Figure 2 Contemporary preclinical drug development process

initiated simultaneously or shortly after tumor cell inoculation (but prior to the development of palpable tumors) or as staged tumor studies (tumor growth delay), where animals are assigned to treatment groups after tumors have reached a defined size.

Once therapy is initiated, tumor size is determined by caliper measurements conducted with defined frequency (weekly to daily, depending on the tumor growth rate). One common representation of tumor size is the calculated tumor volume determined by the formula V = (larger diameter x shorter diameter2) x 0.5. The tumor growth inhibition (TGI) properties of an experimental compound can then be examined by determining the average tumor volume of the treated (T) group relative to the control (C) group (%T/C or %TGI, calculated as 100% - %T/C).18 While this is generally determined at the end of the dosing period, it can also be valuable to analyze %T/C values throughout the study (early in the treatment cycle and posttreatment). This additional information about efficacy can be useful if the %T/C changes dramatically through the duration of a chronically dosed trial, as it may be indicative of either tumor resistance or changes in the PK or PD of the drug within the animal. This is especially informative with the novel targeted agents. An additional efficacy parameter, tumor growth delay, is another important measure of activity. This value is determined by calculating the mean or median number of days required for the treated versus vehicle groups to reach a specific endpoint (e.g., a tumor volume of 1000 mm3), statistically compared using parametric or nonparametric tests and reported as percent increased lifespan (%ILS).

7.02.4.1.2 Liquid/hematologic tumor models

Net log cell kill is an additional parameter of efficacy evaluation that is commonly used for the evaluation of cytotoxic agents in leukemia models but is less frequently determined in solid tumor models.18

7.02.4.1.3 Orthotopic/metastatic tumor models

In addition to these flank models, tremendous progress has also been made in the development of more sophisticated orthotopic and metastatic tumor models that provide an opportunity for the potential evaluation of some stroma-tumor interactions and extravasation/intravasation processes.19-21 The expression of molecular targets can be variable depending on the organ or stromal environment of the growing tumor and this can be helpful when working with targeted agents. In addition, the PK/PD relationships in different tumor-containing organs can be evaluated. However, relative to flank models, orthotopic and metastatic model systems often have increased complications associated with staging of trials and, without sensitive imaging technologies and/or genetic alteration of tumors cells, the inability to monitor tumor sizes remains a drawback. Furthermore, without tumor staging, these models can require large cohorts of tumor-bearing animals for convincing statistical analysis. These models also often require a high degree of technical skill and can be quite time-consuming, thus potentially restricting the size of trials that can be conducted. However, the evaluation of biomarkers in selected orthotopic or metastatic models, such as prostate-specific antigen (PSA) in the LuCap 23.1 (human prostrate carcinoma) or CA15-3 in the MDA-MB-231 (human breast carcinoma) models can greatly aid staged studies in these xenograft models (see Section 7.02.8). In addition, advances in small-animal imaging technologies (for instance, dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), computed tomography (mCT), etc.) will allow these approaches to be an increasingly important component of the preclinical evaluation of experimental cancer agents.

7.02.4.1.4 Transgenics, knockouts, and in vivo gene targeting

With the advent of transgenic and gene targeting technologies in mice, there are now available an increasingly sophisticated variety of genetic tumor models.22 While scientifically allowing the precise control of disease-associated genes, these genetic model systems often have increased complications associated with phenotypic analysis. Further, these investigations require large cohorts of tumor-bearing animals for efficacy studies that require analysis of multiple agents, doses, and schedules. Currently, given the time and expense of these models, only very specific targets or dedicated biology-driven projects are likely to find this approach helpful. However, as our ability to generate these models and analyze their phenotypes increases, these approaches will be an increasingly important component of the preclinical evaluation of experimental cancer agents.

7.02.4.1.5 Predictive value of murine in vivo models

During the last several years, there has been much argument and counterargument regarding the validity of current tumor model systems (particularly subcutaneous or flank models) as predictors of clinical activity.23-27 A recent review demonstrated that, for 39 anticancer agents, those with activity in at least one-third of xenograft models tested also had significant activity in phase II trials.28 Further, all currently registered cancer drugs have shown activity in preclinical animal models. However, antitumor activity in a xenograft model representing a specific histological type within a given organ (e.g., colon adenocarcinoma) does not necessarily translate into clinical activity in that organ type.28 While animal models are most important in showing the spectrum of activity, they have limitations and may not be fully predictive due to many confounding factors, outlined below.

It should come as no surprise that immortalized tumor cell lines grown for many generations on a plastic surface (or even in vivo) are not likely to reflect perfectly the complex biology of spontaneously arising tumors in humans. Clearly there are limitations with how extensively xenograft models can mimic factors such as tumor blood supply, tumor-stromal interactions, genomic integrity, and other processes that influence the ability of any potential drug to impact tumor growth. Another problem unique to model development for cancer drug discovery is that the disease is an outgrowth from normal cells and tissues, yet effective therapy requires killing (or at least significant inhibition of growth) of cancer cells without dramatically compromising the well-being of the host cells. Part of the tremendous challenge of developing neoplastic biological model systems is that cancer represents more than 100 diseases, each with biology unique to the site from which it arose. For experimental agents where there is precise knowledge of the tumor type(s) that are likely to be highly dependent on the therapeutic target, it is possible (in some cases) to develop quite sophisticated tumor models that would be more likely to reflect the clinical setting.29 However, in the vast majority of cases the prediction of tumor types likely to be dependent on the target of interest is associated with a high degree of uncertainty. This problem, in fact, may be addressed by one of the significant benefits of xenograft model systems in that literally hundreds of cell lines representing most major and minor tumor types exist for potential study.

