Ormeval provides decision-support intelligence derived from public sources. It is not investment, medical, legal or regulatory advice, and states nothing about the value of any security. Forward figures are deterministic model estimates, not guidance.
Keytruda (pembrolizumab) · anti-PD-1 · Merck & Co.
This page covers the 1L advanced biliary tract cancer indication only (the KEYNOTE-966 regimen), not the flagship 1L-NSCLC or other Keytruda settings. See all Keytruda indications →
Label 1.14: 'KEYTRUDA, in combination with gemcitabine and cisplatin, is indicated for the treatment of patients with locally advanced unresectable or metastatic biliary tract cancer (BTC).' The row's combination partners match the label exactly. Approved 2023-10-31 on KEYNOTE-966 as a traditional approval - absent from all three FDA accelerated-approval registries. No PD-L1 or other biomarker requirement, and no line-of-therapy restriction in the label text.
All markets
2026 net revenue · modelled, WW
$260M
CONCLUDED · 2026 modelled net revenue - indication slice, all markets; Merck & Co. reports no indication revenue. Modelled net revenue · sourced funnel
No model change since the 20 Sep 2026 snapshot.
The 15-second read
This line tops out the moment it arrives: $260M in 2026, the peak, then a slide to $83M by 2034. KEYNOTE-966 is a real positive pivotal, OS 12.7 vs 10.9 months, HR 0.83, NCCN Category 1 and preferred, but almost every structural factor works against monetising it. Durvalumab's TOPAZ-1 opened the category roughly 14 months earlier with a near-identical result (HR 0.80), so pembrolizumab enters as the co-standard taking the minority share of a shared IO pool. The absolute gain of about 1.8 months drew low-end HTA ratings, which caps price. Biliary tract cancer is rare in the West and progresses fast, so realised exposure is short and revenue per patient is low, while the large Asian burden sits largely behind self-pay access in China. And pembrolizumab's US biologic exclusivity lapses around 2028, three years ahead of durvalumab's, so the later entrant also loses its molecule first.
Threats land:China stays self-pay for both IO agents with no NRDL listing, removing the only large-volume lever this indication has, while biosimilars erode the US and EU from 2028–2029, leaving a curve that peaks in 2026 and never finds a second source of growth.
Modelled path:Pembrolizumab holds a minority co-share of the 1L BTC immunotherapy pool behind durvalumab, peaks at $260M in 2026 on a short realised course at HTA-constrained pricing, then erodes from ~2028 as pembrolizumab biosimilars enter.
Levers land:Merck's oncology footprint narrows durvalumab's first-mover lead in real-world 1L use faster than modelled, and perioperative or earlier-line BTC trials extend pembrolizumab into larger, longer-treated pools before the molecule's exclusivity lapses.
Pembrolizumab's ~40–45% co-share of the 1L BTC immunotherapy pool at maturity, an Ormeval scoping estimate of how two clinically indistinguishable regimens split a market when one arrived 14 months earlier. The fit-for-1L-chemo-IO fraction (~55–70%) is the other soft term, since BTC patients are frequently frail or cholestatic and no clean public treated-fraction exists; Merck reports Keytruda as a single product line with no indication breakout.
At a glance
Mechanism anti-PD-1
Clinical
Commercial
Development to approval
These are regulatory approval dates (FDA/EMA/NMPA/PMDA), not commercial launch; launch follows after pricing & reimbursement (see the access-lag column in the HTA table). All four regulators have approved this indication. Cross-checked against Merck & Co. disclosures and regulator databases.
KEYNOTE-966 extended pembrolizumab into a brand-new IO category and reinforced Keytruda's reach across GI tumours, but commercially it is a SMALL line, not a franchise driver. Three constraints define it: a small Western population, a short treatment course, and the fact that durvalumab (TOPAZ-1) got there first (FDA Sep 2022) and still runs ahead in real-world 1L use. Pembrolizumab's counter-leverage is Merck's vast oncology footprint and prescriber entrenchment, which let it take a meaningful co-share (~40-45% of the IO pool) despite the later entry, but the modest, HTA-discounted benefit and the ~2028 Keytruda biologic LOE cap the upside.
