Effect of Surface Topography on In Vitro Osteoblast Function and Mechanical Performance of 3D Printed Titanium

Abstract

Critical-sized defects remain a significant challenge in orthopaedics. 3D printed scaffolds are a promising treatment but are still limited due to inconsistent osseous integration. The goal of the study is to understand how changing the surface roughness of 3D printed titanium either by surface treatment or artificially printing rough topography impacts the mechanical and biological properties of 3D printed titanium. Titanium tensile samples and discs were printed via laser powder bed fusion. Roughness was manipulated by post-processing printed samples or by directly printing rough features. Experimental groups in order of increasing surface roughness were Polished, Blasted, As Built, Sprouts and Rough Sprouts. Tensile behavior of samples showed reduced strength with increasing surface roughness. MC3T3 pre-osteoblasts were seeded on discs and analyzed for cellular proliferation, differentiation, and matrix deposition at 0, 2, and 4 weeks. Printing roughness diminished mechanical properties such as tensile strength and ductility without clear benefit to cell growth. Roughness features were printed on mesoscale, unlike samples in literature in which roughness on microscale demonstrated an increase in cell activity. The data suggest that printing artificial roughness on titanium scaffold is not an effective strategy to promote osseous integration.

Introduction:

Unlike most human tissues, bone has an innate ability to heal.⁽¹⁾ Bone is constantly undergoing remodeling in response to stress. However, there is a limit to the regenerative capacity of bone. Here we define the limit of bone’s regenerative capacity as a critical-sized defect (CSD), or the minimum amount of bone injury at which bone will no longer heal on its own despite appropriate alignment and fixation.⁽²⁾ In general, the literature defines CSD as bone loss greater than 1-2 cm in length or greater than 50% loss in circumference of bone.⁽³⁾ These large defects generally result from trauma, oncological resection, infection, nonunion after fracture, or congenital anomalies.⁽⁴⁾ The prognosis for patients with a CSD is poor, often undergoing multiple surgical procedures in an effort to reconstruct the defect, or amputation in severe cases.⁽⁵⁾ Further research is necessary to tackle this unmet need in orthopedics to find solutions which can restore the defect and integrate with native bone for long-term clinical success.

Autograft, which is bone harvested from the same patient from a different site, is considered the gold standard for the treatment of a CSD. However, autografts have drawbacks which limit their clinical use including limited size, donor site morbidity, need for intraoperative modifications to fit the defect, and issues with integration of a different bone architecture from a distant donor site.⁽⁶⁾ Allograft, which is disease-screened bone harvested from a deceased donor, has been an alternative for patients with contra-indications to autografts. However, allografts also come with a host of drawbacks including limited availability, poor revascularization, risk of transplant rejection, infection, and long term mechanical failure.⁽⁶⁾

Titanium has been frequently used in orthopaedics as an alternative to autografts and allografts. Unlike autografts and allografts, titanium does not require harvesting nor lead to immunological reaction, and its mechanical properties provide immediate structural support.⁽⁷⁾ The use of titanium implants have been limited by an increased risk for bacterial infections, and the metals mismatch in stiffness to native bone results in stress shielding and potential aseptic loosening of the implant.^(8),(9) With advances in 3D printing technology, porous titanium scaffolds now provide a unique alternative to allografts and autografts for the treatment of CSDs. Porous titanium scaffolds are often manufactured by laser powder bed fusion (L-PBF). L-PBF creates three-dimensional parts in a layer by layer process consisting of distribution of a layer of powder, typically 20-100 μm thick, which is selectively melted by a laser energy source in 2D layer. The process is repeated layer by layer to build up the final part which can have complex geometry due to the spatial control afforded by L-PBF.⁽¹⁰⁾ Utilizing L-PBF techniques enables fabrication of complex porous architectures which were not previously possible with traditional molding or scaffold manufacturing methods. Additionally, using 3D printing enables creation of implants with anatomically matched geometry for a specific patient. This combination allows for the potential integration of surrounding bone tissue to repair CSDs. Porous titanium scaffolds have already proven clinically successful in spinal fusion.^(11),(12) In foot and ankle surgical procedures, porous titanium scaffolds are used to treat failed fusion surgeries, nonunion, and traumatic bone loss.⁽¹³⁾

