Evaluation of single-agent therapy in human colorectal tumour xenografts

Differential Effects of Trp53 Alterations in Murine Colorectal Cancer

BACKGROUND

Colorectal cancer (CRC) development is a multi-step process resulting in the accumulation of genetic alterations. Despite its high incidence, there are currently no mouse models that accurately recapitulate this process and mimic sporadic CRC. We aimed to develop and characterize a genetically engineered mouse model (GEMM) of Apc/Kras/Trp53 mutant CRC, the most frequent genetic subtype of CRC.

METHODS

Tumors were induced in mice with conditional mutations or knockouts in Apc, Kras, and Trp53 by a segmental adeno-cre viral infection, monitored via colonoscopy and characterized on multiple levels via immunohistochemistry and next-generation sequencing.

RESULTS

The model accurately recapitulates human colorectal carcinogenesis clinically, histologically and genetically. The Trp53 R172H hotspot mutation leads to significantly increased metastatic capacity. The effects of Trp53 alterations, as well as the response to treatment of this model, are similar to human CRC. Exome sequencing revealed spontaneous protein-modifying alterations in multiple CRC-related genes and oncogenic pathways, resulting in a genetic landscape resembling human CRC.

CONCLUSIONS

This model realistically mimics human CRC in many aspects, allows new insights into the role of TP53 in CRC, enables highly predictive preclinical studies and demonstrates the value of GEMMs in current translational cancer research and drug development.

A minimum core outcome dataset for the reporting of preclinical chemotherapeutic drug studies: Lessons learned from multiple discordant methodologies in the setting of colorectal cancer

In vivo studies in animal models are critical tools necessary to study the fundamental complexity of carcinogenesis. A constant strive to improve animal models in cancer exists, especially those investigating the use of chemotherapeutic effectiveness. In the present systematic review, colorectal cancer (CRC) is used as an example to highlight and critically evaluate the range of reporting strategies used when investigating chemotherapeutic agents in the preclinical setting. A systematic review examining the methodology and reporting of preclinical chemotherapeutic drug studies using CRC murine models was conducted. A total of 45 studies were included in this systematic review. The literature was found to be highly heterogeneous with various cell lines, animal strains, animal ages and chemotherapeutic compounds/regimens tested, proving difficult to compare outcomes between similar studies or indeed gain any significant insight into which chemotherapeutic regimen caused adverse events. From this analysis we propose a minimum core outcome dataset that could be regarded as a standardised way of reporting results from in vivo experimentation.

Advances in patient-derived tumor xenografts: from target identification to predicting clinical response rates in oncology

Most oncology compounds entering clinical development have passed stringent preclinical pharmacology evaluation criteria. However, only a small fraction of experimental agents induce meaningful antitumor activities in the clinic. Low predictability of conventional preclinical pharmacology models is frequently cited as a main reason for the unusually high clinical attrition rates of therapeutic compounds in oncology. Therefore, improvement in the predictive values of preclinical efficacy models for clinical outcome holds great promise to reduce the clinical attrition rates of experimental compounds. Recent reports suggest that pharmacology studies conducted with patient derived xenograft (PDX) tumors are more predictive for clinical outcome compared to conventional, cell line derived xenograft (CDX) models, in particular when therapeutic compounds were tested at clinically relevant doses (CRDs). Moreover, the study of the most malignant cell types within tumors, the tumor initiating cells (TICs), relies on the availability of preclinical models that mimic the lineage hierarchy of cells within tumors. PDX models were shown to more closely recapitulate the heterogeneity of patient tumors and maintain the molecular, genetic, and histological complexity of human tumors during early stages of sequential passaging in mice, rendering them ideal tools to study the responses of TICs, tumor- and stromal cells to therapeutic intervention. In this commentary, we review the progress made in the development of PDX models in key areas of oncology research, including target identification and validation, tumor indication search and the development of a biomarker hypothesis that can be tested in the clinic to identify patients that will benefit most from therapeutic intervention.

