Integrative Biomedical Research

1. Introduction

Rebuilding dental tissues after disease or injury is rarely simple. Alveolar bone, periodontal ligament, cementum and the dentin–pulp complex are interdependent structures, each organised hierarchically from organic fibres and mineral crystals, and restoring them together remains one of the more stubborn problems in oral and maxillofacial surgery (Elkhenany et al., 2024; Prathap et al., 2025). Advanced periodontitis, large periapical lesions, trauma, congenital defects and tumour resection can all remove mineralised and fibrovascular tissue on a scale the body cannot replace (Elkhenany et al., 2024; Prathap et al., 2025). Bone and mineralised dental tissues do heal on their own, but only up to a point; critical-sized defects exceed that capacity (Elkhenany et al., 2024; Prathap et al., 2025). Left without support, such defects tend to fill with collapsing soft tissue and fibrous scar, and the alveolar ridge resorbs progressively, often ending in tooth loss or implant failure (Elkhenany et al., 2024; Prathap et al., 2025). For this reason, surgeons generally turn to osteoconductive and, ideally, osteoinductive biomaterials that guide anatomical repair and integrate functionally with host tissue (Elkhenany et al., 2024; Prathap et al., 2025).

Autologous bone has long been regarded as the gold standard because it is osteogenic, osteoinductive and osteoconductive at once (Diogo et al., 2018; Elkhenany et al., 2024). Its drawbacks are equally familiar: a second surgical site, bleeding, postoperative pain, longer recovery and a limited volume of available graft (Diogo et al., 2018; Elkhenany et al., 2024). Allografts and xenografts, mostly of bovine and porcine origin, were adopted partly to avoid these problems (Diogo et al., 2018; Prathap et al., 2025; Radulescu et al., 2022). They bring their own liabilities, however, including a residual risk of pathogen transmission (bovine spongiform encephalopathy and other transmissible spongiform encephalopathies, as well as porcine viral disease), structural variation between batches, residual immunogenicity, and religious or cultural restrictions that make them unacceptable to some patients (Binlateh et al., 2022; Coppola et al., 2020; Diogo et al., 2018; Saeed et al., 2025). Synthetic bioceramics and inert metallic or polymeric substitutes avoid the biological risks but frequently degrade at the wrong rate, show little osteoinductivity, provoke foreign-body giant-cell reactions and offer cells few points of attachment (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025). Taken together, these shortcomings explain the growing interest in alternative materials that are biocompatible, sustainable and free of mammalian disease risk.

A possible source of such materials sits, somewhat unexpectedly, in the waste stream of the fishing industry. Fisheries and aquaculture processing generate very large volumes of solid biological waste (Coppola et al., 2021; Elkhenany et al., 2024). Roughly 20 to 30 million tonnes of by-products, amounting to 50%–70% of harvested biomass and consisting largely of bones, skins, scales, fins and heads, are discarded each year (Binlateh et al., 2022; Coppola et al., 2021; Sharma et al., 2024). Landfilling or dumping this material at sea causes microbial putrefaction, odour, eutrophication and emissions of methane and carbon dioxide (Elkhenany et al., 2024; Sharma et al., 2024). Within the frameworks of the circular bioeconomy and the European Commission’s “Blue Growth” strategy, converting these side-streams into high-value biomaterials has come to be seen as a practical “waste-to-wealth” opportunity (Khrunyk et al., 2020; Maschmeyer et al., 2020; Núñez-Tapia et al., 2026).

Two of the most promising products of this valorisation are marine collagen (mCol) and biogenic nano-hydroxyapatite (mHAp), and their combination is particularly attractive for dental tissue engineering (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025). The rationale is biomimetic. Bone, dentin and cementum are themselves organic–inorganic nanocomposites, built from type I collagen triple helices interleaved with non-stoichiometric, carbonated hydroxyapatite nanocrystals (Elkhenany et al., 2024; Mudhafar et al., 2021; Radulescu et al., 2022). Collagen extracted from the skins, bones and scales of species such as Nile tilapia, salmon, cod and catfish closely resembles mammalian type I collagen in amino acid composition, triple-helical structure and extracellular matrix (ECM) function (Binlateh et al., 2022; Coppola et al., 2020; Saeed et al., 2025). It binds cells through RGD-type motifs, supports adhesion and proliferation of osteoblasts and fibroblasts, promotes haemostasis, has low antigenicity and is more water-soluble than collagen from land animals (Binlateh et al., 2022; Saeed et al., 2025; Sharma et al., 2024).

Collagen alone, though, is not enough. Pure marine collagen hydrogels and sponges degrade quickly in vivo and are mechanically soft, which limits their use in load-bearing dental and maxillofacial sites (Binlateh et al., 2022; Saeed et al., 2025; Zhang et al., 2025). Biogenic nano-hydroxyapatite [Ca₁₀(PO₄)₆(OH)₂], recovered from fish bones and scales by controlled calcination, alkaline hydrolysis or subcritical water extraction, can serve both as a mechanical reinforcement and as an osteoconductive nucleation surface (Kongthong et al., 2024; Saeed et al., 2025; Zainol et al., 2024). Unlike reagent-grade synthetic hydroxyapatite, fish-derived mHAp retains naturally incorporated trace ions, including strontium (Sr²⁺), magnesium (Mg²⁺), carbonate (CO₃²⁻), silicon (Si⁴⁺) and fluoride (F⁻) (Diogo et al., 2018; Radulescu et al., 2022; Zainol et al., 2024). These ions are biologically active. Sr²⁺ and Mg²⁺ stimulate osteoblastogenesis through alkaline phosphatase (ALP) activation and restrain osteoclastic resorption through the RANK/RANKL/osteoprotegerin axis (Diogo et al., 2018; Saeed et al., 2025; Zainol et al., 2024), whereas fluoride and carbonate substitution improve acid resistance, apatite reactivity and physiological remodelling in the oral environment (Diogo et al., 2018; Habib et al., 2025).

When the two components are combined in three-dimensional scaffolds, membranes or injectable hydrogels, they create a microenvironment that resembles native tissue more closely than either does alone (Elkhenany et al., 2024; Parisi et al., 2020; Radulescu et al., 2022). The collagen network provides interconnected porosity, spatial organisation and anchorage for human periodontal ligament cells (hPDLCs) and dental pulp stem cells (DPSCs) (Elkhenany et al., 2024; Sharma et al., 2024), while dispersed nano-apatite stiffens the construct, slows its degradation and raises local calcium and phosphate concentrations (Diogo et al., 2018; Elkhenany et al., 2024). Recent in vitro and in vivo studies report that fish-derived collagen–hydroxyapatite nanocomposites upregulate osteogenic markers (RUNX2, ALP, COL1A1 and osteocalcin), support vascularisation, help preserve bone in critical-sized calvarial and mandibular defects and show good biocompatibility without cytotoxicity (Elkhenany et al., 2024; Prathap et al., 2025; Saeed et al., 2025).

Even so, the path from laboratory to clinic is far from clear (Habib et al., 2025; Saeed et al., 2025; Zhang et al., 2025). Seafood side-streams are inherently variable; mineral and amino acid profiles differ between species; collagen from cold-water fish is thermally fragile; some crosslinkers are cytotoxic; and there is no agreed set of green processing protocols (Binlateh et al., 2022; Habib et al., 2025; Zhang et al., 2025). What seems to be missing is a single synthesis that connects sustainable extraction, microstructural characteristics, fabrication technologies such as 3D bioprinting and freeze-casting, and the specific therapeutic needs of dentistry (Elkhenany et al., 2024; Habib et al., 2025; Zhang et al., 2025).

