Search

Cetuximab is an approved treatment for metastatic colorectal carcinoma (mCRC) with codon 12/13-KRAS mutations, recently questioned for its validity, and alternative mutation-based biomarkers were proposed. We set out to investigate whether an expression signature can also predict response by utilizing a cetuximab mouse clinical trial (MCT) dataset on a cohort of 25 randomly selected EGFR+ CRC patient-derived xenografts (PDXs). While we found that the expression of EGFR and its ligands are not predictive of the cetuximab response, we tested a published RAS pathway signature, a 147-gene expression signature proposed to describe RAS pathway activity, against this MCT dataset. Interestingly, our study showed that the observed cetuximab activity has a strong correlation with the RAS pathway signature score, which was also demonstrated to have a certain degree of correlation with a historic clinical dataset. Altogether, the independent validations in unrelated datasets from independent cohort of CRCs strongly suggest that RAS pathway signature may be a relevant expression signature predictive of CRC response to cetuximab. Our data seem to suggest that an mRNA expressing signature may also be developed as a predictive biomarker for drug response, similarly to genetic mutations.

Anecdote clinical observations hint that non-small cell lung cancer (NSCLC) with exon-20 insertions might respond poorly to epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs), contrasting to those with classic mutations. Lack of patient-derived experimental models has been a major hurdle for the discovery of new treatment for the diseases. We established two NSCLC-PDXs harboring two different exon-20 insertions, LU0387-adenocarcinoma (ADC) with a nine-base insertion at 2319 (H773-V774insNPH) and LU3075-squamous cell carcinoma (SCC) with a nine-base insertion at 2316 (P772-H773insDNP). Both insertions immediately follow the regulatory C-helix of the kinase domain. Contrary to the generally good responses to EGFR inhibitors observed in PDXs with classic mutations, both exon-20 insertions are largely resistant to cetuximab and TKIs in vivo, suggesting fundamental difference from the classic EGFR mutations, consistent with the poor response rate to TKI seen in anecdotal clinic reports. It is worth noting that although responses are generally poor, they differ between the two exon-20 mutants depending on the type of TKI. In vitro drug sensitivity assays using established primary cell lines from our two PDXs largely confirmed the in vivo data. Our data from patient-derived experimental models confirmed that exon-20 insertions in domain immediately following the C-helix confer poor response to all known EGFR inhibitors, and suggested that these models can be utilized to facilitate the discovery of new therapies targeting NSCLC harboring exon-20 insertions.

The Cancer Genome Atlas (TCGA) project has generated abundant genomic data for human cancers of various histopathology types and enabled exploring cancer molecular pathology per big data approach. We developed a new algorithm based on most differentially expressed genes (DEG) per pairwise comparisons to calculate correlation coefficients to be used to quantify similarity within and between cancer types. We systematically compared TCGA cancers, demonstrating high correlation within types and low correlation between types, thus establishing molecular specificity of cancer types and an alternative diagnostic method largely equivalent to histopathology. Different coefficients for different cancers in study may reveal that the degree of the within-type homogeneity varies by cancer types. We also performed the same calculation using the TCGA-derived DEGs on patient-derived xenografts (PDX) of different histopathology types corresponding to the TCGA types, as well as on cancer cell lines. We, for the first time, demonstrated highly similar patterns for within- and between-type correlation between PDXs and patient samples in a systematic study, confirming the high relevance of PDXs as surrogate experimental models for human diseases. In contrast, cancer cell lines have drastically reduced expression similarity to both PDXs and patient samples. The studies also revealed high similarity between some types, for example, LUSC and HNSCC, but low similarity between certain subtypes, for example, LUAD and LUSC. Our newly developed algorithm seems to be a practical diagnostic method to classify and reclassify a disease, either human or xenograft, with better accuracy than traditional histopathology. Cancer Res; 76(16); 4619–26. ©2016 AACR.

Engrafting the bone marrow cells of a patient with M5 acute myeloid leukemia into immunocompromised mice (AM7577) resulted in serially transferrable stable AML and eventual mortality. The disease starts in the bone marrow and then expands to peripheral areas, which is typical of M5 leukemogenesis, where high leukemic burden in blood is coincident with symptoms/mortality. The leukemic cells in the mice had myeloid morphology, phenotypes, and genotypes (including the internal tandem duplication of FMS-like tyrosine kinase receptor 3 gene [FLT3-ITD]) similar to those of the original patient. Autocrine mechanisms of human granulocyte–macrophage colony-stimulating factor/interleukin-3 likely support AM7577 growth in mice. Treatment with FLT3 TKI (AC220) caused complete remission in peripheral blood, spleen, and bone, along with relief of symptoms and extended life, hinting that FLT3-ITD may be a key leukemogenic driver maintaining the disease. Interestingly, withdrawal of AC220 (high dose) did not result in relapse of disease, suggesting cure. These results, however, are in contrast to cytarabine (AraC) induction treatment: First, although AraC significantly suppresses the diseases in blood, and to a lesser degree in bone marrow and spleen, the suppression is temporary and does not prevent eventual onset of disease/death. Second, the withdrawal of AraC always resulted in rapid relapse in peripheral blood and eventually death. Our observation in this patient-derived model may provide useful information for clinical applications of the two drugs.

Immune checkpoint inhibitors, i.e., PD1 and CTLA4 antibody therapeutics, have led to long-term survival in many late-stage solid tumor patients. Despite this revolutionary clinical impact, the overall response rate is still low. One major roadblock is that compensatory immune inhibitory pathways were turned on to protect tumor cells from being attacked by T cells. T cell immunoglobulin- and mucin domain-containing molecule 3 (Tim-3) is a type I transmembrane protein expressed in T cells, monocytes, macrophages, and dendritic cells. PD-1 and Tim-3 coexpression is often associated with an exhausted phenotype of tumor-infiltrating lymphocyte (TIL) in various tumor types. TIM-3 expression has been reported as a compensatory mechanism in the PD1 nonresponsive patients. Simultaneously targeting PD1 and TIM3 is now being actively tested in clinical trials. However, we still do not have an animal model that can efficiently evaluate the efficacy of therapeutic TIM3 antibodies or combinations. Here we report generation of a human TIM3 knock-in C57BL/6 mouse model, in which we replaced mouse exon 2-5 of HAVCR2 gene with human counterparts. The engineered knock-in mice express a chimeric TIM3 where its extracellular and transmembrane domains are human sequence, while the signal peptide and intracellular domain are kept intact. We confirmed that homozygous knock-in mice only express this chimeric TIM3, recognizable by human TIM3 antibody in NK cells, macrophages, and stimulated T cells. In an efficacy study, we inoculated the TIM3 knock-in mice with MC38 syngeneic tumor cells. Treating the mice with 5 mg/kg of human TIM3 antibody resulted in modest tumor growth inhibition (TGI), while in combination with anti-mouse PD1 antibody, the TGI improved to 62% following 2 weeks of treatment. We are now breeding the human PD1 and human TIM3 double knock-in mice to test the combinatory therapeutic antibodies. We have also developed human PD-L1 expressing MC38 cells, so that when inoculated in human TIM3 knock-in mice, PD-L1 and TIM3 therapeutic combination can be evaluated. Citation Format: Lei Zheng, Xuesong Huang, Wenyi Ouyang, Gang Chen, Annie Xiaoyu An, Xin Dong, Jay Liu, Jean Pierre Wery, Qian Shi, Davy Xuesong Ouyang. Generation of human TIM3 knock-in mice for preclinical efficacy assessment of therapeutic antibodies [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A203.

Little progress has been made in treatment of acute myeloid leukemia (AML) for years. Following the success of immune checkpoint inhibitors in solid tumors, immunotherapy also offers huge promise in many hematologic malignancies, including AML. Anti-PD1 antibody is currently being investigated in combination with azacitidine in patients with relapsed AML. Other emerging immuno-oncology (I/O) therapeutics are also at different stages of preclinical research. However, we are lacking relevant animal models with intact immune system for the preclinical in vivo pharmacologic studies and mechanism-of-action analysis of I/O agents. To meet this need, we set out to develop a series of transplantable mouse AML models carrying Nras mutations and/or a set of clinical relevant fusion proteins, like AML1/ETO. Using a similar approach reported by Scott Lowe and colleagues (1), we constructed retroviral vectors expressing human AML1/ETO9a IRES GFP and NrasG12D, generated retroviruses, co-transduced the fetal liver cells, and performed syngeneic intravenous injections in C57BL/6 mice. The mice developed AML in the following months as the GFP-positive cells enriched as high as around 90% in PBMC. Leukemia cells also dominatde the spleen and bone marrow in moribund mice. The splenocytes from moribund AML1/ETO9a mice were banked and transplanted into new C57BL/6 mice for several rounds. The later passages were more malignant than the early passages. To validate the immune profiling of this model, FACS analysis was applied using mouse PD-L1, Gr-1, CD45, Mac-1, c-kit, and sca-1 antibodies for GFP-positive cells. In PBMC of moribund AML1/ETO9a mice, PD-L1 and c-kit were highly expressed, while the expression levels of Gr-1, CD45, Mac-1, and sca-1 were low. For efficacy test, we performed standard-of-care (SOC) assay with cytarabine, doxorubicin, sorafenib, and anti-PD1 antibody. The dosing started around 10 days post inoculation of 1 x 106 leukemia cells, when GFP-positive cells were 2-10%. While doxorubicin showed good response, cytarabine and sorafenib resulted in modest tumor growth inhibition, and anti-PD1 antibody failed to show any meaningful efficacy. We are currently testing more combinatory treatments, including anti-TIM3 and anti-CTLA4 antibodies. Taken together, we have developed a mouse AML model with clinical relevant mutations for preclinical efficacy assessment of combinatory chemotherapies, target therapies, and immunotherapies. Reference: 1. Zuber J, Radtke I, Pardee TS, et al. Mouse models of human AML accurately predict chemotherapy response. Genes Dev 2009. Citation Format: Xin Tang, Lily Tong, Annie Xiaoyu An, Likun Zhang, Jie Cai, Qian Shi, Jean Pierre Wery, Davy Xuesong Ouyang. Developing an AML mouse syngeneic model for combinatory chemotherapy and immunotherapy [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A003.

OX40 is a costimulatory molecule of TNF receptor family, primarily expressed on activated effector T cells and regulatory T cells. Engagement of OX40 by ligand leads to dramatic expansion, effector function, cytokine production, and survival of T cells. Several OX40-agonistic antibodies are in early stages of clinical development and some have shown promising antitumor immunity and improved tumor-free survival in the patients compared with standard-of-care therapies. Preclinical studies using surrogate OX40-agonistic antibodies have also shown robust tumor growth inhibition (TGI) efficacies, especially when combined with other checkpoint inhibitors like PD1 and CTLA-4 antibodies. However, we are lacking reliable preclinical models to assess the efficacy of human therapeutic antibodies. To address this need, we created a human OX40 knock-in GEMM (OX40 HuGEMM), where the entire 2.5 Kb mouse OX40 coding region is replaced by human OX40. We confirmed the expression of human OX40 in the 95% of activated T cells from the homozygous knock-in mice, while the mouse OX40 is completely eliminated. In the heterozygous knock-in mice, 58% of activated T cells express both mouse and human OX40. We then used both heterozygous and homozygous knock-in mice for efficacy studies of OX40 antibodies. In the heterozygous knock-in mice bearing MC38 syngeneic tumors, human and mouse OX40 antibody monotherapies led to 49% and 20% TGI, respectively, while their combination with PD1 antibody was synergistically effective, resulting in ~90% TGI with multiple complete tumor regressions. When we grafted the homozygous knock-in mice with B16F10 syngeneic tumors, single 5 mg/kg dose of an investigational human OX40 antibody led to 47% TGI. Tumor-killing efficacy of OX40 antibody is also correlated with significantly increased tumor infiltrated T cells and NK cells. Taken together, our OX40 HuGEMM provides an urgently needed tool for testing in vivo efficacy of human OX40 therapeutics, both as a single agent and as various combination strategies with other treatment modalities. Citation Format: Xuesong Huang, Lei Zheng, Wenyi Ouyang, Mingkun Zhang, Annie Xiaoyu An, Jing Zhao, Xiang Gao, Jean Pierre Wery, Qian Shi. Utilizing human OX40 knock-in mice (HuGEMMTM) to assess antitumor efficacy of OX40-agonistic antibodies [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A207.

