J Clin Periodontol 2008; 35: 630–639 doi: 10.1111/j.1600-051X.2008.01246.

Modeling and remodeling of Leonardo Trombelli1, Roberto

Farina1, Andrea Marzola2, Leopoldo
Bozzi3, Birgitta Liljenberg4 and

human extraction sockets Jan Lindhe4

1
Research Centre for the Study of
Periodontal Diseases, University of Ferrara,
Ferrara, Italy; 2Department of Experimental
Trombelli L, Farina R, Marzola A, Bozzi L, Liljenberg B, Lindhe J. Modeling and and Diagnostic Medicine, Section of
remodeling of human extraction sockets. J Clin Periodontol 2008; 35: 630–639. doi: Histologic and Citologic Anatomy, University
of Ferrara, Ferrara, Italy; 3Private Practice,
10.1111/j.1600-051X.2008.01246.x.
Comacchio, Italy; 4Department of
Periodontology, Faculty of Odontology, The
Abstract
Sahlgrenska Academy at Göteborg
Introduction: The available studies on extraction wound repair in humans are University, Göteborg, Sweden
affected by significant limitations and have failed to evaluate tissue alterations
occurring in all compartments of the hard tissue defect.
Aim: To monitor during a 6-month period the healing of human extraction sockets
and include a semi-quantitative analysis of tissues and cell populations involved in
various stages of the processes of modeling/remodeling.
Material and Methods: Twenty-seven biopsies, representative of the early (2–4
weeks, n 5 10), intermediate (6–8 weeks, n 5 6), and late phase (12–24 weeks, n 5 11)
of healing, were collected and analysed.
Results: Granulation tissue that was present in comparatively large amounts in the
early healing phase of socket healing, was in the interval between the early and
intermediate observation phase replaced with provisional matrix and woven bone. The
density of vascular structures and macrophages slowly decreased from 2 to 4 weeks
over time. The presence of osteoblasts peaked at 6–8 weeks and remained almost
stable thereafter; a small number of osteoclasts were present in a few specimens at
each observation interval.
Key words: alveolar ridge; bone healing;
Conclusions: The present findings demonstrated that great variability exists in man histomorphometric analysis; tissue modeling/
with respect to hard tissue formation within extraction sockets. Thus, whereas a remodeling; tooth extraction
provisional connective tissue consistently forms within the first weeks of healing, the
interval during which mineralized bone is laid down is much less predictable. Accepted for publication 6 April 2008

The alveolar process forms during tooth amount of tissue that is lost in the this model, the socket already after 4
eruption (Schroeder 1986) and under- lingual/palatal portion. The alteration weeks of healing was filled with woven
goes atrophy after the loss of single or of the ridge occurs concomitantly with bone and that after 2 months this imma-
multiple teeth (e.g. Pietrokovski & the healing of the soft and hard tissue ture bone had been replaced with lamel-
Massler 1967, Johnson 1969, Schropp wound but the process of remodeling lar bone and marrow. Concomitant with
et al. 2003, Pietrokovski et al. 2007). As may continue also after the termination these intra-alveolar healing events, the
a rule, the resorption of the buccal of de novo bone formation in the buccal wall of the socket underwent
compartment of the ridge – after tooth socket(s) (Schropp et al. 2003). marked resorption. Thus, in comparison
removal – is more pronounced than the Processes involved in the healing of to the lingual wall of the socket, the
an extraction socket have been studied height of its buccal counterpart was
in different animal models (e.g. Kuboki reduced on the average 2.5 mm (Araujo
Conflict of interest and source of et al. 1988, Lin et al. 1994, Lekic et al. & Lindhe 2005).
funding statement 2001, Cardaropoli et al. 2003, 2005, Tissue formation after tooth extrac-
The authors declare that they have no Kanayama et al. 2003, Sato & Takeda tion was also studied in human models
conflict of interest. 2007). In a recent study using a dog (Claflin 1936, Mangos 1941, Christopher
This study was supported by the Research model, Cardaropoli et al. (2003) 1942, Amler et al. 1960, 1964, 1969,
Centre for the Study of Periodontal Dis- explored the healing of the coronal, Boyne 1966, Evian et al. 1982). Amler
eases, University of Ferrara, Italy, and by a central and apical compartments of fresh (1969) examined new tissue formation
grant from the University of Ferrara, Italy extraction sockets during a 6-month in the marginal portion of extraction
(STAMINA). interval. The authors observed that in sites from human volunteers at healing
630 r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
 Healing of human post-extraction sockets 631

