Why immune aging,
why now

What ZEUS showed

In 2017, CANTOS showed that blocking interleukin-1β reduced recurrent cardiovascular events in patients with a prior heart attack and elevated hsCRP, without lowering cholesterol.¹ It was the first randomized evidence that an inflammatory pathway contributes causally to a common age-related disease. Two colchicine trials, COLCOT and LoDoCo2, later pointed the same way.²

The natural next step was to target interleukin-6, which sits downstream of IL-1β and drives production of CRP in the liver,³ using a long-acting antibody in patients selected for inflammation.

That trial was ZEUS. It enrolled 6,376 patients with atherosclerotic cardiovascular disease, chronic kidney disease and hsCRP of at least 2 mg/L, the same inflammation threshold used in CANTOS.^1,4 Ziltivekimab lowered free IL-6 and hsCRP as expected.⁴

On July 31, 2026, Novo Nordisk reported the primary result: hazard ratio 0.99, 95% confidence interval 0.88–1.11.⁴ Serious infections were more common on the drug, and all-cause mortality did not differ. Full results will be presented at the American Heart Association meeting on November 7, 2026.⁴

The biomarker moved as intended, but patients did not benefit. Explaining that gap is now the central problem for anti-inflammatory therapy in aging.

What ZEUS does and does not rule out

ZEUS does not on its own end the inflammation hypothesis, for two reasons that will be central to the first night.

The first is the population. ZEUS participants had a mean age of 69.5 years and a mean eGFR of 44.5, and 41% already had heart failure.⁴ In chronic kidney disease, heart failure, arrhythmia, sudden cardiac death and vascular calcification are major contributors to cardiovascular risk,⁵ and anti-inflammatory therapy may not prevent these events. A null result in this group does not show that the underlying biology is wrong.

The second is where in the pathway to intervene. CANTOS blocked IL-1β, one of the two cytokines released by the NLRP3 inflammasome.⁶ ZEUS blocked IL-6 further downstream, which lowers hsCRP most directly but leaves IL-1β and IL-18 signaling in place.³ Oral NLRP3 inhibitors, now in Phase 2 trials, act further upstream and block the release of both.^6,7 Human genetic studies support a causal role for IL-6 signaling in coronary disease,⁸ so whether ZEUS reflects the target, the drug or the population is unresolved.

In September 2026 Novo Nordisk stopped HERMES, its heart failure trial, after the data monitoring committee judged a different outcome from ZEUS unlikely.¹¹ ARTEMIS, which starts treatment soon after a heart attack, continues and is expected to report in the first half of 2027.¹¹ TECTONIC, a small Phase 2 trial at Mass General, tests a narrower version of the hypothesis. In an exploratory CANTOS analysis, the benefit of IL-1β blockade was concentrated in carriers of TET2 clonal hematopoiesis (hazard ratio 0.38), and TECTONIC tests that idea prospectively using an imaging measure of coronary inflammation.¹²

Immune reconstitution had a better year

While the anti-inflammatory programs struggled, work on rebuilding the aged immune system advanced. Suppressing inflammation and restoring immune function are different problems with different evidence, which is why they get separate nights.

In December 2025, Friedrich, Zhang and colleagues used mRNA-lipid nanoparticles to make the liver temporarily produce three factors that decline in the aging thymus: DLL1, FLT3L and IL-7. In aged mice this increased thymic output and naïve T cells and improved vaccine and antitumor responses. The effects faded within weeks of stopping treatment.¹³

Human data also suggest the adult thymus still matters. Adults who underwent thymectomy had higher five-year mortality than matched patients who had other cardiothoracic surgery (8.1% versus 2.8%), more cancer, lower thymic output and higher levels of inflammatory cytokines.¹⁴

Investment has followed. In July 2026 United Therapeutics acquired Thymmune, a Cambridge company developing iPSC-derived thymic cells with earlier support from ARPA-H, for $140M upfront and up to $160M in milestones.¹⁵ In August, Tolerance Bio licensed NeoImmuneTech's long-acting IL-7 for thymic indications, with up to $260M in milestone payments.¹⁶ In June, NewLimit raised $435M at a valuation of about $3.1B to develop epigenetic reprogramming therapies. The company has also reported, without peer review, transcription-factor combinations that restore killing activity in aged human T cells.¹⁷

In 2026, the field's clearest negative result and some of its most encouraging preclinical and commercial progress arrived within months of each other.

Measurement is still the bottleneck

ZEUS also raises the bar for biomarkers. If hsCRP can fall sharply without clinical benefit, then markers with less validation than hsCRP, which includes most aging biomarkers, need more scrutiny before they are used as trial endpoints.

hsCRP is the most established inflammatory marker in cardiovascular prevention and trials,¹⁸ and clonal hematopoiesis is a strong candidate to join it.¹⁹ Epigenetic clocks such as GrimAge and DunedinPACE, immune-age scores such as iAge and IMM-AGE, proteomic organ and cell-type clocks, and immune resilience scores all predict outcomes in cohort studies,²⁰ but they remain mostly research tools, and their clinical translation is limited.²¹

There is more to build on than a few years ago. In July 2026, Cipriano and colleagues proposed five criteria in Nature Medicine for judging immune-aging biomarkers as trial endpoints; measures of multidimensional immune function, inflammaging scores and functional assays met them best.²² The Allen Institute's Human Immune Health Atlas (2024) and a companion study of more than 16 million immune cells from donors aged 25 to 90 (2025) give the field reference data on how immune cells change with age.²³ What is still missing is agreement on which measurement would change a regulator's decision.

Why this format

Most meetings are built for presenting finished work. The questions above are still open, and they are easier to discuss frankly over dinner than from a podium.

Each night brings about 100 people together in Cambridge. It opens with a keynote on the night's question, followed by short lightning talks and a panel organized around five questions that recur every night: modality, stratification, endpoints, biomarkers and trial design. At dinner, each table takes one of those questions, and someone at each table takes notes.

Talks are recorded and published. Dinner is held under the Chatham House rule so that people can speak openly about unpublished and negative results.

Afterward we will publish a written playbook that summarizes the best current answer to each question, without attributing views to individuals.