Evaluation of fiber posts vs metal posts for restoring severely damaged endodontically treated teeth_ a systematic review and meta-analysis

Prévia do material em texto

8 QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Evaluation of fiber posts vs metal posts for 
restoring severely damaged endodontically treated teeth: 
a systematic review and meta-analysis
Xiaodong Wang, MDS*/Xin Shu, MDS*/Yingbin Zhang, MDS/Bin Yang, PhD/Yutao Jian, PhD/Ke Zhao, PhD
Objectives: This review was undertaken to answer a controver-
sial clinical question with high-quality evidence: When severely 
damaged teeth are restored, which type of post (metal or fiber) 
demonstrates superior clinical performance? Data sources: The 
meta-analysis was conducted according to the guidelines in the 
Cochrane handbook. Electronic databases (MEDLINE, EMBASE, 
CENTRAL) and gray literatures were screened up to January 
2018. Only randomized controlled trials (RCTs) with follow-up of 
at least 3 years were included. The quality of included studies 
was assessed by the Cochrane Collaboration’s tool. Meta-analy-
sis compared survival, success, post debonding, and root frac-
ture incidence of teeth restored with fiber and metal posts. The 
GRADE system (Grading of Recommendations, Assessment, 
Development and Evaluations) was used to assess the strength 
of the evidence. Of 1,511 records, 14 full texts were obtained. 
Only four RCTs with follow-up times of 3 to 7 years met the se-
lection criteria. The methodologic quality of included RCTs was 
low risk of bias. Fiber posts presented significantly higher sur-
vival rates than did metal posts (RR 0.57, 95% CI: 0.33 to 0.97, 
P = .04), while no difference was observed in success rates, post 
debonding rates, or root fracture rates. The GRADE assessment 
indicated a high quality of evidence for survival rates and a 
moderate quality for success rates. Conclusion: It was con-
cluded that fiber posts displayed higher medium- term (3 to 
7 years) overall survival rates than did metal posts when used in 
the restoration of endodontically treated teeth with no more 
than two coronal walls remaining. (Quintessence Int 2019;50: 
8–20; doi: 10.3290/j.qi.a41499)
Key words: meta-analysis, post, success rate, survival rate, tooth defect
The retention of severely damaged teeth is sometimes contro-
versial, especially when the value of endodontic treatment is 
limited due to a questionable prognosis. Dental implants have 
been promoted as an acceptable but more non-conservative 
alternative for clinicians.1 However, according to a published 
systematic review, compromised but adequately treated and 
maintained teeth showed long-term (over 15 years) survival 
rates comparable with those of implant-supported single 
crowns.2 Additional advantages of endodontic treatment and 
subsequent restorations are that they are less invasive, more 
psychologically acceptable, and of relatively low cost. These 
may provide an opportunity for the retention of natural teeth, 
even with severe hard-tissue loss, and represent a cost-effective 
alternative to tooth extraction and implant placement.3
Endodontically treated teeth (ETT) are brittle and prone to 
fracture.4 Therefore, post application is essential in the resto-
ration of ETT with substantial tissue loss, increasing the frac-
ture resistance of restored teeth, and providing retention for 
the core material and coronal restoration.5 With proper design, 
noble alloy cast posts and cores are considered the “gold stan-
dard,” due to high clinical success rates,6,7 and demonstrate a 
higher fracture resistance for restored teeth compared with 
fiber posts.5 Conversely, the incidence of root fracture is rela-
tively lower for fiber posts, since their elastic moduli are com-
parable with that of natural dentin.8-10 Compared with cata-
strophic vertical root fractures related to the use of cast metal 
posts for restoring teeth, bevel and horizontal fractures in the 
coronal aspects of the roots are reportedly the most common 
 RESTORATIVE DENTISTRY
 
9
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
failure mode related to the use of fiber posts for restoring 
teeth.11,12
A recent systematic review of randomized controlled trials 
(RCTs) compared survival rates between fiber and metal posts, 
but only descriptive results were reported due to the high het-
erogeneity.13 Other relevant systematic reviews were based on 
observational studies, presenting a high risk of bias and a rela-
tively lower power of evidence.14,15 To the present authors’ 
knowledge, without a definite conclusion, selection of the 
most adequate type of post for the restoration of ETT remains 
controversial and a major concern in dentistry.
The amounts of residual cavity walls should be considered 
as the predominant factor affecting the survival of both ETT 
and the restorations.16 The survival rates increased when more 
residual tooth tissue remained, for both metal- and fiber-post-
supported restorations.13 A prospective clinical study with 
5-year follow-up showed that, despite the types of posts and 
cores used, coronal restorations with “substantial dentin height” 
appeared to have significantly higher survival rates than those 
with “minimal dentin height” (98% vs 93%).17 However, post 
placement did not seem to affect the survival of ETT in cases of 
sufficient coronal tissue remaining.18 Results of a systematic 
review showed that although post placement appeared to 
have a significant influence on reducing the catastrophic failure 
of ETT with severe defects, when three or four coronal walls 
remained, post placement seemed to have no influence.19 
Therefore, in light of the positive effects, deleterious side-ef-
fects, and extra costs of posts, only in cases of severe coronal 
tissue loss should posts be used. Comparing the clinical perfor-
mance of post systems without identifying the remaining cor-
onal walls could result in substantial heterogeneity.
The aim of this systematic review was to answer the follow-
ing question: When severely damaged ETT are restored, which 
type of intraradicular post (metal or fiber) demonstrates supe-
rior clinical outcomes in terms of survival and success rates?
Data sources
The meta-analysis was conducted according to the guidelines of 
the Cochrane Handbook (http://handbook-5-1.cochrane.org/). 
The study protocol was registered after the initial screening stage 
(Prospero CRD42015017644). Slight modification was made to 
the initially registered protocol. For the reasons described previ-
ously, only the clinical performance of ETT with no more than 
two coronal walls was included in this systematic review.