intervals extending from 2 to 32 days. intention. When needed, partial thick- tion site’’). The tissue was obtained
He concluded that the blood clot that ness dissection in the most apical por- from the centre of the socket, with the
initially filled the entrance of the socket tion of the flap was performed in order insertion axis of the trephine kept par-
was first replaced with granulation tis- to coronally advance the flap. Local allel to the long axis of the adjacent
sue. After 1 week of tissue modeling, antimicrobial therapy (chlorhexidine tooth. The apical portion of the speci-
osteoid formation had begun and after 0.12% mouthrinse; 10 ml for 60 s t.i.d. men was marked with a fine indelible
about 6 weeks, the marginal portion of for 7 days) was prescribed. Sutures were chine pen.
the socket harboured islands of imma- removed 1 week postoperatively. If a
ture woven bone. However, the human local complication (e.g. rupture of soft
Histological and immunohistochemical
studies cited above (i) included tissue tissue wound) occurred during healing processing
samples from systemically diseased and/or systemic antibiotic therapy had to
individuals or even cadavers (Claflin be prescribed, the site was excluded The specimens were immediately
1936, Mangos 1941), (ii) had an unclear from the study. placed in a 10% formaldehyde fixative,
experimental protocol (Christopher decalcified in ethylene diaminetetraacetic
1942), (iii) evaluated only few tissue Localization of the extraction site
acid dehydrated in increasing con-
samples and/or (iv) short observation centrations of ethanol, embedded in
intervals (Amler et al. 1960, Boyne Before tooth extraction, an alginate paraffin and cut in the sagittal plane
1966, Evian et al. 1982). impression (Xantalgins; Heraeus Kulzer with the microtome set at 5 mm. From
On the basis of the above concerns, a S.r.l., Milan, Italy) was obtained of the each biopsy, between 10 and 20 sections
study was designed to (i) monitor the experimental area. On a cast model, a representing the central part of the sock-
healing of human extraction sockets and resin/silicon stent was prepared and et were selected for histological and
(ii) include a semi-quantitative analysis adapted on the dentition adjacent to the immune-histochemical examinations.
of tissues and cell populations involved tooth to be removed. A canal parallel One section was stained in haema-
in various stages of the healing process. with the long axis of the tooth was toxyline–eosine, and examined in a
prepared in the stent. This allowed the microscope (Leitzs DM-RBE Micro-
operator to properly identify the socket scope; Leica, Wetzlar, Germany)
Material and Methods site during biopsy. equipped with an image system (Q-500
Patient and site selection Immediately after tooth extraction, MCs; Leica).
the distance from the stent to the bottom The total cross-section area of the
Subjects were considered eligible for the of the socket was measured. This was tissue sample was outlined and the size
study if they were X18-years old and if made with a UNC-15 periodontal probe. determined by planimetry. Subse-
they did not suffer from systemic dis- This recording was used in order to quently, the areas (proportions of the
ease. Pregnant or lactating women, and ensure that, at the time of biopsy, the cross-section area) occupied by (1) clot
subjects taking (past or currently) drugs trephine bur would collect the newly (erythrocytes, leukocytes embedded in a
influencing bone metabolism were formed tissue only. fibrin network), (2) granulation tissue
excluded. The study was approved by (rich in newly formed vascular struc-
the Ethical Committee of the University tures, abundance of inflammatory cells
Collection and storage of the specimens
of Ferrara. All included subjects such as neutrophils, macrophages, lym-
provided an informed consent before The timing of the biopsy procedure was phocytes as well as erythrocytes), (3)
participation. selected in accordance with the clinical provisional matrix (densely packed
The following tooth sites were con- need for the surgical procedure (i.e. mesenchymal cells, collagen fibres and
sidered: single rooted-teeth, molars or periodontal surgery or implant position- vessels but no or only scattered inflam-
premolars with fused roots or single ing) involving the extraction socket. At matory cells), (4) woven bone (finger-
roots of molars. Biopsy was consistently the time of biopsy retrieval, local infil- like projections of immature bone
performed in conjunction with perio- tration of anaesthetic was administered. embedded in a primary spongiosa) and
dontal surgery or implant placement in Whenever the concomitant surgical pro- (5) lamellar bone and marrow, were
the extraction site. Teeth with (i) peria- cedure required a full-thickness flap, a calculated.
pical lesion were excluded as well as (ii) crestal incision was performed and a A series of paraffin sections were de-
sites with limited depth (o5 mm) of the mucoperiosteal flap was elevated on paraffinized, treated with TE-buffer (pH
remaining sockets after tooth removal. both buccal and lingual/palatal side 9) or Proteinase K (Dakacytomation,
with a microsurgical periosteal elevator Glostrup, Denmark) and washed in
Extraction and post-extraction protocol
(P-TROM; Hu-Friedy, Chicago, IL, Tris-buffered saline (TBS) for 3 min.
USA). When this was not the case Endogenous peroxidase activity was
Full thickness buccal and lingual muco- (e.g. implant positioning with a flapless blocked with a peroxidase solution
periosteal flaps were elevated. The tooth approach), the soft tissues overlying the for 10 min. and subsequently rinsed
was carefully removed with the use of extraction socket were included in the for 3 min. with TBS. A panel of mono-
elevators. No ostectomy or osteoplasty biopsy. The tissue specimens were col- clonal antibodies diluted in TBS
was performed. No graft or membrane lected with a trephine bur (internal was applied for 30 min. and washed in
device was used to augment the site. diameter 2 mm; Hu-Friedy) from the TBS for 2  3 min. The sections were
Flaps were sutured with 5/0 silk or healing sockets. The depth of the tre- incubated with a peroxidase labelled
Vycrils suture (ETHICON, Johnson & phine bur insertion was related to the polymer for 30 min. and washed in
Johnson, Cincinnati, OH, USA). The measurements previously made by the TBS for 3  3 min. Diaminobenzidine
wound was left to heal for primary stent (see ‘‘Localization of the extrac- tetrahydrochloride (DAB) was used as
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
632 Trombelli et al.