It is worth discussing in some detail the factors that contribute to the perception that existing tumor models have limited value as predictors of clinical activity. Part of the problem lies with the lack of rigor with which efficacy studies are often conducted. It is not uncommon to see a report claiming significant activity in a tumor model, yet examination of the data shows that, at best, no conclusion about activity can be drawn.10 In addition to the factors described above (appropriate sample size, accounting for tumor variability and ulceration), it is important that the evaluation of efficacy is conducted in a biologically relevant fashion. Reporting a difference in tumor volume between treated and control groups when the control groups have only grown to a volume of a few hundred cubic millimeters or less is unlikely to be informative. The method of bilateral inoculation of tumors in order to increase sample size should also be avoided, as it appears that the growth of one tumor can sometimes influence the growth of the other (perhaps due to effects on tumor angiogenesis), in ways that are not predictable.30 For many tumor lines there now also exist a variety of cytotoxic agents that are known to show significant antitumor activity in a particular xenograft model (Table 1).31 Use of these compounds side by side with the experimental agent not only serves as a powerful benchmark to judge activity, but they are also important positive controls for the numerous variables that can affect tumor growth from one trial to the next.

Another important criterion is defining the concept of 'significant activity' for an experimental agent in a tumor model. Depending on the treatment group sample size and the variability of growth in a given model, one could expect that a treatment-induced %T/C of 50-60% (40-50% TGI) would register as statistically significant. If this level of TGI is achieved at only a single measurement during the course of the trial, the effect should be considered of limited value. However, if this level of inhibition is observed throughout the treatment period and can be accurately reproduced from one trial to the next, then a significant biological effect can be argued more convincingly. Does this level of inhibition signify a level of efficacy that could translate to a clinical response? For a purely cytostatic agent that has limited toxicological side effects and can therefore be administered for long duration, the answer may be 'yes,' although the ultimate benefit of such agents in clinic may be in combination with cytotoxic drugs. Numerous clinical trials are currently ongoing to test this hypothesis.32'33 However, in the context of the more traditional way in which clinical activity is measured for cytotoxic agents, the answer is almost certainly 'no.' Given that a 50% reduction in tumor mass is generally considered the clinical standard for partial response (PR), then a 40-50% inhibition of tumor growth rate would almost certainly translate to progressive disease.10,26 A more robust measure of significant efficacy for a bona fide drug candidate in xenograft models is tumor regression (or at least complete tumor stasis), observed at the end of the treatment period.18 It seems logical to argue that the more pronounced the regression and the more durable the response, the more likely the activity is to translate to clinical effect. Of course, even when tumor regression is observed, a reasonable therapeutic window must exist and the translation of activity for agents used in combination can be much more complex. Toxicological evaluations of experimental anticancer agents are typically quite rudimentary during the early preclinical stages. Dose-dependent antitumor activity is usually reported relative to the frequency of deaths or significant weight loss (typically 15-20% of total body weight) within the efficacy trial. However, in some cases where modulation of target activity is expected to have a specific (target-related) toxicological liability, it is worth examining these potential limitations extensively early in the discovery process.34,35

Although 40-50% inhibition of tumor growth is generally not sufficient efficacy to declare an experimental agent ready for clinical trial, it nonetheless can be quite informative to identify appropriate model(s) that are critical for subsequent medicinal chemistry refinement of the lead. Identifying an appropriate model system is vital for efficient evaluation of potential improvements in potency and PK that ultimately result in a drug candidate with more robust efficacy.

In addition to defining the significance of the level of activity in a given model, one must also consider how many models need to be examined in order to try to extrapolate preclinical efficacy to clinical response. This question, of course, needs to be considered in the context of the pathway that is being targeted. Thus, molecules designed to inhibit receptor tyrosine kinase (RTK) signaling critical in angiogenesis are likely to be active in a variety of tumor types, as these pathways have been shown to be important in a broad range of neoplasms.36,37 Indeed, a number of these RTK inhibitors have been described that do show broad activity in xenograft tumor models. Several of these inhibitors have entered clinical evaluation and are showing promising activity as monotherapy and/or in combination with other drugs.36,37 For other targets, efficacy may be expected to be limited to more specific type(s) of tumors. In some cases, however, it has been shown that significant activity in one model that expresses the target is likely to be predictive of clinical response for that class of cancer (e.g., Herceptin).

As has already been mentioned, identification of the appropriate xenograft tumor type(s) for efficacy evaluation is critical for effective lead identification and refinement. This process can be greatly assisted by learning as much as possible about the biology and genetics of the xenograft tumors. The pediatric oncology community has recently established a consortium for evaluation of promising therapeutic agents in fairly large panels of xenograft models representing a variety of tumor types as a key component for selection of drugs for clinical evaluation.26 Although it is not always possible to trace the clinical origin of each line, it is often helpful to know if the line was isolated from a primary or metastatic lesion (and if so, from what site), and the treatment history prior to isolation. For example, in some cases multiple tumor lines have been established from the same patient: this is an important factor if one wishes to evaluate efficacy in a panel of models of a certain tumor type.38,39 In addition, long-term culture of cell lines also carries with it the inherent risk that cells may develop changes in phenotype or may become cross-contaminated 40

over time.

10 Ways To Fight Off Cancer

10 Ways To Fight Off Cancer

Learning About 10 Ways Fight Off Cancer Can Have Amazing Benefits For Your Life The Best Tips On How To Keep This Killer At Bay Discovering that you or a loved one has cancer can be utterly terrifying. All the same, once you comprehend the causes of cancer and learn how to reverse those causes, you or your loved one may have more than a fighting chance of beating out cancer.

Get My Free Ebook


Post a comment