Biliary tract cancer (cholangiocarcinoma + gallbladder cancer) is rare in the West but common in Asia, usually diagnosed advanced (~two-thirds unresectable/metastatic) with a poor prognosis. Gem/cis chemotherapy had been the unchanged 1L standard since 2010. KEYNOTE-966 added pembrolizumab → median OS 12.7 vs 10.9 mo (HR 0.83). Durvalumab (TOPAZ-1) was first to market (~14 months ahead); pembrolizumab is a near-identical co-standard that splits the IO pool, a modest GI line within the much larger Keytruda franchise.
Biliary tract cancer spans intrahepatic and extrahepatic cholangiocarcinoma plus gallbladder carcinoma, a heterogeneous, aggressive group that is RARE in the West (~2–6/100k) but markedly more common in Asia (Japan among the highest national rates; China carries the largest absolute burden; Thailand's liver-fluke belt is the global extreme). It is usually diagnosed late, ~two-thirds present unresectable or metastatic, and gemcitabine/cisplatin had been the unchanged 1L standard since the 2010 ABC-02 trial. KEYNOTE-966 added pembrolizumab for all comers (no biomarker). Across the 8 modelled markets ~55,000 patients/year present with advanced BTC and are fit for 1L chemo-immunotherapy (the majority in Asia), the same pool TOPAZ-1 sizes, which pembrolizumab now splits with durvalumab. The chemo-IO course is short and progression is rapid, so revenue per patient is low.
KEYNOTE-966 was the second Phase 3 to show that adding a checkpoint inhibitor to gemcitabine/cisplatin improves survival in advanced biliary tract cancer, a rare, aggressive cancer where gem/cis had been the unchanged standard since the 2010 ABC-02 trial. Pembrolizumab + gem/cis improved median OS (12.7 vs 10.9 mo; HR 0.83) for all comers (no biomarker). The benefit is modest in months but, as in TOPAZ-1, durable in the tail. Pembrolizumab entered the IO category that durvalumab (TOPAZ-1) opened ~14 months earlier, and the two are near-identical co-standards that split the 1L-IO pool.
Pembrolizumab (KEYNOTE-966) and durvalumab (TOPAZ-1) are co-equal Category 1 preferred 1L options with near-identical OS (HR 0.83 vs 0.80, all-comers), durvalumab's edge is its ~14-month earlier approval, so it holds the larger real-world share. Biomarker-targeted agents (pemigatinib/futibatinib for FGFR2 fusions, ivosidenib for IDH1, zanidatamab for HER2) are all 2L+ and do not compete in 1L all-comers.
Randomized, double-blind, placebo-controlled Ph3; n=1069 (pembrolizumab 533 / placebo 536); previously-untreated unresectable locally-advanced or metastatic biliary tract cancer (intrahepatic + extrahepatic cholangiocarcinoma + gallbladder); pembrolizumab 200 mg q3w + gem/cis (cisplatin up to 8 cycles, gemcitabine could continue) then pembrolizumab maintenance up to 35 cycles; all-comers (no PD-L1 selection).
Load-bearing 4/5 sourced. ASSUMED uptake is yours to change.
This indication only, not Keytruda's total revenue
First-line pembrolizumab + gemcitabine/cisplatin (gem/cis up to 8 cycles) then pembrolizumab maintenance, in previously-untreated unresectable/locally-advanced or metastatic biliary tract cancer (the KEYNOTE-966 regimen), all-comers, no PD-L1 selection.
Reported actuals are always drawn and need no chip: a cited reported figure is a fact, not a claim. Everything forward of them is one of the 4 kinds above.
$260M
8 modelled markets (US, EU5, JP, CN) · modelled net revenue · 2026 (forecast)
Peak $260M · launch 2018 → 2034
modeled history (left of today)·click any year to decompose its funnel below.
Measured band: none - no measured band: n<8
Bull / bear elsewhere in the product are a sensitivity (fixed multipliers), not this band.
