Surface Roughness as a Determinant of the Immunological Tissue Response to Additively Manufactured Titanium Implants

Abstract

Background/Aim

Surface roughness strongly affects host responses to titanium implants. Additive manufacturing (AM) by selective laser melting (SLM) allows controlled surface generation but the immune consequences of AM-derived roughness remain unclear.

Materials and Methods

Three AM titanium disks with distinct roughness were analyzed by scanning electron microscopy and profilometry. At 10 and 30 days after subcutaneous implantation of disks in rats, tissue responses were evaluated according to ISO 10993-6. A novel phagocyte accumulation index (PAI) was applied, integrating mononuclear infiltration, multinucleated giant cells, coverage, and morphology.

Results

All specimens induced only mild to moderate inflammation, confirming biocompatibility. Histology showed fibrosis, macrophage persistence, and rare multinucleated giant cells, without intergroup differences by ISO scoring. In contrast, the PAI detected significant differences: high-roughness specimens showed the highest values at day 30, significantly exceeding those of medium-roughness surfaces (p<0.05), while low-roughness surfaces were intermediate in value.

Conclusion

The PAI revealed surface roughness-dependent differences in phagocytic activity not captured by ISO scoring. High-roughness AM surfaces promoted the strongest response, highlighting AM and its process parameters as a biological determinant and PAI as a sensitive tool for biomaterial evaluation.

Introduction

Titanium and its alloys are widely applied in dental, orthopedic, and craniofacial implantology due to their excellent combination of mechanical stability, corrosion resistance, and biocompatibility (1, 2). These favorable characteristics are largely attributed to the spontaneous formation of a stable titanium oxide layer, which protects against corrosion and reduces adverse host-tissue interactions. However, the long-term performance of titanium implants is not determined solely by bulk material properties but is strongly influenced by their surface characteristics at the micro- and nanoscales. Among these, surface roughness has emerged as a critical parameter, as it affects protein adsorption, immune cell adhesion, and the cascade of biological processes that regulate tissue integration (3, 4). Implants with moderately roughened surfaces typically achieve superior osseointegration compared to smooth or excessively rough surfaces (5, 6). Despite extensive research, the immunological mechanisms by which surface roughness directs tissue responses remain incompletely understood.

In addition to conventional surface modifications, additive manufacturing (AM), particularly selective laser melting (SLM), has become a transformative technology in titanium implant design. SLM builds components layer by layer by selectively melting fine titanium powders, thereby enabling not only the fabrication of complex, patient-specific geometries but also the intrinsic adjustment of surface roughness, porosity, and microstructure (7, 8). Recent work has highlighted that AM Ti-6Al-4V alloys can be precisely characterized and modified to enhance both their mechanical properties and their biological performance, underscoring the potential of AM as a basis for next-generation implant surfaces (9). Moreover, AM-derived surfaces have been shown to actively modulate immune responses, influencing macrophage polarization, cellular infiltration, and overall integration (8, 10-12). Thus, AM should be regarded not merely as a manufacturing technique but also as a determinant of host-implant interactions that warrants systematic investigation.

The implantation of any biomaterial triggers a foreign-body reaction, a dynamic process involving both innate and adaptive immune cells (8, 13). Immediately after implantation, plasma proteins rapidly adsorb to the material surface, forming a provisional matrix that guides early cellular interactions. Neutrophils and monocytes dominate the acute inflammatory phase, while macrophages become the central regulators at later stages. Depending on their polarization state, macrophages can sustain inflammation and fibrotic encapsulation or, alternatively, drive angiogenesis and regenerative remodeling (13, 14). Lymphocytes, granulocytes, and multinucleated giant cells (MGNCs) also participate in shaping the local environment, emphasizing the multifactorial nature of host-implant interactions (15, 16).

Despite this complexity, most histological assessments of titanium surfaces remain descriptive or focus on single immune cell types, thereby overlooking subtle but biologically relevant differences. This limitation is particularly relevant when analyzing AM-derived topographies, where the immunomodulatory potential may manifest in graded rather than binary patterns. As highlighted by Ratner and Bryant (17), conventional biocompatibility testing frameworks often fail to capture the full dynamics of host-material interactions. To address these shortcomings, integrative evaluation strategies have been developed that combine multiple histological parameters into weighted indices. In parallel, emerging computational approaches such as machine learning have shown promise in identifying macrophage response patterns to titanium surfaces, further underscoring the need for multidimensional assessment tools (18, 19). Such methods allow a more comprehensive evaluation of the phagocytic response and may reveal otherwise hidden differences in how surface characteristics, including those generated by AM, shape the host immune reaction.

