Keywords ceramics - composite resins - dental marginal adaptation - mechanical tests - finite
element analysis
Introduction
When restoring endodontically treated teeth, clinicians need to select the restoration
design and material to achieve long-lasting oral rehabilitations. It is known that
one of the main precursors for the longevity of the treatment is the maintenance of
remnant tooth tissue, which many situations today challenge the clinician between
choosing to rehabilitate with a full-coverage crown, usually associated with an intraradicular
post, or an endocrown.[1 ]
[2 ]
[3 ]
Digital dentistry workflow has become a daily routine in many clinical practices worldwide,
employing computer-aided design-computer-aided manufacturing (CAD-CAM) systems to
efficiently deliver predictable long-lasting restorations.[4 ]
[5 ]
[6 ] In this sense, the use of CAD-CAM lithium disilicate-based ceramics (LD) and resin
composite blocks (RC) are widespread options, based on their inherent adequate properties
in mechanical, functional, optical, and aesthetical aspects.[7 ]
[8 ] LD ceramics are characterized by crystal particle-filled glass, whereas the crystalline
content acts as a reinforcing arrangement within the glass' main structure.[9 ] The clinical longevity of CAD-CAM LD crowns is already shown to be adequate, above
80% survival rate, on follow-ups over 15 years.[7 ] Similarly, endocrowns also showed survival rates above 80% on a 10-year follow-up.[3 ] Indirect RC restorations are also already validated as a long-lasting material.[10 ]
[11 ] However, the performance of CAD-CAM resin-based restorations remains more challenging
and is still under discussion, with a higher risk for complications in the short term
(up to 3 years).[12 ] Despite that, their indications are being explored and their use has been intensified
based on their inherent properties of more compatible elastic modulus to the tooth
structure, lesser potential for antagonist wear, and easier fabrication and repair
than dental ceramics.[13 ]
[14 ]
Few studies compare differences in restoration design (e.g. crown or endocrown) and
restoration material (e.g. CAD-CAM LD or RC) in terms of marginal gap, internal gap,
interfacial volume, and fatigue behavior.[15 ]
[16 ]
[17 ] Literature supports that marginal gaps should be less than 120 μm for clinically
acceptable performance.[18 ] Despite that, the American Dental Association recommends that the luting thickness
should not exceed 40 μm.[19 ] It becomes clear the lack of consensus in the literature regarding which gap limit
is acceptable from a clinical standpoint, but it is consensual that increased gaps,
in other words, poor adaptation of restoration, can predispose the patient to poor
periodontal health maintenance,[20 ] facilitate dissolution of the luting agent, and trigger a worse load distribution
of the restorative set.[21 ] Another important aspect directly related to the gap size between restoration and
the tooth substrate is the thickness of the luting agent. It is known that thicker
intaglio surfaces are detrimental to the performance of the luting agent, increasing
the risk of bubble occurrence, which can act as trigger points for stress concentration
and restoration fracture.[21 ]
[22 ]
[23 ]
[24 ] Summed, there is also the fact that CAD-CAM milling has also been known to induce
surface/subsurface damage and residual stresses, which could favor restoration fracture.[25 ]
[26 ]
[27 ]
[28 ]
Based on the aforementioned presupposes, it becomes clear the need for more studies
that compare and characterize the performance of different restoration designs and
materials, on marginal and internal gaps, and interfacial volume of the bonding interface,
and correlate those outcomes to the mechanical fatigue behavior of such restorations.
Thus, this study aims to evaluate the marginal gap, internal gap, interfacial volume,
and fatigue behavior of CAD-CAM restorations with different designs (endocrowns or
crowns) made of different CAD-CAM materials (lithium disilicate-based ceramic [LD],
or resin composite [RC]). Regarding the scarce existence of guiding literature on
the theme this study adopted the null hypothesis that: the marginal gap, internal
gap, and interfacial volume would not be affected by the restoration design (1); and
by the restoration material (2). Additionally, it was pondered that the fatigue behavior
of such restorations also would not be affected by both factors (hypotheses 3 and
4, respectively).
Material and Methods
The study design and materials used are described in [Table 1 ]. An illustration of the study flow, from specimen manufacturing to positioning on
a typodont model (AC 103 model, Pronew Odonto, São Gonçalo, Brazil), scanning, milling,
and obtaining the final restoration, is shown in [Fig. 1 ].