Table 1. Isotypes and dilutions of the monoclonal antibodies used for immuno-histochemical disease (n 5 5), or for prosthetic reasons
stainings (n 5 2).
Antibodies (clone) Specificity Dilutions Isotype Source Of the 27 biopsies that were available
for analysis, 10 were sampled to repre-
CD 31 (JC70A) Endothelial cell 1:20 IgG1 Dakocytomation, sent early healing (2–4 weeks), six
Glostrup Denmark intermediate (6–8 weeks), and 11 a later
BMP7 (164313) BMP7 1:20 IgG1 R&D Systems, Europe Ltd
phase (12–24 weeks) of healing.
Oxon, UK
Osteocalcin (190125) Osteoblast 1:10 IgG1 R&D Systems, Europe Ltd
CD68 Macrophage 1:20 IgG1 Diagnostic Biosystems, Gross morphological characteristics of
Pleasanton, CA, USA the tissue samples
RANK/TNFRSF11A Osteoclast 1:20 IgG2A R&D Systems, Europe Ltd
(80707) In most specimens examined, it was
possible to distinguish between different
types of tissues. Thus, in some regions
the biopsy could disclose features typi-
cal of granulation tissue whereas in
substrate/chromogen (EnVision1s 3 450% buto75% other regions of the same specimen,
System-HRP (DAP) Kit; Dakocytoma- 4 480% tissue modeling was more advanced
tion). The sections were rinsed in TBS and areas including provisional matrix
and distilled water, counterstained in and/or woven bone could be identified.
Mayer’s haematoxyline stain and Granulation tissue was characterized
mounted. For each antibody tested, one CD68 by the presence of large numbers of
section was incubated in serum and vascular structures in a connective tissue
served as a negative control. CD68 score was related to the number
comprising mesenchymal cells and infil-
The various antibodies were used to of CD68-stained cells in the entire tissue
trates of leukocytes (neutrophilic leuko-
identify vascular (endothelial cells) sample.
cytes, macrophages and lymphocytes)
structures (CD31), osteoblasts (BMP-7, (Fig. 1). The inflammatory cells were
Osteocalcin) osteoclasts (RANK/ 1 o20 cells mainly present in peri-vascular compart-
TNFRSF 11A) and macrophages, giant 2 30–50 cells ments (Fig. 2) but occurred also in areas
cells (CD68). The monoclonal antibo- 3 50–100 cells more distant to vascular structures.
dies used, their isotypes and dilutions 4 4200 cells Provisional matrix contained densely
are presented in Table 1. packed mesenchymal cells present in a
All the histologic evaluations and histo-
morphometric measurements were car- collagen-rich connective tissue matrix
Semi-quantitation ried out by a single trained and (Fig. 3). Vascular structures (Fig. 4)
calibrated examiner (B. L.). Data were were abundant but only few infiltrates
The morphometric assessment on expressed as mean  SD. of leukocytes (macrophages and lym-
CD31-stained sections was conducted phocytes) could be observed. Osteo-
according to methods described earlier blasts (osteocalcin-stained cells)
(Schroeder & Münzel-Pedrazzoli 1973). occurred in peri-vascular locations.
Briefly, the cross-section area of the Results Woven bone occurred as fingerlike
tissue samples was determined (1.6) Twenty-four patients (10 males and 14 projections of mineralized tissue (Fig.
and vascular structures were subse- females, range 34–72 years) were 5) in a connective tissue matrix. The
quently identified and enumerated included. Twenty-two patients contribu- ridges of woven bone were lined with
(10). The numerical density of vascu- ted with one, one patient with two and one osteoblasts and contained large numbers
lar units per area unit (mm2) of available patient with three extraction sites. All the of osteocytes (Fig. 6). The trabeculae
tissue was then calculated. biopsy procedures were performed with- of mineralized bone occurred in the
A semi-quantitative analysis of BMP- out any major intra-operative complica- vicinity of, often surrounding, one or
7-, osteocalcin- and CD68-positive cells tions. Occasionally, the sampled tissue several vascular structures (Fig. 7).
was conducted. For each investigated fragmented into two to three separate Osteoclasts (RANK/TNFRSF 11A-
protein/cell type, a score ranging from pieces during biopsy harvesting, espe- positive cells) occurred in discrete areas
0 (absent) to 4 (large amounts) was cially when newly formed hard tissue at the surface of woven bone trabeculae
determined. was present in the extraction socket. and were consistently present in How-
When this was the case, all the pieces ship’s lacunae (Fig. 8).
BMP-7 and osteocalcin were retrieved, oriented, stored together Only one section (12 weeks of heal-
and analysed as a single sample. ing) (Fig. 9) comprised lamellar bone
BMP-7 and osteocalcin scores were The location of the extraction sites and marrow. The mineralized bone
related to the percentage of the periph- was as follows: 1 lateral incisor, 1 harboured secondary osteons whereas
ery of the woven bone projections that canine, 19 premolars, 6 molars. Twelve the bone marrow was rich in vessels,
showed the presence of BMP-7- and teeth were obtained from the upper jaw adipocytes, mesenchymal cells and
osteocalcin-stained cells, respectively. and fifteen from the lower jaw. Teeth inflammatory cells (macrophages, lym-
were extracted due to caries (n 5 8), phocytes). Osteoclasts were present in
1 o10% vertical root fracture (n 5 9), failure of several areas at the surface of the miner-
2 425% buto40% endodontic therapy (n 5 3), periodontal alized bone.
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
 Healing of human post-extraction sockets 633

could be observed (Table 2). In these

specimens, the richly vascularized, hea-
vy infiltrated tissue occupied between
67% and 100% of the samples exam-
ined. Six of the sections contained no
granulation tissue. The overall mean
area occupied by granulation tissue in
this healing interval was 35.9  47.2%
(Fig. 10).
Only one of the six biopsies repre-
senting the intermediate healing interval
(6–8 weeks; Table 2) contained a typical
granulation tissue, whereas no specimen
in the 12–24 week interval harboured
this type of immature tissue.