CD137 belongs to the TNF receptor super family, and its activation is essential to the function of T cells and NK cells. Multiple agonistic CD137 antibodies are currently being tested in humans with advanced cancer, which expected to become an important new immuno-oncology therapeutics in clinic. It is important to establish preclinical models for efficacy assessment of investigational CD137 antibodies before entering clinic. Currently available animal models cannot meet this need. Syngeneic mouse tumor models provide a useful platform for testing surrogate immuno-oncology therapies, but cannot be used for testing human therapeutic antibodies due to the species specificity; human immunity reconstituted models, by human PBMC or hematopoietic stem cell inoculation into immune deficient mice, suffering from highly variable responses, are not robust enough to provide a reliable system for efficacy studies. Establishing target humanized models by replacing mouse therapeutic target with human counterpart while maintaining normal mouse immunity is a practical approach to evaluate human therapeutic antibodies in vivo. Using CRISPR-Cas9 gene editing, we have engineered human CD137 knock-in model expressing chimeric human/mouse CD137 composed of human extracellular and transmembrane domains (Exon 4-7), with intact mouse signal peptide and intracellular domain. FACS analysis of splenocytes derived from homozygous knock-in mice showed that CD3/CD28 stimulated T cells only express chimeric CD137 reactive to human CD137 antibodies, but not mouse CD137; Yet the expression levels of the chimeric CD137 in the knock-in mice are generally lower than endogenous mouse protein in wild-type C57BL/6 mice. At present, we are testing the efficacy of reference human CD137 agonistic antibodies by treating these homozygous knock-in mice grafted with syngeneic MC38 tumor cells. Data will be presented at the conference. Citation Format: Davy Xuesong Ouyang, Gang Chen, Zhensheng Wang, Lei Zheng, Annie Xiaoyu An, Jean-Pierre Wery, Jay Liu, Xin Dong, Henry Q. X. Li. Establishment of CD137 humanized mouse model for efficacy assessment of agonistic anti-CD137 therapeutic antibodies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 1658. doi:10.1158/1538-7445.AM2017-1658

A preclinical trial identified 4 of 20 (20%) gastric cancer (GC) patient-derived xenografts responded to cetuximab. Genome-wide profiling and additional investigations revealed that high EGFR mRNA expression and immunohistochemistry score (3+) are associated with tumor growth inhibition. Furthermore, EGFR amplification were observed in 2/4 (50%) responders with average copy number 5.8 and >15 respectively. Our data suggest that a GC subtype with EGFR amplification and overexpression benefit from cetuximab treatment.

The recent clinical successes of immune checkpoint inhibitors have fueled the intense interest in novel immuno-oncology (I/O) therapeutics. The lack of relevant animal models remains a major hurdle in understanding the mechanism of action and evaluating the efficacy of such therapeutics. Patient derived xenograft (PDX), considered to most closely mimic patient tumors in both histo- and molecular pathology1,2, is however rarely used in I/O studies because it grow only in immune-compromised hosts. In reality, many PDXs grow well in nude mice where certain immune functions remain intact, excluding T-cells/T-cell functions. Therefore, PDX could still potentially be of practical use for studying T-cell independent I/O therapy. This study set out to evaluate a biologics for the treatment of a patient derived xenograft disease, by activating mouse natural killer (NK). NK and CD8 T cells are two major immune effector cell types that mediate cytotoxicity to tumor cells in vivo. One of the immunomodulatory agents, an anti-PD-L1 antibody-based IL-15 immunocytokine, was recently tested as a novel I/O treatment of cancer3. This molecule was engineered to be cross-species for both human and mouse PD-L1 and IL-15 that antagonize PD1/PD-L1 checkpoint-mediated immune suppression and also target the PD-L1-expressing tumor with IL-15 stimulated NK and CD8 T effector cells into local tumor sites, thus synergizing tumor-located anti-tumor immunity. In fact, our previous studies have demonstrated a greatly enhanced anti-tumor activity in the PDL-1-expressing syngeneic mouse tumor models via the mobilization of tumor infiltrating lymphocyte (TILs), over the PD-L1 antibody alone3. Since IL-15 also stimulates NKs in addition to T-cells, we reasoned that this bifunctional agent could also have potential activity against a PD-L1-expressing PDX in nude mice where NK remains functional. LU1901 is previously described NSCLC-PDX1, and was confirmed to express high level of PD-L1 by both RNAseq and flow analysis, after screening of a large panel of PDXs. We implanted LU1901 in Balb/c nude mice, and started to treat the mice by the bifunctional agent when the tumor reached to 150 mm3. Our result clearly demonstrated a significant inhibition of LU1901 growth by the bifunctional agent in nude mice, in the apparent absence of T-cells. When the treated tumors were examined at the termination, significantly infiltrate NK cells were found inside the treated tumors, as measured by both flow cytometry and immunohistochemistry (IHC). The number of infiltrating NK also statistically correlates to the amplitude of the tumor responses. Together, our data suggest that one of important mechanisms of action of this bifunctional agent relies on the tumor-targeting NK activation, and also that PDX could be a useful model suitable for in vivo efficacy analysis of T-cell independent immunotherapy. References 1. Yang, M., et al. Overcoming erlotinib resistance with tailored treatment regimen in patient-derived xenografts from naive Asian NSCLC patients. International journal of cancer. Journal international du cancer 132, E74-84 (2013). 2. Guo, S., Wubin Qian, Jie Cai, Likun Zhang, Jean-Pierre Wery, Qi-Xiang Li. Molecular pathology of patient tumors, patient derived xenografts and cancer cell lines EORTC-NCI-AACR. (2015). 3. Yan Wu, Zhaojing Zhong, Stella Martomo, Dan Lu, Zhanna Polonskaya, Xenia Luna, Haifan Zhang, Zhikai Zhang, Zhun Wang*, Leo Liu*, Jeegar Patel, James Tonra, Henry Li*, Larry Witte, Sam Waksal, Zhenping Zhu. Anti-PD-L1 antibody-based IL-15 immunocytokine has enhanced antitumor immunity. EORTC-NCI-AACR Abstract (2015). Citation Format: Henry Q. Li, Zhun Wang, Xiaoyu An, Jinping Liu, Likun Zhang, Jean-Pierre Wery, Yan Wu, James Tonra, Sam Waksal, Zhenping Zhu. Modeling an immunotherapy of NK mechanism on a NSCLC patient derived xenograft. [abstract]. In: Proceedings of the AACR Special Conference: Patient-Derived Cancer Models: Present and Future Applications from Basic Science to the Clinic; Feb 11-14, 2016; New Orleans, LA. Philadelphia (PA): AACR; Clin Cancer Res 2016;22(16_Suppl):Abstract nr A30.

We previously generated the MuPrime® mBR6004 model, by engrafting the breast adenocarcinoma derived from MMTV-PyVT transgenic mice (GEMM)1 into the syngeneic FVB/N mice2,3. The allograft grows robustly, maintains the histopathology features of the original GEMM, and metastasizes to the lungs in all examined cases when implanted orthotopically2. Additionally, histopathology shows that the models is HER2++, but ER-/PR-. We profiled drug responses to standard of care (SoC)2 and checkpoint inhibitors3. Interestingly, among all checkpoints inhibitors tested, our model is insensitive to PD1/PDL-1 inhibitors, but responds to CTLA-4 inhibitors3. To further explore the underlying mechanism of response, we performed extensive tumor immuno-profiling for the presence of infiltrating immune cells, e.g. TIL, CTL, Treg, immune-suppressive macrophages, NK, etc3. We observed a good pharmacodynamic (PD) correlation between the presence of tumor infiltrating T-cells, particularly CD8+ TIL, and NK and a positive response to therapy, regardless the treatment type. Given this preliminary observation, we attempted animal vaccination with tumor lysates prior to tumor engraftment or treatments. Interestingly, while vaccination had minimal effects on the engraftment take rate (100% take for all animals) and on the growth kinetics of the engraftments, it had profound effects on the tumor response to anti-PD1/anti-PDL-1 antibody treatments, significantly enhancing tumor response to these agents in a model that was otherwise unresponsive. Our results suggest that the immunological status of the animal, at least with regard to the specific anti-tumor immunity, is critically important to determine the response to checkpoint inhibitors. Our observations suggest that vaccination is critical for the overall success of immunotherapy (I/O) treatments utilizing checkpoint inhibitors as well other treatment strategies. References 1. Guy, C.T., Cardiff, R.D. & Muller, W.J. Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Molecular and cellular biology 12, 954-961 (1992). 2. Annie Xiaoyu An1, Jinping Liu1, Jie Cai1, Jean Pierre Wery1, and Henry Q.X. Li1,. Building mouse tumor derived allogragfts for immune-oncology research. (2015). 3. Zhun Wang1, A.X.A., 2*, Jinping Liu1, Gavin Jiagui Qu1, Likun Zhang, Jie Cai1, Bin Chen1, Davy Xuesong Ouyang1, Jean Pierre Wery1, and Henry Q.X. Li1,2. Response to checkpoint inhibition by GEMM breast cancer allograft EORTC-AACR-NCI Abstract (2015). Citation Format: Zhun Wang, Xiaoyu An, Jinping Liu, Xuesong Ouyang, Jiagui Qu, Likun Zhang, Jie Cai, Jean-Pierre Wery, Henry Li. Prior vaccination is critical to PD1/PDL-1 treatment efficacy in a mouse breast cancer allograft. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 3219.

Patient-derived xenograft (PDX) models closely mimic the pathological phenotypes of the original tumors, thus they are ideal surrogates for preclinical anticancer drugs efficacy testing. However, there is an increasing demand to rapidly and reliably screen out candidate drugs at earlier stages, and in vivo animal studies usually take months to produce efficacy data. Compared to traditional two-dimensional (2D) culture, three-dimensional (3D) culture systems more closely reflect in vivo responses. Therefore, we developed an ex vivo 3D culture platform using freshly isolated PDX-derived tumor cells, cultured on the natural methylcellulose polymer as an ex vivo drug screening system to meet the need for patient relevant, lower cost and high-throughput methods. Single-cell suspensions freshly isolated from PDX tumors are seeded in semisolid methylcellulose media and the drug efficacy is quantitatively evaluated by CellTiter-Glo® (CTG) luminescent assay. The main advantage of this method is the possibility of testing cytotoxicity within days (7 days on average), rather than months. Also, the process is carried out in mild condition and the scaffold material is inert to avoid unwanted cellular response. Moreover, cell spheroids can be easily retrieved for further analysis. Dissociated primary tumor cells derived from either fresh or frozen PDX tumors remain highly viable and form colonies when cultured in 3D. The success rate of generating ex vivo 3D culture from PDX tumors is higher than 90%, and over 60 PDX models were validated from cancer of various origins. We monitored tumor colony formation and generated a fitted drug concentration-response curve with small variations in CTG cell viability assays. Most importantly, the ex vivo 3D drug response profiling correlated well with the response obtained in vivo from PDX models (over 80% in average), for anticancer drugs with a variety of mechanisms of action including kinase inhibitor, antineoplastic agent and antibodies. Compared with a commercialized 3D drug testing plates, similar results were obtained in our system, with the advantage of a significant reduction in the cost. These data suggest that our 3D ex vivo cell culture platform from freshly isolated or frozen PDX-derived cells is a feasible and authentic method for testing drug efficacy in a short time period before transitioning to multiple PDX models. To sum up, our ex vivo 3D culture platform can save time and reduce costs, and is extremely useful for preclinical evaluation of anticancer drug, filling the gap between cell line-based and tumor bearing mice-based drug studies, allowing low cost, high throughput in vitro drug screening and the clinically relevant in vivo drug efficacy studies. Citation Format: Xiaoxi Xu, Wenna Zhang, Bingying Xie, Zheng Zheng Bao, Zhu Mei, Jean-Pierre Wery, Jinying Ning. An established ex vivo 3D culture platform using patient-derived xenograft (PDX) tumors replicate the anti-tumor efficacy observed in vivo. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 609.