Table 1
Study design and materials
Group code
Restorative design
Material
Analysis
LD-C
Crown
Lithium disilicate (IPS e.max CAD, Ivoclar AG)
- Marginal and internal gaps, interfacial volume using uCT
- Fatigue behavior
- Topography and fractography via SEM
- Finite element analysis
RC-C
Resin composite (Tetric CAD, Ivoclar AG)
LD-E
Endocrown
Lithium disilicate
RC-E
Resin composite
Abbreviations: C, crown; CAD, computer-aided design; E, endocrown; LD, lithium disilicate;
uCT, computed microtomography; RC, resin composite; SEM, scanning electron microscopy.
Fig. 1 Illustration of the flow for specimen manufacturing, from the fiberglass-reinforced
epoxy resin dies, simulating a tooth preparation for crown or endocrown, positioning
on the typodont model, its scanning and milling, to obtaining the final restoration.
Fiberglass-reinforced epoxy resin rods (10 mm diameter, Protec Produtos Técnicos Ltda.,
São Paulo, Brazil) were milled into dies using a lathe (Diplomat 3001; Nardini, Americana,
Brazil) to simulate simplified tooth preparations for a crown or an endocrown.[15 ] For crowns, the tooth preparation presented a conical shape with axial walls at
an inclination of 8 degrees, and a uniform occlusal and axial space of 1.5 mm, the
height of the final tooth preparation was 5.32 mm.[25 ] For endocrowns, the same external dimensions were used, but the differences were
that the height of the final tooth preparation was set as 2 mm, and there was a deepening
in the center of the occlusal surface with 4 mm deep, simulating the intrapulpal preparation.
The axial walls of such entrance presented the same 8-degree inclination of the external
axial wall, and the thickness of the axial wall was set at 2 mm.[15 ]
For restoration digital planning, each fiberglass-reinforced epoxy resin die was settled
into a typodont model ([Fig. 1 ]), and then, using the CEREC Primescan intraoral scanner (Dentsply Sirona, Charlotte,
United States), the preparation was digitalized and the restorations were designed
individually in the design software (CEREC 4.5.2, Dentsply Sirona). Simplified crowns
(n = 10) were planned with a 1.2-mm thickness at the occlusal surface, which was designed
flat, and a cement space of 120 µm, according to the manufacturer's standard instructions.
Meanwhile, simplified endocrowns (n = 10) were planned with the same design, but the occlusal thickness was set at 1.5 mm,
which is the minimal thickness preconized for this design by the manufacturer of the
used restorative materials (Ivoclar AG, Schaan, Liechtenstein). Both crowns and endocrowns
were wet milled in a 4-axis machine with brand new set of 2 burs each (Step bur 12S
and Cylinder pointed bur 12S) at 42,000 revolutions per minute (rpm) (CEREC MC XL,
Dentsply Sirona), considering the two restorative materials (lithium disilicate-based
ceramic, LD, IPS e.max CAD, Ivoclar AG; or resin composite, RC, Tetric CAD, Ivoclar
AG). Milling was done according to the manufacturer's standards. Subsequently, LD
restorations were crystallized in a furnace according to the manufacturer's instructions
(speed crystallization in Programat CS4, Ivoclar AG), and RC ones remained untouched.
Each restoration was tested into its corresponding die before bonding procedures,
to guarantee optimal setting.
After, the restorations and dies were cleaned in an ultrasonic bath with 70% alcohol
for 5 minutes and bonded according to the manufacturer's guidelines and considering
each substrate's intrinsic composition. The epoxy dies were treated with approximately
equal to 5% hydrofluoric acid (IPS Ceramic Etching Gel, Ivoclar AG) for 60 seconds,
followed by air-water rinsing for 30 seconds, air-drying, and an active application
for 20 seconds of an adhesive (Adhese Universal, Ivoclar AG), which was not light-cured.
LD restorations were etched with approximately equal to 5% hydrofluoric acid for 20 seconds,
followed by air-water rinsing for 30 seconds, followed by the active application of
a silane-based primer (Porcelain Silane, B.J.M. Laboratories Ltd, Or Yehuda, Israel)
for 15 seconds, maintained reacting for 45 seconds, and then air-dried. RC restorations
were air-abraded with aluminum oxide powder (50 µm at 10 mm distance, 1 bar pressure;
Ossido di Alluminio, Henry Schein, New York, United States), and then received an
adhesive application (Adhese Universal, Ivoclar AG), as descripted previously. Lastly,
each restoration was adhesively bonded in each respective epoxy die using a dual cure
resin cement (Variolink Esthetic DC, Ivoclar AG) under a standardized load (500 gr)
in a specific device. All resin cement excess was removed, and a light unit (Starlight
Uno, Mectron, Carasco, Italy) with over 1,500 mW/cm2 power, and 440 to 465 nm light wavelength, was used to cure the resin cement for
40 seconds in each direction (0, 90, 180, and 270 degrees, and on the top).
