treatment continuity, which is essential for improving survival outcomes and preventing infections in patients with NSCLC.\u003c/p\u003e \u003cp\u003ePlatinum-based agents, such as cisplatin and carboplatin, require dose adjustments according to renal function [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Cisplatin dose adjustment based on renal function is well known to reduce the risk of myelosuppression and nephrotoxicity [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Similarly, carboplatin clearance is strongly correlated with glomerular filtration rate (GFR), and a linear relationship has been demonstrated between the area under the concentration-time curve (AUC) and carboplatin dose [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e, \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Based on these pharmacokinetic (PK) characteristics, Calvert et al. developed a formula to calculate the appropriate carboplatin dose from the patient\u0026rsquo;s measured GFR and target AUC [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. Further, no standardized dose adjustment guidelines currently exist for pemetrexed, despite its primary renal excretion [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Whether renal function assessed by creatinine clearance (CLcr) or estimated GFR (eGFR) can predict pemetrexed-induced neutropenia remains controversial [\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. However, a recent study demonstrated that dose-adjusted pemetrexed can be safely used in patients with renal impairment, indicating that effective treatment may be feasible even in this traditionally excluded population [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. Therefore, appropriate dose adjustment according to renal function is essential. Studies have reported a significant association between total systemic exposure to pemetrexed and an increased incidence of severe hematological adverse events [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Thus, considerable interindividual variability in the risk of pemetrexed-induced neutropenia may be influenced by patient-specific clinical and PK characteristics. However, blood pemetrexed levels are not routinely measured in clinical practice, which complicates the calculation of PK parameters.\u003c/p\u003e \u003cp\u003eSeveral semi-mechanistic population PK/pharmacodynamic (PD) models have previously characterized pemetrexed-induced neutropenia in detail. For example, Latz et al. developed an integrated PK/PD model incorporating a physiological description of hematopoiesis within a nonlinear mixed-effects (NLME) framework [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. More recently, Boosman et al. demonstrated that pemetrexed-related toxicity in patients with renal impairment is driven by a threshold exposure metric rather than cumulative AUC [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. These semi-mechanistic approaches provide biologically grounded insights into exposure\u0026ndash;toxicity relationships and allow simultaneous estimation of PK and PD parameters. Although these models were developed using well-controlled clinical trial datasets and provide valuable mechanistic insights, real-world clinical practice involves patients with heterogeneous backgrounds, comorbidities, and variable renal function, often without therapeutic drug monitoring. Such variability may result in substantial interindividual differences in neutrophil dynamics, underscoring the need for modeling strategies that remain applicable in routine clinical settings.\u003c/p\u003e \u003cp\u003eIn this study, we first investigated whether neutrophil dynamics could be predicted using routinely available clinical data from patients undergoing pemetrexed‒platinum chemotherapy, without relying on measured plasma concentration of pemetrexed. To achieve this, we employed physiologically based PK (PBPK) modeling, a mathematical approach that describes and predicts a drug\u0026rsquo;s behavior across various physiological tissues [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. NLME modeling is a widely used pharmacometrics approach in PK and PK/PD analyses of longitudinal data [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. It enables the evaluation of various structural models and covariate effects, uses all available observation points, and facilitates model validation through simulation. Using this framework, we aimed to identify patient-specific risk factors, including patients\u0026rsquo; characteristics obtained from medical records as well as PK parameters of pemetrexed derived from a PBPK model, that influence changes in neutrophil counts during the first cycle of pemetrexed‒platinum chemotherapy in patients with lung cancer, by constructing a population model of neutrophil dynamics using an NLME modeling approach. The first cycle was selected for analysis because empirical dosing is generally applied at this stage, and patients have not yet been influenced by dose adjustments or supportive measures, allowing for an objective evaluation of baseline risk factors. Furthermore, we validated whether the developed model is generalizable to other patients receiving combination regimens that include ICIs or molecularly targeted agents.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy population\u003c/h2\u003e \u003cp\u003eIn this retrospective study, we analyzed data from two patient cohorts at Saiseikai Kumamoto Hospital from January 2015 to December 2023. The detailed inclusion and exclusion criteria for the modeling dataset were as follows: (1) received pemetrexed-platinum chemotherapy for at least one cycle; (2) had accurate records of treatment initiation and discontinuation; (3) underwent at least one hematological examination during the first treatment cycle; (4) had no prior exposure to pemetrexed or platinum agents; and (5) provided written informed consent. The development cohort comprised 86 patients with NSCLC (62 men and 24 women; mean age, 68.24\u0026thinsp;\u0026plusmn;\u0026thinsp;7.85 years) who received pemetrexed‒platinum therapy, including 38 patients treated with cisplatin plus pemetrexed and 48 patients treated with carboplatin plus pemetrexed. To evaluate the applicability of the model to regimens combined with antibody drugs, such as pembrolizumab or bevacizumab, which are currently used as standard treatments, the validation cohort included 83 patients with NSCLC (51 men and 32 women, 65.52\u0026thinsp;\u0026plusmn;\u0026thinsp;7.04 years), comprising 39 patients treated with carboplatin plus pemetrexed plus pembrolizumab, 20 with cisplatin plus pemetrexed plus bevacizumab, and 24 with carboplatin plus pemetrexed plus bevacizumab. In all regimens, chemotherapeutic agents and antibody drugs were administered on day 1 of the 21-day cycle. The standard doses of the chemotherapeutic agents were as follows: 500 mg/m\u003csup\u003e2\u003c/sup\u003e of pemetrexed, 75 mg/m\u003csup\u003e2\u003c/sup\u003e of cisplatin, and multiplying an AUC of 5 mg\u0026middot;min/mL by [25\u0026thinsp;+\u0026thinsp;GFR (mL/min)] for carboplatin (mg); however, the actual doses of pemetrexed and platinum agents were determined empirically by physicians based on clinical judgment, taking into account patient-specific factors such as age, renal function, performance status, and comorbidities. For the validation cohort, pembrolizumab was administered at a fixed dose of 200 mg, and bevacizumab at 15 mg/kg, according to standard clinical protocols at the time. Neutrophil data from all measurement points in the first treatment cycle were used in the analysis, excluding any time points after the administration of granulocyte colony-stimulating factor. The final dataset comprised 316 and 375 data points from the development and validation cohorts, respectively, after excluding 6 data points (from 3 patients) and 5 data points (from 1 patient), respectively. The following information was collected from patient medical records: sex, age, height, weight, body mass index (BMI), creatinine, blood urea nitrogen (BUN), chemotherapy regimens, doses of anticancer agents, and other concomitant medications. Renal function was estimated using the Cockcroft‒Gault formula for CLcr and the Japanese equation for individualized eGFR [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. The eGFR values were de-indexed from 1.73 m\u003csup\u003e2\u003c/sup\u003e to each patient\u0026rsquo;s body surface area (BSA), calculated using the Du Bois formula, to obtain absolute values (mL/min).