This review was written to address that gap, and it has five aims. First, it compares mechanical, enzymatic, acid-solubilisation, hydrothermal and calcination routes for recovering type I collagen and biogenic nano-hydroxyapatite from fish skins, scales and bones, with attention to yield, purity, structural integrity and scalability (Binlateh et al., 2022; Habib et al., 2025; Sharma et al., 2024). Second, it examines phase composition, trace-element substitution, Ca/P stoichiometry, mechanical behaviour, degradation and thermal stability of marine-derived nanocomposites relative to terrestrial and synthetic alternatives (Binlateh et al., 2022; Diogo et al., 2018; Zainol et al., 2024). Third, it appraises crosslinking (EDC/NHS, enzymatic and dehydrothermal), freeze-drying, electrospinning and additive manufacturing as means of controlling porosity, swelling and the collagen–apatite interface (Diogo et al., 2018; Habib et al., 2025; Zhang et al., 2025). Fourth, it synthesises preclinical evidence on cytocompatibility, macrophage polarisation, osteoblast adhesion, angiogenesis and the osteogenic or odontogenic differentiation of DPSCs and hPDLCs in periodontal repair, guided bone regeneration (GBR), socket preservation and pulp capping (Elkhenany et al., 2024; Prathap et al., 2025; Saeed et al., 2025). Finally, it considers the translational, regulatory and economic barriers, including standardisation, good manufacturing practice (GMP), biosafety and sterilisation, that stand between marine bio-waste and approved dental products (Habib et al., 2025; Saeed et al., 2025).

2. Fish Waste-Derived Collagen and Nano-Hydroxyapatite Composites for Dental and Craniofacial Bone Regeneration

2.1 Biomimetic Rationale and the Circular Bioeconomy in Regenerative Dentistry

Craniofacial and dental tissues share a common design principle: organic fibrous networks tightly interwoven with mineral phases, arranged across several length scales (Elkhenany et al., 2024; Prathap et al., 2025). When periodontitis, periapical osteolysis, dentoalveolar trauma or surgical resection removes enough of this architecture, the defect exceeds the healing capacity of the host (Elkhenany et al., 2024; Prathap et al., 2025). Unassisted healing is then typically derailed by fibrous collapse and epithelial down-growth, leading to chronic alveolar atrophy and compromised dentition (Elkhenany et al., 2024; Prathap et al., 2025). Osteoconductive, biomimetic scaffolds have therefore become central to reconstructive dentistry (Diogo et al., 2018; Radulescu et al., 2022).

The clinical benchmark, autogenous bone, combines osteogenicity, osteoinductivity and osteoconductivity, but donor-site morbidity, limited supply, bleeding, pain and unpredictable resorption restrict its use (Diogo et al., 2018; Elkhenany et al., 2024). Bovine and porcine xenografts ease the supply problem but carry documented risks of

Table 1. Extraction parameters, yields, and structural characteristics of type I collagen recovered from fish processing side-streams. The table compares acid-soluble (ASC), pepsin-soluble (PSC) and CO₂-acidified water extraction of collagen from the skins, scales and bones of seven teleost and elasmobranch species. Pretreatment with dilute NaOH removes non-collagenous proteins and pigments, and EDTA or HCl demineralisation is required for mineralised scales and bones. Pepsin digestion of telopeptides generally increases yield and lowers immunogenicity. Denaturation temperature (Td) correlates with imino acid content, so warm-water species are closer to body temperature (37 °C) than cold-water species. Yields are as reported by the original authors (dry- or wet-weight basis indicated where available).

Table 2. Extraction conditions, physicochemical properties, and trace-element profiles of biogenic nano-hydroxyapatite from fish processing waste. The table summarises biogenic hydroxyapatite (mHAp) recovered from the bones, scales and teeth of eight fish species by thermal calcination (500–1100 °C) or alkaline hydrolysis (70–250 °C). Unlike stoichiometric synthetic HAp (Ca/P = 1.67), fish-derived apatite is non-stoichiometric (Ca/P ≈ 1.30–1.74) and retains Mg²⁺, Sr²⁺, Na⁺, F⁻ and B-type CO₃²⁻ substitutions. Higher calcination temperatures increase crystallinity and phase purity but remove carbonate and may generate β-TCP. Elemental data were obtained by ICP-OES, EDX or FTIR as reported in each source.

zoonotic transmission, including bovine spongiform encephalopathy and swine viruses (Coppola et al., 2020; Diogo et al., 2018; Saeed et al., 2025), and religious, cultural and dietary proscriptions limit their acceptance in many patient groups (Binlateh et al., 2022; Diogo et al., 2018). Purely synthetic ceramics and inert polymers, by contrast, often lack physiological recognition motifs, dissolve at mismatched rates and may trigger persistent foreign-body reactions (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025).

Fish processing waste offers a different starting point. Industrial offal, composed mainly of skins, scales, skeletons, heads and viscera, accounts for 50%–70% of harvested biomass, or about 20–30 million tonnes per year (Binlateh et al., 2022; Coppola et al., 2021; Sharma et al., 2024). Most is landfilled or dumped offshore, where it drives eutrophication, anaerobic putrefaction and greenhouse gas release (Elkhenany et al., 2024; Sharma et al., 2024). Valorising this biomass into regenerative products fits naturally within the circular bioeconomy (Khrunyk et al., 2020; Maschmeyer et al., 2020; Núñez-Tapia et al., 2026), and co-extracting mCol and mHAp in particular provides an abundant, pathogen-safe basis for nanocomposites that echo the composition and ultrastructure of alveolar bone, cementum and dentin (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025). The overall valorisation pathway, from side-stream to dental application, is summarised in Figure 1.

2.2 Marine Collagen: Sources and Extraction Methodologies

Collagen makes up roughly 30% of total vertebrate protein and forms the structural backbone of the ECM in connective tissue, bone and teeth (Elkhenany et al., 2024; Saeed et al., 2025; Sharma et al., 2024). Fish skins, scales and skeletons, from species such as tilapia (Oreochromis niloticus), rohu (Labeo rohita), carp (Cyprinus carpio), catfish (Pangasianodon hypophthalmus), salmon and blue shark (Prionace glauca), are concentrated sources of predominantly type I collagen (Binlateh et al., 2022; Diogo et al., 2018; Jafari et al., 2020; Saeed et al., 2025; Sharma et al., 2024).

Isolation begins with pretreatment. Tissues are homogenised, washed in dilute sodium hydroxide (0.1–0.2 M NaOH) to strip non-collagenous proteins and pigments, defatted with butyl alcohol, isopropanol or chloroform/acetone mixtures and, for scales and bones, demineralised with ethylenediaminetetraacetic acid (EDTA) or dilute hydrochloric acid (HCl) (Elkhenany et al., 2024; Kongthong et al., 2024; Sharma et al., 2024). Extraction then follows one of two classical routes: acid-soluble collagen (ASC) or pepsin-soluble collagen (PSC) (Binlateh et al., 2022; Jafari et al., 2020; Saeed et al., 2025; Sharma et al., 2024). The specific parameters reported for individual species are compiled in Table 1, and the generic workflow is illustrated in Figure 2A.