Marginal gap, internal gap, and interfacial volume assessments of each sample were
made using a computed microtomography analysis (SkyScan 1172 Micro-CT, Bruker, Billerica,
United States) with 100 Kv, 100 µA, source–object distance = 89.510 mm, source–detector
distance = 217.578 mm, pixel binning = 9.01 µm, exposure time/projection = 846 ms,
aluminum and copper (Al + Cu) filter, pixel size = 14.83 µm, averaging = 5, and rotation
step = 0.6 degrees.[29 ] The images were obtained using NRecon software (Bruker) with different parameters
for each restorative material, LD/RC, respectively, as follows: smoothening = 0/2,
misalignment compensation = 5/4.5, ring artifacts reduction = 10/2, and beam-hardening
correction = 30/40%. Finally, the images were uploaded onto the Data Viewer software
(Bruker), and three sagittal and coronal slices were randomly selected in each sample
to be analyzed at ImageJ 1.53t (National Institutes of Health, Bethesda, United States)
to obtain the marginal and internal gap values.[30 ] For crowns a single vertical measurement was made to define the marginal gap, while
three different locations were considered for the internal gap, resulting in a total
of 42 measurements per specimen. Meanwhile, for endocrowns, two regions of interest
(ROIs) for the marginal gap were considered, and three ROIs for the internal gap,
resulting in 54 measurements per specimen. Additionally, through a three-dimensional
method, in a specific software (Mimics Medical, v. 23.0; Materialise, Leuven, Belgium),
the volumetric measurement of the resin cement layer of each restoration was evaluated
by checking the dimensions of the bonding interface. Volumetric calculation of the
resultant mask was collected in mm3 and statistically analyzed.[31 ]
For assessing mechanical fatigue behavior using an electric mechanical testing machine
(Instron ElectroPuls E3000, Instron, Norwood, United States), each restorative set
was positioned onto a base, submerged onto distilled water, and cyclic loading was
applied with a stainless steel hemispherical piston (Ø = 40 mm), positioned in the
center of the occlusal surface of the specimen.[32 ] An adhesive tape (110 µm) was interposed between the piston and specimen. Using
a frequency of 20 Hz, an initial load of 100 N for 5,000 cycles and incremental steps
of 50 N for every 10,000 cycles, the test was carried out until failure was detected
or a threshold of 1,500 N was reached; in case of survival up to this step (1,500 N),
the step was increased to 100 N for every 10,000 cycles, until failure or test completion
at 2,800 N.[32 ] After finishing each testing step, the specimens were transilluminated to look for
potential cracks or fractures. Fatigue failure load (FFL) and number of cycles for
failure (CFF) were collected for statistical purposes. Fractography was executed,
first using a stereomicroscope to define the representative failure pattern of each
group, which was sputter-coated with gold, and later further analyzed in a scanning
electron microscopy (VEGA-3G; Tescan, Brno, Czech Republic) at secondary electrons
mode, with 20 kV, under 30× and 200× magnification, for crowns and endocrowns, respectively.
Complementary finite element analysis (FEA) was performed to simulate and map stress
concentration at the mean FFL observed for each restorative setup during the fatigue
test. Three-dimensional models were created, replicating the tested groups while considering
the Young's modulus (E) and Poisson's ratios (v) of each material: LD E = 95 GPa,
v = 0.25; RC E = 11.61 GPa, v = 0.3; tooth preparation in fiberglass-reinforced epoxy
resin E = 18 GPa, v = 0.3; resin cement E = 7.5 GPa, v = 0.3. All solids were assumed
to be isotropic with linear behavior, and all contacts were considered perfectly bonded.
After meshing the models, the setup was fixed at the same points corresponding to
the in vitro tests, and the mean FFL values (in Newton) recorded from the in vitro tests were applied to each model. The maximum principal stresses were measured (in
MPa), and the regions of concentration were mapped in a two-dimensional illustration.
After assuring parametric and homoscedastic distribution, the two-way analysis of
variance (ANOVA) and Tukey's post hoc tests were employed, using Statistix 10 (Analytical
Software, Tallahassee, United States), with α = 0.05, for marginal gap, internal gap, and interfacial volume outcomes. Fatigue
data (FFL and CFF) was submitted to survival analysis by means of Kaplan–Meier with
log-rank (Mantel–Cox) tests, at the IBM SPSS Software v.21 (IBM, New York, United
States), with α = 0.05. FEA data was descriptively analyzed.
