ISPOR Europe 2026: five shifts for market access teams in AI, JCA, MFN pricing and evidence

The meeting at a glance

ISPOR Europe 2026 ran from 8 to 11 November at the Austria Center Vienna under the theme "Shaping the Future of Access Through Science and Policy". The programme listed more than 140 sessions and almost 3,000 posters. The co-chairs were Malina Müller (WifOR Institute), Sandra Nestler-Parr (BioCryst Pharmaceuticals) and Rasmus Prior Gjesing (WHO Europe).^[1] More than 5,000 delegates registered.^[2]

The three plenaries set the agenda: "The Science of What Matters: Rethinking Value in Healthcare", "We Optimize Treatments, But How Do We Optimize the System?" and "Fit for Purpose: What HEOR Evidence Must Become".^[3] Read together, they asked one question of every access team: is your evidence built for the decisions that will actually be made with it?

Five shifts stood out. Each comes with a short technical note on what it means for submissions.

1. AI in evidence synthesis moved from pilots to good practice

For the past two years, AI sessions at ISPOR have mostly been demonstrations. This year the discussion turned to standards. The clearest marker is the ISPOR Special Task Force report on generative AI for systematic literature reviews, published in Value in Health.^[4]

The task force reviewed 115 empirical studies published between November 2022 and July 2025, across seven review tasks.^[4] Its conclusions are measured:

• Strongest support: High-recall title and abstract screening, and structured first-pass data extraction, within human-in-the-loop workflows with explicit oversight.^[4]
• Conditional: Full-text screening, qualitative synthesis and report writing, depending on workflow design and oversight.^[4]
• Lowest readiness: Risk of bias assessment.^[4]
• Not supported: Autonomous deployment, and end-to-end autonomous review generation.^[4]

Supported, with a human in the loop

• High-recall title and abstract screening
• Structured first-pass data extraction
• Both with explicit oversight and verification

Conditional

• Full-text screening
• Qualitative synthesis
• Report writing
• Depends on workflow design and oversight

Lowest readiness or not supported

• Risk of bias assessment
• Autonomous deployment
• End-to-end autonomous review generation

Source: ISPOR Special Task Force, 115 empirical studies published November 2022 to July 2025.^[4]

On the agency side, NICE used the meeting to introduce its AI Methods Framework, a best-practice guide to the use of AI across the HTA process, with draft principles and use-case guidance open for stakeholder feedback.^[5]

Technical insight

The practical line is now clear. AI can speed up screening and first-pass extraction if the workflow is documented, a human verifies the output and the method is reported. It should not be used to judge risk of bias or to write conclusions unsupervised. For JCA and NICE submissions, describe any AI use in the methods section with the same transparency as any other screening tool. We set out the wider picture in our analysis of AI in HTA.

2. The EU JCA became a practice, not a forecast

A year ago the joint clinical assessment was a set of guidance documents. By ISPOR Europe 2026 it had a body of published reports. The Commission published the first, on tovorafenib (Ojemda) for paediatric low-grade glioma, on 9 June 2026. At that point 18 JCAs had been initiated since the Regulation started to apply in January 2025, all for cancer medicines and advanced therapies.^[6]

Sources: European Commission.^[6,7,8,9]

1. 9 June 2026Tovorafenib (Ojemda)First JCA report: paediatric low-grade glioma
2. 8 July 2026Tarlatamab and lurbinectedinTwo reports in small cell lung cancer
3. 7 September 2026Intrathecal onasemnogene abeparvovec (Itvisma)First report for an advanced therapy
4. 6 November 2026Camizestrant (Etcamah)Fifth report: ESR1-mutated breast cancer
5. 13 January 2028Orphan medicines join the scopeAlongside cancer medicines and advanced therapies
6. 13 January 2030All new medicines in scopeEvery new centrally authorised medicine

Sources: European Commission.^[6,7,8,9,10]

The first report showed what the process asks of evidence. Member State requests for tovorafenib were consolidated into eight PICOs. Only one produced a comparative analysis, and its results were inconclusive: after weighting, the effective sample size fell to between 5.81 and 6.64.^[11] The authors of an independent review stress that an inconclusive JCA is not a reason to deny access, because reimbursement decisions stay national.^[11]

