This Week in Oncology — Sep 16, 2026

Welcome to This Week in Oncology. This week we're covering 10 notable papers spanning a wave of new antibody-based drugs in lung, breast and head and neck cancer, circulating tumour DNA and artificial intelligence as tools for risk stratification, and a cluster of care-delivery studies that ask how symptom monitoring, palliative care relationships and industry money shape what our patients actually receive. Let's dive in.

We start with novel antibody constructs, and the headline comes from the New England Journal of Medicine, where Zhao and colleagues report the interim analysis of the phase 3 TAISHAN-302 trial of tambotatug pelitecan, an antibody-drug conjugate directed at the immune checkpoint molecule B7-H3, in relapsed small-cell lung cancer [1]. Four hundred fifty-one patients whose disease had progressed after first-line platinum were randomised openly to the conjugate or to topotecan. Median overall survival rose by roughly four months, to just over 13 months, and the risk of death was cut by more than half. Progression-free survival was more than doubled, and confirmed responses occurred in roughly six in ten patients on the conjugate against about one in ten on topotecan. Just as striking, grade 3 or higher adverse events were less common with the new agent, about 55 percent versus nearly 80 percent. For a disease where topotecan has been a thin and poorly tolerated standard for decades, this is the most convincing second-line signal we have seen, though it remains an interim, open-label analysis from a largely Chinese trial network and cross-trial generalisability will need scrutiny.

Staying with antibody-drug conjugates, Annals of Oncology publishes a phase 1b study from Zhang and colleagues of iza-bren, a first-in-class bispecific conjugate targeting both EGFR and HER3, in 162 heavily pretreated patients with metastatic breast cancer, seven in ten of whom had already had three or more lines [2]. Confirmed response rates clustered in the mid-thirties to mid-forties percent across HER2-positive, hormone receptor-positive and triple-negative cohorts, with median progression-free survival of roughly six to seven months. The safety story is the caveat: treatment-related adverse events were essentially universal, grade 3 or higher toxicity affected just over three quarters of patients, mostly neutropenia and anaemia, and there were two treatment-related deaths. Notably, no interstitial lung disease was reported, which distinguishes it from several HER2- and TROP2-directed conjugates. Bispecific targeting is plausible here, but this is a safety-primary phase 1b, not an efficacy verdict. In Clinical Cancer Research, Kao and colleagues report cohort 2 of OrigAMI-4, testing subcutaneous amivantamab, the EGFR-MET bispecific antibody, plus pembrolizumab as first-line therapy in 39 patients with recurrent or metastatic head and neck squamous cell cancer and a PD-L1 combined positive score of at least one [3]. Confirmed responses occurred in about 59 percent of participants, including seven complete responses, with median progression-free survival of around eight months and median survival not yet reached. That response rate is well above what single-agent pembrolizumab delivers in this population, but with 39 patients and no randomisation, treat this as hypothesis-generating. Rounding out the targeted theme, a narrative review in the Journal of Clinical Oncology from Monaca and colleagues maps the increasingly crowded landscape of EGFR P-loop and alpha-C-helix compressing mutations and exon 20 insertions [4]. The practical message is that these are not classical EGFR mutations: for the compressing variants such as G719X and S768I, second-generation afatinib remains the most consistent option, compound mutations that include a classical alteration do better with osimertinib, and for exon 20 insertions amivantamab and mutant-selective inhibitors now dominate, with sunvozertinib having shown first-line superiority over platinum-pemetrexed. If your laboratory report says EGFR mutation and you reach reflexively for osimertinib, this review is a useful corrective.

The second theme is biomarker-guided risk stratification, and here the results are more sobering. In Nature Cancer, Marsoni and colleagues report PEGASUS, a single-arm phase 2 trial that used post-surgical circulating tumour DNA to either de-escalate or escalate adjuvant therapy in 135 patients with resected high-risk stage II or stage III microsatellite-stable colon cancer [5]. About a quarter of patients were ctDNA-positive at the landmark timepoint, and their three-year disease-free survival was 58 percent against 83 percent for ctDNA-negative patients, so the marker's prognostic power is not in doubt. But the primary endpoint was not met: the two-year relapse-free rate among ctDNA-negative patients who were de-escalated to single-agent capecitabine was 88 percent, below the pre-specified threshold of 92 percent. Against a matched historical control cohort, disease-free survival looked similar with substantially less neurotoxicity, and relapses clustered in stroma-rich consensus molecular subtype 4 biology. The honest reading is that ctDNA-guided de-escalation is operationally feasible but not yet validated, and that ctDNA-negative does not mean cured. Contrast that with the Journal of Clinical Oncology report from Shulman and colleagues, which prospectively embedded ctDNA analysis into Children's Oncology Group and LEOPARD studies in Ewing sarcoma, covering 140 patients with localised and 255 with metastatic disease [6]. Elevated pretreatment ctDNA burden roughly doubled the hazard of an adverse outcome in both localised and metastatic cohorts, and combining ctDNA with tumour size, site and TP53 status defined genuinely discrete groups, including a localised low-risk group that had no events at all and a metastatic subgroup with TP53 variants and persistent on-therapy ctDNA whose outcomes were dismal. That is the strongest argument yet for baseline liquid biopsy at Ewing diagnosis, and it should shape the design of the next generation of risk-stratified trials. The third strand here is artificial intelligence: in Nature Medicine, Prelaj and colleagues report I3LUNG, an international real-world study of nearly 2,400 patients with non-small cell lung cancer receiving immunotherapy [7]. Models built on clinical and blood data alone reached an area under the curve of up to 0.77 and significantly outperformed PD-L1, performance status, neutrophil-to-lymphocyte ratio, lactate dehydrogenase and the Lung Immune Prognostic Index. Importantly, both lung specialists and non-specialists improved their own predictions when using the explainable version of the tool, but performance dropped on external validation and adding CT imaging, digital pathology and genomics did not deliver reproducible incremental gain. Multimodal is not automatically better, and prospective validation is ongoing.