New criteria for analyzing the statistical relationships between biological parameters and therapeutic responses of xenografted tumor models

Analysis of preclinical studies using human tumors xenografted into rodents is commonly performed with Tumor Growth Index (TGI) and Tumor Growth Delay index (TGDi). To circumvent the limitations of these parameters, two new parameters, Time To Relapse (TTR) and Tumor Growth Speed (TGS), were developed using a mathematical modeling approach based on an exponential tumor growth. TTR is similar to progression free survival used in human clinical trials and TGS characterizes the pattern of tumor cell proliferation. Parameters were estimated for each rodent by the maximum likelihood method and statistical analyses were performed using ANOVA. These criteria can be used when tumor growths are assessed by repeated measures of their volume. As an example, we used data from a previously published study, which aimed to evaluate the relationship between histology, genetic parameters, and response to alkylating agents in a series of twelve gliomas. The group treated with temozolomide was reanalyzed using our criteria. This group presented a significantly longer TTR than the control group. TTR was also different according to tumor type: oligodendrogliomas relapsed later than glioblastomas. The TGS was different according to the tumor type. Loss of heterozygosity (LOH) 1p ± 19q, LOH 10p ± 10q, telomerase activity, PTEN mutation, and EGFR amplification were related to temozolomide efficacy. Our criteria provide additional information to those given by TGI and TGDi. Due to statistical properties of TTR and TGS, some relations between the parameters such as tumor type or genetic alterations can be studied with TTR and TGS and not with TGI or TGDi.

Effects of diarylsulfonylurea antitumor agents on the function of mitochondria isolated from rat liver and GC3/c1 cells

Diarylsulfonylureas, such as N-(4-chlorophenyl)aminocarbonyl-2,3-dihydro-1-indene-5-sulfonamide (LY186641, Sulofenur) and N-(4-chlorophenyl)aminocarbonyl-4-methylbenzene sulfonamide (LY181984), have been shown to be effective antitumor agents in a variety of in vivo and in vitro animal models. Their mechanism of action is unknown but does not appear to be the result of nonselective destruction of actively dividing cell populations. Mitochondria have been shown to accumulate Sulofenur and therefore may be targets of drug action. The purpose of these investigations was to examine the effects of a variety of diarylsulfonylureas in mitochondria and attempt to determine the relevance of these changes to antitumor activity. Many of the diarylsulfonylureas which were effective antitumor agents in animal models were also uncouplers of mitochondrial oxidative phosphorylation. They increased state 4 respiration and dissipated the mitochondrial membrane potential in a concentration-related fashion. The mechanism of uncoupling appeared to be related to a dissociable hydrogen ion as these molecules had pKa values that ranged from 6.0 to 6.2 and were highly lipophilic. Thus, the uncoupling action appears to be the result of hydrogen ion translocation. The mechanism of antitumor activity does not appear to be the result of uncoupling as no correlation was evident between inhibition of cell growth and uncoupling action of a variety of active and inactive diarylsulfonylureas. In vitro, Sulofenur is cytotoxic at high concentrations and inhibits cell growth at lower concentrations in the absence of any overt cell kill. The inhibition of cell growth also did not appear to be related to the uncoupling action of these drugs. In contrast, uncoupling may have played a partial role in the early, high exposure cell kill that can occur with these compounds.

Comparison of 5-FU versus FUDR activity in human colorectal cancer using an in vitro clonogenic assay (HTCA)

Comparative in vitro drug testing was performed in 72 of 183 surgically removed human colorectal cancer specimens (34 primary lesions, 38 metastases). In 10 of these tumors, comparative dose-response curves were obtained. Given a greater than or equal to 70% ICF (inhibition of colony formation) as threshold for in vitro sensitivity, 5-FU was active in 16/62 specimens, and FUDR in 14/62. Significantly discordant sensitivity results were observed in 8/62 tests, 5-FU being the more active agent in 5 of these cases. These data are supported by the finding of 3 considerably differing dose-response curves in 10 additional comparative studies of human primary tumors.

Predictive tests in cancer chemotherapy. A reappraisal

The evolution of medical oncology so far owes much to the preclinical and clinical development of antineoplastic agents. Prognostic factors and empiric treatment strategies have guided the clinician in his choice of drugs. In the light of increasing ethical restrictions met with phase I-II clinical trials and major advances in propagating human tumor cells outside the donor patient, a reappraisal of predictive tests in cancer chemotherapy is warranted. Among 'short-term assays' only the determination of steroid-hormone receptor content in tumor tissues has gained clinical acceptance, whereas other methods still suffer from theoretical or practical shortcomings. Both the human tumor stem cell assay and the xenograft model have revealed unique patterns of sensitivity for each individual tumor line. While interindividual heterogeneity among tumors sharing a common site of origin justifies efforts to develop predictive tests, microheterogeneity among tumor samples from the same donor patient limits the potential of this approach. Predictive tests should be performed in conjunction with clinical trials to ensure optimal extraction of information. As additional prognostic factors, they should in the near future accelerate drug development and reduce the hazard of unnecessary drug toxicity without therapeutic benefit.