The present study was therefore designed to evaluate titanium disks with defined levels of surface roughness, produced by AM, in a standardized in vivo setting. Classical histopathological parameters were complemented by use of an integrative phagocyte accumulation index (PAI). This dual approach was intended to overcome the limitations of purely descriptive analyses and to provide deeper insights into the relationship between AM-derived titanium surfaces and phagocyte-driven responses. Moreover, the PAI provides a novel, reproducible, quantitative measure of phagocyte-material interaction that is not captured by conventional ISO-based histopathology. This principle may also extend to other implant materials, including biodegradable materials and surface coatings, where the extent and dynamics of cellular degradation processes must be carefully quantified to assess both functionality and safety.

Materials and Methods

Sample manufacturing. The samples were designed as disc-shaped specimens with a diameter of 12 mm and a thickness of 2 mm. This geometry was selected to provide a large specific surface area. The specimens were manufactured using SLM on a Lasertec 12 SLM system (DMG MORI Co., Ltd., Nagoya, Japan) equipped with a fiber laser (wavelength 1070 nm, continuous wave operation, minimum spot diameter 35 µm). New Ti-6Al-4V Grade 23 powder (ECKART TLS GmbH, Hartenstein, Germany) was used as the feedstock material. The powder exhibited a nearly spherical particle morphology with a particle size distribution ranging from 20.0 µm to 53.0 µm. All specimens were fabricated in an upright build orientation (Figure 1). A uniform layer thickness of 0.03 mm and a hatch distance of 0.08 mm were applied for all samples. To generate three different levels of surface roughness, R1 (medium roughness), R2 (high roughness) and R3 (low roughness), different contour processing parameters were employed (Table I). After fabrication, the specimens were cleaned in an ultrasonic bath with isopropanol for 20 min to remove residual and loosely adhered powder particles. In accordance with ISO 20160 requirements for medical implants, heat treatment was applied to reduce internal stress and to refine the microstructure. For this purpose, the samples were heat treated at 1,050°C for 4 h.

Figure 1.

Additively manufactured specimens.

Table I. Process parameters for manufacture of specimens.

Scanning electron microscopy (SEM). SEM was performed using a TESCAN MIRA3 XMU (Tescan Group, a.s., Brno, Czech Republic). High vacuum mode was used with 13-kV acceleration voltage. All samples were sputter-coated with 2-nm gold layer before observation of surface morphology.

Profilometry. To ensure representative sampling and account for process-related variability, surface roughness was analyzed on 10 heat-treated AM specimens per surface-roughness group, with measurements performed at two defined locations per specimen using laser scanning microscopy, in order to investigate how process parameters influence surface topography. Measurements were taken using a Zeiss LSM 900 KMAT system (Carl Zeiss AG, Oberkochen, Germany) at 10× magnification with a vertical step size of 0.1 µm. The specimens were aligned consistently throughout all measurements. Additionally, the working distance and vertical step size were carefully adjusted for each individual measurement to ensure optimal conditions. Raw measurement data were processed using ZEISS ConfoMap software (version 10.1). Noise reduction and artefact removal were applied prior to extracting surface roughness parameters. The following parameters were measured by profilometry: (i) the root mean square height (Sq), representing the standard deviation of all measured height values on the surface and serving as an indicator of the mean roughness; (ii) the maximum peak height, corresponding to the highest measured point above the mean surface plane; (iii) the maximum pit depth, representing the deepest measured point below the mean surface plane; (iv) the maximum height of the surface, defined as the total vertical range from maximum peak height to maximum pit depth; and (V) the arithmetical mean height, quantifying the mean absolute deviation of all heights from the mean surface plane of titanium specimens. The root mean square height, providing the most direct measure of overall surface roughness, was assessed across all 10 specimens per group. The remaining parameters were determined once for one representative specimen per group.

In vivo study, histology and analysis. In vivo implantation and histopathological evaluation were conducted according to previously established protocols described by Barbeck et al. (21). The animal study was approved by the Veterinary Directorate of the Ministry of Agriculture, Forestry and Water Management of the Republic of Serbia and Ethical Committee for the Protection of the Welfare of Experimental Animals at the Faculty of Medicine, University of Niš, Serbia (decision number 323-07-01762/2019-05/9, date: 01 March 2019).