In ASC extraction, tissue is held in dilute organic acid, typically 0.5 M acetic acid at pH 2.5–3.0 and 4 °C, for 48–72 h (Binlateh et al., 2022; Sharma et al., 2024). The acid cleaves labile intermolecular aldimine crosslinks and induces electrostatic repulsion between positively charged tropocollagen molecules, which brings the protein into solution (Binlateh et al., 2022; Sharma et al., 2024). The method is simple and inexpensive, but yields from mature tissue are limited by dense, non-reducible trivalent crosslinks (Sharma et al., 2024). PSC extraction addresses this by adding porcine or marine pepsin to the acetic acid (Binlateh et al., 2022; Saeed et al., 2025; Sharma et al., 2024). Pepsin removes the non-helical telopeptides while leaving the central triple helix intact (Binlateh et al., 2022; Sharma et al., 2024), producing atelocollagen that dissolves more readily, is purer and is less immunogenic because the principal antigenic sites sit in the telopeptide regions (Binlateh et al., 2022; Saeed et al., 2025). Greener variants are also emerging; collagen has been recovered from Atlantic cod skin using CO₂-acidified pressurised water, avoiding corrosive organic acids altogether (Sousa et al., 2020) (Table 1).

2.3 Biochemical Structure and Functional Biocompatibility of Marine Collagen

Marine type I collagen is a heterotrimer of two α1 chains and one α2 chain, wound into a right-handed triple helix stabilised by repeating Gly–X–Y triplets in which X and Y are most often proline and hydroxyproline (Binlateh et al., 2022; Saeed et al., 2025). Accessible Arg–Gly–Asp (RGD) integrin-binding sequences support attachment, spreading and signalling in osteoblasts, fibroblasts and mesenchymal stem cells (Elkhenany et al., 2024; Saeed et al., 2025). Marine collagen also promotes platelet adhesion and aggregation at bleeding surgical margins and binds water readily, swelling in a way that helps nutrient diffusion and cell migration in the regenerative niche (Elkhenany et al., 2024; Prathap et al., 2025; Sharma et al., 2024). Characterisation of collagens from

Figure 1. Valorisation pathway from fish processing waste to marine collagen–hydroxyapatite nanocomposites for dental regeneration. Fish side-streams (skins, scales, bones, fins and heads; ≈ 20–30 million tonnes per year) are processed along two parallel routes. The organic route (blue) recovers type I collagen by NaOH pretreatment, defatting, demineralisation and acid- or pepsin-soluble extraction. The inorganic route (orange) recovers ion-substituted biogenic nano-hydroxyapatite by calcination or alkaline hydrolysis. Both components are assembled and crosslinked into biomimetic nanocomposites (green) for periodontal, alveolar, pulpal and craniofacial applications (violet). Compiled from Binlateh et al. (2022), Coppola et al. (2021), Elkhenany et al. (2024), Habib et al. (2025) and Sharma et al. (2024).

Figure 2. Extraction workflow and thermal stability of marine collagen. (A) Generic workflow for isolating type I collagen from fish tissue: sequential alkaline pretreatment, defatting and demineralisation, followed by acid-soluble (ASC) or pepsin-soluble (PSC) extraction and purification; PSC removes antigenic telopeptides and generally raises yield. (B) Denaturation temperatures (Td) of collagens from mammalian and fish sources; bars show reported ranges and dots single values. The dashed red line marks body temperature (37 °C), which all marine collagens fall below, with cold-water species the most vulnerable. Data from Abbas et al. (2022), Ahmed et al. (2019), Diogo et al. (2018), Saeed et al. (2025), Song et al. (2019) and Sousa et al. (2020).

tilapia, catfish, carp and tuna by SDS-PAGE, circular dichroism (CD) and Fourier transform infrared spectroscopy (FTIR) has consistently confirmed the native (α1)₂α2 heterotrimer and intact helical structure (Abbas et al., 2022; Ahmed et al., 2019; Lim et al., 2023; Pal et al., 2015; Song et al., 2019) (Table 1).

2.4 Thermal Stability and Mechanical Vulnerabilities

Biological affinity aside, pristine marine collagen has physical weaknesses that limit its use alone in load-bearing reconstruction (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025). The most important is a lower denaturation temperature (Td) than mammalian collagen (Binlateh et al., 2022; Saeed et al., 2025). Bovine and porcine collagens typically denature at 39–41 °C, whereas marine collagens denature at 15–28 °C in cold-water species such as cod and salmon and at 32–36.5 °C in warm-water species such as tilapia, gourami and catfish (Elkhenany et al., 2024; Saeed et al., 2025; Sharma et al., 2024) (Figure 2B).

The difference tracks closely with imino acid (proline plus hydroxyproline) content, because pyrrolidine rings and hydroxyproline-mediated hydrogen bonding are the main forces holding the triple helix together (Binlateh et al., 2022; Saeed et al., 2025). At 37 °C, uncrosslinked marine collagen can relax and collapse prematurely into random-coil gelatin (Binlateh et al., 2022; Elkhenany et al., 2024). Native marine collagen hydrogels and sponges are also highly susceptible to matrix metalloproteinases (MMPs) and collagenases, degrading within days, and their compressive modulus is typically below 100 kPa (Elkhenany et al., 2024; Radulescu et al., 2022). Mineral hybridisation and exogenous crosslinking are therefore needed to provide the mechanical stability required for space maintenance in dental defects (Diogo et al., 2018; Elkhenany et al., 2024; Radulescu et al., 2022).

2.5 Biogenic Nano-Hydroxyapatite: Extraction by Calcination and Alkaline Hydrolysis

Bone and tooth minerals are not stoichiometric ceramics; they are poorly crystalline, carbonated apatites with multiple ionic substitutions (Elkhenany et al., 2024; Habib et al., 2025; Radulescu et al., 2022). Fish bones and scales contain 45%–70% mineral by dry weight, which makes them a rich precursor for biogenic nano-hydroxyapatite (Habib et al., 2025; Suresh et al., 2026; Zainol et al., 2024). Extraction conditions and resulting properties for representative species are listed in Table 2, and the temperature-dependence of the main routes is summarised in Figure 3A.

Thermal calcination is the most widely used route (Dermawan et al., 2022; Habib et al., 2025; Zhang et al., 2025). Cleaned, defatted bones or scales are heated in a muffle furnace at 600–1100 °C for 2–9 h (Dermawan et al., 2022; Habib et al., 2025; Núñez-Tapia et al., 2026). The heat burns off the collagen and lipid fraction, eliminates biological pathogens and leaves white, phase-pure apatite powder (Dermawan et al., 2022; Habib et al., 2025). Temperature largely determines crystallite size, specific surface area and phase composition (Dermawan et al., 2022; Radulescu et al., 2022). Treatment at 600–800 °C produces nanoparticles of 20–80 nm that retain B-type carbonate, whereas temperatures of 900–1000 °C or higher drive crystal growth, complete decarbonation, agglomeration and phase purities above 95%–99% (Dermawan et al., 2022; Habib et al., 2025; Jaffri et al., 2022). In some teleost bones, calcination above 850 °C partially converts non-stoichiometric apatite into β-tricalcium phosphate [β-TCP; Ca₃(PO₄)₂], yielding a biphasic calcium phosphate (BCP) with greater bioreactivity and tunable resorption (Piccirillo et al., 2013; Radulescu et al., 2022).

Alkaline hydrolysis, sometimes described as subcritical wet-chemical extraction, offers a lower-temperature alternative (Gnanasekaran et al., 2024; Habib et al., 2025; Kongthong et al., 2024). The raw material is digested in 1–2 M NaOH at 70–250 °C for 4–8 h under ambient or hydrothermal pressure (Gnanasekaran et al., 2024; Kongthong et al., 2024; Suresh et al., 2026). The alkali hydrolyses the protein matrix without the sintering and grain growth seen during dry calcination (Gnanasekaran et al., 2024; Suresh et al., 2026). The resulting powders consist of fine rod-like or spherical particles of 20–80 nm, with high Brunauer–Emmett–Teller (BET) surface areas of 80–120 m²/g and a lower crystallinity that more closely resembles biological apatite (Gnanasekaran et al., 2024; Kongthong et al., 2024; Suresh et al., 2026).

