Anti-Drug Antibodies (ADA)

Anti-drug antibodies are host antibodies that bind an administered peptide or protein. Regulators expect them to be measured in tiers: a sensitive screening assay, a confirmatory step showing that binding is displaced by excess drug, a titre obtained by serial dilution, then characterisation of positives for neutralising activity and for cross-reactivity with any endogenous counterpart. Every neutralising antibody is a binding antibody; most binding antibodies never neutralise anything.

Incidence varies enormously with the molecule. Twice-daily exenatide, a 39-residue exendin-4 sequence of lizard origin, provoked antibodies in a large fraction of treated patients, with a small high-titre subset showing attenuated glycaemic response. Human-sequence analogues such as liraglutide, semaglutide and dulaglutide report low single-digit incidences with little efficacy effect. Assay behaviour matters as much as biology: circulating drug masks antibody, so a long half-life analogue sampled anywhere but a deep trough under-reports.

What matters clinically is consequence, not positivity. The questions are whether the antibodies neutralise, whether the titre is high and persistent rather than transient, whether exposure or effect actually fell, and whether they cross-react with an endogenous hormone. The last is the serious one: antibody-mediated pure red cell aplasia in patients given recombinant erythropoietin happened because the antibodies also destroyed the native protein.

The reliable mistake is treating ADA incidence as comparable between products. Assay format, sensitivity, drug tolerance and sampling schedule differ, which is why prescribing information states that incidence cannot be compared across products. A related trap is reading a low reported incidence as a low-immunogenicity molecule when it may only reflect a less sensitive assay.