PM360 March 2010
OPTIMIZING BRAND PERFORMANCE

FORECASTING A NEW ONCOLOGY PRODUCT
BY DAVID ROBINSON

THE PROCESS OF FORECASTING FOR ONCOLOGY PRODUCTS IS MORE COMPLEX than for other therapeutic areas. When forecasting new oncology products, you should consider a number of basic steps that can add layers of complexity, including:

• The caveats of the different types of data used to quantify the relevant patient populations or subpopulations
• Accounting for disease recurrence and disease progression
• Lines of therapy (for example, first-, second-, and third-line)
• Significance of combination therapy and possible off-label usage
• Duration of therapy
• Complexity in dosing regimens (for instance, multiple cycles or variable dosing per cycle)

The most critical step in oncology forecasting is defining and quantifying the most accurate measure of the target patient
population.

In general, the starting point for a patient-based forecast is investigating the epidemiology of the target tumor type. To analyze the relevant patient population, a “top-down” approach is recommended. Oncology data sources are referenced for either number of new cases (incidence) or number of existing cases (prevalence) during a given interval. Each source has its own caveats.

By definition, incidence is the number of patients who contract a disease in a given period (usually a year), theoretically regardless of whether they are diagnosed. Because cancer patients are not tracked from the first appearance of a malignant cell, they are regarded as incident when they are diagnosed. Cancer incidence underestimates the treatable patient population because it accounts for only newly diagnosed patients, thus missing recurrent or progressed patients.

Prevalence refers to all surviving patients who have a disease. Survival in oncology is quoted in terms of years from diagnosis. So, five-year prevalence refers to all patients who have been diagnosed within the last five years. In this case, prevalence can overestimate the treatable population because it includes patients with early-stage disease who may be in remission and thus not requiring treatment.

The distribution by stage in standard prevalence is biased toward early-stage disease, where drug use is relatively low compared with late-stage disease.

Neither incidence nor prevalence necessarily provides a realistic starting point in oncology. It is critical to take into account disease progression and recurrence. Progressed and recurrent patients need to be reclassified by stage. Yet in clinical practice patients are not generally restaged when they present with disease progression. Therefore, calculating the progression needs to incorporate a number of factors. By calculating progression and recurrence, we can more accurately depict a patient’s status at that point in time.

However, just to add another layer to the complexity of the oncology forecast, this restaging of prevalence is not the most accurate portrayal of treatable patients because it too includes patients in early-stage remission. Therefore, a measure of the true treatable population is still needed.

True Treatable Population
In general, cancer patients typically receive treatment:

• When they are newly diagnosed
• When disease progresses or recurs
• When they have survived with their disease and new or different treatment options become available

Kantar Health uses proprietary projection models to measure just those patients who are true treatment opportunities, including incident (newly diagnosed) patients in Stages I through IV, patients whose disease has progressed or recurred at the same or a different stage or site, and any patient who has been diagnosed as having Stage IV disease in the past (by definition, a Stage IV patient is never disease-free). Figure 1 (above) illustrates the flow diagram of Kantar Health’s Active Disease.

Another distinct advantage of the top-down epidemiology-based methodology is that it can accurately assess those patients not treated at any given time. Forecasts that project up to a treatable patient universe using data derived from patients seeking treatment do not account for this untreated patient segment. This is an important component in assessing drug treatment opportunity and is absolutely essential to assess first-line therapy potential.

Using this top-down approach, we are able to assess a realistic number of actual treatable patients to further segment by stage and line of therapy, having accounted for:

• The understatement of incidence
• The overstatement of prevalence
• Those who don’t receive treatment
• Those who have moved into remission and are not eligible for treatment at this point in time

We also take into account those patients who reenter the treatable patient pool due to disease recurrence. Thus, calculating an Active Disease figure gives you the most accurate starting point for considering or forecasting the potential treatable patient population. This is just the beginning. Subsequent steps require knowledge of disease and treatment, such as stage and line of therapy, which must be taken into consideration to build an accurate forecast of appropriate subpopulations.

To download Optimizing Brand Performance Tools for this issue’s topic, go to www.PM360online.com/tools

David Robinson is Senior Director of Epidemiology Services with Kantar Health. He has headed up the Epidemiology Services group since 2004. He can be contacted at david.robinson@kantarhealth.com