Search

Neutropenia is the most common adverse event (AE) of palbociclib, an oral CDK4/6 inhibitor for breast cancer. Neutropenia increases the risk of infection and is even life threatening. Asian patients generally suffer more severe neutropenia from palbociclib treatment, but the label does not recommend a reduction in the starting dose for Asian patients. Therefore, the study aimed to explore the exposure-response (E-R) relationship in Chinese patients and preliminarily generate a scale for starting dose selection of palbociclib in Chinese patients. After comparing the kinetic-pharmacodynamic (K-PD) and the pharmacokinetic/pharmacodynamic (PK/PD) model, a semi-mechanistic K-PD model was selected and developed on the basis of real-world data from 28 patients with breast cancer to describe the time course of longitudinal absolute neutrophil counts (ANC). The longitudinal ANC data were well described by the population K-PD model with reasonable parameters: mean transit time (MTT) of 198 h, feedback parameter (γ) of 0.317, baseline ANC level (Circ

Pregnant women are often excluded from routine clinical trials. Consequently, appropriate dosing regimens for majority of drugs are unknown in this population, which may lead to unexpected safety issue or insufficient efficacy in this un-studied population. Establishing evidence through the conduct of clinical studies in pregnancy is still a challenge. In recent decades, physiologically-based pharmacokinetic (PBPK) modeling has proven to be useful to support dose selection under various clinical scenarios, such as renal and/or liver impairment, drug-drug interactions, and extrapolation from adult to children. By integrating gestational-dependent physiological characteristics and drug-specific information, PBPK models can be used to predict PK during pregnancy. Population pharmacokinetic (PopPK) modeling approach also could complement pregnancy clinical studies by its ability to analyze sparse sampling data. In the past five years, PBPK and PopPK approaches for pregnancy have made significant progress. We reviewed recent progress, challenges and potential solutions for the application of PBPK, PopPK, and exposure-response analysis in clinical drug development for pregnancy.

SI-B001 is a new EGFR/HER3 bi-specific antibody showing encouraging anti-tumor efficacy in the preclinical studies and was ready for further clinical research. To help with the dose design, human pharmacokinetics (PK) and clinical effective doses of SI-B001 were predicted by PK and PK/PD modeling and simulation. A Michaels-Menten (M-M) PK model was first used to describe the PK of SI-B001 in cynomolgus monkeys, whose parameters were allometrically scaled to humans for the simulation of human PK profiles. Besides, the anti-tumor efficacy of SI-B001 on different xenografts in tumor-bearing mice was quantitatively described by PK/PD models. The clinical effective doses were predicted by comparing the effective exposure (AUCs) in mice with simulated human AUCs. The clinical effective doses of SI-B001 were predicted to be over 16 mg/kg, 5-7 mg/kg or 5-6 mg/kg per week for colon cancer, head and neck cancer or esophageal cancer, respectively, which may help with the optimization of dose escalation schemes and the selection of indications for SI-B001.

Aim The aim of this study was to build and verify a preliminary physiologically based pharmacokinetic (PBPK) model of Chinese pregnant women. The model was used to predict maternal pharmacokinetics (PK) of 6 predominantly renally cleared drugs. Method Based on SimCYP Caucasian pregnancy population dataset, the preliminary Chinese pregnant population was built by updating several key parameters and equations according to physiological parameters of Chinese (or Japanese) pregnant women. Drug-specific parameters of 6 renally cleared drugs were validated through PBPK modeling of Caucasian non-pregnant, Caucasian pregnant and Chinese non-pregnant population. The preliminary PBPK model of Chinese pregnant population was then developed by integrating the preliminary Chinese pregnant population and the drug-specific parameters. This model was verified by comparing the predicted maternal PK of these 6 drugs with the observed in vivo data from the literature. Results The preliminary Chinese pregnant population PBPK model successfully predicted the PK of 6 target drugs for different pregnancy stages. The predicted plasma concentrations time profiles fitted the observed data well, and most predicted PK parameters were within 2-fold of observed data. Conclusions The preliminary Chinese pregnant population PBPK model provided a useful tool to predict the maternal PK of 6 predominantly renally cleared drugs in Chinese pregnant women.