\u003c/p\u003e \u003cp\u003e The protocol was approved by the Ethics Committee of Saiseikai Kumamoto Hospital (Approval Nos. 1187 and 1319). The study was performed in accordance with the Declaration of Helsinki.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDeveloping a PBPK model for pemetrexed\u003c/h3\u003e\n\u003cp\u003eA PBPK model for pemetrexed was developed using PK-Sim (version 11.3, Open Systems Pharmacology Suite). Physicochemical and PK parameters (including molecular weight, logP, pKa, plasma protein binding, and renal clearance) were obtained from the literature [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e] and the PubChem database [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e] (Supplementary Table\u0026nbsp;1). Model parameters not directly available were optimized using the Levenberg‒Marquardt algorithm implemented in PK-Sim. Pemetrexed is a hydrophilic antifolate primarily eliminated by renal clearance (~\u0026thinsp;70\u0026ndash;90%), with limited hepatic metabolism [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The renal excretion process was modeled to account for glomerular filtration and active secretion, reflecting the known involvement of transporters.\u003c/p\u003e\n\u003ch3\u003eEvaluation of the PBPK model for pemetrexed\u003c/h3\u003e\n\u003cp\u003eModel performance was evaluated by comparing predicted and observed values [\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] for clearance (CL), maximum plasma concentration (C\u003csub\u003emax\u003c/sub\u003e), and AUC from time zero to infinity (AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e). Predicted values for these PK parameters were obtained through simulations in which pemetrexed was administered to virtual populations generated in PK-Sim. In creating virtual populations, the number of individuals, race, proportion of females, and age (mean and range) were set to match those reported in the literature [\u003cspan additionalcitationids=\"CR27\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. Model performance was assessed using the average fold error (AFE) between predicted and observed PK parameters. AFE was calculated using Eq.\u0026nbsp;1:\u003c/p\u003e \u003cp\u003eAFE\u0026thinsp;=\u0026thinsp;Predicted value / Observed value (Eq.\u0026nbsp;1)\u003c/p\u003e \u003cp\u003eA two-fold error range (AFE between 0.5 and 2) was considered acceptable for model evaluation [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFirst, the predictive performance of the PBPK model was evaluated without accounting for differences in varying renal function. Predicted PK parameters were obtained for eight pemetrexed doses (300, 500, 600, 700, 800, 900, 1,000, and 1,200 mg/m\u003csup\u003e2\u003c/sup\u003e) for which observed values were reported in the literature [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. These PK parameters were predicted based on simulations using healthy adult populations, as patients with impaired renal or hepatic function were excluded from the studies reporting the observed data [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. To evaluate model performance, AFE was calculated for each PK parameter at each dose simulation, and the geometric mean of the AFEs was subsequently computed.\u003c/p\u003e \u003cp\u003eSubsequently, model performance was assessed across populations with varying renal function by comparing predicted and observed [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] PK parameters for three groups: GFR\u0026thinsp;\u0026ge;\u0026thinsp;80, 60\u0026ndash;79, and 40\u0026ndash;59 mL/min. The mean GFR for each group was set to match previously reported values [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. PK parameters for the GFR 40\u0026ndash;59 mL/min group were predicted using simulations of a chronic kidney disease population, whereas those for the other groups were predicted using simulations of healthy adult populations. The AFE for the PK parameters was calculated for each group, and the predicted plasma concentration-time profiles (arithmetic mean and 5th\u0026ndash;95th percentiles) were visually compared with the observed data [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] for each GFR group. Observed concentration‒time data [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] were digitized from the published figures using WebPlotDigitizer (version 4.8, Ankit Rohatgi, Pacifica, CA). Plasma concentration‒time profiles were illustrated using R (version 4.1.1, R Foundation for Statistical Computing, Vienna, Austria).\u003c/p\u003e\n\u003ch3\u003eDeveloping a population model for changes in neutrophil counts over time\u003c/h3\u003e\n\u003cp\u003eNONMEM (version 7.5.1, ICON Dev Soln, Ellicott City, MD) was used to model changes in neutrophil count during the first cycle of pemetrexed‒platinum therapy. The model was fitted to neutrophil data using the first-order conditional estimation with interaction method. Changes in neutrophil count during the first cycle were best described visually by a quadratic function. The base model comprised two quadratic functions: one for the decrease (Eq.\u0026nbsp;2) and the other for the recovery (Eq.\u0026nbsp;3) of the neutrophil count.\u003c/p\u003e \u003cp\u003ey\u0026thinsp;=\u0026thinsp;A1 \u0026times; (WEEK\u0026thinsp;\u0026minus;\u0026thinsp;B)\u003csup\u003e2\u003c/sup\u003e + Base\u0026thinsp;\u0026minus;\u0026thinsp;A1 \u0026times; B\u003csup\u003e2\u003c/sup\u003e \u0026times; exp(ε) if WEEK is lower than B (Eq.\u0026nbsp;2)\u003c/p\u003e \u003cp\u003ey\u0026thinsp;=\u0026thinsp;A2 \u0026times; (WEEK\u0026thinsp;\u0026minus;\u0026thinsp;B)\u003csup\u003e2\u003c/sup\u003e + Base\u0026thinsp;\u0026minus;\u0026thinsp;A1 \u0026times; B\u003csup\u003e2\u003c/sup\u003e \u0026times; exp(ε) if WEEK is greater than B (Eq.\u0026nbsp;3)\u003c/p\u003e \u003cp\u003eWEEK is the elapsed time in weeks, calculated as the number of days since pemetrexed administration divided by seven. Base denotes neutrophil counts before pemetrexed administration (typically within 1 week). A1 and A2 represent the slopes of neutrophil decline and recovery, respectively. B is the WEEK when neutrophil recovery starts, and ε denotes intra-individual variabilities with mean zero and variance σ\u003csup\u003e2\u003c/sup\u003e. To identify factors influencing interindividual variability in neutrophil counts following pemetrexed and platinum administration, the effects of various covariates on parameters A1, A2, and B were assessed.\u003c/p\u003e \u003cp\u003e \u003cb\u003eStep 1\u003c/b\u003e: A model was constructed using routinely available clinical variables in standard medical practice. The following covariates were evaluated: age, BMI, BSA, renal function parameters (CLcr, eGFR, and BUN), pemetrexed dose, type of platinum-based agent (cisplatin or carboplatin), and the use of concomitant medications, including non-steroidal anti-inflammatory drugs, renin‒angiotensin system (RAS) inhibitors, and diuretics. Covariate values were obtained at the initiation of chemotherapy during the first cycle.\u003c/p\u003e \u003cp\u003e \u003cstrong\u003eStep 2\u003c/strong\u003e \u003cp\u003eIn the second step, instead of the pemetrexed dose, PBPK model-derived PK parameters of pemetrexed (CL, C\u003csub\u003emax\u003c/sub\u003e, AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e, and AUC from time zero to 24 h [AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e]) were also included as covariates to capture interindividual differences in drug exposure beyond dose.