Pre-2026 figures are MODELLED (Merck reports Keytruda as one product line, never by indication × country) and calibrated so the BTC slice RISES from the Oct 2023 US launch through the 2024 ex-US launches to a modest ~$0.50-0.52bn WW plateau (~2026-27), then erodes from ~2028 as the Keytruda molecule loses US exclusivity. BTC is a MINOR Keytruda line, rare in the West, short-course, and sharing the IO pool with the first-mover durvalumab, well under 2% of the ~$31.7B FY2025 Keytruda franchise. The 'today' marker splits modelled history from forecast.
No reported anchor exists at this grain: companies publish this asset as one product line and never split it by indication × country, so every year above is modelled, including the years already past. The claims below are scored against reported actuals wherever a program has them.
A forecast year closes when an actual is reported for it. A miss is the distance from that actual, signed as the claim fell: under it, or over it.
| Claim | Closed years | Worst miss | Worst miss % |
|---|---|---|---|
| Ormeval modelMODELLED · Ormeval · live | - | no closed year yet | - |
A superseded team version keeps its own row: what it was off by is still a fact about that version. A row with no closed year is not a good score.
What it shows. A modelled base-case net-revenue curve for this indication, year by year, split at 'today' into modelled history and forecast, with the branded curve shown against its in-class rivals.
What to look for. The peak year and how steep the decline is after loss of exclusivity, and where a rival's curve crosses yours, that crossing is the share the model has you losing.
Common misread. Reading the pre-'today' portion as reported actuals (it's modelled), or treating the base case as a company target. It is one deterministic Ormeval-modelled scenario anchored to cited actuals, not guidance and not analyst consensus.
Key model assumptions · The headline rests on the co-share split against durvalumab and on the fit-for-treatment gate, applied to an Asia-weighted incidence base (China ~50k, Japan ~18k, the West rare at ~2–6/100k) with ~67% advanced at diagnosis, and on a short-course price assumption set roughly in line with the TOPAZ-1 BTC nets, since rapid progression means realised exposure is a fraction of the 35-cycle label.
The decline is defended, not certain
The post-2027 fall to $83M is the pembrolizumab molecule losing US exclusivity around 2028 (EU ~2029) with biosimilar erosion following, a patent date rather than a clinical judgement, and one that bites this line earlier than it bites the rival, whose molecule runs to ~2031–32. That asymmetry is the defensible core of the shape. It is not certain in depth: in a short-course chemo-IO regimen delivered at specialist centres, biosimilar substitution can be fast on price but the whole category may also compress together, so relative share against durvalumab could hold even as absolute revenue falls.
What would change our mind
The base case is wrong if China lists either IO agent on the NRDL for 1L BTC, the model explicitly assumes self-pay access in the market with the largest disease burden, so a reimbursement listing would add volume the curve has no allowance for and break the 2026 peak. In the other direction, the near-term falsifier is share: if real-world 1L IO use in BTC does not converge toward the modelled 40–45% for pembrolizumab within two or three years of launch, first-mover advantage in a tie-on-efficacy category is stronger than assumed, and even the peak year is too high.
A scenario is your own assumptions layered on Ormeval’s cited baseline. We never change the baseline; save as many named scenarios as you like and click one to load it back later.
