Acta Biomaterialia 10 (2014) 394–405

Contents lists available at ScienceDirect

Acta Biomaterialia
journal homepage: www.elsevier.com/locate/actabiomat

Bone healing and the effect of implant surface topography

on osteoconduction in hyperglycemia
E. Ajami a, E. Mahno a, V.C. Mendes a,b, S. Bell a, R. Moineddin c, J.E. Davies a,b,⇑
a
Institute of Biomaterials and Biomedical Engineering, University of Toronto, 164 College Street, Toronto, Ontario M5S 3G9, Canada
b
Dental Research Institute, Faculty of Dentistry, University of Toronto, 124 Edward Street, Toronto, Ontario M5G 1G6, Canada
c
Department of Family and Community Medicine, Faculty of Medicine, University of Toronto, 263 McCaul Street, Toronto, Ontario M5T 1W7, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Dental implant failures that occur clinically for unknown reasons could be related to undiagnosed hyper-
Received 25 June 2013 glycemia. The exact mechanisms that underlie such failures are not known, but there is a general consen-
Received in revised form 12 September sus that bone growth is compromised in hyperglycemia. Nevertheless, contradictory findings exist
2013
related to peri-implant bone healing in hyperglycemia. We hypothesized that hyperglycemia delays early
Accepted 18 September 2013
Available online 26 September 2013
bone healing by impeding osteoconduction, and that the compromised implant integration due to hyper-
glycemia could be abrogated by using nanotopographically complex implants. Thus we undertook two
parallel experiments, an osteotomy model and a bone in-growth chamber model. The osteotomy model
Keywords:
Bone healing
tracked temporal bone healing in the femora of euglycemic and hyperglycemic rats using micro com-
Hyperglycemia puted tomography (microCT) analysis and histology. The bone in-growth chamber model used implant
Implant surfaces of either micro- or nanotopographical complexity and measured bone–implant contact (BIC)
Surface topography using backscattered electron imaging in both metabolic groups. Quantitative microCT analyses on bone
volume, trabeculae number and trabeculae connectivity density provided clear evidence that bone heal-
ing, both reparative trabecular bone formation and remodeling, was delayed in hyperglycemia, and the
reparative bone volume changed with time between metabolic groups. Furthermore, fluorochrome label-
ing showed evidently less mineralized bone in hyperglycemic than euglycemic animals. An increased
probability of osteoconduction was seen on nano-compared with microtopographically complex surfaces,
independent of metabolic group. The nanotopographically complex surfaces in hyperglycemia outper-
formed microtopographically complex surfaces in euglycemic animals. In conclusion, the compromised
implant integration in hyperglycemia is abrogated by the addition of nanotopographical features to an
underlying microtopographically complex implant surface.
Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

1. Introduction Uncontrolled diabetes has been associated with an increased

Diabetes mellitus has been described as a ‘‘group of metabolic
disorders sharing the common underlying feature of hyperglyce-
mia’’ [1], rather than a single disease entity. The International
Diabetes Federation (IDF) reports an estimated 25 million diabetics
in the USA, of which 7 million have not been diagnosed [2]. While
the impact of diabetes can be seen in many different tissues and
physiological systems, such as the kidneys, eyes, nerves and blood
vessels, its effect on bone is notable [3,4], although not entirely
understood.
⇑ Corresponding author at: Institute of Biomaterials and Biomedical Engineering,
University of Toronto, 164 College Street, Toronto, Ontario M5S 3G9, Canada.
Tel.: +1 416 978 1471; fax: +1 416 946 5639.
E-mail addresses: jed.davies@utoronto.ca, jed@verypowerfulbiology.com
(J.E. Davies).
risk of dental implant failures [5–7], and it is possible that undiag-
nosed diabetes may be related to the 5% of dental implants that fail
clinically for unknown reasons. Although the mechanisms that
underlie such failures have not been clearly explained, the general
consensus is that bone growth is compromised in hyperglycemia.
The bone in hyperglycemic individuals is weaker and more fragile
than healthy bone [4,8]. Affected individuals seem to have higher
rates of osteoporotic fractures [9], slower wound healing [10],
decreased skeletal growth during adolescence [11], and delayed
or impaired fracture healing [12]. Nevertheless, the reports related
to peri-implant bone formation and remodeling in hyperglycemia
are contradictory, mostly due to obvious differences in animal
models, implants employed, experimental procedures, and, most
importantly, study time points.
Some reports show consistent decreases in bone volume adja-
cent to an implant [13], while others report an increase in the
amount of bone [14]. Reports of decreased mechanical retention