2.6 Stoichiometry and Multi-Elemental Ionic Substitution

Reagent-grade synthetic hydroxyapatite (sHAp) has a fixed Ca/P molar ratio of 1.67, few lattice defects and slow resorption (Elkhenany et al., 2024; Habib et al., 2025; Suresh et al., 2026). Fish-derived mHAp, in contrast, is intrinsically non-stoichiometric, with Ca/P ratios that range from about 1.45 to 1.74 depending on species, age, diet and processing temperature (Dermawan et al., 2022; Habib et al., 2025; Suresh et al., 2026; Zainol et al., 2024) (Figure 3B; Table 2).

Its principal advantage lies in its natural content of trace substituents, including Mg²⁺, Sr²⁺, Na⁺, K⁺, Zn²⁺, F⁻ and CO₃²⁻ (Diogo et al., 2018; Habib et al., 2025; Suresh et al., 2026; Zainol et al., 2024). These ions occupy the Ca(I) and Ca(II) sites or replace phosphate and hydroxyl groups within the apatite lattice (Habib et al., 2025; Suresh et al., 2026), and each has a recognisable biological role (Figure 3C). Strontium and magnesium stimulate osteoblast proliferation and ALP activity while downregulating osteoclastic resorption through suppression of receptor activator of nuclear factor-κB ligand (RANKL) signalling (Diogo et al., 2018; Suresh et al., 2026; Zainol et al., 2024). Carbonate substitution strains the lattice and reduces crystallite size, increasing surface reactivity and accelerating osteoclast-mediated resorption and remodelling (Habib et al., 2025; Radulescu et al., 2022). Fluoride, which is concentrated in shark and fish skeletal tissue, strengthens the lattice against acid dissolution in the oral cavity, forming fluorhydroxyapatite solid solutions that support odontogenic and osteogenic differentiation while resisting bacterial demineralisation (Diogo et al., 2018; Habib et al., 2025). Compared with synthetic HAp, catla-scale apatite has been reported to contain about 3.5 times more Mg²⁺ and 10 times more Sr²⁺ (Deb et al., 2019; Radulescu et al., 2022), and salmon and trout bone apatites rich in Mg²⁺ and CO₃²⁻ stimulated ALP activity in osteoblasts (Shi et al., 2018; Zainol et al., 2024) (Table 2).

2.7 Fabrication Technologies for Biomimetic Nanocomposite Assembly

In natural hard tissue, plate- or rod-like apatite nanocrystals nucleate within the periodic gap zones of self-assembled collagen fibrils (Elkhenany et al., 2024; Mudhafar et al., 2021; Radulescu et al., 2022). Several processing approaches have been used to approximate this organisation with marine components (Table 3).

Freeze-drying is the most common. Acid-solubilised marine collagen is blended with dispersed mHAp nanoparticles, frozen under controlled conditions and lyophilised to form highly porous three-dimensional sponges (Elkhenany et al., 2024; Mudhafar et al., 2021; Prathap et al., 2025). Adjusting the freezing kinetics produces interconnected pores of 100–350 µm, a range that permits cell infiltration, nutrient diffusion and capillary ingrowth (Diogo et al., 2018; Elkhenany et al., 2024; Mudhafar et al., 2021).

Additive manufacturing offers more control. Extrusion bioprinting and robocasting use viscous inks of fish collagen, biogenic mHAp and natural rheology modifiers such as alginate, chitosan or methylcellulose to produce patient-specific scaffolds with programmed pore geometry, macroporosity and reproducible channel interconnectivity (Jorge Sousa et al., 2025; Schippa, 2023). Fused filament fabrication of fish-scale HAp/poly(lactic acid) composites and cryogenic printing of fish-bone HAp/chitosan/polycaprolactone (PCL) scaffolds extend this approach to stiffer, slower-resorbing constructs (Liu et al., 2024; Thomas et al., 2024).

Electrospinning, finally, blends marine collagen and nano-apatite with biodegradable synthetic carriers such as PCL or poly(lactic-co-glycolic acid) (PLGA) to form non-woven nanofibrous membranes that mimic the dense architecture of periosteum and periodontal basement membranes (Elkhenany et al., 2024; Radulescu et al., 2022). Marine-derived calcium phosphates from other sources, notably cuttlefish bone, have been processed in similar ways and offer useful points of comparison (Al-Rawe et al., 2024; Liu et al., 2022).

2.8 Interfacial Interactions and Crosslinking Strategies

In simple blends, collagen and apatite are held together mainly by weak electrostatic attraction and hydrogen bonding between the phosphate and hydroxyl groups of the mineral and the carboxyl and amine groups of the protein (Mudhafar et al., 2021; Radulescu et al., 2022). Such constructs swell considerably, retain particles loosely and disaggregate rapidly in water (Mudhafar et al., 2021; Saeed et al., 2025). Chemical and physical crosslinking is used to build durable networks and prolong survival in vivo (Diogo et al., 2018; Oryan et al., 2018; Saeed et al., 2025), as summarised in Figure 4B and Table 3.

Carbodiimide chemistry, using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) with N-hydroxysuccinimide (NHS), is the leading zero-length route (Diogo et al., 2018; Oryan et al., 2018). EDC activates free carboxyl groups on aspartic and glutamic acid residues, which then react with lysine ε-amino groups to form stable isopeptide bonds without introducing a bridging molecule into the matrix (Diogo et al., 2018; Oryan et al., 2018). EDC/NHS-crosslinked mCol–mHAp scaffolds show substantially higher hydrated compressive moduli (about 5–15 MPa), less swelling and resistance to

Table 3. Formulation, crosslinking, microstructure, and mechanical performance of marine collagen–hydroxyapatite and related biogenic apatite scaffolds. The table compares seven scaffold systems built from fish- or cuttlefish-derived apatite, alone or with marine collagen and natural or synthetic polymers, produced by freeze-drying, extrusion bioprinting, fused filament fabrication, cryogenic printing or solution casting. Zero-length EDC/NHS crosslinking and polymer reinforcement increase compressive stiffness from below 0.1 MPa (uncrosslinked collagen) into the megapascal range and slow swelling and enzymatic degradation. Porosity and pore size were measured by different methods across studies and are therefore indicative rather than directly comparable. Moduli refer to the hydrated state unless otherwise stated.

Table 4. Preclinical in vitro and in vivo performance of fish-derived collagen and biogenic hydroxyapatite constructs in bone and craniofacial regeneration models. The table summarises seven preclinical studies that tested marine collagen and/or biogenic apatite constructs in osteoblast-like cell cultures, antimicrobial assays and critical-sized calvarial or tibial defects in rats and rabbits. Constructs were consistently cytocompatible (> 90% viability; < 2% haemolysis) and upregulated osteogenic markers (ALP, RUNX2, COL1A1, OCN, BSP). In vivo, they supported defect bridging and radiodensity approaching a bovine xenograft (Bio-Oss). Note that the Prathap et al. (2025) composite used eggshell-derived, not fish-derived, HAp, and that all in vivo models were small-animal, non-dental sites.

collagenase digestion over several weeks (Diogo et al., 2018; Elkhenany et al., 2024) (Figure 4A).

Physical methods are an alternative. Dehydrothermal (DHT) treatment under vacuum at 100–120 °C forms condensation crosslinks and sterilises the matrix at the same time, though excessive heat can denature the helix if conditions are not carefully calibrated (Binlateh et al., 2022; Saeed et al., 2025). A third strategy incorporates secondary biopolymers, such as crustacean chitosan, hyaluronic acid or fucoidan, to form interpenetrating polyelectrolyte complexes that bridge collagen and apatite, improving cohesion and adding antimicrobial activity (Elkhenany et al., 2024; Mudhafar et al., 2021; Radulescu et al., 2022).