The final theme is how care is organised, and it may matter as much as any drug. In The Lancet Oncology, Hassing and colleagues report long-term post-hoc survival results from SYMPRO-Lung, a stepped-wedge cluster-randomised trial across 14 Dutch centres in which 446 patients with stage I to IV lung cancer either completed weekly online symptom assessments for a year or received usual care [8]. At a median follow-up of just over four years, median overall survival was five months longer in the monitoring group, and after adjustment the mortality reduction was of borderline statistical significance; progression-free survival showed no significant difference. This was a post-hoc analysis of a trial whose primary endpoint was quality of life, so it should not be oversold, but it extends the survival signal from patient-reported outcome monitoring beyond advanced disease and into routine practice. Complementing that, a multicentre qualitative study in JAMA Internal Medicine from Hua and colleagues used a positive-negative deviance design across six United States palliative care programmes, with 93 interviews and more than 400 hours of direct observation, to ask why hospice use varies so much [9]. The differentiator was not staffing or protocols but the relationship between palliative care and oncology teams: high-performing programmes showed mutual trust, physical proximity, real-time communication and a service-oriented reputation, while low-performing programmes showed mistrust and unfamiliarity, and it was at the high-performing sites that oncologists themselves introduced hospice earlier. Finally, and less comfortably, Mitchell and colleagues in the Journal of Clinical Oncology linked Medicare claims for more than 15,000 patients across 14 cancer scenarios to industry payments received by their medical oncologist in the year before diagnosis [10]. Payments tied to a guideline-preferred drug modestly increased the chance that the patient received the preferred treatment, while payments tied to a non-preferred drug modestly decreased it, and the effect grew with the dollar value of the payment. Industry money moved prescribing toward the promoted product regardless of whether that product was the better choice for that patient.

If you only have time for one paper this week, make it the TAISHAN-302 trial of tambotatug pelitecan in relapsed small-cell lung cancer [1]. A more than halving of the risk of death against topotecan, with less severe toxicity, is the kind of result that redefines second-line standard of care in a disease that has resisted progress for thirty years.

Here are the key takeaways from this week in Oncology. First, a B7-H3-directed antibody-drug conjugate substantially improved survival over topotecan in relapsed small-cell lung cancer, with a better toxicity profile, and should be on your radar as a new second-line standard. Second, bispecific antibody approaches are gaining traction across tumour types, with an EGFR-HER3 conjugate active but myelosuppressive in heavily pretreated breast cancer and amivantamab plus pembrolizumab producing high response rates in first-line head and neck cancer, both in early-phase settings. Third, not all EGFR mutations are osimertinib mutations; check whether your patient has a compressing variant or an exon 20 insertion before choosing a tyrosine kinase inhibitor. Fourth, circulating tumour DNA is now robustly prognostic in both colon cancer and Ewing sarcoma, but de-escalating adjuvant chemotherapy on the basis of a negative result missed its efficacy threshold in PEGASUS, so hold that strategy to trials. And fifth, structural factors matter: weekly patient-reported symptom monitoring was associated with longer survival in lung cancer, the quality of the working relationship between palliative care and oncology teams shapes hospice transitions, and industry payments measurably pull prescribing toward the promoted drug, preferred or not.

That's your roundup for This Week in Oncology. The full transcript and references are available on the episode page in your AudioScholar library. This is an AI-curated summary — for clinical decisions, always consult primary sources and current guidelines. See you next week.

If this weekly briefing is useful, follow the show in your podcast app so new episodes arrive automatically. And a quick rating on Apple Podcasts or Spotify helps other physicians discover the show.

This is an automated summary generated by artificial intelligence, which can make mistakes. Always review the original source materials.