Portfolio · Project 05 · Cancer medicines · October 2026
Launch sequence strategy
Prices travel. Once a cancer medicine has a price in one country, other countries use it to cap, floor or negotiate their own. So the order in which a company launches decides not only when patients get the drug, but what it earns everywhere. Part 1 shows how the access rules connect, from the moment Phase 3 reads out. Part 2 is an engine: enter the drug’s epidemiology, comparator, PFS and OS, safety and price, and it recommends where to launch first and next, and why.
From Phase 3 readout to a launch sequence
Three decisions sit between a positive Phase 3 and the first patient in each country. Each one changes what the next one can achieve.
Primary endpoint met, survival immature
Often PFS or response is mature; overall survival (OS) is still early. The licence is reachable now.
OS more mature, months later
A clearer survival estimate, which most HTA bodies weigh most heavily, at the cost of a later start.
More than one indication planned
Typically a smaller, high-need population and a larger, broader one, reading out at different times.
The benefit rating unlocks the price rules
- Germany: a minor or non-quantifiable added benefit means a price no higher than the comparator’s.
- France: ASMR I–III brings the European price guarantee; ASMR IV–V does not.
- Brazil: proven therapeutic gain (category 1) can price up to the lowest of 14 countries; no added benefit (category 2) is capped at the comparator’s treatment cost.
Its price is the starting point
- Japan prices most new drugs from a comparator’s price plus premiums (innovativeness 70–120%, usefulness 5–60%), or from a cost build-up.
- Against generics, the comparator-based caps in Germany and Brazil sit far below a brand price, so a small gain is costly.
- Against a branded drug, parity is easier to defend, but the incumbent defends its share.
Volume changes how hard payers push
- Germany treats an orphan drug’s added benefit as proven until sales pass €30M a year.
- France needs a CEESP efficiency opinion once ASMR I–III is claimed and sales reach €20M a year.
- Japan’s cost-effectiveness review for high budget impact has trimmed oncology prices by about 0.2–15%; orphans get priority review automatically.
Earlier licence, weaker HTA position
- Regulators have early routes: FDA priority review and accelerated approval, EMA conditional MA (Project 01, US and EU cards).
- Germany: OS, symptoms, quality of life and side effects carry the rating; PFS alone rarely earns added benefit.
- England: immature data with plausible cost-effectiveness lead to the Cancer Drugs Fund, then re-appraisal.
- Canada pairs time-limited recommendations with temporary access for NOC/c products.
Later licence, stronger benefit ratings
- More mature OS supports a higher German benefit category and a better French ASMR, which drives price and the liste en sus.
- Fewer managed-access conditions, but every month of waiting is a month without sales anywhere.
- The JCA PICOs still need an answer for every comparator; more mature data make indirect comparisons more robust.
Early where speed pays, later where survival sets price
File early with the regulator whose market prices freely and pays from approval, and bring mature survival data to the HTA bodies that rate on it. Since January 2025 the EU joint clinical assessment narrows this for EU markets: its report is completed within 30 days of authorisation, and national bodies must give it due consideration, so a later data cut usually reaches them through reassessment rather than a later launch. The engine in Part 2 makes this call market by market.
The first indication sets the price
In interviews across nine OECD countries, payers named the highest price as the defining feature of first indications, which were mostly smaller populations. No example was found of a price rising after a new indication (Mills and Kanavos, 2023).
Each system handles it differently
- France, Spain, Belgium, Canada: the price is renegotiated as a weighted average.
- England, Switzerland: different discounts per indication.
- Italy: a mandatory discount scaled to the extra population.
- Germany: each new indication is a new AMNOG cycle. United States: one free price.
Markets others look at, that pay early
The US prices freely and covers from approval. Germany pays the company price from launch, with the negotiated price from month 7. Their prices then sit in other countries’ baskets.
Launch once their reference prices exist
Japan checks the average of US, UK, German and French prices. France’s price guarantee looks at Germany, Spain, Italy and the UK. Canada caps at the highest of 11. Switzerland compares nine countries.