Across the five reports, the scope ran from 1 to 12 PICOs, and the share answered with comparative evidence varied as widely. Randomised trials against a relevant comparator, or networks that could be anchored, answered most of their PICOs. Single-arm evidence answered almost none.^[12,7,8,9]

• Tovorafenib: 1 of 8 PICOs (single-arm trial; one unanchored MAIC)13%
• Tarlatamab: 7 of 7 PICOs (RCT vs topotecan plus indirect comparisons; some outcomes missing)100%
• Lurbinectedin: 1 of 1 PICO (RCT vs the single comparator)100%
• Intrathecal onasemnogene abeparvovec: 7 of 12 PICOs (sham-controlled RCT plus MAICs)58%
• Camizestrant: 2 of 5 PICOs (one RCT analysis vs one of three comparators, applied to everyone and to women)40%

Sources: JCA summary reports, European Commission.^[12,7,8,9] PICOs in the final assessment scope.

1. 7 PICOsThe first assessment scope, for the originally claimed indication
2. 5 PICOsAfter the indication was narrowed in January 2026, across three populations: everyone, women, men
3. 3 comparatorsCDK4/6 inhibitor + aromatase inhibitor, CDK4/6 inhibitor + fulvestrant, aromatase inhibitor alone
4. 2 PICOs answeredOne SERENA-6 analysis, against continued aromatase inhibitor + CDK4/6 inhibitor

• AnsweredDirect comparison in PICO 1, with the women-only PICO covered by the same data (99% of patients)
• Not answeredFulvestrant and aromatase inhibitor monotherapy comparators; men (three patients, all in the control arm)
• Left out of the reportPatient-reported outcomes and response rate, for missing data and few events

Source: JCA summary report on camizestrant.^[9]

The scope widens on a fixed timetable: orphan medicines from 13 January 2028 and all new centrally authorised medicines from 13 January 2030.^[10]

Technical insight

Single-arm evidence will keep meeting multiple PICOs with comparators the trial never studied. The planning implications are early PICO simulation, an indirect comparison plan for each likely comparator, and honest reporting of effective sample size. Our guide to JCA PICO scoping covers the method, and JCA Pulse tracks every ongoing assessment.

3. US MFN pricing became a European launch question

Most-favoured-nation pricing was everywhere in Vienna, from posters mapping European responses to payer surveys in the EU5.^[13,14] Three US models are in play:

Sources: King & Spalding; Skadden.^[15,16]

GENEROUS

• Medicaid, voluntary, five years
• Announced November 2025
• Second-lowest net price across 8 countries
• Adjusted by GDP per capita (PPP)

GLOBE

• Medicare Part B, mandatory
• Finalised 30 September 2026
• Voluntary data collection from 1 January 2027
• 19 reference countries; seven treatment categories over $100m Part B spend

GUARD

• Medicare Part D, mandatory
• Proposed; under White House review in October 2026
• 19 reference countries
• Rebates calculated annually

Sources: King & Spalding; Skadden.^[15,16]

GLOBE is narrower than first proposed. It covers single-source drugs and biologics in seven treatment categories with more than $100 million in Part B spending, and excludes orphan-only drugs, cell and gene therapies, plasma-derived products and drugs with a negotiated maximum fair price.^[16]

Technical insight

When US prices reference European net prices, a confidential discount in one market can carry a cost in another. Launch sequencing, the choice between list-price and net-price agreements, and the design of managed access all need a global view. Our MFN pricing analysis sets out the reference baskets and the GDP (PPP) adjuster.

4. "Fit for purpose" evidence replaced "more" evidence

The third plenary asked what HEOR evidence must become.^[3] The recurring answer was evidence designed for a decision, rather than evidence collected and then fitted to one. The JCA experience makes the point sharply: a pivotal single-arm study, however well run, cannot answer eight PICOs on its own.^[11]

For real-world evidence, that means stating the question first, choosing data that can answer it, and reporting how bias was handled. External control arms and population-adjusted comparisons will be judged on their transparency as much as their results.