Results
Considering the marginal gap, two-way ANOVA indicates that the factor “design” was
not statistically significant, meanwhile the factor “material” and the associated
factors “design*material” were ([Table 2 ]). It can be noted that the lower gap was obtained with the RC crown, while the larger
gap was verified in the LD crown. Endocrowns showed intermediate gap values, with
resin-based restorations presenting lower gaps than lithium disilicate ones ([Table 3 ]).
Table 2
Two-way ANOVA tables for marginal gap, internal gaps (cervical, axio-occlusal, occlusal
regions), interfacial volume, and fatigue data
Marginal gap
Cervical gap
Source
DF
SS
MS
F
p
-Value
Source
DF
SS
MS
F
p
-Value
Design
1
2595
2595
1.05
0.31
Design
1
144075
144075
77.02
0.00
Material
1
184946
184946
74.86
0.00
Material
1
1888
188
1.01
0.32
Design* material
1
55513
55513
22.47
0.00
Design*material
1
1172
1172
0.63
0.43
Error
476
1175925
2470
Error
476
890369
1871
Total
479
1418978
Total
479
1037504
Axio-occlusal gap
Occlusal gap
Source
DF
SS
MS
F
p
-Value
Source
DF
SS
MS
F
p
-Value
Design
1
165680
165680
43.56
0.00
Design
1
166690
166690
15.81
0.00
Material
1
7362
7362
1.94
0.16
Material
1
59630
59630
5.65
0.02
Design* material
1
34301
34301
9.02
0.00
Design*material
1
129875
129875
12.31
0.00
Error
716
2723079
3803
Error
236
10546
10546
Total
719
Total
239
Interfacial volume
FFL/CFF
Source
DF
SS
MS
F
p
-Value
Source
DF
SS
MS
F
p
-Value
Design
1
0.23
0.23
0.009
0.93
Design
1
68062.50
68062.50
1.22
0.28
Material
1
59.48
59.48
2.316
0.14
Material
1
19113062.50
19113062.50
343.65
0.00
Design* material
1
10.48
10.48
0.408
0.53
Design*material
1
14062.50
14062.50
0.25
0.62
Error
36
924.44
25.68
Error
36
2002250
55618.06
Total
40
19488.57
Total
40
1171142500
Abbreviations: ANOVA, analysis of variance; CFF, cycles for failure; DF, Degrees of
Freedom; FFL, fatigue failure load; MS, Mean Square; SS, Sum of Squares.
Table 3
Mean, standard deviation (SD), and 95% confidence interval (95 CI) of marginal gap,
cervical-axial angle, axio-occlusal angle, and occlusal/pulpal space adaptation values
(in µm) and of the volume of the adhesive interface (in mm3 )
Groups
Marginal gap
Cervical-axial angle
Axio-occlusal angle
Occlusal/Pulpal space[a ]
Interfacial volume
Mean (SD)
95% CI
Mean (SD)
95% CI
Mean (SD)
95% CI
Mean (SD)
95% CI
Mean (SD)
95% CI
LD-C
113.6 (78.8)A
99.3–127.9
136.5 (43.7)B
128.5–144.4
155.7 (51.6)B
146.3–165.0
185.0 (76.2)B
165.1–204.9
22.1 (5.4) A
18.2–26.0
RC-C
52.7 (21.2)D
48.8–56.5
136.7 (33.7)B
130.6–142.9
177.0 (51.6)A
167.8–186.3
262.3 (76.2)A
242.9–281.7
20.7 (2.5) A
18.9–22.5
LD-E
95.6 (40.5)B
88.2–102.9
167.7 (42.4)A
160.0–175.4
138.0 (77.2)BC
128.1–147.8
283.9 (174.3)A
238.5–329.3
23.3 (7.5) A
17.9–28.7
RC-E
78.4 (39.3)C
71.3–85.5
175.0 (51.7)A
165.5–184.3
130.1 (53.5)C
123.3–136.9
268.7 (33.9)A
259.8–277.6
19.8 (10.4)A
17.5–22.1
Abbreviations: ANOVA, analysis of variance; C, crown; E, endocrown; LD, lithium disilicate;
RC, resin composite.
Note: Distinct uppercase letters in each column indicate statistical differences according
to two-way ANOVA test with Tukey's post hoc (α = 0.05).
a Occlusal space was considered at crowns, and compared with the pulpal space on endocrowns,
which are the interfacial surfaces parallel to the occlusal surface.