The interaction between the c-MET tyrosine kinase receptor (RTK) on the surface of cancer cells with its ligand, hepatocyte growth factor (HGF) secreted from surrounding tumor stroma, results in paracrine signal transduction in cancer cells to promote growth, invasion, angiogenesis and metastases. c-MET gene amplification/over-expression, or activating mutations, frequently function as oncogenic driver in several types of cancers. c-MET is thus considered a key target(1) for both TKIs or biologics. However, lack of suitable animal models has represented an obstacle for the development of new cMET inhibitors. While patient derived xenografts (PDXs) mirror patients’ histopathological/genetic profiles(2-6), they usually lose human stroma, the source of HGF; moreover the HGF originated from murine stroma is incapable of triggering a paracrine signaling because of lack of cross reactivity with the human c-MET. Some PDX grow independent of paracrine mechanisms because of their constitutive c-MET activation by the presence of activating mutations, gene amplification or autocrine signaling, and can be used to evaluate the efficacy of TKIs(5, 7, 8). However they are not suitable to assess biologics, e.g. antibodies disrupting receptor-ligand interaction. Alternatively it is possible to grow tumors in hu-HGF-transgenic immunocompromised mice (humanization) where an artificial paracrine signaling is created. Interestingly, we have identified a unique HCC-PDX, LI0801, which maintains human stroma throughout passages, as demonstrated by immunochemistry staining for human vimentine and human HGF, both markers of stromal components(7, 8). The mechanisms of human stroma maintenance in this PDXremain unclear. We showed that LI0801 responds to different TKIs(7, 8). In addition we evaluated the efficacy of an anti-human c-MET monoclonal antibody in reducing tumor burden in LI0801, and we were able to demonstrate great tumor response to anti-human c-MET antibody, which is significantly more effective than crizotinib. To the best of our knowledge, LI0801 is the first paracrine c-MET PDX reported, and represents an important model to study paracrine cMET/HGF signaling and to evaluate new c-MET inhibitors, both TKIs, and biologics. References 1. Comoglio PM, et al. Nat Rev Drug Discov. 2008;7:504-16. 2. Ding L, et al. Nature. 2010;464:999-1005. 3.Marangoni E, et al. A new model of patient tumor-derived breast cancer xenografts for preclinical assays. Clin Cancer Res. 2007;13:3989-98. 4.Chen D, Oncotarget. 2015. 5. Yang M, et al. International journal of cancer Journal international du cancer. 2013;132:E74-84. 6. Zhang L, et al. A subset of gastric cancers with EGFR amplification and overexpression respond to cetuximab therapy. Sci Rep. 2013;3:2992. 7. Bladt F, et al. Cancers (Basel). 2014;6:1736-52. 8. Chen D. AACR Annual Meeting 2012. 2012;Poster. Citation Format: Dawei Chen, Ziyong Sun, Likun Zhang, Jie Cai, Davy X. Ouyang, Jean-Pierre Wery, Henry Li. Paracrine c-MET/HGF HCC PDX: evaluation of a biologics targeting c-MET. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 747.

The recent clinical success of novel therapeutics blocking immune checkpoints cytotoxic T-lymphocyte associated antigen 4 (CTLA-4) and programmed cell death 1 (PD-1) has fueled an intense interest around immuno-oncology. However, the lack of relevant animal models is a major bottleneck for understanding the mechanism of action and evaluating the efficacy of such therapeutics. Syngeneic mouse tumor models, despite being widely used as experimental models for efficacy studies, are limited by the restricted number of models available and responsive to current checkpoint inhibitors. Genetically engineered mouse models (GEMMs) of human cancer are effective tools for mechanistic analysis, but they are often not suitable for efficacy studies due to usually un-synchronized tumor progression. Syngeneic allografts of spontaneous mouse tumors derived from GEMMs (MuPrime®) may be used as a new type of immuno-oncology models with the following advantages: 1) their primary nature of “stem cell diseases” and relevant tumor microenvironment mirror the patient derived xenograft (PDX) models; 2) the availability of diverse cancer types and oncogenic drivers deriving from a wide range of GEMMs. We set out to build a library of allografts of spontaneous mouse tumors, including those derived from GEMM, in an immunocompetent host to support pharmacological investigation, particularly of immuno-oncology agents. Recently, we observed the growth of spontaneous cutaneous tumors on the neck of the C57BL/6J-ApcMin/J GEMM, which is heterozygous for the Apc (APCMin/+). Histopathology suggests the tumors are well-differentiated skin squamous cell carcinoma. We subcutaneously transplanted these skin tumors into the syngeneic C57BL/6J mice. The allografts grow well, are serially transplantable and maintain the histopathology of the primary tumor. Transcriptome sequencing revealed that the allografts maintain the original ApcMin mutation, express high levels of HER2 (also confirmed by immunochemistry, or IHC), and present frameshift mutations in both c-met and EGFR. In vivo pharmacological assessment indicate that tumors respond to 5-FU, paclitaxel, gemcitabine, docetaxel, and cisplatin chemotherapy. They is also sensitive to anti-mouse CTLA-4 antibody, and the response is associated with an increased number of tumor infiltrating immune cells, including TILs. Alltogether, here we show that we have established an allograft model suitable for in vivo efficacy analysis of immunotherapy using surrogate anti-mouse antibodies. Citation Format: Gavin Jiagui Qu, Annie Xiaoyu An, Jinping Liu, Davy Xuesong Ouyang, Likun Zhang, Jie Cai, Jean Pierre Wery, Henry Q. X. Li. Establishment of a mouse skin squamous cell carcinoma allograft model for in vivo pharmacological analysis of immunotherapy. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4043.

Overall benefits of EGFR-TKIs are limited because these treatments are largely only for adenocarcinoma (ADC) with EGFR activating mutation. The treatments also usually lead to development of resistances. We have established a panel of patient-derived xenografts (PDXs) from treatment naive Asian NSCLC patients, including those containing "classic" EGFR activating mutations. Some of these EGFR-mutated PDXs do not respond to erlotinib: LU1868 containing L858R/T790M mutations, and LU0858 having L858R mutation as well as c-MET gene amplification, both squamous cell carcinoma (SCC). Treatment of LU0858 with crizotinib, a small molecule inhibitor for ALK and c-MET, inhibited tumor growth and c-MET activity. Combination of erlotinib and crizotinib caused complete response, indicating the activation of both EGFR and c-MET promote its growth/survival. LU2503 and LU1901, both with wild-type EGFR and c-MET gene amplification, showed complete response to crizotinib alone, suggesting that c-MET gene amplification, not EGFR signaling, is the main oncogenic driver. Interestingly, LU1868 with the EGFR L858R/T790M, but without c-met amplification, had a complete response to cetuximab. Our data offer novel practical approaches to overcome the two most common resistances to EGFR-TKIs seen in the clinic using marketed target therapies.

Background. Activated B cell-like (ABC) is a subtype of DLBCL difficult to treat. Activation of B-cell receptor (BCR) signaling pathway, including CD79 transmembrane proteins (A/B subunits) and Bruton's tyrosine kinase (BTK), are required for the disease maintenance. The MYD88 (an adapter for Toll-like receptors in a separate pathway from BCR pathway) activating mutation, L265P, has been suggested to activate BCR pathway (29% of ABC-DLBCL) and thus drives the disease. These DLBCLs seem responsive to ibrutinib (a BTK inhibitor) in the clinics1,2. Majority of MYD88-L265P mutant DLBCL patients also have CD79B mutation (e.g. Y197N), i.e. double mutants (21% of total ABC-DLBCL), hinting collaborative oncogenesis, while some are MYD88-L265P single mutant with wild type CD79B. It remains unclear whether the single and double mutant ABC-DLBCL differ in their molecular pathogenesis as well as response to ibrutinib. Method. We created DLBCL patient derived xenografts (PDX) as experimental system mimicking patients in order to investigate the molecular pathogenesis/response to ibrutinib. We first transcriptome-sequenced the established PDXs. We categorized them into ABC/GCB subtypes along with mutation status of CD79B/MYD88 per transcriptome profiles. We then assessed these PDXs for their response to ibrutinib and trying to correlate antitumor activity to molecular signature. Results. We have created 20 NHL-PDX, with 5 of them transcriptome-sequenced. Clustering analysis per expression profile of 14-genes have classified all these 5 into subclass ABC-DLBCL. Among the 5, one is MYD88-L265P/CD79B-Y197N double mutants (LY2298), one MYD88-L265P single mutant (LY0257), and three wild types (LY2345, LY2214, LY2266). Among the 3 models tested so far, LY2214 (wildtype) and LY0257 (single mutation) did not respond to ibrutinib, but LY2298 (double mutations) did. These results, although small sample sizes, seem to suggest that double mutations are required for the response to ibrutinib, while single mutant and wild type does not. These observation seem to be consistent with the implication from recent clinical study1. We are currently profile and test the remaining models to further confirm these observations. We will also further elucidate the downstream signaling along with the drug response. Conclusion. Our preliminary results confirmed that ibrutinib sensitivity correlates with the concomitant double mutation of CD79B and MYD88. References 1. Wilson, W.H., et al. Targeting B cell receptor signaling with ibrutinib in diffuse large B cell lymphoma. Nature medicine (2015). 2. Aalipour, A. & Advani, R.H. Bruton's tyrosine kinase inhibitors and their clinical potential in the treatment of B-cell malignancies: focus on ibrutinib. Therapeutic advances in hematology5, 121-133 (2014). Disclosures No relevant conflicts of interest to declare.