\u003c/p\u003e \u003c/p\u003e \u003cp\u003eThe covariate model was developed through univariate analysis in NONMEM, followed by a stepwise method using forward inclusion and backward elimination. The effect of covariates on parameters A1, A2, and B was assessed using the likelihood ratio test, wherein changes in the objective function value (OFV) computed by NONMEM were compared against critical values from the chi-squared (χ\u003csup\u003e2\u003c/sup\u003e) distribution. In the forward step, covariates that exhibited significant effects in the univariate analysis were added to the model in descending order of their impact on OFV. The covariates were included in the model when the OFV decreased by \u0026gt;\u0026thinsp;3.84 (χ\u003csup\u003e2\u003c/sup\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, one degree of freedom) and 5.99 (χ\u003csup\u003e2\u003c/sup\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, two degrees of freedom), thereby developing the full model. In the backward step, models with only one covariate excluded from the full model were developed. The covariates were retained in the model when the OFV increased by \u0026gt;\u0026thinsp;3.84 (χ\u003csup\u003e2\u003c/sup\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, one degree of freedom) and 5.99 (χ\u003csup\u003e2\u003c/sup\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;\u0026lt;\u0026thinsp;0.05, two degrees of freedom), resulting in the final model.\u003c/p\u003e\n\u003ch3\u003eModel evaluation of the development cohort\u003c/h3\u003e\n\u003cp\u003eThe model evaluation was performed using the same procedures previously reported [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e]. The final model was internally validated using goodness-of-fit (GOF) plots and visual predictive checks (VPCs). A stratified nonparametric bootstrap analysis was also performed to investigate model robustness and parameter precision. The GOF plots used scatter plots of observed versus population-predicted values, observed versus individual-predicted values, and conditional weighted residuals versus population-predicted values and time after the pemetrexed dose. A random sampling of the original data generated 1,000 replicated datasets. In the VPC approach, the 5th, 50th, and 95th percentiles of simulations from the final model were compared with the observed values. The 95% confidence intervals (CIs) for the parameters after successful bootstrap convergence were compared with the final model estimates. In addition, we generated a receiver operating characteristic (ROC) curve for a decrease in neutrophil count to \u0026lt;\u0026thinsp;1,000/\u0026micro;L, and we confirmed prediction accuracy using the area under the ROC curve (AUROC).\u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eExternal validation of the final model\u003c/h2\u003e \u003cp\u003eTo assess the predictive performance and generalizability of the final model, external validation was conducted using data from the validation cohort. PK parameters of pemetrexed\u0026mdash;specifically CL, C\u003csub\u003emax\u003c/sub\u003e, AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e, and AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e\u0026mdash;were estimated for each patient in the validation cohort using the PBPK model constructed in the development cohort. Subsequently, individual neutrophil counts during the first treatment cycle were predicted using Bayesian estimation based on the final model implemented in NONMEM. Model performance was assessed using GOF plots and the VPC approach, following the same procedure as in the model validation of the development cohort. Furthermore, an ROC curve for a decrease in neutrophil count to \u0026lt;\u0026thinsp;1,000/\u0026micro;L was generated, and its discriminative ability was assessed using the AUROC. Statistical analyses and graphics were generated using R.\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003ePatient demographics\u003c/h2\u003e \u003cp\u003eA comparison of the demographic and clinical characteristics of patients in the development (n\u0026thinsp;=\u0026thinsp;86) and validation (n\u0026thinsp;=\u0026thinsp;83) cohorts is summarized in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Patients in the validation cohort received higher doses of pemetrexed, had a higher proportion of concomitant carboplatin use, were younger, and had higher BMI values than those in the development cohort. No significant differences were observed in other baseline characteristics.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eParticipant clinical characteristics\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eAge (year)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDevelopment cohort\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;86)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eValidation cohort\u003c/p\u003e \u003cp\u003e(n\u0026thinsp;=\u0026thinsp;83)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cem\u003eP\u003c/em\u003e value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e68.24\u0026thinsp;\u0026plusmn;\u0026thinsp;7.85\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e65.52\u0026thinsp;\u0026plusmn;\u0026thinsp;7.04\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.019\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSex (male/female)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e62 (72.1) /24 (27.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e51 (61.4) /32 (38.6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.191\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHeight (cm)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e161.77\u0026thinsp;\u0026plusmn;\u0026thinsp;9.10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e161.88\u0026thinsp;\u0026plusmn;\u0026thinsp;8.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.939\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWeight (kg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e58.65\u0026thinsp;\u0026plusmn;\u0026thinsp;11.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e62.10\u0026thinsp;\u0026plusmn;\u0026thinsp;13.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.065\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBMI (kg/ m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e22.34\u0026thinsp;\u0026plusmn;\u0026thinsp;3.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e23.54\u0026thinsp;\u0026plusmn;\u0026thinsp;3.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.034\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBSA (m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.61\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.65\u0026thinsp;\u0026plusmn;\u0026thinsp;0.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.170\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCreatinine (mg/dL)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.76\u0026thinsp;\u0026plusmn;\u0026thinsp;0.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.76\u0026thinsp;\u0026plusmn;\u0026thinsp;0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.985\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCLcr (mL/min)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e77.32\u0026thinsp;\u0026plusmn;\u0026thinsp;24.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e82.93\u0026thinsp;\u0026plusmn;\u0026thinsp;24.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.135\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eeGFR (mL/min)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e71.09\u0026thinsp;\u0026plusmn;\u0026thinsp;19.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e70.78\u0026thinsp;\u0026plusmn;\u0026thinsp;16.