Literature published in English and Chinese was obtained 
through an electronic search of the following databases (up to 
January 2018): MEDLINE via PubMed, EMBASE via embase.com, 
and CENTRAL via the Cochrane Library. The search strategy 
included the following key word combinations (MeSH and free-
text terms in Title/Abstract): “Post and Core Technique [MeSH 
Terms]”, “fiber*”, “fibre*”, “glass-fib*”, “quartz-fib*”, “carbon-fib*”, 
“non-metal*”, “metal*”, “cast*”, “wrought”, “screw”, “titanium”, “Ti”, 
“cobalt-chromium”, “Co?Cr”, “stainless”, “gold”, “alloy”, “prefabri-
cate*”, “custom”, “endodontic*”, “root canal”, “post*”, “dowel*”, 
and “pin*”. Gray literature was searched for potentially suitable 
unpublished clinical trials in ClinicalTrials.gov, Open Grey, and 
Google Scholar. Additional relevant studies were identified by 
hand-searches of the references in the retrieved articles.
Resource selection
Initially, two reviewers independently examined the titles and 
abstracts according to the following inclusion criteria:
 ■ Patients (P): patients with endodontically treated perma-
nent teeth restored with posts and cores. For restored teeth, 
residual coronal walls should ≤ 2. There was no restriction 
by gender or tooth position.
 ■ Intervention (I): endodontically treated teeth restored with 
prefabricated or custom fiber posts, followed by coverage 
with indirect fixed permanent restorations (crowns or fixedpartial dentures [FPDs]).
 ■ Control (C): endodontically treated teeth restored with pre-
fabricated or custom metal posts, followed by coverage 
with indirect fixed permanent restorations (crowns or FPDs).
 ■ Outcomes (O): survival and success rates of restored teeth 
(primary outcomes). The survival criterion was defined as 
an in situ tooth–post complex without extraction. The suc-
cess criterion was defined as both the restored tooth and 
restoration being present and clinically acceptable, without 
intervention or repair needed. Secondary outcomes 
included incidence of complications including root fracture 
and debonding of posts.
 ■ Study design (S): RCTs with follow-up time of at least 3 years. 
The full texts were analyzed when the titles or abstracts indi-
cated that the inclusion criteria were fulfilled. When necessary, 
the original author was contacted for further clarification of 
missing information. Any disagreement between the two 
reviewers was resolved by consensus. If the disagreement 
could not be resolved, a third reviewer was involved. Two 
reviewers independently extracted the data using a data col-
lection form.
10
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Initial electronic search results:
MEDLINE via PubMed: n = 901
EMBASE via embase.com: n = 641
CENTRAL via Cochrane Library: n = 366
Duplicates removed: n = 397
Handsearching and gray literature searching: 
n = 0
Excluded by title and abstract: 
n = 1,497
Articles included in meta-analysis: 
n = 4
Potentially relevant publications: 
n = 1,511
Potentially relevant articles for full text analysis: 
n = 14
Articles excluded with reasons: 
n = 10 (Table 2)
Fig 1 Flowchart of the selection of studies.1
Table 1 Characteristics of included studies
Study 
(year)
Study 
design
Follow- 
up Region Coronal wall Restorations Post cement
Post No.
Cloet 
et al20 
(2017)
RCT 5 y 68 anterior (28 fiber, 
40 metal*); 123 
posterior (63 fiber, 
60 metal*)
Fewer than 2 dentin 
walls (≥ 2 mm) with 
wide pulp chambers
Single crowns Dual-curing 
adhesive cement 
(Panavia F 2.0)
65 prefabricated 
fiber posts (Parapost 
FibreLux); 26 
custom-made glass 
fiber posts 
(EverStick)
91
Sarkis- 
Onofre 
et al21 
(2014)
RCT 3 y 40 anterior (21 fiber, 
19 metal); 32 
posterior (16 fiber, 
16 metal)
No remaining 
coronal wall, or the 
enamel portion of 
one wall with no 
dentinal support 
(ferrule height, 0–0.5 
mm)
Metal-ceramic single 
crowns
Regular resin cement 
(RelyX ARC) or 
self-adhesive resin 
cement (RelyX U100)
Glass fiber post 
(White Post DC)
37
Sterzenbach 
et al22 
(2012)
RCT 7 y 27 incisors (15 fiber, 
12 metal); 17 canines 
(7 fiber, 10 metal); 37 
premolars (18 fiber, 
19 metal); 10 molars 
(5 fiber, 5 metal)
Two or fewer cavity 
walls remaining
52 single crowns (24 
fiber, 28 metal); 23 
FPDs (13 fiber, 10 
metal); 4 single 
crown-RPDs (1 fiber, 
3 metal); 8 FPD-RPDs 
(3 fiber, 5 metal)
Self-adhesive resin 
cement (RelyX 
Unicem)
Glass fiber post 
(Fiberpoints Root 
Pins Glass)
45
Schmitter 
et al23 
(2011)
RCT 5 y Anterior and 
posterior (sample 
size unclear)
Coronal tooth 
destruction ≥ 40% 
Crowns, FPDs, 
crowns integrated 
into RDPs
Metal post: zinc 
phosphate cement 
(Harvard Dental). 
Fiber post: 
composite cement 
(Variolink II)
Fiber-reinforced 
posts (ER dentin 
post)
50
*About 90% (90/101) of the samples in the metal post group had insufficient (≤ 2 coronal walls) tooth tissue remaining, and therefore, the study was included in the analysis. 
†”Absolute failure” refers to catastrophic complications (root fracture, post fracture in root canal, caries, endodontic failure, and periodontal failure) that resulted in tooth extraction. 
“Relative failure” refers to repairable complications (loss of retention of the post, endodontic failure, and post fracture requiring post replacement). 
11
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Review
Assessment of risk of bias and evidence quality
Methodologic qualities of retrieved studies were evaluated inde-
pendently by two reviewers according to the guidance provided 
by the Cochrane Collaboration. The following domains were as-
sessed: sequence generation, allocation concealment, blinding, 
incomplete outcome data, selective outcome reporting, and 
other sources of bias. Additionally, quality of the body of evi-
dence and strength of recommendation were assessed accord-
ing to the GRADE approach by GRADEprofiler 3.6.1 (McMaster 
University and Evidence Prime).