Elucidation of pathways of 5-fluorouracil metabolism in xenografts of human colorectal adenocarcinoma

Hypoxanthine (Hx) and allopurinol (HPP) have been shown experimentally to reduce the conversion of 5-fluorouracil (FUra) to FUMP by orotate phosphoribosyltransferase (OPRTase). This study was designed to elucidate the major pathway by which FU ra was metabolized to ribonucleotides by human colorectal tumors. Consequently, the effect of Hx and HPP on the metabolism of [6-3H]-FUra was examined in 5 human colorectal adenocarcinomas maintained as xenografts in immune-deprived mice. In 2 tumors the formation of ribonucleotides from FUra was depressed by Hx and HPP in combination during the first hour after treatment, while in 3 other lines ribonucleotide concentrations were not reduced. The data suggested that these 5 xenograft lines may be divided into 2 groups: (1) group 1 tumors formed relatively high levels of FUrd and low levels of fluorinated ribonucleotides after the injection of FUra, with no decrease in ribonucleotide concentrations after the administration of Hx and HPP. These tumors possessed high ratios of uridine (Urd) phosphorylase/orotate phosphoribosyltransferase (OPRTase: 7-24) and ribose-1-phosphate (R-1-P)/5-phosphoribosyl-1-pyrophosphate (PRPP;5), and thus appeared to metabolize FUra by the U rd phosphorylase and U rd kinase pathway; (2) group 2 tumors formed low levels of FU rd, higher concentrations of fluorinated ribonucleotides and a reduction in levels of these nucleotides after administration of the purine combination. Group 2 tumors demonstrated a lower enzyme ratio (1-2), higher endogenous levels of PRPP, a lower R-1-P/PRPP ratio (1) and appeared to metabolize FUra predominantly by the activity of OPRTase. Hypoxanthine and HPP, alone or in combination, caused a rapid depletion of PRPP in each tumor line examined. In group 2 tumors this may be responsible for the decreased formation of FUra ribonucleotides observed.

The effect of derivatives of folic acid on the fluorodeoxyuridylate-thymidylate synthetase covalent complex in human colon xenografts

This study was designed to examine the endogenous concentrations of 5,10-methylenetetrahydrofolate (CH2FH4) in human colorectal adenocarcinoma xenografts, and to determine the ability of other folate derivatives to increase the formation of the ternary covalent complex between CH2FH4, [6-3H]-5-fluorodeoxyuridylate (FdUMP) and thymidylate synthetase (TS, EC 2.1.1.45). Levels of CH2FH4 were determined by measuring the release of [3H]2O from [5-3H]-dUMP using TS from Lactobacillus casei. The reaction was linear from 1.9 X 10(-13) to 2.4 X 10(-11) mol of CH2FH4 assayed. Concentrations of CH2FH4 were low, ranging from 66 to 233 nM in cell water. Tetrahydrofolate (FH4) and dihydrofolate (FH2) increased complex formation, while 5-formyltetrahydrofolate (5-CHOFH4) and 5-methyltetrahydrofolate (5-CH3FH4) decreased the covalent binding of [6-3H]-FdUMP in vitro. Administration of FH4 or FH2 to tumor-bearing mice reduced subsequent formation of the covalent complex in vitro. Since 5-CH3FH4 is a major derivative of folate in mammalian tissues, its effect on the covalent binding of [6-3H]-FdUMP was examined further; even in the presence of homocysteine and cyanocobalamin (B12), the formation of the covalent complex was not increased. The fate of [5-14CH3]-FH4 was subsequently examined in vivo. In tumors at 1 hr after injection, 72% of the radiolabel remained as [5-14CH3]-FH4, while 17% had been converted to [14C]-methionine or incorporated into protein. By contrast, however, the incorporation of radiolabel into the protein fraction of liver was almost 30-fold greater at this time. At 4 hr, radioactivity in tumors (dpm/g) and in the fraction associated with [5-14CH3]-FH4 was decreased by over 60%, while metabolism was increased by only 13%. No polyglutamate forms of [5-14CH3]-FH4 were detected in tumors at 4 hr after treatment.

Development of resistance to vincristine in a childhood rhabdomyosarcoma growing in immune-deprived mice

