Responsiveness of human gastric tumors implanted in nude mice to clinically equivalent doses of various antitumor agents

How translational modeling in oncology needs to get the mechanism just right

Translational model‐based approaches have played a role in increasing success in the development of novel anticancer treatments. However, despite this, significant translational uncertainty remains from animal models to patients. Optimization of dose and scheduling (regimen) of drugs to maximize the therapeutic utility (maximize efficacy while avoiding limiting toxicities) is still predominately driven by clinical investigations. Here, we argue that utilizing pragmatic mechanism‐based translational modeling of nonclinical data can further inform this optimization. Consequently, a prototype model is demonstrated that addresses the required fundamental mechanisms.

Opportunities for Quantitative Translational Modeling in Oncology

A 2-day meeting was held by members of the UK Quantitative Systems Pharmacology Network () in November 2018 on the topic of Translational Challenges in Oncology. Participants from a wide range of backgrounds were invited to discuss current and emerging modeling applications in nonclinical and clinical drug development, and to identify areas for improvement. This resulting perspective explores opportunities for impactful quantitative pharmacology approaches. Four key themes arose from the presentations and discussions that were held, leading to the following recommendations: Evaluate the predictivity and reproducibility of animal cancer models through precompetitive collaboration. Apply mechanism of action (MoA) based mechanistic models derived from nonclinical data to clinical trial data. Apply MoA reflective models across trial data sets to more robustly quantify the natural history of disease and response to differing interventions. Quantify more robustly the dose and concentration dependence of adverse events through mathematical modelling techniques and modified trial design.

Targeted sequencing reveals genetic variants associated with sensitivity of 79 human cancer xenografts to anticancer drugs

Although there has been progress moving from a 'one-size-fits-all' cytotoxic approach to personalized molecular medicine, the majority of patients with cancer receive chemotherapy using cytotoxic anticancer drugs. The sequencing analysis of 409 genes associated with cancer was conducted in the present study using 59 DNA sequences extracted from human cancer xenografts implanted into nude mice, of which sensitivity to 9 cytotoxic anticancer drugs [5-fluorouracil, nimustine, adriamycin, cyclophosphamide, cisplatin, mitomycin C (MMC), methotrexate, vincristine (VCR), and vinblastine] was examined. The present study investigated the association between the sensitivities of the xenografts to the 9 anticancer drugs and the frequency of single nucleotide variants (SNV). The correlation between the expression level of the genes and sensitivities to the 9 drugs in the above xenografts was also estimated. In the screening study using 59 xenografts, 3 SNVs (rs1805321, rs62456182 in PMS1 Homolog 2, Mismatch Repair System Component and rs13382825 in LDL Receptor Related Protein 1B), were associated with sensitivity to VCR and MMC, respectively (P<0.001). A replication study of 596 SNVs was subsequently performed, which indicated P<0.05 in the screening study using independent samples of 20 xenografts. A combined result of the screening and replication studies indicated that 35 SNVs were potentially associated with sensitivities to one or more of the nine anticancer drugs (P_combined=0.0011-0.035). Of the 35 SNVs, rs16903989 and rs201432181 in Leukemia Inhibitory Factor Receptor α and Adhesion G Protein-Coupled Receptor A2 were commonly associated with sensitivity to 2 or 4 anticancer drugs, respectively. These findings provide novel insights which may benefit the development of personalized anticancer therapy for patients with cancer in the future.

Comparison of hypoxia-activated prodrug evofosfamide (TH-302) and ifosfamide in preclinical non-small cell lung cancer models

Evofosfamide (TH-302) is a hypoxia-activated prodrug of the cytotoxin bromo-isophosphoramide. In hypoxic conditions Br-IPM is released and alkylates DNA. Ifosfamide is a chloro-isophosphoramide prodrug activated by hepatic Cytochrome P450 enzymes. Both compounds are used for the treatment of cancer. Ifosfamide has been approved by the FDA while evofosfamide is currently in the late stage of clinical development. The purpose of this study is to compare efficacy and safety profile of evofosfamide and ifosfamide in preclinical non-small cell lung cancer H460 xenograft models. Immunocompetent CD-1 mice and H460 tumor-bearing immunocompromised nude mice were used to investigate the safety profile. The efficacy of evofosfamide or ifosfamide, alone, and in combination with docetaxel or sunitinib was compared in ectopic and intrapleural othortopic H460 xenograft models in animals exposed to ambient air or different oxygen concentration breathing conditions. At an equal body weight loss level, evofosfamide showed greater or comparable efficacy in both ectopic and orthotopic H460 xenograft models. Evofosfamide, but not ifosfamide, exhibited controlled oxygen concentration breathing condition-dependent antitumor activity. However, at an equal body weight loss level, ifosfamide yielded severe hematologic toxicity when compared to evofosfamide, both in monotherapy and in combination with docetaxel. At an equal hematoxicity level, evofosfamide showed superior antitumor activity. These results indicate that evofosfamide shows superior or comparable efficacy and a favorable safety profile when compared to ifosfamide in preclinical human lung carcinoma models. This finding is consistent with multiple clinical trials of evofosfamide as a single agent, or in combination therapy, which demonstrated both anti-tumor activity and safety profile without severe myelosuppression.