However, 3D printed titanium implants are not yet optimized. In a subset of cases, implants need to be removed due to infection or poor osseous integration where surrounding bone does not grow onto or through the scaffold. There is no one clear reason for poor osseous integration and is likely due to a multitude of factors including implant-related factors (e.g. implant fit, porous architecture, surface roughness and chemistry, pore size, pore geometry), patient specific factors (e.g. diabetes, smoking status, vascular disease, underlying infection), and surgical technique (e.g. osteotomies, use of biologics, additional fixation methods).⁽¹⁴⁾

One hypothesized factor impacting osseous integration of titanium implants is surface roughness. By focusing on surface roughness, this study evaluates the effect of surface roughness while minimizing confounding variables. In vitro studies have demonstrated that increasing the surface roughness of 3D printed titanium discs with acid etching between R_a values of 0.29 – 3.26 significantly increased proliferation and activity of osteoblasts.⁽¹⁵⁾ Micro and nano roughness is theorized to impact osseointegration by providing local asperities for osteoblast attachment, or by influencing the adsorption of proteins on the surface, inducing the expression of certain integrins, such as α₂β₁, which in turn directly regulates osteoblast differentiation.^(16)-(18) However, there is a tradeoff between implant surface roughness and mechanical performance. Increased surface roughness introduces surface mediated crack initiation sites, reducing strength and fatigue life of 3D printed as well as traditionally manufactured titanium implants.^(19),(20) Intentionally incorporating roughness on titanium implants for improved friction and osseointegration comes at a cost to mechanical behavior as the additional topography creates stress concentrations that can lead to fatigue of the construct.⁽²¹⁾ Furthermore, in addition to theoretically enhancing osteoblast activity, roughness has been shown to increase the potential for bacterial biofilm formation.⁽²²⁾ In order to engineer an optimal titanium scaffold, tradeoffs associated with modulating surface roughness must be clearly understood both in terms of biological function and mechanical integrity.⁽²³⁾ Therefore, the purpose of this study was to understand how changing the surface roughness of 3D printed titanium either by processing printed samples or artificially printing rough topography impacts the mechanical and biological properties of the titanium. We would expect that printing rough topography would enhance the proliferation and differentiation of cultured pre-osteoblasts while modestly diminishing the mechanical properties.

Materials and Methods:

Sample Fabrication

Both tensile test samples and discs for cell culture were produced via L-PBF using a 3D Systems ProX DMP320 system. Parts were produced with medical grade titanium alloy powder (Ti6Al4V ELI) with an average diameter of 35 μm, conforming with chemistry appropriate for implanted medical device per ASTM F3001. To create samples with varied surface topography, samples underwent either a post-processing surface treatment or had topographical roughness features designed and printed directly onto the surface. Post-processed samples were either micro-blasted (“Blasted”) using the same powder used in fabrication in order to not change the surface chemistry or mechanically polished (“Polished”) using successively finer grit paper and a final diamond polishing paste. Three-dimensionally printed topographical roughness features were designed to have 500 μm height as tetrahedral (“Sprouts”) or irregular (“Rough Sprouts”) extruded from the surface. Samples which underwent no post-processing or additional surface features were used as a control (“As Built”). Images of tensile samples are shown in Figure 1A.

Figure 1:

A) Titanium samples were printed via Laser Powder Bed Fusion. Samples either underwent post printing surface modification such as polishing and blasting to decrease surface roughness or had Sprouts or Rough Sprouts printed on surface to add topography. Tensile specimens with analogous surface topography were tested under tensile loading to failure. Discs from each experimental group were seeded with MC3T3 pre-osteoblast and cultured for 0,2 and 4 weeks. In vitro assays assessed for cell proliferation, differentiation, and matrix deposition at each time point. B) The left panel shows tensile specimens and cell disc samples with varied surfaces produced via post-processing, or designed meso-scale topographical features. The black scale bar represents 10 mm . The right panel shows SEM images of disc samples are at 50X. The black scale bar represents 2mm.