Provisional matrix
A tissue rich in mesenchymal cells,
collagen fibres and vascular structures
was found in 7 out of 10 specimens from
the early healing interval. In these speci-
mens, the provisional matrix occupied
between 33% and 100% of the tissue
examined with mean value 57.2 
44.2% (Table 2). All sections from the
intermediate healing period harboured
differently large zones (between 27%
Fig. 1. A biopsy obtained after 3 weeks of healing. The tissue is rich in vessels, fibroblasts and 98%; mean value 62.2  23.8%)
and inflammatory cells and is characterized as granulation tissue. Original magnification of provisional matrix. Ten out of eleven
2.5. Inset: large amounts of inflammatory cells can be observed immediately beneath the biopsies from the late healing interval
oral epithelium. Original magnification 10. H & E stain. contained a typical provisional matrix
tissue. In these sections, between 35%
and 92% of the area examined contained
this type of tissue with mean value
58.5  24.5%.

Woven bone and lamellar bone

Areas occupied by woven bone could be
observed already in specimens from the
2–4 week interval. Thus, in 6 out of 10
biopsies (Table 2) fingerlike projections
of newly formed bone could be identi-
fied. When present in the sections, this
immature bone occupied between 2%
and 34% of the tissue area (mean
6.9  10.5%; Fig. 10).
In practically all specimens represent-
ing the intermediate and late healing
interval, woven bone was present and
occupied between 2% and 73% of the
Fig. 2. Detail from Fig. 1 illustrating vascular structures that have been stained with CD 31.
tissue examined (mean value 6–8
Note the large amount of vessels of varying diameters including several vascular sprouts
(arrows). Original magnification 10. weeks: 34.0  24.6%; 12–24 weeks:
32.4  18.4%). Lamellar bone and mar-
row was found in only one biopsy and
Histomorphometric measurements the sections. However, no definitive clot this tissue represented the 12–24 week
formation could be distinguished. healing interval (Table 2).
The composition of each tissue sample
is shown in Table 2. Only in samples
Granulation tissue Immunohistochemical analysis
representing the early interval of healing
(2–4 weeks) could few, scattered ery- In 4 out of 10 samples representing the Table 3 shows the presence of positive
throcytes be observed in various parts of early healing interval, granulation tissue cells for each staining.
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
634 Trombelli et al.

(one to three cells) osteoclasts were

found in five out of the six samples. In
one sample, however, a large number of
osteoclasts (410 cells) were stained. In
specimens representing 12–24 weeks,
few osteoclasts were found in 6 out of
11 samples; i.e. five were negative.

Discussion
The overall findings from the present
study demonstrated that the healing of
an extraction socket, as disclosed in our
panel of subjects, followed a pattern
similar to that previously described
from histological examinations of tis-
sues harvested from socket sites in man
and animals (for review see Chen et al.
2004, Cardaropoli 2006). The process of
bone modeling/remodeling described in
the present biopsy samples has several
features in common with intra-membra-
neous bone formation (Whistson 1994)
and bone formation that occurs in mem-
Fig. 3. A biopsy obtained after 4 weeks of healing. The provisional matrix comprises brane protected defects in the alveolar
mesenchymal cells densely packed fibres and vessels. Only few inflammatory cells can be ridge (Schenk et al. 1994). Data also
observed. Original magnification 2.5. Inset: at the lateral border of the section, minute areas disclosed that (i) the rate of healing
of woven bone formation can be identified. Original magnification 10. H & E stain. varied markedly between subjects; (ii)
the process of remodeling, i.e. the repla-
cement of woven bone with lamellar
bone and marrow was slow. Only 1 of
the 11 specimens representing 12–24
weeks of healing comprised mature
bone, i.e. lamellar bone and marrow;
(iii) density of vascular structures was
high at 2–4 and 6–8 weeks and tended to
decrease in the 12–24 week specimens;
(iv) presence of osteoblasts peaked at 6–
8 weeks and remained almost stable
thereafter; (v) macrophages slowly
decreased from 2 to 4 weeks over time
and (vi) osteoclasts were present in
only few specimens at each observation
interval.
Immediately after tooth extraction, the
socket fills with blood and clot formation
occurs (Amler 1969, Cardaropoli et al.
2003). In the present material, erythro-
cytes scattered in between mesenchymal
Fig. 4. Detail from Fig. 3 illustrating vascular structures that have been stained with CD 31. cells were frequently observed in the
Original magnification 10. tissue from biopsies representing 2–4
weeks, although typical clot formations
(erythrocytes, platelets and leukocytes
The number of CD31 positive cells 6–8 weeks, and 1.9  0.8 at 12–24 entrapped in a dense fibrin network)
per area unit (mm2) was 33.0  12.2 at weeks. could not be observed in any of the
2–4 weeks, 27.2  5.6 at 6–8 weeks and CD68 score shifted from 2.8  1.0 at tissues examined. This finding is in
19.0  8.8 at 12–24 weeks. 2–4 weeks to 2.1  1.1 at 6–8 weeks, agreement with data presented by Amler
BMP-7 score shifted from 1.5  0.6 and 1.1  0.4 at 12–24 weeks. et al. (1960) and Amler (1969) who
at 2–4 weeks to 2.8  1.0 at 6–8 weeks, RANK/TNFRSF 11A-positive cells documented that the blood clot that first
and 2.1  1.2 at 12–24 weeks. were detected in low numbers (one to filled the socket space was almost
Osteocalcin score shifted from two cells) in only 3 out of 10 specimens entirely remodeled within the first
1.4  0.6 at 2–4 weeks to 2.5  1.0 at at 2–4 weeks. At 6–8 weeks, only few week after tooth removal. Furthermore,
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
 Healing of human post-extraction sockets 635