2.9 Preclinical Applications in Dental and Craniofacial Regeneration

Combining marine collagen with biogenic nano-apatite produces a microenvironment that is both osteoconductive and, to some degree, osteoinductive (Elkhenany et al., 2024; Prathap et al., 2025; Saeed et al., 2025). The main application areas investigated so far are summarised in Figure 1 and Table 4.

In vitro cellular responses. Studies using DPSCs, hPDLCs and osteoblast-like lines (MG-63, MC3T3-E1 and hFOB) consistently report good cytocompatibility (Elkhenany et al., 2024; Mudhafar et al., 2021; Núñez-Tapia et al., 2026; Prathap et al., 2025). Alamar Blue and MTT assays show viabilities above 90%–95% (Mudhafar et al., 2021; Núñez-Tapia et al., 2026; Prathap et al., 2025), and fish-derived nano-HAp produces less than 2% haemolysis, within the ISO 10993-4 threshold for blood-contacting devices (Habib et al., 2025). Ions released from the mineral phase (Ca²⁺, PO₄³⁻, Mg²⁺ and Sr²⁺) appear to act together with collagen–integrin signalling to upregulate early markers such as ALP and runt-related transcription factor 2 (RUNX2), followed by type I collagen (COL1A1), osteopontin (OPN), osteocalcin (OCN) and dentin sialophosphoprotein (DSPP) (Elkhenany et al., 2024; Jorge Sousa et al., 2025; Radulescu et al., 2022; Saeed et al., 2025).

Periodontal regeneration and GBR membranes. The goal in advanced periodontitis is coordinated regeneration of alveolar bone, cementum and functionally oriented periodontal ligament fibres (Binlateh et al., 2022; Elkhenany et al., 2024). Crosslinked bilayer or functionally graded mCol–mHAp membranes have been developed for guided tissue regeneration (GTR) and GBR (Binlateh et al., 2022; Elkhenany et al., 2024). A dense collagenous outer layer blocks the apical migration of epithelial cells, while a porous, apatite-rich inner layer stabilises the clot and recruits osteoprogenitors (Binlateh et al., 2022; Elkhenany et al., 2024; Zhang et al., 2025) (Figure 4B).

Alveolar bone repair and socket preservation. These composites have been tested mainly in rodent critical-sized defect models (Elkhenany et al., 2024; Parisi et al., 2020; Prathap et al., 2025). In rat calvarial defects, an injectable composite of fish-scale collagen and eggshell-derived hydroxyapatite in glycerol produced osseous bridging, higher radiodensity and greater trabecular volume than untreated defects or hydroxyapatite alone over 90 days, approaching a commercial bovine xenograft (Bio-Oss) with less chronic inflammation (Prathap et al., 2025). In rat tibial defects, marine collagen/hydroxyapatite composites accelerated woven bone formation, early vascular invasion and lamellar remodelling (Parisi et al., 2020).

Pulp capping and dentin remineralisation. mCol–mHAp nanocomposites have also been proposed for direct and indirect pulp capping (Elkhenany et al., 2024; Radulescu et al., 2022). Placed over exposed pulp, the material is thought to form a bioactive seal that buffers acidic bacterial products, encourages tertiary reparative dentin and drives DPSCs towards odontoblast-like cells without causing irreversible pulpal necrosis (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025).

3. Methods

3.1 Information Sources

Electronic searches were run in PubMed/MEDLINE, Scopus, Web of Science Core Collection and Google Scholar (first 200 results ranked by relevance). The reference lists of key reviews on fish collagen (Jafari et al., 2020; Sharma et al., 2024), marine-derived hydroxyapatite (Habib et al., 2025; Suresh et al., 2026), marine biomaterials (Elkhenany et al., 2024; Khrunyk et al., 2020; Saeed et al., 2025) and waste valorisation (Coppola et al., 2021; Maschmeyer et al., 2020; Zhang et al., 2025) were hand-searched for additional eligible studies. Grey literature, including theses relevant to 3D printing of fish-derived bioceramics (Schippa, 2023), was considered when it reported primary experimental data. The final search was completed on [insert date], covering records published between January 2013 and [insert month] 2026.

3.2 Search Strategy

Search strings combined controlled vocabulary (MeSH terms in PubMed) with free-text keywords grouped into four concept blocks joined by the Boolean operator AND, with synonyms within each block joined by OR:

1. Source: “fish” OR “marine” OR “seafood” OR “aquaculture” OR “fish scale*” OR “fish skin” OR “fish bone*” OR “by-product*” OR “side-stream*” OR “waste”.

2. Material: “collagen”[MeSH] OR “atelocollagen” OR “Durapatite”[MeSH] OR “hydroxyapatite” OR “bioapatite” OR “calcium phosphate*”.

3. Construct: “nanocomposite*” OR “composite*” OR “scaffold*” OR “Tissue Scaffolds”[MeSH] OR “membrane*” OR “hydrogel*” OR “3D print*”.

4. Application: “Tissue Engineering”[MeSH] OR “Bone Regeneration”[MeSH] OR “Guided Tissue Regeneration, Periodontal”[MeSH] OR “dental” OR “periodontal” OR “alveolar” OR “pulp capping” OR “craniofacial” OR “maxillofacial”.

The PubMed string was adapted to the syntax of each database. Searches were limited to English-language publications. The full database-specific strings are provided in Supplementary Table S1.

3.3 Eligibility Criteria

Studies were included if they (a) used fish or other marine processing by-products as the source of collagen, hydroxyapatite or both; (b) reported extraction conditions together with at least one quantitative physicochemical outcome (yield, Td, Ca/P ratio, crystallite size, crystallinity, porosity, swelling, degradation or mechanical properties); or (c) evaluated a marine collagen and/or biogenic apatite construct in vitro or in vivo for bone, dental or craniofacial regeneration. Peer-reviewed reviews were retained as contextual sources. Studies were excluded if they used only mammalian or fully synthetic materials, lacked experimental characterisation, were conference abstracts without full data, or were not available in full text. A small number of non-dental studies were retained when they provided otherwise unavailable characterisation data for fish-derived apatite, such as its use as a filler in packaging (Anbuchezhian et al., 2024; Kongthong et al., 2024), and studies of eggshell- or cuttlefish-derived apatite were included only as comparators (Al-Rawe et al., 2024; Liu et al., 2022; Prathap et al., 2025).

4. Extraction Efficiency, Physicochemical Signatures, Scaffold Mechanics, and Regenerative Outcomes of Fish-Derived Collagen–Hydroxyapatite Systems

4.1 Extraction Yields and Biophysical Profiles of Marine Collagen

How much collagen can be recovered from fish waste depends, it turns out, on three things more than any other: which tissue is used, how it is pretreated and which solubilisation route follows (Sharma et al., 2024) (Table 1). Skin is reliably the richest source. Scales and mineralised bone lag well behind (Ahmed et al., 2019; Sharma et al., 2024). With conventional ASC extraction in 0.5 M acetic acid at 4 °C, reported yields were about 19.07% for Nile tilapia skin (Song et al., 2019), 10.5%–22.4% for catfish skin (Abbas et al., 2022; Hukmi & Sarbon, 2018) and, rather strikingly, 46.13%–63.40% for the skins of the tropical carps Labeo rohita and Catla catla (Mahboob, 2015; Pal et al., 2015) (Table 1). Mineralised tissues gave far less. Bigeye tuna scales and bones yielded only 4.6% and 2.6%, respectively (Ahmed et al., 2019), probably because dense calcium phosphate deposits hinder acid penetration.