Roughness Measurements

Surface roughness measurements were made using an Olympus LEXT OLS4000 Confocal Microscope at 20x magnification and wavelength cutoff of 100 μm. Line measurements were taken along the long axis of tensile specimens (n=6/group). Surface roughness is reported as both R_a, the arithmetical average of absolute distances of the profile from the centerline, and R_z, the maximum peak to valley distance of the profile.

Mechanical Testing

The effects of surface topography were assessed with tensile testing. Tension is the worst case loading mode and highlights the relative difference in mechanical properties caused by difference in surface topography. ⁽¹⁸⁾

Tensile testing samples were printed in the direction of loading. Samples were loaded under tension at a constant displacement rate per ASTM E8M until failure using a Test Resources TR830 axial testing frame with 50kN capacity. Strain was captured using a video extensometer. The diameters of tensile samples were measured with digital calipers prior to testing and used to calculate strength. Yield strength was calculated using the 2% offset method.

Cell culture

MC3T3 (ATCC CRL-2593) pre-osteoblasts were obtained from Duke University’s Cell Culture Facility at passage three. MC3T3 cell line was chosen to focus on the early phase of cell progression to model osteoblast proliferation and differentiation on a Ti implant. The cells cultured to 80% confluence in culture media consisting of MEM α (Gibco) with 10% FBS and 1% P/S. Thirty-one Ti discs from each group were autoclaved at 250°F for 85 minutes and placed in a sterile 48 well plate. Wells were filled with culture media (MEM α with 10 % FBS, and 1 % P/S) to pre-wet discs. Cells were trypsinized with 0.25% Trypsin-EDTA and resuspended to an average cell density of 2.3 x 10⁴ cells/ml. The pre-wet media was aspirated from each well after twenty-four hours and a droplet of cell solution (250uL) was pipetted on each disc for cell density of 2 x 10⁴ cells / cm². Discs were incubated at 37 °C and 5% CO₂ for 5 hours to allow for cell adherence. Osteogenic media made up of MEM α with 10% FBS, 1% P/S, 5 mM Beta Glycerol Phosphate, and 50 μg / ml ascorbic acid was added to each well. Media was changed every three days. Week 0, 2, and 4 endpoints took place 1, 15, and 29 days after initial seeding respectively.

Alkaline Phosphatase assay

Supernatant was collected from each well at the defined endpoints above. Twenty-four hours before collection, media was aspirated off wells and osteogenic media (500 μL) was pipetted into each well. On day of collection, 300 μL of supernatant was pipetted into 1.5 ml tube and stored in −80 °C. Media was replaced into well and culture plate was put back into incubator. After all samples were collected and frozen, they were thawed for Alkaline Phosphatase (ALP) assay (ab83369, ABCAM). Undiluted samples (80 μL) and reagents were added to a 96 well plate per manufacturer’s instruction. The plates were incubated for 1 hour and stop solution was added to each well. Plate was shaken for 90 seconds and absorbance was measured at 405 nm using a SPARK® Multimode Microplate Reader (Tecan). ALP activity was calculated from absorbance values and standard curve using manufacturer’s instructions.

Cell Viability

Three discs from each group were placed in a new 48 well plate. Fresh osteogenic media was added to each well with MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (12mM) per Vybrant® MTT Cell Proliferation Assay Kit instructions. Plate was incubated for 6 hours and media removed from each well. Dimethyl sulfoxide (500 μL) (Sigma Aldrich) was added to each well. Plate was shaken for 90 seconds with a SPARK® Multimode Microplate Reader (Tecan). Titanium discs were removed from each well and plate was shaken for 90 seconds. Absorbance was measured at 470 nm and normalized by dividing by average absorbance calculated for As Built group.

En bloc staining

Discs used in MTT assay were placed in a new 24 well plate. Each replicate was stained with Picrosirius Red or Alizarin Red solutions. Each disc was stained with respective dye (500 μL) for 30 minutes. Dye was removed and samples were rinsed with 1x Phosphate-buffered saline (PBS) until PBS was clear. Discs were imaged with Leica M60 stereo microscope and Leica IC80 HD camera.