1742-7061/$ - see front matter Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.actbio.2013.09.020
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405
of implants [15] and bone–implant contact (BIC) [14,16–18], which
is the product of osteoconduction and de novo bone formation,
provide support for the notion of compromised peri-implant heal-
ing in diabetic subjects, yet no mechanistic explanations exist for
such differences. We believe that chronic hyperglycemia has an
impact on the early stages of peri-implant healing, specifically
osteoconduction and de novo bone formation, which consequently
affects long-term endosseous implant stability. This could lead to
an increased number of implant failures in the undiagnosed hyper-
glycemic population compared with a truly healthy population.
In the present study we monitored the effects of hyperglycemia
on early bone healing and, specifically, tested the effect of hyper-
glycemia on osteoconduction, in the presence of candidate implant
surfaces, using a previously established model [19]. We hypothe-
size that hyperglycemia could delay early bone healing by imped-
ing osteoconduction. Since osteoconduction, together with bone
formation, results in contact osteogenesis, and since nano-
topographically complex implant surfaces have previously been
shown to accelerate osteoconduction [19], we also hypothesize
that compromised implant integration due to hyperglycemia could
be abrogated by using nanotopographically complex endosseous
implants. To address our hypotheses we undertook two parallel
experiments. In the first, an osteotomy model, we created femoral
drill holes in both hyperglycemic and healthy rats and tracked
bone healing with time by micro computed tomography (microCT)
and histology. In the second, an implant model, we measured BIC
in both hyperglycemic and healthy rats using custom bone
in-growth chambers modified with either micro- or nanotopo-
graphically complex surfaces.
2. Materials and methods
2.1. Design of ‘‘T-plant’’ bone in-growth chambers
200 titanium (commercially pure grade IV) bone in-growth
chambers, termed T-plants (Fig. 1A), were custom fabricated by
Biomet 3i (Palm Beach Gardens, FL) for this study. The implants
were machined in two modular parts and then assembled. The
external dimensions of each T-plant are approximately
5  3  2 mm (height  width  depth), and the internal chamber
has dimensions of 3  3  1 mm (height  width  depth)
(Fig. 1B).
Four different surface treatments were applied to generate
micro- and nanotopographically complex surface designs on 200
implants. Initially all chambers were acid etched by a dual acid
etch (DAE) method in 8% HF solution followed by 78% H2SO4/3%
HCl solution (wt.%). No further modification was made to 50
Fig. 1. (A) View of the bone in-growth chamber, which is assembled in modular parts and has the shape of a ‘‘T’’. The external dimensions of the T-plant chamber are 5 (H)  3
(W)  2 (D) mm. (B) Side view of the T-plant showing the chamber into which bone grows. The internal dimensions are 3 (H)  3 (W)  1 (D) mm. (C) The plane of grinding,
which is shown dotted, was along the long axis of the femur to produce a cross-section showing both walls of the T-plant chamber. The double arrow shows the area that
accommodated eight cross-sectional layers with 250 ± 50 lm intervals.
395
implants (DAE group). Of the remaining implants, 50 implants
were further etched in 3% KOH/17% H2O2 (wt.%) at 52 °C for
1 min (MAE group), while another 50 were etched in 22% KOH at
60 °C for 65 min (NAT group). Both groups were then treated with
HNO3 at 60 °C for 10 min, rinsed in water and dried in an oven at
110 °C. The remaining 50 implants were modified by the deposi-
tion of discrete crystalline calcium phosphate nanocrystals (DCD
group). For the DCD treatment implants were dipped in an alco-
hol-based suspension containing 1% nanocrystals of stoichiometric
hydroxyapatite (20–100 nm in size, P95% crystalline) at room
temperature and dried in an oven at 100 °C.
Thus a total of four groups, each containing 50 samples, were
generated.
2.2. Characterization of candidate implant surfaces
2.2.1. Surface roughness
Surface roughness analysis was performed using an optical
interferometer (MicroXAM-100, ADE Phase Shift, KLA-Tencor,
Milpitas, CA). The data was obtained at 312.5 magnification, over
a 52,400 mm2 implant area. A 50 mm Gaussian filter and inverse
fast Fourier transformation were employed.
2.2.2. Field emission scanning electron microscopy (FE-SEM)
Two additional T-plants from each surface group were opened
by carefully prizing the two walls of the chambers apart with a
scalpel blade. This procedure was done with extreme caution so
as not to damage or contaminate the internal walls of the cham-
bers. The internal surface topography of the chamber walls was ob-
served by FE-SEM (Hitachi S-5200, Japan) at an accelerating
voltage of 5 keV and increasing magnification (up to 100,000)
at the Centre for Nanostructure Imaging, University of Toronto,
without coating with an electron-conducting medium.
2.3. Animals
The experimental protocol was approved by the Ethics Commit-
tee of Animal Research at the University of Toronto. 180 Young
male Wistar rats (200–250 g, Charles River Laboratories, Canada)
were housed at the animal facility of the Faculty of Dentistry,
University of Toronto. Animals were allowed to adapt for 1 week
prior to commencement of the study. They had free access to rat
chow and water throughout the study. 80 Animals were used for
the osteotomy model and 100 for the implantation model. In each
study the animals were divided into two metabolic groups, hyper-
glycemic (H) and healthy control (C). Thus the final animal groups
396 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405
consisted of four groups: osteotomy model, 40 °C and 40 H (10 per
time point in each group); implantation model, 50 °C and 50 H.
2.4. Streptozotocin-mediated hyperglycaemia induction
Animals from the H groups were injected intraperitoneally in
the abdominal upper right quadrant with 65 mg kg 1 streptozoto-
cin (STZ) (Sigma Aldrich Co., Canada) dissolved in 0.9% sterile sal-
ine. Age-matched control animals were injected with the same
volume of saline only.
Weight and blood glucose levels of all animals were monitored
prior to injection, 24 and 48 h post-induction, at surgery following
anesthesia, and at death. To measure blood glucose a drop of blood
was collected in a glucose test strip (FreeStyle Lite, Abbott Diabetes
Care Inc., Alameda, CA) following standard tail vein puncture,
which was loaded into a conventional glucometer (FreeStyle Lite).
Animals with blood glucose levels of P16 mmol l 1 in the first 48 h
were considered hyperglycemic and underwent surgery 1 week
post-induction. If hyperglycemia was not achieved a second STZ
injection was given 1 week after the first. A maximum of three
injections (at 1 week intervals) were applied.
2.5. Surgical procedures
Animals underwent general anesthesia with isofluorane in
nitrous oxide and oxygen (5% induction, 2–2.5% maintenance).
Buprenorphin 0.01–0.15 mg kg 1 was administered subcutane-
ously as an analgesic both pre- and post-operatively. Aseptic con-
ditions were maintained throughout the surgical procedures. The
antero-lateral aspect of each hind limb of the animals was shaved
and an incision was made through the skin, parallel to the femur.
After blunt dissection of the underlying muscles the periosteum
was incised and the lateral border of the femur, proximal to the
knee capsule, on the widest aspect of the femur, was exposed.
2.5.1. Osteotomy
Bilateral mono-cortical round defects were created in the ante-
rior cortex of the femur using a round dental bur (2.3 mm diame-
ter, Brasseler, USA) attached to a dental handpiece (ImplantMED
DU 900 and WS-75, W&H Dentalwerk, Austria) and were left to fill
with blood.
2.5.2. Implantation
The length of the bone defect was bilaterally demarcated with a
round bur (1.4 mm diameter, Brasseler, USA) in the anterior cortex
of the femur. A 1.5 mm custom made cylindrical drill (Biomet 3i,
USA) was used to create an osteotomy, which resulted in a bone
defect with approximate dimensions 3  2 mm (length  width)
in both cortices of the femur. A T-plant was initially positioned just
above the defect, so that the blood emerging from the bony orifice
filled the chamber, and finally it was press-fitted into the defect.
The samples were assigned randomly for each femur.
In both models the muscle tissues were sutured with biode-
gradable sutures (4–0 Polysorb, Syneture, USA) and the cutaneous
tissues were re-apposed using staples (9 mm wound clips, Becton
Dickinson, Franklin Lakes, MD). Animals were maintained in the
animal care facility of the Faculty of Dentistry, University of
Toronto, allowed free access to water and rat chow, and observed
daily during recovery. Surgical sites were observed and inspected
for signs of infection, and animals were checked for signs of
compromised ambulatory ability.
2.6. Fluorochrome labeling
Tetracycline hydrochloride (30 mg kg 1) and alizarin red
(50 mg kg 1) (Sigma Aldrich Co., Canada) were dissolved in sterile
saline and injected into the lower right abdominal quadrant in
alternating sequence on the seventh post-operative day and at
7 day intervals thereafter for 30 days.
2.7. Harvesting and embedding of specimens
Death was achieved by exposure to CO2 followed by cervical
dislocation. Implanted animals were killed 9 days post-
implantation, while those in the osteotomy wound model were
killed at 5, 10, 15, and 30 days. After harvesting the femora were
initially fixed in 10% neutral formalin for at least 24 h.
The samples were then trimmed to smaller bone segments, to
the width of the defect in the osteotomy model and to the width
of the T-plant in the implantation model. The specimens were then
dehydrated in ascending concentrations of ethanol.
For the segments containing T-plants a rectangular mounting
plate was fabricated from aluminum with 10 curved slots designed
to accept the curved cap of the T-plant (see Fig. 2 in Mendes et al.
[19]). A maximum of 10 T-plants with the surrounding bone seg-
ment were mounted on each plate, parallel to each other and per-
pendicular to the mounting plate, and embedded as a block in
polymethyl methacrylate (PMMA) resin (OsteoBed, Polysciences,
Warrington, PA) according to the supplier’s protocol. The samples
from the osteotomy model collected 30 days after operation were
also embedded in the same resin.
2.8. MicroCT analysis
Samples were scanned using MicroCT coupled with a custom
software package (MicroCT40, Scanco Medical, Basserdorf,
Switzerland) at 70 kVp and 114 lA at high resolution (6 lm) in
three planes. This created 600–1000 axial cut slices of 6 lm
thickness. A region of interest (ROI) was selected and highlighted
on the cross-sectional images from each specimen, based on the
healing phase at each time point. ROIs were then reconstructed
within the full three-dimensional (3-D) reconstruction of the
specimen.
Bone was highlighted within ROIs using threshold segmentation
based on gray level distribution, allowing for bone volume and per-
cent bone composition calculations. The threshold value remained
constant for all samples at the same time point, but was altered at
different time points to account for bone at different stages of heal-
ing. The following were measured within the medullary space on
samples harvested at 5 and 10 days after operation: bone volume
per total volume (BV/TV%), connectivity density (the interconnectiv-
ity of reparative bone trabeculae within the medullary space), tra-
beculae number, thickness, and spacing, and bone density.
2.9. Microscopy
2.9.1. Repetitive block face microscopy
The blocks with implanted bone segments were ground and
polished on an automated microgrinder machine (Exakt 400 CS,
Exakt Technologies Inc., USA). A protocol was developed to deter-
mine the grinding and polishing time based on the paper grit
roughness (320, 500, 800, 1000, 1200, 2400, and 4000 grit, Buehler,
USA), so that approximately the same amount of grinding and pol-
ishing was achieved for every layer of each block. The plane of
grinding was the long axis of the femur to produce a cross-section
showing both walls of the T-plant chamber (Fig. 1C).
Samples were carbon sputter coated and backscatter electron
imaged (Hitachi S-570, Japan) at the Faculty of Engineering,
University of Toronto. Grinding, polishing, and imaging were re-
peated numerous times on several surface layers of each sample
block to acquire images at sequential planes through the T-plant
chamber.
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405 397