Pepsin helped in every matrix examined (Pal et al., 2015; Sharma et al., 2024). By trimming the telopeptides while leaving the Gly–X–Y helical domain intact (Binlateh et al., 2022; Sharma et al., 2024), PSC extraction raised yields to 19.61% in tilapia skin (Song et al., 2019), as much as 27.3% in catfish skin (Hukmi & Sarbon, 2018) and 64.94%–69.53% in rohu and catla skin (Pal et al., 2015) (Table 1). The gain in tilapia was modest, admittedly; in tuna skin and in the carps it was larger. CD spectra of these collagens typically showed a positive peak near 220–221 nm, a negative trough near 197–198 nm and a crossover near 213 nm, which suggests that the native triple helix had survived extraction without denaturing to random coil (Lim et al., 2023; Song et al., 2019).

The thermal data were more sobering (Figure 2B). Differential scanning calorimetry showed a clear split by habitat (Saeed et al., 2025; Sharma et al., 2024). Collagens from tropical and warm-water freshwater fish denatured not far below body temperature: 34.4–36.1 °C in Nile tilapia (Song et al., 2019), 32.5–36.0 °C in catfish (Abbas et al., 2022; Hukmi & Sarbon, 2018) and 36.5 °C in giant gourami (Saeed et al., 2025). These species also had relatively high imino acid contents of roughly 82–88 hydroxyproline residues per 1000 (Abbas et al., 2022), consistent with the stabilising role of pyrrolidine rings and water-bridged hydrogen bonds (Binlateh et al., 2022; Saeed et al., 2025). Cold-water species told a different story. Atlantic cod collagen began to denature at around 16.5 °C (Sousa et al., 2020), and cold-water species more generally fell in the 15–28 °C range (Saeed et al., 2025). Whatever its other merits, collagen of this kind cannot be implanted unmodified at 37 °C.

4.2 Structural, Stoichiometric, and Trace-Element Hallmarks of Biogenic Hydroxyapatite

Fish-derived apatite differed from stoichiometric synthetic HAp (Ca/P = 1.67) in ways that were consistent across studies (Habib et al., 2025; Suresh et al., 2026) (Table 2). Calcination between 600 and 1100 °C reliably removed fat and collagen and left crystalline apatite (Dermawan et al., 2022; Habib et al., 2025). The effect of temperature was visible in striped catfish bone: raising the calcination temperature from 500 to 900 °C transformed broad, poorly defined diffraction patterns into sharp reflections, with crystallinity of 95.5%–99.5% and a mean crystal size of 71.38 nm by transmission electron microscopy (TEM) (Anbuchezhian et al., 2024). Red snapper scales calcined at 1100 °C gave hexagonal hydroxyapatite with a minor β-TCP component, a recovery of 25.8% ± 2.15% and a Ca/P ratio of 1.668–1.708 (Núñez-Tapia et al., 2026). Black tilapia scales and bones calcined at 700–1000 °C produced hexagonal nanorods of 30–65 nm, with crystallinity above 90% at 900 °C (Dermawan et al., 2022; Jaffri et al., 2022).

Across all sources, Ca/P ratios ran from 1.30 to 1.74 (Figure 3B; Table 2), although the lowest values belonged to catfish bone calcined at only 500 °C and should probably be read as incompletely processed material rather than a true species signature (Anbuchezhian et al., 2024). Trace-element analyses, mostly by inductively coupled plasma optical emission spectrometry (ICP-OES) and energy-dispersive X-ray spectroscopy, consistently detected functional dopants (Table 2; Figure 3C):

• Magnesium (Mg²⁺) was present at roughly 3.5 times the level in synthetic HAp in catla-scale apatite (Deb et al., 2019; Radulescu et al., 2022), where it partially replaces calcium and is associated with osteoblast adhesion and ALP activation (Shi et al., 2018; Zainol et al., 2024).

• Strontium (Sr²⁺) was enriched up to about tenfold relative to synthetic controls in the same material (Deb et al., 2019), and is linked to bone deposition and suppression of osteoclastic resorption (Diogo et al., 2018; Zainol et al., 2024).

• Carbonate (CO₃²⁻) produced characteristic FTIR bands at 873–875 cm⁻¹ (ν₂) and 1410–1455 cm⁻¹ (ν₃), indicating B-type substitution for phosphate (Anbuchezhian et al., 2024; Gnanasekaran et al., 2024). The resulting lattice strain is thought to raise chemical reactivity and promote resorption in vivo (Habib et al., 2025; Radulescu et al., 2022).

• Fluoride (F⁻) reached about 1.0 wt% in shark teeth and skeletal apatite, alongside Ca 44.36 wt%, P 22.8 wt%, Na 0.9 wt%, Mg 0.65 wt% and Sr 0.25 wt% (Diogo et al., 2018). Fluorhydroxyapatite of this kind would be expected to resist acid attack in the mouth (Habib et al., 2025).

Alkaline hydrolysis, using for example 2 M NaOH at 250 °C (Gnanasekaran et al., 2024) or 2%–20% NaOH at 70–100 °C (Kongthong et al., 2024), gave quite different powders. Red big eye bone yielded unagglomerated spherical particles of 50–80 nm (Gnanasekaran et al., 2024), and Asian sea bass scales produced material with a BET surface area of 90.80 m²/g and 79% crystallinity (Kongthong et al., 2024). Because the process avoids high-temperature sintering, carbonate groups appear to be retained (Gnanasekaran et al., 2024; Suresh et al., 2026).

4.3 Microstructure, Porosity, and Mechanics of Crosslinked Nanocomposites

On their own, marine collagen hydrogels and sponges are soft and dissolve quickly (Elkhenany et al., 2024; Radulescu et al., 2022). Adding biogenic apatite, and especially crosslinking the result, changed this substantially (Table 3; Figure 4A).

The clearest evidence comes from freeze-dried blue shark collagen–bioapatite sponges (Diogo et al., 2018). Crosslinking chemistry largely determined architecture and stiffness. EDC/NHS-treated scaffolds had porosities of 48.8%–85.4%, interconnectivity of 49.8%–96.8% and mean pore diameters of 50.5–126.3 µm. The 50:50 formulation crosslinked with 12.5% EDC/NHS reached a hydrated compressive modulus of about 8.95 MPa, compared with less than 0.1 MPa for uncrosslinked collagen (Diogo et al., 2018; Elkhenany et al., 2024). It also retained a swelling capacity of 231%, which is arguably useful for nutrient transport, and remained intact in collagenase solution for 21 days (Diogo et al., 2018) (Table 3).

Scale-derived systems behaved somewhat differently. In fish-scale HAp/collagen/silver nanoparticle composites,

Figure 3. Processing, stoichiometry, and ionic substitution of fish-derived biogenic hydroxyapatite. (A) Temperature windows for alkaline hydrolysis (green), moderate calcination (blue) and high-temperature calcination (orange), with the typical particle size, crystallinity, carbonate retention and phase outcome of each. (B) Reported Ca/P molar ratios of apatite from nine fish sources (bars, ranges; dots, single values) relative to stoichiometric synthetic HAp (dashed line, 1.67). (C) Principal trace ions in biogenic apatite and their proposed biological roles in osteogenesis, resorption and acid resistance. Data from Table 2 and Habib et al. (2025).