Background. Cancer patient's response to immunotherapy, e.g. checkpoint inhibitors, differ greatly, likely due to their genetic composition (e.g. neo-mutation) or immunological profile of tumors. Experimental models may facilitate the identification of responders or of the approach to enhance response. Syngeneic mouse tumor models have been widely used as experimental model for testing surrogate immunotherapy, but limited by the fact that only a few are available and respond to the current checkpoint inhibitors. Allografts of mouse spontaneous tumors (MuPrimeTM) may be used as new type of I/O model with advantages: 1) its primary natures with “stem cell diseases” and relevant tumor microenvironment as seen in PDX1-5; 2) diverse cancer types and deriving from wider range of available spontaneous mouse tumors. Method. We created a MuPrime, MBR6004, by allografting the breast adenocarcinoma derived from GEMM-MMTV-PyVT in FVB/N mice1. We then characterized the model by examine it subcutaneous and orthotopic growth, and its ability to metastasis. We also transcriptome-sequenced and immune-profiled the tumor, and assessed its response to different treatments, including I/O therapy. Results. MBR6004 was found to over-express HER2 but negative for ER and PR. It maintains original primary tumour histopathology, grows robustly, and is confirmed to express its original transgene. It is also resistant to docetaxel. The orthotopic implantation resulted in lung metastasis. We can detected and quantitatively confirmed tumor-infiltrating immune cells, e.g. TIL, CTL, Treg, immune-suppression macrophage, NK, etc. We confirmed that the tumor cell expressed relatively low PD-L1, but inducible, per flow analysis. Our preliminary data indicated that it partially respond to anti-mouse PD1 and anti-mouse CTLA-4 antibodies when mice are preconditioned. Currently, we are investigating whether the responses are associated with the reduced Treg and increased CD8+ TIL in tumors, and also whether the combination therapy can enhance its tumor response synergistically. Conclusion. Our data seem to suggest that MuPrime could be another type of I/O model alternative to commonly used syngeneic cell line derived models at present. References 1. Guy, C.T., Cardiff, R.D. & Muller, W.J. Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Molecular and cellular biology 12, 954-961 (1992). 2. Guy, C.T., et al. Expression of the neu protooncogene in the mammary epithelium of transgenic mice induces metastatic disease. Proc Natl Acad Sci U S A 89, 10578-10582 (1992). 3. Moser, A.R., Dove, W.F., Roth, K.A. & Gordon, J.I. The Min (multiple intestinal neoplasia) mutation: its effect on gut epithelial cell differentiation and interaction with a modifier system. The Journal of cell biology 116, 1517-1526 (1992). 4. Moser, A.R., et al. ApcMin, a mutation in the murine Apc gene, predisposes to mammary carcinomas and focal alveolar hyperplasias. Proc Natl Acad Sci U S A 90, 8977-8981 (1993). 5. Moser, A.R., Pitot, H.C. & Dove, W.F. A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science 247, 322-324 (1990). Citation Format: Zhun Wang, Xiaoyu An, Jinping Liu, Jie Cai, Likun Zhang, Jiagui Qu, Davy Ouyang, Bin Chen, Jean-Pierre Wery, Henry Li. Response to checkpoint inhibition by GEMM breast cancer allograft. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr B97.

Background. The Cancer Genome Atlas (TCGA) and many other genomic profiling of patient samples generated abundant human cancer genomic datasets of various histological types and individual patients, enabling development of molecular pathology methodology per big data fashion1-4. Patient derived xenograft (PDX) is believed to mimic patient tumors with anecdotal data, and assumed to be predictive experimental models (avatar) for human trial. There is also a need to investigate tumor pathology among different histopathology types, either patient or xenograft diseases, per “big data” approach. Methods. We first systematically compared transcriptome expression similarity within and between specific histological types using either global or specific subsets of genes using the datasets from TCGA. We next investigated PDXs (also cell lines) using the same systemic approaches. Results. We demonstrated the presence of highly consistent correlation in transcriptome expression within a given histology types, but more distinct between the types, establishing molecular specificity alternative to histological type based on morphology, or an alternative diagnostic approach. We performed the same systematic examination of PDXs of different histological types corresponding to the TCGA types. The similar general pattern of similarity, also somewhat reduced in degree, have been observed. These suggest the high relevance of PDX to human cancer tissues in general from “big data” perspective. In contrast, this seems generally untrue for cancer cell lines: quite similar among themselves even for different histological types, but less correlation to either TCGA or PDXs per disease type. The similarity among cell lines of all types may suggest a “new type of cancer, or tissue culture type”. Our studies also revealed certain specific correlations, with several examples: 1) colon and rectal adenocarcinoma are highly correlated, without apparent difference; 2) NSCLC SCC are quite different from NSCLC ADC, as seen morphologically; 3) NSCLC SCC are also quite closely related to HNSCC. Conclusion. There are generally good correlation between transcriptome expression based molecular pathology and histopathology; the molecular pathology of PDXs are generally similar to that of human's, not so to human cell lines. References 1. Comprehensive molecular characterization of gastric adenocarcinoma. Nature 513, 202-209 (2014). 2. Hoshida, Y., et al. Integrative transcriptome analysis reveals common molecular subclasses of human hepatocellular carcinoma. Cancer research 69, 7385-7392 (2009). 3. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068 (2008). 4. Ge, L., et al. Integrated analysis of gene expression profile and genetic variations associated with ovarian cancer. European review for medical and pharmacological sciences 19, 2703-2710 (2015). Citation Format: Sheng Guo, Wubin Qian, Jie Cai, Likun Zhang, Jean-Pierre Wery, Henry Li. Molecular pathology of patient derived xenografts. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr B101.

Background. Anaplastic lymphoma kinase (ALK) is a receptor tyrosine kinase, and its mutation, over-expression and gene fusion have been associated with various cancers. EML4-ALK fusion has been confirmed to be an oncogenic driver for a small subset of the lung cancers (∼4% of NSCLC)1, and becomes an excellent drug target that led to the development of the first effective therapy, crizotinib, a tyrosine kinase inhibitor for anaplastic lymphoma kinase, for this disease 2,3. However, like any other cancer therapy so far, the treatment always led to the development of resistance, rendering them ineffective in the end4. Understanding the mechanisms causing these resistances can potentially facilitate overcoming the resistance and extend patients' life5,6. However, lack of experimental model hinders this understanding. Method. We have recently established a large collection of patient derived xenografts (∼350 PDXs)7, and we screened some of them for alk gene fusions. We then tested the positive model for sensitivity to crizothinib, and keep it under several rounds of crizotinib treatments to select resistant model. The induced resistant model was subjected to genomic analysis to identify the changes that might potentially be responsible for the resistance. Results. We identified one model, NSCLC LU1656, containing EML4-ALK fusion with elevated ALK expression. It has also been shown to respond well to crizotinib in vivo. The continued treatment eventually led to the development of resistance to crizotinib (LU2445), a situation that might occur in patients in the clinic under the same treatment. We have performed transcriptome sequencing of the parental sensitive tumor (LU1656) and the selected resistant tumor (LU2445). So far, we found both models expressed similar high levels of expression of ALK, but no additional mutations in ALK gene. There are other alterations either genetically or epigenetically, some seemingly related to the ALK signaling pathways. However, their role in resistance still need to be confirmed. Conclusions. Induced crizotinib resistance in ALK-fusion lung PDX can be useful to investigate crizotinib resistant mechanism and future drugs overcoming the resistance. References 1. Koivunen, J.P., et al. EML4-ALK fusion gene and efficacy of an ALK kinase inhibitor in lung cancer. Clinical cancer research : an official journal of the American Association for Cancer Research 14, 4275-4283 (2008). 2. Kwak, E.L., et al. Anaplastic lymphoma kinase inhibition in non-small-cell lung cancer. N Engl J Med 363, 1693-1703 (2010). 3. Takeuchi, K., et al. RET, ROS1 and ALK fusions in lung cancer. Nature medicine 18, 378-381 (2012). 4. Katayama, R., et al. Mechanisms of acquired crizotinib resistance in ALK-rearranged lung Cancers. Science translational medicine 4, 120ra117 (2012). 5. Katayama, R., et al. Two novel ALK mutations mediate acquired resistance to the next-generation ALK inhibitor alectinib. Clinical cancer research : an official journal of the American Association for Cancer Research 20, 5686-5696 (2014). 6. Friboulet, L., et al. The ALK inhibitor ceritinib overcomes crizotinib resistance in non-small cell lung cancer. Cancer discovery 4, 662-673 (2014). 7. Yang, M., et al. Overcoming erlotinib resistance with tailored treatment regimen in patient-derived xenografts from naive Asian NSCLC patients. Int J Cancer 132, E74-84 (2013). Citation Format: Jianyun Deng, Zhun Wang, Jie Cai, Sheng Guo, Jean-Pierre Wery, Henry Li. Induction of resistances to crizotinib in lung patient derived xenograft. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr C82.

Background. Blockage of immune checkpoints, e.g. by anti-PD1 antibody, becomes an important new cancer therapy1. Experimental models are important to evaluate new investigational therapy or new combination strategy. Syngeneic mouse tumor models have been widely utilized as an experimental model for testing surrogate immune-oncology (I/O) therapy by using its competent mouse immunity2, but cannot be used for testing human biologic therapeutics, due to the species specificity. The direct replacing mouse therapeutic target by human counterpart in mouse while maintaining the normal mouse immune-functions could be a potential practical preclinical approach to evaluate human biologic therapeutics in vivo. Methods. We have engineered a chimeric human/mouse PD1 gene (h/mPD1) composed of human exome 2&3 and mouse exome 1&4. We expressed the recombinant proteins and tested their bindings to their binding partner of PDL1 of both mouse and human origins, and also anti-human PD1 antibody. We knock-in the recombinant gene into C57Bl/6 mouse to create homozygous HuGEMM-h/mPD1 mouse, which is tested for growth of MC38 syngeneic mouse cell line tumor graft and for the growth inhibition by anti-human PDL1 antibody. Results. Our data demonstrated that chimeric protein h/mPD1 can interchangeably interact with mPDL1 or hPDL1 efficiently as efficiently as mouse PD1, and it also recognizes anti-human PD1 antibody as expected. The binding of anti-human PD1 antibody blocks its binding to mouse or human PDL1. The knock-in mice express the chimeric gene in the T-cells of the engineered mice both in vivo and ex vivo, however at significantly lower levels than mouse PD1 in the wild type C57BL/6 mice (1/10) under induction. When syngeneic MC38 cell line was subcutaneously engrafted in HuGEMM-h/mPD1 mice, the tumor were found to grow significantly slower with increased T-cell infiltration, as compared to those in the wild type mice. MC38 tumor did not respond to anti-human PD1 antibody well either. These observations can apparently be attributed to the low level PD1 and the associated high autoimmunity that also inhibits tumor growth. Interestingly, a specific condition can be artificially created to enhance MC38 tumor growth in the chimeric mice, likely contributed by the enhanced h/mPD1 expression. The enhanced tumor growth seems to also be suppressible, at least partially, by anti-human PD1 antibody as shown in our preliminary study. Currently, we are re-engineering our chimeric gene (version 2) in order to increase their expression to the wild type gene level, so to create a model for more optimal drug evaluation. In the meantime, we are also engineering HuGEMM-hu-CKPT (e.g. CTLA4 PDL-1, OX40, 4-1BB, etc.) for evaluating other checkpoint inhibitors. Conclusions. Our data suggests the conditioned version 1-HuGEMM-h/mPD1 mouse can be explored to evaluate anti-human antibody. References 1. Pardoll, D.M. The blockade of immune checkpoints in cancer immunotherapy. Nature reviews. Cancer 12, 252-264 (2012). 2. Allard, B., Pommey, S., Smyth, M.J. & Stagg, J. Targeting CD73 enhances the antitumor activity of anti-PD-1 and anti-CTLA-4 mAbs. Clinical cancer research : an official journal of the American Association for Cancer Research 19, 5626-5635 (2013). Citation Format: Zhun Wang, Bin Cai, GANG Chen, Jinping Liu, Xiaoyu An, Zhengsheng Wang, Davy Ouyang, Jean-Pierre Wery, Jay Liu, Xin Dong, Henry Li. HuGEMM-h/mPD1 mouse models for assessing anti-human PD1 therapeutics. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2015 Nov 5-9; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2015;14(12 Suppl 2):Abstract nr A11.