| Driver · 2026 | US | FR | DE | IT | ES | UK | JP | CN | Total | Source |
|---|---|---|---|---|---|---|---|---|---|---|
| Biliary tract cancer incidence | 14k | 2.5k | 3.2k | 2.8k | 1.5k | 2.2k | 18k | 50k | 94k | ACS · GBD 2021 |
New BTC cases/yr (cholangiocarcinoma + gallbladder); Asia-heavy. ~flat trend. BTC incidence is markedly Asia-weighted: China carries the largest absolute burden (~50k/yr, GBD 2021), Japan has one of the highest national rates (~18–22k/yr), while the West is rare (~2–6/100k). Western counts blend cholangiocarcinoma + gallbladder + extrahepatic ducts (registry overlap), so country figures are approximate. Same epidemiology base as the TOPAZ-1 (durvalumab) model in this indication. | ||||||||||
| × Advanced / unresectableest | 67% | 67% | 67% | 67% | 67% | 67% | 64% | 70% | share, not summed | literature (~67%) |
Unresectable advanced or metastatic at diagnosis (vs resectable early disease). BTC is usually diagnosed late, most series report two-thirds to ~90% unresectable or metastatic at presentation. A conservative ~67% is used; slightly higher where early detection is weaker. Same basis as TOPAZ-1. | ||||||||||
| × Fit for 1L chemo-IOest | 68% | 66% | 68% | 65% | 65% | 64% | 70% | 55% | share, not summed | modelled estimate |
Advanced patients fit for (and receiving) 1L gem/cis + immunotherapy, ECOG 0–1, adequate organ function. Many advanced BTC patients are frail, jaundiced or have poor performance status, so a meaningful share receive gem/cis alone or best supportive care rather than chemo-IO. Modelled at ~64–70% in the West/Japan, lower in access-limited China (~55%); flagged estimated. Same basis as TOPAZ-1. | ||||||||||
| = Addressable 1L-treated pool | 6.3k | 1.1k | 1.4k | 1.2k | 647 | 934 | 8.0k | 19k | 39k | computed |
Incidence × the rates above, advanced BTC patients fit for 1L gem/cis + immunotherapy this year (all-comers; no PD-L1 gate). This is the SHARED IO pool that pembrolizumab splits with durvalumab. BTC incidence is roughly stable (rising NASH/biliary-disease drivers offset by other trends); the mild forward drift reflects competition (biomarker-targeted later-line agents, IO-pool sharing with durvalumab) rather than a shrinking pool. | ||||||||||
| × Keytruda uptake (branded) | 20% | 19% | 20% | 19% | 19% | 19% | 21% | 7% | share, not summed | ASSUMED |
Per-market branded uptake of the addressable pool this year (launch ramp net of competitors & biosimilars). See the competitive curve below; pick a market for its access basis. | ||||||||||
| = Treated with Keytruda | 1.3k | 210 | 294 | 225 | 120 | 179 | 1.7k | 1.3k | 5.3k | computed · assumed input |
Addressable pool × branded uptake. | ||||||||||
| × Net price / patient (2026) | $90K | $42K | $46K | $57K | $21K | $54K | $36K | $29K | per patient, not summed | derived (course-adj.) |
Same molecule and per-mg net as the other Keytruda models, but realized exposure in advanced BTC is short (rapid progression; the chemo-IO course is brief), so per-patient revenue is modelled at a short-course fraction of the standard Keytruda per-market nets, set roughly in line with the TOPAZ-1 BTC nets for the same indication. Low-end HTA ratings further constrain price. China is self-pay. −4%/yr forward; flagged modelled. SHOWN FOR 2026: the per-market figures in "Net pricing: assumptions & basis" below are the 2024 basis; this row applies -4%/yr from 2024 (×0.9216), so it is the same source value at a later date, not a second price. | ||||||||||
| = Net revenue / yrOutput | $116M | $8.9M | $13.6M | $12.8M | $2.5M | $9.8M | $59.9M | $37.2M | $260M | computed · assumed input |
Treated patients × realized net price. | ||||||||||
Regional revenue mix · 2026 modelled
Model estimate, sourced from the funnel table above
No reported competitor-share series exists for this setting, so every trajectory below is modelled from approval timing, label breadth and the funnel, not reported data.
In-class share, 1L immunotherapy in advanced biliary tract cancer: 8 modelled markets (US, EU5, JP, CN)
Each agent's share of the addressable 1L pool (advanced BTC fit for gem/cis + a checkpoint inhibitor, all-comers). Durvalumab was first (2022) and holds the larger share; the branded pembrolizumab line ramps from its 2023 launch to a minority co-share (~40-45% of the IO pool) as it matures, then erodes from ~2028 as the Keytruda molecule loses exclusivity. The gap below the top lines is patients on chemo alone / best supportive care.
Share of the 39k-patient addressable 1L-treated pool this regimen is modelled to treat, by year
Keytruda is modelled at 14% of this pool in 2026, peaking at 15% in 2027; each point of share is revenue on the chart above.
Scope: 8 modelled markets (US, EU5, JP, CN) · follows the market selector above
Rivals' share, same pool
Durvalumab (Imfinzi, TOPAZ-1) + gem/cis holds the largest rival share in 2026, 9% of the pool; that is the share Keytruda has to take or hold.