On the development of models in mice of advanced visceral metastatic disease for anti-cancer drug testing

It is well known clinically that advanced, bulky visceral metastatic disease is generally much less responsive to most anti-cancer therapies, compared to microscopic metastatic disease. This problem is exacerbated when treating cancers that have been previously exposed to multiple lines of therapy, and which have acquired a 'refractory' phenotype. However, mimicking such clinical treatment situations in preclinical mouse models involving the testing of new or existing cancer therapies is extremely rare. Treatment of 'metastasis', in retrospect, usually involves minimal residual disease and therapy naïve tumors. This could account in many instances for the failure to reproduce highly encouraging preclinical results in subsequent phase I or phase II clinical trials. To that end, we have embarked on an experimental program designed to develop models of advanced, visceral metastatic disease, in some cases involving tumors previously exposed to various therapies. The strategy first involves the orthotopic transplantation of a human cancer cell line, such as breast cancer cell line, into the mammary fat pads of immune deficient mice, followed by surgical resection of the resultant primary tumors that develops. Recovery of distant macroscopic metastases, usually in the lungs, is then undertaken, which can take up to 4 months to visibly form. Cell lines are established from such metastases and the process of orthotopic transplantation, surgical resection, and recovery of distant metastases is undertaken, at least one more time. Using such an approach highly metastatically aggressive variant sublines can be obtained, provided they are once again injected into an orthotopic site and the primary tumors removed by surgery. By waiting sufficient time after removal of the primary tumors, about only 1 month, mice with extensive metastatic disease in sites such as the lungs, liver, and lymph nodes can be obtained. An example of therapy being initiated in an advanced stage of such disease development is illustrated. Metastases that eventually stop responding to a particular therapy can be removed as a source of variant cell lines which have both 'refractory' and highly metastatic phenotypes. Such models may provide a more accurate picture of the potential responsiveness to an experimental therapy so that a high degree of responsiveness observed could be a factor in deciding whether to move a particular therapy forward into phase I/phase II clinical trial evaluation. An example of this is illustrated using doublet metronomic low-dose chemotherapy for the treatment of advanced metastatic breast cancer, using two conventional chemotherapy drugs, namely, cyclophosphamide and UFT, a 5-FU oral prodrug.

Practices and Pitfalls of Mouse Cancer Models in Drug Discovery

Mouse models of cancer are critical tools for elucidating mechanisms of cancer development, as well as for assessment of putative cancer therapies. However, there are ongoing concerns about the value of mouse cancer models for predicting therapeutic efficacy in humans. This chapter reviews the most commonly used transplanted tumor models, including subcutaneous and orthotopic tumors in mice. It also reviews commonly utilized in vivo study endpoints. Even small improvements in predictive value achieved through careful selection of models and endpoints have the potential to have large impacts on productivity and overall drug development costs.