Statistical analysis
Pooled data from all outcomes were subjected to meta-analysis to 
estimate the risk ratio (RR) and 95% confidence intervals (CI) using 
the Review Manager (Ver. 5.3). For each outcome, Cochrane’s Q 
test was applied for analysis of the heterogeneity among 
included studies. To test the reliability of evidence, outcomes of 
fixed-effects and random-effects models were compared, but 
only random-effects estimates were reported to be more con-
servative. If the heterogeneity of outcomes was higher than 
40%, sensitivity analysis was carried out by subgroup analysis, 
testing for excess studies with significant results. If heterogeneity 
could not be eliminated, meta-analysis was abandoned and nar-
rative analysis was applied instead. Additionally, publication 
bias could not be assessed due to the limited number of studies.
Search results
An electronic search identified 1,511 potentially relevant arti-
cles. The evaluation of titles and abstracts led to the selection 
of 14 articles, and finally, only four articles met the inclusion 
criteria. The search strategy is described in Fig 1.
Characteristics of included studies
The characteristics of included studies are detailed in Table 1. 
 Fiber post Metal post
Drop-
out Failure
Survival 
(success) 
rate
Post No. Drop-out Failure
Survival 
(success) 
rate
Unclear Absolute failure†: 6 prefabricated 
fiber posts (1 anterior, 5 
posterior); 2 custom fiber posts 
(2 anterior). Relative failure†: 
7 prefabricated fiber posts 
(3 anterior, 4 posterior); 3 custom 
fiber posts (3 anterior)
91.21% 
(80.22%)
Gold alloy-based 
wrought posts and cast 
cores (Parapost and 
Medior 3 Centres + 
Métaux)
101* 5* Absolute failure†: 14 (6 anterior, 
8 posterior). Relative failure†: 10 
(6 anterior, 4 posterior)
86.14% 
(76.24%)
2 2 debonded (1 premolar, 1 
anterior); 1 debonded with root 
fracture (premolar)
97.30% 
(91.89%)
Co-Cr metal post 41 4 1 root fracture (molar) 97.56% 
(97.56%)
11 1 cervical root fracture 
(premolar); 1 middle root fracture 
(incisor); 1 enhanced tooth mobil-
ity (canine); 1 core fracture 
(premolar)
91.11% 
(91.11%)
Titanium post 
(Fiberpoints Root Pins 
Titanium)
46 11 3 endodontic failures (2 incisors, 
1 molar)
93.48% 
(93.48%)
11 6 extracted (post core-crown 
loosened and severe decay); 2 
post core-crown dislodgement 
and recemented; 2 crowns 
cracked or severely chipped; 1 
apical alteration
88.00% 
(78.00%)
Titanium screw post 
(BKS)
50 8 17 extracted (root fracture or 
perforation); 1 post dislodgement 
and recemented; 1 crown 
dislodgement and recemented; 1 
core-crown complex loosened 
and restored; 1 apical alteration
66.00% 
(58.00%)
12
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
All four included studies were RCTs (parallel design) with fol-
low- up of 3 to 7 years.20-23 Characteristics and reasons for exclu-
sion of 10 studies are listed in Table 2.24-33
For the same clinical trial with multiple publications, only 
the most recent was included in the analysis. Initially, a study 
by Naumann et al24 was included in the survival rate calcula-
tion, and resulted in marked heterogeneity of meta-analysis 
(P = .06, I2 = 59%). Sensitivity analysis indicated the following 
possible reasons: first, the dropout rate of this RCT was as high 
as 49.4%, which is not acceptable in analysis. In addition, itwas 
the only long-term (11-year) RCT, and the author reported that 
the survival rates for severely damaged ETT decreased signifi-
cantly after 8 years, leading to a significantly higher failure rate 
and heterogeneity compared with results of other included 
studies. Therefore, the 11-year long-term results were excluded, 
and data from the 7-year report for the same study were 
included instead.22
Risk of bias
The quality assessments are presented in Fig 2. Generally, most 
studies showed relatively low risk of bias, except for selection 
bias, since half of the included studies did not conceal alloca-
tion. Sarkis-Onofre et al21 and Sterzenbach et al22 mentioned 
the blinding of participants and personnel or outcome assess-
ment. However, the objective outcome was not likely to be 
influenced by lack of blinding; therefore, low risk of both per-
formance bias and detection bias was determined.
Table 2 Characteristics of excluded studies and reasons for exclusion
Study 
(year)
Study 
design
Remaining 
coronal 
wall
Fiber post Metal post
Reasons for 
exclusion
Post information No. Drop-outs (n)
Failures 
(n) Post information No.
Drop-
outs (n)
Failures 
(n)
Naumann 
et al24 
(2017)
RCT (11 y) ≤ 2 cavity 
walls
Glass fiber posts 45 16 12 Titanium posts 46 19 7 Multiple 
reports
Naumann 
et al25 
(2007)
RCT (3 y) ≤ 2 cavity 
walls
Glass fiber posts 45 4 0 Titanium posts 46 0 0 Multiple 
reports
Zicari et 
al26 (2011)
RCT (3 y) < 2 cavity 
walls 
(height ≥ 2 
mm)
65 prefabricated 
fiber posts; 26 
custom-made 
glass fiber posts
91 Unclear 2 Gold alloy-based 
wrought posts 
and cast cores 
88 Unclear 1 Multiple 
reports
Qian et 
al27 (2017)
RCT (2 y) Unclear Custom-made 
glass fiber posts
48 Unclear 0 Co-Cr alloy cast 
posts-cores
49 Unclear 5 No description 
of coronal 
destruction
Gbadebo 
et al28 
(2014)
RCT (0.5 y) ≥ 2 mm 
ferrule
Glass fiber posts 20 2 0 Stainless steel 
posts
20 4 0 Coronal 
destruction is 
inconsistent 
with our PICO
King et 
al29 (2003)
Prospec-
tive (80 to 
100 mo)
Unclear Carbon fiber 
posts
16 2 0 Wrought precious 
alloy 
11 2 0 Non-RCT
Ma et al30 
(2013)
Retro-
spective 
(> 5 y)
Unclear Glass fiber posts 73 Unclear 2 Cast titanium 
posts-cores
40 Unclear 5 Non-RCT
Aquilino 
et al31 
(2002)
Retro-
spective
Unclear Unclear Unclear Unclear Unclear Unclear Unclear Unclear Unclear Non-RCT
Turker et 
al32 (2007)
Prospec-
tive (10 to 
73 mo)
Unclear Polyethylene 
fiber ribbon-rein-
forced posts
42 Unclear 0 NA NA NA NA Non-RCT; 
“No metal 
posts” group
Ellner et 
al33 (2003)
RCT (10 
years)
Unclear NA NA NA NA Cast tapered 
posts; Prefabri-
cated posts; Cast 
posts and cores; 
Threaded 
titanium posts
50 Unclear 3 “No fiber 
posts” group
NA, not applicable.
