Abstract / Summary
Background. In 2006, Imai and Takaoka set out a reference model contrasting monoclonal antibodies (mAbs) with small-molecule inhibitors, stating that mAbs “can only act on molecules that are expressed on the cell surface or secreted” whereas small molecules “can therefore be developed to target any molecules regardless of their cellular location.” Their Table 1 listed 14 approved agents as illustration. The approved landscape has since grown by roughly an order of magnitude and now includes antibody–drug conjugates, bispecific antibodies and allosteric kinase inhibitors. Individual departures from the model are described in the specialist literature, but the extent to which the model still holds across approved agents has not been quantified. Methods. Scoping review of regulatory records, reported per PRISMA-ScR, with the protocol fixed before extraction; the unit of analysis is an approved agent, not a study. Antibody-class agents were enumerated from Drugs@FDA (162 distinct molecular entities under a Biologics License Application, biosimilars and co-formulations collapsed) with one further agent recovered by hand from Federal Register records, and small-molecule agents from the BRIMR compilation of FDA-approved protein kinase inhibitors (93); agents with an oncology indication were retained, giving 79 small-molecule protein kinase inhibitors and 70 antibody-class agents, 149 in total. The two comparator sets are deliberately not symmetrical — the antibody side is the whole approved antibody class, the small-molecule side is the approved protein kinase inhibitors only — and this asymmetry limits what the comparison can show. Subcellular location of the primary target, binding–action location, Fc effector modification and year of first US approval were charted for both classes; molecular weight, kinase binding mode, ATP-site occupancy and route of administration were charted for the small-molecule class only. Results. Median molecular weight among the 79 small-molecule agents was 482.2 Da (interquartile range 440.5–530.4), and 76 of them (96.2 %) fell within 300–600 Da. No antibody-class agent has a cytoplasmic target as its charted primary target (0/70), whereas 34 small molecules do; the reciprocal cells are also empty, but because those targets are not kinases rather than because of size — an artefact of how the small-molecule comparator set was drawn, not evidence about size. On membrane receptors bearing both extracellular and intracellular domains the two classes coexist (45 vs 15). Departures from the model’s stated premises were: intracellular payload delivery, 16 of 70 antibody-class agents (22.9 %); and small molecules not occupying the ATP site, 6 of 79 (7.6 %) observed, with an inclusive upper bound of 11 of 79 (13.9 %) that additionally counts three agents whose assignment is disputed between sources and two for which ATP-site occupancy does not apply because their primary target is not a kinase. Two further features were charted that the model does not state as premises: 13 of 70 (18.6 %) are bispecific in format, and 11 of 70 (15.7 %) have reduced or inherently low Fc effector function — the latter measuring heterogeneity within the antibody class rather than contradicting the model, whose effector claim is one of exclusivity relative to small molecules. One departing agent — gemtuzumab ozogamicin, approved in 2000 — already existed when the reference model was published, and appears in the model’s own Table 1; the remainder accumulated from 2011 onwards, and the largest one-year increase anywhere in the series is among bispecific agents, from 5 to 9 between 2022 and 2023. Conclusions. In this charted set the two classes are fully separated where the target has no extracellularly accessible domain, which is what a size constraint predicts; where a target presents both an extracellular domain and an intracellular one, size excludes neither class. The reference model was not an unconditional description even when it was written — one agent in its own illustrative table already acted inside the cell — and what has changed since is the pace of departure rather than its existence. Two limits on the inference are stated rather than left implicit: a cross-tabulation of charted primary targets cannot by itself establish impossibility, and the small-molecule comparator set is restricted to protein kinase inhibitors, so the empty cells on that side carry no information about size.