Fig. 2. (A) Summary of the healing process in the H and C metabolic groups over time (arrow) as portrayed by representative microCT images. Trends in medullary (B) bone
volume, (C) trabeculae number, (D) connectivity density, (E) trabecular thickness, (F) trabecular spacing, and (G) bone density between the H (red) and C (blue) groups at 5, 10
and 15 days post operation. ⁄Statistical significance at P < 0.05; ⁄⁄statistical significance at P < 0.01. Means are labeled as percentages.

2.9.2. Fluorescent microscopy
For qualitative analysis of fluorochrome labeling the
resin-embedded samples from the osteotomy model were cut in
cross-section through the center of the defect and trimmed using
a cutting system (Exakt 300 CL, Exakt Technologies Inc., USA).
Sections were fixed to back-up slides and ground to a final thick-
ness of 15 lm using an Exakt 400 CS microgrinding system and
decreasing paper grit roughnesses (800, 1200, 2000, and 4000 grit,
Buehler, USA). Images were acquired using a Ti-E inverted micro-
scope equipped with a Nikon C1si laser scanning confocal system
398 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405
and EZ-C1 software (Nikon, USA). Image viewing and analysis was
performed using EZ-C1 free viewer (Nikon, USA).
2.10. Histology
To check for unmineralized reparative bone formation in the
medullary compartment at early time points of healing selective
5 day post-operative samples from the osteotomy model were col-
lected, decalcified, and embedded in paraffin blocks. 6 lm thick
sections were cut from the blocks, stained with haematoxylin
and eosin, and examined using a light microscope (Olympus
BX51) (Olympus, Melville, NY) interfaced with a digital camera
(SPOT RT, Digital Instruments, now Veeco Instruments, Woodbury,
NY) running Qcapture software.
Selective resin-embedded blocks from the implant model were
chosen for histological examination of bone inside the chambers.
Once the final layer of each block was backscatter electron imaged
the blocks were fixed on a microscopy slide with their imaged side
facing the slide. They were then cut using a cutting system (EXAKT)
through the mounting plate and approximately 1 mm away from
the slide. The remaining mounting plate was then lifted off the
resin, leaving the bone segments with the T-caps of the T-plants
intact inside the resin. The T-caps were then milled using a high
speed dental handpiece (DCI) until flattened. The section was then
ground and polished using an Exakt 400 CS microgrinding system
and decreasing paper grit roughnesses (as above) until the T-caps
were fully removed and the chambers were exposed.
2.11. Data acquisition and statistical analysis
Numerical data were obtained following microCT analysis of
medullary bone, and BIC measurements were performed with the
software Sigma Scan Pro 4 (SPSS, USA) calibrated to generate val-
ues in micrometer units for each backscatter electron image. For
each micrograph the length of the implant wall on each side of
the chamber was measured to allow normalization of the data,
by transforming the measured BIC values into percentages, to cor-
rect for any obliquity of the planes of section which would other-
wise introduce error.
Box plots were used for visual comparisons. The box stretches
from the 25th percentile to the 75th percentile. The median is
shown as a line across the box, and the average is shown as a solid
symbol inside the box. If the median is not equidistant from the
borders then the data is skewed. The ends of the vertical lines show
the minimum and maximum values. If outliers are presented then
lines are extended to a maximum of 1.5 times the interquartile
range (the height of the box), and the points outside the ends of
the lines are outliers or suspected outliers.
For medullary bone analysis the generalized estimation equa-
tions (GEE) method was used for correlated measurements and
comparing metabolic groups and time points. If there were no cor-
related measurements analysis of variance was used to compare
the groups and Pearson correlation was used to measure the corre-
lation among continuous measurements.
To calculate the percentage of BIC for all surfaces, metabolic
groups, and legs the Newton–Cotes formulae were used, which
approximate the integration of a complicated function by replacing
the function by many polynomials across the integration interval.
The GEE method was used to analyze correlated measurements
and compare surface groups and metabolic groups.
For statistical analysis of BIC probability the BIC were catego-
rized as ‘‘No’’ if the measurement was zero and ‘‘Yes’’ otherwise.
Then the frequencies of ‘‘Yes’’ for each group (MAE, DAE, NAT,
and DCD) in both metabolic groups were plotted. These frequen-
cies showed the probability of BIC for each material surface group
in each metabolic group.
A distribution analysis of BIC was performed for all surface
groups in both metabolic groups by categorizing BIC as being
0–24.2% (overall median) or above 24.2%. The analyses were then
done using logistic regression, using the GEE method to adjust
for repeated measurements per rat.
The statistical software SAS 9.1 (SAS Institute, USA) was used
for data analysis, and a P value of <0.05 was considered significant.
3. Results
All animals survived the operative procedure and recovered
uneventfully from the anesthesia without any evident complica-
tions. In the osteotomy wound model three rats in the H group
did not become hyperglycemic after three attempts at induction
and were excluded from the study. Therefore, the resulting sample
sizes were: 5 days, C 10 and H 10; 10 days, C 10 and H 9; 15 days,
C 10 and H 8; 30 days, C 10 and H 10. These groups will be denoted
C5, C10, C15, and C30 and H5, H10, H15, and H30 for the C and H
animals at 5, 10, and 30 days, respectively.
Similarly, five rats in the C group and one in the H group in the
implant groups were found to have fractured femora at harvest,
and these femora were excluded from the study. Two femora, both
from the H group, were also excluded from the study due to grind-
ing problems during sample preparation for backscatter electron
imaging. Therefore, the resulting sample sizes were: DAE-H 24;
MAE-H 24; NAT-H 24; DCD-H 25; DAE-C 25; MAE-C 25; NAT-C
23; DCD-C 22.
3.1. Osteotomy model
3.1.1. MicroCT analysis
A summary of temporal healing in both the H and C groups is
illustrated in Fig. 2A using representative microCT images obtained
at each time point, which reveal evident differences between the
metabolic groups. At 5 days a significantly lower medullary bone
volume was observed in the H compared with the C group. Com-
paring the amount of medullary bone between 5 and 10 days in
both metabolic groups, a considerable increase was observed from
5 to 10 days in the H group, while a slight decrease in the C group
was observed up to 10 days. Furthermore, the amount of medullary
bone in C10 and H10 seemed to be comparable. At 15 days more
medullary bone was observed in the H compared with the C group,
and, finally, at 30 days all the medullary bone had been resorbed in
both metabolic groups.
These differences were of statistical significance when analyz-
ing the quantitative microCT data. The H5 group had a significantly
lower BV/TV% compared with C5, with no statistical differences be-
tween the two groups at either 10 or 15 days (Fig. 2B). Comparing
BV/TV% within the metabolic groups at different time points re-
vealed differences only between 10 and 15 days for both groups
(H, P = 0.0001; C, P = 0.0001). Interestingly, there was an increase
in BV/TV% in the H group between 5 and 10 days, reaching a peak
on day 10. This was not observed in the C group, which showed a
maximum BV/TV% at 5 days with a decrease at 10 and 15 days. Fur-
thermore, BV/TV% in H10 was not significantly different from that
in C5.
There was a greater number of trabeculae in the C5 group com-
pared with the H5 group (P = 0.03), with no differences at 10 and
15 days (Fig. 2C). Within the metabolic groups an increase in
trabeculae number was seen from H5 to H10 (P = 0.006), with a
decrease from H10 to H15 (P = 0.0001). In the C group there was
no difference in the number of trabeculae from C5 and C10, but
there was a decrease from C10 to C15 (P = 0.001). The decrease
in trabeculae number from 10 to 15 days in C was not statistically
significantly different from that in the H group, which indicated a
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405 399

