IPG Health @ ESMO 2025 | Key takeaways for healthcare marketers

By Sommer Bazuro, PhD, Chief Medical Officer, IPG Health; Heather McCune, Group Director, Medical Strategy, FCB Health New York; Ann Gordon, Senior Medical Director, CMC Connect; Louise Picken, Senior Medical Strategy Director, CMC Connect

Our experts from IPG Health, CMC Connect and FCB Health New York share their insights after attending ESMO Congress 2025, the annual event that brings together experts from around the world with the goal of driving progress in cancer diagnosis, treatment and survivorship.

• AI and LLMs take center stage at ESMO 2025
• Changing perspectives of the patient experience

• ADCs stole the show across breast cancer settings

• Enhertu is the ADC that continues to redefine the treatment of HER2+ breast cancer

• New hope for patients with GU cancer with ADCs showing big gains here

• Radioligands: From niche modality to mainstream oncology

• Tumor-agnostic medicine: A new frontier

AI and LLMs take center stage at ESMO 2025

This year’s ESMO congress highlighted a pivotal shift toward intelligent innovation. AI and large language models (LLMs) are transforming how oncology data is captured, interpreted and applied across research and clinical practice. AI is no longer theory—it is tangible practice. Across multiple sessions, oncologists presented real-world data showing how LLMs are improving data quality, reducing manual errors and reshaping everything from tumor boards to clinical trials. The publication of ESMO’s first formal guidance, the ESMO Guidance on the Use of Large Language Models in Clinical Practice (ELCAP), marked a pivotal regulatory moment, setting the boundaries for responsible AI adoption in oncology.

Here’s what we learned:

LLMs improve real-world data accuracy and speed.

In the LUCC lung cancer cohort, Mihaela Aldea’s team at Gustave Roussy showed a fine-tuned Mistral-24B LLM reduced data abstraction errors by >50%, cutting processing time from 17.5 to 1.7 minutes per patient, with hybrid human AI review achieving 4.4% error rates.

This proves LLMs can outperform trained research staff across 10 centers, enabling scalable, high-quality real-world evidence collection.

Agentic AI shifts from passive to autonomous reasoning.

Daniel Truhn’s "Agentic AI in Oncology" illustrated AI agents capable of independently sourcing PubMed, guidelines and imaging data to recommend evidence-based treatments, like trastuzumab emtansine for HER2+ early breast cancer.

It reframes AI from a static assistant to a co-pilot—an agent that acts, reasons and self-corrects, prompting discussions about accountability and trust.

ESMO’s ELCAP defines three AI categories for clinical safety.

The ELCAP framework—published in Annals of Oncology—formally delineated Type 1 (patient-facing), Type 2 (HCP-facing) and Type 3 (background institutional) AI systems. Each comes with consensus guidance (e.g., maintain human oversight, avoid autonomous diagnosis, validate continuously).

ESMO is establishing global governance norms, mirroring European Medicines Agency (EMA)-level precision for AI regulation.

LLMs accelerate drug discovery and trial recruitment.

Loïc Verlingue (Centre Léon Bérard) presented how generative AI models such as Cell2Sentence (C2S) can identify novel immunotherapy targets, while trial-matching LLMs boosted patient eligibility prediction rates from <8% to >20%, improving efficiency across early-phase trials.

AI isn’t just summarizing data, it’s now directly influencing pipeline acceleration and trial access.

For oncologists, AI’s growing role in data curation, clinical reasoning and patient interaction may transform the rhythm of care. As LLMs automate documentation and streamline data retrieval, oncologists can spend more time in direct patient dialogue and less on administrative load. However, the emergence of agentic AI systems introduces a new layer of clinical responsibility requiring oncologists to evolve from "information interpreters" to AI supervisors, ensuring that recommendations align with patient context, ethics and emotional nuance. Over time, this may redefine what constitutes medical expertise blending clinical judgment with digital literacy, trust calibration and shared decision-making alongside machines.

While AI promises unprecedented precision, we were also reminded throughout the event that technology is only as impactful as the human experience it supports. As data becomes cleaner and faster, understanding what that data means for patients becomes even more critical.