13
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Data synthesis
Comparison of survival rates All four studies were included 
in the analysis of survival rates with follow-up times from 3 to 
7 years.20-23 As meta-analysis showed (Fig 3a), fiber posts pre-
sented significantly higher overall survival rates than did metal 
posts (RR 0.57, 95% CI: 0.33 to 0.97, P = .04). No heterogeneity 
was detected among the studies (P = .40, I2 = 0%).
Two studies (Fig 3b) were included for the analysis of sur-
vival rates of anterior teeth (5 to 8 years),20,22 and three studies 
(Fig 3c) were included in posterior tooth survival rate analysis 
(3 to 7 years).20-22 The results of meta-analysis indicated that no 
statistical significance was detected in the anterior zone (RR 
0.80, 95% CI: 0.27 to 2.32, P = .68) or the posterior zone (RR 0.76, 
95% CI: 0.31 to 1.89, P = .56), with no heterogeneity in both 
groups (anterior, P = .77, I2 = 0%; posterior, P = .62, I2 = 0%).
Comparison of success, post debonding, and root fracture 
rates The same four studies were also included in the analysis 
of 3- to 7-year success rates.20-23 No statistically significant differ-
ence was observed between fiber posts and metal posts (RR 
0.78, 95% CI: 0.48 to 1.27, P = .31) with low heterogeneity 
(P = .27, I2 = 23%) (Fig 4). 
Studies from Sarkis-Onofre et al21 (3 years) and Schmitter et 
al23 (5 years) were included in comparisons of post debonding 
rates. The results indicated that fiber posts and metal posts 
showed no differences in rates of post dislodgement (RR 1.64, 
95% CI: 0.33 to 8.28, P = .55) (Fig 5). No heterogeneity was 
detected between the two studies (P = .34, I2 = 0%).
Sarkis-Onofre et al21 (3 years) and Sterzenbach et al22 (7 years) 
reported the incidence of root fracture. Meta-analysis pre-
sented no statistical significance between fiber and metal posts 
(RR 2.21, 95% CI: 0.29 to 16.75, P = .44) (Fig 6). No heterogeneity 
was detected (P = .46, I2 = 0%).
GRADE assessment
SoF (summary of findings) tables were generated with GRADE-
profiler. Among the six outcomes analyzed in this review, over-
all survival rates showed high quality of evidence (Fig 7); ante-
rior and posterior tooth survival rates and success rates (Fig 8) 
showed moderate quality of evidence (all downgraded one 
level for imprecision), but incidences of post debonding and 
root fracture showed low quality of evidence (downgraded two 
levels for imprecision).
Discussion
In the present meta-analysis, all the included studies were 
well-designed RCTs with low risk of bias, resulting in satisfac-
tory quality of evidence for the primary outcomes.
A published descriptive systematic review comparing fiber 
and metal posts did not conduct a meta-analysis of RCTs 
because of high heterogeneity.13 In any exploration of the ori-
gin of heterogeneity, residual coronal walls are expected to be 
one of the determinant factors. A previous systematic review 
showed that the high variation in success/survival rates of ETT 
crowns with the absence of a ferrule or only one coronal wall 
ranged between 0% and 97%, while higher success/survival 
rates with lower variation between 66.7% and 100% were pres-
ent for the three- and four-wall groups.34 Based on the guide-
line for the type of reconstruction for ETT according to the 
residual coronal wall, it is recommended that with two to four 
remaining coronal walls, a post would not be necessary. But 
with no or one remaining wall, the core material could provide 
limited retention as well as fracture resistance, and a post is 
required.35 Consequently, studies on ETT with fewer than two 
walls were included in the current meta-analysis, and very low 
Ra
nd
om
 s
eq
ue
nc
e 
ge
ne
ra
tio
n 
(s
el
ec
tio
n 
bi
as
)
A
llo
ca
tio
n 
co
nc
ea
lm
en
t (
se
le
ct
io
n 
bi
as
)
Bl
in
di
ng
 o
f p
ar
tic
ip
an
ts
 a
nd
 p
er
so
nn
el
 (p
er
fo
rm
an
ce
 b
ia
s)
Bl
in
di
ng
 o
f o
ut
om
e 
as
se
ss
m
en
t (
de
te
ct
io
n 
bi
as
)
In
co
m
pl
et
e 
ou
tc
om
e 
da
ta
 (a
tt
rit
io
n 
bi
as
)
Se
le
ct
iv
e 
re
po
rt
in
g 
(r
ep
or
tin
g 
bi
as
)
O
th
er
 b
ia
s
Chloet et al20
Sarkis-Onofre et al21
Schmitter et al23
Sterzenbach et al22
? ? + + ? + ?
+ – + + + + ?
+ – + + + + ?
+ + + + + + ?
Fig 2 Risk of bias summary.
2
14
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Fig 3b Comparison of survival rates of anterior teeth.
Chloet et al20
Sarkis-Onofre et al21
Schmitter et al23
Sterzenbach et al22
heterogeneity was detected. Therefore, it is suggested that in 
comparisons of the application and performance of different 
post and core systems, consideration of remaining coronal 
walls should be regarded as a prerequisite.
Besides remaining coronal walls, a ferrule is another import-
ant factor,36 since ETT restored with ferrules showed survival 
rates superior to those of ETT restored without ferrules during 
5-year observation.16 Unfortunately, most included studies did 
not report detailed information on ferrules and the height of 
remaining coronal walls. Therefore, the relationship between 
ferrules and survival/success ratescould not be clarified by sub-
group analysis. It is suggested that the influence of the ferrule 
Chloet et al20
Sterzenbach et al22
Fig 3a Comparison of survival rates.