Dexamethasone (DEX), a synthetic glucocorticoid, exhibited anti-cancer efficacy in pancreatic xenografts derived from patient tumor tissue or cancer cell lines. The aim of this study was to establish pharmacokinetic/pharmacodynamic (PK/PD) models to quantitatively characterize the inhibitory effect of DEX on tumor growth as well as its discrepancy among 3 xenograft models. Data of tumor growth profiles were collected from a patient-derived xenograft (PDX) model in NOD/SCID mice and 2 cell line-derived (PANC-1 and SW1990) xenograft models in BALB/c nude mice. Empirical PK/PD models were developed to establish mathematical relationships between plasma concentration of DEX and tumor growth dynamics after integrating PK parameters extracted from literature. Drug effect in each model was well described by a linear inhibitory function with a potency factor of 4.67, 3.14, and 2.35 L/mg for PDX, PANC-1, and SW1990 xenograft, respectively. Human efficacious doses of DEX were preliminarily predicted through model-based simulation, and 60% tumor growth inhibition at clinical exposure corresponded to a daily dose range of 26-52 mg intravenously. This modeling work quantified the preclinical anti-cancer effect of DEX and demonstrated the feasibility of its medication in pancreatic cancer, which would be conductive to future translational research.

Dopamine D1 receptor (D1DR) is proved to be a promising target to prevent tumor metastasis, and our previous studies showed that QAP14, a potent anti-cancer agent, exerted inhibitory effect on lung metastasis via D1DR activation. Therefore, the purpose of the study was to establish count data models to quantitatively characterize the disease progression of lung metastasis and assess the anti-metastatic effect of QAP14. Data of metastatic progression were collected in 4T1 tumor-bearing mice. Generalized Poisson distribution best described the variability of metastasis counts among the individuals. An empirical PK/PD model was developed to establish mathematical relationships between steady plasma concentrations of QAP14 and metastasis growth dynamics. The latency period of metastasis was estimated to be 12 days after tumor implantation. Our model structure also fitted well to other D1DR agonists (fenoldopam and l-stepholidine) which had inhibitory impact on breast cancer lung metastasis likewise. QAP14 40 mg/kg showed the best inhibitory efficacy, for it provided the longest prolongation of metastasis-free periods compared with fenoldopam or l-stepholidine. This study provides a quantitative method to describe the lung metastasis progression of 4T1 allografts, as well as an alternative PD model structure to evaluate anti-metastatic efficacy.

Breast cancer is the second leading cause of cancer-related mortality in women, mainly due to metastasis, which is strongly associated with cancer stemness. Our previous studies showed that the eradication of cancer stem-like cells (CSCs) may be related to the activation of dopamine D1 receptor (D1DR). This study aimed to explicitly demonstrate the target-role of D1DR activation in antimetastatic therapy and to investigate the potential efficacy and the underlying D1DR-related mechanisms of QAP14, a new oral compound. 4T1, MDA-MB-231, and D1DR-knockout 4T1 (4T1-D1DR) cells were selected for in vitro study, while 4T1 and 4T1-D1DR cells were further used to establish a mouse allograft model for in vivo study. Our results showed that D1DR is abundantly expressed in both 4T1 and MDA-MB-231 cells and that knocking out D1DR in 4T1 cells accelerated migration and invasion in vitro as well as lung metastasis in vivo. QAP14 inhibited colony formation, cell motility, mammosphere formation and CSC frequency, induced CSC apoptosis and D1DR expression, and increased cAMP/cGMP levels. Additionally, QAP14 showed inhibitory effects on tumor growth and lung metastasis with acceptable safety in vivo. Knocking out D1DR almost completely abolished the efficacy, confirming that QAP14 exhibits its anti-CSC and antimetastatic effects through D1DR activation. The underlying mechanisms involved suppression of the nuclear factor κB (NF-κB)/protein kinase B (Akt) pathway and consequent downregulation of both epithelial-to-mesenchymal transition (EMT) process and cancer stemness. In summary, our findings suggest a potential candidate compound, QAP14, as well as a potential target, D1DR, for metastatic breast cancer therapy.