Later, or behind a confidential deal
Markets whose published prices are low and widely referenced pull others down, so companies often launch them later or protect the list price with confidential discounts. Brazil caps at the lowest of 14 countries.
The access mesh: who looks at whose price
Each arrow points from a market to a market whose price it uses. Select a market to see what it looks at (oxblood) and who looks at it (navy). The rules come from the sourced Project 01 cards and the Reference pricing page.
The clocks: how soon each market pays
Speed matters as much as price. A market that pays from launch earns from day one; a market with a long funding clock adds little in the first five years however high its price. All figures come from the Project 01 cards.
Across Europe, EFPIA’s 2025 W.A.I.T. survey found an average of 597 days from EU approval to availability for medicines approved in 2021–24, and 655 days for oncology medicines.
Medicare covers medically accepted indications from approval; the company sets the launch price.
Sickness funds pay the company’s price from launch; the negotiated price applies from month 7.
NHI listing usually within 60 days of approval (90 at most); listings four times a year.
Aligned pathway: NICE guidance publishes with the licence; positive guidance funded within 90 days.
Specialty List benchmark 60 days after MA, met for 43% of files in 2024; case-by-case payment before that.
Non-generic negotiations averaged about 7 months; until then the drug can be sold unreimbursed (class Cnn).
Real timelines run well past the legal target; early access (AP) bridges the gap for many cancer drugs.
About 70% of innovative medicines reached patients before the funding decision through special-situation access.
Average time from approval to public funding; pre-NOC filing and Ontario’s FAST pilot shorten it.
The Minister decides on the CRM proposal within 180 days; confidential MEA conventions are routine for oncology.
The parallel pilot targets 150 days from approval to listing; early rounds ran longer. Risk-sharing agreements are usual for oncology.
Drugs costing €20M a year, or €10M with €50,000 per patient, wait in the lock until Zorginstituut advises and a price is agreed.
PBAC meets in March, July and November; a positive recommendation is not a listing, and price talks can still fail.
CMED price within 90 days of registration; Conitec 180 + 90 days for the public system; ANS 120 + 60 days for cancer in private plans.
What companies have actually done
Launches follow expected price and volume
In 25 markets in the 1990s, the chance of launch rose with expected price and volume. Low prices spill over through referencing and parallel trade, so firms accepted delay or no launch: 73 new drugs launched in the US, 66 in Germany and 64 in the UK, against 13 in Japan and 26 in Portugal.
Danzon, Wang and Wang, Health Economics 2005 (NBER w9874)Submissions held back to keep a price out of baskets
Companies systematically delayed Belgian dossiers so its price would not feed other countries’ calculations. Across seven EU countries, 11 products were not launched to avoid expected low prices.
LSE, impact of external reference pricing (2017)Germany: leaving rather than accept a low price
Among 139 medicines assessed from 2011 to June 2016, there were 10 opt-outs and 12 supply terminations; 21 of the 22 had been rated “no added benefit”.
Staab et al., Health Economics Review 2018Multi-indication drugs launch the high-value use first
Payers see the first indication as the price-setting one; current systems still reward launching niche indications first. Some later indications were withheld in France, Italy and Belgium when price talks failed.
Mills and Kanavos, PharmacoEconomics Open 2023US most-favoured-nation pricing reshapes Europe
In the year after the May 2025 US order, launches fell 37.8% in 11 EU states facing both MFN and the JCA, against 10.5% in the US. Insmed held back a European launch approved in November 2025 while awaiting MFN clarity.
EU Perspectives, Aug 2026 (King’s College study); BioSpace, Apr 2026Germany now counts only after AMNOG
Japan’s reference basket is the US, UK, Germany and France. Since the FY2026 reform, only Germany’s post-AMNOG negotiated price counts, so a German free launch price no longer lifts Japan’s comparison. Japanese industry asked in June 2026 to ease the foreign price adjustment because of MFN.
Windrose CG on the FY2026 reform; PharmaJapan, Jun 2026Rules of thumb
- Set high, visible prices first in markets that pay early and that others reference: typically the US, Germany and the UK.