first and third week of socket healing

there was a progressive replacement of
granulation tissue with connective tis-
sue. Provisional matrix and woven bone
also dominated in biopsies from the late
phase of healing (12–24 weeks) of the
current study, and only one specimen
(representing 12 weeks) included sub-
stantial amounts of lamellar bone and
marrow. In other words, whereas tissue
modeling in the human extraction sites
appeared to be a comparatively fast
process, remodeling of the newly
formed hard tissue was seemingly slow.
The observation that bone remodeling
of extraction sites in man is a slow
process is in many respects consistent
with data presented by Schropp et al.
(2003). They used clinical and radio-
graphic means to monitor tissue altera-
tions within and outside extraction sites
in the premolar and molar regions of 46
patients. The data described by the
authors disclosed that the dimensions
of the extraction sites were markedly
reduced after tooth removal. Thus, at
3 months 430% and at 12 months 450%
of the buccal–lingual/palatal width of
the marginal portion of the ridge was
Fig. 5. Decalcified section obtained from a biopsy sampled after 6 weeks of healing. Note the lost after tooth removal. Moreover, it
presence of trabeculae of immature woven bone that occur in a cell and fibre-rich provisional was noted that large amounts of miner-
matrix. Original magnification 2.5. Inset: the projections of woven bone are enclosing alized tissue formed in the sockets dur-
vascular structures to form primary osteons. Original magnification 10. H & E stain. ing the first 6 months whereas in the
interval between 6 and 12 months, the
amount of newly formed mineralized
tissue was substantially reduced (woven
bone was replaced with lamellar bone
and marrow). The findings by Schropp
et al. (2003) thus demonstrate that heal-
ing of an extraction site in man proceeds
with the resorption of the marginal
socket walls as well as with varying
degree of hard tissue fill within the
socket. It is important to emphasize
that the biopsy technique used in the
current study limited the analysis to the
intra-alveolar events.
In a recent publication, Chen et al.
(2004) reported on osseous regeneration
in human extraction sockets and
included four studies with histologic
documentation in their review (Amler
et al. 1960, 1969, Boyne 1966, Evian
et al. 1982). Amler (1969) claimed that
Fig. 6. Detail from Fig. 5 illustrating osteoblasts that line the newly formed woven bone as after about 40 days of healing two-third
well as osteocytes within the immature bone. Osteocalcin staining. Original magnification of the sockets were filled with miner-
20.
alized bone, whereas Evian et al. (1982)
stated that the sockets were completely
most of the granulation tissue that weeks) apparently been replaced with filled with bone after 10 weeks of heal-
occurred in samples representing the provisional matrix and woven bone. ing. The findings from the current study
early phase (2–4 weeks) of healing, This observation corroborates findings could only in part corroborate the above
had in the interval between the early by Amler et al. (1960) and Evian et al. findings. Thus, in specimens obtained
and intermediate healing phase (6–8 (1982) who reported that between the between 6 and 8 weeks woven bone
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
636 Trombelli et al.

from the human extraction sites

included in the current study, this rather
distinct pattern of tissue modeling/remo-
deling was difficult to identify. In con-
trast, the results from the histological
examinations revealed great variation
between samples with respect to tissue
formation and maturation. The reason
for this variation is presently not under-
stood but may be linked to subject-
related factors, pre-existing damage to
the tooth and its supporting tissues (rea-
son for tooth extraction), trauma
inflicted to the tissues in conjunction
with tooth removal, the size of the
hard tissue defect, etc.
In our study, immunohistochemical
stainings were used to investigate the
time-dependent changes in density of
Fig. 7. Detail from Fig. 5. The tissue was stained with CD 31 and illustrates vascular the main cell types involved in post-
structures that reside in the centre of primary osteons. Original magnification  20. extraction wound healing. The density
of endothelial cells, assumed as corre-
sponding to the number of CD31-
positive cells, was decreased at 12–24
weeks with respect to 2–4 weeks and
6–8 week observations. This finding
can be explained by the progressive
remodeling of the highly vascularized
granulation tissue and its subsequent
replacement by provisional matrix.
The histomorphometric analysis
included the use of antibodies against
BMP-7 and osteocalcin. For these mea-
surements, only positive cells lining the
newly formed trabeculae of woven bone
were considered in order to limit the
study to the cells directly involved in the
bone forming and remodeling process.
For both antibodies, the density of posi-
tive cells increased from 2–4 weeks to
6–8 weeks and tended to decrease at
12–24 weeks. The increase in BMP-7
Fig. 8. Detail from Fig 5. Note the presence of osteoclasts on the surface of newly formed positive cells between the early and
bone. Stain: Rank/TNFRSF 11A. Original magnification  40. intermediate healing phase can be attrib-
uted to an increased bone modeling and
remodeling activity, leading to the
occupied about 35% of the tissue sample, formation in various parts of the socket deposition of woven bone from provi-
whereas after 12 and 24 weeks of wound followed a well-defined course. Thus, sional matrix. In this respect, previous
healing only about 41% of the biopsy the clot that formed immediately after studies demonstrated that BMP-7 is
harvested from the socket comprised tooth removal was in the first few days highly expressed during the early stages
mineralized bone. consistently replaced with granulation of bone repair after a fracture (Spector
Various aspects of healing of extrac- tissue that already after 1 week in part et al. 2001, Kloen et al. 2003). This
tion sockets were studied in the dog had been replaced with a cell rich protein continues to be expressed during
model (for review see Chen et al. provisional matrix. After 2 weeks of the process of bone remodeling, albeit
2004, Cardaropoli 2006). In such care- healing, large portions of the apical less prominently, and the return of
fully monitored animal experiments, (i) and lateral compartments of the socket BMP-7 density to baseline value coin-
the teeth were carefully removed from were occupied by woven bone and after cides with the histological appearance of
periodontally non-compromised sites 4 weeks, the process of modeling was mature lamellar bone (Spector et al.
and (ii) healing was analysed in biopsies advanced. Hence, in the interval 2001). In this respect, the limited pre-
containing not only the hard tissue between 4 and 8 weeks, the newly sence of lamellar bone in the 12–24
defect but in addition the surrounding formed immature woven bone was week specimens observed in the current
bone walls (mesial, distal, buccal, lin- replaced with lamellar bone and mar- study was associated with a tendency for
gual). In such experiments, new tissue row. In the tissue samples harvested decrease in BMP-7 levels.
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
 Healing of human post-extraction sockets 637