Fig 3c Comparison of survival rates of posterior teeth.
Chloet et al20
Sarkis-Onofre et al21
Sterzenbach et al22
P
P
P
P
P
P
3a
3b
3c
15
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
effect should be based on relatively sufficient support from 
coronal walls. The results of a 6-year RCT demonstrated that, for 
restored teeth without coronal walls, failure risks were similar, 
regardless of the presence or absence of ferrules.37
Differences in remaining dental tissue, post depth and 
preparation, materials, and processing techniques for posts/
cores, as well as cement selection, etc, resulted in survival rates 
with a large variation among included studies (88%23 to 97.3%21 
for fiber posts, 66%23 to 97.6%21 for metal posts). Therefore, for 
comparisons of clinical performance of different post systems, 
meta-analysis with only RCTs is the best choice to minimize pos-
sible influencing factors. The results showed that fiber posts pres-
Chloet et al20
Sarkis-Onofre et al21
Schmitter et al23
Sterzenbach et al22
Fig 4 Comparison of success rates.
Fig 5 Comparison of post debonding rates.
Sarkis-Onofre et al21
Schmitter et al23
Fig 6 Comparison of root fracture rates.
Sarkis-Onofre et al21
Sterzenbach et al22
5
6
4
P
P
P
P
P
P
16
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Fiber posts compared to metal posts for severely damaged ETT
Patient or population: patients with severely damaged ETT
Settings: University
Intervention: Fiber posts
Comparison: Metal posts
Outcomes Illustrative comparative risks* (95% CI) Relative 
effect
No. of participants 
(studies)
Quality of the 
evidence 
(GRADE)
Comments
Assumed risk Corresponding risk
Metal posts Fiber posts
Survival Study population RR 0.57 (0.33 
to 0.97)
461 (4 studies) ++++ high
147 per 1,000 84 per 1,000 (49 to 143)
Moderate
103 per 1,000 59 per 1,000 (34 to 100)
*The basis for the assumed risk (eg, the median control group risk across studies) is provided in footnotes. 
The corresponding risk (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 
95% CI). 
(CI, confidence interval; RR, relative risk).
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the 
estimate.
Very low quality: We are very uncertain about the estimate.
Fig 7 Summary of findings for survival rates.
Fiber posts compared to metal posts for severely damaged ETT
Patient or population: patients with severely damaged ETT
Settings: University
Intervention: Fiber posts
Comparison: Metal posts
Outcomes Illustrative comparative risks* (95% CI) Relative 
effect
No. of participants 
(studies)
Quality of the 
evidence 
(GRADE)
Comments
Assumed risk Corresponding risk
Metal posts Fiber posts
Survival Study population RR 0.78 (0.48 
to 1.27)
461 (4 studies) +++− moderate†
206 per 1,000 161 per 1,000 
(99 to 261)
Moderate
153 per 1,000 119 per 1,000 
(73 to 194)
*The basis for the assumed risk (eg, the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% CI) is 
based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). (CI, confidence interval; RR, relative 
risk).
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the 
estimate.
Very low quality: We are very uncertain about the estimate.
†Downgraded one level for imprecision (total sample size [461] is relatively small).
Fig 8 Summary of findings for success rates.
7
8
17
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
ent higher 3- to 7-year overall survival rates compared with metal 
posts for ETT with severe coronal damage. Among the included 
studies, four different metal post systems were involved (gold-
alloy-based wrought posts,20 Co-Cr casting posts,21 titanium pre-
fabricated pin posts,22 and titanium screw posts23), three of which 
demonstrated comparable survival rates without statistically 
significant differences compared with fiber post groups (86.14% 
[metal] vs 91.21% [fiber] for 5 years;20 97.56% [metal] vs 97.30% 
[fiber] for 3 years;21 93.48% [metal] vs 91.11% [fiber] for 7 years22). 
In contrast, in the study by Schmitter et al,23 it was found that the 
failure rate of prefabricated screw titanium posts was signifi-
cantly higher than that of fiber posts (88.00% vs 66.00% for 
5 years). Interestingly, all six failures in the fiber post group were 
due to post core-crown loosening and severe decay, while the 14 
teeth in the metal post group were extracted because of root 
fracture or perforations. Root canals were enlarged and tapped 
by means of a tapper (Brasseler), followed by insertion of paral-
lel-sided screw posts, and root fractures or perforation were 
observed subsequently.23 It has been proved that threaded metal 
posts represent a high risk of root fracture,38 since tapping inside 
the root exerts more stress on the tooth.39 Moreover, clinicians 
should pay more attention to parallel threaded posts, which dis-
played two times greater root strain and stress concentration 
compared with tapered threaded posts.40 In summary, the use of 
parallel-sided screw metal posts should be avoided if possible. 
According to methodology, although titanium parallel-sided 
screw posts demonstrated much lower survival rates, there is 
insufficient evidence to exclude them from the meta-analysis, 
especially when heterogeneity is negligible. Additionally, accord-
ing to a published systematic review, no statistically significant 
difference was detected between the 5-year success rates of pre-
fabricated titanium and those of cast posts.13
Further, it was found that ETT restored with metal posts 
were extracted more often due to endodontic failure. The elas-
tic modulus mismatch between roots and metal posts led to a 
stress concentration at the bonding interface between the post 
and the luting material,41 which therefore resulted in microgaps 
and could increase bacterial colonization as well as risk for api-
cal periodontitis.22 An in-vitro study reported that glass fiber 
posts and resin-supported polyethylene fiber posts exhibited 
less microleakage compared with stainless steel posts and zir-
conia posts.42 With no access to root canal retreatment, teeth 
restored with a cast post-and-core restoration but with a peri-
apical lesion present are very likely to be extracted.43
Three studies reported survival rates of anterior and poste-
rior teeth, but only two studies were included in anterior tooth 
analysis, because, in the study by Sarkis-Onofre et al,21 no failure 
was found in the anterior zone for both groups (both-arm-0-
event, BA0E). Also, the study by Schmitter et al23 was excluded, 
since detailed information for anterior and posterior teeth was 
unclear. With the limited number of included studies and mod-
est sample size, no statistical significance was detected between 
fiber and metal posts for both anterior and posterior teeth. 
