All experiments were conducted by inoculation of cells or tumor brei into the flanks of immunocompromised mice (nude, scid, or scid-bg). Trials were conducted either as early treatment (ET) studies with therapy beginning prior to the development of measurable tumors, or as staged tumor (ST) studies, with therapy initiated after tumors attained a specific size (typically 150-300 mm3). Tumor volumes were determined by two or three times weekly measurements using digital calipers.

aTumor growth inhibition. Calculated as 100 - %T/C based on the average determined from between one and more than 10 trials.

bTime for untreated or vehicle-treated tumors to reach an average tumor volume of 1000 mm3. Determined from an analysis of between one and more than 10 trials per tumor type.

Immunohistochemical (IHC) markers and/or DNA fingerprinting are methods that can be used to ensure cell line integrity. Most xenograft tumor lines used to establish models have been maintained in culture for many years and have highly aneuploidic genomes.41 Utilization of methods such as comparative genomic hybridization (CGH) can be informative for determining chromosomal regions of interest that may be amplified or deleted in the tumor lines. Mutational analysis of specific genes of interest may also be warranted, such as for examination of specific mutations in the epidermal growth factor receptor that confer higher degrees of sensitivity to inhibitors of this protein.42-44 Expression profiling of tumors and the cell lines from which they originated can help elucidate expression of target protein pathways, subclassify lines both within a given tumor type and between different tumor types, and identify changes in gene expression between cells grown in vitro versus in vivo that may help explain differences in compound sensitivity between these two settings. In cases where differential sensitivity (in vitro and/or in vivo) is observed for a particular agent (or class of agents) between different tumor lines, expression profiling can be used to identify genes that may be potentially used as predictive markers of activity in a patient population in the clinical setting.45 Histological analysis of basic morphologic characteristics such as degree of necrosis at various tumor sizes, relative extent of tumor angiogenesis (e.g., microvessel density or MVD), and proliferative/apoptotic index provides a more sophisticated assessment of tumor growth properties. Furthermore, IHC methods can be used to quantify, at the cellular level, drug-mediated effects on proliferation and/or apoptosis and MVD.46 IHC can also be used to examine target gene expression within a tumor as well as study downstream proteins that can be used as pharmacodynamic markers of activity in vivo. The histological and immunohistochemical analysis of tumors has been greatly aided by the development of tissue microarray technology.47 With this method, small cores of multiple (as many as several hundred) tumors are collected and reassembled in a single paraffin block. In this way many different tumor samples can be analyzed on a single slide.

Concern about the predictive ability of xenograft models has been reinforced by some notable examples of agents that have shown (to date) relatively limited clinical activity despite quite significant antitumor activity (including tumor regression) in xenograft models.48,49 In preclinical xenograft models, several farnesyl transferase inhibitors (FTIs) have demonstrated significant preclinical activity in colon, bladder, lung, breast, and other tumors but, for the most part, have shown limited clinical activity.50 Although these results certainly raise concerns about interpretations of xenograft efficacy (note that at least one FTI also showed significant activity in a transgenic model expressing oncogenic ras), one must be cautious to dismiss the efficacy observed in the preclinical models as uninformative. As discussed above, analysis of a broader set of models of each tumor type may help elucidate determinants that contribute to clinical resistance. In addition, the stratification of patients based on these selective criteria could significantly improve chances of a clinical response.

In the case of EGFR inhibitors, clinical evaluation of a very large set of patients revealed that specific mutations in the target were likely to underlie sensitivity.51 In other cases the disconnect between preclinical activity and poor clinical response may not be reflective of the model but rather due to differences in PK and/or metabolism of the experimental agent in humans compared to rodents. Peterson and Houghton26 describe several examples of agents (e.g., irofulven/MGI-114: MGI Pharma) that showed significant activity in xenograft models, yet failed in the clinic for the simple reason that it was not possible to obtain drug exposures in humans equivalent to exposures in mice that were required for activity. PK and metabolic interactions must also be considered when evaluating combination therapies. As it has now become standard practice to combine experimental agents with standard therapies early in both preclinical and clinical development, it is critical to demonstrate that potentiation or synergistic activity is not simply the result of an indirect enhanced exposure of the standard therapy.

0 0

Post a comment