maturing bone matrix, osteocalcin

represents a late marker of bone forma-
tion. In fact, a greater expression of
osteocalcin-positive cells can be
observed in the mature bone matrix
than in newly formed bone (Ivanovski
et al. 2000, Reinhardt et al. 2005). It was
demonstrated that osteocalcin, used as a
marker to define alveolar bone turn-over
in humans, was negatively correlated
with increasing radiographic density of
experimentally produced bone cores, i.e.
the higher the osteocalcin expression the
less dense was the amount of new
mineralized bone (Reinhardt et al.
2004).
For obvious reasons, the current study
had a cross-sectional design. Most
patients contributed with only one
biopsy from one healing interval. Thus,
it can only be assumed that an immature
Fig. 9. Biopsy obtained from an extraction wound representing 12 weeks of healing. The tissue
tissue identified in biopsies from 2 to 4
comprises more mature bone; woven bone and lamellar bone that reside in a non-mineralized
matrix. Original magnification 2.5. H & E stain. Inset A: note the presence of secondary weeks interval would have matured to
osteons in lamellar bone. Original magnification 40. Inset B: large number of inflammatory form bone if continued healing would
cells among adipocytes in the newly regenerated bone marrow. Original magnification 40. have been allowed. Furthermore, cells
identified as osteoblasts or pre-osteoblasts
Table 2. Composition (percentage) of each sample with respect to clot, granulation tissue, – in sections stained with monoclonal
provisional connective tissue, woven bone and lamellar bone/bone marrow antibodies for osteocalcin and BMP-7 –
Sample Clot Granulation Provisional connective Woven bone Lamellar bone/bone varied in numbers from one biopsy to
no. (%) tissue (%) tissue (%) (%) marrow (%) the next (Table 3) but reached peak
values in sections representing 6–8
2–4 weeks weeks of healing. In such sections,
#1 0 100 0 0 0 however, a large number of typical
#2 0 92 0 8 0 mesenchymal cells were negative for
#3 0 0 89 11 0
the monoclonal antibodies used. This
#4 0 0 89 11 0
#5 0 0 97 3 0 may indicate that such cells were not
#6 0 0 100 0 0 in an active bone-forming phase or
#7 0 0 66 34 0 represented populations of cells not
#8 0 100 0 0 0 involved in the formation of bone. It is
#9 0 67 33 0 0 a well-known fact that cells that re-
#10 0 0 98 2 0 populate a wound determine the nature
6–8 weeks of the tissues formed (Melcher 1976).
#1 0 0 27 73 0
Thus, it cannot be taken for granted that
#2 0 23 57 20 0
#3 0 0 53 47 0 tissue portions in the present biopsy
#4 0 0 76 24 0 sample were occupied by provisional
#5 0 0 62 38 0 matrix would become replaced with
#6 0 0 98 2 0 woven bone.
12–24 weeks In our study, osteoclasts, when
#1 0 0 52 48 0 detected, were present in low numbers
#2 0 0 0 0 100 (one to three cells) at each observation
#3 0 0 92 8 0
#4 0 0 84 16 0
interval. In a dog study by Cardaropoli
#5 0 0 67 33 0 et al. (2003), a substantial remodeling
#6 0 0 35 65 0 activity with an increased number of
#7 0 0 64 36 0 osteoclasts was observed at 1 month
#8 0 0 58 42 0 after tooth extraction, leading to a
#9 0 0 60 40 0 decrease in the area occupied by miner-
#10 0 0 64 36 0 alized bone. In humans, a similar remo-
#11 0 0 68 32 0 deling activity was observed not before
6 months after tooth removal, i.e. at a
The osteocalcin expression paralleled tially confined to osteoblasts actively longer observation interval with respect
BMP-7 expression during the entire laying down new osteoid or remodeling to our study (Schropp et al. 2003).
observation interval. Osteocalcin is an bone (Paccione et al. 2001). Because of Therefore, it may be hypothesized that
osteoblast-specific marker, being spa- its progressive accumulation in the our study, due to earlier observation
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
638 Trombelli et al.

