Fibroin is the reference point we all know, a crystalline β-sheet with two decades of literature behind it and a handful of devices already reaching the clinic. Yet a slight shift in intended use is enough for the same protein, or the sericin that accompanies it in the cocoon, to change legal regime and fall under radically different authorities, procedures, and evidentiary thresholds. Anyone developing or validating a silk material in Europe does not face a single regulatory framework but a bundle of parallel regimes competing for the same molecule, and the first task is not to demonstrate conformity but to establish which conformity applies. This is the part of the path that is systematically underestimated, and where silk, by its nature as a versatile and degradable animal protein, is almost a textbook borderline case.
Legal qualification comes before conformity
The medical device regulation and the novel food regulation are not chosen, they are imposed as a function of the declared intended use. A fibroin hydrolysate proposed as a supplement, a fibroin matrix for tissue regeneration, and a sericin film in a cream are, from a chemical standpoint, variations of the same protein family, but the finality the manufacturer attributes to them routes them toward Regulation (EU) 2017/745, toward Regulation (EU) 2015/2283, or toward Regulation (EC) 1223/2009 on cosmetic products. Qualification is not a formal step downstream of development, it is the decision that determines the entire architecture of evidence, the timelines, the costs, and the institutional counterpart. Getting it wrong, or deferring it, means building a dossier for the wrong regime.
The difficulty with silk materials arises from the fact that the same molecule can justify more than one credible intended use. Fibroin is structural and supports tissues, so it lends itself to the device. It is degradable and its peptides have documented biological activity, so it is tempted by the food route. It is film-forming and hydrating on the skin, so it has an established home in cosmetics. None of these readings is contrived, and precisely for this reason the borderline zone is wide. The Commission has a dedicated tool for these cases, the manual on borderline and classification, and a coordination procedure among competent authorities, a sign that the legislator anticipated the contest but left it to be resolved case by case.
The medical device route under regulation 2017/745
When the intended use is therapeutic or one of structural support, the silk material enters the scope of Regulation (EU) 2017/745, applied since 26 May 2021 in replacement of Directives 93/42 and 90/385. The regulation broadened the definition of a device by including the modification of a physiological or pathological state among medical purposes, and explicitly introduced so-called devices made of substances, governed by specific general safety and performance requirements and by rule 21 of Annex VIII. An absorbable fibroin, designed to be degraded and reabsorbed at the implant site, may find itself assessed precisely as a substance-based device, with the consequence of a close confrontation with the medicinal product framework whenever the principal action risks appearing pharmacological rather than physical and structural.
A rule that at first glance seems tailor-made to capture silk is rule 18, which classifies as class III all devices manufactured utilising tissues or cells of human or animal origin, or their derivatives, rendered non-viable, save for those intended to come into contact with intact skin only. Read against the bare text, fibroin and sericin, proteins recovered from the cocoon of Bombyx mori, look like animal-derived materials that the rule would sweep into class III without effort. This is precisely the reading that the guidance corrects. In its note to rule 18, MDCG 2021-24 confines the notion of derivative to substances processed from animal tissues, and sets expressly apart the products made by animals, among which it lists milk, silk, beeswax, honey, propolis, royal jelly, hair and lanolin. Silk is therefore excluded by name. A material spun by the silkworm is not a tissue of the animal worked into a derivative, in the way collagen or gelatine are extracted from skin and bone; it is a secretion, and the regulatory logic files it alongside milk and honey rather than alongside porcine collagen. The consequence is decisive: rule 18 does not apply to fibroin and sericin, and a silk device is not class III by virtue of its animal origin.
It is worth understanding why the line falls exactly here, because the rationale confirms the outcome rather than fighting it. The European framework on animal tissues in devices was built around the risk of transmissible agents, and in particular around spongiform encephalopathies. Regulation (EU) 722/2012, which the notified body applies to devices manufactured with animal tissues rendered non-viable, has in mind the TSE-risk species, ruminants and a few other mammals, not a lepidopteran bred for millennia for its fibre. Letter of the guidance and purpose of the norm point the same way, and an insect secretion sits outside the perimeter that rule 18 was designed to police. What remains, therefore, is not an argumentative exercise to escape class III, but the ordinary task of classifying the silk device on its own merits, by intended use, duration, invasiveness and any biological or absorbable action, under the rules that actually govern it. An absorbable fibroin scaffold may still reach class III, but through rule 8, on account of its biological effect or its resorption, not through a contested reading of animal origin; a non-absorbable structural implant would sit in class IIb under the same rule. The class may be high, but its justification runs through the ordinary logic of risk, and the apparatus of animal-tissue documentation does not attach.
The material itself, however, still had to be made acceptable, and industrial history shows how. The strategy that made fibroin acceptable as an implant consisted in removing its sericin, considered the immunogenic component of raw silk, and in purifying the fibroin filament to a high degree of definition. This is the logic behind the purified fibroin scaffold developed from the studies of David Kaplan's laboratory at Tufts University, which then reached the market as a device for soft tissue support, with CE marking in addition to its U.S. clearance, and passed through several industrial owners. On the European side, an electrospun fibroin nerve conduit is another example of a silk device actually placed on the market. These precedents demonstrate that the device route is viable, and that it passes through a stringent definition of the material, control of the extraction process, and a documentation burden that grows as the class rises, though that burden is driven by the intended use of the finished device rather than by the animal provenance of its raw material.
It is worth recalling the asymmetry between the two sides of the Atlantic, because it conditions the way these materials arrive in Europe. In the United States, almost all fibroin devices have gone through the simplified route of substantial equivalence, which allows reliance on an already authorised predicate. The European regulation offers no analogous shortcut: a silk device must be classified and documented on its own terms, and where its intended use places it in a high class, for instance an absorbable implant that falls into class III under rule 8 rather than under rule 18, clinical evidence and post-market surveillance follow their own demanding requirements. A silk material validated elsewhere is not automatically ready for the European framework, and the most laborious part of the realignment is often precisely the justification of the classification, though that justification now runs through the ordinary rules on intended use rather than through a contested reading of animal origin.