The continuous speech separation (CSS) is a task to separate the speech sources from a long, partially overlapped recording, which involves a varying number of speakers. A straightforward extension of conventional utterance-level speech separation to the CSS task is to segment the long recording with a size-fixed window and process each window separately. Though effective, this extension fails to model the long dependency in speech and thus leads to sub-optimum performance. The recent proposed dual-path modeling could be a remedy to this problem, thanks to its capability in jointly modeling the cross-window dependency and the local-window processing. In this work, we further extend the dual-path modeling framework for CSS task. A transformer-based dual-path system is proposed, which integrates transform layers for global modeling. The proposed models are applied to LibriCSS, a real recorded multi-talk dataset, and consistent WER reduction can be observed in the ASR evaluation for separated speech. Also, a dual-path transformer equipped with convolutional layers is proposed. It significantly reduces the computation amount by 30% with better WER evaluation. Furthermore, the online processing dual-path models are investigated, which shows 10% relative WER reduction compared to the baseline., Accepted by ICASSP 2021

Modules in all existing speech separation networks can be categorized into single-input-multi-output (SIMO) modules and single-input-single-output (SISO) modules. SIMO modules generate more outputs than input, and SISO modules keep the numbers of input and output the same. While the majority of separation models only contain SIMO architectures, it has also been shown that certain two-stage separation systems integrated with a post-enhancement SISO module can improve the separation quality. Why performance improvements can be achieved by incorporating the SISO modules? Are SIMO modules always necessary? In this paper, we empirically examine those questions by designing models with varying configurations in the SIMO and SISO modules. We show that comparing with the standard SIMO-only design, a mixed SIMO-SISO design with a same model size is able to improve the separation performance especially under low-overlap conditions. We further validate the necessity of SIMO modules and show that SISO-only models are still able to perform separation without sacrificing the performance. The observations allow us to rethink the model design paradigm and present different views on how the separation is performed.

Overall survival (OS) has traditionally been the primary endpoint to evaluate drug efficiency in oncology but is often limited by the long observation period and high cost. The aims of this study were to perform a comprehensive meta-analysis at a clinical trial level to investigate the potential surrogate endpoints of OS in patients with castration-resistant prostate cancer (CRPC), and to predict OS based on the relationships associated with the potential surrogate endpoints. A systematic literature search was conducted in the PubMed database up to August 2018. Correlations between OS and potential surrogate endpoints were determined by linear regression analysis weighted by the square roots of sample size. Simulations were conducted to assess the effect of covariates on the relationships between OS and surrogate endpoints. A total of 233 studies including clinical trials and real-world data were included in our dataset. The correlations between median OS and potential surrogate endpoints for androgen-targeting therapy (R2 = 0.58–0.92) were generally stronger than those for taxane chemotherapy (R2 = 0.37–0.71). Median radiographic progression-free survival (rPFS) showed the strongest correlations with median OS (R2 = 0.94) in patients treated with novel androgen-targeting therapy. The meta-analysis demonstrated that rPFS might serve as a potential surrogate endpoint of OS and offer opportunity to facilitate the interim analyses and decision-making during the early stage of clinical trials for androgen-targeting agents.

Autism Spectrum Disorder (ASD), a neurodevelopmental disability, has become one of the high incidence diseases among children. Studies indicate that early diagnosis and intervention treatments help to achieve positive longitudinal outcomes. In this paper, we focus on the speech and language abnormalities of young children with ASD and present an automated assessment framework in quantifying atypical prosody and stereotyped idiosyncratic phrases related to ASD. For detecting atypical prosody from speech, we propose both the hand-crafted feature based method as well as the end-to-end deep learning framework. First, we use the OpenSMILE toolkit to extract utterance level high dimensional acoustic features followed by a support vector machine (SVM) backend as the conventional baseline. Second, we propose several end-to-end deep neural network setups and configurations to model the atypical prosody label directly from the constant Q transform spectrogram of speech. Third, we apply cross-validation on the training data to perform segments selection and enhance the subject level classification performance. Fourth, we fuse the deep learning based methods with the conventional baseline at the score level to further enhance the overall system performance. For detecting the stereotyped idiosyncratic usage of words or phrases from speech transcripts, we adopt language model, dependency treebank and Term Frequency–Inverse Document Frequency (TF–IDF) in addition to Linguistic Inquiry and Word Count software (LIWC) methods to extract a set of text features followed by a standard SVM backend. We collect a database of spontaneous Mandarin speech recorded during the Autism Diagnostic Observation Schedule (ADOS) Module 2 and Module 3 sessions. The Module 2 part consists of 118 children while the Module 3 part includes 71 children. Experimental results on this database show that our proposed methods can effectively predict the atypical prosody and stereotyped idiosyncratic phrases codes for young children with the risk of ASD. On the two categories classification task, the unweighted accuracy of the aforementioned two tasks are 88.1% and 77.8%, respectively.