consistent decrease in trabeculae in both groups. Similar to the
pattern observed in BV/TV%, the mean trabeculae number in C5
was not statistically different from H10, while the maximum value
for the H group occurred at 10 days.
Connectivity density was greater in the C5 (P = 0.002) and lower
in the C10 (P = 0.01) groups compared with the H5 and H10 groups,
respectively, yet there was no difference after 15 days (Fig. 2D).
Connectivity density within metabolic groups at different time
points revealed a decrease with time (C5 to C10 (P = 0.001) and
C10 to C15 (P = 0.0003); H10 to H15 (P = 0.0002)).
There were no significant differences in density, trabecular
thickness and trabecular spacing of the reparative medullary bone
between the H and C groups at any time points (data not shown).
3.1.2. Histological assessment of medullary bone
The histological assessment of medullary bone correlated with
the quantitative data obtained from microCT analysis. Apparent
morphological differences were observed between the H and C
groups at 5 days post-operation (Fig. 3A vs. B), at which time a
fibrin clot with entrapped red blood cells was still visible in the
H sample (Fig. 3A1), but not in the C sample (Fig. 3B1). New bone
formation was evident in both the C and H samples. In the C sam-
ple new bone was seen throughout the osteotomy defect, whereas
it was only observed at the periphery of the residual blood clot in
the H sample (Fig. 3A2 vs. B1).
3.1.3. Qualitative fluorochrome labeling
Fluoresence images of samples in both the C and H groups taken
with identical microscope settings, including equal photographic
exposure, are shown in Fig. 4. At low magnification autofluore-
sence was observed from the cells of the medullary space in both
metabolic groups. Intense fluorescence from both the tetracycline
and alizarin red labels was clearly visible in the cortex of the
C sample (Fig. 4A). In contrast, the cortical bone in the H sample
exhibited considerably less fluorescence (Fig. 4B). At higher magni-
fication (Fig. 4A1) organized patterns of parallel fluorescent bands
were apparent close to the endosteal surface, with a more random
pattern close to the center of the cortex. At the defect site staining
followed the disorganized pattern of reparative bone formation
(Fig. 4A2). In the H sample only weak fluorescence was visible at
the defect site from either tetracycline or alizarin red labeling
(Fig. 4B1). Fluorescence from the H sample was less intense than
from the C sample even at increased exposure (Fig. 4B1s).
3.2. Implantation model
3.2.1. The implant surfaces
Optical interferometry returned values for each group of: DAE,
0.47; MAE, 0.48; NAT, 0.49; DCD, 0.46.
FE-SEM photomicrographs of implant surfaces are shown in
Fig. 5. An overview of all surfaces revealed micron scale

Fig. 3. Light micrographs of samples collected at 5 days post operation from (A) the H and (B) the C groups. In the H sample the medullary cavity was still occupied by a clot
(field width 5.4 mm), as is evident by (A1) entrapped red blood cells in a fibrous network (field width 0.22 mm). (A2) New bone formation (light pink) is evident peripheral to
the clot (field width 0.87 mm). (B) The medullary cavity of the C sample was mostly occupied by new bone (light pink, field width 5.4 mm), encompassing (B1) darker staining
pink bone chips created during surgical drilling (arrows, field width 2.2 mm).
400 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405

Fig. 4. Samples labeled with tetracycline HCl (green) and alizarin red (red) and harvested 30 days post operation from both metabolic groups. (A) A cross-section near the
middle of the defect in C group. (A1) Magnified image of region 1 in (A). Clear fluorescent labeling is observed in the region opposing the defect. Disorganized labeling is seen
closer to the center of the cortex, but is rather organized near the endosteal surface. (-2) Magnified image of region 2 in (A). Vibrant labeling is observed in the defect region.
(B) A cross-section near the middle of the defect in the H group. (B1) Magnified images of region 1 in (B). Very faint tetracycline HCl labeling is observed in the defect region.
The images in (B) and (B1) were acquired under the same microscopy settings and equal exposures to those in (A), (A1) and (A2). (B1s) The same image as in (B1) taken at
increased exposure settings for easier visualization. The magnified image of region 2 in (B), which is the region opposite to the defect, had very little tetracycline HCl labeling
that could not be visualized and is thus not shown here.