Immuno-oncology is an area under intensive investigation, however, many questions remain to be answered due to lack of sufficient number of adequate experimental models: 1) why only subsets of patients respond? 2) who will respond (signature)? 3) what can we do to make more patients to respond? Patient derived xenograft (PDX or HuPrime®) mirrors patients’ histopathological/genetic profiles, considered more predictive than traditional human cell line derived xenograft (CDX) and also syngenic mouse cell derived models, due to the primary natures of PDX. On the other hand, both xenografts (PDX/CDX) grow in absence of immunity, thus unsuitable for immune-oncology research. Limited types/numbers of available syngenic cells also restrict its application. Genetically engineered mutant mouse cancer model (GEMM) are diverse (types/numbers)1-5, grow in immune-environment, and are primary tumors. However, GEMM has logistic limitation for pharmacology research for high cost and high variations in tumor development (spontaneous) in type/time. We set out to create a library of primary mouse tumor allografts (MuPrimeTM) as standard experimental models, aiming at overcoming these limitations. We established a number of such allografts and are profiling them for growth, histopathology, genomic (RNAseq), and standard of care (SOC). We are also charactering their immune-oncology properties: tumor-infiltrating immune cells, including different subtypes of T-cells (CTL, Treg), as well as their functional activation/suppression and tumor responses to immune-modulating agents. MBR6004 is an allograft derived from a breast adenocarcinoma of GEMM-MMTV-PyVT in FVB/N mice1, which expressed medium level Her2 but negative for ER and PR, and is resistant to Docetaxel(SOC). It expresses low-level PD-L1(CD274)/L2 and seems insensitive to PD-1 antibody treatment. MCR6013, an allograft of a colon adenocarcinoma derived from GEMM-C57BL/6J-ApcMin/JNju3-5, are currently being profiled. MPR6003, an allograft of transgenic adenocarcinoma mouse prostate (TRAMP) in C57BL/6 mice6-8, grow both orthotopically and subcutaneously, and are also being profiled. M6011, an allograft derived from a spontaneous tumor in Balb/c mouse with high Her2 expression, has apparent infiltration of T-cells, including Treg, but negative for PD-L1. We are currently characterizing its histopathology and other property. MLY1014, an allograft of a spontaneous lymphoma derived from Balb/c, is also being profiled. In general, these allografts keep the original primary tumor histopathology, but distinct from conventional syngenic cell line derived tumors, a contrast seen between PDX vs. CDX. The models grow robustly after engraftment, thus ideal for pharmacology evaluation. Our ongoing studies will provide a useful platform with normal immunity for drug evaluation, particularly for immune-oncology agents. Citation Format: Xiaoyu An, Jinping Liu, Jie Cai, Jean-Pierre Wery, Henry Qixiang Li. Building mouse tumor derived allogragfts for immune-oncology research. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 2281. doi:10.1158/1538-7445.AM2015-2281

The PI3K-mTOR pathway is one of the major signaling cascades that promote tumor growth and cell survival. Multiple inhibitors targeting this important axis are currently under clinical investigation for cancer treatment and some are under preclinical development. Here we examined the anti-proliferation activity of BEZ235, a PI3K and mTOR dual inhibitor, in 243 human cancer cell lines of different caner types. We identified 35 lines that are sensitive to the inhibitor with IC50 values less than 0,04uM and 57 insensitive lines with IC50 values more than 0.4uM. We compared gene expression of the two groups by Affymetrix U219 arrays. With a p-value cutoff set at 0.00001, we identified 14 genes that are differentially expressed. These genes include epithelial membrane glycoprotein EPCAM and serine threonine kinase 31 (STK31). Furthermore, the mutation status of tubulin tyrosine ligase TTL is significantly correlated with sensitivity to BEZ235. These results suggested a potential genomic signature that could be predictive of response to BEZ235 and provided information that may help selecting patients that are likely to receive clinical benefit. Citation Format: Jing Zhang, Sheng Guo, Zhengzheng Bao, Limei Shang, Xiao Wen, Jean-Pierre Wery, Jinying Ning. Cell-based screening identifies gene expression signature correlated with sensitivity to PI3K-mTOR dual inhibitor BEZ235. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1038. doi:10.1158/1538-7445.AM2015-1038

RET encodes a receptor tyrosine kinase (RTK) involved in cellular mechanisms of proliferation, migration and differentiation. RET fusions with different genes at 10q11.2 result in constitutively activation and drive tumor development in various cancers: 10-20% of sporadic papillary thyroid cancer1, spitzoid neoplasms, CMML and 1% of lung adenocarcinomas2,3. RET fusions have been suggested to be important therapeutic targets. Ponatinib, a multi-kinase inhibitor, was approved for the treatment of Ph+ chronic myeloid leukemia (CML) and acute lymphoblast leukemia (ALL), including CML with T315I/imatinib-resistance. Ponatinib has recently been found to potently inhibit the common NSCLC fusion variant, KIF5B-RET at clinically-achievable concentrations. We have established large collection of colorectal patient derived xenografts (PDXs)4,5. Two of the poorly differentiated adenocarcinoma, CR1520 and CR2518, contain two different chromosome 10 in-frame RET-fusions, CCDC6-RET (CR2518) and NCOA4-RET (CR1520), as revealed by RNAseq and Sanger Sequencing6. Both models have also demonstrated over-expressing ret gene at mRNA levels. More importantly, they both responded Ponatinib significantly along with dephosphorylation of RET and downstream AKT, confirming these two types of Ret fusions as oncogenic drivers in these two models6. In addition, we also observed that Ponatinib treatment also rapidly drives CR2518 into its resistance (The resistant derivative is called CR2545). To investigate the underlying mechanism of the resistance, we performed RNAseq analysis of CR2545, which revealed retaining of the ret-fusion, but also the introduction of a previously described gate-keeper mutation, V804M, at ret kinase domain. V804M, a mutation frequently found in familial medullary thyroid carcinoma (FMTC), introduces bulky amino acid at position 804 and found to resistant to vandetanib. This may contribute to the resistance. However, this notion apparently contradicts to the previous observation/claim that V804M is sensitive to Ponatinib7, and ponatinib should be used to treat RET-V804M/L cancers. It is possible that the described V804M containing vandetanib resistant cells are sensitive to ponatinib, but via a non-RET mechanism for the multi-kinase inhibition nature of ponatinib; or in our case, there are other non-RET Ponatinib resistant mechanisms been introduced in CR2545 where RET is no longer the key oncogenic driver. We are currently testing other RET inhibitors and also assessing pharmacodanamic (PD) parameters to further investigate RET resistant mechanism. The further findings from the ongoing studies will be presented at the meeting. Citation Format: Mengmeng Yang, Jie Cai, Sheng Guo, Jean-Pierre Wery, Henry Qixiang Li. Rapid conversion to resistance, of a colon PDX with ret-fusion, by ponatinib treatment could potentially be attributed to the introduction of the gate keeper mutation V804M. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 3581. doi:10.1158/1538-7445.AM2015-3581

Patient derived xenografts (PDXs) mirrors patients’ pathology and genetic profiles, thus valued as predictive experimental models for studying oncogenesis and personalized treatments. Cancer is not a single disease but diseases of complex genetic components and oncogenic processes. Limited number of PDX models with minimal genetic characterization is insufficient to meet current research needs. For this, we have built the largest and most comprehensive PDX library with full genetic profiles. By far, our PDX library contains over 1,100 models derived from patients of both Asian and Western origins, covering over 20 major cancer types, including large panels (over 100 models each) of NSLCL(1), CRC(2), gastric(3), HCC(4), and pancreatic, and smaller panels ( Our large library of different disease panels are particularly useful in conducting mouse clinical trial (MCT of HuTrialTM) (2, 5, 12), which can be used to discover predictive biomarker (2, 5, 13) and guide clinical study design. Citation Format: Jie Cai, Dawei Chen, Rajendra Kumari, Sheng Guo, Jie Yang, Mengmeng Yang, Andrew McKenzie, Zhun Wang, Xuesong Huang, Xiaoyu An, Jinping Liu, Jean-Pierre Wery, Henry Li. Building comprehensive and fully annotated patient tumor derived xenogragft (PDX) library mirroring cancer patient population. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1472. doi:10.1158/1538-7445.AM2015-1472

Patient derived xenografts (PDXs) without in vitro manipulation are believed to mirror original patients’ histopathologic and genetic profiles, thus to be predictive surrogate models for patients, with superiority over conventional cell lines. We have built the largest and comprehensive PDX library of >1,100 models with genetic profiles of major cancer types, including some major cancer types: 200 NSLCL1, 200 CRC2, 200 gastric3, 100 HCC4, 100 pancreatic, 30 ovarian, 10 brain tumors, and many other cancer types. We set out to compare these 6 types of our PDXs with the corresponding TCGA5 patient tumor samples and CCLE6 cancer cell lines on their genomic expression by calculating pairwise Spearman rank correlation coefficient ρ, in order to further explore/confirm the similarity and difference among the three collections. PDX were profiled by both RNAseq and microarray (Affymetrix Human Genome U219 Array); CCLE were from microarray (Affymetrix Human Genome U133 Plus 2.0 Array); and TCGA were from RNAseq. For convenience, these 4 gene expression datasets are called “PDX, PDXchip, CCLE and TCGA”. Only genes common to all 4 datasets were used to compute ρ. The preliminary data seems to lead to two observations. First, for all 6 types, there is always at least one different tumor CCLE-TCGA pair that has higher ρ than the same tumor CCLE-TCGA pair. For example, the TCGA colorectal, lung, and ovarian cancers have higher ρ with CCLE pancreatic cancer than the TCGA pancreatic cancer does. In contrast, for both PDX and PDXchip, only lung and pancreatic cancers show such behavior, and notable, both to TCGA colorectal cancer. Seocnd, for same tumor type comparisons, ρ is 0.73-0.83 (average 0.772) between PDX and TCGA, 0.67-0.76 (average 0.698) between PDXchip and TCGA, 0.67-0.7 (average 0.682) between CCLE and TCGA. Third, about 20% of CCLE lung cancers have much lower ρ with TCGA, PDX, and PDXchip lung cancers than the other CCLE lung cancers. In summary, these observations show that our PDX models are close in genomic expression profile to TCGA patient tumors per tumor types specifically, and more so than CCLE cancer cell lines, which is also less specific. References 1. Yang, M., et al. Overcoming erlotinib resistance with tailored treatment regimen in patient-derived xenografts from naive Asian NSCLC patients. Int J Cancer 132, E74-84 (2013). 2. Chen, D. Cetuximab response in CRC patient-derived xenografts is predicted by RAS pathway activation rather than KRAS mutation status. ASCO 2013 Annual Conference (2013). 3. Zhang, J. et al. A more efficient RNAi inducible system for tight regulation of gene expression in mammalian cells and xenograft animals. Rna 13, 1375-1383 (2007). 4. Guo, S. et al AACR-Annual Conference (2013). 5. http://cancergenome.nih.gov/. 6. Barretina J, et al. http://www.broadinstitute.org/ccle/home. Nature 28, 603-607 (2012). Citation Format: Sheng Guo, Wubin Qian, Jie Cai, Jean-Pierre Wery, Henry Qixiang Li. Patient-derived xenografts seem to have closer global expression profile to that of the patient tumors of the corresponding cancer types, than the equivalent cell lines do. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1926. doi:10.1158/1538-7445.AM2015-1926

e15078 Background: Esophageal squamous cell carcinoma (ESCC) is a deadly malignance prevalent in parts of China and no target therapy option. Cetuximab, approved for treating subset of metastatic c...

e18505 Background: Patient-derived xenograft (PDX) is increasingly used to evaluate oncology drugs. PDX, reflective of original patient pathology and genomics, is predictive of response to treatmen...

