I now have comprehensive data across all three categories. Let me compile the full answer.
Here is a curated list of PubMed articles organized by which fabrication method showed superior surface roughness (measured by profilometer), across occlusal splint and closely related denture base PMMA studies.
Category 1: Heat-Cure (Conventional) Shown to Be Better or Comparable
1. Ozden et al. (2025) - PMID: 39895617
Title: "Impact of Fabrication Techniques and Polishing Procedures on Surface Roughness of Denture Base Resins"
Journal: Journal of Visualized Experiments (JoVE)
DOI: 10.3791/67844
Key Finding: Before and after polishing, 3D-printed specimens had the highest surface roughness (Ra). Milled and heat-polymerized PMMA showed no statistically significant difference from each other after polishing - both outperformed 3D printed. The 3D-printed group remained significantly rougher than both heat-cured and milled groups (p < 0.001) even after polishing.
Relevance: This study directly supports heat-cure as being better than 3D-printed in surface roughness. When comparing unpolished surfaces, heat-cured also performed comparably to milled - making heat-cure a strong performer.
2. Uma, Lertyingyos & Lilitsuvan (2026) - PMID: 41777279
Title: "Analyzing Color, Surface Roughness, and Microhardness on the Unpolished and Polished Surfaces of Occlusal Splint Materials From Conventional and CAD-CAM Fabrication Methods"
Journal: International Journal of Dentistry
DOI: 10.1155/ijod/9002663
PMC: PMC12951543
Key Finding (Occlusal Splints specifically): Among the six materials tested (SC, HC, ML-A, ML-B, 3D-A, 3D-B), the heat-cured (HC) group ranked 3rd lowest in Ra on unpolished surfaces (Ra order from lowest to highest: ML-A 0.085 µm > ML-B 0.117 µm > HC > 3D-B > SC > 3D-A). Crucially, 3D-A had the highest roughness, and HC outperformed both 3D-printed brands. After polishing, HC also maintained strong microhardness (18.85 VHN, second only to ML-B). The paper notes: "CAD-CAM fabrication does not necessarily result in superior surface roughness."
Relevance: Directly on occlusal splints. Heat-cure outperformed both 3D-printed brands in surface roughness.
Category 2: 3D-Printed Shown to Be Better
3. Nejatidanesh et al. (2025) - PMID: 40694394
Title: "Comparative Evaluation of Surface Properties of Milled, 3D-Printed, and Conventional Denture Base Materials: Implications for Clinical Use"
Journal: International Journal of Prosthodontics
DOI: 10.11607/ijp.9285
Key Finding: Using a digital laser profilometer, 3D-printed (DentaBase) achieved Ra of 0.12 µm, which was the lowest among all groups - outperforming both milled (0.15 µm) and conventional heat-polymerized (0.16-0.20 µm). The study concluded: "Milled and 3D-printed resins demonstrated better surface properties" with 3D printed showing the absolute best roughness value.
Relevance: One of the clearest papers where 3D printing produces the lowest Ra, suggesting superiority for surface smoothness.
4. Rueda, Sepsick & Hammamy et al. (2025) - PMID: 40181636
Title: "The Effect of Different Surface Treatments on the Roughness, Translucency, and Staining of 3D-Printed Occlusal Device Materials"
Journal: Journal of Esthetic and Restorative Dentistry
DOI: 10.1111/jerd.13476
PMC: PMC12159789
Key Finding (Occlusal Devices specifically): Using a contact profilometer on occlusal device materials specifically. When 3D-printed specimens were polished or resin-coated, they achieved surface roughness comparable to or better than conventional/milled reference specimens. Optical polish tank printing and glycerin curing further optimized smoothness. Conclusion: 3D-printed occlusal devices with proper post-processing produce competitive (and in some treatments, superior) surface quality.
Relevance: Directly on occlusal devices. Supports 3D-printed superiority when proper surface treatment is applied.
Category 3: Milled (CAD/CAM Subtractive) Shown to Be Better
5. Uma, Lertyingyos & Lilitsuvan (2026) - PMID: 41777279 (also cited above)
Title: Same as above
Key Finding here: ML-A had the lowest Ra of all six groups (0.085 µm) on unpolished occlusal splint surfaces, and ML-B had the second lowest (0.117 µm). Both milled groups outperformed heat-cured, self-cured, and both 3D-printed brands. The conclusion directly states: "The ML-B occlusal splint showed the best overall performance... low surface roughness."
Relevance: The strongest occlusal splint-specific paper showing milled is best for surface roughness.
6. Singh, Jain & Bhasin et al. (2025) - PMID: 40585711
Title: "Comparative Evaluation of Surface Roughness, Wettability, and Hardness of Conventional, Heat-Polymerized, CAD/CAM-Milled, and 3D-Printed PMMA Denture Base Resins: An In Vitro Study"
Journal: Cureus
DOI: 10.7759/cureus.85008
PMC: PMC12205262
Key Finding: Contact profilometer (Surftest SJ-210, Mitutoyo). CAD/CAM-milled specimens showed the smoothest surfaces: Ra = 0.16 ± 0.014 µm, vs 3D-printed (0.18 ± 0.016 µm) and conventional heat-polymerized (0.21 ± 0.019 µm, the roughest). Conclusion: "CAD/CAM-milled PMMA denture bases demonstrated superior surface smoothness and hardness."
Relevance: Clear ranking - milled best, heat-cure worst for surface roughness in this study.
7. Di Fiore, Meneghello et al. (2022) - PMID: 34853238
Title: "Comparison of the Flexural and Surface Properties of Milled, 3D-Printed, and Heat Polymerized PMMA Resins for Denture Bases"
Journal: Journal of Prosthodontic Research
DOI: 10.2186/jpr.JPR_D_21_00116
Key Finding: Contact profilometer used. Milled (CAD) group had the lowest Ra = 0.29 ± 0.16 µm before polishing, vs heat-polymerized (CV) and 3D-printed (3D). Bacterial adhesion was also lowest for milled at 90 minutes. After polishing, all three groups showed similar Ra values. Conclusion: "The CAD group displayed... the lowest roughness before polishing."
Relevance: Well-cited study (2022) confirming milled superiority before polishing.
Summary Table
| Study (PMID) | Method Used for Roughness | Which Was Best | Specific Finding |
|---|
| Ozden 2025 (39895617) | Contact profilometer | Heat-cure ≈ Milled (both better than 3D) | 3D-printed significantly rougher before and after polishing |
| Uma 2026 (41777279) | Profilometer (Ra, µm) | Milled (ML-A = 0.085 µm) | HC outperformed 3D-printed; milled best overall |
| Nejatidanesh 2025 (40694394) | Digital laser profilometer | 3D-printed (0.12 µm) | Lowest Ra of all groups |
| Rueda 2025 (40181636) | Contact profilometer | 3D-printed (post-treated) | Polished/resin-coated 3D achieves best roughness |
| Singh 2025 (40585711) | Contact profilometer (Mitutoyo SJ-210) | Milled (0.16 µm) | Milled < 3D-printed < heat-cure for Ra |
| Di Fiore 2022 (34853238) | Contact profilometer | Milled (0.29 µm) | Milled lowest Ra before polishing |
Important clinical note: The outcome varies depending on material brand, whether specimens are polished, and the specific profilometer protocol. No single fabrication method dominates universally - context and post-processing matter significantly, which is why the literature is distributed across all three categories.