Mouse Patient Avatars Serve as Improved Models of Cancer Therapy Resistance

Physicians still cannot accurately and consistently predict whether a cancer patient will completely respond to a certain therapy, have a partial response or not respond at all. However, Cleveland Clinic researchers, led by Mohamed Abazeed, MD, PhD, are beginning to pinpoint certain genetic abnormalities that contribute to making specific types of tumors resistant to treatment.

The team is about 18 months into its research using patient-derived cell lines from solid tumors grown in culture and, in some cases, mouse xenografts as patient avatars. Dr. Abazeed says the cell-line information indicates which therapy may be most effective, and each xenograft provides a specific individual test bed. The team now has proof of concept and will begin to ramp up its testing efforts, he says.

The benefits of scaling up

“We want to be able to go from 500 cancer cell lines to 5,000,” Dr. Abazeed says. “And using the resources here at Cleveland Clinic — the patient volume, the incredible infrastructure — I think we’ll be able to do that.”
By scaling up, the team hopes to learn more about the complexity of resistance and develop better models to predict the efficacy of cancer therapies.

“Is it a single mutation that confers resistance? That seems unlikely. There are probably multiple genetic events interacting in a complex network to promote resistance. Only by scaling will we have the power to determine the contours of this network,” he says.

The team’s effort is focused on identifying “omic” determinants — otherwise known as changes in DNA sequence (mutations, rearrangements, copy number alterations), the transcriptome (coding and non-coding RNA), and the epigenome (DNA methylation and histone modifications) — that correlate with responses to cancer therapeutics.

Bell-curve results in lab reflect natural underlying biologic diversity

Dr. Abazeed’s research approach initially used more than 500 genetically diverse, patient-derived cell lines from a variety of solid tumors. Grown in culture, these cells lines were comprehensively genetically profiled and tested to see how they respond to radiation or particular drugs. “We and collaborators have developed computational algorithms based in information theory that allow us to couple the resistance of these tumors with particular genetic alterations,” he says.

The researchers analyzed the presence, type and frequency of genetic changes in 26 cancer lineages, including cancers from multiple anatomical site (like breast, lung, uterine, et cetera), and used that information to estimate the likelihood that a patient would respond to therapy. They found a probabilistic — or bell curve — distribution of the prospects of a therapeutic response.

“Our results demonstrate that it is almost by chance whether a cancer patient is going to be a strong responder, a weak responder or an average responder,” Dr. Abazeed says. “It’s something that I think most physicians suspected all along. When we irradiate a patient, we don’t know whether their tumor is going to disappear, progress or show a partial response. The data actually bear that out.”

Many variables likely regulate whether a patient is going to respond to a particular therapy. “To identify these variables, we search for the genomic features of the cancer that correlate with sensitivity and resistance,” says Dr. Abazeed.

Approaching more detailed results

In their genetic analysis of cancer lineages, the researchers have found that expression and activity of the androgen receptor in a subset of breast cancers confers resistance to chemotherapy and radiation. In other cancers, the extent of chromosomal copy number alterations correlates with resistance to chemotherapy and radiation, Dr. Abazeed says. There are additional intriguing correlations that help map the genomic landscape of resistance to current therapies. These results are awaiting publication, he says.

Part of the current effort includes the identification of lower-frequency events that confer resistance. The team combined cancer genomic data with their recently developed high-throughput platform for measuring radiation survival. In adenocarcinoma of the lung — the most aggressive type of non-small cell lung cancer (NSCLC) — the researchers identified BRAF mutations as predictors of resistance to radiation therapy, Dr. Abazeed says. These data indicate a potential role of BRAF pathway inhibitors as radiation sensitizers in the treatment of BRAF-mutated lung adenocarcinoma. The Cleveland Clinic team presented these results at the 2015 American Association for Cancer Research annual meeting in Philadelphia.

Mouse models aid study of resistance in small cell lung cancer therapy

To test the results of the genomic cell line studies in a more physiologically relevant setting, the team created patient avatars by injecting samples obtained from patients’ tumors into mice. They now have an inventory of more than 40 primary lung cancer xenografts.

Craig Peacock, PhD, who is leading this effort in Dr. Abazeed’s lab, is using these models to study resistance of small cell lung cancer (SCLC) to chemotherapy. SCLC comprises about 10 to 15 percent of lung cancers and tends to grow and spread quickly. In the clinic, SCLC responds initially to chemotherapy but eventually almost invariably comes back.

Dr. Peacock and his team have been able to model that clinical behavior in their mouse model using cisplatin/etoposide chemotherapy.

Currently, the researchers have 15 SCLC samples injected and growing in xeongrafts. Some become resistant to chemotherapy very quickly, while others respond to the therapy even up to the third cycle. “This tells us that the patients are different from the very beginning. We see primary resistance in which the cells are resistant from the start, and secondary resistance, which develops after one or two cycles of chemotherapy,” explains Dr. Abazeed.
“If you can predict the group of patients that are going to be resistant to a particular chemotherapy regimen upfront, you may want to adapt the clinical paradigm for these patients,” he says.

In cancer in general and lung cancer especially, a very low percentage of drugs from the initial phase I studies receive FDA approval (Nat Biotechnol. 2014 Jan;32(1):40-51).

“The failure rate is very high from the point when you think you have a great therapy in the preclinical setting to eventually making it into the clinic,” says Dr. Abazeed. “There are multiple factors, including toxicity. But the phase II level is where most clinical trials fail because the current models are poor at predicting efficacy in a patient.”
Better models in cancer therapy are urgently needed.

“We think we can learn a great deal about why cancer patients become resistant using our avatar models,” explains Dr. Abazeed. “By targeting these resistance factors, we hope we can prevent resistance and improve the efficacy of current therapies.”