Patient derived xenograft (PDX or HuPrime®) is believed to best mimic original human cancers from which the models were derived. PDX becomes popular to evaluate cancer drug for better predictive power. Cancers are diverse for their molecular pathogenesis and different in response to treatment. Multiple subjects are always needed in clinical evaluation, which is also true in PDX. Mouse clinical trial (MCT), or HuTrialTM, mimics phase II clinical trial, involving multiple subjects. MCT can be used for drug position and biomarker discovery. A large number of PDXs have been created globally from diverse cancer patients, enabling MCT. MCT offers advantages over human trial: a) complex tests (e.g. new agents, high dose, complex combinations, the same subject repeatedly, and no need for early stop before completion,...) can readily be performed with significant less regulatory constraints; b) any single subject can be assigned to different treatment arms, enabling internal control and enhancing statistic power; c) the endpoints can be objectively defined and measured; d) healthy inbred mice can guarantee consistent drug exposure, avoiding PK and/or other non-tumor related variations; and e) the potential savings in later clinical trial due to reduction in size and improving success. Randomized and controlled trial (RCT) with statistic power, as a gold standard in the clinic, can also be applied in MCT design and implementation, in order to enable reliable data analysis and conclusion. In addition, the commonly used clinical endpoints can also been tested in MCT, including Response Evaluation Criteria in Solid Tumors (RECIST: CR, PR, SD, PD) and survival (PFS, OS, HR, MST), besides those commonly used in mouse study, e.g. T/C, TGI, etc. We have created a large library of PDX models (1, 2), and also been testing them for drug evaluation in a MCT format (2) for confirming indication and discovering predictive biomarker. In this study, we have specifically compared the clinic endpoint and the endpoints used in MCT, and also investigated the potential parameters that can influence the study design by retrospectively analyzing our historic MCT studies. Our result can help us to design clinic-relevant endpoint and reveal the parameters influencing trial data integrity: 1) # of models and # mice per arm needed; 2) the potency of test article, as measured by hazard ratio or HR, on mouse # requirement; 3) tumor growth rate on response (ΔT/ΔC) or on “survival” endpoint (time to event). Our data so far seem to point to that response endpoint are correlated with “survival endpoint”in general. HR seems not correlated with tumor growth rate. The smaller HR, the lower the # mice per arm is needed. These observations can be useful in facilitating future MCT design. References 1. Yang et al. Int J Cancer. 2013;132:E74-84. 2. Zhang et al. Sci Rep. 2013;3:2992. Citation Format: Sheng Guo, Zhun Wang, Dawei Chen, Jie Yang, Mengmeng Yang, Xuesong Huang, Jean-Pierre Wery, Henry Li. Parameters influencing the design of mouse clinical trial (HuTrialTM). [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 1206. doi:10.1158/1538-7445.AM2014-1206

e15009 Background: For many cancers, chemotherapy is a major or even the only treatment option, but unfortunately without ultimate proof of clinical benefits in many cases, e.g. improved PFS/OS. In...

Gastric carcinoma (GC) is a common cancer of high mortality and particularly prevalent in East Asia. There are only a few treatment options: surgery, chemotherapy, as well as Herceptin® for a small subset of patients with erbb2 amplification. Cetuximab has been approved for CRC (colorectal) and SCCHN, but not yet for gastric carcinoma. Cetuximab, combined with capecitabine/cisplatin, has recently been tested in Phase III clinical trial (EXPAND) in GC patients (1), but failed to demonstrate benefit over the existing chemotherapy (capecitabine/cisplatin). One of the reasons for the failure is only small subset of patients responding to cetuximab, which constitutes little effects on large trial populations, while the identity of the responders yet to be revealed. One of the keys to a successful search for effective treatment is the development of experimental model that truly mimics patient conditions. Patient derived xenograft (PDX or HuPrime®) is believed to best mimic human cancers (2). Recently, we have established a cohort of ∼70 GC xenograft (PDX) models from Asian and Caucasian patients. In the present study, we conducted a mouse clinical trial (MCT, or PDX trial) to test cetuximab in a cohort of GC-PDX, aiming at identifying those responsive to cetuximab. Interestingly, our trial result identified 4 of 20 (20%) GC-PDX responded to cetuximab. Genome-wide profiling and additional investigations revealed that high EGFR mRNA expression and immunohistochemistry score (3+) are associated with tumor growth inhibition. Furthermore, EGFR amplification were observed in 4/4 (50%) responders by DNAchip analysis or 2/4 confirmed with FISH analysis, with average copy number 5.8 and >15 respectively. Our data suggest that a GC subtype with EGFR amplification and overexpression benefit from cetuximab treatment. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):B43. Citation Format: Lianhai Zhang, Jie Yang, Yiyou Chen, Jie Cai, Xiaoming Song, Xuesong Huang, Jianyun Deng, Dawei Chen, Mengmeng Yang, Shuangxi Li, Aiwen Wu, Ziyu Li, Zhongwu Li, Yiqian Liu, Jean-Pierre Wery, Henry Li, Jiafu Ji. A subset of gastric cancers with EGFR amplification and overexpression respond to cetuximab therapy. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr B43.

Gastric carcinoma is a common cancer of high mortality and is particularly prevalent in East Asia. There are only a few treatment options: surgery, chemotherapy, as well as Herceptin® for a small subset of patients with erbb2 amplification. Metastasis usually renders surgery ineffective. Drug therapy becomes only meaningful option. One of the keys to a successful search for effective treatment is development of experimental model that truly mimics patient conditions. Patient derived xenograft (PDX or HuPrime®) is believed to be among the best mimicking human diseases (1). Recently, we have established a cohort of ∼70 gastric cancers from Asian and Caucasian patients. This report describes that two of them (GA0046, GA0087), while established as subcutaneous xenografts, were confirmed to metastasize to lung at high frequency (100% for GA0087). GA0087 is a neuroendocrine cancer of gastric origin. Its high metastasis potential, as measured by counting the colony nudules in lung of subcutaneous tumor bearing mice, suggests that it would be a useful experimental model for studying cancer metastasis and exploring inhibitory agents. The observed metastasis is particularly significant since metastasis is rarely occurin subcutaneous xenograft. Circulating tumor cells, or CTCs, have been identified in the blood of some cancer patients and believed to be responsible for cancer metastasis (2, 3). We recently developed a new method, based on subtraction enrichment combined with immunofluorescence staining (anti-human cytokeratin 18 and CD45 monoclonal antibodies) and FISH (iFISH), for effective identification and enumeration of CTCs enriched from bloods of patients (4). In this study, we investigated the presence of CTCs in GA0087 metastatic PDX and their potential correlation to metastastic potential. As a result, we, for the first time, detected CTCs in PDX (GA0087), with a range of 1-61 CTCs per 200μL mouse blood. Our preliminary observation seems to also indicate that the CTC frequency is correlated to the numbers of metastasis nodules in lung. Furthermore, a chemotherapy agent can inhibited both CTC and metastasis. Our data seem to confirm the proposed correlation of metastasis and CTC in this first experimental metastasis/CTC PDX model. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):A157. Citation Format: Jiahua Jiang, Daisy D. Wang, Mengmeng Yang, Dawei Chen, Sheng Guo, Jie Cai, Linda Li, Jean-Pierre Wery, Peter Ping Lin, Henry Li. Metastatic PDX modeling displays correlation between CTC and metastatic potential, both inhibited by chemotherapy. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr A157.

Cetuximab was approved for advanced colorectal carcinoma (CRC), but yet for NSCLC, although some trials [e.g. FLEX (1) and BMS099 (2)] demonstrated benefit in subsets of patients. Recent retrospective study on Caucasian patients among the FLEX study cohort revealed that high EGFR expressors, uniquely defined by high-IHC score, are likely to benefit from the treatment (3). In particular, this was confirmed in a 50-patient derived xenografts model trial, mouse clinical trial (MCT or “Patient Avatar Trial”), in a similar format of FLEX, first time demonstrating relevance of human trial and MCT. The goal of the present study is identify the subset of Asian NSCLC patients that respond to cetuximab in a MCT. We have established a large panel of PDXs from NSCLC Asian patients of different histology (4): squamous cell carcinoma (SCC), adenocarcinoma (ADC), pleiomorphic carcinoma (PLC), and large cell carcinoma (LCC). We tested cetuximab activity in a randomly selected 32 PDX in a MCT. While NSCLC had overall engraftment rate of 25%, SCC interestingly, had higher rate of ∼31%, near-double ∼17% of ADC, which may have metastasis and prognosis implications. A subset of PDXs were found to respond to cetuximab (10/32) (31%, defined by %ΔT/ΔC < 25%). Surprisingly, all 10 responders exclusively belong to SCC and PLC (9/10 and 1/10 respectively), contrasting none to ADC (0/10). Non-/poor-responders (%ΔT/ΔC > 25%) include ADC (7/22), LCC (2/22) and SCC (12/22). Among the 12 non-/poor-responsive SCC, one has KRAS-G12D mutation, one has ALK-EML4 fusion and one has c-MET gene amplification. Among the SCC responders are ones with activating EGFR mutations. SCC has strikingly higher cetuximab response rate (RR) than ADC, with possible positive factors of EGFR mutation and negative factors of KRAS mutation, ALK fusion and c-met amplification. If proven true clinically, cetuximab could become a new therapy for NSCLC-SCC. Citation Information: Mol Cancer Ther 2013;12(11 Suppl):C14. Citation Format: Mengmeng Yang, Jie Cai, Sheng Guo, Xuesong Huang, Jie Yang, Xiaoming Song, Dawei Chen, Likun Zhang, Jiahua Jiang, Jean-Pierre Wery, Henry Li. Squamous non-small cell lung cancer (NSCLC-SCC) patient-derived xenografts (PDX) from Asian patients have high response rate (RR) to cetuximab than those of adenocarcinoma. [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2013 Oct 19-23; Boston, MA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(11 Suppl):Abstract nr C14.

3544 Background: Cetuximab was approved for treating EGFR-expressing metastatic colorectal carcinoma (mCRC) without patient stratification. Subsequent retrospective clinical studies resulted in an exclusion criterion for patients with KRAS mutations at codons 12 and 13. However, only a fraction of patients with wtKRAS benefit from the treatment. Recent analyses indicated that patients with KRAS mutation at codon 13 can also benefit. We set out to investigate whether KRAS mutations, or any other factors, are good biomarkers for CRC-cetuximab therapy patient stratification. Methods: We have established patient derived xenografts (PDX) from treatment-naive Asian CRC patients to discover biomarker predictive of cetuximab response. We conducted a clinical trial style study (“patient avatar trial”) of cetuximab using a randomly selected cohort of 26 EGFR+ PDXs. The antitumor activities were analyzed against KRAS mutation status and a published expression-signature. Results: We identified 12/26 (~46%) as cetuximab responders and 14/26 non-responders (defined by 50% tumor growth inhibition threshold). All 14 non-responders are mutated for one or more of KRAS, BRAF (V600E), EGFR, AKT and PIK3CA oncogenes. In contrast, 5/11 responders analyzed are wild-type for all these genes. Importantly, 5/11 responders have activating KRAS mutations at codons 12/13, contradictory to the currently practiced clinical exclusion criteria, but consistent with more recent clinical findings. Most interestingly, the observed cetuximab activity in this cohort of 26 PDXs has a surprisingly strong correlation to a published RAS pathway score. Conclusions: Our results indicated strong predictive power of cetuximab-CRC-PDX trial and RAS signature, and warrant prospective clinical studies for further confirmation.

