How Image Analysis Can Improve The Results Of Drug Development And Clinical Trials
By Thomas Nifong, MD, Definiens Inc.
In recent years there have been several potentially life-saving medications approved for cancer treatment, including targeted molecular entities and biologics such as Opdivo (nivolumab) and Keytruda (pembrolizumab). Oncology drugs remain a pharmaceutical priority and investments into cancer account for 30% of all pre-clinical and phase 1 clinical development expenditures. There is an impressive list of close to 800 drugs and vaccines currently in the industry-wide development pipeline, many with promising results in early-stage clinical trials. However by historical measures only 10% or fewer of these drugs will ever make it through FDA approval and become part of routine patient care.
The cost of failed drug trials places an enormous burden on the pharmaceutical industry, especially in the current environment where targeted therapies focus on smaller populations of patients. Each approved drug has a smaller market to recoup drug development costs, which leads to higher priced drugs that face patient and payor pushback and reimbursement uncertainty.
The cost of drug development increases substantially as it moves down the pipeline, with the average cost for an oncology drug increasing from $5M for a phase 1 trial to over $20M for a phase 3 trial. Costs can be reduced by eliminating unsuccessful drugs early in the development process, and by focusing trials for drugs most likely to succeed on patients with tumors that are most responsive to the therapy. This can be accomplished through improved biomarker studies which can profile drug targets and pharmacodynamic markers, improve understanding of drug mechanisms, and predict which patients and tumor types are most likely to respond to a drug. Image analysis can help to optimize biomarker studies and improve drug development through these mechanisms, particularly in the field of immuno-oncology.
Immuno-oncology uses drugs referred to as immunotherapies that target and modulate the patient’s immune system to help it fight cancer. The immune cell composition and immunomodulatory biomarker expression levels in the local tumor microenvironment offer clues about the tumor-immune cell interaction and how they may be affected by immunotherapies. These clues can be unlocked on immunohistochemically stained tissue sections by using image analysis to accurately identify and quantify tumor cells, immune cells such as cytotoxic T cells (CD8+), regulatory T cells (FoxP3+) and macrophages (CD68+), specific immunomodulatory biomarkers such as PD-L1, and then calculating the spatial relationships between cells. Image analysis can measure these cellular parameters across tissue sections to characterize heterogeneous tumor and immune infiltrates and export all of the cellular variables into a database, along with clinical measures such as drug response, disease recurrence and patient survival. Tissue Phenomics® can then be used to mine this rich quantitative digital information to discover biomarkers that improve the drug development process and clinical trial success.
For example an immuno-oncology therapy may show an increase in cytotoxic T cells infiltrating the tumor in preclinical models as an important mechanism of action. Image analysis can be used to quantitatively compare pre- and post-treatment cytotoxic T cell infiltrates in early stage clinical trial samples to further understand the drug mechanism and provide actionable information on drug effect much early than can be seen by clinical measures. If the expected change in the pharmacodynamic biomarker is not seen then adjustments in dosing or drug prioritization can be made earlier in the drug development cycle, thus saving costs and allowing resources to be concentrated in the drug programs with the greatest likelihood of success.
Immuno-oncology is notable for combining two or more therapeutic compounds that modulate different immunologic pathways. Image analysis can measure several immune-related biomarkers and their relationships within the tumor context and may help researchers to rationally design combinations of drugs that have complementary mechanisms.
Many early phase immuno-oncology trials are “basket studies” meaning that patients with many different types of cancer are enrolled without prior knowledge as to which indications are most likely to respond to the therapy. Often image analysis can be used in advance of expensive clinical trials to analyze banked tissue samples to determine which cancer indications express higher levels of the drug target or have more favorable tumor microenvironments. Pharmacodynamic biomarkers can also be analyzed by image analysis in pre- and post-treatment samples across indications to discover much earlier which ones are biologically responding to the therapy in the expected manner. These image analysis applications allow researchers to focus drug trials on the cancer indications most likely to respond to therapy and minimize enrolling patients with tumor types that are not likely to respond.
Discovering predictive biomarkers, or companion/complementary diagnostics, to select the right patient for the right drug is the holy grail of personalized medicine. Even though the current immuno-oncology drugs are helping many patients through durable responses and improved survival, the benefits can only be achieved in a minority of patients. In order to obtain FDA approval it is necessary to develop predictive biomarkers that can stratify patients into treatment groups that are enriched for the likelihood of therapeutic response and improved survival. Tissue Phenomics can be used to evaluate image analysis-based variables in the tumor microenvironment along with clinical measures to discover novel predictive biomarkers that are associated with positive patient outcome. These predictive biomarkers can result in smaller trials with higher drug response and survival rates among treated patients saving substantial money and time and improving chances for drug approval and commercialization.
In summary, image analysis can improve the results of drug development and clinical trials, especially in the field of immuno-oncology. Image analysis can be used to discover and measure pharmacodynamic and predictive biomarkers that can help to advance or fail drug programs earlier, rationally design immune-oncology drug combinations, determine which tumor indications are most suitable for a given drug, and stratify patients in clinical trials to enrich for drug response. These decisions can lower the cost of drug development and improve the likelihood of getting the right drug to the right patient at the right time.