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.908\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBUN (mg/dL)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e15.25\u0026thinsp;\u0026plusmn;\u0026thinsp;5.16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e14.44\u0026thinsp;\u0026plusmn;\u0026thinsp;3.93\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.257\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePemetrexed dose (mg)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e760.64\u0026thinsp;\u0026plusmn;\u0026thinsp;127.05\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e802.64\u0026thinsp;\u0026plusmn;\u0026thinsp;116.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.027\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePemetrexed dose per BSA (mg/m\u003csup\u003e2\u003c/sup\u003e)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e469.54\u0026thinsp;\u0026plusmn;\u0026thinsp;51.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e484.27\u0026thinsp;\u0026plusmn;\u0026thinsp;30.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.026\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCo-administration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCisplatin/Carboplatin\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e38 (44.2) / 48 (55.8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e20 (24.1) / 63 (75.9)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.009\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBevacizumab/Pembrolizumab\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e44 (53.0) / 39 (47.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026ndash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNSAIDs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e23 (26.7)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12 (14.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.058\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRAS inhibitor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e25 (29.1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e17 (20.5)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.216\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDiuretics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e8 (9.3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5 (6.0)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.566\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eData are presented as means\u0026thinsp;\u0026plusmn;\u0026thinsp;standard deviations or proportions for categorical variables.\u003c/td\u003e\u003c/tr\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eBMI, body mass index; BSA, body surface area; BUN, blood urea nitrogen; CLcr, creatinine clearance; eGFR, estimated glomerular filtration rate; NSAIDs, non-steroidal anti-inflammatory drugs; RAS, renin‒angiotensin system.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003ePK parameters estimation using the PBPK approach\u003c/h2\u003e \u003cp\u003eSupplementary Table\u0026nbsp;2 summarizes the predicted PK parameters, the corresponding observed values reported in the literature [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e], and the AFE for each pemetrexed dose. The developed PBPK model predicted PK parameters within a two-fold error range across the wide range of pemetrexed dose levels. Moreover, the geometric mean (95% CI) of the AFE for CL, C\u003csub\u003emax\u003c/sub\u003e, and AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e were 1.04 (0.86\u0026ndash;1.26), 0.91 (0.80\u0026ndash;1.05), and 1.00 (0.84\u0026ndash;1.20), respectively, suggesting the favorable predictive performance. Model evaluation across three virtual populations with varying GFRs similarly showed PK predictions within a two-fold error range for each group (Supplementary Table\u0026nbsp;3). The developed PBPK model successfully captured the expected PK trend: as GFR declines, CL decreases, and AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e increases, while C\u003csub\u003emax\u003c/sub\u003e generally remains relatively unchanged [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In addition, simulated 24-h plasma concentration‒time profiles of pemetrexed closely matched observed data [\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e] across groups (Supplementary Fig.\u0026nbsp;1), supporting the validity of the developed PBPK model. Subsequently, individual CL, C\u003csub\u003emax\u003c/sub\u003e, AUC\u003csub\u003e0\u0026ndash;inf\u003c/sub\u003e, and AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e values were calculated for each participant using the developed PBPK model and patient-specific information, including race (Japanese), age, sex, height, weight, BMI, BSA, eGFR, and pemetrexed dose.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eDevelopment of the final model for the change in neutrophil count\u003c/h2\u003e \u003cp\u003eSupplementary Table\u0026nbsp;4 summarizes the impact of tested covariates on the OFV in the neutrophil count variation model. In Step 1, the model was developed using routinely available information, without including PK parameters. The model parameters at this step were as follows:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eA1: 0.717 \u0026times; (Dose/760)\u003csup\u003e1.27\u003c/sup\u003e \u0026times; exp(η) (Eq.\u0026nbsp;4)\u003c/p\u003e\u003cp\u003eA2: 0.0319 \u0026times; exp(η) (Eq.\u0026nbsp;5)\u003c/p\u003e\u003cp\u003eB: 1.32 \u0026times; exp(η) (Eq.\u0026nbsp;6)\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere η represents interindividual variability with a mean of zero and variance ω\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eStep 1 yielded a predictive model that may explain changes in neutrophil counts solely in terms of the pemetrexed dose. However, as we retrospectively obtained real-world clinical data, pemetrexed doses may have been adjusted based on individual patient characteristics, such as age and renal function. Furthermore, the model included only the pemetrexed dose, making it challenging to predict individualized dosing before starting chemotherapy. Therefore, we incorporated the PK parameters of pemetrexed derived from the PBPK model (Step 2) to capture interindividual variability in pharmacokinetics beyond empirically adjusted doses. This extended analysis identified BUN, concomitant use of RAS inhibitors, and pemetrexed AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e as statistically significant covariates (Supplementary Table\u0026nbsp;5). The final model equations incorporating these covariates were as follows:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003eA1: 0.334 \u0026times; (BUN/15.2)\u003csup\u003e\u0026minus;0.975\u003c/sup\u003e \u0026times; exp(η) (Eq.\u0026nbsp;7)\u003c/p\u003e\u003cp\u003eA2: 0.0443 \u0026times; 0.0451\u003csup\u003eRAS inhibitor\u003c/sup\u003e \u0026times; exp(η) (Eq.\u0026nbsp;8)\u003c/p\u003e\u003cp\u003eB: 1.39 \u0026times; 1.44\u003csup\u003eAUC0\u0026ndash;24\u0026gt;175\u003c/sup\u003e \u0026times; exp(η) (Eq.\u0026nbsp;9)\u003c/p\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere RAS inhibitor\u0026thinsp;=\u0026thinsp;1 if RAS inhibitor is administered concomitantly, otherwise 0; AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e\u0026gt;175\u0026thinsp;=\u0026thinsp;1 if pemetrexed AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e exceeds 175 \u0026micro;g\u0026middot;h/mL, otherwise 0; and η represents interindividual variability with a mean of zero and variance ω\u003csup\u003e2\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe effects of individual patient factors incorporated into the final model are illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e The model suggested that patients with lower BUN levels experienced a more rapid decline in neutrophil counts (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA), those receiving concomitant RAS inhibitors showed delayed recovery (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB), and patients with a pemetrexed AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e \u0026gt;175 \u0026micro;g\u0026middot;h/mL exhibited a delayed onset of neutrophil recovery (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eEvaluation of the final model\u003c/h2\u003e \u003cp\u003eThe GOF plots suggested a correlation between the neutrophil counts predicted by the final model and the actual measurements (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB). The distribution of residuals showed no significant pattern across population-predicted values and time after pemetrexed administration (Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD). The VPCs showed that the final model described the observed neutrophil data without substantial misspecification (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Among 1,000 bootstrap runs, 622 exhibited successful minimization and were included in the bootstrap analysis. The 95% CIs for all parameters obtained using the bootstrap approach were generally consistent with the NONMEM estimates (Supplementary Table\u0026nbsp;6). The final model demonstrated high performance in predicting a decrease in neutrophil count to \u0026lt;\u0026thinsp;1,000/\u0026micro;L, with an AUROC of 0.92 (95% CI: 0.75\u0026ndash;1.00).