Changing perspectives of the patient experience

Many oncology trials report "manageable" safety profiles, but manageability does not always translate to a tolerable experience for patients. What resonated most across sessions:

• Shining a light on the frequent disconnect between clinician perceptions and patient experience: Dr. Anne-Marie Dingemans highlighted two studies with “manageable” safety profiles, despite high rates of diarrhea, while Dr. M. Fernanda Mosele shared historical data revealing low rates of adverse event reporting by clinicians and poor agreement between patient and clinician assessment of adverse event severity.
• Objective assessment of patient quality of life must become the norm in clinical trials: Both discussants offered solutions to closing this gap. Dr. Mosele highlighted the low probability that studies will publish patient-reported outcome (PRO) data and the need for instruments that “capture granular, patient-relevant and clinically meaningful quality of life differences.” Dr. Dingemans discussed best practices for reporting toxicity in clinical trials, particularly avoiding the use of Investigator-assigned Subjective or Judgmental Efficacy and Toxicity (ISJET) terms.

Robust PRO assessment and avoidance of subjective descriptors can provide a more realistic picture of how patients experience adverse events and what being on treatment will look like for them. Empowering patients with knowledge of what to expect and preparing them to manage side effects can help to keep them on their life-changing treatment.

ADCs stole the show across breast cancer settings

Nowhere was the intersection of data-driven innovation and patient impact more visible than in the rapid evolution of antibody–drug conjugates (ADCs).

From curative-intent HER2+ early disease to first-line (1L) metastatic triple-negative breast cancer (TNBC), ADCs dominated ESMO. ADCs utilize a targeting system (the antibody) with a payload (a powerful cancer drug to deliver the toxic drug directly inside the cancer cell, sparing most healthy cells). Readouts for Enhertu, Datroway and Trodelvy reshaped near-term standards and the messaging chessboard. Multiple tumor settings, two targets (HER2, TROP2) and hard endpoints invasive disease-free survival (IDFS) and overall survival (OS) all moved in the right direction, turning "standard chemo-sparing" from slogan into a data-backed story clinicians can act on.

We learned:

• Enhertu ups the ante in early HER2+: In DESTINY-Breast05, Enhertu reduced the risk of invasive disease recurrence or death by 53% vs trastuzumab emtansine (T-DM1), signaling a new post-neoadjuvant (treatment before surgery) benchmark.
• Datroway delivers first 1L TNBC OS win: TROPION-Breast02 showed longer OS 23.7 months for Datroway vs 18.7 months of investigator’s choice of chemotherapy. The OS benefit in this population vaults TROP2 ADCs to the front line.
• Trodelvy challenges chemo in 1L TNBC: ASCENT-03 improved progression-free survival (PFS) 9.7 vs 6.9 months (≈38% risk reduction) in PD-L1-ineligible TNBC; OS is maturing, but the bar for “standard chemo” is clearly moving.

For healthcare/oncology marketing, expect a shift from "last-line rescue" to front-line ADC positioning with clear standard chemo-sparing narratives and value dossiers anchored in OS/IDFS. Readiness now means diagnostics + safety pathways (e.g., interstitial lung disease (ILD) vigilance for topoisomerase-I ADCs) and dosing-convenience messaging (e.g., Datroway’s q3w schedule vs biweekly competitors) to differentiate in crowded TROP2 lanes. These data also strengthen access stories for health technology assessment/US payers by tying hard endpoints to quality-of-life–leaning claims.

With increased ADC presence comes growing questions around sequencing the ADCs relative to each other. We expect more weight to be placed on unique targets and payloads, along with amplified urgency to be used first in the treatment algorithm.

Enhertu is the ADC that continues to redefine the treatment of HER2+ breast cancer

Among them, Enhertu stood out, continuing to push the boundaries of HER2+ treatment and redefining therapeutic benchmarks.

As mentioned, the unstoppable momentum of new innovations in the treatment of HER2+ breast cancer was fully on display at ESMO 2025, with potentially practice-changing data from Enhertu presented across treatment settings.