With regards to internal gap outcomes, it was seen that the factor “design” was statistically
significant for all regions (cervical gap, axio-occlusal gap, and occlusal gap), the
factor “material” was statistically significant only for occlusal gap, and the associated
factors “design*material” were statistically significant for axio-occlusal gap and
occlusal gap ([Table 2 ]). Another important aspect that should be noted is that although a space of 120 µm
was standardized during restoration planning, all internal regions exceeded this threshold.
At the cervical-axial angle, the lowest gap was seen at crowns, regardless of material;
even though no difference was found between LD and RC for both crowns and endocrowns.
At the axio-occlusal angle, LD crowns presented a lower gap than RC, but for endocrowns,
there was no difference between LD and RC. When comparing occlusal/pulpal space, LD
crowns showed the lowest values, and RC-C, LD-E, and RC-E were statistically similar
([Table 3 ]). For interfacial volume, there was no statistical influence for any of the factors,
or when they were considered in association ([Tables 2 ] and [3 ]).
For fatigue outcomes (FFL and CFF), only the factor “material” showed statistical
influence whereas RC restorations were superior to LD ones, independently of the restoration
design ([Tables 2 ] and [4 ]). It can be noted that when using LD restorations (both crowns and endocrowns) there
is some risk of failure when the applied load surpasses 700 N or 125,000 cycles ([Table 5 ] and [Fig. 2 ]). Besides, LD restorations presented a 100% failure rate when they reached loads
of 1,000 N or 185,000 cycles. RC restorations required at least 1,800 N or 315,000
cycles to start to show any failure risk (20%), requiring loads above 2,000 N for
at least 335,000 cycles to present a higher than 50% risk of failure ([Fig. 2 ]).
Table 4
Mean, standard deviation (SD), and 95% confidence interval (95 CI) of fatigue failure
load (FFL, in Newton), and number of cycles for failure (CFF)
Groups
FFL[a ]
CFF[a ]
Stress calculated through FEA at the mean FFL
Mean (SD)
95% CI
Mean (SD)
95% CI
Maximum principal stress (MPa)
LD-C
888 (82)B
825–951
162,000 (16,414)B
150,160–175,395
215.4
RC-C
2255 (296)A
2,027–2,483
360,000 (29,627)A
337,781–383,329
547.0
LD-E
850 (103)B
770–929
155,000 (20,615)B
139,153–170,846
59.8
RC-E
2133 (308)A
1,896–2,370
349,000 (30,867)A
325,717–373,171
61.7
Abbreviations: ANOVA, analysis of variance; C, crown; E, endocrown; FEA, finite element
analysis; LD, lithium disilicate; RC, resin composite.
Note: The maximum principal stress (in MPa) calculated through finite element analysis
(FEA) at the mean FFL is presented.
a Distinct uppercase letters in each column indicate statistical differences according
to the Kaplan–Meier log-rank (Mantel–Cox) test (α = 0.05).
Table 5
Survival rates, that is, specimens' probability to exceed the respective fatigue failure
load (FFL, in Newton), and number of cycles for failure (CFF), with their respective
standard error measurements
Groups
FFL (N)/CFF
100/5,000
…
700/125,000
750/135,000
800/145,000
850/155,000
900/165,000
950/175,000
1,000/185,000
1,050/195,000
…
1,800/315,000
1,900/325,000
2,000/335,000
2,100/345,000
2,200/355,000
2,300/365,000
2,400/375,000
2,500/385,000
2,600/395,000
2,700/405,000
LD-C
1
…
1
0.9 (0.1)
0.7 (0.2)
0.6 (0.2)
0.2 (0.1)
…
0.0
–
–
–
–
–
–
–
–
–
–
–
–
RC-C
1
…
1
1
1
1
1
1
1
1
…
0.8 (0.1)
…
…
0.7 (0.2)
…
0.5 (0.2)
0.4 (0.2)
0.2 (0.1)
0.1 (0.1)
0.0
LD-E
1
…
0.9 (0.1)
0.6 (0.2)
0.5 (0.2)
0.3 (0.2)
0.2 (0.1)
0.1 (0.1)
…
0.0
–
–
–
–
–
–
–
–
–
–
–
RC-E
1
…
1
1
1
1
1
1
1
1
…
0.8 (0.1)
0.7 (0.2)
0.6 (0.2)
0.4 (0.2)
…
…
0.2 (0.1)
…
0.1 (0.1)
0.0
Abbreviations: C, crown; E, endocrown; LD, lithium disilicate; RC, resin composite.