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003eExternal validation\u003c/h2\u003e \u003cp\u003eVisual inspection of the GOF plots and VPCs confirmed that the final model accurately predicted changes in neutrophil counts among patients in the validation cohort, demonstrating performance comparable to that observed in the development cohort (Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). Notably, the model maintained a strong predictive ability for identifying patients with neutrophil counts\u0026thinsp;\u0026lt;\u0026thinsp;1,000/\u0026micro;L, achieving an AUROC of 0.85 (95% CI: 0.57\u0026ndash;1.00) in the validation cohort.\u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we developed a predictive model for neutrophil dynamics following pemetrexed‒platinum chemotherapy using PBPK and NLME modeling approaches (Equations 7‒9). The model demonstrated high accuracy in predicting neutrophil counts\u0026thinsp;\u0026lt;\u0026thinsp;1,000/\u0026micro;L (AUROC: 0.92, 95% CI: 0.75\u0026ndash;1.00), enabling early identification of high-risk patients and supporting timely dose adjustment to reduce the risk of severe neutropenia. Furthermore, the model was externally validated in a separate cohort of patients receiving combination therapies that included ICIs or molecular-targeted agents, supporting the model\u0026rsquo;s broader applicability and potential clinical utility across diverse treatment settings beyond conventional chemotherapy alone. Traditionally, the risk factors associated with neutropenia following pemetrexed‒platinum therapy remain unclear, with dose adjustments often being made empirically. Our findings suggest the potential for data-driven, dose-individualized pemetrexed dosing based on quantitative predictions, even during the first treatment cycle, when neutrophil dynamics are not yet known, offering a path toward more personalized and safer chemotherapy regimens.\u003c/p\u003e \u003cp\u003eIn the model constructed using only routinely available clinical data, pemetrexed dose was the only significant covariate (Equations 4‒6). However, the pemetrexed dose showed a weak but significant correlation with age and renal function (Supplementary Fig.\u0026nbsp;2). For example, patients aged\u0026thinsp;\u0026ge;\u0026thinsp;75 years received significantly lower doses of pemetrexed than those aged\u0026thinsp;\u0026lt;\u0026thinsp;75 years (477.44\u0026thinsp;\u0026plusmn;\u0026thinsp;39.93 mg/m\u003csup\u003e2\u003c/sup\u003e vs. 441.69\u0026thinsp;\u0026plusmn;\u0026thinsp;74.51 mg/m\u003csup\u003e2\u003c/sup\u003e, \u003cem\u003eP\u003c/em\u003e\u0026thinsp;=\u0026thinsp;0.007). These findings suggest that pemetrexed dosing was likely adjusted based on various patient characteristics, guided by clinical judgment and experience, as the data were collected retrospectively in a real-world setting. Therefore, the influence of different patient factors may have been indirectly reflected in the administered doses, which likely explains why only the dose was included in the initial model.\u003c/p\u003e \u003cp\u003eIn contrast to previous semi-mechanistic PK/PD models primarily developed using controlled clinical trial datasets [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], our approach was designed to accommodate the substantial variability encountered in real-world clinical practice. By directly modeling neutrophil dynamics using routinely collected clinical data and PBPK-derived PK parameters, our framework enhances applicability in settings where therapeutic drug monitoring is unavailable. Furthermore, we investigated the influence of individual PK differences, beyond empirical dose adjustments, on neutrophil dynamics following pemetrexed‒platinum chemotherapy. The final model demonstrated that a higher pemetrexed AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e was associated with a delayed onset of neutrophil recovery (Eq.\u0026nbsp;9), consistent with the semi-mechanistic physiological PD model of pemetrexed proposed by Latz et al., which quantitatively attributed delayed neutrophil recovery to prolonged maturation time of precursor cells as they transition from the maturation compartment to peripheral blood [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. In high-exposure settings, the degree of precursor cell depletion increases, leading to a marked delay in peripheral blood recovery [\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. It should be noted that previous work by Boosman et al. suggested that pemetrexed-related toxicity, particularly in patients with renal impairment, is more closely driven by a threshold exposure metric than by cumulative AUC [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e]. Therefore, the association between AUC\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e and delayed neutrophil recovery observed in our study should be interpreted as a pragmatic exposure surrogate within our dataset rather than a definitive mechanistic driver of toxicity. We aimed to determine whether a parsimonious empirical exposure\u0026ndash;response relationship could support clinically actionable risk prediction in settings where plasma concentration measurements are unavailable. Importantly, the exposure metrics in our model were estimated from patient characteristics without direct concentration measurements, supporting its applicability in real-world clinical settings.\u003c/p\u003e \u003cp\u003eMoreover, prolonged dosing intervals, potentially resulting from delayed neutrophil recovery, can increase mortality in patients undergoing first-line treatment for NSCLC [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. By leveraging PBPK and NLME modeling approaches, this study provides a novel framework for detailed evaluation of exposure‒toxicity relationships, suggesting the possibility of designing individualized dosing regimens that maximize efficacy and safety, even without the need for plasma pemetrexed concentration measurements. Collectively, these findings emphasize the importance of optimizing pemetrexed dosing from the first treatment cycle onward to minimize the risk of severe toxicities, maintain the intended treatment schedule, and ultimately improve therapeutic outcomes.\u003c/p\u003e \u003cp\u003eIn addition to drug exposure metrics, patient-specific factors such as BUN levels also influenced neutrophil dynamics (Eq.\u0026nbsp;7). This study demonstrated that the rate of neutrophil decline after pemetrexed administration was greater in patients with lower BUN levels. When participants were stratified by median BUN, the low-BUN group was younger, had better renal function, and had a higher rate of concomitant cisplatin use than the high-BUN group (Supplementary Table\u0026nbsp;7). Notably, no significant between-group differences in pemetrexed or platinum doses were observed, indicating that the observed decline in neutrophils was not attributable to differences in chemotherapy intensity. Given that cisplatin is generally associated with a lower frequency of severe neutropenia than carboplatin [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e], the greater neutrophil decline observed in the low BUN group is unlikely to be explained solely by differences in chemotherapy regimens.