The classification in widest use distinguishes hypoplastic, hypomaturation and hypocalcified forms, with a fourth group combining hypomaturation and hypoplasia with taurodontism (Witkop, 1988). The distinction matters at the chairside. In hypoplastic forms the enamel is thin but well mineralised, and it etches and bonds much as normal enamel does. In hypomaturation and hypocalcified forms the enamel retains excess protein and is poorly mineralised, etching patterns are irregular, and bonds are less dependable (Seow and Amaratunge, 1998). Pretreatment with sodium hypochlorite to remove surface protein has improved bond strength to hypocalcified enamel in laboratory testing (Saroğlu et al., 2006), though clinical confirmation is limited.

Direct composite restorations in young people with amelogenesis imperfecta fail sooner than in unaffected peers, and failure is more frequent in the hypomineralised forms (Pousette Lundgren and Dahllöf, 2014). A randomised trial of ceramic crowns placed in adolescents and young adults with the condition reported good short-term survival and reduced sensitivity, with few pulpal complications (Pousette Lundgren et al., 2015). These findings support indirect restoration earlier than was once customary. They do not show which material or preparation design is best, and follow-up remains short in relation to the lifetime of the patient.

Intraoral scanners capture tooth and soft-tissue surfaces without impression material, which is an advantage when teeth are sensitive and gag reflexes strong (Mangano et al., 2017). Design software allows the proposed tooth form to be related to facial photographs, as in the digital smile design approach (Coachman and Calamita, 2012), and the design can be printed or milled at each stage. A systematic review of complete digital workflows in fixed prosthodontics found the clinical evidence encouraging but sparse (Joda et al., 2017). The protocol below treats digital methods as a means of keeping one design consistent from planning to delivery. It makes no assumption that they are more accurate than conventional ones.

Treatment starts with the diagnosis. The clinician records the enamel phenotype, the family history and any associated findings such as taurodontism, open bite or unerupted teeth, and considers referral for genetic assessment. Radiographs show pulp size and enamel thickness. Baseline measures of sensitivity, plaque, gingival health and oral health-related quality of life are taken so that later change can be judged against them. Caries and gingival inflammation are brought under control before any restorative planning.

Both arches and the intercuspal relationship are scanned, and a standard series of facial and intraoral photographs is taken. Where the vertical dimension is to be changed, a registration in centric relation at the proposed dimension is scanned as well. Rough, pitted enamel and moisture can disturb image stitching, so the scan is checked for gaps before the patient leaves. The files become the reference against which every later stage is compared.

The new tooth forms are designed on the scans with reference to the photographs. Because enamel is thin or worn, the design is largely additive: material is added to restore the original volume of the teeth, and the occlusal vertical dimension is commonly increased to create room for it. A moderate increase is generally well tolerated when it is tested first in provisional form (Abduo and Lyons, 2012). The principle follows the adhesive rehabilitation sequence described for eroded dentitions by Vailati and Belser (2008), in which restorative space is gained by opening the bite and not by cutting teeth.

The design is printed and transferred to the mouth with a silicone index and a temporary resin. The mock-up lets patient, family and clinician see the proposed appearance, hear the effect on speech and check the lip line before anything irreversible is done. Changes requested at this visit are made in the design file, not by adjusting the resin alone, so that the file remains the single source for later steps.

Trial restorations are milled from polymer blocks to the approved design and bonded with little or no preparation. They serve as a test of the new occlusion and vertical dimension over a period of months, protect sensitive teeth, and give an adolescent an acceptable appearance while growth and eruption continue. For a patient who is still growing, this phase may be deliberately prolonged, with the definitive stage deferred until gingival margins are stable. Problems that emerge, such as fractures, discomfort or phonetic difficulty, are corrected in the design, as the return loop in Figure 1 indicates.

Definitive preparation is carried out through the trial restorations or a fresh mock-up, so that reduction is measured from the intended final surface and not from the existing tooth (Magne and Belser, 2004). This keeps removal of tooth tissue to the minimum that the material requires, an important consideration when pulps are large; conventional full-coverage preparations remove far more of the crown than partial-coverage designs do (Edelhoff and Sorensen, 2002). The preparations are scanned, and the scan is aligned to the approved design so that the definitive restorations reproduce it.

The definitive restorations are milled, tried in and cemented adhesively under isolation. Choice of material depends on available thickness, enamel quality and occlusal load; glass-ceramics, zirconia and resin-based blocks each have a place, and the evidence in this condition does not single out one. The bonding procedure is matched to the substrate. Where sound enamel remains, etch-and-rinse adhesion is the reference approach (Van Meerbeek et al., 2003). Where enamel is hypomineralised, margins are placed on better tissue if possible and deproteinisation is considered, in the knowledge that its clinical benefit is not established.

A protective splint is provided where parafunction is suspected, and recall intervals are set by periodontal and caries risk. At review the arches are rescanned and the scans superimposed on the delivery scan, which gives an objective record of wear, chipping and gingival change. The same baseline measures taken at Step 1 are repeated.