- Launch referencing markets after their baskets are filled with your best prices. Japan wants US, UK, French and (negotiated) German prices; France’s guarantee wants German and UK prices above Spain and Italy.
- Hold back or shield low visible prices. Confidential discounts keep the list price high; a later launch keeps a low price out of baskets while it matters most.
- Lead with the indication that earns the highest price. In most systems a later, broader indication lowers the price; it rarely raises it. Over a short horizon a much larger indication can still win on volume.
- Weigh IA1 against IA2 market by market. Survival-driven HTA bodies reward waiting; markets that pay from approval reward speed.
- Model the US as part of the mesh. With most-favoured-nation pricing, a low price abroad can now lower the US price too.
Sources
- Danzon, Wang, Wang: price regulation and launch delay (NBER w9874; Health Economics 2005)
- LSE: the impact of external reference pricing within and across countries (2017)
- Staab et al.: “market withdrawals” in Germany after AMNOG (2018)
- Mills and Kanavos: payer perspectives on multi-indication oncology products (2023)
- EU Perspectives: US pricing and EU assessment push new drugs out of Europe (Aug 2026)
- BioSpace: MFN pricing expected to delay more European launches (Apr 2026)
- Windrose CG: Japan’s FY2026 drug pricing reform
- PharmaJapan: FPMAJ asks to ease foreign price adjustment (Jun 2026)
- EFPIA Patients W.A.I.T. Indicator 2025
- Market rules and clocks: the sourced Project 01 cards (US, EU, Germany, France, UK, Italy, Spain, Japan, Canada, Switzerland, Brazil) and the Reference pricing page.
Launch sequence engine
Enter what you know about the drug: how many patients it can treat, what it replaces, its PFS and OS results at each analysis, its safety and the price you have in mind. The engine tries every market at approval, at later months, on IA2 data or held back, chooses which indication to lead with, and keeps the plan worth the most over the horizon you pick. It explains each step, shows what would change each decision, and reruns the plan under 200 sets of varied assumptions to show how often each choice holds up.
Where to launch, and in what order
What would change the plan
Each input is moved on its own, with everything else fixed, to find where a market’s decision flips.
Market by market: why
Launch timeline
Net sales by market
What your inputs mean in each market
Germany’s column shows the IQWiG-style category. Elsewhere the benefit score is each payer or HTA body’s weight on OS times the OS score and the OS credit for statistical maturity, plus the rest on the PFS score, plus a safety adjustment.
What the mesh did to your prices
Build your sequence
Markets launch in the order listed. Each one launches no earlier than its approval (plus any wait you choose) and no earlier than the market above it, plus the months between launches. Move markets with the arrows, choose when each launches, or hold it out of the plan. Everything recalculates at once.
Your sequence against the recommendation
Your launch timeline
What the mesh did to your prices
Market assumptions (editable)
Change any market’s numbers, or leave a market out. Patient ratios come from 2022 cancer deaths and new cases (WCRF country tables); Brazil’s reach is private-plan members, about 53 million people (ANS, late 2025), roughly a quarter of the population. Price levels, approval months and funding delays are illustrative.
Checks against what companies have done
Each check runs the engine on a profile and tests whether it reproduces a pattern documented in the Part 1 precedents. These are qualitative checks of direction, not a back-test against real launches.
| Precedent and pattern | Profile | What the engine did | Result |
|---|---|---|---|
| Running… | |||
Test yourself
How the engine works, and what it leaves out
Patients and uptake. Other markets’ patients scale from the US figure by each country’s 2022 cancer deaths (or new cases) relative to the US, times the share the launch can reach. Peak share of eligible patients is 5% plus 30% times the average of the PFS and OS scores, times 1.25 for superior safety or 0.7 for worse, times 0.85 against a branded incumbent. Uptake peaks two years after launch; patients stay on treatment nine months. The second indication peaks at 60% of the first’s share.