Fluorescent Microscopy

Discs with cells were fixed with 4% paraformaldehyde (PFA) for 20 minutes after aspirating media. Upon fixation, samples were rinsed with PBS for 30 minutes. Fresh PBS was added and samples were stored at 4 °C. PBS was removed and Hoechst dye (600 μM, Thermo Fisher) was added to each well to cover discs. Samples were incubated for 5 minutes. Hoechst dye was removed and replaced with PBS. Discs were imaged with a confocal microscope (Leica SP8 Upright Confocal) using a 405 nm Diode laser. Images were taken using a 2.5X dry objective.

Reverse-Transcription Quantitative Polymerase Chain Reaction (RT-qPCR)

RT-QPCR was performed to measure expression levels of Osteocalcin (Bglap) and Sclerostin (Sost) relative to a housekeeping gene (β-actin). Cell lysate and RNA isolation was performed using the RNeasy mini kit (Qiagen) per manufacture’s instructions. Cell lysates were collected at each time point and stored at −80°C until completion of experiment and then RNA isolated for all samples at one time. The concentration and purity of RNA was assessed using NANODROP 2000 (Thermo Scientific). RNA (250 ng) from week 2 and 4 was reverse transcribed into cDNA using iScript cDNA Synthesis Kit (Bio-Rad Laboratories) as there was not sufficient RNA for week 0 samples. RT-QPCR was perform using SsoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad Laboratories) and CFX96™ Real-Time System Thermal Cycler (Bio-Rad Laboratories) according to manufacture’s instructions with two technical replicates.

The following primers (Integrated DNA Technologies) were used in RT-QPCR reactions:

β-actin forward: 5’-AGATGTGGATCAGCAAGCAG –3’

β-actin reverse: 5’- GCGCAAGTTAGGTTTTGTCA-3’

Bglap forward: 5’-CTTGAAGACCGCCTAAC-3’

Bglap reverse: 5’-GCTGCTGTGACATCCATA-3’

Sost forward; 5’-TCCTCCTGAGAACAACCAGAC-3’

Sost reverse; 5’-TGTCAGGAAGCGGGTGTAGTG-3’

The primers spanned different exons to ensure there was no amplification of the genomic DNA. Relative gene expression compared to reference gene, β-actin, was calculated from quantitative cycles (Cq) by following formula:

$$RelativeRNAExpression=2^{C_{q_{BetaAction}}-C_{q_{Target}}}$$

Scanning Electron Microscopy (SEM)

Fixed samples were washed twice with 1x PBS for 10 minutes each. Samples were dehydrated with increasing ethanol concentrations and washed 3 times for 10 minutes each in respective concentrations (10, 30, 50, 70, 90, 100%). Samples were washed with Hexamethyldisilazane (Electron Microscopy Sciences) three times and left in the chemical fume hood until reagent evaporated. Samples were sputter coated with gold for 400 seconds at 12 v (Denton Desk V). Samples were imaged at 50x and 1000x magnifications with a Tabletop Scanning Electron Microscope (Hitachi TM3030Plus).

Statistical Analysis

One-way ANOVA was used to compare average measurements between the roughness groups. Tukey HSD test was done if one-way ANOVA indicated differences between groups for tensile testing and in vitro assay results. Student’s t-test was performed to compare means between 2 groups. Statistical analysis was conducted with RStudio. Alpha was set at 0.05.

Results:

Surface roughness and tensile testing results are summarized in Table 1. Representative stress-strain curves are shown in Figure 2A. As Built samples had an average R_a of 6.7 μm as a result of the partially adhered powder particles characteristic of the L-PBF process. Surface roughness measurements showed an increased surface roughness with the addition of surface topography features for both the sprout and rough sprout groups. Conversely, blasting and polishing of the surface reduced surface roughness, as expected. Although Polished samples had the highest average strength and elongation, no statistically significant difference was observed in the ultimate strength, yield strength, or elongation of As Built, Blasted and Polished samples (Figure 2B). However, the increased surface roughness for the spout and rough sprout samples significantly reduced ultimate strength, yield strength, and elongation at failure. The additional printed topography increased sites for crack initiation at the surface leading to reduced elongation.

Table 1:

Summary of surface roughness and tensile properties of titanium samples with varied topography.

