Forecasting Oncology Sales in a Changing Environment

1
Forecasting Oncology Sales in a Changing
Environment

• Pre-Conference
  WorkshopSan Diego, 22nd May 2004
• Petra MärtensAssociate Director International
  Studies
• - Basel (Switzerland)

2
Topics covered

• Specific challenges in forecasting oncology sales
  based on the complex nature of this particular
  market
• Using secondary data criteria to assess the
  reliability of secondary data sources
• Introduction to workshop case studies
• Five common pitfalls in oncology forecasting
• Five primary steps of oncology forecasting
• Case Study oxaliplatin in colorectal cancer
• Questions Answers

3
Clinical Research
WHO / IARCGLOBOCANEUCAN
Proprietary Cancer Patient Databases,e.g. GCPM
?
Know-How
Primary Research
Statistics
We will provide you with a road map to find the
optimal sources of information on the oncology
marketplace
4
Definition of a Tumor
What is Cancer?
5
Progression of Cancer
6
Most Common Cancer Sites in ? and ?- Mortality
in Europe and USA -

• Female deaths/ year
• Breast 170000
• Lung 130000
• Colorectal 100000
• Ovary 40000
• Stomach 30000
• Pancreas 20000

Male deaths/ year Lung 270000 Colorectal 96000 P
rostate 89000 Stomach 58700 Bladder 36000 Oesop
hagus 35000
7
Incidence of the Most Common Cancers
worldwide
Australia
Germany
Canada
Finland
Japan
China
USA
UK
Source WHO Cancer Incidences, Vol. VIIcrude
rate per 100000 population of the corresponding
country given in ()
8
The Global Demographic Development
Source US Bureau of the Census. International
Programs Center,International Data Base
By 2025 the population in those age categories
susceptible for cancer in developing countries
will outwigh the population in industrial
countries.
9
Cancer is not one single disease, but represents
a variety of different diseases requiring
individual therapeutic solutions
ER/PR positive breast cancer in postmenopausal
women
Refractory multiple myeloma
Hormone refractory prostate cancer
Indolent non hodgkins lymphoma
locally advanced or metastatic nonsmall-cell
lung cancer after failure of both platinum-based
and docetaxel chemotherapies
EGFR expressing metastatic colorectal
cancerrefractory or intolerant to
irinotecan-based chemotherapy
10
New Drug Classes entering the Market
Monoclonal Antibodies
Tyrosine Kinase Inhibitors

• Rituximab (MABTHERA Genentech/Roche)
• Trastuzumab (HERCEPTIN Genentech/Roche)
• Alemtuzumab(MAbCAMPATH Schering)
• Cetuximab (ERBITUX BMS/E.Merck)
• Bevacizumab (AVASTIN Genentech/Roche)

• Imatinib (GLIVEC Novartis)
• Gefitinib (IRESSA AZ)
• Erlotinib (TARCEVA Genentech/Roche)
• Vatalanib (Schering/ Novartis)

11
Factors that might Influence the Oncology
Marketplace in the Future - I
Demographics

• The worlds ageing population will lead to an
  increase in cancer incidence and prevalence
  (especially in developing countries)

Screening

• A broader use of screening procedures could lead
  to an earlier detection of certain tumors which
  might result in a higher number of patients
  eligible for treatment but also in higher cure
  rates
• Cervical Cancer Pap Test
• Breast Cancer Mammography
• Prostate Cancer PSA Test
• Colorectal Cancer Endoscopic procedures

12
Factors that might Influence the Oncology
Marketplace in the Future - II
Life Style

• Changes in life style might reduce or increase
  the risk to develop cancer, eg.
• Anti-smoking campaigns in various countries might
  stop lung cancer
• Connection between Sushi bars and stomach cancer?