In accordance with DIN EN ISO 10993-6, the AM titanium discs with differing surface roughness were subcutaneously implanted individually, one disc per animal, into the rostral subscapular region of 30 randomly assigned Wistar rats, i.e., five experimental animals were assigned to each study group per time point (20). Animals were euthanized at 10 and 30 days post-implantation. Following euthanasia of the animals at the afore-mentioned time points, the implants with attached tissue were explanted and then dispatched from the surrounding tissue, so that only soft tissue itself was further histologically processed for histopathological analysis. Afterwards, the tissues were fixed in 4% neutral buffered formalin, dehydrated through graded alcohols, cleared in xylene, embedded in paraffin, and sectioned at 3-5 µm using a rotation microtome (Slee cut 6062; SLEE Medical, Mainz, Germany). Sections were stained with hematoxylin and eosin.

Histological evaluation focused on biocompatibility, following the scoring criteria of DIN EN ISO 10993-6:2017, as already described in detail by Barbeck et al. (21). Parameters included the presence of granulocytes, lymphocytes, plasma cells, macrophages, and MNGCs, as well as tissue reactions such as fibrosis, vascularization, and adipose infiltration.

The respective scores for each section were then calculated and statistically compared.

Cell classification and counting rules. As already defined in the DIN EN ISO 10993-6 scoring system, six non-overlapping high-power fields (HPFs: 400× magnification) along the biomaterial-tissue interface were chosen per slide using a systematic grid scheme. The region of interest was defined as the first 10 µm of peri-implant tissue adjacent to the biomaterial surface, including pore walls of porous materials. Cells were classified as adherent if their membrane was in direct contact with the biomaterial or if clear spreading (filopodia, lamellipodia) across the surface was evident. Only two phagocyte classes were considered:

• Macrophages: Mononuclear phagocytes with direct contact with the biomaterial surface.

• MNGCs: Phagocytes with ≥3 nuclei in direct contact with the biomaterial surface.

Cells not in contact with the surface, or separated by a visible gap >1 µm, were excluded.

Primary metrics (per HPF). The following scoring parameters were included in the newly developed system:

• Surface coverage (%): Proportion of the interface length covered by adherent phagocytes.

• Mononuclear fraction (%): Percentage of adherent phagocytes that were mononuclear macrophages.

• MNGC fraction (%): Percentage of adherent phagocytes that were MNGCs.

• Morphology/spreading: Qualitative assessment of cell shape and spreading behavior.

The complete scoring rubric is shown in Table II, by which scores of 0-4 were assigned for each metric based on the thresholds shown. These categories allowed normalization of quantitative and qualitative features for the subsequent calculation of the PAI.

Table II. Scoring rubric for phagocyte adherence at biomaterial surfaces.

MGNC: Multinucleated giant cell.

PAI. The host immune response to biomaterials critically influences the long-term success of implants. While established standards such as DIN EN ISO 10993-6 focus on histopathological evaluation of peri-implant tissue reactions, they do not provide a standardized method to quantify direct phagocyte adherence to implant surfaces. Given that macrophages and MNGCs at the immediate biomaterial interface modulate inflammation, remodeling, and eventual tissue integration, a PAI, a reproducible scoring system to quantify surface-adherent phagocytic cells (macrophages, MNGCs), was newly developed for this study by Mike Barbeck based on biological relevance and previous experience with ISO-based scoring systems. This scoring system complements conventional assessment by enabling comparative immunomodulatory readouts of surface modifications and refining links between surface design and host response.

For the PAI, the individual primary metrics were weighted according to their biological relevance and reproducibility:

• Surface coverage (weighted by 0.35): Considered the most critical parameter, as it directly reflects the extent of biomaterial surface colonization by phagocytes.

• Mononuclear fraction vs. MNGC fraction (each weighted by 0.25): These two parameters represent distinct but equally important aspects of the foreign-body reaction. The proportion of mononuclear macrophages is indicative of an acute, potentially pro-regenerative response, while the fraction of MNGCs reflects the transition toward a chronic, degradative stage.

• Morphology (weight=0.15): Provides additional qualitative information on cell spreading and epithelioid transformation. Due to the subjective nature of morphological grading and lower reproducibility compared to quantitative parameters, this metric was assigned a lower weight.