Rivals, rescaled - note the axis maximum (10%), which is smaller than the plot above. Same data, a linear axis of its own, so a low-share entrant is readable rather than a hairline on the baseline.
Category timing = per-market approval/access (durvalumab first, 2022–24; pembrolizumab KEYNOTE-966 from Oct 2023 / 2024 ex-US). China share reflects self-pay dynamics (no confirmed NRDL listing for either agent). Pembrolizumab biosimilars timed to the Keytruda molecule LOE (~2028 US / ~2029 EU). Competitor trajectories are stated assumptions, not reported shares.
Readouts, approvals and LOE events for pembrolizumab and its rivals in this setting. Estimated timing is flagged; confirmed dates cited from Merck & Co. disclosures and regulator databases.
Emerging FGFR/HER2-directed and ADC 1L combinations could fragment the all-comers pool toward biomarker-selected regimens
Keytruda composition-of-matter expiry (~2028 US / ~2029 EU); pembrolizumab biosimilars erode this line earlier than durvalumab's ~2031-32 molecule LOE
✓ = reported public fact, cited. Upcoming timing is estimated from company guidance and trial registries, a projection, not a commitment. Approvals = US (FDA) · EU (EMA) · Japan (PMDA) · China (NMPA). ● = pembrolizumab, competitor tag = a same-setting rival.
Residual unmet need: why the field stays alive
Even with chemo-immunotherapy, median OS is only ~13 months and most patients have died by 3 years, BTC remains one of the worst-prognosis GI cancers. The modest absolute gain (~1.8-month median in KEYNOTE-966) and the unmet need in biomarker-defined subsets (FGFR2, IDH1, HER2, addressed only in later lines) keep the field active and constrain pembrolizumab's pricing power against an equally-priced durvalumab.
The synthesis for this indication, grounded in the cited trial, guideline and access facts above.
Rivals in this setting (1L advanced biliary tract cancer)
Swipe to explore →
| Rival / regimen | Developer | Position in this setting | Approved (date, trial, n) | Head start vs this regimen (months) | Source |
|---|---|---|---|---|---|
| Durvalumab + gem/cis (TOPAZ-1) (Live threat) | AstraZeneca | Live threatwhyPOSITIVE & APPROVED, OS 12.8 vs 11.5 mo, HR 0.80 (FDA Sep 2022); first IO ever in BTC, ~14 months ahead of pembrolizumab, holds the larger share of the IO pool. | not in the record | not in the record | TOPAZ-1, durvalumab + gem/cis, 1L BTC (Oh, NEJM Evidence 2022) |
| Gemcitabine/cisplatin (ABC-02) (Low) | - | LowwhyThe prior 1L standard since 2010 (OS 11.7 vs 8.1 mo); now the chemo backbone both IO agents are added to. | not in the record | not in the record | ABC-02, gemcitabine/cisplatin 1L standard (Valle, NEJM 2010) |
| Pemigatinib / futibatinib (Watch) | Incyte / Taiho | WatchwhyApproved 2L+ for FGFR2-fusion cholangiocarcinoma only, biomarker-restricted, not a 1L competitor. | not in the record | not in the record | FIGHT-202 / FOENIX-CCA2 |
| Ivosidenib (Tibsovo) (Watch) | Servier | WatchwhyApproved 2L+ for IDH1-mutant cholangiocarcinoma, biomarker-restricted, later-line. | not in the record | not in the record | ClarIDHy |
| Zanidatamab (Ziihera) (Watch) | Jazz / Zymeworks | WatchwhyAccelerated approval (Nov 2024) for 2L+ HER2-positive BTC, biomarker-restricted, later-line. | not in the record | not in the record | HERIZON-BTC-01 |
Not every rival is approved in every region, so a share here is the modelled position in the markets where it competes, not a global one.