Treatments

• New treatments might increase cure rates.
• New anti-cancer drugs might be used over a longer
  period and thereby
• contribute to extended survival time and increase
  in both quality of life and number of treatable
  patients

13
Increase of Costs
In use since 1960s 1984 1996 2000
Source Cancer Patient Monitor 2002, PROPHARES
(formerly IG Suisse) Rote Liste 2003
14
Topics covered

• Specific challenges in forecasting oncology sales
  based on the complex nature of this particular
  market
• Using secondary data criteria to assess the
  reliability of secondary data sources
• Introduction to workshop case studies
• Five common pitfalls in oncology forecasting
• Five primary steps of oncology forecasting
• Case Study oxaliplatin in colorectal cancer
• Questions Answers

15
Key Sources of Information to assess the Oncology
Market
Databases

• on the treatment of cancer
• The Global Cancer Patient Monitor(covers the
  treatment of all cancer types in the 7 Key
  markets US / Europe / Japan)
• Other Oncology Databases(available for Europe
  and/or the US and/or Japan)
• Specific Reports / Research Projects covering
  cancer treatment for a single disease or in a
  single country

0ther Sources

• Epidemiology Data
• Cancer Research Groups (Universities)
• Health Insurances
• Pharmaceutical Companies
• Health Research Agencies(for ad-hoc research
  projects)

16
Epidemiology Data
17
Sources for Epidemiology Data
Publications
Proprietary Sources

• Individual Cancer Registries
• Very detailed information available for the US
  (SEER)
• In other countries only parts of a country are
  covered and the available information is less
  detailed
• WHO / International Agency for Research on Cancer
  (IARC), Lyon
• Based on co-operation with the International
  Association of Cancer Registries
• Cancer Incidence Vol I-VIII, Globocan and Eucan ?
  do not cover subgroups (e.g. for certain cell
  types, tumor stages)

• Cancer Epidemiology Database (CEDA), Prophares
• Based on WHO data
• Includes breakdown by tumor subgroup, country,
  gender
• Epi Database, Mattson Jack Group
• Includes breakdown by stage

18
The Ideal Cancer Epidemiology Database

• Allows differentiation of tumor by morphology
• NSCLC vs. SCLC, AML vs. CML, ALL vs. CLL,
  astrocytoma vs. glioblastoma
• Allows differentiation by patient demographics
• Gender
• Age
• Allows differentiation by stage of disease
• Provides information on incidences, prevalences,
  survival data, mortality and population data

19
Three alternative and complementary Research
Options

• Desk Research
• Qualitative Research
• Quantitative Research

20
Option 1Desk Research
To gather as much relevant statistical data and
background information as available from
secondary sources
Would this information answer your questions?
21
Option 2Qualitative Research
To develop an accurate understanding of the
disease area under research
To achieve representative and projectable data on
incidences, prevalences or mortality of a certain
disease area
Option 3Quantitative Research
22
Analysis and Interpretation of Data
Different statistical methods and procedures can
be used to project data
Considerations
23
Road Map for Forecasting Oncology Sales

• Quantify the current Market Potential
• Estimate the future Market Potential
• Forecast market / patient shares

24
Road Map StepQuantification of the current
Market Potential
1
Example for

• Literature
• Ad-hoc research

• Published or non-published epidemiology data,
  depending on availability
• Ad-hoc research

EGFR expressingmetastaticcolorectal
cancer,refractory or intolerant
toirinotecan-based chemotherapy
Published or non-published epidemiology data,
depending on availability
Ad-hoc research
25
Road Map Steps
2
3
26
Topics covered

• Specific challenges in forecasting oncology sales
  based on the complex nature of this particular
  market
• Using secondary data criteria to assess the
  reliability of secondary data sources
• Introduction to workshop case studies
• Five common pitfalls in oncology forecasting
• Five primary steps of oncology forecasting
• Case Study oxaliplatin in colorectal cancer
• Questions Answers

27
Five Common Pitfallsin Oncology Forecasting
28
Five Common Pitfalls in Oncology Forecasting

• Epidemiology (incidence/prevalence) does not
  equal anticancer drug treatable patient
  population
• Line of therapy migration the oncology product
  life cycle
• Pent-up demand
• Off-label usage
• Oncology reimbursement

29
Pitfall ?Incidence vs. Prevalence vs. Treated
Patient Population

• New cancer incidence is a number used by
  epidemiologists to estimate the total number of
  patients newly diagnosed with cancer every year
  this does not directly translate into
  opportunities for anticancer drugs
• Untreated
• Watchful Waiting
• Other treatment modalities like radiation and/or
  surgery
• Cancer prevalence is a number used by
  epidemiologists to estimate the total number of
  people living with cancer or cured over a
  defined period of time (typically 5 years) this
  does not directly translate into opportunities
  for anticancer drugs
• Untreated
• Watchful Waiting
• Other treatment modalities like radiation and/or
  surgery
• Cure
• Only anticancer drug treated patient population
  represents patients who are receiving anticancer
  drug therapy in a given time period (typically
  monthly and 1 year)

