ESMO 2026 Insight | Part 2

Raising the Bar, Raising the Stakes

This is the second of four installments from ESMO 2026, released one per day over four days. ESMO 2026 Insight | Part 1 – Lumanity was made available on Tuesday and set the stage for the series.

This October, Lumanity’s oncology experts joined the crowds in Madrid for Europe’s largest cancer conference. Across tumor types, we heard the central question shift from “does it work?” to “where does it belong, for whom, and can patients get it?” Potent new agents are moving into earlier lines of therapy. That makes patient selection, sequencing, evidence quality, pricing and access as important as efficacy.

Executive summary: Today’s installment focuses on next-generation therapies in lung, pancreatic and ovarian cancer. The data show major improvements within existing drug classes, which is exciting for patients and clinicians. It also raises the pressure on sponsors to show a clear difference from competitors.

Below, our experts cover three themes:

• New combinations in lung cancer, built on an existing backbone of care
• Next-generation KRAS inhibitors in lung and pancreatic cancers
• New era of ADCs in ovarian cancer

What’s coming next:

• Part 3 / Thursday: Refining and redefining standard of care. Bladder cancer, glioblastoma, lymphoma and myeloma, and how to find the right treatment intensity for each patient.
• Part 4 / Friday: Looking forward to new paradigms. New immunotherapy modalities, radiopharmaceuticals and delivery, and new models of evidence generation.

---

Multiple Modalities Will Redefine Care in Lung Cancer

Tyler Nellos, MBS, Consultant, and Pamela Chiesi, MBA, SVP, Strategy Consulting

ESMO 2026 saw substantial progress in lung cancer with T-cell engagers showing transformative promise in ES-SCLC and EGFRm NSCLC patients likely to benefit from both biomarker-selected and agnostic post-TKI treatment options.

Dr. Rebecca Heist called it “a first step” toward a 1L practice change.

Tarlatamab sets a new bar in SCLC

Frontline ES-SCLC has historically been defined by modest gains, with chemo-immunotherapy delivering a median OS of roughly 12–13 months and most patients progressing soon after induction, leaving a substantial unmet need for more effective maintenance therapy. The phase 3 DeLLphi-305 study evaluated Amgen’s tarlatamab (Imdelltra; DLL3 × CD3 T-cell engager) plus AstraZeneca’s durvalumab (Imfinzi) as 1L maintenance after durvalumab plus platinum-etoposide induction. In 563 patients, the combination reduced the post-induction risk of death by 58% vs durvalumab alone (median OS 21.1 vs 12.7 months; HR 0.42) and nearly quadrupled the post-induction median PFS (7.4 vs 1.9 months; HR 0.34), with benefit across all subgroups, including patients with brain or liver metastases. The data earned a standing ovation at Presidential Symposium I and position tarlatamab plus durvalumab as a likely new standard of care in the 1L maintenance setting.

The caveats in this study are mostly practical. Survival was measured from maintenance randomization, so the trial captures only patients who did not progress during induction. Cytokine release syndrome occurred in half of patients (49.6%), although nearly all cases were grade 1–2. On-site monitoring is also required after the first two doses, which may slow uptake in community practices. Still, treatment-related discontinuations were low (3.2%). The next question is whether adding tarlatamab to induction as well can further improve outcomes for 1L ES-SCLC patients—a hypothesis being tested in the phase 3 DeLLphi-312 study.

Options abound in EGFRm NSCLC post-TKI

MET overexpression (OE) or amplification is found in roughly a third of patients progressing on osimertinib, yet there are no approved therapies targeting this resistance mechanism outside China. ESMO 2026 delivered the data to change that. The global Ph3 SAFFRON trial  enrolled 338 patients with MET-OE and/or amplification after osimertinib. HUTCHMED/AstraZeneca’s savolitinib (Orpathys; MET TKI) plus osimertinib (Tagrisso) improved both PFS (7.6 vs 5.4 months; HR 0.55) and OS (17.8 vs 12.9 months; HR 0.68) vs platinum-pemetrexed, and it did so with fewer grade ≥3 TRAEs (40% vs 52%). SAFFRON is the first global Ph3 after 1L osimertinib to demonstrate a statistically significant OS benefit, confirming China-only SACHI results, and an NDA has already been submitted to the FDA. Discussant Dr. Natasha Leighl declared it a new standard of care.

