DNTH: Claseprubart CAPTIVATE response and randomized-withdrawal evidence

Active C1s and peripheral nerve injury

Claseprubart targets activated C1s in the classical complement pathway. The rationale in CIDP is to limit antibody-associated complement injury while leaving alternative and lectin pathway initiation intact. This is selective pathway inhibition, not preservation of every immune function: classical-pathway host defense and immune-complex handling may still be affected. [1] [2]

Clinical interpretation: CIDP combines potentially reversible inflammatory conduction failure with established axonal loss. The same degree of biochemical complement suppression need not produce the same functional recovery in both conditions. An independent diagnostic review panel improves case consistency, but neither a diagnosis nor a plausible target identifies which individuals have complement-dependent disability.

Exposure, dosing and reversibility

The antibody is engineered for prolonged exposure, with a company-reported half-life of roughly eight weeks. CAPTIVATE Part A uses an intravenous loading dose followed by 300 mg/2 mL subcutaneously every two weeks. The revised Part B retains this dose versus placebo, removing the originally planned 600 mg arm. [1] [4]

Analytical judgment: a long half-life can reduce dosing burden but complicates a withdrawal trial. Participants switched to placebo may retain pharmacologically active antibody for weeks, delaying separation of relapse curves. The interpretation therefore needs drug concentrations, complement activity and time since the last active dose; an early flat placebo curve is not necessarily spontaneous remission.

What the interim response actually established

The first 40 participants completing open-label Part A had a 75% response rate, disclosed in June 2026 and reiterated in August. Response requires at least a one-point improvement in adjusted INCAT from Part A baseline. This is an induction signal without a concurrent control; the early GO decision was a development decision, not a positive randomized Phase 3 result. [1] [2] [5]

The relevant denominator is participants completing that interim assessment, not every screened or ultimately enrolled patient. To estimate effectiveness in an unselected clinical population, screening exclusions, treatment starters, nonresponders, rescues and early withdrawals all need accounting. The first 40 may differ from later recruits in prior treatment and disease severity.

Randomized withdrawal and estimand

Only Part A responders enter the 52-week randomized, blinded, placebo-controlled Part B. The primary question is time to relapse, defined through worsening adjusted INCAT, among patients who initially improved on claseprubart. The program includes previously treated, treatment-naïve and standard-care-refractory CIDP; it does not require every participant to first worsen during a formal IVIg withdrawal. [3] [4]

Clinical interpretation: the randomized estimate will establish maintenance benefit conditional on induction response. It cannot directly provide the probability that a newly presenting patient will both respond and stay relapse-free. That broader probability combines induction success, retention and randomized maintenance, with uncertainty at each step rather than one headline responder percentage.

A withdrawal trial also changes how absolute benefit should be expressed. Calculating a number needed to treat from the open-label75% response would be invalid because there is no randomized induction counterfactual. A later maintenance estimate would apply only to responders at a specified follow-up, not everyone starting therapy. Relapse plots should include numbers at risk and distinguish rescue-triggered failure from simple censoring. Residual antibody can delay loss of pharmacodynamic protection after switching to placebo, so short follow-up could underestimate maintenance benefit. Conversely, excluding induction nonresponders can make the maintenance population appear more broadly responsive than the disease population actually is. Previously untreated, previously stable and refractory patients should retain their original strata. Combining these strata after observing different response patterns could conceal clinically meaningful heterogeneity rather than establish a single treatment effect for all CIDP.

Scales, rescue and readout timing

Adjusted INCAT is a disability scale with discrete steps. Grip strength and I-RODS can support functional consistency, but responder definitions are not interchangeable across scales. Relapse adjudication, confirmation visits and rescue initiation must be aligned; censoring people when rescue begins could conceal treatment failures if handled incorrectly. [2] [3]

As of the August 4 update, the year-end milestone is guidance on when Part B topline results will arrive. The June induction analysis is already available. No randomized relapse hazard ratio, confidence interval or arm-level relapse count was reported in that update; treating the existing 75% induction result as a prospective year-end efficacy readout would duplicate the event.

Safety and development execution

The interim update reported no related serious infections, drug-induced lupus symptoms, related serious adverse events or safety discontinuations. This is reassuring within limited exposure, not a precise estimate of rare-event risk. ANA laboratory findings and symptomatic autoimmune disease must be separated; laboratory positivity alone is not drug-induced lupus. [1] [2]

The company reported about $1.2 billion in cash at June 30 and projected runway into 2030. This supports ability to execute multiple long trials but provides no clinical validation. Relevant oversight includes implementation of the revised protocol, preservation of blinding, consistent rescue access and complete safety surveillance during the extended exposure created by a long-lived antibody.

Sources / References

1. Dianthus June 2026 corporate presentation filed with SEC
2. Dianthus Q2 2026 financial and clinical update, August 4
3. CAPTIVATE registry NCT06858579
4. Dianthus March 2026 CAPTIVATE design update, SEC 8-K
5. Dianthus full-year 2025 results and CAPTIVATE GO decision