topographical features in the z-plane, represented by a combina-
tion of peaks and valleys of 1–3 lm in size. However, at higher
magnification sub-micron scale structures were visualized within
the valleys, except for the DAE group. The latter was smooth
within the craters and no undercuts were observed at the reso-
lution limit of the microscope (Fig. 5C). Compared with the DAE
surface the MAE treatment superimposed sub-micron scale pits
on the underlying micron scale craters, which could be clearly
visualized at high magnifications. While these nano-sized pits
increased the sub-micron scale complexity of the underlying
surface, they created a z-plane reduction in the underlying mic-
rotopography (Fig. 5F). In contrast, the NAT treatment resulted in
the addition of a reticulate network of tangled nanofibers
(Fig. 5I), which covered the surface of the underlying substrate,
while the DCD treatment also superimposed features with sub-
micron scale undercuts (Fig. 5L). Thus, NAT and DCD were both
additive treatments and did not alter the microtopography of the
underlying substrate. Therefore, we divided these four surfaces
into two operational groups of micro (DAE and MAE) and nano
(DCD and NAT) surfaces.
3.2.2. Histological, histomorphometric, and statistical analysis of BIC
Qualitative analysis of backscatter electron micrographs
showed new bone formation at both ends of the chambers
advancing toward the center in all groups. An osteoconductive
pattern was characterized by BIC along the walls of the chamber
(Fig. 6A), whereas bone growth within the chamber volume,
with minimal or no contact with the walls of the implant, char-
acterized a non-osteoconductive pattern (Fig. 6B). Histological
assessment of bone healing revealed no obvious signs of unmin-
eralized bone inside the bone in-growth chamber and/or on the
walls of the implant. The backscatter electron micrographs of the
samples correlated with the light micrographs (Supplementary
Fig. 1).
For each surface under both the C and H conditions the percent-
age probability of having zero bone growth (BIC% = 0) was recorded
and the value inverted (Fig. 7A) to generate a probability of osteo-
conduction. Both nano surfaces (NAT and DCD) had a significantly
higher probability of osteoconduction compared with the two mi-
cro surfaces (MAE and DAE), independent of metabolic group.
While the differences in the probability of osteoconduction on
the nano surfaces under both metabolic groups was negligible,
the micro surfaces presented higher probabilities of osteoconduc-
tion in the H compared with the C group.
Box plot presentations of mean BIC% for all surface groups
showed slightly higher mean BIC% values for C compared with
H animals (Fig. 7B). However, significant differences between the
surface groups in both metabolic groups were observed (Table 1).
Both nano surfaces had significantly higher BIC% compared with
the two micro surfaces in both the C and H groups, with the excep-
tion of NAT and DAE in the H group (Table 1A and B).
Comparing surface groups between metabolic groups
(Table 1C), there were no statistical differences between DCD-H
and all the surfaces in the C group, except for MAE-C, which had
a significantly lower mean BIC% than DCD-H. There were no differ-
ences between NAT-H and DAE-C, however, the mean BIC% for
NAT-H was lower than for DCD-C and higher than for MAE-C.
The DAE-H group presented no statistically significant differences
in mean BIC% in comparison with miro-C, but the mean BIC% for
this group was significantly lower in comparison with nano-C.
Finally, MAE-H presented a significantly lower mean BIC% value
in comparison with all other surfaces in the C group.
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405 401

Fig. 5. FE-SEM micrographs of the implant surfaces at different magnifications. (A–C) Dual acid etched (DAE) surfaces; (D–F) micro acid etched (MAE) surfaces; (G–I) alkali-
treated (nanotitanate, NAT) surfaces; (J–L) CaP-treated (DCD) surfaces. The low magnification (10,000) micrographs (A, G, J) show the general microtopography of the DAE
surfaces, while the micrograph in (D) shows the microtopography of the original DAE surface. The surfaces were characterized by prominent features in the z-plane (B, E, H, K),
represented by peaks and valleys. At higher magnifications (100,000) the DAE surface was shown to be smooth with no undercuts (C), the MAE surface showed reduced
microtopography with some nano features due to the secondary etching process (F), the NAT surface was shown to have a network of tangeled nanofibers (I), and the DCD
surface was characterized by an enhanced implant surface nanotopography due to the deposition of 20–100 nm calcium phosphate nanocrystals (L), while leaving the
underlying original microtopography of the surface visible. Field widths 12.55 lm (left column), 2.51 lm (middle column) and 1.25 lm (right column).

Fig. 6. Backscattered SEM micrographs of the implant surfaces showing different patterns of bone in-growth. (A) Uniform bone growth along both walls of the T-plant
chamber is clearly visualized and represents an osteoconductive pattern. (B) Bone in-growth inside the chamber occurs without any bone contact with the walls of the
T-plant chamber and represents a non-osteoconductive pattern.
402 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405

the BIC% values in the two metabolic groups (Fig. 8). By visualizing
these maps we found evidence of the nano surfaces showing higher
BIC% values compared with the micro surfaces. Therefore, the
probability distribution of BIC% values falling above, or below, a
median value (24.2%) was plotted for all surface types in both met-
abolic groups (Fig. 8B).
The probability of BIC% measurements falling above the median
in the C group is higher than for the H group within all surface
groups, however, the differences are not statistically significant.
DCD had a significantly higher probability of BIC% values falling
above the median than both micro surfaces, independent of meta-
bolic groups, except DCD-H and DAE-C, which were not signifi-
cantly different (Fig. 8C). However, the NAT surface in both
metabolic groups significantly outperformed MAE in both the
C and H groups. No differences were observed between NAT-H
and DAE-C and DAE-H.