However, Dr. Leighl raised questions on whether MET OE alone is enough for patient selection, noting that only 45% of patients were both IHC3+ and FISH-amplified, while the rest were positive by just one test (27% IHC3+ only; 28% FISH-amplified only). Drawing on prior EGFR + MET TKI studies, she argued that true MET “addiction” (i.e., high-level amplification) may be required for benefit. Patient selection depends on a tissue biopsy at progression, which is not yet routine. Also, the trial comparator was chemotherapy rather than newer regimens now approved in the post-1L osimertinib setting.

Two other phase 3 trials in post-3rd generation TKI EGFRm setting read out with impressive data, though both are China-only trials. PANKU-Lung01 pitted Biokin/BMS’s izalontamab brengitecan (EGFR x HER3 bispecific ADC) against chemo and cut the risk of progression by 60%. In OptiTROP-Lung04 Kelun/Merck’s sac-TMT faced chemo and reported a final OS benefit of 25.9 vs 17.4 months. Both ADCs are being evaluated in global trials in the hope that the China-only data translate to post-TKI benefits for biomarker-agnostic patients.

In 1L EGFRm NSCLC, the Ph3 China-only SANOVO trial tested the savolitinib plus osimertinib combination in 1L MET-OE patients where the all oral regimen improved PFS over osimertinib (ITT population, 24.9 vs 13.8 months; HR 0.56) with an early OS signal (HR 0.48 at 40.5% maturity). But grade ≥3 TEAEs nearly doubled (77% vs 38%), and patients with IHC2+ tumors showed no clear PFS benefit (HR 0.85). Discussant Dr. Rebecca Heist called it “a first step” toward a 1L practice change. To make an impact outside China, the regimen will need global data—likely against the 3rd generation TKI combinations that are replacing monotherapy as 1L standard of care.

Takeaways

ESMO 2026 marked a shift toward more targeted and more potent strategies in lung cancer, with T-cell engagers moving into frontline ES-SCLC and biomarker selection extending MET targeting in EGFRm NSCLC. The next 12–18 months will test whether these gains translate into practice. Watch for regulatory decisions for tarlatamab maintenance and savolitinib plus osimertinib after osimertinib, and global ADC approaches that aim to transform lung cancer care.

---

Breaking the Undruggable: The Evolution of RAS Pathway Therapies

Viraj Parekh, PhD, Senior Advisor, Strategy Consulting

As we’ve seen with earlier conferences like ASCO earlier this year, ESMO 2026 marked a significant milestone in targeting RAS in oncology, showcasing paradigm-shifting data for next-generation RAS pathway inhibitors. These presentations underscored a clear evolution and noteworthy progress from the 1^st-gen KRAS G12Cs, sotorasib and adagrasib (approved in 2021-22), toward highly potent next-gen mutant-specific molecules, combination regimens, and pan-RAS coverage. These advancements offer the potential for robust survival advantages in late-stage settings while showing early-stage pipeline progress against previously untargetable alleles. However, as the field transitions into this next era, each breakthrough brings distinct clinical issues, sequencing challenges, and future operational considerations that drug developers must navigate to truly maximize patient benefit. 

Krascendo-1: Next-gen G12C inhibition for NSCLC

Headlining data from the registrational phase 3 Krascendo-1 trial established Roche’s next-generation KRAS G12C inhibitor divarasib as the new benchmark in advanced 2L+ NSCLC. Divarasib demonstrated clear superiority over sotorasib (Lumakras; Amgen) and adagrasib (Krazati; BMS), extending median PFS to 9.8 months (vs. 5.7 months; HR: 0.58) and median OS to 17.7 months (vs. 12.0 months; HR: 0.56) with an impressive ORR of 58.6%. Another key differentiator was the doubling of DOR, indicating that the optimization in its biochemical design, structural binding kinetics, and target occupancy has paid off.

Importantly, key subgroup analyses from Krascendo-1 provide vital clinical clarity, proving that divarasib’s structural modifications translate directly into meaningful neuro-oncological control and partial resistance bypass. Historically, significant CNS progression has been associated with KRAS G12Cm NSCLC, with sotorasib demonstrating poor BBB penetration. Divarasib’s impressive intracranial ORR of 54.8% and median intracranial PFS of 10.4 months effectively match or exceed the CNS benchmarks previously established by adagrasib. By demonstrating such strong CNS disease control, divarasib may offer a viable systemic alternative that can defer or minimize the need for aggressive whole-brain radiation therapy, thereby directly preserving cognitive function and quality of life in this population. 