Surface	Ra (μm)	Rz (μm)	Diameter of Tensile Sample (mm)	UTS (MPa)	Yield Strength (MPa)	Elongation at Failure (%)
Polish	0.5 (0.0)	2.6 (0.1)	5.96 (0.02)	1078.2 (11.1)	989.6 (59.4)	14.4 (1.7)
Blasted	3.2 (0.7)	22.6 (3.2)	6.09 (0.04)	1039.1 (16.5)	952.6 (38.8)	11.9 (1.0)
As Built	6.7 (1.1)	33.0 (4.5)	6.05 (0.02)	1053.7 (8.2)	992.1 (8.1)	12.6 (1.3)
Sprouts	5.8 (1.9)	41.4 (13)	6.59 (0.02)	970.0 (12.6)	907.1 (41.2)	6.7 (1.3)
Rough Sprouts	9.9 (1.8)	65.1 (14)	6.45 (0.03)	951.9 (9.8)	875.2 (12.3)	8.3 (1.2)

Figure 2:

A) Representative stress strain curves from tensile testing of samples with varying surface topography. B) Ultimate tensile strength and fracture strain for varied surfaces of Ti6Al4V samples produced via L-PBF. (1-way ANOVA and Tukey’s HSD, † denotes statistically significant difference from Polished, Blasted, and As Built, p<0.05).

ALP assay measured the activity of ALP in the supernatant of each of the discs and cell samples. One-way ANOVA showed no statistical difference in ALP activity among all groups within each time point. There was a marked decrease in ALP activity between 2 and 4 weeks across all groups. (Figure 3A). One-way ANOVA of MTT assay results did not show any statistical difference in cell metabolic activity among all groups for week 2 and week 4 time points. At the week 4 time point, the As Built group displayed statistically higher metabolic activity than Sprouts group. (Figure 3B).

Figure 3:

A) Activity of ALP in supernatant of MC3T3 osteoblasts cultured on titanium disc with varying topography and increasing roughness across groups. Results are presented as the mean ± standard error (S.E.) (n=5 per group). B) Absorbance at 540 nm reflective of MTT activity of MC3T3 osteoblast cultured on titanium disc with varying topography and increasing roughness across groups. A) Cells cultured for 0 weeks (1 day). B) Cells cultured for 2 weeks. C) Cells cultured for 4 weeks. Results are presented as the mean ± standard error (S.E.) (n=3 per group). P value from one-way ANOVA is reported in the top right corner.

RT-qPCR showed roughness had a significant effect on gene expression of Bglap (Figure 4A,C). By grouping rough and non-rough discs together, results showed that the rougher groups had a decrease in Bglap expression at both time points (Figure 4C). No difference among roughness groups was seen for Sost expression (Figure 4B,D). Fluorescent microscopy shows that cells adhered to disc surface after 1 day of culture. (Figure 5A). By week 2, the cells appeared to reach confluence. By week 4, the cultured cells formed a film over the disc (Figure 5A)). Qualitatively, the studied rough surfaces did not impact cell growth. It appears that there is less fluorescent signal from the rough groups, Sprouts and Rough Sprouts. However, the topography of disc is large enough where all the cells cannot be imaged on one focal place with confocal microscope.

Figure 4:

Relative expression of Bglap (left column) and Sost (right column) compared to β-actin of MC3T3 cells cultured on 3D printed titanium discs for 2 and 4 weeks. First row displays relative gene expression for individual groups in order of increasing roughness. Bottom rows shows relative gene expression for disc grouped based on the addition of artificial topography. Polished, Blasted and As Built groups were classified as Not Rough, and Sprouts and Rough Sprouts are classified as Rough. Results are presented as the mean ± standard error (S.E.) p value from Students T Test is reported for each comparison.

Figure 5:

A) Confocal fluorescent images of titanium discs with varying topography stained with Hoechst, a nuclear stain. White scale bar represents 500 μm. B) SEM images of titanium discs with varying topography imaged at 1000x. Black scale bar represents 100μm.

SEM images showed that the extracellular matrix formed by week 2 and had become more fibrous at week 4 (Figure 5B). Cell matrix was qualitatively more visible on the rougher titanium discs indicating that either there was more ECM deposition on rougher discs or that the ECM film on rough disc was more resistant to multiple rounds of washing needed to process SEM samples. En bloc staining of discs with Picrosirius Red to highlight collagen show ECM covering the discs by week 2 and becoming more prominent at week 4 (Figure 6A). En bloc stain with Alizarin Red show scant mineralization at week 2 and widespread mineralization at week 4 (Figure 6B). As with other imaging techniques done in this study, en bloc staining did not show clear difference between experimental groups. The collagen and calcium deposition confirm that the time points and MC3T3 cell line used in the study captured the progression of preosteoblast differentiation ending with the deposition of calcium in the extracellular matrix.