4. Discussion

We have demonstrated, using our osteotomy model, that bone

healing, both reparative bone formation and remodeling, were
delayed in hyperglycemic animals. Furthermore, these delays
occurred in the early stages of healing. This diaphyseal drill hole
bone healing model, in which there is initially negligible amounts
of medullary trabecular bone, is characterized by a reparative
trabecular bony response to injury through remodeling, by
re-establishment of the bony cortex and resorption of the medul-
lary trabeculae, to restore homeostasis. As an example, the most
significant differences in bone volume, medullary connectivity
density and trabeculae number between metabolic groups were
Fig. 7. (A) The probability of osteoconduction (BIC% values not zero) in both the H seen after 5 days, all of which were considerably less in hypergly-
and C groups for all the implant surfaces. ⁄Statistical significance at P < 0.05; cemic animals compared with the controls. Histological examina-
⁄⁄
statistical significance at P < 0.01. (B) Box plot of BIC for all implant surfaces in tion of the hyperglycemic samples did not reveal any obvious
both metabolic groups. The box stretches from the 25th percentile to the 75th
percentile. The median is shown as a line across the box, and the average is shown
signs of unmineralized bone 5 days post operation. Such differ-
as a symbol inside the box. ences diminished by 10 days, since in the control group reparative
bone was starting to be resorbed, while in the hypergylcemic group
delayed reparative bone was still forming. Clearly, such temporal
Table 1 changes demonstrate that a hyperglycemic animal may display
Statistical differences in BIC% values between all implant surfaces (A) in the C group, either less or more bone than a healthy animal, depending on the
(B) in the H group, and (C) between the C and H groups. time of examination, which may provide an explanation for the
C animals MAE DAE NAT contradictory findings on bone healing in hyperglycemic and
euglycemic populations.
(A)
MAE – – –
However, while the restoration of homeostasis was clearly
DAE 0.1261 – – delayed in hyperglycemic animals, fluorescence microscopy of
NAT 0.0024 0.2063 – reparative bone in these animals also showed considerably less
DCD <.0001 0.0155 0.1494 labeling by both alizarin red and tetracycline. Since these fluoro-
phores both label the mineralizing front in newly forming bone,
H animals MAE DAE NAT
and although this study did not examine bone mineralization
(B) dynamics in particular, a consistently lower fluorescence would
MAE – – –
DAE 0.3719 –
suggest that in hyperglycemic animals the reparative bone was less
NAT <.0001 0.0203 – mineralized than in the control group. A similar absence of fluoro-
DCD <.0001 0.0043 0.3200 chrome labels in the bone of diabetic animals has been reported
MAE-H DAE-H NAT-H DCD-H previously [20], and severe mineralization disorders, such as a
(C)
decrease in all fluorochrome-based parameters of mineralization,
MAE-C 0.6383 0.0849 0.0002 <.0001 apposition, formation, and timing of mineralization, have been
DAE-C <.0001 0.1735 0.6437 0.2950 observed within the first 14 days in diabetic animals [21].
NAT-C <.0001 <0.00021 0.6134 0.6095 There are several mechanisms that could contribute to such com-
DCD-C <.0001 <.0001 0.0003 0.3134
promised healing, specifically delayed osteoconduction, in a hyper-
Numerical data are P values. Significant P values are shown in bold. glycemic environment. In normal wound healing the blood clot is
populated by cells from the peripheral blood, red blood cells, leuko-
cytes and platelets. The latter have been implicated in diabetic
To further confirm these results we designed an additional pathologies [22–24], but it has been shown that hyperglycemia
method of analysis to look at the bone growth distribution inside has deleterious effects on neutrophils [25,26], macrophages [27],
individual chambers. For this two maps of bone growth distribu- and lymphocytes [28], all of which populate the wound site at early
tion were generated per implant (a total of 384 maps) based on healing times. In addition, those local vessels that remain intact will
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405 403

Fig. 8. (A) (Top) A representation of the T-plant with the two detached walls. Bone that grows in contact with the walls of the implant is shown in yellow and the cross-
sectional layers subjected to backscatter electron imaging are shown as lines along the walls. (Bottom) Maps of bone distribution on the walls of a T-plant. Two maps were
generated per T-plant (one map per wall) based on the BIC% values measured on each wall from both anatomical sides, and the maps were visualized for bone distribution
phenomena. (B) Histogram plot of the percent probability of BIC% values being above or below the median (24.2%) for all implant surfaces in both the H (red) and C (blue)
groups. (C) Statistical differences in the probability of BIC% falling above the median (24.2%) between all implant surfaces in both metabolic groups. Numerical data are P
values. Significant P values are shown in bold.