\u003c/p\u003e \u003cp\u003eNotably, younger age and preserved renal function are typically considered protective against severe myelosuppression [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e, \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e]. Therefore, this finding appears counterintuitive and suggests the involvement of other underlying factors. BUN may have been incorporated into the model not solely as a marker of renal clearance, but also as a surrogate for broader physiological conditions\u0026mdash;such as nutritional status or metabolic reserve\u0026mdash;that influence hematologic sensitivity [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. Thus, the observed association between low BUN and greater neutrophil decline may reflect a complex interplay of the aforementioned patient-specific factors. Nevertheless, further studies are warranted to confirm these findings and elucidate the underlying mechanisms.\u003c/p\u003e \u003cp\u003eFurther, this study suggests that neutrophil recovery after pemetrexed administration is slower in patients taking RAS inhibitors (Eq.\u0026nbsp;8). Angiotensin II reportedly regulates hematopoiesis \u003cem\u003ein vivo\u003c/em\u003e at the level of stem cells [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Moreover, angiotensin II analogs reportedly promote hematopoiesis and facilitate the recovery of circulating cells [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Therefore, RAS inhibitors may disrupt the hematopoietic regulatory function of angiotensin II, leading to delayed neutrophil recovery following pemetrexed administration. In addition, some RAS inhibitors, such as olmesartan and valsartan, have been reported to inhibit the organic anion transporter 3 (OAT3) [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], which is crucial for pemetrexed excretion. Concomitant use of OAT3 inhibitors increases pemetrexed AUC, suggesting that RAS inhibitors could impair pemetrexed elimination and potentially increase drug exposure and toxicity risk [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. However, the detailed mechanisms underlying this interaction remain unclear and warrant further investigation.\u003c/p\u003e \u003cp\u003eRecent advances in NSCLC treatment have demonstrated that the addition of antibody drugs, such as bevacizumab or pembrolizumab, to pemetrexed‒platinum regimens can enhance therapeutic efficacy [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Although the addition of antibody drugs can cause a range of side effects, their impact on hematologic toxicity is generally minimal [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e]. Given the increasing clinical use of these combination therapies, we performed external validation of the developed model using a separate cohort of patients receiving pemetrexed‒platinum chemotherapy in combination with these antibody drugs. The final model demonstrated the ability to predict neutrophil count fluctuations after pemetrexed‒platinum administration, even in the presence of concomitant antibody drug use. Therefore, the results of this study may be applicable across a wide range of pemetrexed‒platinum regimens.\u003c/p\u003e \u003cp\u003eThis study has some limitations. First, as blood samples for the PK analysis were not collected from our study participants, the predictive performance of the PBPK model could not be evaluated. Therefore, future studies incorporating clinical PK data will be necessary to verify its accuracy. In addition, the developed PBPK model focused solely on pemetrexed and did not consider the effects of concomitant medications on its pharmacokinetics. In particular, the potential impact of drugs that interact with OAT3, which plays a significant role in pemetrexed renal elimination [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e], was not accounted for in the pemetrexed PBPK model. Accordingly, future enhancements to the PBPK model should incorporate the effects of concomitant medications, particularly OAT3 inhibitors, to better predict pemetrexed PK. Moreover, the model did not explicitly include the type of platinum agent used (cisplatin or carboplatin). As previously mentioned, the choice and dosage of platinum agents may have been adjusted based on patient-specific factors, such as renal function. In addition, this study did not consider the pharmacokinetics of platinum agents. To enhance the model's accuracy and applicability, future studies should consider these factors. Furthermore, this study did not assess the relationship between pemetrexed exposure and antitumor efficacy. Although the model focused on hematologic toxicity, evaluating exposure‒efficacy relationships is also crucial for dosage optimization. Future studies should investigate whether pemetrexed exposure is correlated with clinical outcomes, such as response or survival. Furthermore, although this observational study was designed to reflect real-world clinical conditions, its findings may have limited generalizability because of the relatively small sample size, the single-institution setting, and the exclusive inclusion of Japanese patients. These factors may restrict the applicability of the results to broader, more diverse populations. Many patient-specific factors may influence the pharmacokinetics and pharmacodynamics of pemetrexed. Therefore, it is essential to confirm that the study findings are not limited to the study population through retrospective analyses in other patient groups with comparable backgrounds. Subsequently, prospective validation in larger, more diverse populations will be necessary to assess the generalizability of our findings.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study demonstrated the potential to predict neutrophil dynamics after pemetrexed‒platinum chemotherapy by integrating exposure metrics estimated from patient characteristics, without requiring measured plasma concentrations of pemetrexed or patient-specific factors such as BUN levels or concomitant RAS inhibitor use. Our findings suggest that data-driven, individualized pemetrexed dosing may be feasible, starting with the first treatment cycle, by accounting for individual patient characteristics. Although further validation with actual PK data and consideration of concomitant drug effects are needed, these results highlight the promise of model-informed strategies to optimize pemetrexed dosing and improve therapeutic outcomes.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors thank all of the study participants. We would like to thank Editage (www.editage.jp) for English language editing.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKM, KS, HN, TK, AKM, KO, and JS contributed to the conceptualization and study design. KM, KT, AS, HT, and KI contributed to data curation and investigation. KM, KS, HN, TK, NS, and JS performed formal analysis and interpreted the data. KM, KS, HN, AKM, KO, and JS drafted the manuscript. TK, RT, HJ, HI, DK, KT, AS, HT, KI, AKM, KO, and JS critically reviewed and revised the manuscript for important intellectual content. All authors approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by JST SPRING (No. JPMJSP2127). None of the funders played a role in the design, implementation, analysis, and interpretation of the data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflict of Interest\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAuthors do not report any conflict of interest for this work.