Previous studies show that dopamine D2-like receptor (D2DR) antagonist sulpiride (SUL) enhances the antitumor efficacy of dexamethasone (DEX) in drug-resistant breast cancer involving cancer stem-like cells (CSCs). In this study, we investigated the pharmacokinetic (PK) properties of SUL in nude mice and developed a semi-mechanism PK/PD model to quantitatively characterize the synergistic effect of DEX and SUL in preclinical breast cancer xenografts. After nude mice received oral administration of a single dose of SUL (50 mg/kg, ig), plasma concentrations were assessed using LC-MS/MS. A two-compartment model with double first-order absorption rate was developed to describe the PK profiles of SUL. The pharmacodynamic (PD) study was conducted in nude mice bearing human breast cancer MCF-7/Adr xenografts, which received oral administration of DEX (1, 8 mg·kg−1·d−1) or SUL (25, 50 mg·kg−1·d−1) alone or in various combination. Tumor volumes were measured every other day. The PK model of SUL as well as that of DEX with a time-dependent clearance were integrated into the final PK/PD model both using Hill’s function, where DEX exerted its antitumor efficacy by inhibiting the proliferation of tumor cells, and SUL enhanced DEX responses by decreasing the sensitivity parameter EC50. The PK/PD model was evaluated and subjected external validation. Finally, simulations were performed to predict the antitumor efficacy of DEX combined with SUL under various dose regimens, where changing dosing frequency of SUL had little effect, while the antitumor efficacy was predicted to be improved when DEX was given more frequently. The established PK/PD model in this study quantitatively characterizes the antitumor efficacy of the DEX combined with SUL as well as their synergism, and the simulations could provide reference for dose optimization of the combination in future studies.

Continuous speech separation (CSS) is an arising task in speech separation aiming at separating overlap-free targets from a long, partially-overlapped recording. A straightforward extension of previously proposed sentence-level separation models to this task is to segment the long recording into fixed-length blocks and perform separation on them independently. However, such simple extension does not fully address the cross-block dependencies and the separation performance may not be satisfactory. In this paper, we focus on how the block-level separation performance can be improved by exploring methods to utilize the cross-block information. Based on the recently proposed dual-path RNN (DPRNN) architecture, we investigate how DPRNN can help the block-level separation by the interleaved intra- and inter-block modules. Experiment results show that DPRNN is able to significantly outperform the baseline block-level model in both offline and block-online configurations under certain settings.

This paper describes the Microsoft speaker diarization system for monaural multi-talker recordings in the wild, evaluated at the diarization track of the VoxCeleb Speaker Recognition Challenge(VoxSRC) 2020. We will first explain our system design to address issues in handling real multi-talker recordings. We then present the details of the components, which include Res2Net-based speaker embedding extractor, conformer-based continuous speech separation with leakage filtering, and a modified DOVER (short for Diarization Output Voting Error Reduction) method for system fusion. We evaluate the systems with the data set provided by VoxSRCchallenge 2020, which contains real-life multi-talker audio collected from YouTube. Our best system achieves 3.71% and 6.23% of the diarization error rate (DER) on development set and evaluation set, respectively, being ranked the 1st at the diarization track of the challenge., 5 pages, 3 figures, 2 tables

Deep CNN networks have shown great success in various tasks for text-independent speaker recognition. In this paper, we explore two approaches for modeling long temporal contexts to improve the performance of the ResNet networks. The first approach is simply integrating the utterance-level mean and variance normalization into the ResNet architecture. Secondly, we combine the BLSTM and ResNet into one unified architecture. The BLSTM layers model long range, supposedly phonetically aware, context information, which could facilitate the ResNet to learn the optimal attention weight and suppress the environmental variations. The BLSTM outputs are projected into multiple-channel feature maps and fed into the ResNet network. Experiments on the VoxCeleb1 and the internal MS-SV tasks show that with attentive pooling, the proposed approaches achieve up to 23-28% relative improvement in EER over a well-trained ResNet.