This weighting scheme balances objective quantitative metrics with relevant qualitative descriptors, ensuring that the index captures both the magnitude and the quality of the phagocytic response at the biomaterial surface.

The PAI was thus calculated as:

PAI = (0.35×coverage score) + (0.25×mononuclear score) + (0.25×MNGC score) + (0.15×morphology score)

Scores were averaged per HPF, section and specimen, and the results are reported as the mean ± standard deviation for each experimental group.

This scoring system provides a standardized and reproducible means of quantifying the adherence of phagocytes, macrophages and MNGCs to biomaterial surfaces. It complements established ISO-based evaluations by focusing on the direct cell-surface interaction, thereby enabling comparative assessment of surface modifications and their influence on the early and chronic phases of the foreign body reaction.

Statistics. Statistical analyses were performed using GraphPad Prism 10.6.1 (GraphPad Software Inc., La Jolla, CA, USA). Normality of data distribution was assessed using the Shapiro-Wilk test. For group comparisons, one-way analysis of variance was applied, followed by Tukey’s multiple comparisons post hoc test. Data are presented as mean ± standard deviation. Differences were considered statistically significant at p≤0.05.

Results

SEM. Distinct surface morphologies of the three titanium disc types were revealed by SEM, showing distinct nominal roughness levels across the R1, R2 and R3 groups (Figure 2). At low magnification, all samples showed a uniform circular disc geometry (Figure 2A). Intermediate magnification demonstrated a gradual increase in the density and size of topographical features from the low roughness (R3) through the medium roughness (R1) to the high-roughness (R2) surfaces (Figure 2B). While the surface of R3 appeared relatively smooth with only minor irregularities, the R1 disks exhibited more pronounced depressions and elevated areas distributed across their surface (Figure 2B). Samples of R2 disks displayed the most complex microtopography, characterized by numerous pits, ridges, and terrace-like structures (Figure 2C). Figure 3 provides a color-coded visualization of the different surface roughness levels.

Figure 2.

Scanning electron microscopy images showing the different views of the surfaces of the additively manufactured specimens of three types, R1-R3. Each row shows the same specimen of roughness type across three scales. A: Macroscopic overview of the disc geometry (scale bar: 10 mm). B: Medium magnification, highlighting the distribution and size of surface texture elements (scale bar: 500 μm). C: High magnification showing detail of the microtopography (scale bar: 50 μm), with pits, rims and terrace-like features that distinguish the levels of roughness.

Figure 3.

Surface roughness measured with laser scanning confocal microscopy at 20× magnification. A: Medium roughness (R1); B: high roughness (R2); C: low roughness (R3); scale bar: 500 μm.

As is typical for AM components, the as-built surfaces exhibited an inherently rough morphology, characterized by adhered features such as partially melted powder particles and visible melt pool tracks. The resulting surface topography was strongly governed by the applied contour process parameters (Table I) and led to different melting behaviors of the Ti-6Al-4V powder. The lowest energy input, a laser power of 40 W (R2), caused insufficient melting, leading to an increased presence of partially sintered and incompletely fused powder particles and, accordingly, the highest surface roughness (Sq=19.28 µm). In contrast, the higher energy inputs of 120 W (R3) and 200 W (R1) promoted more complete melting of the powder particles, resulting in markedly lower and closely comparable roughness values (Sq = 14.16 µm and 14.68 µm, respectively) with R3 yielding the lowest overall roughness.

Profilometry. To complement the qualitative SEM observations, the surface roughness of the AM titanium specimens was quantified by optical profilometry (Table III). Five height parameters were evaluated: the arithmetical mean height, the root mean square height, the maximum peak height, the maximum pit depth, and the maximum height of the surface.

Table III. Overview of the roughness parameters measured by profilometry for the additively manufactured titanium specimens.

Sa: Arithmetical mean height; Sp: maximum peak height; Sq: root mean square height; Sv: maximum valley depth; Sz: maximum height.

With regard to the mean height parameters, R2 specimens exhibited the highest roughness, R3 the lowest, and R1 showed intermediate values (Figure 4). Sq values differed by approximately 2-5 µm between groups. The maximum peak height varied more distinctly between groups, being lowest for R1 (60.16 µm) and highest for R2 (79.12 µm), with R3 showing an intermediate-to-high value (75.10 µm), indicating that the tallest surface asperities were more pronounced in R2 and R3 than in R1. In contrast, the maximum pit depth (Sv) and the total vertical range (Sz) were both highest for R3, while R2 showed the lowest pit depth and R1 the lowest total vertical range, indicating that the low-roughness R3 specimens nonetheless exhibited the deepest valleys and the greatest overall vertical extent of the surface profile.