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe protocol was approved by the Ethics Committee of Saiseikai Kumamoto Hospital (Approval Nos. 1187 and 1319). The study was performed in accordance with the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cbr\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74(3):229-263.10.3322/caac.21834\u003c/li\u003e\n\u003cli\u003eAriyasu R, Kakuto S, Miyadera K, Akita T, Kiritani A, Tsugitomi R, et al. Real-World outcome analysis of patients with stage IV NSCLC treated with tyrosine kinase and immune checkpoint inhibitors. JTO Clin Res Rep. 2023;4(6):100524.10.1016/j.jtocrr.2023.100524\u003c/li\u003e\n\u003cli\u003eGandhi L, Rodriguez-Abreu D, Gadgeel S, Esteban E, Felip E, De Angelis F, et al. Pembrolizumab plus chemotherapy in metastatic non-small-cell lung cancer. N Engl J Med. 2018;378(22):2078-2092.10.1056/NEJMoa1801005\u003c/li\u003e\n\u003cli\u003eMeyer ML, Peters S, Mok TS, Lam S, Yang PC, Aggarwal C, et al. Lung cancer research and treatment: global perspectives and strategic calls to action. Ann Oncol. 2024;35(12):1088-1104.10.1016/j.annonc.2024.10.006\u003c/li\u003e\n\u003cli\u003eScagliotti GV, Parikh P, von Pawel J, Biesma B, Vansteenkiste J, Manegold C, et al. Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advanced-stage non-small-cell lung cancer. J Clin Oncol. 2023;41(14):2458-2466.10.1200/JCO.22.02544\u003c/li\u003e\n\u003cli\u003eWang XJ, Wong M, Hsu LY, Chan A. Costs associated with febrile neutropenia in solid tumor and lymphoma patients - an observational study in Singapore. BMC Health Serv Res. 2014;14:434.10.1186/1472-6963-14-434\u003c/li\u003e\n\u003cli\u003eCrawford J, Denduluri N, Patt D, Jiao X, Morrow PK, Garcia J, et al. Relative dose intensity of first-line chemotherapy and overall survival in patients with advanced non-small-cell lung cancer. Support Care Cancer. 2020;28(2):925-932.10.1007/s00520-019-04875-1\u003c/li\u003e\n\u003cli\u003eKidney Disease: Improving Global Outcomes CKDWG. KDIGO 2024 clinical practice guideline for the evaluation and management of chronic kidney disease. Kidney Int. 2024;105(4S):S117-S314.10.1016/j.kint.2023.10.018\u003c/li\u003e\n\u003cli\u003eYanagita M, Muto S, Nishiyama H, Ando Y, Hirata S, Doi K, et al. Clinical questions and good practice statements of clinical practice guidelines for management of kidney injury during anticancer drug therapy 2022. Clin Exp Nephrol. 2024;28(2):85-122.10.1007/s10157-023-02415-0\u003c/li\u003e\n\u003cli\u003eZhang J, Ye ZW, Tew KD, Townsend DM. Cisplatin chemotherapy and renal function. Adv Cancer Res. 2021;152:305-327.10.1016/bs.acr.2021.03.008\u003c/li\u003e\n\u003cli\u003eCalvert AH, Newell DR, Gumbrell LA, O\u0026apos;Reilly S, Burnell M, Boxall FE, et al. Carboplatin dosage: prospective evaluation of a simple formula based on renal function. J Clin Oncol. 1989;7(11):1748-1756.10.1200/JCO.1989.7.11.1748\u003c/li\u003e\n\u003cli\u003ede Rouw N, Beunders R, Hartmann O, Schulte J, Boosman RJ, Derijks HJ, et al. A comparison of the renal function biomarkers serum creatinine, pro-enkephalin and cystatin C to predict clearance of pemetrexed. Cancer Chemother Pharmacol. 2024;94(6):799-806.10.1007/s00280-024-04717-w\u003c/li\u003e\n\u003cli\u003eSaito Y, Taniguchi O, Takekuma Y, Sakakibara-Konishi J, Shimizu Y, Kinoshita I, et al. Impact of baseline renal impairment on severe neutropenia development in pemetrexed and carboplatin thoracic cancer treatment. Support Care Cancer. 2024;32(12):829.10.1007/s00520-024-09006-z\u003c/li\u003e\n\u003cli\u003eShima Y, Yoshida H, Suminaga K, Yoshida H, Hashimoto K, Ogimoto T, et al. Safety and efficacy of pemetrexed for patients with non-small cell lung cancer in the setting of renal impairment: a retrospective study. BMC Cancer. 2025;25(1):388.10.1186/s12885-025-13785-x\u003c/li\u003e\n\u003cli\u003ede Rouw N, Otten LS, Kicken MP, Piet B, Biesma B, van Veggel B, et al. Folinic acid prophylaxis and dose adjustments enable safe treatment with pemetrexed in patients with renal impairment. Clin Pharmacol Ther. 2025;118(3):715-722.10.1002/cpt.3735\u003c/li\u003e\n\u003cli\u003eLatz JE, Schneck KL, Nakagawa K, Miller MA, Takimoto CH. Population pharmacokinetic/pharmacodynamic analyses of pemetrexed and neutropenia: effect of vitamin supplementation and differences between Japanese and Western patients. Clin Cancer Res. 2009;15(1):346-354.10.1158/1078-0432.CCR-08-0791\u003c/li\u003e\n\u003cli\u003eVisser S, Koolen SLW, de Bruijn P, Belderbos HNA, Cornelissen R, Mathijssen RHJ, et al. Pemetrexed exposure predicts toxicity in advanced non-small-cell lung cancer: A prospective cohort study. Eur J Cancer. 2019;121:64-73.10.1016/j.ejca.2019.08.012\u003c/li\u003e\n\u003cli\u003eLatz JE, Karlsson MO, Rusthoven JJ, Ghosh A, Johnson RD. A semimechanistic-physiologic population pharmacokinetic/pharmacodynamic model for neutropenia following pemetrexed therapy. Cancer Chemother Pharmacol. 2006;57(4):412-426.10.1007/s00280-005-0077-5\u003c/li\u003e\n\u003cli\u003eBoosman RJ, Dorlo TPC, de Rouw N, Burgers JA, Dingemans AC, van den Heuvel MM, et al. Toxicity of pemetrexed during renal impairment explained-Implications for safe treatment. Int J Cancer. 2021;149(8):1576-1584.10.1002/ijc.33721\u003c/li\u003e\n\u003cli\u003eKuepfer L, Niederalt C, Wendl T, Schlender JF, Willmann S, Lippert J, et al. Applied concepts in PBPK modeling: How to build a PBPK/PD model. CPT Pharmacometrics Syst Pharmacol. 2016;5(10):516-531.10.1002/psp4.12134\u003c/li\u003e\n\u003cli\u003eJian W, Xue J, Yao Q, Chen R, Yao Y, Wang M, et al. Starting dose selection of palbociclib in Chinese patients with breast cancer based on population kinetic-pharmacodynamic model of neutropenia. Cancer Chemother Pharmacol. 2022;90(6):489-497.10.1007/s00280-022-04484-6\u003c/li\u003e\n\u003cli\u003eTang F, Ng CM, Horn J, Bada HS, Leggas M. Pharmacokinetic modeling and model-based hypothesis generation for dose optimization of clonidine in neonates with neonatal opioid withdrawal syndrome. Clin Pharmacol Ther. 2025;117(5):1254-1263.10.1002/cpt.3507\u003c/li\u003e\n\u003cli\u003eMatsuo S, Imai E, Horio M, Yasuda Y, Tomita K, Nitta K, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis. 2009;53(6):982-992.10.1053/j.ajkd.2008.12.034\u003c/li\u003e\n\u003cli\u003ePosada MM, Bacon JA, Schneck KB, Tirona RG, Kim RB, Higgins JW, et al. Prediction of renal transporter mediated drug-drug interactions for pemetrexed using physiologically based pharmacokinetic modeling. Drug Metab Dispos. 2015;43(3):325-334.10.1124/dmd.114.059618\u003c/li\u003e\n\u003cli\u003eNational Center for Biotechnology Information. PubChem database. 2025 18 September. Available from: https://pubchem.ncbi.nlm.nih.gov/compound/Pemetrexed.\u003c/li\u003e\n\u003cli\u003eNakagawa K, Kudoh S, Matsui K, Negoro S, Yamamoto N, Latz JE, et al. A phase I study of pemetrexed (LY231514) supplemented with folate and vitamin B12 in Japanese patients with solid tumours. Br J Cancer. 2006;95(6):677-682.10.1038/sj.bjc.6603321\u003c/li\u003e\n\u003cli\u003eKavathiya K, Gurjar M, Patil A, Naik M, Noronha V, Joshi A, et al. A comparative pharmacokinetic study of 2 pemetrexed formulations in Indian adult chemonaive patients with adenocarcinoma stage III/IV non-small cell lung cancer. Clin Pharmacol Drug Dev. 2017;6(3):234-239.10.1002/cpdd.347\u003c/li\u003e\n\u003cli\u003eMita AC, Sweeney CJ, Baker SD, Goetz A, Hammond LA, Patnaik A, et al. Phase I and pharmacokinetic study of pemetrexed administered every 3 weeks to advanced cancer patients with normal and impaired renal function. J Clin Oncol. 2006;24(4):552-562.10.1200/JCO.2004.00.9720\u003c/li\u003e\n\u003cli\u003eRostami-Hodjegan A. Reverse Translation in PBPK and QSP: Going Backwards in Order to Go Forward With Confidence. Clin Pharmacol Ther. 2018;103(2):224-232.10.1002/cpt.904\u003c/li\u003e\n\u003cli\u003eShigetome K, Egashira T, Tomita T, Higa N, Iwashita K, Morita K, et al. Effect of cumulative exposure on the efficacy of paroxetine: A population pharmacokinetic-pharmacodynamic and machine learning analyses. CPT Pharmacometrics Syst Pharmacol. 2025;14(6):1119-1127.10.1002/psp4.70032\u003c/li\u003e\n\u003cli\u003eGriesinger F, Korol EE, Kayaniyil S, Varol N, Ebner T, Goring SM. Efficacy and safety of first-line carboplatin-versus cisplatin-based chemotherapy for non-small cell lung cancer: A meta-analysis. Lung Cancer. 