- Top of page
- Materials and Methods
- Disclosure Statement
Most of the cancer xenograft models are derived from tumor cell lines, but they do not sufficiently represent clinical cancer characteristics. Our objective was to develop xenograft models of bladder cancer derived from human tumor tissue and characterize them molecularly as well as histologically. A total of 65 bladder cancer tissues were transplanted to immunodeficient mice. Passagable six cases with clinico-pathologically heterogeneous bladder cancer were selected and their tumor tissues were collected (012T, 025T, 033T, 043T, 048T, and 052T). Xenografts were removed and processed for the following analyses: (i) histologic examination, (ii) short tandem repeat (STR) genotyping, (iii) mutational analysis, and (iv) array-based comparative genomic hybridization (array-CGH). The original tumor tissues (P 0) and xenografts of passage 2 or higher (≥P2) were analyzed and compared. As a result, hematoxylin and eosin staining revealed the same histologic architecture and degree of differentiation in the primary and xenograft tumors in all six cases. Xenograft models 043T_P2 and 048T_P2 had completely identical STR profiles to the original samples for all STR loci. The other models had nearly identical STR profiles. On mutational analysis, four out of six xenografts had mutations identical to the original samples for TP53, HRAS, BRAF, and CTNNB1. Array-CGH analysis revealed that all six xenograft models had genomic alterations similar to the original tumor samples. In conclusion, our xenograft bladder cancer model derived from patient tumor tissue is expected to be useful for studying the heterogeneity of the tumor populations in bladder cancer and for evaluating new treatments.
Bladder cancer (BCa) is also the second most common genitourinary malignancy in the United States, with an expected 69 000 newly diagnosed cases in 2008, and 14 000 deaths, and it also is the second most common urologic cancer in South Korea. Bladder cancer is a heterogeneous disease, with 70% of patients presenting with superficial cancers that tend to recur but are generally not life-threatening, and 30% presenting with muscle-invasive disease associated with a high risk of death from distant metastases.
Animal models of BCa allow for the investigation of aspects of BCa that cannot be studied clinically, such as evaluating new chemotherapeutic agents or other treatments as well as investigating the basic mechanisms of tumor biology. Different animal models of preclinical BCa have been developed. Most of these models consist of chemically induced BCa,[3-5] implanting xenografts generated from well-established human BCa cell lines that have adapted to in vitro growth into immunodeficient mice,[6-9] or transplanting carcinogen-induced BCa in syngeneic, immunocompetent mice.[10-12] Although such models have been useful, they also have limitations. The chemically induced BCa model takes at least 8–11 months to develop tumors. Models generated by planting human tumor cell lines subcutaneously in immunodeficient mice do not sufficiently represent clinical cancer characteristics, especially with regard to metastasis and drug sensitivity. The cell lines are mainly undifferentiated, have only a minor relation to the tissue of origin, and show less physiological conditions in their micro-environment.
Recently, a few groups have developed more relevant models based on xenografting primary human tumor tissue in immunodeficient mice, including non-small-cell lung, gastric, ovarian, and colon cancers.[14-18] These patient-derived tumor tissue xenograft models retain morphology, architecture, and molecular signatures similar to the original cancers.
Therefore, the aim of our study was to develop and characterize a xenograft model of BCa using patient-derived tumor tissues with heterogeneous clinico-pathological features, which would be useful for investigating BCa biology and developing new anticancer therapeutics.
- Top of page
- Materials and Methods
- Disclosure Statement
Xenograft models increasingly use patient-derived tumor tissues implanted subcutaneously in immunodeficient mice to reflect the biological characteristics of human tumors more accurately than tumor cell lines. Fichtner and colleagues mentioned that cell lines used for preclinical studies as xenografts mainly represented poorly differentiated carcinomas, which lacked similarity to the original tumor and therefore the clinical situation. There are several inherent problems with cell line xenograft models. To survive, tumor cells in culture adjust to a microenvironment devoid of stromal and endothelial elements. This adjustment likely occurs gradually over serial passages and involves genetic changes that make these tumor cells differ from the original tumor. In addition, cell lines do not reflect tumor heterogeneity, as xenografts derived from a cell line typically have homogenous histologies lacking architectural organization. A study by Johnson et al. using 39 anticancer drugs reported that, with the exception of lung, histological matches were not found between cell line xenograft models and clinical response. Moreover, they also showed that the relationship between chemo-response levels of various cancer cell line xenografts and those in the clinical trials was limited. On the other hand, studies of colon cancer and sarcoma have shown that the drug responsiveness between the early-passage tissue xenografts and the original cancers was closely correlated.[24, 27] Our study definitely showed that early passages (P2) of xenografts almost preserved the characteristics of the original human cancer tissues, especially with referring the array-CGH results. Therefore, although there has been no study that directly compared between the tissue and cell line xenografts, it is suggested that tissue-based xenografts may be more representative of the original tumor than the cell line xenografts. Further comparative analysis is required to clarify this issue.