100 a well-defined tendency to change over

time. Apparently, the bone organization
and architecture was not completed at
24 weeks after tooth extraction.

% Acknowledgements
This study was supported by the
Research Centre for the Study of Perio-
dontal Diseases, University of Ferrara,
Italy, by a grant from the University of
0 Ferrara, Italy (STAMINA).
2−4 weeks 6−8 weeks 12−24 weeks

Fig. 10. Distribution (mean %) of the tissue components (clot, granulation tissue, provisional
connective tissue, woven bone, lamellar bone/bone marrow), calculated on all samples for
each observation interval. The standard deviation is shown in parenthesis.
References
Amler, M. H. (1969) The time sequence of
Table 3. Density of positive cells for immunohistochemical stainings (CD31, BMP-7, osteocalcin, tissue regeneration in human extraction
CD68) wounds. Oral Surgery Oral Medicine Oral
Pathology 27, 309–318.
Sample no. CD31 (no. of cells/mm2) BMP-7 (score) Osteocalcin (score) CD68 (score) Amler, M. H., Johnson, P. L. & Salman, I.
(1960) Histological and histochemical inves-
2–4 weeks
tigation of human alveolar socket healing in
#1 38.6 0.5 1.5 4.0
#2 27.8 1.5 1.0 3.5 undisturbed extraction wounds. Journal of
#3 25.6 2.0 2.0 1.0 the American Dental Association 61, 32–44.
#4 23.5 2.0 2.0 1.5 Amler, M. H., Salman, I. & Bungener, H.
#5 30.4 2.0 2.0 3.0 (1964) Reticular and collagen fiber character-
#6 35.8 0.5 1.0 3.5 istics in human bone healing. Oral Surgery
#7 16.7 1.5 1.0 2.0 Oral Medicine Oral Pathology 17, 785–796.
#8 61.5 1.0 2.0 3.0 Araujo, M. G. & Lindhe, J. (2005) Dimensional
#9 31.3 1.5 0.5 3.5 ridge alterations following tooth extraction.
#10 38.7 2.0 0.5 3.0 An experimental study in the dog. Journal of
Mean 33.0 1.5 1.4 2.8 Clinical Periodontology 32, 212–218.
SD 12.2 0.6 0.6 1.0 Boyne, P. J. (1966) Osseous repair of the
6–8 weeks postextraction alveolus in man. Oral Surgery
#1 36.1 3.0 3.0 1.5 Oral Medicine Oral Pathology 21, 805–813.
#2 24.0 3.0 2.0 2.0 Cardaropoli, G. (2006) Ridge alterations
#3 20.9 3.0 3.0 1.0 following tooth extraction and implant
#4 23.0 4.0 4.0 2.0 placement. Thesis. Institute of Odontology,
#5 29.7 3.0 2.0 Not available The Sahlgrenska Academy at Göteborg
#6 29.4 1.0 1.0 4.0 University.
Mean 27.2 2.8 2.5 2.1 Cardaropoli, G., Araujo, M., Hayacibara, R.,
SD 5.6 1.0 1.0 1.1 Sukekava, F. & Lindhe, J. (2005) Healing of
12–24 weeks extraction sockets and surgically produced –
#1 27.0 4.0 3.0 Not available augmented and non-augmented – defects in
#2 13.7 4.0 3.0 0.5 the alveolar ridge. An experimental study in
#3 38.7 1.0 1.0 0.5 the dog. Journal of Clinical Periodontology
#4 12.7 1.0 1.0 1.0 32, 435–440.
#5 23.3 3.0 2.0 1.5 Cardaropoli, G., Araujo, M. & Lindhe, J. (2003)
#6 10.3 0.5 1.0 Not available Dynamics of bone tissue formation in tooth
#7 14.8 1.5 2.0 Not available extraction sites. Journal of Clinical Perio-
#8 13.7 1.0 2.0 1.5 dontology 30, 809–818.
#9 26.0 2.0 3.0 1.5 Chen, S. T., Wilson, T. G. & Hämmerle, C. F.
#10 17.8 3.0 2.0 Not available (2004) Immediate or early placement of
#11 10.6 2.0 1.0 1.0 implants following tooth extraction: review
Mean 19.0 2.1 1.9 1.1 of biologic basis, clinical procedures, and
SD 8.8 1.2 0.8 0.4 outcomes. The International Journal of Oral
and Maxillofacial Implants 19, 12–25.
Christopher, E. R. (1942) Histological study of
intervals, could have anticipated this forms within the first weeks of healing, bone healing in relation to the extraction of
teeth. Northwestern University Bulletin 45, 5.
remodeling phase. the interval during which mineralized
Claflin, R. S. (1936) Healing of disturbed
The present findings demonstrated bone is laid down is much less predict- and undisturbed extraction wounds. Journal
that great variability exists in man with able. Despite the high heterogeneity that of the American Dental Association 23,
respect to hard tissue formation within was observed in the composition of the 945–959.
extraction sockets. Thus, whereas a pro- specimens within each observation Evian, C. I., Rosenberg, E. S., Coslet, J. G. &
visional connective tissue consistently interval, each tissue component shows Corn, H. (1982) The osteogenic activity of
r 2008 The Authors
Journal compilation r 2008 Blackwell Munksgaard
 Healing of human post-extraction sockets 639