Net pricing: assumptions & basis
Pembrolizumab is dosed 200 mg q3w (or 400 mg q6w) IV with chemo then maintenance up to 35 cycles. But BTC progresses fast (median OS ~12.7 mo), so realized exposure is short, per-patient revenue is modelled from each market's published per-mg net × a ~7-month course, and low-end HTA benefit ratings hold prices down. Set roughly in line with the TOPAZ-1 BTC nets, since the indication and regimen exposure are the same.
Figures below are 2024 net prices, the basis the model is anchored on. The funnel applies -4%/yr from 2024, so the price it consumes for a later year is lower than the figure quoted here. Both come from this one value; the funnel row states the year it is showing.
Same molecule per-mg economics as the other Keytruda models, but a short course: advanced BTC progresses rapidly, so realized exposure is well below a long-running 1L-NSCLC course. Per-patient net is modelled at a short-course fraction of the standard Keytruda per-market nets and aligned to the TOPAZ-1 BTC values (same indication, same regimen exposure): US ~$98k, France ~$46k, Germany ~$50k (AMNOG), Italy ~$62k, Spain ~$23k (lowest Western), UK ~$59k (confidential PAS), Japan ~$39k (NHI). The modest ~1.8-month median OS benefit drives low-end HTA ratings that constrain pricing. China is self-pay (no confirmed NRDL listing), ~$31k net. All figures are modelled: no indication-specific net price is publicly disclosed.
What it shows. The approval-to-reimbursement journey by market: when each payer granted access, on what terms (PD-L1 or comparator restrictions), and the lag between regulatory approval and reimbursed revenue.
What to look for. The access-lag column, approval is not revenue and some markets take a year or more. And the restrictions that narrow the eligible population below the full label.
Common misread. Assuming a US price or approval transfers to other markets. Ex-US payers benchmark against comparators and reference prices, so the net price and the covered population are usually smaller.
Rows ordered by access speed. The approval→reimbursement lag is commercially load-bearing, and it is not one number per drug: a monotherapy indication and a later combination extension file separately with a payer and can reach routine commissioning years apart, even within the same asset’s first-line setting.
| Market | HTA body | Decision | Access lag | PD-L1 | Comparator | Key outcome / MEA note |
|---|---|---|---|---|---|---|
| US | FDA | 31 Oct 2023 | reference | None (all-comers) | Gem/cis | Second IO regimen approved in BTC (~14 mo after durvalumab); Medicare Part B (J9271); no biomarker gate |
| EU | EMA / EC | 2024 | ~0.5 yr | None | Gem/cis | CHMP positive then EC approval; all-comers, co-standard with durvalumab |
| Japan | PMDA / NHI | 2024 | ~0.6 yr | None | Gem/cis | Unresectable/advanced BTC; NHI-listed; Japan is a high-incidence BTC market |
| Germany | G-BA / IQWiG | 2024 | ~0.8 yr | None | Gem/cis | G-BA likely minor added benefit (mirrors TOPAZ-1 'geringer'); AMNOG-rebated |
| France | HAS | 2024 | ~0.8 yr | None | Gem/cis | SMR important; ASMR IV (minor) expected, modest benefit caps price |
| China | NMPA | 2024 | ~1.0 yr | None | Gem/cis | NMPA approval; NOT confirmed on the NRDL, largely self-pay despite high incidence (as for durvalumab) |
| UK | NICE | 2024-25 | ~1.4 yr | None | Gem/cis | Appraised after TOPAZ-1 (TA944); recommended with a confidential commercial arrangement |
How much each assumption could move the forecast, one at a time, over the same cited baseline.
This is a sensitivity illustration, not a probability: it shows how far each assumption could move the forecast, never how likely it is.
Set a target and solve for one lever, what it would have to reach, benchmarked against real comparable launches.
One lever at a time: the other assumptions stay at baseline, so the trade-off stays visible.
A model that shows where its own risk is, not a number to trust. Never investment advice.
Cite this forecast as recorded on Q2 2026 → A permanent record of these figures that does not change when the model does.
Every reported figure here traces to a public source below; the model's own assumptions are labelled where they appear.
Reported figures are public facts from company filings; the forward peak is Ormeval's own modelled estimate, not company guidance and not analyst consensus. Approval and readout dates are cross-checked against public regulator and company disclosures. Not investment advice.