Aim M701 is a bispecific CD3/EpCAM T-cell engager antibody to treat malignant ascites. This study aimed to predict in vivo exposure-cytotoxicity relationship and human pharmacokinetics (PK) characteristics of M701, as well as to design optimal starting dose and effective dose for M701 first-in-human (FIH) study. Method Mechanistic in vitro PK/PD model was firstly developed based on in vitro data of M701’s cytotoxicity and binding affinities with targeting receptors. The cell killing effect of M701 in vitro was driven by tri-molecular synapse, which formed by binding drug to both CD3 receptor on T cells and EpCAM receptor on tumor cells. Human exposure-response (E-R) curve in ascites was estimated using the same model structure with clinical systemic model parameters. Human PK was predicted by allometrically scaling monkey PK data, which was characterized using a two compartment model. Human PK model was integrated into in vivo synapse-based cell killing model to provide human PK/PD characteristics. Integrated human PK/PD model was applied in FIH dose design. Clinical starting dose and effective dose were suggested as the simulated drug concentration in human ascites that achieved the estimated in vivo minimally anticipated biological effect level (MABEL) and pharmacologically active level. Other approaches including PK-driven and receptor occupancy calculation were also employed in this study to verify the starting dose prediction. Results In vitro M701 cytotoxicity curves under 24, 48, 72 h incubations were well captured by mechanistic synapse-based cell killing model. Human E-R curve in ascites was obtained based on in vitro model structure and clinical systematic parameters. We defined 10~20% and 80% of maximum cytotoxicity effect as in vivo MABEL and pharmacologically active level. Human E-R curve indicated in vivo EC10, EC20 and EC80 were 0.56, 1.26 and 31.6 ng/mL. For human PK model, clearance (CL, CLd), distribution volumes (Vc, Vp) and absorption rate were allometrically scaled using exponent of 0.9, 1 and -0.25. Predicted clearance and volume were 0.53- and 1.19-fold of observed data. Simulated average ascites M701 concentrations (calculated as Cave_ ascites = AUCτ/τ) were 0.81 and 32.5 ng/mL under dose of 5 and 200 μg within 2-hour i.p. infusion. By integrating human E-R curve and the simulated PK profile in ascites, we suggested 5 and 200 μg within 2-hour i.p. infusion as MABEL dose and pharmacologically active dose (PAD) for M701 FIH study. PK-driven approach predicted a starting dose of 5 μg, which was comparable to that predicted via PK/PD-driven approach. Conclusions This study predicted human ascites PK and E-R curve by integrating human PK model into in vivo synapse-based cell killing model. Optimal clinical MABEL dose and PAD of bispecific T cell engager antibody M701 were suggested based on current integrated PK/PD approach.

Glucocorticoid receptor (GR) modulates extensive biological and pathological processes including tumor progression through diverse mechanisms. The regulatory effects of dexamethasone (DEX), a synthetic glucocorticoid, as well as its interaction with GR have been recognized beyond hematologic cancers. In the present study, we investigated the anti-cancer efficacy of DEX and the correlation with GR in pancreatic cancer, a most aggressive malignancy threatening human health. The differential levels of GR expression were examined in two human pancreatic cancer cell lines, PANC-1 and SW1990, as well as in xenografts and patient tumor tissues. DEX significantly inhibited colony formation, migration, and tumor growth of PANC-1 cells expressing abundant GR. The underlying mechanisms involved suppression of nuclear factor κB (NF-κB) phosphorylation and down-regulation of epithelial-to-mesenchymal transition (EMT), interleukin 6 (IL-6) and vascular endothelial growth factor (VEGF). The anti-cancer effects of DEX were partially reversed by GR silencing or combinational administration of GR antagonist, RU486. The dose-dependent efficacy of DEX in tumor growth inhibition was also demonstrated in a GR-positive patient-derived xenograft model along with safety in mice. DEX was less potent, however, in SW1990 cells with poor GR expression. Our findings suggest that DEX effectively inhibits pancreatic tumor growth partially through GR activation. The potential correlation between GR expression and anti-cancer efficacy of DEX may have some clinical implications.

QAP14 is a novel anti-cancer compound exhibiting good efficacy and safety in preclinical study. To investigate its pharmacokinetic properties, a rapid and sensitive HPLC-MS/MS method was developed and validated to quantify the concentration of QAP14 in rat plasma. QAP14 was separated on ZORBAX Eclipse XDB-C8 column with a gradient elution. Erlotinib was selected as internal standard and plasma samples were prepared by precipitation with acetonitrile. In the pharmacokinetic study, rats were treated with QAP14 at 2.4 mg/kg (i.v.) or 6 mg/kg (p.o.). This method provided good linearity in the range of 0.5-1000 ng/mL in rat plasma. The accuracy, precision, matrix effect, recovery, stability and carryover fulfilled the criteria. The pharmacokinetic profiles of QAP14 in rats were described by non-compartmental analysis. The oral bioavailability was 50.29 %. The assay is reliable and requires only little sample volume, and the pharmacokinetic properties of QAP14 in rats may provide reference for future studies.

