Epiphany

epic oncology™

 
epic oncology™
Advanced Methodologies
THIS HAS TO CHANGE!!!!
Advanced methodologies employed by Epiphany are what really set epic oncology™ apart from other cancer epidemiology data sources. Following is a brief overview of several of our novel approaches.

Patient Flow Modeling

Epiphany’s patient flow model is different from other vendors because we have separated the concept of relapse/recurrence from progression. While the two clinical functions are inter-related, they should be modeled separately: some patients may relapse and progress at the same time (coupled), but some will also relapse and not progress (locoregional relapse, separate functions). It is important to have both functions in a detailed flow model.

The following graphic shows how the different concepts of response, relapse, and progression are aggregated into a flow model design. In some instances, the concept of progression is dependent upon relapse following response; for other patients, relapse is not an option because the patients did not respond, but progression is still an independent variable.

image_group

The following is a detailed flow model graphic that presents the actual flow used in Epiphany’s 2012 model structure.

image_group

Line of Therapy vs. Course of Treatment

Epiphany’s flow model is constructed using “course of treatment” as the primary order for the flow: first course or initial treatment, second course after first relapse or non-response, third-course after second relapse or non-response, etc. The course of treatment is counted irrespective of the modality(ies) utilized. It is simply the count of the number of treatments over time.

However, when talking about a specific modality (for example, drug or chemotherapy), we use the phrase “line of therapy.” In the context of a specific modality, the line of therapy is the order in which the patient receives each round of that modality. It is important to understand that the course of treatment and line of therapy may not correspond.

For example, a breast cancer patient may receive the following modalities in her first three courses of treatment. When looking specifically at the line of chemotherapy, the first course has no chemotherapy. The first time the patient receives chemotherapy (first-line chemotherapy) is in the second course of treatment.

Course of Treatment Modalities Line of Chemotherapy
First Course
Surgery, Radiotherapy
Second Course Surgery, Radiotherapy, Chemotherapy First Line
Third Course
Chemotherapy only Second Line

 

image_group
Epiphany's model takes into account these levels of detail to provide the most rigorous patient estimates available.

Incidence Forecasting

Current incidence data source, including SEER, have source-specific issues. As such innovative approaches to define the “best available” data are important.

Epiphany uses both meta-analyses and mortality-based incidence forecasting, depending on the country, to define the best historical incidence data. In addition, Epiphany uses a more recent time-based analysis to generate the best incidence trend (given that many incidence data sources are from the 2000 time frame).

SCREENSHOT OF META-ANALYSIS
Example of SEER Meta-analysis to Define “Best Historical Data”

SCREENSHOT OF TREND ANALYSIS
Example of Mortality-Based Incidence Trend Analysis

Subset Analysis

Given that newer indications are becoming more niche and can be driven by the specific biomarkers, histologies, and/or mutations, it is important to understand that each subset may have different incidence and survival trends. The following graphic illustrates why subset analysis is critical to understand future populations. The dotted line is the outcome when the U.S. female non-small cell lung cancer incidence is forecast in aggregate compared to the green line which shows the incidence forecast when conducted by subset and then added together. Without subset analyses, future forecasts can have significant additional error inherent in the estimates.

SCREENSHOT OF LUNG CANCER CHART

Historical Data

Many modelers use only the most recent survival data resulting in skewed estimates of historical populations. As the following graphic illustrates, CML survival in the U.S. improved dramatically after the adoption of tyrosine kinase inhibitors. If a modeler only uses the most recent survival (~50% 5-year observed survival), it would significantly overestimate the number of patients in earlier years when the document survival was only around 30% 5-year observed survival.

LEUKEMIA CHART
Trend Line illustrates the need to use both current and historical survival data

Beyond Standard Survival

Epiphany uses several additional analyses to take “average” survival and generate course-specific outcomes. Since cancer patients who respond to their initial treatment have better outcomes than those that are multiply relapsed, Epiphany has made its best attempt to replicate this behavior in our detailed epidemiology models.

SURVIVAL BEHAVIOR CHART
Separating survival and other assumptions by treatment course best replicates how cancer patients actually behave.

For a more detailed discussion of our methodologies, please contact Epiphany Partners at support@time4epi.com.

Epiphany Partners, Inc.
1900 South Norfolk Street, Suite 260
San Mateo, California 94403
Main 650-242-4626
Fax 650-288-0036
info@time4epi.com 
Driving Directions
Follow us: LinkedIn
©2013 Copyright Epiphany Partners, Inc.  •  Terms of Use  •  Privacy Policy  •  Site Map