demonstrate endothelial abnormalities that limit leukocyte adhesion
and diapedesis [29]. In addition, diabetes has been referred to as a
hypercoagulative state [30], in which there is enhanced thrombin for-
mation, platelet activation, and fibrin clot resistance to lysis [31], all of
which would lead to a delay in clot resolution. Such a delay in clot res-
olution was further confirmed by histological examination of our
hyperglycemic samples from the osteotomy model, in which the
medullary compartment was occupied by a blood clot in the early
stages of healing, while it was filled with the new bone in the control
samples, with no signs of residual clot.
The delay in clot resolution will impair osteogenic signaling
pathways by obstructing the chemokine transmission path,
specifically those released from platelets. The latter have been
shown to stimulate the migration of pericytes, the tissue-resident
mesenchymal stem cells that in bone differentiate into osteoblasts
and elaborate the reparative bony tissue. Upon injury platelets are
activated and release cytokines and growth factors which stimu-
late the recruitment and migration of osteogenic cells [32], and
thus play a key role in up-regulating osteogenic responses by tak-
ing part in the signaling mechanisms of osteogenic cells. Secondary
to platelet activation, recruitment and migration of pericytes (oste-
ogenic precursors) to the implant surface is a critical step in
peri-implant bone healing as, together with peri-implant bone for-
mation, it results in contact osteogenesis [33]. Indeed, there is a
404 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405
clear evidence that diabetes is associated with changes in both the
platelet and pericyte populations.
It has been reported that platelets in diabetes are ‘‘hyperactive’’,
i.e. they show intensified adhesion, activation, and aggregation
[34,35], which we speculate could cause premature pre-implant
platelet activation of platelets, which together with reduced endo-
thelial cell migration and angiogenesis could cause chemokines to
be depleted before the pericyte population was available to insti-
gate the process of osteoconduction.
Indeed,diabetes has also been shown to be associated with per-
icyte loss in retinal capillaries [36], but it is still controversial
whether this is due to local disorders in the retina (possibly hemo-
dynamic) or is associated with hyperglycemia, as it has been
shown that glycemic control in dogs inhibited pericyte loss [37].
Apoptosis [38] and destructive signaling pathways within peri-
cytes [39] are among many reasons for retinal pericyte loss leading
to retinopathy under hyperglycemic conditions. Although the exact
causes of pericyte loss during early diabetic retinopathy remain
unclear, some of the mechanisms involved in pericyte recruitment
and depletion have been discussed [40]. Apoptosis of pericytes has
also been observed in diabetic skin wounds [41] and, thus, it is rea-
sonable to consider that the pericyte population in the peri-
implant bone healing compartment in hyperglycemia may be com-
promised in a similar manner to that reported in retina and skin.
For our bone in-growth implant model we chose the 9 day time
point based on our previous experiments carried out in healthy
animals [19], and observed no differences in mean BIC between
hyperglycemic and healthy animals for all surface groups. Clearly,
the single time point chosen for our implant study was less than
ideal, because, as seen in the osteotomy model, both metabolic
groups had comparable amounts of bone. Indeed, while it is gener-
ally agreed that bone healing is compromised in diabetes, contro-
versies exist in terms of peri-implant bone healing, which mainly
arise from the differences in the time points studied. Most of the
existing literature has addressed such effects at later stages of
peri-implant bone healing (P10 days) and have shown less BIC
in hyperglycemic compared with control animals after 21 [18],
28 [16,42,43], 56 [16,43,44] and 84 [43] days healing, while others
have observed no differences after 90 days [45]. However, there are
a few investigations of the early stages of peri-implant healing
(<10 days) [13,18,43] that have reported an increased total bone
volume [13], no change [18], or a lower BIC in hyperglycemia
[43]. Similarly, we did not see any differences in BIC between
hyperglycemic and healthy animals at 9 days post operation, nev-
ertheless, the differences in the implant surface topography did re-
sult in changes in osteoconduction in both metabolic groups. There
was a higher BIC in hyperglycemic compared with control animals
because of the delayed healing associated with hyperglycemia,
while bone remodeling had already commenced in the controls,
as deduced from the osteotomy model. Nevertheless, there were
clear differences between the nano and micro surfaces. This is in
line with our previous findings in which the addition of DCD
nanocrystals to the implant surface resulted in an increase in BIC
compared with those without DCD in healthy conditions [19].
The increased probability of osteoconduction on the nano surfaces
compared with the micro surfaces could not be due to the chemis-
try of CaP nanocrystals for two reasons. First, the performance of
the NAT surface, which was devoid of CaP, was statistically indis-
tinguishable from that of the DCD surface. Second, we have previ-
ously shown that DCD nanocrystals are stable and, while providing
a complex nanotopography, do not dissolve when exposed to body
fluid in vivo for up to 1 month [46]. Nevertheless, surface chemis-
try could have played a role in the marginal, but statistically
insignificant, differences between the two nano surfaces.
Another important observation is that the MAE surface showed
the poorest performance. It should be noted that the nanotopo-
graphic features of MAE are subtractive, unlike the additive nanot-
opography of the NAT and DCD surfaces, which obliterates some of
the features of the initial microtopographic surface. This empha-
sizes the significance of maintaining a certain degree of microto-
pography while increasing the nanotopographic complexity of
the implant surface, and therefore demonstrates the predominant
role of different scales of surface topography in the phenomenon
of osteoconduction, which has recently been shown to be a critical
determinant in the bone bonding behavior of the implant surface
[47]. Finally, it is salutary to point out that the differences we dem-
onstrate between the performances of the nano and micro surfaces
are not reflected in the results of the optical interferometry. This is
to be expected, since mathematical ‘‘roughness’’ values bear little
relation to the biological reaction to surfaces, as we have discussed
in detail elsewhere [47].
5. Conclusion
This study showed delayed bone formation and remodeling in
hyperglycemia and demonstrated the dependence of the amount
of reparative bone on the study time point, which emphasizes
the importance of tracking temporal changes when elucidating
the effects of hyperglycemia on bone healing. Increased implant
surface topographical complexity resulted in enhanced osteocon-
duction in hyperglycemia. Not only did the nano surfaces perform
equally well under both healthy and hyperglycemic conditions, the
nano surfaces under hyperglycemic conditions outperformed the
micro surfaces under healthy conditions.
Disclosures
This work was supported by Biomet 3i (Palm Beach Gardens,
FL). S.B. is the recipient of an Industrial Postgraduate Scholarship,
provided by the National Sciences and Engineering Research
Council of Canada (412028).
Acknowledgements
The authors would like to thank Biomet 3i for financial support
and the manufacture of all custom implants. The assistance of
Susan Carter with animal care, Dr Mucio S. Reis Junior with the
preparation of resin blocks, and Dr Zhenmei Liu, Dr Ashish Khadka,
and Mr Brian Merchant with the histology is greatly appreciated.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.actbio.2013.09.
020.
Appendix B. Figures with essential color discrimination
Certain figures in this article, particularly Figs. 1–4, 7 and 8, are
difficult to interpret in black and white. The full color images can
be found in the on-line version, at doi: http://dx.doi.org/10.1016/
j.actbio.2013.09.020.
References