Figure 4.

Surface roughness as shown by the root mean square height (Sq) of additively manufactured titanium specimens across the three surface-roughness groups (medium, R1; high, R2; and low, R3). Data are presented as the mean ± standard deviation. ***Significantly different at p≤0.001.

In summary, R2 displayed the highest average roughness, R3 the lowest and R1 occupied an intermediate position, with the latter also showing a less pronounced topography in terms of valley depth and total height. Statistically significant differences (p<0.001) in surface roughness (Sq) were detected across all three fabricated surface types, with Tukey’s post hoc tests revealing distinct values across the three roughness levels. Nevertheless, when considering the full range of Sq values measured across all specimens per group, a substantial overlap in the distribution of roughness values was observed.

Together, these qualitative SEM observations, corroborated by quantitative profilometric analysis, confirm that the applied surface treatments successfully generated reproducible and distinct roughness profiles on titanium disks, which are expected to differentially influence subsequent biological interactions.

Histopathological analysis. Histological evaluation revealed distinct host tissue responses to the titanium specimens of different surface roughness at day 10 post-implantation (Figure 5).

Figure 5.

Histological evaluation of tissue reactions to titanium disks with different surface roughness (R1-R3) at 10 days after implantation into Wistar rats; hematoxylin and eosin staining. Images of the implant-tissue interface for specimens with medium (A: R1), high (B: R2), and low roughness (C: R3) at 100× (upper panels) and 400 × (lower panels). The titanium disks are surrounded by connective tissue (CT). Higher magnification of the interface region shows macrophages (black arrows), lymphocytes (yellow arrows), granulocytes (green arrows), blood vessels (red arrows), and reactive tissue zones (double-ended arrows). The R1 specimen shows moderate infiltration with macrophages and scattered lymphocytes and granulocytes, together with limited reactive tissue and few vascular structures. The R2 specimen exhibits denser infiltrates with increased numbers of immune cells, pronounced vascularization, and a broader zone of reactive tissue. The R3 specimen displays compact and homogeneously distributed infiltrates dominated by macrophages, with interspersed lymphocytes and granulocytes as well as clearly discernible areas of reactive tissue. MT: Muscle tissue. Scale bars: 100 μm (upper panel), 50 μm (lower panel).

In the medium-roughness group (R1), the implant-tissue interface appeared relatively well demarcated, with moderate infiltration of mononuclear cells and a limited number of MNGCs (Figure 5A). The surrounding connective tissue exhibited an organized architecture with only sparse neovascularization.

In contrast, animals with implants with high roughness (R2) displayed more pronounced cellular infiltration at the implant surface (Figure 5B). Numerous macrophages and fibroblasts were observed, accompanied by higher numbers of MNGCs and newly formed blood vessels. The connective tissue in this group appeared more actively remodeled, indicating an intensified foreign body response compared to R1.

Specimens with low roughness (R3) demonstrated dense and homogeneously distributed cellular infiltrates (Figure 5C). Mononuclear cells were abundant, and MNGCs were present in moderate numbers. The connective tissue surrounding these implants appeared compact and well-organized, suggesting a balanced but active foreign body reaction.

At day 30 post-implantation, histological analysis demonstrated distinct host responses around the different titanium specimens (R1-R3), including varying degrees of fibrotic encapsulation (Figure 6).

Figure 6.

Histological evaluation of tissue responses to titanium disks with different surface roughness after implantation into Wistar rats. Overview images of specimens with medium (A: R1), high (B: R2), and low roughness (C: R3) at 100× (upper panels) and 400 × (lower panels), showing the implant-tissue interface and adjacent tissue structures. Higher magnification of the interface region shows macrophages (black arrows), lymphocytes (yellow arrows), granulocytes (green arrows), blood vessels (red arrows), and fibrotic/reactive tissue zones around the implants (double-ended arrows). The R1 specimen shows moderate macrophage infiltration within a broad fibrotic layer containing scattered lymphocytes and blood vessels. The R2 specimen exhibits a dense fibrotic capsule rich in macrophages, lymphocytes, and granulocytes, with extensive neovascularization. The R3 specimen shows a thinner but compact fibrotic layer dominated by macrophages and lymphocytes with moderate vascularization. CT: Connective tissue; MT: muscle tissue. Scale bars: 100 μm (upper panel), 50 μm (lower panel).