Note: The sign “-” indicates absence of specimen tested on the respective step. The
sign “…” indicates absence of failure on the respective step.
Fig. 2 Survival plots of the different tested conditions considering fatigue failure load
(FFL – in Newton) and number of cycles for failure (CFF). It is noticeable that lithium
disilicate restorations presented a higher survival rate compared to resin composite,
regardless of the restorative design.
With regards to the pattern of failure, crowns, regardless of material (LD or RC),
fractured from surface defects located at the restoration/cement intaglio surface,
which cracks then propagated onto the top, occlusal, opposite surface ([Fig. 3 ]). As for endocrowns, the crack initiated at surface defects located at the restoration/cement
intaglio surface juxtaposed to the pulpal axial angle, at the entrance of the pulpal
chamber where the restoration prolonged itself. Such cracks then propagated toward
the occlusal surface.
Fig. 3 Representative fractographic analysis of the restorations, according to their design
and respective materials. Lithium disilicate (LD-C and LD-E) and resin composite (RC-C,
and RC-E). The yellow asterisk indicates the failure origin in the crowns' bonding
surface, and at the endocrowns pulpal angle, while white arrows indicate the direction
of crack propagation.
FEA data ([Fig. 4 ] and [Table 4 ]) showed that the amount of stress concentration required to unleash the restoration
failure varies between restorative materials and designs, whereas crowns required
more stress concentration to unleash their failure than endocrowns. Besides that,
RC crowns required also more stress concentration to unleash their failure than LD
ones. The regions where stress concentrated also differed among restorative designs.
In endocrowns, the maximum stress was observed at the pulpal angle between the occlusal
surface of the tooth preparation and the entrance to the pulpal chamber. For crowns,
the maximum stress was seen on the occlusal surface, a few millimeters lateral to
the center of the crown. The material factor did not influence this aspect.
Fig. 4 Representative images of the finite element analysis (FEA) simulating stress concentration
at the mean fatigue failure load observed for each restorative setup during the fatigue
test. The first image on the left shows an exploded view of each restorative setup,
including tooth preparation, the cement layer, restoration, and load applicator. The
second image from the left displays the mesh for each restorative design. The two
images on the right present the FEA results (stress distribution) for the two restoration
designs (endocrowns at the top and crowns at the bottom) and the two restorative materials
(lithium disilicate – LD, on the left; and resin composite – RC, on the right).
Discussion
The present findings revealed that the design factor significantly influenced all
internal gap regions (cervical, axio-occlusal, and occlusal gaps). The material factor
impacted marginal gap and fatigue outcomes (FFL and CFF), while the interaction between
design and material factors influenced marginal gaps and some internal gap regions
(axio-occlusal and occlusal gaps). Consequently, the study's null hypotheses 1, 2,
and 4—asserting no influence of design and material on these outcomes—were rejected.
Only null hypothesis 3, suggesting that fatigue behavior is not affected by restoration
design, was accepted.
It is established that various factors during restoration manufacturing can affect
its fit to the tooth preparation.[16 ]
[17 ]
[33 ]
[34 ]
[35 ]
[36 ]
[37 ] Distortions can occur during tooth impression; however, literature shows that intraoral
scanners have advanced to a level where they match or surpass traditional impression
techniques.[38 ]
[39 ]
[40 ] In this study, the CEREC Primescan (Dentsply Sirona) was used for direct digitalization
of the preparations. This scanner employs both active triangulation and confocal microscopy,
aligning with existing literature that suggests optimal accuracy is achieved when
multiple imaging principles are used. Active triangulation estimates object position
based on the known positions and angles of two other points, while confocal microscopy
correlates object position and distance with the focal length of the lens.[35 ] Therefore, potential distortions related to the tooth impression and digitalization
process were likely minimal, representing the best possible current performance for
such procedures.
Despite this, it is important to note that the planned cement space during restoration
was set at 120 µm, a threshold traditionally accepted as adequate for maintaining
clinically acceptable biological responses in surrounding tissues.[18 ] May et al[21 ] further demonstrated that the benefits of an adhesive cementation strategy, which
enhances the reinforcement of ceramic restorations, are only realized when the internal
gap remains below 300 μm, with thinner cement intaglio surfaces optimizing performance.
Studies directly correlating the planned space, and the actual gap post-CAD-CAM milling
are scarce, indicating a need for further exploration.[33 ]
[41 ]
[42 ] In this study, RC restorations exhibited smaller marginal gaps compared to LD restorations;
however, no clear trend was observed when comparing the internal gaps between the
two materials, suggesting that there is the influence of other variables evolved.