Patient derived xenograft (PDX) is closely reflective of tumor in patient for both histopathological and genetic profiles 1, 2. It thus becomes widely accepted as model of choice to evaluate anticancer agents with improved predictive power. However, current PDX (HuPrime® 1.0/2.0) still has limitations, including growth in an environment lacking functional immunity, particularly that of human. Human immune functions have been successfully reconstituted in the same types of immunodeficient mice that were also used for engrafting PDX. This type of humanized mice with functional human immune system are established usually through engraftment of stem cells containing human CD34+ cells isolated from fetal liver, peripheral blood or umbilical cord blood (UCB). It is therefore also plausible to engraft both patient tumor and normal functional human immune system to a single recipient mouse so that tumors would grow in an environment even more similar to that of patients, and yield an even more predictive experimental model. This is particularly so for evaluating biologics or any immune modulating agents. In the present study, we engrafted CD34+ UCB (hematopoietic stem cell enriched population) into NOD/SCID mice and demonstrated full reconstitution of human hemopoiesis, including myelopoiesis, lymphopoiesis (both T and B lineages), growth of dentritic cells (DCs), macrophage and monocytes, and growth of natural killer cells (NKs). These cells can be found in abundance in bone marrow, spleen as well as peripheral blood. We also further demonstrated the normal immune functions in the reconstituted mice (HuMICE™), including production of human cytokines such as GM-CSF and IL-3, etc. Meanwhile, we also engrafted a previously established non-small cell lung cancer PDX (HuPrime®) model (LU25033 and LU0387) into the CD34+ UCB reconstituted NOD/SCID mice. Our preliminary results demonstrated that the tumor in the normal human immune system grow well, with maybe slight slower rate than the tumors in mice without reconstitution. We have observed significant infiltration of various human immune cells into the xenograft tumors. We are presently evaluating this type of new PDX models (HuPrime™ 3.0) by treating them with existing antibody therapeutics. Our preliminary data showed that the presence of human immunity, such as human NK-mediated ADCC, may have slight enhanced anti-tumor effect of antibody drug in HuPrime™ 3.0 than in HuPrime™ 2.0 (no human immune-reconstitute). We are currently assessing the role of the reconstituted immune function played in a biologic drug effect. We believe that HuPrime™ 3.0 could represent a more predictive model than those without functional immunity. Citation Format: Xiaoyu (Annie) An, Mengmeng Yang, Ran Wu, Jinping Liu, Taiping Chen, Jean-Pierre Wery, Yiyou Chen, Henry Li. Growth of patient derived tumor xenograft (PDX) in NOD/SCID mice reconstituted with human umbilical cord blood (UCB) CD34+ cells. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2784. doi:10.1158/1538-7445.AM2013-2784

Drug resistance is one of important factors limiting effect of most cancer therapies. Understanding mechanisms causing resistance can facilitate overcoming resistance. c-MET gene activation, via either activating mutations or gene amplification, is an oncogenic driver for some non-small cell lung cancers (NSCLC) in patients as well as in patient derived xenograft (PDX) (1). c-MET gene activation is also one of the two most dominant TKI-resistance seen among the patients treated with EGFR-TKI (tyrosine kinase inhibitors) (1, 2). Therefore, c-MET inhibitor could become an important treatment option for NSCLC patients whose c-MET is either de nova activated or activated during TKI treatment (1). We recently demonstrated that Crizotinib (PF2341066), an approved drug against NSCLC with ALK-fusion and a known c-MET inhibitor, can effectively treat NSCLC with c-MET gene amplification, including those resistant to EGFR-TKIs (1), in PDXs. Specifically, we described two models LU1901 and LU2503 – both with amplified gene, accompanied with over-expression, of c-MET, and both sensitive to Crizotinib (complete regression in LU2503, while near-complete regression in LU1901) (1), confirming the c-MET activation as dominant oncogenic driver for these two tumors. In the present study, interestingly, the treated LU1901 tumors subsequently re-emerge and re-grow even in the presence of continuous treatment. In particular, the re-growed LU1901 become completely resistant to Crizotinib (we call LU1901R), even after it was re-engrafted. Pharmacodynamic analysis revealed that the LU1901R tumor has similar level of phosphorylation (or activation) of c-MET as the parental LU1901. Crizotinib treatment caused nearly-similar inhibition of c-MET in both LU1901 and LU1901R, as well as the downstream signaling, including pMEK and pAKT by Crizotinib in both LU1901 and LU1901R tumors along with c-MET inhibition. All these suggest that the threshold of c-MET activation for prompting tumor growth/survival is significantly lower in LU1901R than LU1901. We are currently characterizing the LU1901R in order to identify the mechanisms of resistance, and attempt to find approach to overcome them. In summary, we have successfully developed c-MET inhibitor resistant models that may reflect resistance emergence in the clinics. These models could be particularly useful in investigating clinical resistant mechanisms and evaluating new therapy that can overcome resistance. Citation Format: Mengmeng Yang, Xiaoming Song, JIe Cai, Youzhu Wang, Taiping Chen, Jean-Pierre Wery, Yiyou Chen, Henry Li. Treatments of NSCLC-PDX with c-MET gene amplification by c-MET inhibitors lead to rapid development of drug resistance. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 5630. doi:10.1158/1538-7445.AM2013-5630

Non-small cell lung cancer (NSCLC) has the highest mortality among all malignances while with limited treatment options. Among the most significant treatment advancement in recent years are the introduction of target therapies of EGFR-TKIs (Erlotinib and Gefitinib) for adenocarcinoma with EGFR activating mutations, and Crizotinib for NSCLC with ALK activation (EML4-ALK fusion)1. ALK activation are associated with a subset of NSCLC patients (3∼5%).Crizotinib is considered not a particularly potent ALK inhibitor, with IC50 at range of tens of nanomoles. It may not be an ideal agent for treating NSCLC with activating ALK. Therefore, a new ALK inhibitor with better potency and specificity could be of particularly medical importance. Currently, the industry is actively searching for such new agents. A relevant experimental model could also greatly facilitate evaluating and validating these new drug candidates. Patient derived xenografts (PDX) (including NSCLC PDX) mirror patients’ histopathology and genetic profiles2, thus known as “patient avatar”. It has improved predictive power as preclinical cancer models, and enables discovery of predictive biomarkers for targeted therapeutics. We have established a large collection of NSCLC PDX models (NSCLC-HuPrime®) by engrafting naïve patient tumor tissues into immune-compromised mice2. Among these models, we have identified one model, LU1656, of NSCL carcinosarcoma histology and with EML4-ALK fusion mutation. This mutation apparently causes over-expression of ALK as measured at mRNA level. This model has wild-type KRAS, c-MET and EGFR genes. Treatment of LU1656 with Crizotinib caused significant but not complete tumor response, suggesting ALK activation is one of the oncogenic drivers of this model. In addition, LU1656 also responded to Cetuximab, but respond poorly to Erlotinib. It would be interesting to examine the combination of these treatments. This model is, to our knowledge, the first reported PDX model with EML4-ALK fusion mutation. We believe that LU1656 would be a useful model to evaluate investigational ALK inhibitors, and new combination treatment regimens of Crizotinib and standard of care chemotherapies. Citation Format: Jie Cai, Mengmeng Yang, Jianyun Deng, Xiaoming Song, Xuesong Huang, Taiping Chen, Jean Pierre Wery, Yiyou Chen, Henry QX Li. A NSCLC PDX with EML4-ALK fusion mutation responds to crizotinib. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 337. doi:10.1158/1538-7445.AM2013-337

Hepatocellular carcinoma (HCC) is a disease of diverse types. Transcriptome analysis recently classified HCC into three major categories with distinct clinical parameters as well as cellular differentiation1. They are S1 (stem-like group with activation of the WNT pathway and TGF-β), S2 (activation of MYC and AKT), and S3 (hepatocyte differentiation). Patient derived xenografts (PDXs) have been suggested to be predictive experimental models (“Patient avatars”) because they have not been in vitro manipulated, and have the original patients’ histopathological and genetic profiles2,3. In order to assess the resemblance of these “avatar” to patients, we performed transcriptome profiling on a collection of HCC-PDXs using the same algorithm1. Twenty-two HCC-PDX models were classified into 3 groups, among which S1 and S2 are more closely related. Using the 572 genes generated from the algorithm, the same classifications were obtained by hierarchical clustering and principal component analysis (PCA), except for two outliers. The results demonstrate that our cohort of HCC-PDXs can be divided into the same three subclasses as those from patient samples1. In addition, Luk et al., recently classified HCC using expression levels of microRNA cluster at 14q32.2 locus, in which a stem-like HCC phenotype was associated with high expression4. We thus also used this criterion to classify our collection of HCC PDXs. The two resulting categories, 14q32.2-hi and 14q32.2-lo, can be compared to the S1, S2, S3 subclasses determined by mRNA profiling described above, in which S1 belongs to 14q32.2-lo and S2/3 belong to 14q32.2-hi (except LI1025, LI1078). The serum α-fetoprotein (AFP) and tumor AFP-mRNA levels were found to be strongly and positively correlated, and also associated with S21 and 14q32.2-hi4, consistent with previous reports1. Six stem-like markers were found not to be associated with S11, nor with 14q32.2-hi4, thus different from the report by Luk et al4. Activation of c-MET was found exclusively in S1 models, consistent with one of previous observations5, but not the other4. When all these models were treated with Sorafenib, a multi-kinase inhibitor, there seemed to be no correlation in the tumor responses between the subclasses. In summary, the data provided insights into the classifications of HCC by mRNA expression1 and 14q32.2 microRNA expression4, partially consistent with existing reports1,2. The results also show distinctive characteristics of our HCC-PDXs that merit further studies. Citation Format: Sheng Guo, Wubin Qian, Jie Cai, Dawei Chen, Jie Yang, Zhun Wang, Xiaoming Song, Taiping Chen, Jean Pierre Wery, Yiyou Chen, Henry QX Li. Classification of HCC patient derived xenograft (PDX) models by transcriptome and microRNome profiling, and their resemblance to patients. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2731. doi:10.1158/1538-7445.AM2013-2731

Recent advances in targeted anticancer drug discovery and development has brought about many new promising IND entities to the frontline of preclinical and clinical research. One of the major challenges is to identify and select appropriate patient population for further validation of the new drug efficacy in preclinical and clinical settings. Thus, patient stratification is a critical step in developing personalized cancer therapeutic drugs. This requires comprehensive understanding of genetic background and biomarker profiles of the patients. To meet this highly desired need, we have built up a robust preclinical oncology pharmacology platform, establishoing over 250 human primary tumor xenograft models (HuPrimeTM), including more than 20 common cancer types. All HuPrimeTM models are annotated with clinical information, diagnosis pathology, genome-wide transcription profiling, genome-wide SNP-copy number variation (CNV), hot-spot mutations of key oncogenes, key biomarker IHC information, and responses to standard-of-care (SOC) agent treatment. All the HuPrimeTM model information, along with cancer cell line genomic information, were collected into a proprietary interactive database, HuBaseTM, which offers intuitive navigation control and searchable function, enabling cancer researchers to analyze hundreds of cancer cell lines and HuPrimeTM tumor models for their need of drug development programs. With easily accessible genomic information, HuPrimeTM models or cell lines which express specific target genes of interest can be identified by transcription profiling and gene copy number variation (CNV) status. Biomarkers that are potentially predictive to drug response can then be validated in HuPrimeTM tumor models based on genomic annotation prior to pharmacological studies. Models with appropriate genomic signatures and biomarker information can be selected for more efficient drug efficacy tests with higher success rate. Using the genomic-biomarker guided approaches, we have validated standard-of-care drug responses in HuPrimeTM tumor models that showed significantly elevated expression of molecules involved in drug target related signaling pathways, conducted PD biomarker identification and discovery studies, including biomarkers in various cellular response upon drug treatment.In this presentation, examples of HuPrimeTM tumor models, application of HuBase® data and predictive biomarker IHC analysis in model selection for several targeted therapy agents, including Trastuzumab (Herceptin), Sorafenib, Cetuximab, and Tarceva, will be discussed. We believed that this genomics-biomarker guided approach can greatly improve the effectiveness of preclinical evaluation of targeted drug development. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2667. doi:1538-7445.AM2012-2667