Germany. The OS hazard ratio’s 95% confidence interval comes from the number of deaths (standard error of log HR ≈ 2/√deaths for 1:1 randomisation). IQWiG’s thresholds for all-cause mortality set the category: upper limit below 0.85 major, below 0.95 considerable, below 1.00 minor. Superior safety moves the rating up one step and worse safety down one. PFS alone earns no German benefit. A minor or unproven benefit caps the price at the comparator’s cost, except under orphan-type rules, where added benefit is deemed proven.
Other markets. The PFS score is (1 − HR) ÷ 0.45 and the OS score (1 − HR) ÷ 0.30, both capped at 1.25. OS counts in full once it is statistically convincing (log HR at least 2.5 standard errors from zero) and not at all below one standard error. Each market weights OS differently (US 0.35, Japan and Brazil 0.5, Korea 0.55, UK, Canada, Switzerland and Australia 0.6, Italy, Spain and Belgium 0.65, Netherlands 0.75, France 0.8). Safety adds 0.25 or subtracts 0.3.
Which data each market sees. A market sees IA2 data if it launches at least two months after IA2 reads out. When EU evidence is fixed at the joint clinical assessment, EU markets (Germany, France, Italy, Spain, the Netherlands, Belgium) are assessed on IA1 data whenever they launch, because the JCA report is completed within 30 days of authorisation and national bodies must give it due consideration. With reassessment on, Germany reassesses on IA2 data at the later of 12 months after launch and IA2 plus two months, with the new price six months later; England funds immature OS through the Cancer Drugs Fund at a price halfway between IA1 and IA2 value until re-appraisal six months after IA2 data are ready.
Prices. Each market has a brand price level per patient-year (US $240k, Switzerland $150k, Germany $140k, Japan and France $125k, Canada and the Netherlands $120k, UK and Belgium $115k, Australia $110k, Italy $80k, Spain and Korea $70k, Brazil $55k). The comparator costs your standard-of-care figure scaled to each market. A value-based market pays the comparator’s cost plus the benefit score times a premium of one brand level over a cheap comparator, or half a brand level over a branded one, and never more than your ask. Orphan-type populations (under 1,500 US patients) earn 1.5 times the premium. In the US, a list price above value cuts uptake by (value ÷ price)1.5.
Reference rules (from the Project 01 cards unless noted). Japan: within 25% of the average of US, UK, French and post-AMNOG German prices; an upward move needs at least two prices and is capped at twice Japan’s own price (MHLW). France: ASMR I–III earns early access and the guarantee of not falling below the lowest of Germany, Spain, Italy and the UK; otherwise no liste en sus. Canada: list capped at the highest PMPRB11 price set (nine of the eleven are in the engine). Switzerland: half its value-based price, half the average of its basket (Belgium, France, Germany, the Netherlands and the UK of its nine; Interpharma). Brazil: proven gain prices up to the lowest of its references, otherwise the comparator’s cost. Netherlands: maximum price is the average of Belgium, France and the UK (Norway, the fourth, is not in the engine; Taylor Wessing), and the lock holds drugs costing €20M a year, or €10M with €50,000 per patient, until a price is agreed. Korea: small-population cancer drugs listed in at least three A7 countries skip economic evaluation and price at the lowest A7 price (Frontiers in Pharmacology, 2024). The UK, Italy, Spain, Canada, Australia, Korea, the Netherlands and Belgium keep net prices confidential, so others see the list.
The search. For each market the engine tries launching at approval; three, six, nine, 12, 18 or 24 months later; when IA2 data are ready; one month after each other market’s price is set; or not at all. It also tries leaving out the second indication in each market whose price would fall, and leading with either indication (leading with B moves every approval six months later). It changes one choice at a time, keeps any change worth more than 0.2% of value, and repeats until nothing improves.
Robustness. The engine redraws its assumptions 200 times: patients overall (±20%) and by market (±15%), peak uptake (±20%), brand price levels (±12%), the value premium (±15%), funding delays (±30%), each market’s OS weight (±0.1), the US uptake penalty (exponent 1.2 to 1.8), and the IA2 hazard ratio from its predictive distribution given IA1 (log HR variance 4/d₁ − 4/d₂). Each decision is compared with its best alternative of another kind. This shows how sensitive the plan is to the model’s assumptions; it is not a probability that the plan is right.