However, the trial also highlighted the complex biological boundaries of single-allele mutant-selective inhibition when confronted with high-risk co-mutational profiles. While divarasib achieved a notable clinical victory by doubling median PFS to 6.4 months within the ultra-poor prognosis KEAP1/NFE2L2-mutant stratum, this sub-population still experienced rapid attrition compared to the overall intention-to-treat cohort. This discrepancy underscores the vulnerability of selective G12C inhibitors to rapid upstream reactivation and wild-type HRAS/NRAS bypass signaling. Furthermore, post-progression genomic profiling mapped clear structural acquired resistance limitations specific to divarasib monotherapy. Tumors ultimately escape target engagement via pocket-occluding Type I (Y96C, H95Q) and severe Type II (Y96D) structural mutations, alongside secondary KRAS-activating alterations (G12D, Q61H) in trans that reactivate downstream MAPK signaling and bypass direct G12C inhibition entirely. These subgroup dynamics suggest that the future of precision oncology will not belong to a single class; rather, clinical practice will require a nuanced, biomarker-driven triage to determine the utilization and combination of mutant-selective agents like divarasib and multi-RAS platforms for heterogenous, bypass-prone co-mutational profiles. 

Looking at the safety, while divarasib successfully minimized the severe liver toxicities common to first-generation drugs, it shifted the safety burden to the gut, with 9% grade 3/4 diarrhea (vs. 2.9% in the control arm) that places an additional supportive care and dehydration-management burden.

Furthermore, the trial protocol’s lack of a crossover mechanism upon progression does not allow for evaluation of divarasib’s efficacy post-progression. Additionally, from the trial design perspective, there appears to be a sharp control arm imbalance, with 91% of patients receiving sotorasib rather than providing a balanced head-to-head comparison against adagrasib, which historically tracks with slightly higher single-agent efficacy. Finally, the data is strictly from fit participants with stable, treated brain metastases, and will likely face performance attrition when translated to frailer, real-world oncology patients with active CNS disease. For divarasib, a primary future consideration will be determining optimal, data-driven sequencing strategies to manage this resistance, alongside determining how it will compete with emerging frontline combination regimens. 

Other data in KRAS-mutant NSCLC

Divarasib also faces significant competition from multiple KRAS G12C inhibitors in the 1L setting. One example showcased at ESMO was Lilly’s olomorasib, which achieved durable, clinically relevant activity when paired with pembrolizumab and pembrolizumab + chemotherapy (LOXO-RAS-20001 [P1/2] & SUNRAY-01 [P3 lead-in] trials), which compared favorably with historical data for sotorasib/adagrasib. Additionally, the data demonstrates that KRAS G12C inhibitors can be safely combined with pembro and chemotherapy. The rate of severe liver toxicity (grade 3+ ALT elevation: 15.3%) appears to be manageable, potentially opening the door for a safe, dual-targeted frontline regimen in this high-unmet-need KRASm population. However, the critical issue here is potential long-term toxicity, and we do not yet know if low-grade liver inflammation will compound over time, possibly leading to late-stage drug discontinuations or permanent liver injury. Hence, future considerations for these early-stage frontline combinations must focus on identifying optimized dosing schedules, such as lead-in or intermittent dosing, to maintain a safe therapeutic index.

Continued advances in KRAS-mutant PDAC

For KRAS G12D, in the phase 1 evaluation of Jiangsu Hengrui’s HRS-4642, a liposomal nanoparticle-formulated non-covalent inhibitor, monotherapy yielded a 20.8% ORR and a 79.2% DCR in R/R KRAS G12Dm PDAC patients, with an ORR of 43.2%-48.5% when paired with anti-PD-L1. In the frontline PDAC setting, the agent plus gemcitabine/nab-paclitaxel produced a robust 63.3% ORR. Concurrently, FIH data for Incyte’s INCB161734, an oral non-covalent dual “ON/OFF” state inhibitor, showcased a 37% monotherapy ORR and 78% DCR in R/R PDAC, and when combined with FOLFIRINOX in frontline PDAC, INCB161734 drove an impressive 89% ORR and a 100% DCR, paving the way for the global phase 3 DAWN-303 trial. Despite these encouraging responses, a central future hurdle remains navigating deep tissue drug delivery through the dense stroma of pancreatic malignancies to maximize target exposure and prevent the rapid activation of bypass signaling pathways like PI3K/AKT. Additionally, Revolution Medicines is already in phase 3 trials RASolute-305 and 309 for its KRAS G12D(ON) inhibitor, zoldonrasib, in 1L PDAC combinations, and has opened a phase 3 in 1L NSCLC as well.