Figure 6:

Titanium discs with varying surface topography stained en bloc with A) Picrosirius Red to highlight collagen deposition and B) Alizarin Red to highlight calcium deposition.

Discussion:

While current 3D printed titanium implants have been successfully used to treat CSD, the implant fails in a subset of these patients resulting in revision surgery and potentially amputation when all other therapeutic options are exhausted. The significant morbidity of implant failure highlights the unmet need to optimize implant design parameters including roughness. In this study, surface roughness was shown to impact the tensile behavior of titanium samples produced via L-PBF. The observed reduction in ductility with increasing surface roughness for present results as well as comparable studies in the literature is shown in Figure 7. High surface roughness (in the absence of internal voids or other stress risers) leads to increased initiation of fracture from surface defects, reducing ductility as well as fatigue life of titanium.^(19),(20) Although no significant increase in strength or ductility was seen for the Blasted or Polished samples as compared to the As Built control, previous results have shown an increase in fatigue life for surface treated titanium ⁽¹⁰⁾. It should also be noted that due to the increase in sample diameter by the protrusion of the sprout and rough sprout features, the area used in stress calculations was larger than other groups. Due to lack of interconnectivity of the surface features, load bearing area was not increased. Therefore, when adding additional surface roughness to orthopedic implants it is important to understand the tradeoffs to mechanical properties of topographical features. Further investigation will include the effect of post-processing surface treatments and additional topographical features on tensile fatigue behavior.

Figure 7:

Tensile failure strain as a function of surface roughness for current experiments and results from the literature for Ti6Al4V fabricated via L-PBF.

The results of the cell study did not definitively demonstrate that one roughness group was superior in terms of cell proliferation or differentiation. MC3T3 cells undergo three stages of growth in culture: proliferation (days 0-9), differentiation (days 10-20) and mineralization (days 20+).⁽²⁴⁾ The various assays done in this study were designed to assess the effect of surface topography at all three stages. However, specific timepoints were primarily chosen to examine markers with clinically meaningful impacts on mineralization. Proliferation was quantitatively measured using a MTT assay.⁽²⁵⁾ Cellular metabolic activity was not significantly different in the earliest two time points (Figure 3B). Although, there was an increased MTT activity in the As Built group compared to sprout group at the week 4 time point, the primarily conclusion should be based on data from the proliferation time point.⁽²⁶⁾ The results of MTT assay are corroborated by the florescent images of the discs that show robust cell growth in all groups. The differentiation phase of the MC3T3 occurs between 10 – 20 days where matrix formation and maturation takes place and is marked by increased ALP activity followed by a drop in ALP as cells progress into the mineralization phase.⁽²⁴⁾ The presence of cell matrices was reflected by the ECM film where collagen stained with picrosirius red that was present at week 2 and became more prominent at week 4 (Figure 6A). The ALP assay demonstrated no difference between roughness group, but the reduction of ALP at week 4 followed trends reported in literature indicating that cells have progressed past differentiation phase. The transition to the mineralization phase is further illustrated by the large increase in Alizarin red staining, indicating calcium deposits⁽²⁷⁾ (Figure 6B). The inclusion of sprouts appreciably increases the surface area of the disc. The ability to increase the surface area without significantly increasing the volume of a construct is a theoretical advantage of printing topography. Therefore, the quantitative MTT and ALP results were not normalized by surface area so any change in those values due to variation in the surface area could be captured.