Nonetheless, it is noteworthy that in this study, internal gaps, although larger than
planned ([Table 3 ]), remained within the 300-μm threshold,[21 ] and marginal gaps were consistently below the 120-μm threshold.[18 ] These findings suggest that the variations observed in marginal and internal gaps
are likely clinically insignificant ([Table 3 ]) and clinical decisions regarding material and design may not be primarily influenced
by these outcomes, though this conclusion should be cautiously considered given the
limitations of this in vitro study.
Another factor potentially influencing marginal and internal gaps is distortion during
the CAD-CAM milling process.[33 ]
[43 ] Milling accuracy is affected by the characteristics of the burs used, the milling
modes (e.g., slow or fast), the number of burs employed, the number of axes in the
milling system, and the material being milled.[33 ]
[43 ]
[44 ]
[45 ]
[46 ] A recent scoping review indicated that accuracy and precision between planned and
final restoration dimensions are optimized when finer burs, longer milling times,
and machines with more axes are used.[33 ] In this study, we utilized the CEREC MC XL (Dentsply Sirona), a 4-axis machine operating
at 42,000 rpm with two burs (bur 12S and Cylinder pointed bur 12S, that reaches 1.00 mm,
with 65 μm grit size). Despite being a 4-axis system, CEREC MC XL is one of the most
used, accurate, and precise systems observed in scientific literature.[33 ]
[45 ] Thus, we believe that any potential distortions related to the milling process were
minimal and consistently distributed across the groups, with variations attributed
solely to the study design and the factors considered within it.
With regard to the different characteristics within the existent material options
for CAD-CAM processing, the LD material is composed of lithium disilicate crystals,
larger than 1 µm and needle-like in shape, randomly distributed within a vitreous
matrix at a 70% volume ratio. In contrast, RC consists of a cross-linked dimethacrylate
matrix filled with barium aluminum silicate glass particles (< 1 µm in size) and silicon
dioxide fillers (< 20 nm in size), also with a filler content of approximately 70%
by volume. Despite the similarity in filler volume ratios, the microstructural differences
between these materials lead to distinct mechanical behaviors under load. RC exhibits
a lower elastic modulus and greater resilience, whereas LD, as a ceramic material,
is more brittle and incapable of sustaining plastic deformation.[47 ]
[48 ]
[49 ]
[50 ] This characteristic leads to RC being more easily milled than LD, which cannot be
directly understood of something beneficial. Furthermore, it is true that LD milling
has being classified as hard milling, which is known to induce a cascade of events
on the ceramic surface and subsurface resulting in radial and lateral cracks, chipping,
damage, and residual stress introduction,[25 ]
[28 ]
[51 ]
[52 ] and that all of these factors constitute potential sites for fracture initiation
and consequent failure of the respective restoration in a clinical environment.[26 ]
[27 ]
[28 ]
[51 ]
[53 ] Despite that, RC are so less resistant than LD that the system can, in some regions,
generate CAD-CAM overmilling, thus removing unplanned material regions and causing
distortions from what was initially planned.[33 ]
[41 ] Hence, we believe that the reasons for the greater performance of RC over LD is
basically microstructural differences, higher resilience of RC, lesser brittleness,
and enhanced compatibility between restorative material and substrate, which present
more similar elastic modulus,[47 ]
[48 ]
[49 ]
[50 ]
[54 ] summed to lesser damage during milling, with lesser residual stresses being incorporated
into the material structure.[15 ]
[33 ] Moreover, RC has an enhanced bonding performance to the resin cement, inducing enhanced
stress distribution through the restorative set.[55 ]
Our findings indicate that RC restorations exhibited lower marginal gap, although
thicker gaps were observed in the occlusal/pulpal space, likely due to overmilling.