In the R1 group (Figure 6A), the implant-tissue interface was characterized by a relatively broad and dense fibrous layer separating the implant. Cellular infiltration was moderate, consisting predominantly of macrophages with scattered lymphocytes and occasional blood vessels.

In the R2 group (Figure 6B), the most pronounced inflammatory and fibrotic reaction was observed. A dense fibrous capsule with abundant macrophages, lymphocytes, and scattered granulocytes surrounded the implant. Numerous blood vessels penetrated the reactive tissue zone, reflecting active tissue remodeling and angiogenesis.

In the R3 group (Figure 6C), an intermediate response was found, with a thinner but compact fibrotic layer between the implant and connective tissue. Infiltration consisted mainly of macrophages and lymphocytes, with fewer granulocytes and moderate vascularization.

Analysis of biocompatibility scoring. Semi-quantitative histopathological evaluation demonstrated that macrophages represented the dominant immune cell type across all groups and both time points, with consistently high scores independent of surface roughness (Figure 7). Lymphocyte infiltration was low and stable at both times. Polymorphonuclear cells were also present at low levels at day 10, with a significant (p≤0.05) increase to moderate levels in the R2 group after 30 days compared with R1 at day 10 and R3 at day 10. Plasma cells and giant cells were rarely observed with scores remaining minimal in all groups and at both time points. Necrosis scores were likewise low across all groups and time points, confirming the overall biocompatibility of the titanium implants.

Figure 7.

Semi-quantitative histopathological scoring of tissue reactions to titanium disks at 10 and 30 days after implantation into Wistar rats. Macrophages were the predominant immune cells (D), while lymphocyte infiltration was low and stable over time (B). Polymorphonuclear cells were present at low levels at day 10, with a significant increase to moderate levels in the R2 group at day 30 (A). Plasma cells (C) and giant cells (E) were minimal in all groups. Necrosis scores were low (F) and fatty infiltrates were minimal to absent across all groups and time points (I). Neovascularization (G) and fibrosis (H) showed a tendency to increase over time, particularly for rougher surfaces (R2, R3); fibrosis ranged from low to moderate levels. Data are presented as the mean ± standard deviation. *Significantly different at p<0.05.

Regarding tissue remodeling, neovascularization was present at low levels and showed a tendency to increase over time, particularly in the R2 group and the R3 group after 30 days, although this trend did not reach statistical significance. Fibrotic encapsulation was evident around all implants, with the highest scores observed for the R3 group at 30 days, followed by the R2 group, while the R1 group generally exhibited the lowest degree of fibrosis. This difference reached statistical significance only between the R2 group at day 10 and the R3 group at day 30 (p≤0.05), while all other pairwise comparisons of fibrosis scores were non-significant. Fatty infiltrates were minimal to absent in all specimens.

In summary, the histological scoring revealed a macrophage-dominated tissue response that was accompanied by stable lymphocytic infiltration and progressive fibrotic remodeling. R2 and R3 surfaces were associated with stronger fibrotic reactions and slightly enhanced angiogenesis, whereas the medium-roughness surface (R1) elicited a weaker overall response. Statistical comparison across all groups and time points identified two isolated significant differences; an increase in polymorphonuclear cell infiltration in the R2 group at day 30 and a difference in fibrosis scores between R2 at day 10 and R3 at day 30, while all remaining comparisons were non-significant.

PAI. The calculation of the PAI, which integrates phagocyte coverage, mononuclear infiltration, multinucleated giant cells, and morphological activation into a weighted score, revealed distinct differences between the groups (Figure 8). At day 10, medium-rough titanium surfaces (R1) exhibited the lowest PAI (1.48), reflecting limited macrophage and granulocyte infiltration. In contrast, high roughness (R2) and low roughness (R3) specimens showed higher values (2.10 each), indicating stronger phagocyte accumulation and broader coverage of the implant surface compared to R1.

Figure 8.