2019;135:196-204.10.1016/j.lungcan.2019.07.010\u003c/li\u003e\n\u003cli\u003eKulkarni PM, Chen R, Anand T, Monberg MJ, Obasaju CK. Efficacy and safety of pemetrexed in elderly cancer patients: results of an integrated analysis. Crit Rev Oncol Hematol. 2008;67(1):64-70.10.1016/j.critrevonc.2008.01.011\u003c/li\u003e\n\u003cli\u003eOkada S, Yamazaki S, Kaiga T, Funada T, Kochi M, Takayama T. Impact of nutritional status in the era of FOLFOX/FIRI-based chemotherapy. World J Surg Oncol. 2017;15(1):162.10.1186/s12957-017-1226-0\u003c/li\u003e\n\u003cli\u003eKim S, Zingler M, Harrison JK, Scott EW, Cogle CR, Luo D, et al. Angiotensin II regulation of proliferation, differentiation, and engraftment of hematopoietic stem cells. Hypertension. 2016;67(3):574-584.10.1161/HYPERTENSIONAHA.115.06474\u003c/li\u003e\n\u003cli\u003eGaffney K, Weinberg M, Soto M, Louie S, Rodgers K. Development of angiotensin II (1-7) analog as an oral therapeutic for the treatment of chemotherapy-induced myelosuppression. Haematologica. 2018;103(12):e567-e570.10.3324/haematol.2018.193771\u003c/li\u003e\n\u003cli\u003eSato M, Iwanaga T, Mamada H, Ogihara T, Yabuuchi H, Maeda T, et al. Involvement of uric acid transporters in alteration of serum uric acid level by angiotensin II receptor blockers. Pharm Res. 2008;25(3):639-646.10.1007/s11095-007-9401-6\u003c/li\u003e\n\u003cli\u003eWatanabe T, Kusuhara H, Watanabe T, Debori Y, Maeda K, Kondo T, et al. Prediction of the overall renal tubular secretion and hepatic clearance of anionic drugs and a renal drug-drug interaction involving organic anion transporter 3 in humans by in vitro uptake experiments. Drug Metab Dispos. 2011;39(6):1031-1038.10.1124/dmd.110.036129\u003c/li\u003e\n\u003cli\u003eKurata T, Iwamoto T, Kawahara Y, Okuda M. Characteristics of pemetrexed transport by renal basolateral organic anion transporter hOAT3. Drug Metab Pharmacokinet. 2014;29(2):148-153.10.2133/dmpk.dmpk-13-rg-042\u003c/li\u003e\n\u003cli\u003eSweeney CJ, Takimoto CH, Latz JE, Baker SD, Murry DJ, Krull JH, et al. Two drug interaction studies evaluating the pharmacokinetics and toxicity of pemetrexed when coadministered with aspirin or Ibuprofen in patients with advanced cancer. Clin Cancer Res. 2006;12(2):536-542.10.1158/1078-0432.CCR-05-1834\u003c/li\u003e\n\u003cli\u003eGarassino MC, Gadgeel S, Speranza G, Felip E, Esteban E, Domine M, et al. Pembrolizumab plus pemetrexed and platinum in nonsquamous non-small-cell lung cancer: 5-year outcomes from the phase 3 KEYNOTE-189 study. J Clin Oncol. 2023;41(11):1992-1998.10.1200/JCO.22.01989\u003c/li\u003e\n\u003cli\u003ePatel JD, Socinski MA, Garon EB, Reynolds CH, Spigel DR, Olsen MR, et al. PointBreak: a randomized phase III study of pemetrexed plus carboplatin and bevacizumab followed by maintenance pemetrexed and bevacizumab versus paclitaxel plus carboplatin and bevacizumab followed by maintenance bevacizumab in patients with stage IIIB or IV nonsquamous non-small-cell lung cancer. J Clin Oncol. 2013;31(34):4349-4357.10.1200/JCO.2012.47.9626\u003cstrong\u003e\u003c/strong\u003e \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"cancer-chemotherapy-and-pharmacology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ccap","sideBox":"Learn more about [Cancer Chemotherapy and Pharmacology](http://link.springer.com/journal/280)","snPcode":"280","submissionUrl":"https://submission.nature.com/new-submission/280/3","title":"Cancer Chemotherapy and Pharmacology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Cancer chemotherapy, Neutropenia, Nonlinear mixed-effects model, Pharmacometrics, Physiologically based pharmacokinetic model","lastPublishedDoi":"10.21203/rs.3.rs-8989784/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8989784/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003ePurpose\u003c/h2\u003e \u003cp\u003ePemetrexed‒platinum chemotherapy is a key treatment option for non-squamous non-small cell lung cancer (NSCLC); however, its use is often limited by hematologic toxicity, particularly neutropenia. We aimed to model neutrophil dynamics following pemetrexed‒platinum chemotherapy using clinical data routinely available for patients and pharmacokinetic parameters of pemetrexed.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e \u003cp\u003eThis real-world investigation included 86 patients with NSCLC who received pemetrexed‒platinum chemotherapy for model development, and an additional 83 patients who received the same chemotherapy plus pembrolizumab or bevacizumab as a validation cohort. We developed a nonlinear mixed-effects (NLME) model to predict neutrophil dynamics during the first cycle following pemetrexed‒platinum chemotherapy, using patient-specific clinical data collected before chemotherapy initiation and pemetrexed pharmacokinetic parameters derived from physiologically based pharmacokinetic (PBPK) modeling.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e \u003cp\u003eThe final model suggested that the area under the curve (AUC)\u003csub\u003e0\u0026ndash;24\u003c/sub\u003e of pemetrexed, blood urea nitrogen value, and concomitant use of renin‒angiotensin system inhibitors influenced neutrophil suppression and delayed recovery. The goodness-of-fit plots, visual predictive check, nonparametric bootstrap analysis, and receiver operating characteristic (ROC) curves demonstrated good fit and high performance, with an area under the ROC curve (AUROC) of 0.92 for predicting severe neutropenia. The model maintained good performance in the validation cohort receiving antibody-containing regimens (AUROC 0.85).\u003c/p\u003e\u003ch2\u003eConclusion\u003c/h2\u003e \u003cp\u003eThis study demonstrated the importance of considering exposure to pemetrexed for an early prediction of neutropenia following pemetrexed‒platinum chemotherapy. Our model supports proactive pemetrexed dose optimization to prevent severe neutropenia, minimize treatment delays or interruptions, and preserve overall treatment intensity in patients undergoing pemetrexed‒platinum chemotherapy.\u003c/p\u003e","manuscriptTitle":"Prediction of Neutropenia Risk Based on Exposure Information Without Plasma Concentration Measurement in Pemetrexed‒Platinum‒Based Chemotherapy: A Modeling Approach Using Real-World Clinical Data","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-08 17:05:25","doi":"10.21203/rs.3.rs-8989784/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-20T06:16:05+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-19T09:22:20+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-30T15:12:56+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"270895703131605390228691315171657938423","date":"2026-03-18T07:04:30+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"339342522595223895215248103880769291256","date":"2026-03-03T07:51:39+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-03T06:19:03+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-02-28T07:41:28+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-02-28T07:39:48+00:00","index":"","fulltext":""},{"type":"submitted","content":"Cancer Chemotherapy and Pharmacology","date":"2026-02-27T15:26:05+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"cancer-chemotherapy-and-pharmacology","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"ccap","sideBox":"Learn more about [Cancer Chemotherapy and Pharmacology](http://link.springer.com/journal/280)","snPcode":"280","submissionUrl":"https://submission.nature.com/new-submission/280/3","title":"Cancer Chemotherapy and Pharmacology","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"fbe5d8bc-ab2d-483a-b317-7b0afaebfc2f","owner":[],"postedDate":"March 8th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-17T06:38:12+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-08 17:05:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8989784","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8989784","identity":"rs-8989784","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}