Let’s dig deeper:

• Primed for use in the curative-intent setting: The results of DESTINY-Breast11 and DESTINY-Breast05 trials of Enhertu regimens in the neoadjuvant and post-neoadjuvant settings were among the most anticipated of the congress. DESTINY-Breast11 reported a “statistically significant and clinically meaningful improvement in pathologic complete response (pCR)” rate over a standard of care regimen of dose-dense doxorubicin and cyclophosphamide followed by paclitaxel, trastuzumab, and pertuzumab (ddAC-THP). The DESTINY-Breast05 data were described as “practice-changing” for patients with high-risk residual disease.
• Effective in the 1L metastatic setting across subgroups: Results of the DESTINY-Breast09 study in the 1L showing that Enhertu gives a “highly statistically significant” improvement in PFS over the standard of care (THP) were presented this year at ASCO. Analysis presented at ESMO showed consistent Enhertu benefit across three critical subgroups: prior treatment, hormone receptor (HR) status and PI3KCA mutation status.

For many years, the HER2+ breast cancer treatment algorithm across early-stage and metastatic settings was relatively straightforward and dominated by Herceptin-based regimens. Therapeutics like Enhertu are changing that, and the algorithm will only become more complex, particularly as additional targeted regimens enter the picture and treatment selections become more complex. For marketers, understanding the patient population where a product will shine, and communicating that story in a simple and straightforward way, is the key to navigating a complex landscape.

New hope for patients with GU cancer with ADCs showing big gains here

The promise of ADCs didn’t stop at breast cancer. Similar breakthroughs are now reshaping outcomes for patients with genitourinary (GU) cancers.

There was a strong focus on GU cancers throughout the event. Important findings included:

• Disitamab vedotin + toripalimab nearly doubles OS vs chemo with median OS 31.5 vs 16.9 months; PFS 13.1 vs 6.5 months; in HER2-expressing locally advanced metastatic urothelial cancer (la/mUC).
• Few moments at ESMO stopped the room like the EV-303/KEYNOTE-905 data, looking at enfortumab vedotin in combination with pembrolizumab in the perioperative setting for patients with muscle-invasive bladder cancer (MIBC) who were cisplatin-ineligible. It received three rounds of applause as the data were being presented; the results were classed as transformational and may offer a potential new standard of care for this patient population.
• Biomarker testing and the role of ctDNA was explored across GU cancers to look at personalizing treatment options in the same way that has already been seen for breast and lung cancer.

This was an inspiring meeting for the treatment of GU cancers—some long-standing questions were finally answered. What was even more exciting was that the GU oncology community gained insights into which direction to take to personalize care and avoid unnecessary treatments being administered.

Radioligands: From niche modality to mainstream oncology

Beyond ADCs, ESMO 2025 underscored that precision delivery is expanding in multiple directions, including the rapidly growing field of radioligand therapy. Radioligand therapies represent another face of targeted innovation, fusing diagnostics and treatment in ways that mirror the ADC revolution.

Theranostics—the fusion of diagnostic imaging and targeted radionuclide therapy—is moving from a specialist niche to a central pillar of oncology innovation. At ESMO 2025, the field showed rapid expansion and diversification, powered by both established players and emerging biotech entrants.

We observed:

• ESMO discussions showed a surge in new radionuclides, ligands and tumor targets beyond prostate and neuroendocrine cancers. Isotopes such as actinium-225 and lead-212 are being paired with new ligands targeting DLL3, FAP and mesothelin signaling a maturing, multi-tumor nuclear medicine ecosystem.
• Once led by niche radiopharma firms, the space now includes AstraZeneca, Bayer, Johnson & Johnson and Sanofi, accelerating translation and driving new diagnostic–oncology partnerships. 
• The PSMAddition study of Pluvicto in metastatic hormone-sensitive prostate cancer reflects a trend toward earlier-line use. Trials are also exploring radioligands in combination with immuno-oncology and targeted agents, positioning them as core components of multimodal therapy.
• A consistent theme was the need to expand isotope production, radiopharmacy logistics and clinical delivery infrastructure to ensure patient access to these highly targeted and effective therapies.

The radioligand evolution is both a storytelling and systems challenge. The narrative must link diagnostics and therapy in a unified value story, connecting innovation to diagnostics, logistics and infrastructure and communicating the transformative potential of this modality in clear, scalable ways across clinical, payer and policy audiences.