Reported reasons for failurewere similar for anterior and poste-
rior teeth, including root fracture, core fracture in fiber posts, and 
endodontic failure, as well as root fracture for metal posts.20-22
Debonding was considered the most common type of fail-
ure with post-retained restorations.44,45 Only two studies were 
included in the meta-analysis,21,23 since the other two studies 
did not report the incidence of debonding.20,22 No statistically 
significant difference was found when fiber and metal posts 
were compared, but in the study by Sarkis-Onofre et al,21 dis-
lodgement was observed in two of a group of 37 prefabricated 
fiber posts, compared with none in a group of 43 cast Co-Cr 
posts. Post retention was determined by micromechanical 
interlocking, chemical bonding, and sliding friction. Therefore, 
perhaps the types of fabrication rather than the materials 
should count for the retention of posts. This was supported by 
an in-vitro study showing that cast metal posts had greater 
retention than fiber and zirconia posts because of superior fit-
ting.46 Similarly, customized fiber posts also showed superior 
bond strength compared with prefabricated metal or fiber 
posts.47,48 The above results could be explained by cement thick-
ness, since thicker cement might lead to increased polymeriza-
tion stress and decreased bending strength.48 Simultaneously, 
with a lack of direct visualization and control of adhesive proce-
dures, the predictability of cement could be further jeopar-
dized.48 For some prefabricated posts, since their adaptation to 
root canal anatomy was not always good (eg, oval canal), lateral 
compaction of smaller accessory posts was required after place-
ment of a master post. A well-prepared root canal for post space 
and the selection of post taper/diameter are essential to acquire 
initial friction retention during a prefabricated post try-in, which 
will contribute to the prevention of post debonding.
In the studies by Cloet et al20 and Schmitter et al,23 the exact 
number of root fractures is unclear. No statistically significant 
differences were found in the incidence of fracture between 
the two post systems in meta-analysis of the remaining two 
studies.21,22 In accordance with these results, a previous review 
indicated that root fracture incidence for metal posts was 
5.13% per year, which is comparable with that reported for 
fiber posts (4.78% per year).14 Actually, root fractures are usually 
more unfavorable and catastrophic with metal posts,14,23 since 
potential fracture locations are expected to be apical with 
18
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
metal posts, but coronal (cervical) with fiber posts.49 In addition 
to similar elastic moduli and better stress dissipation in ETT 
restored with fiber posts, a thick cement layer can also act as 
stress absorption under functional loading, reducing the possi-
bility of root facture.5 Finite Element Analysis indicated that the 
use of a post-and-core material with a lower elastic modulus 
and a cement with a higher elastic modulus could result in high 
stress concentrations in the cement layer, leading to a reduc-
tion in stress and deformation in the residual root.50 However, it 
must be kept in mind that the placement of a post, regardless 
of its composition, produces a potential risk to the integrity of 
the tooth. Filling the root canal space with a material with stiff-
ness that varies from that of pulp will make it impossible to 
replicate the stress distribution of the original tooth, thus 
increasing the incidence of root fracture.50
In the inclusion criteria, the follow-up time was stipulated to 
be at least 3 years, since observation of less than 3 years was 
generally regarded as a short-term investigation.51 Although the 
quality of evidence was relatively high and credible, it should be 
noted that all the results of the present review were based on 
medium-term studies (3 to 7 years), and the conclusions should 
not be extrapolated to long-term situations. Long-term prospec-
tive controlled clinical trials are very rare because they are diffi-
cult to conduct and have high dropout rates. Up to now, only 
one long-term RCT could be found on this topic, in which cumu-
lative survival probability for glass-fiber posts (58.7%) was sig-
nificantly lower than that for metal posts (74.2%) after 11 years’ 
observation.24 Failure reasons for fiber posts were root fracture 
(4/29), endodontic failure (2/29), core fracture (1/29), tooth frac-
ture (1/29), secondary caries (1/29), and tooth mobility (1/29) 
after 11 years’ follow-up, while failures for metal posts were 
attributed to endodontic failure (5/27) and root fracture (1/27). 
Apart from the high dropout rate (approximately 50%), an inter-
esting trend found in this study was that, for severely damaged 
teeth (defects in more than two walls), the Kaplan-Meier curve 
dropped rapidly after 8 years for both groups.24 Long-term sur-
vival of ETT with extensive coronal tissue loss is highly uncertain 
and could be quite different from medium-term results. 
Conclusions
Based on the high level of evidence from RCTs, fiber posts display 
higher medium-term (3 to 7 years) survival rates than do metal 
posts in the restoration of ETT with fewer than two coronal walls 
remaining. With the limited included studies and sample sizes, no 
differences are evident between fiber and metal posts in anterior 
teeth, posterior teeth, post debonding, and root fracture.
Acknowledgment
This work was supported by the National Natural Science Foun-
dation of China under grant No. 81470767 and No. 81600907.
References
1. Chércoles-Ruiz A, Sánchez-Torres A, 
Gay-Escoda C. Endodontics, endodontic re-
treatment, and apical surgery versus tooth 
extraction and implant placement: a system-
atic review. J Endod 2017;43:679–686.
2. Levin L, Halperin-Sternfeld M. Tooth 
preservation or implant placement: 
a systematic review of long-term tooth 
and implant survival rates. J Am Dent Assoc 
2013;144:1119–1133.
3. Pennington MW, Vernazza CR, Shackley 
P, Armstrong NT, Whitworth JM, Steele JG. 
Evaluation of the cost-effectiveness of root 
canal treatment using conventional 
approaches versus replacement with an 
implant. Int Endod J 2009;42:874–883.
4. Sokol DJ. Effective use of current core 
and post concepts. J Prosthet Dent 1984;52: 
231–234.
5. Sreedevi S, Sanjeev R, Raghavan R, 
Abraham A, Rajamani T, Govind GK. An in 
vitro study on the effects of post-core design 
and ferrule on the fracture resistance of end-
odontically treated maxillary central incisors. 