The ResNet-based architecture has been widely adopted to extract speaker embeddings for text-independent speaker verification systems. By introducing the residual connections to the CNN and standardizing the residual blocks, the ResNet structure is capable of training deep networks to achieve highly competitive recognition performance. However, when the input feature space becomes more complicated, simply increasing the depth and width of the ResNet network may not fully realize its performance potential. In this paper, we present two extensions of the ResNet architecture, ResNeXt and Res2Net, for speaker verification. Originally proposed for image recognition, the ResNeXt and Res2Net introduce two more dimensions, cardinality and scale, in addition to depth and width, to improve the model's representation capacity. By increasing the scale dimension, the Res2Net model can represent multi-scale features with various granularities, which particularly facilitates speaker verification for short utterances. We evaluate our proposed systems on three speaker verification tasks. Experiments on the VoxCeleb test set demonstrated that the ResNeXt and Res2Net can significantly outperform the conventional ResNet model. The Res2Net model achieved superior performance by reducing the EER by 18.5% relative. Experiments on the other two internal test sets of mismatched conditions further confirmed the generalization of the ResNeXt and Res2Net architectures against noisy environment and segment length variations., Comment: to be published in IEEE Spoken Language Technology Workshop(SLT) 2021

This paper proposes a new loss function called the “quartet” loss for the better optimization of the neural networks for matching tasks. For such tasks, where neural network embeddings are the key component, the optimization of the network for better embeddings is critical. The embeddings are required to be class discriminative, resulting in minimal inter-class variation and maximal intra-class variation even for unseen classes for better generalization of the network. The quartet loss explicitly computes the distance metric between pairs of inputs and increases the gap between the similarity score distributions between the same class pairs and the different class pairs. We evaluate on the speaker verification task and demonstrate the performance of the loss on our proposed neural network.

Text-independent speaker verification imposes no constraints on the spoken content and usually needs long observations to make reliable prediction. In this paper, we propose two speaker embedding approaches by integrating the phonetic information into the attention-based residual convolutional neural network (CNN). Phonetic features are extracted from the bottleneck layer of a pretrained acoustic model. In implicit phonetic attention (IPA), the phonetic features are projected by a transformation network into multi-channel feature maps, and then combined with the raw acoustic features as the input of the CNN network. In explicit phonetic attention (EPA), the phonetic features are directly connected to the attentive pooling layer through a separate 1-dim CNN to generate the attention weights. With the incorporation of spoken content and attention mechanism, the system can not only distill the speaker-discriminant frames but also actively normalize the phonetic variations. Multi-head attention and discriminative objectives are further studied to improve the system. Experiments on the VoxCeleb corpus show our proposed system could outperform the state-of-the-art by around 43% relative.

In this work, a semimechanistic pharmacokinetic/pharmacodynamic (PK/PD) model to quantitatively describe the antitumor activity of docetaxel (Doc) and cabozantinib (Cab) under monotherapy, concurrent therapy, interval therapy, and different sequential therapy in mouse xenograft models of castration-resistant prostate cancer was developed and evaluated. The pharmacokinetics (PK) of Doc and Cab when administered separately and simultaneously were investigated in nude mice, and PD study was conducted in tumor-bearing mice treated with different dosing schedules. The PK interaction between Doc and Cab was expressed by adding the effect of Cab on the clearance of Doc in the PK model. And the PD interaction between the two drugs was demonstrated by the developed PK/PD model through the combination index "φ" Our results showed that the concurrent therapy and Doc followed by Cab (Doc ∼ Cab) sequential therapy exhibited better tumor inhibitory efficacy than monotherapy. The Cab followed by Doc (Cab ∼ Doc) sequential schedule was less effective than monotherapy, and the interval therapy did not enhance the antitumor efficacy compared with the concurrent therapy. The parameter φ estimated from the PK/PD model quantitatively characterized the action between Doc and Cab. There was no significant PD interaction between Doc and Cab in both the concurrent schedule and the interval schedule, whereas the effect of the two drugs in the "Doc ∼ Cab" and "Cab ∼ Doc" sequential schedule was synergistic and antagonistic, respectively. The proposed model properly described the antitumor effects of Doc and Cab under different treatment schedules and could be used for dose optimization through model-based simulation.