bone removed from healing extraction sockets healing after tooth extraction in the rat. protected defects: a histological study in the
in humans. Journal of Periodontology 53, Anatomical Records 1240, 492–506. canine mandible. The International Journal
81–85. Mangos, J. F. (1941) The healing of extraction of Oral & Maxillofacial Implants 9, 13–29.
Ivanovski, S., Li, H., Daley, T. & Bartold, P. M. wounds. An experimental study based on Schroeder, H. E. (1986) The Periodontium.
(2000) An immunohistochemical study of microscopic and radiographic investigations. Berlin/Heidelberg/New York/Tokyo: Springer-
matrix molecules associated with barrier New Zealand Dental Journal 37, 4–24. Verlag, 12.
membrane-mediated periodontal wound heal- Melcher, A. H. (1976) On the repair potential of Schroeder, H. E. & Münzel-Pedrazzoli, S. D.
ing. Journal of Periodontal Research 35, periodontal tissues. Journal of Perio- (1973) Correlated morphometric and bio-
115–126. dontology 47, 256–260. chemical analysis of gingival tissue. Journal
Johnson, K. (1969) A study of the dimensional Paccione, M. F., Warren, S. M., Spector, J. A., of Microscopy 99, 301–329.
changes occurring in the maxilla following Greenwald, J. A., Bouletreau, P. J. & Longaker, Schropp, L., Wenzel, A., Kostopoulos, L. &
tooth extraction. Australian Dental Journal M. T. (2001) A mouse model of mandibular Karring, T. (2003) Bone healing and soft
14, 241–244. osteotomy healing. Journal of Craniofacial tissue contour changes following single-tooth
Kanayama, M., Kuboki, T., Akiyama, K., Surgery 12, 444–450. extraction: a clinical and radiographic 12-
Nawachi, K., Miyauchi, F. M., Yatani, H., Pietrokovski, J. & Massler, M. (1967) Alveolar month prospective study. International Jour-
Kubota, S., Nakanishi, T. & Takigawa, M. ridge resorption following tooth extraction. nal of Periodontics and Restorative Dentistry
(2003) Connective tissue growth factor Journal of Prosthetic Dentistry 17, 21–27. 23, 313–323.
expressed in rat alveolar bone regeneration Pietrokovski, J., Starinsky, R., Arensburg, B. & Spector, J. A., Luchs, J. S., Mehrara, B. J.,
sites after tooth extraction. Archives of Oral Kaffe, I. (2007) Morphologic characteristics Greenwald, J. A., Smith, L. P. & Longaker,
Biology 48, 723–730. of bony edentulous jaws. Journal of Prostho- M. T. (2001) Expression of bone morphoge-
Kloen, P., Di Paola, M., Borens, O., Richmond, dontics 16, 141–147. netic proteins during membranous bone heal-
J., Perino, G., Helfet, D. L. & Goumans, M. J. Reinhardt, R. A., Lee, H. M., Schmid, M., ing. Plastic Reconstructive Surgery 107,
(2003) BMP signaling components are Payne, J. B. & Golub, L. (2005) Relationship 124–134.
expressed in human fracture callus. Bone between gelatinases and bone turnover in the Whistson, S. W. (1994) Bone. In: Ten Cate, A.
33, 362–371. healing bone defect. Journal of Oral and E. (ed.) Oral histology, Development, Struc-
Kuboki, Y., Hashimoto, F. & Ishibashi, K. Maxillofacial Surgery 63, 1455–1460. ture and Function. St. Louis: Mosby.
(1988) Time-dependent changes of collagen Reinhardt, R. A., Sanderfer, V. J., Meinberg, T.
crosslinks in the socket after tooth extraction A., Nummikoski, P., Lee, H. M. & Marx, D.
in rabbits. Journal of Dental Research 67, B. (2004) Local biochemical markers of bone
944–948. turnover: relationship to subsequent density Address:
Lekic, P., Rojas, J., Birek, C., Tenenbaum, H. & of healing alveolar bone defects. Journal of Dr. Leonardo Trombelli
McCulloch, C. A. (2001) Phenotypic com- Clinical Periodontology 31, 223–228. Research Center for the Study of Periodontal
parison of periodontal ligament cells in vivo Sato, H. & Takeda, Y. (2007) Proliferative Diseases
and in vitro. Journal of Periodontal Research activity, apoptosis, and histogenesis in the University of Ferrara
36, 71–79. early stages of rat tooth extraction wound Corso Giovecca 203
Lin, W. L., McCulloch, C. A. & Cho, M. I. healing. Cells, Tissues, Organs 186, 104–111. 44100 Ferrara
(1994) Differentiation of periodontal ligament Schenk, R. K., Buser, D., Hardwick, W. R. & Italy
fibroblasts into osteoblasts during socket Dahlin, C. (1994) Healing pattern in membrane- E-mail: [email protected]

Clinical Relevance woven bone with lamellar bone and Practical implications: The great
Scientific rationale for the study: marrow, is slow. Limited morpholo- variability observed in the rate of
Studies on post-extraction wound gical alterations seem to occur new bone formation from patient to
repair are affected by significant lim- between 6–8 and 12–24 weeks of patient may be of clinical relevance
itations and do not evaluate all com- healing. High inter-individual het- when to determine the most suitable
partments of the bone defect. erogeneity was noted within each time for implant placement in post-
Principal findings: The process of observation interval. extraction sockets.
remodeling, i.e. the replacement of

r 2008 The Authors

Journal compilation r 2008 Blackwell Munksgaard