Similar overmilling/distortions were observed in endocrowns within the pulpal chamber,
where scanner accuracy decreases and milling becomes more challenging, fact that results
in larger cement intaglio thicknesses.[15 ]
[30 ]
[33 ]
[56 ] Besides misfit, the potential of milling in introducing surface defects and residual
stresses should also be considered, which could facilitate crack propagation when
the restoration is submitted to the oral functional stimuli afterwards.[25 ]
[28 ]
[51 ]
[52 ] Even so, adhesive cementation may heal surface defects induced during milling and
optimize mechanical performance of such restoration.[55 ]
The condition of tooth structure is a major point to consider and to give this substrate
an adequate form is a critical step. Errors in tooth preparation, or insertion axis
during placement,[29 ]
[37 ] or neglecting fundamental principles of tooth preparation (e.g., appropriate wall
convergence and/or parallelism, preparation height, rounded angles, and adequate finish
line),[57 ] can increase cement thickness, leading to potential failures. Larger cement surfaces
increase the likelihood of critical defects, stress concentration, and premature failure
(Griffith law).[58 ] Studies have shown that increased cement thickness also increases air bubbles, reduces
adhesion, increases cement solubility, and compromises long-term performance.[59 ]
[60 ]
Our findings revealed that fatigue behavior was influenced solely by the restorative
material, with RC outperforming LD ([Tables 2 ], [4 ], and [5 ]). RC's lower brittleness allows it to endure higher loads and better distribute
stresses onto the remaining tooth structure.[47 ]
[48 ]
[49 ]
[50 ] According to the manufacturer and corroborated by the literature,[61 ] the flexural strength of the RC used in this study (Tetric CAD) is approximately
272 MPa, while lithium disilicate (IPS e.max CAD) is around 530 MPa.[50 ] The reduction in strength observed between these literature values and the FEA results
supports the slow crack growth mechanisms triggered by fatigue testing, which induce
cumulative damage.[48 ]
[62 ] This mechanism is logical and directly related to the brittleness of ceramic materials.[47 ]
[49 ] It is further supported by fractography analysis, which shows failures initiating
at the cement intaglio surface ([Fig. 3 ]), where the FEA indicates maximum stress concentration.
The performance of RC in endocrowns also aligns with this fatigue mechanism, and FEA
confirms the regions where fractures begin. Furthermore, RC crowns demonstrated optimal
performance, likely due to the resin composite's elastic modulus, which is more compatible
with the tooth structure, allowing greater deformation without causing fractures and
improving stress distribution throughout the restorative assembly.[48 ]
[62 ] Additionally, the enhanced support provided by the tooth preparation for crowns
likely contributes to their superior performance compared to endocrowns in terms of
requiring more stress concentration to unleash fracture as shown by the FEA data.[63 ]
[64 ] An aspect that could have influenced such behavior is the fact that the area of
contact between the crown and the support tooth is external, meanwhile endocrowns
has its prolongation into the pulpal chamber, thus distributing the load that is applied
onto it at the long axis of the whole teeth. Such biomechanics could be responsible
for the differences of stress distribution considering the restorative design factor,
showing optimized performance of endocrowns in distributing the stress than crowns,
which denoted more stress concentration.
With regard to the FFLs endured by the restorative setup and clinical function requirements,
it is important to note that the observed failure loads far exceeded the normal functional
loads typically seen in clinical settings, where the mean chewing load values range
between 285 and 462 N for men and between 254 and 446 N for women.[65 ]
[66 ] They were also higher than parafunctional loads seen in bruxist patients, which
can reach 627 N.[67 ] Following this point of view, it seems a good option for clinicians to opt for RC
restorations, over LD, when rehabilitating bruxist patients.
A previous report assessed stress distribution in full crowns and endocrowns, finding
that endocrowns reduced tensile stress under axial loads, while full crowns performed
better under oblique loads.[68 ] Similarly, the present study applied axial loads, however, the design was not as
significant as the restorative material, probably due to the specimen's shape, which
allowed a less aggressive contact area than when using anatomical crowns as the previous
study. Despite that, both studies corroborate to show less stress concentration in
the restoration when a flexible restorative material is used instead of LD.
Finally, this in vitro study has inherent limitations, such as using only one scanner and a CAD-CAM system,
both of which can impact the final restoration dimensions. Additionally, the anatomy
of the occlusal surface was simplified to induce more stability during the fatigue
testing; on the other hand, a simplified anatomy also induces enhanced support for
the restorations. Thus, the multidirectional incidence of loads resulting from complex
occlusal anatomy may influence the results seen herein. Lastly, the resin cement used
had a similar composition to one of the restorative materials (resin composite), potentially
affecting the accuracy of computed microtomography analysis. Despite these limitations,
this study successfully compared and characterized the performance of different restoration
designs and materials in terms of marginal gaps, internal gaps, interfacial volume,
and their correlation with mechanical fatigue performance, supporting the adequacy
and similar performance of different restorative designs.
Conclusion
Resin composite, crowns, and endocrowns, present smaller gaps at the restoration margins;
differences in other internal gap outcomes exists, but within a potential nonrelevant
threshold.
The fatigue performance was not influenced by the restorative design; namely, crowns
and endocrowns showed a similar FFL, CFF, survival probabilities, and pattern of failure
(fractographical features). Despite that, crowns required more stress concentration
to unleash their failure than endocrowns.
