Imaging of Relevance

• Better expression of clinically relevant genetic drivers and stem cell populations
  
• Localisation of the tumours in environments/locations that result in relevant stromal interactions and drug exposures
  
• The generation of a truly malignant phenotype, including the ability to invade neighbouring tissue and metastasize to distant sites with relevant physiologic consequences
  
• Evaluating human cancers in intact mice by so-called humanisation of the immune system

Model Formats

Key models promoted as more clinically relevant include orthotopic implants of human tumour xenografts, PDX and GEMs. All three of these formats are typically more prone to invade and metastasize, thus better modelling malignant disease.

Orthotopic xenografts, regardless of the source of the tumour sample, place the tumour in sites typically encountered in the clinic – usually in the organ in which it arises – providing more relevant drug exposures, environments and stromal interactions. (2,3).

GEMs are precisely established with carefully chosen genetic drivers that promote the formation of tumours in relevant environments (4). PDX models show strong ability to recapitulate the treatment outcomes of the patients from which they are derived; they are also more clinically relevant in terms of maintenance of relevant histology, stem cell populations, genetic drivers and heterogeneity – especially in an orthotopic format –but, until recently, have not been widely available. Orthotopic PDX models are arguably the most clinically relevant models available today (5).

All of these model formats are showing promise in terms of recapitulating clinical experience, but they share a common property that hinders efficient and large-scale use in development: the tumours and response to treatment are difficult to diagnose and measure.

Overcoming Barriers

Fortunately, the advancement of in vivo imaging technologies, in parallel with these more relevant models, has enabled non-invasive detection and quantification of tumour burden. The application of imaging technologies makes these models much more robust and efficient, with more clinically relevant end-points, and thus more appropriate for broad application in the discovery process (6,7).

Orthotopic models have the advantage of a known location, at least for the primary tumour. GEM models, on the other hand, are stochastic, even if organ specific, and can vary in precise location, time of formation and multiplicity.

Nevertheless, several imaging technologies are applicable and can be honed to deal with both model types. Traditional and clinically translatable anatomical imaging techniques such as magnetic resonance imaging (MRI), computed tomography (CT) and ultrasound can be used to noninvasively locate and quantify tumour burdens in preclinical animal models. Furthermore, they can be used to assess tumour progression over time, response to therapy, and eventual relapse or emergence of resistance. MRI and CT, in particular, offer high spatial resolution and accurate determination of tumour volume. These modalities are most easily applied when the location of the tumour is already known, and therefore are used most with orthotopic models or GEM models with specific organ tropism.

Tracking Metastasis

A hallmark of cancer is metastatic spread. This can be assessed non-invasively in preclinical studies by MRI or CT imaging, if the target tissues for metastasis of the particular model have been predetermined. CT imaging is particularly useful for the assessment of lung metastases, where incidence, multiplicity and mean (as well as individual) lesion sizes can be simultaneously determined. However, other imaging modalities receive greater use for this application. These include positron emission tomography (PET), fluorescence molecular tomagraphy (FMT) and bioluminescence imaging (BLI). Despite having relatively poor spatial resolution, these techniques have high sensitivity, and allow whole body imaging for detecting and tracking primary tumours and metastases.