Radiation therapy delivery is dependent upon accurate identification of the tumor target. Anatomic imaging has limited ability to define tumor borders, particularly within the pelvis. Functional imaging strategies (i.e. Magnetic Resonance imaging (MR) and positron emission tomography (PET)) take advantage of tumor-specific physiologic, micro-environmental and metabolic changes to localize areas of active tumor growth. We use a well annotated clinical database to evaluate the prognostic value of imaging metrics from 18F-fluoro-deoxy-glucose (FDG)-PET and MR imaging in cervical cancer. Samples from a prospectively collected tumor banking study are used explore how tumor biologic changes (i.e., mutations, gene expression, protein modifications and metabolites) are related to PET and MR imaging metrics.

Bioluminescence of female SCID mouse implanted with SiHa-CBG-GFP tumor in the cervix at Day 5, 15 and 63 post implant.

In the lab, we use a variety of preclinical tumor models in mice to study the interface between functional imaging characteristics and response to cancer treatment. Bioluminescent human cervical cancer cells can be implanted directly into the cervix of immune compromised mice. These tumors grow, invade local structures and metastasize in patterns consistent with what we see in cervical cancer in humans. At various stages of cancer progression, we treat these tumors with biologics, chemotherapies +/- focused radiation and monitor treatment responses non-invasively using bioluminescence, microPET and microMRI.

John Floberg, MD, PhD presenting his work at RSNA

Most recently we have used this tumor implant system to study how developing PET imaging tracers, such as CuATSM, can be used to predict the level of oxidative stress present in tumors prior to the administration of radiation, chemotherapy and other metabolically targeted drugs (Floberg et al, JNM 2019).  Most importantly, using tumor implants, we can manipulate the tumor cells in vitro prior to implantation to directly test the function of individual genes and to evaluate the effects of common mutations on imaging and treatment response phenotypes. Using this approach, personalized cancer treatments can be better selected and adapted through the course of treatment via incorporation of both biologic and imaging data.

To complement our tumor cell implant approaches, we have developed a library of patient derived xenograft models (PDXs) using human tumor specimens from our cervix tumor bank.  These models include primary, recurrent and treatment refractory cervical cancers sampled at multiple time points during treatment, which allows us to study how tumors evolve during standard of care cancer treatments.  In addition to our implant models which grow in immunocompromised mice, we are now using genetically engineered mouse tumor models (GEMMs) for more focused study of the impact of standard of care cancer treatments on tumor cells and cells within the tumor immune microenvironment.  In addition to focused study of cervical cancer, through new collaborations, we are now studying radiation and chemotherapy effects using pancreas and ovarian cancer tumor models.

Read more about John Floberg, MD, PhD, and his time in the Schwarz lab as a Holman Pathway Resident here and here.

Spin echo and diffusion weighted images of a mouse 33 days after tumor implantation.
Spin-echo images of a mouse 26 days and 33 days after orthotopic tumor implantation.