Lastly, the late-breaking mature subgroup analysis of the phase 3 RASolute-302 trial establishes daraxonrasib as a pan-genotypic standard of care for R/R PDAC. By demonstrating a consistent OS benefit across all major KRAS variants: including G12D (HR 0.49), G12V (HR 0.32), and G12R, with the data validating the drug’s multi-RAS targeted mechanism and potentially removing the need for mutation-specific patient selection. However, the mature data also highlights a critical efficacy imbalance between mutation variants, with exceptional survival outcomes in KRAS G12V patients (HR 0.32) contrasting against a more modest benefit in the highly prevalent G12D cohort (HR 0.49). Additionally, the drug’s multi-RAS mechanism triggers high rates of wild-type on-target toxicities, including an 85.5% rash rate and 58.1% diarrhea rate, which pose significant compliance challenges for future frontline combination regimens, affecting the drug’s combinability potential. Ultimately, while the therapy near-doubles overall survival over historical chemotherapy, the median OS remained only 13.2 months, limited by downstream pathway bypass resistance. This leaves critical outstanding questions regarding optimal combination partners, cost-effectiveness, and efficacy in sicker patients (ECOG 2) or rare RAS genotypes.

Takeaways

In summary, these datasets presented in Madrid this year demonstrate steady, tangible progress in dismantling the historic “undruggable” barriers of the RAS oncogene, incrementally refining the playbook for precision oncology, with a shifting focus in clinical design away from standalone single-mutation monotherapies in R/R disease towards rational, combination-first frameworks in earlier lines of therapy aimed at mitigating clonal escape.

---

A New Wave of ADCs for Ovarian Cancer

Dennis Chang, PhD, SVP, Strategy Consulting

Ovarian cancer is entering a new era of antibody-drug conjugates (ADCs). Currently only one ADC is approved specifically for ovarian cancer: AbbVie’s Elahere (mirvetuximab sorvatansine) for platinum-resistant ovarian cancer (PROC) with high expression of folate receptor alpha (FRα). But during ESMO 2026, we saw a profusion of clinical data for new agents including a raft of new FRα-targeted ADCs as well as ADCs for new targets including CDH6 and B7-H4.

All of these new ADCs showed promising efficacy, apparently stronger than Elahere’s based on cross-trial comparisons. The new FRα ADCs also seem to be able to do without FRα biomarker restriction. For example, AZ’s Torvu-sam achieved an ORR of 57% and mPFS of 7.6 mo. for FRα-low PROC, better numbers than what Elahere had in FRα-high PROC (42%, 5.6 mo). The results for Torvu-sam in FRα-high PROC were even better (62%,  9.7 mo). Zymeworks’s ZW191 does not have PFS data yet, but showed similarly favorable ORR across FRα expression levels. Genmab’s Rina-S did not feature in presentations at the conference itself, but a positive phase 3 trial (RAINFOL-02) readout for PROC was announced by press release on Oct 23^rd during the meeting, again without the need for FRα selection. Moreover, all of these FRα ADCs can be used after prior Elahere use, given they use a different class of payload: topoisomerase 1 (TOP1) inhibitors instead of Elahere’s antitubulin DM4 payload.

Looking beyond FRα, Daiichi/Merck’s R-DXd, which targets CDH6, showed an ORR of 50% at its go-ahead dose (in combination with bevacizumab) in PROC. GSK’s Mo-rez targets B7-H4 and had ORR of 62% and mPFS of 11.1 mo at the highest dose level, with responses seen across B7-H4 levels.

All of these impressive efficacy findings point to a future where many strong ADCs will be available for ovarian cancer patients. However, there are two big caveats: toxicity and cross-resistance.

Grade 3+ TRAE rates ranged from 40% to over 70% across these studies. Major AEs included neutropenia and other hematologic AEs, as well as fatigue and GI AEs, in line with other ADCs using TOP1 payloads.

All of the new ADCs use a TOP1 inhibitor payload, which means their use will likely be limited by cross-resistance. They may be used before or after Elahere, but likely not after another TOP1-loaded ADC. Thus, these ADCs are now in a race to be used as early in the treatment algorithm as possible. At ESMO, Lilly’s FRα ADC Sofe-M reported an impressive ORR of 79% in platinum-sensitive ovarian cancer (PSOC), but many other ADCs are also in ongoing PSOC trials.

The pipeline for ovarian cancer also has a variety of mechanisms besides ADCs, including novel immunotherapies and novel agents targeting DNA damage response (DDR), as well as other approaches.

Takeaways

Ovarian cancer is the deadliest gynecologic cancer type, with limited treatment options especially in the platinum-resistant setting. However, data readouts at ESMO presage a much larger, more powerful, and more ADC-rich toolkit for treating ovarian cancer in the future, and these tools will be used in earlier lines of therapy.