Marker genes to predict sensitivity to FK228, a histone deacetylase inhibitor

In this study, we detected genes sensitive to an histone deacetylase inhibitor, FK228 [(E)-(1S,4S,10S,21R)-7-[(Z)-ethylidene]-4,21-diisopropyl-2-oxa-12,13-dithia-5,8,20,23-tetraazabicyclo-[8,7,6]-tricos-16-ene-3,6,9,19,22-pentanone; FR901228, depsipeptide] in vitro and identified marker genes to predict sensitivity to FK228 in vivo using Affymetrix GeneChip. Three percent of genes (205/7070) were sensitive to FK228 in vitro, 105 and 100 genes, were up- and down-regulated, respectively, by FK228. Commonly up-regulated genes included p21(WAF1/Cip1), interleukin-8 (IL-8), histone family, JunB, caspase 9, mitogen-activated protein kinase phosphatase 1 (MKP-1) and mitogen-activated protein kinase (MAPK) family, and commonly down-regulated genes included cyclin A and MAPK family. One percent of genes (76/7070) showed native differences in patterns of expression, when FK228-sensitive (PC-3 prostate and SC-6-JCK (SC-6) stomach) and FK228-resistant (ACHN and A-498 renal) tumors implanted in BALB/c nu/nu mice were compared. Twenty-seven and forty nine of those genes were expressed at high or low levels, respectively, in FK228-sensitive tumors. Caspase 9 and MKP-1 genes showed distinct differences in patterns of expression between FK228-sensitive and resistant tumors and have been known to have roles in apoptosis and chromatin remodeling. The expression of caspase 9 gene was higher in FK228-sensitive tumors and the expression of MKP-1 gene was higher in FK228-resistant tumors. Caspase 9 and MKP-1 genes in the other FK228-sensitive tumors had the same patterns of expression as they did in PC-3 and SC-6 tumors. Our results present profiles of gene expression related to FK228 and marker genes to predict sensitivity to FK228, such as caspase 9 and MKP-1 genes.

Experimental chemotherapy against canine mammary cancer xenograft in SCID mice and its prediction of clinical effect

The effectiveness of 6 antitumor agents has been evaluated for canine mammary gland tumor (CMG-6) serially transplanted into severe combined immunodeficiency (SCID) mice. CMG-6 diagnosed as a solid carcinoma was subcutaneously transplanted into SCID mice and six antitumor agents were intravenously given to the mice as a single injection. The effectiveness was evaluated by Treatment group/Control group percent (T/C %) and statistical significance determined by Mann-Whitney's U-test in tumor volume. The minimum effective doses (MEDs; mg/kg) of mice were as follows; cyclophosphamide (CPM) 65, doxorubicin (DXR) 6, cisplatin (CDDP) 5, vincristine (VCR) 1.6, vinblastine (VLB) more than 5.5, 5-fluorouracil (5-FU) 105. Clinical effects of the drugs were predicted based on area under the curve (AUC) of dogs given a clinical dose (AUCdog)/AUC of mice given a MED (AUCmouse) ratios from published references. The AUC ratios were as follows; CPM 2.24, DXR 0.19, CDDP 1.20, VCR 0.04, VLB <1.24 and 5-FU 1.15. Drugs indicating more than 1.0 in AUCdog/AUCmouse ratio were CPM, CDDP and 5-FU, and would be suggested as effective in the original patient with CMG-6. The combination chemotherapy using clinically equivalent doses in CDDP and CPM, which were the two highest values in AUCdog/AUCmouse ratio by single agent therapy, was performed and shown to have additional effects as compared to the responsiveness of each agent against CMG-6.

A Humanized Anti-c-erbB-2 Monoclonal Antibody for the Treatment of Breast Cancer

The c-erbB-2 product is thought to be a unique and useful target for antibody therapy of cancers that overxpress the c-erbB-2 gene. Its overexpression is also speculated to be correlated with chemoresistance to doxorubicin. The in vitro and in vivo anti-tumor effects of a humanized antibody directed against the extracellular domain of the c-erbB-2 gene product, rhu4D5, were examined. Rhu4D5 had direct antiproliferative activity against the SK-BR-3 cell line which overexpresses c-erbB-2. The in vivo treatment, using rhu4D5, of SCID mice carrying xenografts of 4-1ST human gastric carcinoma, which overexpresses c-erbB-2, revealed that the recombinant protein had potent anti-tumor activity. Furthermore, the cytotoxic action of human peripheral blood mononuclear cells against the SK-BR-3 cell line was significantly augmented with the administration of rhu4D5, but not with mu4D5. These results indicate that rhu4D5 might be a more efficacious treatment than previously predicted by preclinical studies.