What it leaves out. France can also re-evaluate ASMR on new data. Competitors, loss of exclusivity, supply, launch-team capacity and patient need are not modelled. Price levels, approval months and funding delays are illustrative; real decisions need each market’s epidemiology, comparator prices and payer research.
Drug X and every number in Part 2 are hypothetical. The access rules follow the sourced Project 01 cards in simplified form. Personal project: not sponsored by or affiliated with any employer (current or former) or any university I attended; views are my own. Developed with AI assistance, so errors or outdated details may exist; check the cited sources before relying on anything here. Not medical, regulatory, reimbursement or investment advice. Full disclaimer.
Portfolio · Project 05 · Guide
How to use the launch engine
The launch engine turns what you know about a cancer drug into a recommended launch sequence across 14 markets. It explains every step, shows which inputs would change the plan, and lets you build and test your own sequence. This guide takes about ten minutes to read.
What the engine does
Once a cancer drug has a price in one country, other countries use that price to cap, floor or negotiate their own. So the order and timing of launches decide not only when patients get the drug, but what it earns everywhere. The engine models this “access mesh” for 14 markets: the US, Germany, Japan, the UK, France, Italy, Spain, Canada, Switzerland, Brazil, Australia, South Korea, the Netherlands and Belgium. Part 1 of the project explains the rules; the engine applies them to a drug you describe.
For each market the engine tries launching at approval, at later months, once the second interim analysis (IA2) is available, or not at all. It also decides which of two indications to lead with. It keeps the combination worth the most over your chosen horizon, net of launch costs. It then reruns that plan 200 times with its assumptions varied, to show how often each decision holds up.
Quick start
- Pick an archetype or enter your drug. In the left panel, choose one of the three archetypes to load a ready-made profile, or type your own numbers into sections 1 to 5. The engine recalculates as you type.
- Read the recommendation. At the top right you see the plan’s value and robustness, then where to launch and in what order, what would change the plan, and a card for every market explaining its timing.
- Try your own sequence. Switch to “Build your own sequence” at the top of the results, reorder markets, set when each launches, and see how your plan compares with the recommendation.
Step 1: Describe the drug
Epidemiology
Enter the number of new patients a year in the US who would be eligible for the first indication. The engine scales every other market from this figure using each country’s 2022 cancer deaths relative to the US, which tracks advanced disease more closely than incidence. You can switch to new cases instead. The note under the field shows the resulting patient numbers per market. Fewer than 1,500 US patients switches on orphan-type rules (German added benefit deemed proven, Japanese priority review, Korea’s evaluation waiver and a higher accepted price per patient). Five thousand or more switches on high-budget-impact rules (Japan’s cost-effectiveness review and harder net-price bargaining).
Standard of care
Choose whether the drug replaces cheap generics or a branded drug, and adjust the annual US cost of that comparator. The comparator matters twice. Its cost is the base of every value-based price, and in Germany and Brazil a drug without a convincing benefit is capped at the comparator’s cost. Against a branded incumbent, uptake is also lower because the incumbent defends its share.
Efficacy
Enter the PFS hazard ratio, then the OS hazard ratio and the number of deaths at IA1 and at IA2, and the months between the two analyses. The number of deaths is what makes OS convincing: the engine estimates the 95% confidence interval from it and applies IQWiG’s thresholds for all-cause mortality (upper limit below 0.85 for a major benefit, below 0.95 for considerable, below 1.00 for minor). The note under the fields shows the interval and the German category at each analysis. For example, a hazard ratio of 0.75 with 180 deaths has an upper limit just above 1.00, so no mortality benefit is proven; with 360 deaths the upper limit falls to about 0.92, which is considerable. Other markets give OS full weight once it is statistically convincing and weigh PFS more heavily than Germany does.
Safety
Superior safety raises Germany’s rating by one step, lifts the benefit score elsewhere and increases uptake. Worse safety does the opposite.