Figure 4. Mechanical reinforcement and interfacial stabilisation of marine collagen–apatite nanocomposites. (A) Compressive modulus or strength of representative scaffolds on a logarithmic scale, showing the increase from below 0.1 MPa for uncrosslinked collagen to roughly 1–13 MPa after EDC/NHS crosslinking, bioprinting with alginate or reinforcement with PCL or PLA. (B) Left: schematic of a bilayer GTR/GBR membrane in which a dense collagen layer excludes epithelium and a porous apatite-rich layer supports osteogenesis. Right: the four main crosslinking and stabilisation strategies, with their mechanisms and limitations. Data from Table 3, Diogo et al. (2018) and Oryan et al. (2018).

the 80:20 HAp:collagen ratio showed the most uniform particle dispersion, the least swelling and the lowest degradation over 14 days in phosphate-buffered saline (PBS), with porosity of 65%–82% and pores of 100–250 µm (Mudhafar et al., 2021). Bioprinted tilapia collagen/fish-bone HAp/alginate scaffolds, ionically crosslinked in CaCl₂, formed macro-channels of 200–400 µm and reached compressive moduli of 1.2–3.4 MPa; they also kept the surrounding pH between 7.2 and 7.4 during 21 days of incubation (Jorge Sousa et al., 2025). Stiffer still were constructs that incorporated synthetic polymers. Fish-bone HAp/chitosan/PCL scaffolds reached 6.8–10.5 MPa and lasted more than 8 weeks (Habib et al., 2025; Liu et al., 2024), and fish-scale HAp/PLA scaffolds printed by fused filament fabrication reached 13.4 ± 1.2 MPa, which falls within the range reported for human trabecular alveolar bone (Thomas et al., 2024; Zhang et al., 2025) (Table 3; Figure 4A).

4.4 Cytocompatibility, Cellular Responses, and In Vivo Osteogenesis

Biological testing, on the whole, was reassuring (Table 4). MTT, Alamar Blue and lactate dehydrogenase (LDH) assays using MG-63 and hFOB 1.19 osteoblast-like cells and MC3T3-E1 pre-osteoblasts gave viabilities above 90%–95% for a range of formulations (Diogo et al., 2018; Mudhafar et al., 2021; Núñez-Tapia et al., 2026). Nano-HAp from red big eye bone caused less than 2% haemolysis while supporting a proliferation rate of 92% relative to control (Gnanasekaran et al., 2024). The silver-containing fish-scale composite went a step further, producing inhibition zones above 14 mm against Escherichia coli and Staphylococcus aureus without measurable cytotoxicity (Mudhafar et al., 2021) (Table 4).

At the biochemical level, ALP activity in osteoblasts cultured on shark bioapatite composites rose at days 7 and 14 compared with pure collagen or synthetic HAp (Diogo et al., 2018; Shi et al., 2018). Quantitative PCR in several studies showed upregulation of RUNX2, COL1A1, OPN and OCN (Elkhenany et al., 2024; Liu et al., 2024), and bone sialoprotein (BSP) was elevated on marine-derived compared with synthetic HAp (Elkhenany et al., 2024).

The in vivo data, while limited in number, were broadly consistent with these findings (Table 4). The most detailed study evaluated a composite of rohu fish-scale collagen and eggshell-derived hydroxyapatite (EHPA/COLL) in 5 mm critical-sized calvarial defects in 24 male Wistar rats over 90 days (Prathap et al., 2025). On cone-beam computed tomography (CBCT) with ImageJ grey-value analysis, the composite reached a mean radiodensity of 134.5 ± 5.6, well above the empty defect (62.2 ± 8.8; p < 0.0001) and eggshell HAp alone (93.8 ± 11.8; p < 0.01), though still somewhat below the Bio-Oss xenograft (159.4 ± 26.5) (Prathap et al., 2025) (Figure 5). Histologically, empty defects were spanned only by thin fibrous tissue, whereas composite-treated defects showed dense trabecular bridging, capillary ingrowth and lamellar remodelling, with little mononuclear infiltration and none of the fibrous encapsulation seen around some xenograft granules (Prathap et al., 2025).

Other models pointed in the same direction. In rat tibial defects, marine collagen/HAp composites accelerated the transition from woven to lamellar bone and increased bone area within 30 days, without foreign-body granulomas (Parisi et al., 2020). Fish-scale HAp/chitosan/PCL scaffolds increased bone volume fraction (BV/TV) about 2.4-fold over pure PCL in rat calvarial defects at 8 weeks (Liu et al., 2024; Zhang et al., 2025), and biogenic HAp outperformed synthetic HAp in mature lamellar bone formation in rabbit calvarial defects at 12 weeks (Elkhenany et al., 2024) (Table 4).

5. Weighing Biological Promise Against Translational Reality

5.1 Overview of Principal Findings

Taken together, the studies reviewed here make a reasonably persuasive case that fish processing waste can yield both components of a bone-like nanocomposite, and that the resulting materials behave, at least in rodents, much as one would hope. Collagen can be extracted in useful quantities, particularly by pepsin digestion of skin (Table 1); biogenic apatite carries a trace-element profile that synthetic HAp lacks (Table 2); crosslinking turns a fragile sponge into a construct with a modulus in the megapascal range (Table 3); and early in vivo results approach those of a widely used xenograft (Table 4; Figure 5). Whether these observations add up to a clinically viable dental product is, however, a separate and much less settled question.

5.2 Source Selection and Extraction: Why Species and Tissue Matter

One practical implication of Table 1 is that the choice of raw material may matter as much as the choice of extraction method. Warm-water skins, especially those of tilapia, catfish and the tropical carps, combine relatively high yields with denaturation temperatures in the mid-30s (Abbas et al., 2022; Pal et al., 2015; Song et al., 2019). That still falls short of 37 °C, but it is a far easier starting point than cod collagen at around 16.5 °C (Sousa et al., 2020). It seems sensible, then, for dental applications to favour tropical aquaculture species, which happen also to be among the most heavily farmed fish worldwide (Coppola et al., 2021; Sharma et al., 2024). The very high carp yields reported by Pal et al. (2015) and Mahboob (2015) deserve some caution, because yield bases (wet versus dry weight, pooled versus single fractions) were not always reported consistently, and comparisons across studies are therefore approximate.

PSC extraction appears preferable for biomedical use not only because of yield but because telopeptide removal reduces antigenicity (Binlateh et al., 2022; Saeed et al., 2025). The CO₂-acidified water route described by Sousa et al. (2020) is interesting from a green-chemistry perspective, though its low yield and the poor thermal stability of cod collagen limit its immediate relevance to dentistry.

5.3 The Thermal Stability Problem

The gap between marine collagen Td and body temperature (Figure 2B) is, in our reading, the single most important materials limitation identified in this review. Uncrosslinked collagen from most species would be expected to begin unfolding soon after implantation (Binlateh et al., 2022; Elkhenany et al., 2024). Crosslinking clearly helps; EDC/NHS treatment both stiffened shark collagen composites and protected them from collagenase for three weeks (Diogo et al., 2018). Yet glutaraldehyde, the traditional alternative, leaves cytotoxic residues, and the safer options, namely genipin, proanthocyanidins and microbial transglutaminase, have not yet been optimised systematically for marine collagen (Binlateh et al., 2022; Oryan et al., 2018; Saeed et al., 2025). Dehydrothermal treatment can sterilise and crosslink at once, but the same heat that forms crosslinks may denature a thermally fragile helix if conditions are not carefully controlled (Binlateh et al., 2022; Saeed et al., 2025). Arguably, the mineral phase itself contributes some thermal protection by restricting chain mobility, though this has been inferred more often than measured directly.