J Int Oral Health 2015;7:37–41.
6. Shillingburg H. Fundamentals of fixed 
prosthodontics (3rd ed). Chicago: 
Quintessence, 1997:73–84.
7. Creugers NH, Mentink AG, Käyser AF. An 
analysis of durability data on post and core 
restorations. J Dent 1993;21:281–284.
8. Malferrari S, Monaco C, Scotti R. Clinical 
evaluation of teeth restored with quartz 
fiber-reinforced epoxy resin posts. 
Int J Prosthodont 2003;16:39–44.
9. Glazer B. Restoration of endodontically 
treated teeth with carbon fibre posts: 
a prospective study. J Can Dent Assoc 
2000;66:613–618.
10. Perel ML, Muroff FI. Clinical criteria for 
posts and cores. J Prosthet Dent 1972;28: 
405–411.
11. Nie EM, Chen XY, Zhang CY, Qi LL, 
Huang YH. Influence of masticatory fatigue 
on the fracture resistance of the pulpless 
teeth restored with quartz-fiber post-core 
and crown. Int J Oral Sci 2012;4:218–220.
12. Soares CJ, Soares PV, de Freitas Santos- 
Filho PC, Castro CG, Magalhaes D, Versluis A. 
The influence of cavity design and glass fiber 
posts on biomechanical behavior of 
endodontically treated premolars. J Endod 
2008;34:1015–1019.
13. Marchionatti AME, Wandscher VF, 
Rippe MP, Kaizer OB, Valandro LF. Clinical 
performance and failure modes of pulpless 
teeth restored with posts: a systematic 
review. Braz Oral Res 2017;31:e64.
14. Figueiredo FE, Martins-Filho PR, 
Faria-E-Silva AL. Do metal post-retainedrestorations result in more root fractures 
than fiber post-retained restorations? A 
systematic review and meta-analysis. 
J Endod 2015;41:309–316.
19
Wang et al
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
15. Theodosopoulou JN, Chochlidakis 
KMA. Systematic review of dowel (post) and 
core materials and systems. J Prosthodont 
2009;18:464–472.
16. Naumann M, Schmitter M, 
Frankenberger R, Krastl G. “Ferrule comes 
first. Post is second!” Fake news and 
alternative facts? A systematic review. 
J Endod 2018;44:212–219.
17. Creugers NH, Mentink AG, Fokkinga 
WA, Kreulen CM. 5-year follow-up of a 
prospective clinical study on various types 
of core restorations. Int J Prosthodont 
2005;18:34–39.
18. Aurélio IL, Fraga S, Rippe MP, Valandro 
LF. Are posts necessary for the restoration 
of root filled teeth with limited tissue loss? 
A structured review of laboratory and clinical 
studies. Int Endod J 2015;49:827–835.
19. Zhu Z, Dong XY, He S, Pan X, Tang L. 
Effect of post placement on the restoration 
of endodontically treated teeth: a systematic 
review. Int J Prosthodont 2015;28:475–483.
20. Cloet E, Debels E, Naert I. Controlled 
clinical trial on the outcome of glass fiber 
composite cores versus wrought posts and 
cast cores for the restoration of endodonti-
cally treated teeth: a 5-year follow-up study. 
Int J Prosthodont 2017;30:71–79.
21. Sarkis-Onofre R, Jacinto RC, Boscato N, 
Cenci MS, Pereira-Cenci T. Cast metal vs. glass 
fibre posts: a randomized controlled trial 
with up to 3 years of follow up. J Dent 
2014;42:582–587.
22. Sterzenbach G, Franke A, Naumann M. 
Rigid versus flexible dentine-like endodontic 
posts: clinical testing of a biomechanical 
concept: seven-year results of a randomized 
controlled clinical pilot trial on endodonti-
cally treated abutment teeth with severe 
hard tissue loss. J Endod 2012;38:1557–1563.
23. Schmitter M, Hamadi K, Rammelsberg 
P. Survival of two post systems: five-year 
results of a randomized clinical trial. 
Quintessence Int 2011;42:843–850.
24. Naumann M, Sterzenbach G, Dietrich T, 
Bitter K, Frankenberger R, von Stein-Lausnitz 
M. Dentin-like versus rigid endodontic post: 
11-year randomized controlled pilot trial on 
no-wall to 2-wall defects. J Endod 
2017;43:1770–1775.
25. Naumann M, Sterzenbach G, Alexandra 
F, Dietrich T. Randomized controlled clinical 
pilot trial of titanium vs. glass fiber prefabri-
cated posts: preliminary results after up to 
3 years. Int J Prosthodont 2007;20:499–503.
26. Zicari F, Van Meerbeek B, Debels E, 
Lesaffre E, Naert I. An up to 3-year controlled 
clinical trial comparing the outcome of glass 
fiber posts and composite cores with gold 
alloy-based posts and cores for the resto-
ration of endodontically treated teeth. 
Int J Prosthodont 2011;24:363–372.
27. Qian YM, Zhong Q, Chen S. Comparison 
of clinical effects of Co-Cr alloy cast post-
core and everStick fiber post in restoration of 
labially or lingually inclined maxillary central 
incisor. Shanghai Kou Qiang Yi Xue 
2017;26:89–93.
28. Gbadebo OS, Ajayi DM, Oyekunle OO, 
Shaba PO. Randomized clinical study com-
paring metallic and glass fiber post in resto-
ration of endodontically treated teeth. 
Indian J Dent Res 2014;25:58–63.
29. King PA, Setchell DJ, Rees JS. Clinical 
evaluation of a carbon fibre reinforced 
carbon endodontic post. J Oral Rehabil 
2003;30:785–789.
30. Ma H, Shen L, Liu K, Su L. Clinical evalu-
ation of residual crowns and roots restored 
by glass fiber post and core, cast metal post 
and core or directly. Hua Xi Kou Qiang Yi Xue 
Za Zhi 2013;31:45–48.
31. Aquilino SA, Caplan DJ. Relationship 
between crown placement and the survival 
of endodontically treated teeth. J Prosthet 
Dent 2002;87:256–263.