Phagocyte accumulation index (PAI) for titanium disks with different surface roughness (R1: medium roughness, R2: high roughness, R3: low roughness) at 10 and 30 days after implantation into Wistar rats. R1 showed the lowest PAI values at both time points, while R2 and R3 showed numerically higher values. At day 30, R2 reached a significantly higher PAI than R1 (*p<0.05, one-way ANOVA with Tukey’s post-hoc test); all other pairwise comparisons, including R2 versus R3, were not statistically significant. Values represent the mean ± standard deviation of 5 animals per group and time point (6 high-power fields evaluated per specimen).

At day 30, the responses diverged further numerically. R1 remained at a similarly low PAI of 1.48, consistent with a weak and stable foreign-body reaction. By contrast, R2 showed the highest PAI (2.48), which was significantly higher than that of R1 at day 30 (p≤0.05), consistent with a pronounced phagocytic response accompanied by intensified cellular activation. R3 also displayed an elevated PAI (2.35), although lower than for R2 and not statistically significant compared with R1; the infiltrates in R3 appeared more compact and homogeneously organized.

Taken together, these findings show that medium-roughness surfaces (R1) elicited the lowest phagocyte activation at both day 10 and day 30, whereas both higher-roughness (R2) and lower-roughness (R3) surfaces were associated with numerically greater phagocyte accumulation. Among these, the highest-roughness surfaces (R2) induced the strongest response, reaching a significantly higher PAI than R1 at day 30 (p≤0.05), while low-roughness surfaces (R3) resulted in an elevated but statistically non-significant response relative to R1.

Discussion

This study evaluated how different levels of surface roughness in AM titanium specimens influence the in vivo foreign-body response, combining traditional ISO-based histopathology with a novel PAI.

SEM and profilometry confirmed surface morphologies typical after SLM, including partially melted particles and sintered satellites. The measured arithmetical mean height values (14-18 µm) fall within the typical as-built roughness range reported for SLM-manufactured Ti-6Al-4V surfaces. These observations build upon the work by Kök et al., who characterized and modeled AM Ti-6Al-4V surfaces and demonstrated that the AM process itself creates microstructural and surface-chemical features influencing biological interactions (9). Their results underscore that AM is not merely a fabrication route but represents a biologically relevant surface variable influencing implant performance.

Barbeck et al. have repeatedly emphasized in recent editorials and reviews the importance of surface treatment and topography in the translational development of tailored implants (22-24). In particular, Barbeck et al.’s editorial discusses how advanced processing and surface modifications (including AM) are central to controlling immune engagement and tissue integration (24). This perspective supports the concept that the intrinsic microtopography generated by AM must be considered an active variable in the host response rather than a passive fabrication artifact.

The biocompatibility analysis based on the DIN EN ISO 10993-6 assessment revealed that all specimens elicited only mild to moderate inflammatory reactions and fulfilled the biocompatibility criteria, confirming the known tolerability of titanium in vivo (24, 25). However, ISO-based scoring did not discern any significant intergroup differences, which highlights a recognized limitation: conventional histopathology often lacks the sensitivity to capture subtle immunological variations, especially among advanced biomaterial surfaces (26, 27). Earlier studies with titanium surfaces have similarly reported good gross biocompatibility but failed to resolve differences in inflammatory cell behavior using standard scoring (28, 29).

Moreover, surface roughness measurements were performed per material group. Significantly different roughness values were detected for all three fabricated surface types revealing distinct mean root mean square height values across the three roughness levels. Nevertheless, substantial overlap in the distribution of roughness values was evident, indicating inherent variability in surface topography that may not be fully captured by the semi-quantitative nature of the ISO scoring system. These findings suggest that, while the manufacturing process successfully generated distinct surface topographies, the current ISO-based assessment is insufficient to translate these physical differences into biologically meaningful distinctions. A more detailed analysis of the local cellular tissue response is therefore required to enable a more differentiated evaluation among the three production groups.

The histopathological analysis documented consistent fibrosis and macrophage presence, with low to moderate levels of foreign-body giant cells (MNGCs). Traditionally, MNGCs are regarded as a hallmark of frustrated phagocytosis and chronic reaction (30). However, mechanistic reviews emphasize that MNGC formation depends on multiple surface cues, including texture, chemistry, and fusogenic signaling pathways (31, 32). The relatively limited presence of MNGCs in this study suggests that the AM-derived surfaces did not strongly stimulate fusion, favoring instead a predominantly mononuclear macrophage response. This is in line with an investigation indicating that not all rough surfaces provoke extensive MNGC formation (33).