Historically, we are not the first group to directly implant a patient-derived BCa tissue into mice. A few studies have developed BCa xenograft models by implanting patient-derived tumor tissues, some even more than 30 years ago.[28-31] Successful passage rates were 9–15%, which are similar to those of our study (15.4%). Although these studies successfully developed passagable BCa xenograft models, the characterization was limited to phenotypic analyses, such as H&E or immunocytochemical staining, without detailed molecular analyses. Although histological analyses are the first step in assuring that xenografts mimic the original tumor, they are not sufficient, as molecular changes could occur in the absence of histological changes. To investigate the molecular alterations in xenografts, further genomic analyses should be performed. Short tandem repeat genotyping, mutational analysis, and array-CGH are three well-established genomic analyses used in many other studies. Thus, to our knowledge, this is the first study to develop a BCa xenograft model from patient-derived tumor tissue and characterize it using these three genomic analyses as well as conventional histologic analysis.
In the present study, we developed six BCa xenograft models with heterogeneous clinico-pathological features that can be routinely passaged. The consistency of the histological patterns between the original tumors and serial passages of the xenografts support the validity of this model. Four of the six xenograft models had a mutation in HRAS, CTNNB1, BRAF, or TP53, with none having mutations in other genes. These results are possibly related to the mutation frequency of each gene. The frequency of TP53 mutation is over 40% in human BCa, whereas that of the RAS family is 13% and PIK3CA is 13 to 27%. One unique finding in our study is the BRAF mutation, which is infrequent in human BCa. Finally, our STR genotyping and array-CGH results confirmed that our models accurately represented their respective donors, because the original samples and xenografts had almost identical results.
Two of the most significant reasons to establish animal models of human cancer are to evaluate new chemo- or other therapeutic agents or treatments, and to investigate the basic mechanisms of tumor biology. Using patient-derived tumor tissue xenograft models for these two purposes assumes that xenografts closely resemble the original tumor. In our study, we demonstrated that early passages of BCa xenograft models are highly similar to the original cancer with regard to histology, mutation status, and genomic alterations.
Our xenograft models not only retain the histopathological features and molecular signatures of the original tumors, but also show the clinico-pathological heterogeneity of BCa. Therefore, our models are pertinent to investigate an anticancer drug response or resistance, or to evaluate the biology of BCa expressing specific clinico-pathological feature. In fact, an experiment to establish a cisplatin-resistant BCa model using our xenografts is ongoing.
Our study has a few limitations. First, although our xenograft models almost preserved phenotypic and genotypic characteristics of the primary human BCa tissues, it has not been proven whether they show the similarity in ‘clinical’ aspects, for example, drug or radiation sensitivity. Further clinical relevance study using our xenografts is necessary. Second, since our models lacked grade I urothelial carcinoma (Table 3), they cannot generalize the BCa as a whole. Although there were a few grade I tumor cases which were passaged to P1, P1 is not defined as a successful xenograft model in our study. Third, reporting array-CGH results was limited to showing the similarity of genetic profile patterns. For further investigations, it is required to analyze specific genetic loci where amplifications or deletions have occurred, and this may reveal novel genes causing mutations in the BCa. Finally, we selected P2 xenografts as a role of indicator models representing patients arbitrarily. But we don't know the exact passage number to prove the superiority of our tissue xenograft model over cell line xenografts. In general, studies for anticancer drug response using tumor tissue xenografts used early-passage models such as our xenografts in other cancers.[24, 27]
In conclusion, our BCa xenograft models derived from human tumor tissue not only retain histopathological features and molecular signatures similar to the original tumors, but also have the clinico-pathological heterogeneity of BCa. Therefore, we expect our model to be useful for studying the heterogeneity of BCa populations and for evaluating new treatments.