In vitro and in vivo anti-tumour effects of a humanised monoclonal antibody against c-erbB-2 product

The c-erbB-2 product is thought to be a unique and useful target for antibody therapy of cancers overexpressing the c-erbB-2 gene. In vitro and in vivo anti-tumour effects of a humanised antibody against the extracellular domain of the c-erbB-2 gene product, rhu4D5, were examined. Rhu4D5 was less effective than its murine counterpart, mu4D5, for the direct antiproliferative activity against the c-erbB-2-overexpressing SK-BR-3 cell line. In vivo treatment of severe combined immunodeficient (SCID) mice carrying the c-erbB-2-overexpressing 4-1ST human gastric carcinoma xenograft with 4hu4D5 revealed that the recombinant protein had potent anti-tumour activity. Furthermore, cytotoxicity of human peripheral blood mononuclear cells against 4-1ST was significantly augmented with rhu4D5, but not with mu4D5. These results indicate that rhu4D5 might perform better in patients than predicted from preclinical studies.

Prolonged survival of mice with human gastric cancer treated with an anti-c-ErbB-2 monoclonal antibody

A monoclonal antibody (MAb), 4D5, specifically recognising an extracellular epitope of the c-ErbB-2 protein, inhibited the growth of human gastric cancer overexpressing c-ErbB-2 severe combined immunodeficient (SCID) mice. This antibody also reduced the mass of established tumours xenografted into SCID mice, whereas gastric cancer not expressing c-ErbB-2 exhibited no regression in response to 4D5 treatment. In addition, administration of 4D5 prevented colonisation of cancer cells and prolonged the survival of host SCID mice inoculated i.v. with c-ErbB-2-overexpressing tumour cells. This is the first reported study to show that treatment with a single antibody specific to c-ErbB-2 prolongs the survival of host SCID mice bearing xenotransplanted tumours.

Rhabdomyosarcoma. From the laboratory to the clinic

The prospect of identifying and developing new agents for treatment of rhabdomyosarcomas is discussed in the light of current prognosis for children with advanced stage disease. Preliminary attempts to identify tumor-specific agents using in vitro cell culture show potential promise, but as yet remain unproven. The more complex system of identifying therapeutically active agents using human tumor xenografts has demonstrated usefulness. The potential problems associated with this system are discussed.

Combination therapy of ACNU and ifosfamide in tumor bearing mice with M2661 breast cancer, B16 malignant melanoma or C38 colon cancer

Three murine tumor lines, B16 melanoma, C38 colon cancer and M2661 breast cancer, were used to evaluate the therapeutic efficacy of the drug combination ACNU and ifosfamide. For each tumor type five treatment groups of 12 mice each were studied, which respectively received 16.5 mg/kg ACNU, 150 mg/kg ifosfamide, 33 mg/kg ACNU, 300 mg/kg ifosfamide or 16.5 mg/kg ACNU + 150 mg/kg ifosfamide. One group served as controls. Growth delay was measured as the endpoint. In the B16 and C38 tumor lines both drugs were active and showed additive antitumor effects. However, no synergism was observed. Neither ACNU or ifosfamide or the combination of both had any activity against the M2661 tumor line.

Prediction of ACNU plasma concentration-time profiles in humans by animal scale-up

Plasma concentration-time profiles of nimustine hydrochloride, 1-[(4-amino-2-methyl-5-pyrimidinyl)methyl]-3-(2-chloroethyl)-3-nitrosour ea hydrochloride (ACNU), in the mouse, rat, rabbit, and dog were determined by high-performance liquid chromatographic analysis. The pharmacokinetic parameters for these four animal species and previously reported clinical data were analyzed for investigation of interspecies correlation. Log-log plots of body weight (W; kg) vs total plasma clearance (CLtot,p; ml/min) and steady-state distribution volume (Vd,ss; l) for the four animal species were linear, with high correlation coefficients (r 0.996 for both parameters), despite the fact that the nonrenal clearance was greater than 97% in these species. Linear regression on the plots excluding human data yielded allometric equations (CLtot,p = 50.6 W0.957; Vd, ss = 1.29 W1.03) that were extrapolated to predict ACNU pharmacokinetic parameters in humans. For both parameters, however, there were 3-fold differences between the predicted and observed parametric values. To investigate these discrepancies, we measured serum protein binding of ACNU in these animal species and in humans. The values of CLtot,p and Vd,ss were converted into those of CLutot,p and Vd,uss, which correspond to the parameters for unbound ACNU. In this case, correlation coefficients of the log-log plots excluding human data (CLutot,p = 71.7 W0.891; Vd,uss = 1.82 W0.966) were also high (r greater than or equal to 0.991). The extrapolated values vs those observed in a 70-kg human were the following: CLutot,p, 3,160 vs 2,290 ml/min; Vd,uss, 110 vs 106 l. Thus, the animal data were successfully extrapolated to yield better predictions of human pharmacokinetic parameters if the analysis was based on the unbound plasma concentration of ACNU. In addition, the predicted plasma concentration-time profile for humans also showed good agreement with the observed ones. These results suggest the importance of measuring unbound fractions of drugs for more accurate prediction of human pharmacokinetic parameters by extrapolation of animal data to the human situation.