Technical insight

Treat every comparative claim as something an assessor will try to reproduce. Pre-specify the analysis, report effective sample size and overlap, and show sensitivity to unmeasured confounding. Our notes on population-adjusted indirect comparisons go into the detail.

5. The definition of value kept widening

The opening plenary, "The Science of What Matters", and the system-level focus of the second plenary both pushed beyond the incremental cost per QALY.^[3] The discussion covered the value of a whole system, not only a single product.

The JCA shows the tension. Its scope excludes caregiver burden and some quality-of-life factors, which remain for national assessment.^[11] Wider value elements will therefore enter decisions country by country, not through one European framework.

Technical insight

Build wider value evidence into the national layer of the evidence plan: carer outcomes, productivity and service impact. Collect them prospectively where possible, because retrofitting them at submission is rarely convincing.

What this means for access teams

1. Write an AI use policy for evidence work: Define which tasks can use AI, who verifies the output and how it is reported.
2. Simulate the JCA scope early: Draft likely PICOs before the pivotal trial reads out, and plan indirect comparisons for each.
3. Model MFN exposure before European pricing: Map which European prices could feed US reference baskets.
4. Design evidence for the decision: Agree the questions that HTA bodies will ask, then choose data and methods that answer them.
5. Plan wider value at national level: Collect carer, productivity and service outcomes where national bodies will weigh them.
6. Keep one evidence plan: Align JCA, NICE and national submissions so that each answers its own question from the same core. Our guide to one evidence plan for NICE and the JCA shows how.

Dates for 2027 and 2028

Sources: ISPOR; HTAi.^[17,18]

The GLOBE data collection period also begins on 1 January 2027.^[16]

If any of these shifts touches a launch you are planning, follow the JCA reports as they land in JCA Pulse, or talk to us about your evidence plan.