The integrative nature of the PAI, combining phagocyte coverage, infiltration, MNGC presence, and morphological activation, provides a higher discriminatory resolution than DIN EN ISO 10993-6 scoring. The significantly higher PAI observed for the high roughness group (R2) at day 30 demonstrates that even moderate topographical differences can alter phagocyte accumulation behavior. This aligns with established evidence that surface roughness modulates cellular adhesion and immune response dynamics (34, 35). Importantly, given the relatively low MNGC counts, much of the PAI difference was attributable to mononuclear macrophage responses rather than giant-cell formation. Such integrated approaches reflect emerging strategies in biomaterials research, where digital image analysis, artificial intelligence, and weighted scoring systems are increasingly applied to unravel complex host-material interactions (36-38). The PAI contributes to this methodological evolution by quantifying phagocyte-driven activity as a unified metric.

The broader utility of the PAI extends to biodegradable implants and surface coatings. In such systems, phagocytes not only mediate inflammation but also regulate degradation and tissue regeneration. Excessive phagocytic activity may accelerate premature resorption and compromise mechanical integrity, whereas insufficient activity may delay remodeling (39, 40). Therefore, the PAI may support the rational design of biomaterials with controlled degradation profiles. Barbeck et al. have similarly underscored the need for immunomodulatory surface strategies in next-generation implants (24), reinforcing that quantitative metrics such as the PAI may help bridge translational gaps.

The relatively small difference in roughness between specimen groups (<5 µm difference in arithmetical mean height) may have limited the magnitude of biological contrasts observed. Future studies should therefore investigate broader roughness gradients (e.g., 10-30 µm differences) to clarify potential threshold effects. Furthermore, additional characterization of macrophage phenotype and MNGC activity using immunohistochemical markers – e.g., cluster of differentiation 68 (CD68), CD163, dendritic cell-specific transmembrane protein (DC-STAMP), and tartrate-resistant acid phosphatase (TRAP) – would provide deeper mechanistic insight.

Overall, this work reaffirms that AM not only enables the creation of complex titanium geometries but also defines biologically relevant surface features. While conventional ISO scoring confirmed general biocompatibility, only the PAI revealed meaningful differences among surface types: high-roughness surfaces elicited the strongest phagocytic response, medium roughness-surfaces the weakest and low-roughness surfaces elicited an intermediate response pattern. The presence of occasional MNGCs suggests a controlled foreign-body reaction rather than chronic inflammation. The PAI therefore emerges as a methodological advancement for implant pathology, with potential applicability to both inert and degradable biomaterials.

Conclusion

Additively manufactured titanium surfaces were biocompatible in vivo, eliciting only mild to moderate tissue reactions. Conventional ISO 10993-6 scoring was unable to distinguish between implants with different surface types, whereas the PAI revealed significant differences, with high-roughness surfaces inducing the strongest phagocyte response. The low frequency of MNGCs suggests a controlled rather than chronic reaction. These findings underline the dual role of AM as both a fabrication method and a determinant of biological response and establish the PAI as a valuable tool for future biomaterial evaluation.

Conflicts of Interest

All the Authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Authors’ Contributions

Conceptualization: M.B., O.J., S.N., S.K. and F.W. Resources: M.B., O.J., S.N., S.K. and F.W. Data curation: S.P., J.S., A.J., S.S. and M.B. Writing – original draft preparation: M.B., S.P., S.S., A.J. and S.S. Writing – review and editing: M.B., S.P., C.R., O.J., S.N., S.K. and F.W. Visualization: S.P., J.S., A.J., S.S. and M.B. Funding acquisition: M.B., O.J., S.N., S.K. and F.W. All Authors have read and agreed to the published version of the manuscript.

Acknowledgements

The Authors gratefully acknowledge the funding by the German Research Foundation (Deutsche Forschungsgemeinschaft, DFG) for the research unit 5250 ”Mechanism-based characterization and modeling of permanent and bioresorbable implants with tailored functionality based on innovative in-vivo, in-vitro and in-silico methods” (project no. 449916462) and the Ministry of Culture and Science of North Rhine-Westphalia (Ministerium fuer Kultur und Wissenschaft des Landes Nordrhein-Westfalen, NRW) within the Major Research Instrumentation Program for the “CLSM-AFM” (project no. 544699841).

Artificial Intelligence (AI) Disclosure

During the preparation of this manuscript, a large language model (Microsoft Copilot) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the Authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.