Antitumor activity of a derivative of mitomycin, 7-N-[2-[[2-(gamma-L-glutamylamino)ethyl]dithio]ethyl]mitomycin C (KW-2149), against murine and human tumors and a mitomycin C-resistant tumor in vitro and in vivo

The antitumor activity of a mitomycin derivative, 7-N-[2[[2-(gamma-glutamylamino)ethyl]dithio]ethyl]mitomycin C (KW-2149), was evaluated in murine and human tumor models, including a mitomycin C (MMC)-resistant tumor in vitro and in vivo. KW-2149 showed a profound effect against i.p. inoculated P388 leukemia on both a single and an intermittent administration schedule. Against s.c. implanted colon adenocarcinoma 38 (colon 38). KW-2149 was as effective as MMC in ILS% and in tumor growth inhibition on a single-administration schedule. Both compounds were similarly effective when an intermittent schedule was used. KW-2149 showed activity against human tumor xenografts and was effective in two of four non-small-cell lung carcinomas but was not effective against three gastric adenocarcinomas on the single-administration regimen. The activity of KW-2149 against gastric adenocarcinoma was inferior to that of MMC on a single-administration schedule. However, the antitumor activity of KW-2149 was higher on an intermittent schedule than on a single-administration regimen. The antitumor activity of KW-2149 against human tumor xenografts was similar to that of MMC on an intermittent schedule, and the former drug was effective against both gastric adenocarcinomas and both non-small-cell lung carcinomas. KW-2149 was more effective than MMC against a subline of P388 leukemia that is resistant to MMC in vitro as well as in vivo.

Evaluation of antitumor activity in a human breast tumor/nude mouse model with a special emphasis on treatment dose

Eight lines of human breast tumors implanted in nude mice were treated with various antitumor agents at two different doses, maximum tolerated doses (MTD) and rational doses (RD) that were pharmacokinetically equivalent to the clinical doses; the response rates to both doses were compared. With MTD, the response rates to mitomycin C and vinblastine were 100%, and those to other agents including cyclophosphamide, nimustine (a water-soluble nitrosourea), vincristine, Adriamycin (doxorubicin; Adria Laboratories, Columbus, OH), 5-fluorouracil (5-FU), and methotrexate were 30%-50%, indicating high responsiveness to the former two agents. In contrast, when the RD were used, the response rates to the majority of these agents were 25%-40%, and those to vincristine and nimustine were 13% and 0%, respectively. These results agree with the reported clinical results compared with those with MTD, suggesting the importance of the use of clinically equivalent doses in the evaluation of antitumor efficacy in a human tumor/nude mouse system.

Responsiveness of human lung cancer/nude mouse to antitumor agents in a model using clinically equivalent doses

The responses of 14 lines of human lung cancer xenografts in BALB/c-nu/nu mice to eight known antitumor agents were investigated. These xenografts consisted of four small-cell carcinomas (SCLC) and ten non-small-cell carcinomas (four large cell, three squamous cell, and three adenocarcinomas; NSCLC). The doses used in the experiments were the maximum tolerated dose (MTD) in nude mice and the "rational dose" (RD), the latter considered to be pharmacokinetically equivalent to the clinical dose. When given at MTDs, all drugs except 5-fluorouracil (5-FU) and methotrexate (MTX) were extremely effective against NSCLC as well as SCLC. The response rates of drug-sensitive SCLC to mitomycin C (MMC), ACNU, and vinblastine (VLB) were 100%, and those to Adriamycin (ADR) and vincristine (VCR) were 75%. In addition, the response rates of even drug-resistant NSCLC to MMC and VLB were 70% and 90%, respectively. In contrast, the response rates of NSCLC to RDs of the drugs were reduced to less than 40% and corresponded well to the respective clinical rates. In SCLC, a good correlation of experimental and clinical response rates was observed with four drugs [cyclophosphamide (CPM), ACNU, VLB, and 5-FU]. As a result, we emphasize that a more reasonable prediction of the clinical effectiveness of antitumor agents can be made by a protocol using clinically equivalent doses.