A cell line of childhood rhabdomyosarcoma (RD) has been grown as a xenograft in immune-deprived mice. The tumor responded to vincristine (VCR), but not to vinblastine, doxorubicin or actinomycin D. The rate and frequency at which resistance developed from administration of VCR once weekly was investigated. Tumor growth could be inhibited for 6 weeks, after which time 11 or 16 xenografts grew progressively despite continued treatment. That this was a tumor-acquired resistance was confirmed by growing both parent and "resistant" lines in the same host. Under these conditions VCR completely inhibited growth of the parent tumor, but not growth of the resistant for 6 weeks, after which time 11 or 16 xenografts grew progressively despite continued treatment. That this was a tumor-acquired resistance was confirmed by growing both parent and "resistant" lines in the same host. Under these conditions VCR completely inhibited growth of the parent tumor, but not growth of the resistant for 6 weeks, after which time 11 or 16 xenografts grew progressively despite continued treatment. That this was a tumor-acquired resistance was confirmed by growing both parent and "resistant" lines in the same host. Under these conditions VCR completely inhibited growth of the parent tumor, but not growth of the resistant line. Continued passage of the resistant line for 10 months in mice either treated with VCR (1.5 mg/kg/wk) or untreated, demonstrated that resistance to VCR was stable in the absence of selection pressure. In addition, the VCR-resistant line acquired a stable change in karyotype with the addition of a number 9 chromosome and an additional, unknown marker.

The human tumour xenograft--a valid model in experimental chemotherapy?

The chemotherapeutic response of a series of patients with bronchial carcinoma has been compared with the response of their xenografts established in immune-suppressed mice. The in situ endpoint of growth delay in subcutaneous tumours was the main parameter used to assess xenograft response, but clonogenic cell survival studies were also performed to assess the extent of cell kill associated with in vivo responses. Histology, chromosome analysis and demonstration of ectopic hormone production indicated that the xenografts retained human morphology and functional behaviour. Clinical response was assessed by serial measurement of metastases. The chemotherapeutic validity of xenografts is supported by this study in which 21 separate responses in 16 patients and their respective xenografts were similar. Metastases regressed completely in 5 out of 7 patients with oat cell carcinoma. Xenografts derived from these were also highly responsive to similar chemotherapy. In contrast, minimal responses were observed in xenografts established from two chemoresistant patients. One large cell anaplastic, 3 squamous and 4 adenocarcinomas were universally chemoresistant in patients and their xenografts. It is concluded that the bronchial carcinoma xenografts broadly retained human morphology and functional behaviour, and reproduced the pattern of chemotherapeutic response of their source tumours. The present work suggests that the incorporation of human tumour xenografts into drug development programmes is justified, although their usefulness in a predictive capacity to select appropriate chemotherapy for individual patients remains limited.

On the mechanism of cytotoxicity of fluorinated pyrimidines in four human colon adenocarcinoma xenografts maintained in immune-deprived mice

Three fluorinated pyrimidines, 5-fluorouracil (FUra), 5-fluorouridine (FUrd), and 5-fluoro-2'-dexoyuridine (FdUrd), have been studied in four human colonic tumor xenograft lines. The preliminary findings may be summarized as follows: 1) after equimolar dosages, the agent reaching the highest concentration in the tumor produced the highest level of fluorodeoxyuridylate (FdUMP); 2) within a tumor line, the order of response to the three agents is related to the order of FdUMP-generation; 3) the tumor-response did not correspond to the level of analogue-incorporation into RNA; 4) the measurement of levels of free FdUMP in tumors is a poor prognostic indicator of drug-response; and 5) the levels of FdUMP in the tumors are maintained for considerable periods and appear to be dependent upon the maintenance of the levels of either FUra or FUrd, irrespective of the parent agent administered.

In vitro chemosensitivity tests on xenografted human melanomas

An in vitro chemosensitivity test has been applied to malignant melanoma cells from 5 patients. The tumour cells were first grown as xenografts in immune-suppressed mice, so that the results of the in vitro test could be compared with precise measurements of the sensitivity of the melanoma cells when exposed to chemotherapeutic drugs in vivo in the mouse. The in vitro assay involved exposing the tumour cells to each of 8 drugs, after which cell survival was determined by colony assay in soft agar. Dose-response curves were obtained and the surviving fraction at drug levels estimated to be achieved in man was used as a measure of in vitro drug sensitivity. Significant differences among the 8 drugs were detected, and these accorded with clinical experience. The correlation of in vivo (in the mouse) and in vitro sensitivities to Melphalan and MeCCNU was also significant.

BCG treatment of human tumour xenografts in athymic nude mice

Xenografts of 3 human malignant cell lines in congenitally athymic nude mice have been examined for susceptibility to BCG. Growth of all 3 tumours, a bladder carcinoma, a melanoma and a colon carcinoma, was suppressed when cells were injected in admixture with BCG. Distant injection of BCG was ineffective. Mice with progressive growths had no detectable anti-human antibody, and rejection of cells and BCG failed to confer protection against subsequent tumour challenge. These studies indicate that human malignant cells are susceptible to local BCG-activated host responses, and that athymic mouse xenografts may be a useful model for assessing the response of human tumours to such agents.