30
Pitfall ?Incidence/Prevalence ? Treatable
Patient PopulationExample Colorectal Cancer in
US
Depending on the tumor, forecasting with
incidence or prevalence figures will overestimate
the market size and the forecast will be
inaccurate.
Treatable Patient Population includes patients
receiving anticancer drug therapy in ALL lines.
31
Pitfall ?Line of Therapy Migration The
Oncology Product Life Cycle
1st Line Contribution
2nd Line Contribution
3rd Line Contribution
4th Line Contribution

• These line of therapy penetration points are
  tantamount to an accurate forecast. Estimate
  their timing by understanding clinical trial
  activity, clinical results and publication
  strategy.
• Rule of Thumb 1 Assume penetration point 1 to 2
  months after positive results publication.
• Rule of Thumb 2 Penetration rate will decrease
  in latter line of therapy as product moves to
  earlier lines.

32
Pitfall ? Line of Therapy Migration The
Oncology Product Life CycleExample Oxaliplatin
Performance by Line of Therapy
Oxaliplatin patient migration started in January
2004 from metastatic 2nd line to metastatic 1st
line.
33
Pitfall ?Pent-Up Demand

• In cancer, there is typically a segment of
  patients who hold off on their next course of
  therapy in anticipation of the newly launched
  product, this is called pent-up demand.
• Actual sales of an oncology product early in the
  launch cycle should not be used as a proxy of
  future product performance.
• Rule of Thumb 3 Allow 2 to 3 months for pent-up
  demand to disappear
• Pent-up demand will overstate actual product use
  of the product.

34
Pitfall ?Off-Label Use

• As early as launch, anticancer drug products
  typically will have patients receiving therapy
  outside their published indications
• Earlier lines of therapy
• Combination Use
• Other tumors
• Although it is difficult to estimate off-label
  use, some quantitative and qualitative assessment
  must be made to forecast its impact
• Ignoring off-label usage will result in
  underestimation of the market
• Off-label use is typically driven by reported and
  published data
• Rule of Thumb 4 Allow up to 3 months after
  publication for significant market reaction

35
Pitfall ? Off-Label UseExample Bevacizumab
Recently Launched in US
36
Five Steps of Oncology Forecasting
37
Five Steps of Oncology Forecasting

• Understand the Oncology Forecasting Continuum
• Develop baseline quantitative projection
• Develop appropriate units of measure
• Employ appropriate forecast techniques
• Monitor Validate

38
Step ?Oncology Forecasting Continuum

• Develop quantitative baseline projection using
  time series methods
• Using historical events, identify analogies to
  help quantify the impact of future events
• Quantify impact of specific events based on
  timing and likelihood of occurrence
• Obtain consensus from internal stakeholders
• Test forecast with management expectations
• Track actual performance against forecast,
  explain variance, fine-tune monthly

39
Step ?Develop the Baseline Quantitative
Projection

• Use basic time-series methods to identify the
  trend in the data
• Use historical sales, units, and/or
  patient/market share
• Exponential smoothing methods to forecast the
  underlying movement in the data that will help to
  identify recent trends
• Data at the most micro level is recommended
• Identify the timing and probability of events
  occurring, and their impact
• Micro level projections should then be
  aggregated to a macro forecast

40
Step ?Develop Appropriate Units of Measure

• Time Period
• Short Term (launch to 1 year) use monthly data
• Medium Term (1 to 3 years) use monthly data
• Long Term (3 to 10 years) use annual data
• Drug Demographics
• Sales, Units, Patients
• Number of completed cycles, dose per
  administration
• Market Segmentation
• Line of Therapy, Stage of Disease
• Practice Setting Private Practice, Hospital
• Country
• Market Life Cycle
• Earlier lines of therapy, adjuvant care
• New, Growth, Mature Products
• Generic Erosion