In contrast to the other cell assays, RT-qPCR revealed that roughness had an impact on Bglap gene expression of the MC3T3 cells. Bglap is a calcium binding protein that is initially expressed in the differentiation phase before becoming highly expressed in the mineralization phase of osteoblast differentiation.^(24),(28) By grouping samples based on similar mechanical properties (post-processed vs artificial topographical features), RT-qPCR results indicate that total RNA encoding for Bglap was significantly diminished in the cells grown on discs with artificially printed roughness compared to the less rough surfaces (Figure 4 A,C). Moreover, the sprout and rough sprout discs did not have the increased Bglap activity from week 2 to week 4 expected as cells move into the mineralization phase and as observed in the discs with reduced roughness. Sost was poorly expressed in all groups across both time points (Figure 4 B,D). Sost is a late stage osteocyte marker needed for the regulation of mineralization.⁽²⁹⁾ The low expression of Sost in all both types suggests the MC3T3 cells have not differentiated into mature osteocyte. Future studies with longer culture timepoints or additional genetic markers of osteoblast differentiation allow for determination of a difference.

The results seemingly indicate that adding roughness either has no affect or a slightly negative affect on osteoblast activity in contrast to past studies that assert that increasing roughness of titanium substrate enhanced cell proliferation or differentiation.^{(15),(29)-(31)} However, many of these studies produce roughness via subtractive manufacturing, often using acid etching techniques which produce roughness on a smaller scale than was able to be produced in this study using additive manufacturing. The minimum feature size for LBF printer used in this study is 100 μm.⁽³²⁾ Acid etching has been show to create hierarchical micro/nano structures that were not recapitulated with L-PBF followed by polishing or blasting. ⁽³³⁾ Microroughness and nanoscale roughness can be sensed by the cell and therefore influence cell behavior, while roughness on the macroscale is important for bone fixation.^(34),(35) The maximum distance between the peak and valley, R_z, indicate that artificially printed topography pushed the boundary of microscale into the meso-scale regime.⁽³⁶⁾ The sprouts and the rough sprouts groups have the microscale topography as the as built group. The SEM images (Figure 5B) show the fused Ti powder particles in all 3 groups that did not undergo any postprocessing. The addition of sprouts did not change the microscale and nanoscale topography. Therefore, no significance biological difference in cell behavior was observed in our model because microscale and nanoscale topography are the most important factors influencing cell behavior.

Similarly, the in vitro data in this study showed artificially printed roughness had either no effect or slightly negative effect on cell behavior, but may suggest that the meso- scale roughness is important for bone fixation to the implant. SEM images showed extensive cell matrix structures for the groups with meso-scale roughness, but less ECM formation on As Built, Blasted, or Polished discs. The florescent images showed extensive cell growth in all the groups. To process the samples for SEM the discs underwent several rounds of washes. The rougher discs may have been more resistant to the sheer force of the washes supporting the claim that macroroughness helps with adhesion to the implant. Further in vitro studies are needed to specifically evaluate the effect of 3D printed meso-scale topographical features on cell matrix fixation. If 3D printed roughness does in fact aid in implant fixation, artificial roughness will be designed on implant surfaces opposed to bone to increase early fixation at the bone-implant interface. In addition, it is well known that porosity enhances osseointegration in metal implants^(37)-(42) and recent work has shown that porosity is a larger factor in controlling osseointegration versus surface roughness alone.^(43),(44) Osseous integration remains a challenging problem in orthopedic implants likely requiring a multifaceted solution. 3D printing has opened the doors to new geometries that were not practical with traditional manufacturing methods.

This study evaluated the strategy of artificially printing rough topography on a substrate. The tensile behavior showed that the additionally printed roughness produced less favorable mechanical properties, specifically diminishing elongation. The in vitro experiment did not show a clear advantage of artificially printed roughness in enhancing osteoblast proliferation and differentiation as hypothesized. This was likely due to scale of the printed features being too large to influence activity at the cellular level but may enhance adhesion of cell matrix at the meso-scale. While in vitro studies are a useful tool to ethically, and efficiently evaluate scaffold design parameters, in vivo models are ultimately needed before results can be translated to clinical practice.

In conclusion, the addition of 3D printed artificial roughness leads to inferior mechanical properties and confers no clear benefit regarding cellular proliferation. Printed topography increases the initiation of fractures resulting in diminished tensile strength and ductility. Concurrently, the resolution of LBF is not fine enough at this time to create surface features that enhance cell behavior. Therefore, data in this study suggest that artificially printing roughness is not an effective strategy to enhance osseous integration into titanium implants for critical sized defects.