References

1. Becaris Publishing. Looking ahead: ISPOR Europe 2026, HEOR heads to Vienna. The Evidence Base [Internet]. London: Becaris Publishing; 2026 [cited 2026 Nov 10]. Available from: https://becarispublishing.com/digital-content/blog-post/looking-ahead-ispor-europe-2026-heor-heads-vienna
2. ISPOR. ISPOR Europe 2026: know before you go [Internet]. Lawrenceville (NJ): ISPOR; 2026 [cited 2026 Nov 10]. Available from: https://www.ispor.org/docs/default-source/ispor-europe-2026/ispor-europe-2026---know-before-you-go.pdf
3. ISPOR. ISPOR Europe 2026 plenaries announced [Internet]. Lawrenceville (NJ): ISPOR; 2026 Sep 30 [cited 2026 Nov 10]. Available from: https://www.ispor.org/heor-resources/news-top/news/view/2026/09/30/ispor-europe-2026-plenaries-announced
4. Fleurence RL, Qureshi R, Aggarwal R, et al. Generative artificial intelligence for systematic literature reviews: a Good Practices Report of an ISPOR Special Task Force. Value Health. 2026 [Epub ahead of print] [cited 2026 Nov 10]. doi:10.1016/j.jval.2026.08.007. Available from: https://www.ispor.org/heor-resources/good-practices/article/generative-artificial-intelligence-for-systematic-literature-reviews--a-good-practices-report-of-an-ispor-special-task-force
5. ISPOR. Responsible AI in HTA: introducing the NICE AI Methods Framework. ISPOR Europe 2026 session [Internet]. Lawrenceville (NJ): ISPOR; 2026 [cited 2026 Nov 10]. Available from: https://www.ispor.org/heor-resources/presentations-database/session-cti/ispor-europe-2026/responsible-ai-in-hta--introducing-the-nice-ai-methods-framework
6. European Commission. Commission publishes first joint clinical assessment on innovative medicine [Internet]. Brussels: European Commission; 2026 Jun 9 [cited 2026 Nov 10]. Available from: https://health.ec.europa.eu/document/download/b853b3f5-39d1-4aa3-8787-b1125d0d06d1_en?filename=hta_20260609_news-first-jca-report_en.pdf
7. European Commission, Directorate-General for Health and Food Safety. Commission publishes joint clinical assessment reports on two new lung cancer medicines [Internet]. Brussels: European Commission; 2026 Jul 8 [cited 2026 Nov 10]. Available from: https://ec.europa.eu/newsroom/sante/items/947372/
8. European Commission. Joint clinical assessment report on onasemnogene abeparvovec (Itvisma) [Internet]. Brussels: European Commission; 2026 Sep 7 [cited 2026 Nov 10]. Available from: https://health.ec.europa.eu/publications/joint-clinical-assessment-report-onasemnogene-abeparvovec-itvisma_en
9. European Commission. Joint clinical assessment report on camizestrant (Etcamah) [Internet]. Brussels: European Commission; 2026 Nov 6 [cited 2026 Nov 10]. Available from: https://health.ec.europa.eu/publications/joint-clinical-assessment-report-camizestrant-etcamah_en
10. European Commission. Joint clinical assessment for medicinal products: factsheet [Internet]. Brussels: European Commission; 2025 Jan [cited 2026 Nov 10]. Available from: https://health.ec.europa.eu/document/download/ced91156-ffe1-472d-85eb-aa6a91dd707e_en?filename=hta_htar_factsheet-jca_en.pdf
11. Gilardino R, Cardone A, Cossio CM. Rigorous, transparent and inconclusive: what Europe's first joint clinical assessment reveals about a framework trying to serve everyone. J Comp Eff Res. 2026;15(10):e260141. doi:10.57264/cer-2026-0141. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC13648050
12. European Commission. Joint clinical assessment report on tovorafenib (Ojemda) [Internet]. Brussels: European Commission; 2026 Jun 9 [cited 2026 Nov 10]. Available from: https://health.ec.europa.eu/publications/joint-clinical-assessment-report-tovorafenib-ojemda_en
13. ISPOR. One policy, many responses: mapping European strategies to U.S. most-favoured-nation pricing. ISPOR Europe 2026 poster [Internet]. Lawrenceville (NJ): ISPOR; 2026 [cited 2026 Nov 10]. Available from: https://www.ispor.org/heor-resources/presentations-database/presentation-cti/ispor-europe-2026/poster-session-1-5/one-policy-many-responses-mapping-european-strategies-to-u-s-most-favoured-nation-pricing
14. ISPOR. Most favoured nation, least favoured markets? EU5 payer perspectives on US pricing reforms. ISPOR Europe 2026 poster [Internet]. Lawrenceville (NJ): ISPOR; 2026 [cited 2026 Nov 10]. Available from: https://www.ispor.org/heor-resources/presentations-database/presentation-cti/ispor-europe-2026/poster-session-3-4/most-favoured-nation-least-favoured-markets-eu5-payer-perspectives-on-us-pricing-reforms
15. King & Spalding. Medicare and Medicaid demos seek to impose MFN pricing: an overview of GENEROUS, GLOBE and GUARD [Internet]. Atlanta: King & Spalding; 2025 Dec 29 [cited 2026 Nov 10]. Available from: https://www.kslaw.com/Insights/Articles/medicare-and-medicaid-demos-seek-to-impose-mfn-pricing-an-overview-of-generous-globe-and-guard
16. Turow R. CMS finalizes the GLOBE model: a most-favored-nation drug payment framework for Medicare Part B [Internet]. New York: Skadden, Arps, Slate, Meagher & Flom LLP; 2026 Oct 5 [cited 2026 Nov 10]. Available from: https://www.skadden.com/insights/publications/2026/10/cms-finalizes-the-globe-model
17. ISPOR. Upcoming conferences [Internet]. Lawrenceville (NJ): ISPOR; 2026 [cited 2026 Nov 10]. Available from: https://www.ispor.org/conferences-education/conferences/upcoming-conferences
18. Health Technology Assessment international. HTAi announces Brussels 2027 annual meeting [Internet]. Edmonton: HTAi; 2026 Jun 25 [cited 2026 Nov 10]. Available from: https://htai.org/htai-announces-brussels-2027-annual-meeting/

Sources accessed October 2026. This article is analysis for information and is not legal, regulatory or tax advice.