Price
Enter the US list price per patient-year. Every other market asks the same position relative to its own brand price level, shown under the field. Value-based markets pay the lower of your ask and what the drug is worth to them, so a price above value mainly slows negotiations there. In the US, a price above value triggers utilisation management that cuts uptake faster than the price rises.
Step 2: Set the access rules and horizon
EU evidence after the joint clinical assessment. The default, “Fixed at the JCA”, assesses EU markets on IA1 data whenever they launch, because the JCA report is completed within 30 days of authorisation and national bodies must give it due consideration. Choose “Can be updated nationally” to let a later EU launch use IA2 data, as companies could before the JCA.
Reassessment after launch. With this on, Germany reassesses on IA2 data and renegotiates the price, and England funds immature OS through the Cancer Drugs Fund until re-appraisal. This makes launching early on immature data less costly.
US most-favoured-nation pricing. With this on, half of US volume is priced at the lowest purchasing-power-adjusted price among the other launched markets, so a low price anywhere can pull the US price down.
A second, broader indication. With this on, a second indication with three times the patients follows 18 months after the first in each market, and the engine chooses which to lead with. Leading with the broader indication (B) moves every approval six months later.
Horizon and costs. Choose a five- or ten-year horizon, an annual discount rate and a fixed launch cost per market. The launch cost is what makes small, low-price markets drop out; the longer horizon shows effects, such as a reassessed price, that a short window hides.
Step 3: Read the recommendation
The headline. The dark panel shows the plan’s value (net sales less launch costs) and how it compares with launching everywhere at approval. The robustness figure is how often the plan beats launching everywhere at approval across 200 sets of varied assumptions, or, if the plan is to launch everywhere at approval, how well its weakest material decision holds up. “Searching…” appears while the engine recalculates.
Where to launch, and in what order. One sentence lists the launches by month and names any market that waits for IA2 data or is held. The blue panel says which indication to lead with and how much the other choice would earn.
What would change the plan. For each decision that matters, the engine moves one input at a time (deaths at IA1, the OS hazard ratio, the months to IA2 and the US price) and reports where the decision flips. These tipping points are often the most useful output: they tell you which evidence or pricing assumption a launch decision really depends on.
Market by market: why. Each card gives the market’s timing, its net sales, its rating and price after the access rules, why the engine chose that timing over the alternatives, and how its price helps or hurts other markets. The pill on the right shows robustness: the share of the 200 draws in which the chosen timing beats its best alternative of another kind (launching at approval against waiting, or launching against holding). “Close call” means the two choices are worth almost the same, so either is reasonable.
Charts and tables. The timeline shows when each market launches, when funding is full (navy), partial access before that (light blue), the second indication (oxblood tick) and any reassessment (gold marker). The dashed line marks when IA2 data are ready. The market table lists patients, approval and launch months, the data each market sees, its rating and its prices. The price log shows, month by month, what each rule did to each price.
Step 4: Build your own sequence
Switch to “Build your own sequence” at the top of the results, or use the button under the recommendation. The builder starts from the recommended plan, so the two values match until you change something.
Order. Markets launch in the order listed. Use the arrows to move a market earlier or later. A market can never launch before the one above it, so moving a market up can pull it earlier, and moving it down can push it later; the note under its name says when that happens.
Timing. For each market, choose to launch at approval, three to 24 months later, or on IA2 data. “Months between launches” spaces the launches out, as a launch team with limited capacity would.
Hold. Hold keeps a market out of the plan for the whole horizon. This is how you test an opt-out, such as keeping a low price out of other countries’ baskets.
Starting points. The buttons at the top load the recommendation, everyone at approval, price setters first (US, UK, Germany, Japan, France, then the rest, two months apart) or the US then low-price markets first. Each button shows that sequence’s value, so you can compare them before you start.
Reading the comparison. The dark panel shows your value and the difference from the recommendation. The table shows, market by market, your launch month against the recommended one, both prices, and the change in value. Your own timeline and price log show what the rules did to your sequence. Occasionally a hand-built plan beats the recommendation by a little: the search changes one choice at a time and ignores gains under 0.2%.