5.4 Biogenic Apatite Chemistry: An Advantage and a Source of Variability

The trace-element signature of fish-derived apatite (Table 2; Figure 3C) is often presented as its main selling point, and with some justification. Mg²⁺ and Sr²⁺ are both implicated in promoting osteoblast activity and restraining osteoclasts (Diogo et al., 2018; Zainol et al., 2024), carbonate increases resorbability (Habib et al., 2025), and fluoride could plausibly help in the acidic oral environment (Diogo et al., 2018). The enhanced ALP activity and bone formation seen with biogenic compared with synthetic HAp (Elkhenany et al., 2024; Shi et al., 2018) is consistent with these mechanisms, although few studies have isolated the contribution of individual ions.

The same chemistry, though, makes the material harder to standardise. Ca/P ratios spanned 1.30–1.74 (Figure 3B), and composition shifted with species, diet, habitat and, most dramatically, processing temperature (Dermawan et al., 2022; Habib et al., 2025). Processing conditions can also work against each other. Higher calcination temperatures produce purer, more crystalline powders and may form β-TCP, which could be useful for resorption (Piccirillo et al., 2013), but they also drive off the carbonate that is supposedly beneficial (Dermawan et al., 2022; Habib et al., 2025). Alkaline hydrolysis preserves carbonate and surface area (Gnanasekaran et al., 2024; Kongthong et al., 2024), but whether it removes all organic residues and pathogens as completely as calcination does has not, to our knowledge, been established. There is probably no single optimal process; the right choice likely depends on whether a fast-resorbing filler or a stable, load-bearing phase is needed.

5.5 Mechanics, Architecture, and the Choice of Fabrication Route

The mechanical data (Table 3; Figure 4A) suggest a rough division of labour among fabrication approaches. Crosslinked freeze-dried collagen–apatite sponges (about 9 MPa) and bioprinted hydrogels (1.2–3.4 MPa) seem best suited to contained defects such as extraction sockets and periodontal intrabony defects, where they need to maintain space rather than bear load (Diogo et al., 2018; Jorge Sousa et al., 2025). PCL- and PLA-reinforced constructs reach the stiffness of trabecular alveolar bone (Liu et al., 2024; Thomas et al., 2024) and may therefore be more suitable for ridge augmentation, although adding synthetic polymer dilutes the “marine” and biomimetic character of the material and slows resorption to months (Zhang et al., 2025). For GTR/GBR, the bilayer concept (Figure 4B) remains attractive, but most supporting data describe the design rationale rather than direct comparisons with commercial collagen membranes (Binlateh et al., 2022; Elkhenany et al., 2024).

The swelling and porosity values also warrant some caution. Pore sizes of 50–400 µm and porosities of 45%–88% were reported (Table 3), but measurement methods varied (micro-CT, mercury porosimetry and SEM image analysis), and hydrated and dry values were not always distinguished. Comparisons across Table 3 should therefore be read as indicative.

5.6 Interpreting the Preclinical Evidence

The biological results (Table 4) are encouraging, but several features of the evidence base limit how far they can be generalised. Most importantly, the strongest in vivo dataset (Prathap et al., 2025) used eggshell-derived, not fish-derived, hydroxyapatite combined with fish collagen. It therefore supports the value of marine collagen in a composite graft but does not, strictly speaking, test a fully fish-derived nanocomposite. It is also worth noting that the composite did not exceed Bio-Oss in radiodensity (Figure 5); it approached it, with lower variance and less inflammation. That is a respectable result, though not evidence of superiority.

Beyond that, nearly all in vivo work used calvarial or long-bone defects in rats or rabbits (Elkhenany et al., 2024; Parisi et al., 2020; Prathap et al., 2025). These sites are not loaded by mastication, are not exposed to the oral microbiome and heal differently from alveolar bone. Follow-up periods were short (30 days to 12 weeks), group sizes were small, and blinding and randomisation were reported inconsistently. Claims about pulp capping and dentin remineralisation (Elkhenany et al., 2024; Radulescu et al., 2022; Zhang et al., 2025) rest largely on extrapolation, and we could find little direct experimental evidence of dentin bridge formation with fish-derived composites. The antimicrobial silver-loaded formulation (Mudhafar et al., 2021) is appealing for peri-implantitis and periodontitis, but long-term silver exposure in oral tissues would need its own safety assessment.

5.7 Translational and Regulatory Bottlenecks

Several obstacles would need to be cleared before routine clinical use. Raw-material heterogeneity. Species, harvest location, season, water temperature and diet all alter imino acid content, collagen Td and apatite trace chemistry (Binlateh et al., 2022; Habib et al., 2025; Zhang et al., 2025). Standardised sourcing, handling and GMP-compliant cleaning would be needed to control batch-to-batch variation.

Sterilisation. Autoclaving, dry heat and high-dose gamma irradiation can cleave peptide chains, disrupt the triple helix and alter the apatite lattice (Saeed et al., 2025; Zhang et al., 2025). Ethylene oxide, supercritical CO₂ and low-dose electron-beam irradiation may be gentler, but each requires validation for sterility assurance and for preservation of mechanical and biological performance (Saeed et al., 2025). Large-animal validation. Orthotopic dental models in dogs, pigs or sheep, including periodontal pockets, sinus floor augmentation and ridge preservation under masticatory loading and in a microbially complex environment, are a necessary step before human trials (Elkhenany et al., 2024; Habib et al., 2025). Economic and regulatory viability. The circular-economy argument is compelling in principle (Coppola et al., 2021; Maschmeyer et al., 2020), but life-cycle and cost analyses comparing fish-derived products with bovine xenografts at commercial scale are, as far as we can tell, still lacking.

5.8 Limitations of This study

This review has limitations of its own. The search was restricted to English-language publications, and some relevant work in other languages may have been missed. The heterogeneity of species, methods and outcomes precluded meta-analysis, and the synthesis is therefore narrative. Several included sources are reviews rather than primary studies, and some values were reported inconsistently across sources (for example, yield bases and hydration states), which limits direct comparison. Publication bias towards positive findings is likely in this young field. Finally, a few comparator studies used non-fish marine or avian sources (Al-Rawe et al., 2024; Liu et al., 2022; Prathap et al., 2025), and their results should not be attributed directly to fish-derived materials.

5.9 Future Directions

Several priorities emerge. Thermally stable collagen could be obtained by prioritising warm-water species and by systematically comparing non-toxic crosslinkers at clinically relevant temperatures. Process–property maps linking calcination or hydrolysis conditions to Ca/P ratio, carbonate retention and ion release would help match apatite to indication. Head-to-head studies of fully fish-derived composites against Bio-Oss and commercial collagen membranes in large-animal dental models are clearly needed. Standardised reporting of yields, mechanical testing conditions and in vivo design, ideally following ARRIVE guidelines, would make future evidence easier to synthesise. Life-cycle assessment and early dialogue with regulators would help determine whether the environmental promise of this approach survives contact with industrial reality.

6. Conclusion

Fish processing waste, discarded in tens of millions of tonnes each year, contains the two principal building blocks of bone and dentin. This review shows that type I collagen and biogenic nano-hydroxyapatite can be recovered from fish skins, scales and bones with reasonable efficiency, and that their combination yields nanocomposites with tunable porosity, megapascal-range stiffness after crosslinking, natural ionic doping and convincing cytocompatibility. Early animal studies suggest bone regeneration approaching that achieved with a commercial bovine xenograft, with comparatively mild inflammation and no signs of cytotoxicity. Real obstacles remain, however, and they should not be understated. Marine collagen is thermally fragile, raw materials vary considerably, sterilisation and crosslinking remain unresolved, and the evidence comes almost entirely from small-animal, non-dental models. With standardised processing, safer crosslinking and rigorous large-animal and clinical testing, fish-derived collagen–hydroxyapatite nanocomposites could become a sustainable, culturally acceptable alternative for periodontal, alveolar and pulp-related regeneration.

References