Hepatocellular Carcinoma (HCC) is a major health problem, as the fifth most common cancer and the third most common cause of cancer death worldwide. Many pharmaceutical companies lead the effort to design new medicines on this cancer type. The overwhelming need to improve preclinical models of HCC requires robust models that closely mimic the disease population and therefore have the potential to better predict clinical outcome with novel therapies. Here we successfully validated subcutaneous and orthotopic xenograft models derived from HepG2, SK-HEP-1, Hep3B, PLC/PRF/5, HUH-1, HUH-7 and MHCC97H cell lines and also directly from HCC patient tumors (HuPrime®). Four HCC serum markers (AFP, α-fetoprotein; AFP-L3; DCP, Des-gamma-Carboxyprothrombin; GP73, Golgi Protein-73) have been used or will be potentially used clinically in detection of this disease. AFP is the most widely used biomarker in HCC, somehow it is not always correlated well with the liver cancer progress, even it gives false result, and some liver lesion or liver cirrhosis can also produce high level of AFP; while GP73 is reported a promising alternative as HCC serum marker. AFP and GP73 were tested at both mRNA level and protein level. They were also compared in the sensitivity and correlation with the tumor burden using the blood samples from the orthotopic xenograft models. Our results demonstrated that biomarkers released by the tumor models could well monitor the tumor take rate, growth and response to treatments. Furthermore, Sorafenib, one popular anti- HCC medicine, was tested, and the serum biomarkers could be utilized for confidence in evaluating the efficacy of Sorafenib. In conclusion, the use of HCC xenograft models combined with the biomarker strategies is helpful in maximizing the value of in vivo research and designing clinical trials Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1611. doi:10.1158/1538-7445.AM2011-1611

Gene expression analysis has led to the identification of 5 subtypes of breast cancer (basal-like, HER2-like, luminal A and B, and normal-like) that have distinct biological features, clinical outcomes, and responses to conventional chemotherapy and targeted therapy. Animal models possessing genetic and other biomarker abnormalities similar to each subtype of the human counterpart are thus of crucial importance in the development of new therapeutic strategies. Human breast cancer xenograft model is one of the more difficult models with transplantation of either established cell lines or primary tumors. We have successfully established 6 xenograft models derived from established cell lines and 2 xenograft models by transplanting primary human tumor fragments. The immunohistochemical results classified the 9 models into 4 major groups: ER negative and HER2 negative (MX-1, MDA-MB-231, MCF-7-CB1, and SKBR3-CB1), ER negative and HER2 positive (BT-474-CB1), ER positive and HER2 negative (MCF-7), and ER positive and HER2 positive (BT-474), which well resembled basal-like/normal-like, HER2-like, luminal A, and luminal B breast cancer types in human. The mutation status of TP53, BRCA1, BRCA2 and EGFR was investigated, and compared with the in vivo efficacy results by using various test drugs including cytotoxic agents, paclitaxel and doxorubicin; ER antagonist, tamoxifen; anti-HER2 monoclonal antibody, trastuzumab; HER2 inhibitor, HKI-272 and EGFR inhibitor, erlotinib. These data closely mimicked the heterogeneity in clinical outcomes of the human disease. Therefore, these models provide valuable platform for the development of personalized therapy for breast cancer. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 647.

The ability to engraft human leukemia cells directly from patient samples into immunodeficient rodents such as NOD/SCID mice would be uniquely valuable in preclinical prioritization of novel therapies for future clinical trial. The aim of this study was to develop an animal model for human acute leukemia in which the kinetics, characteristics and chemosensitivity of leukemia can be sequentially monitored in individual immunodeficient mouse model. Continuous xenografts from 28 acute myeloid leukemia (AML) patients and 18 acute lymphoblastic leukemia (ALL) patients with AML-ETO, CBFβ/MYH11, TAL/SIL1, MLL/AF4 fusion genes were established in NOD/SCID mice by intrasplenic inoculation of human leukemic cells from born marrow. Progression of leukemia was monitored throughout the development of disease by determination of immunophenotyping and karyotyping of the engraftment tumor cells. Chemosensitivity and in vivo efficacy to standard of care therapies were screened in xenografts from both AML and ALL patients. Genetic lesions and special surface markers in different passages of cells in recipient mice were also examined by gene array and FACS analysis. Intrasplenic inoculation of leukemia cells significantly enhanced the engraftment rate and expedited disease development when compared with I.V. route. Relative to engrafted human primary cells, negligible changes in growth rates, chromosomal aberrations and immunophenotype were observed at second, third and fourth passages. The in vivo sensitivity of xenografts to standard of care correlated significantly with clinical course and patient outcome. We have also uncovered key molecular events and serum immunoassay biomarkers which correlate with the efficacy and potency of the drugs for AML and ALL models. The biologic and genetic characteristics and patterns of chemosensitivity of xenografts from acute leukemia patients accurately reflect the clinical disease. The efficient engraftment and rapid expansion of the leukemia within NOD/SCID mice affords a robust model for the preclinical evaluation of new therapies. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 395.

Most anticancer drug candidates, currently in clinical trials, have clear targets in molecular and cellular systems. However, such molecular targets are not always affected (or not correlated with efficacious endpoints) when the candidate is evaluated in a clinical setting. The gap between preclinical and clinical drug evaluation remains the bottle neck for anticancer drug development. There is an urgent need to develop better biological tools and models to ease the transition between in vitro and in vivo, and between preclinical and clinical settings. In our translational platforms, 170 human primary tumor models (HuPrimeTM) have been established directly from tumor fragments of the patients with many cancer types. In order to pick the best models for evaluation of targeted therapeutics, we have conducted a comprehensive molecular profiling of our models. This profiling includes Affymetrix whole genome expression, Illumina 1400 SNP of cancer related cancer genes for amplification, deletion, and allelic imbalance. Methylation statuses for selected gene sets were also analyzed by SuperArray platform. Genomic amplifications were further confirmed with protein analysis by western blot or immunohistochemistry. The models selected in this way have facilitated pharmacodynamic (PD) biomarker strategies. We have developed a unique platform combining our novel HuPrimeTM xenograft models with molecular profiling and PD biomarker exploration. We have now validated immunohistochemistry (IHC) assays on many molecules covering most signaling pathways that are involved in proliferation, apoptotic, necrotic, and cell cycle regulation, G2M phase arrest, DNA damage response, etc. Additional markers including angiogenesis have also been validated for certain therapeutic compounds. To illustrate the usefulness of our platform, we have applied it to understand the pharmacodynamic changes (at the molecular level) which are associated with the activity of Sunitinib, Sorafenib and Tarceva in several HuPrimeTM models. The PD biomarkers validated in this approach have potential for clinical application and patient stratification. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 637.

The hedgehog (Hh) signaling pathway is known to be essential for multiple aspects of embryonic development, and activation of the Hh pathway leads to the development of several human malignancies such as basal cell carcinoma and medulloblastoma. Recent data have implicated a much broader role of the Hh pathway in other human cancer as well as the maintenance of cancer stem cells. Loss-function mutation of Ptc and gain of gunction mutation of Smo in human malignancies make the components in Hh pathway attractive therapeutic targets. However, this pathway is readily deactivated once human tumor cells are propagated in cell culture, a prerequisite of most preclinical human cancer models. In order to identify useful models for the evaluation of novel therapeutics targeting the Hh pathway, we profiled the expression of Ptc, Smo, Gli in a cohort of twenty established human primary gastric tumor models by TaqMan real-time quantitative PCR reaction and immunohistochemistry (IHC). Seven of the twenty models displayed high expression of either Gli-1 or Gli-2. Eight models were found to have high expression level of Shh. And three models were found to have high level of Ptc. The Gli protein expression level is also examined with IHC. High levels of Gli and Shh in the selected models will be very useful to evaluate anticancer drugs against Hh signaling pathway for their drug activity and pharmacodynamic responses. These models, together with the associated clinical and pathological information, provide a unique platform to evaluate the in vivo activities of pharmaceutical agents targeting the Hh pathway. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 713.

e14612 Background: Most anticancer drug candidates, currently in clinical trials, have clear targets in molecular and cellular systems. However, such molecular targets are not always affected (or not correlated with efficacious endpoints) when the candidate is evaluated in a clinical setting. There is an urgent need to develop better biological tools and models to ease the transition between in vitro and in vivo, and between preclinical and clinical settings. Activation of tyrosine kinases in tumor cells has been recognized as key driving force in malignancy; therefore inhibitors of tyrosine kinases (TKI) have often shown efficacy in preclinical in vivo models and beneficial responses in clinical trials. An increasing number of TKI has been approved as anticancer drugs in selected cancer types. Methods: We have developed a unique platform combining our novel HuPrime™ xenograft models with our PD biomarker technologies. This platform is used to better understand the efficacy of novel drug candidates and generate information critical to maximize the chance of a successful clinical development. To illustrate the usefulness of our platform, we have applied it to understand the pharmacodynamic changes (at the molecular level) which are associated with the activity of sunitinib and sorafenib in our sensitive esophageal HuPrime models. Tumor tissues from sensitive esophageal HuPrime models treated with single dose of the respective drugs (and a vehicle control) have been collected at time-points 4, 8, 16, and 24 hour. Results: We have successfully applied immunohistochemistry (IHC) assays on many molecules covering most pathways including proliferation, apoptotic, necrotic, and cell cycle regulation, G2M phase arrest, DNA damage response, etc. Additional markers including angiogenesis can also be included for certain therapeutic compounds. These tissues have been analyzed with our platform and we have uncovered key molecular events which correlate with the efficacy and potency of the drugs. Conclusions: The PD biomarkers validated in this approach have potential for clinical application and patient stratification. [Table: see text]

11008 Background: New oncology drug development has moved from general cytotoxic agents to molecular target-directed therapeutics. Consequently, there is a need to identify tumor types and individual patient tumors that express the target and could benefit from more selective therapies in clinical trials. Therefore, the in vivo models used in preclinical development should be“disease-oriented” and target-directed. Recently, we developed non-small cell lung cancer (NSCLC) xenograft models by transplanting human fresh tumor fragments into nude mice, which have been used for test of cytotoxic and targeted anticancer drugs. Methods: The fresh NSCLC tumors were collected from local hospitals. The tumor fragments were subcutaneously implanted into nude mice. The EGFR and K-ras mutation status of the tumors were investigated and compared with the efficacy results. The test drugs included paclitaxel, gemcitabine, and the epidermal growth factor receptor (EGFR) inhibitor erlotinib. Results: A total of 72 NSCLC samples were implanted, and 13 tumor models were established (tumor taking rate 18%) for the first passage. The tumor taking rates were higher in the second and third passages (80–100%). Paclitaxel and gemcitatbine produced tumor growth inhibition rates of 50–53% regardless of the EGFR and K-ras mutation status. While erlotinib demonstrated a significant antitumor activity only in the tumors bearing EGFR mutation with wild-type K-ras, which were consistent with their clinical findings. The tumor xenografts’ architecture, the cell and histopathological morphology from the three generations mirrored the original patient cancers. Conclusions: These results suggest that human primary tumor xenograft models provide a unique renewable source of tumor material for test of novel anticancer agents. They may predict more relevant clinical response rates and higher correlation with clinical findings than use of xenograft models established from long-term cultured cancer cell lines, especially for test of target-oriented therapeutics in new drugs development programs. No significant financial relationships to disclose.