32. Turker SB, Alkumru HN, Evren B. 
Prospective clinical trial of polyethylene fiber 
ribbon-reinforced, resin composite post-core 
buildup restorations. Int J Prosthodont 
2007;20:55–56.
33. Ellner S, Bergendal T, Bergman B. Four 
post-and-core combinations as abutments 
for fixed single crowns: a prospective up to 
10-year study. Int J Prosthodont 
2003;16:249–254.
34. Sarkis-Onofre R, Fergusson D, Cenci 
MS, Moher D, Pereira-Cenci T. Performance 
of post-retained single crowns: a systematic 
review of related risk factors. J Endod 
2017;43:175–183.
35. Peroz I, Blankenstein F, Lange KP, 
Naumann M. Restoring endodontically 
treated teeth with posts and cores: a review. 
Quintessence Int 2005;36:737–746.
36. Stankiewicz NR, Wilson PR. The ferrule 
effect: a literature review. Int Endod J 
2002;35:575–581.
37. Ferrari M, Vichi A, Fadda GM, et al. 
A randomized controlled trial of endodonti-
cally treated and restored premolars. J Dent 
Res 2012;91:72S–78S.
38. Fox K, Wood DJ, Youngson CC. A clinical 
report of 85 fractured metallic post-retained 
crowns. Int Endod J 2004;37:561–573.
39. Bergenholtz G. Textbook of Endodont-
ology (2nd ed). Oxford: Wiley-Blackwell, 
2010:323–324.
40. Santos Filho PC, Soares PV, Reis BR, 
Veríssimo C, Soares CJ. Effects of threaded 
post placement on strain and stress distribu-
tion of endodontically treated teeth. Braz 
Oral Res 2013;27:305–310.
41. Garbin CA, Spazzin AO, Meira-Júnior 
AD, Loretto SC, Lyra AM, Braz R. Biomechani-
cal behaviour of a fractured maxillary incisor 
restored with direct composite resin only or 
with different post systems. Int Endod J 
2010;43:1098–1107.
42. Usumez A, Cobankara FK, Ozturk N, 
Eskitascioglu G, Belli S. Microleakage of end-
odontically treated teeth with different 
dowel systems. J Prosthet Dent 2004;92: 
163–169.
43. Wenteler GL, Sathorn C, Parashos P. 
Factors influencing root canal retreatment 
strategies by general practitioners and spe-
cialists in Australia. Int Endod J 2015;48: 
417–427.
44. Cagidiaco MC, Goracci C, Garcia-Godoy 
F, Ferrari M. Clinical studies of fiber posts: 
a literature review. Int J Prosthodont 
2008;21:328–336.
45. Bitter K, Noetzel J, Stamm O, et al. 
Randomized clinical trial comparing the 
effects of post placement on failure rate of 
postendodontic restorations: preliminary 
results of a mean period of 32 months. 
J Endod 2009;35:1477–1482.
46. Türker SA, Özçelik B, Yilmaz Z. Evalua-
tion of the bond strength and fracture resis-
tance of different post systems. J Contemp 
Dent Pract 2015;16:788–793.
47. Singh A, Logani A, Shah N. An ex vivo 
comparative study on the retention of cus-
tom and prefabricated posts. J Conserv Dent 
2012;15:183–186.
48. Marcos RM, Kinder GR, Alfredo E, et al. 
Influence of the resin cement thickness on 
the push-out bond strength of glass fiber 
posts. Braz Dent J 2016;27:592–598.
49. Trushkowsky RD. Restoration of end-
odontically treated teeth: criteria and tech-
nique considerations. Quintessence Int 
2014;45:557–567.
50. Oyar P. The effects of post-core and 
crown material and luting agents on stress 
distribution in tooth restorations. J Prosthet 
Dent 2014;112:211–219.
51. Wang XD, Fan DN, Swain MV, Zhao K. 
A systematic review of all-ceramic crowns: 
clinical fracture rates in relation to restored 
tooth type. Int J Prosthodont 2012;25: 
441–450.
20
 RESTORATIVE DENTISTRY
QUINTESSENCE INTERNATIONAL | volume 50 • number 1 • January 2019
Xiaodong Wang Xin Shu
Xiaodong Wang* Clinical Assistant Professor, Department of 
Prosthodontics, Guanghua School of Stomatology, Sun Yat-sen 
University, Guangdong Provincial Key Laboratory of Stomatology, 
Guangzhou, China.
Xin Shu* Staff Dentist, Department of Stomatology, Baoan Ma-
ternal and Child Health Hospital, Jinan University, Baoan District, 
Shenzhen, China.
Yingbin Zhang Staff Dentist, The 1st Clinical Division, Hospital 
of Stomatology, Peking University, Xicheng District, Beijing, China.
Bin Yang Clinical Assistant Professor, Department of Restorative 
Dentistry, University of Illinois at Chicago, Chicago, IL, USA.
Yutao Jian Associate Professor, Institute of Stomatological Re-
search, Guangdong Provincial Key Laboratory of Stomatology, Sun 
Yat-sen University, Guangzhou, China.Ke Zhao Professor, Department of Prosthodontics, Guanghua 
School of Stomatology, Sun Yat-sen University, Guangdong Provin-
cial Key Laboratory of Stomatology, Guangzhou, China.
*These authors contributed equally to the study.
Correspondence: Dr Ke Zhao, Department of Prosthodontics, Guanghua School of Stomatology, Sun Yat-sen University, Guangdong 
Provincial Key Laboratory of Stomatology, 54 Ling-yuan West Street, Guangzhou, China 510055. Email: zhaoke@mail.sysu.edu.cn; 
dr.zhaoke@aliyun.com; 
Dr Bin Yang, Department of Restorative Dentistry M/C 555, 801 South Paulina Street, Room 204j, Chicago, IL 60612, USA. 
Email: yangbin@uic.edu; 
Dr Yutao Jian, Institute of Stomatological Research, Guangdong Provincial Key Laboratory of Stomatology, Sun Yat-sen University, Zhong-
shan Er Road 74, Guangzhou, China 510080. Email: jianyt@mail.sysu.edu.cn.