[1] Maitra A. The endocrine pancreas: diabetes mellitus. In: Kumar VAA, Fausto N,
Aster JC, editors. Robbins and Cotran pathologic basis of disease. Philadeliphia,
PA: Saunders Elsevier; 2010.
[2] Centers for Disease Control and Prevention. National diabetes fact sheet:
national estimates and general information on diabetes and prediabetes in the
United States, 2011. Atlanta, GA: U.S. Department of Health and Human
Services, Centers for Disease Control and Prevention, 2011.
 E. Ajami et al. / Acta Biomaterialia 10 (2014) 394–405
[3] Adami S. Bone health in diabetes: considerations for clinical management. Curr
Med Res Opin 2009;25:1057–72.
[4] Dixit P, Ekstrom R. Decreased breaking strength of diabetic rat bone and its
improvement by insulin treatment. Calcif Tissue Int 1980;32:195–9.
[5] Marchand F, Raskin A, Dionnes-Hornes A, Barry T, Dubois N, Valéro R, et al.
Dental implants and diabetes: conditions for success. Diabetes Metab
2012;38:14–9.
[6] Moy PK, Medina D, Shetty V, Aghaloo TL. Dental implant failure rates
and associated risk factors. Int J Oral Maxillofac Implants 2005;20:
569–77.
[7] Oates TW, Dowell S, Robinson M, McMahan CA. Glycemic control and implant
stabilization in type 2 diabetes mellitus. J Dent Res 2009;88:367–71.
[8] Einhorn TA, Boskey AL, Gundberg CM, Vigorita VJ, Devlin VJ, Beyer MM. The
mineral and mechanical properties of bone in chronic experimental diabetes. J
Orthop Res 1988;6:317–23.
[9] Hofbauer LC, Brueck CC, Singh SK, Dobnig H. Osteoporosis in patients with
diabetes mellitus. J Bone Miner Res 2007;22:1317–28.
[10] Brem H, Tomic-Canic M. Cellular and molecular basis of wound healing in
diabetes. J Clin Investig 2007;117:1219–22.
[11] Salerno M, Argenziano A, Di Maio S, Gasparini N, Formicola S, De Filippo G,
et al. Pubertal growth, sexual maturation, and final height in children with
IDDM. Effects of age at onset and metabolic control. Diabetes Care
1997;20:721–4.
[12] Graves DT, Alblowi J, Paglia DN, O’Connor JP, Lin S. Impact of diabetes on
fracture healing. J Exp Clin Med 2011;3:3–8.
[13] McCracken MS, Aponte-Wesson R, Chavali R, Lemons JE. Bone associated with
implants in diabetic and insulin-treated rats. Clin Oral Implants Res
2006;17:495–500.
[14] Hasegawa H, Ozawa S, Hashimoto K, Takeichi T, Ogawa T. Type 2 diabetes
impairs implant osseointegration capacity in rats. Int J Oral Maxillofac
Implants 2008;23:237–46.
[15] Margonar R, Sakakura CE, Holzhausen M, Pepato MT, Alba JR, Marcantonio JE.
The influence of diabetes mellitus and insulin therapy on biomechanical
retention around dental implants: a study in rabbits. Implant Dent
2003;12:333–9.
[16] Nevins ML, Karimbux NY, Weber HP, Giannobile WV, Fiorellini J. Wound
healing around endosseous implants in experimental diabetes. Int J Oral
Maxillofac Implants 1998;13:620–9.
[17] Schlegel KA, Prechtl C, Most T, Seidl C, Lutz R, von Wilmowsky C.
Osseointegration of SLActive implants in diabetic pigs. Clin Oral Implants
Res 2013;24:128–34.
[18] Siqueira JT, Vavalher-Machado SC, Arana-Chavez VE, Sannomiya P. Bone
formation around titanium implants in the rat tibia: role of insulin. Implant
Dent 2003;12:242–51.
[19] Mendes VC, Moineddin R, Davies JE. Discrete calcium phosphate
nanocrystalline deposition enhances osteoconduction on titanium-based
implant surfaces. J Biomed Mater Res 2009;90A:577–85.
[20] Shyng YC, Devlin H, Sloan P. The effect of streptozotocin-induced experimental
diabetes mellitus on calvarial defect healing and bone turnover in the rat. Int J
Oral Maxillofac Surg 2001;30:70–4.
[21] Follak N, Klöting I, Wolf E, Merk H. Histomorphometric evaluation of the
influence of the diabetic metabolic state on bone defect healing depending on
the defect size in spontaneously diabetic BB/OK rats. Bone 2004;35:144–52.
[22] Aoki I, Shimoyama K, Aoki N, Homori M, Yanagisawa A, Nakahara K, et al.
Platelet-dependent thrombin generation in patients with diabetes mellitus:
effects of glycemic control on coagulability in diabetes. J Am Coll Cardiol
1996;27:560–6.
[23] Gresele P, Guglielmini G, De Angelis M, Ciferri S, Ciofetta M, Falcinelli E, et al.
Acute, short-term hyperglycemia enhances shear stress-induced platelet
activation in patients with type II diabetes mellitus. J Am Coll Cardiol
2003;41:1013–20.
[24] Kakouros N, Rade JJ, Kourliouros A, Resar JR. Platelet function in patients with
diabetes mellitus: from a theoretical to a practical perspective. Int J Endocrinol
2011;2011:1–14.
405
[25] Alba-Loureiro TC, Munhoz CD, Martins JO, Cerchiaro GA, Scavone C, Curi R,
et al. Neutrophil function and metabolism in individuals with diabetes
mellitus. Braz J Med Biol Res 2007;40:1037–44.
[26] de Souza Ferreira C, Araújo TH, Ângelo ML, Pennacchi PC, Okada SS, de Araújo
Paula FB, et al. Neutrophil dysfunction induced by hyperglycemia: modulation
of myeloperoxidase activity. Cell Biochem Funct 2012;30:604–10.
[27] Liu YJ, Saini A, Cohen DJ, Ooi BS. Modulation of macrophage proliferation by
hyperglycemia. Mol Cell Endocrinol 1995;114:187–92.
[28] Sakowicz-Burkiewicz M, Kocbuch K, Grden M, Szutowicz A, Pawelczyk T.
Protein kinase C mediated high glucose effect on adenosine receptors
expression in rat B lymphocytes. J Physiol Pharmacol 2009;60:145–53.
[29] Kolluru GK, Bir SC, Kevil CG. Endothelial dysfunction and diabetes: effects on
angiogenesis, vascular remodeling, and wound healing. Int J Vasc Med
2011;2012:1–30.
[30] Carr ME. Diabetes mellitus: a hypercoagulable state. J Diabetes Complications
2001;15:44–54.
[31] Undas A, Wiek I, Stêpien E, Zmudka K, Tracz W. Hyperglycemia is associated
with enhanced thrombin formation, platelet activation, and fibrin clot
resistance to lysis in patients with acute coronary syndrome. Diabetes Care
2008;31:1590–5.
[32] Oprea WE, Karp JM, Hosseini MM, Davies JE. Effect of platelet releasate on
bone cell migration and recruitment in vitro. J Craniofac Surg
2003;14:292–300.
[33] Davies JE. Understanding peri-implant endosseous healing. J Dent Educ
2003;67:932–49.
[34] Ferreiro JL, Gómez-Hospital JA, Angiolillo DJ. Platelet abnormalities in diabetes
mellitus. Diab Vasc Dis Res 2010;7:251–9.
[35] Ferroni P, Basili S, Falco A, Davi G. Platelet activation in type 2 diabetes
mellitus. J Thromb Haemost 2004;2:1282–91.
[36] Engerman RL. Perspectives in diabetes – pathogenesis of diabetic retinopathy.
Diabetes Care 1989;38:1203–6.
[37] Engerman L, Bloodworth JMB, Nelson S. Relationship of microvascular disease
in diabetes to metabolic control. Diabetes Care 1977;26:760–9.
[38] Mizutani M, Kern TS, Lorenzi M. Accelerated death of retinal microvascular
cells in human and experimental diabetic retinopathy. J Clin Invest
1996;97:2883–90.
[39] Kowluru RA, Odenbach S. Effect of long-term administration of alpha-lipoic
acid on retinal capillary cell death and the development of retinopathy in
diabetic rats. Diabetes Care 2004;53:3233–8.
[40] Ejaz S, Chekarova I, Ejaz A, Sohail A, Lim CW. Importance of pericytes and
mechanisms of pericyte loss during diabetic retinopathy. Diabetes Obes Metab
2008;10:53–63.
[41] Darby IA, Bisucci T, Hewitson TD, MacLellan DG. Apoptosis is increased in a
model of diabetes-impaired wound healing in genetically diabetic mice. Int J
Biochem Cell Biol 1997;29:191–200.
[42] Oliveira PA, Oliveira AM, Pablos AB, Costa FO, Silva GA, Santos JN, et al.
Influence of hyperbaric oxygen therapy on peri-implant bone healing in rats
with alloxan-induced diabetes. J Clin Periodontol 2012;39:879–86.
[43] Takeshita F, Murai K, Iyama S, Ayukawa Y, Suetsugu T. Uncontrolled diabetes
hinders bone formation around titanium implants in rat tibiae. A light and
fluorescence microscopy, and image processing study. J Periodontal
1998;69:314–20.
[44] Ottoni CEC, Chopard RP. Histomorphometric evaluation of new bone
formation in diabetic rats submitted to insertion of temporary implants.
Braz Dent J 2004;15:87–92.
[45] Retzepi MR, Lewis MP, Donos N. Effect of diabetes and metabolic control on de
novo bone formation following guided bone regeneration. Clin Oral Implants
Res 2010;21:71–9.
[46] Pezeshki P, Lugowski S, Davies JE. Dissolution behaviour of calcium phosphate
nanocrystals deposited on titanium alloy surfaces. J Biomed Mater Res
2012;94A:660–6.
[47] Davies JE, Ajami E, Moineddin R, Mendes VC. The roles of different scale ranges
of surface implant topography on the stability of the bone/implant interface.
Biomaterials 2013;34:3535–46.