41
Step ?Employ Appropriate Forecast Technique

• Time series quantitative base-line forecast using
  historical data (usually 12 to 24 months forecast
  using market share, units, sales)
• Assumption Patterns observed in past data can be
  used to predict performance in future periods.
  Warning Time series methods alone will not
  accurately predict turning points in the market
• Identification of events
• Marketing events promotion, pricing
• Competitive events promotion, pricing, new
  products
• Market events regulatory, indications,
  supporting data
• Modification of time series forecast by event set

42
Step ?TIPS

• Always include factors that may affect the
  forecast model in terms of sales
• Completed number of cycles, dosing per cycle
• You must understand the input variables to the
  forecast models, the sources of information, the
  market assumptions, and standard model outputs
  (base case, upside, downside)
• The rate of growth through time will be based on
  either outputs of other models or analog models
  based on secondary data. Peak market share and
  time-to-peak will also be obtained through
  analogs
• Quantitative and qualitative research should
  identify who are the patients and how the various
  types of patients will affect product uptake

43
Step ?Monitor and Validate
If youre not keeping score, youre just
practicing !
44
Rules of Thumb in Oncology Forecasting

• Assume penetration point 1 to 2 months after
  positive results are published
• Penetration rate will decrease in latter line of
  therapy as product moves to earlier lines.
• Allow 2 to 3 months for pent-up demand to
  disappear
• Allow up to 3 months after publication of new
  data for significant market reaction
• Exponential functions are generally more accurate
  indicators for future performance in the oncology
  market

45
US Case Study Oxaliplatin in Colorectal
CancerPrepared by IntrinsiQ Inc.
46
Case Oxaliplatin Forecast Colorectal Cancer

• Oxaliplatin Facts
• Launched September 2002 in relapsed metastatic
  colorectal cancer
• Submitted NDA for 1st line mCrC in July 2003
• Indicated for 1st line mCrC in December 2003
• Data available for adjuvant use in 2H2003
• Oxaliplatin Market Share Assumptions
• Launch share 3
• 1st Peak 30 within 1 ½ years of launch
• Penetrate 1st line, adjuvant therapy with
  availability of data
• 2nd Peak 45 within 1 year of additional
  indication
• Using Decelerated Exponential Curves

47
Case Oxaliplatin Forecast Colorectal Cancer
Top-Line Projection in All Lines and Stages
48
Case Oxaliplatin Forecast Colorectal
CancerActual Market Share Performance vs.
Forecast
What could be the reasons causing the dip in
actual performance vs. forecast? Could this have
been predicted in the forecast model?
49
Case Oxaliplatin Forecast Colorectal
CancerAll Lines Total Share and New Share
Performance
Aggressive monitoring of new patient starts is a
valuable leading indicator of future market share
performance. In colorectal cancer, new patient
starts represent 25 of total patients. New
patient share has significant (.33) influence on
total share.
50
Case Oxaliplatin Forecast Colorectal
CancerMetastatic 2nd Line Total Share and New
Share Performance
2
1
The decline in new patient starts in 2nd line is
(1) due to competitive counter-measures of
irinotecan and (2) a indication that the product
is moving to earlier lines of therapy.
51
Summary and Key Recommendations.
52
Summary Oncology Forecasting Considerations

• Simplicity
• Methods must be simple to use. Methods that are
  not understood will not be used by management
• Documentation
• On-going identification of key events,
  documenting timing impact
• Keeping Score
• Monthly performance should be tracked
  maintained
• New products, new indications, patent expiry,
  generics, regulatory
• Caution Pent-up Demand
• Communications
• Between product managers, marketing research,
  finance, manufacturing, sales management,
  medical/clinical and forecasting

53
Key Recommendations
For short and medium term forecasting (i.e.
launch to 3 years) use monthly data
Monitor new patient starts as leading indicator
for future market share performance
Check your sources whether they meet above
requirements !
54
Questions Answers
55
Working Groups
56
Workshop Objectives

• Group 1 3 Defending market position of Product
  B
• Group 2 4 Launching Product X
• Identify key assumptions in your oncology
  forecast.
• Illustrate the next twelve months of this market
  by drug / combination
• Prepare to explain

57
Thank You ! For further informationplease
contact
PROPHARES GmbH Kleinhueningerstrasse
179 CH-4057 Basel - Switzerland Tel [phone]
80 00 Fax [phone] Email
info_at_prophares.com www.prophares.com
Petra Maertens Tel [phone] Fax
41 61 638 80 10 Email maertens_at_prophares.com