Step 5: Change the market assumptions
Open “Market assumptions (editable)” below the results to change any market’s patient ratio, the share of patients the launch can reach, its brand price level, its approval month or its funding delay, or to leave a market out altogether. Patient ratios come from 2022 cancer deaths and new cases; Brazil’s reach is limited to private-plan members, about a quarter of the population. Price levels, approval months and funding delays are illustrative, so this is where to put your own numbers. “Reset all” in the archetype box restores every default.
Four things to try
1. Watch reassessment do the work
With the default inputs, the engine launches everywhere at approval. Open Germany’s card: on IA1 data no mortality benefit is proven, so after six months at the free price Germany pays only the comparator’s cost, until reassessment on IA2 data lifts the price. Now switch “EU evidence” to “Can be updated nationally”: France, Italy, Spain and the Netherlands start waiting for IA2 data instead. Switch reassessment off, and Germany starts waiting too.
2. Find what Germany’s decision depends on
Load the PFS-driven archetype. Germany now waits a few months after approval, and “What would change the plan” shows the range of OS hazard ratios over which waiting pays. Change the deaths at IA1 or the OS hazard ratio and watch the decision flip.
The flip points show which evidence the German decision depends on.
3. Turn on most-favoured-nation pricing
Go back to the default inputs and switch on US most-favoured-nation pricing. Germany now waits, Brazil becomes a marginal launch, and the tipping points show when holding Brazil would pay. This is the mechanism behind the drop in EU launches reported in 2026.
4. Beat, or fail to beat, the engine
Switch to “Build your own sequence”, start from “US, then low prices first”, and see how much it costs against the recommendation. Then try holding Germany, or moving Japan after Germany, and read the price log to see which rule moved which price.
What the engine is, and is not
The engine is a teaching and thinking tool. Its rules follow the sourced Project 01 cards in simplified form, and its patient ratios and several reference rules are sourced, but its price levels, uptake, approval months and funding delays are illustrative assumptions. Robustness shows how sensitive a decision is to those assumptions; it is not a probability that the plan is right in the real world. The engine does not model competitors, loss of exclusivity, supply, launch-team capacity beyond the spacing you choose, patient need or France’s re-evaluation of ASMR. A real launch decision needs each market’s epidemiology, comparator prices and payer research.
Terms used
- IA1, IA2
- First and second interim analyses of the Phase 3 trial. IA2 has more deaths, so a more precise OS estimate.
- Hazard ratio (HR)
- The relative rate of progression (PFS) or death (OS) on the new drug against the comparator; below 1 favours the drug.
- 95% confidence interval
- The range of plausible hazard ratios; its upper limit sets the IQWiG category.
- JCA
- EU joint clinical assessment under the HTA Regulation, mandatory for new cancer medicines filed since January 2025.
- AMNOG, G-BA
- Germany’s benefit assessment and price negotiation; the Federal Joint Committee rates added benefit.
- ASMR
- France’s rating of added clinical benefit, I (major) to V (none); I–III unlocks the European price guarantee.
- Cancer Drugs Fund
- England’s managed-access route for cancer drugs with immature data, followed by re-appraisal.
- PMPRB11
- The 11 countries whose highest price caps Canada’s patented-drug price.
- CMED
- Brazil’s drug price regulator; proven gain prices up to the lowest of 14 reference countries.
- A7
- Korea’s reference countries: the US, UK, Germany, France, Italy, Switzerland and Japan.
- The lock
- The Netherlands’ holding procedure for high-budget drugs until a price is agreed.
- MFN
- US most-favoured-nation pricing, which links US prices to the lowest prices paid in other wealthy countries.
- Robustness
- How often a decision beats its best alternative across 200 sets of varied assumptions.
Drug X and every number in the engine are hypothetical. Personal project: not sponsored by or affiliated with any employer (current or former) or any university I attended; views are my own. Developed with AI assistance, so errors or outdated details may exist; check the cited sources before relying on anything here. Not medical, regulatory, reimbursement or investment advice. Full disclaimer.