d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620
avai lab le at www.sc iencedi rec t .com
journa l homepage: www. int l .e lsev ierhea l th .com/ journa ls /dema
arginal adaptation in vitro and clinical outcome of Class Vestorations
.D. Heintzea,∗, U. Blunckb, T.N. Göhringc, V. Roussond
R&D, Ivoclar Vivadent AG, Bendererstrasse 2, FL-9494 Schaan, LiechtensteinCharité Berlin, Department of Operative Dentistry, Dental School, Berlin, GermanyUniversity of Zurich, Center for Dental and Oral Medicine, Clinic of Preventive Dentistry, Periodontology and Cariology, Zurich,witzerlandBiostatistics Unit, Institute for Social and Preventive Medicine, University of Lausanne, Switzerland
r t i c l e i n f o
rticle history:
eceived 21 July 2008
ccepted 10 November 2008
eywords:
dhesive system
arginal adaptation
n vitro
n vivo
lass V
a b s t r a c t
Objectives. We examined the correlation between the quantitative margin analysis of two
laboratory test methods (Berlin, Zurich) and the clinical outcome in Class V restorations.
Methods. Prospective clinical studies with an observation period of at least 18 months were
searched in the literature, for which laboratory data were also available. The clinical outcome
variables were retention loss, marginal discoloration, detectable margins and secondary
caries. Forty-four clinical studies matched the inclusion criteria, including 34 adhesive sys-
tems for which laboratory data were also present. For both laboratory test methods and the
clinical studies, an index was formulated to better compare the in vitro and in vivo results.
Linear mixed models which included a random study effect were calculated. As most clini-
cal data were available for 12 and 24 months, the main analysis was restricted to these recall
intervals.
Results. The comparative analysis revealed a weak correlation between the clinical index
and both in vitro indices. The correlation was statistically significant for the Berlin method
but not for the Zurich method and only present if studies were compared which used the
same composite in the in vitro and in vivo study. When defining specific cut-off values, the
prognosis for the good clinical performance of an adhesive system based on in vitro results
was 78% (Berlin) or 100% (Zurich). For poor performance it was 67% and 60%, respectively.
No correlation was found between both in vitro methods.
Significance. The surrogate parameter “marginal adaptation” of restorations placed in
extracted teeth has a mediocre value to predict the clinical performance of an adhesive sys-
tem in cervical cavities. The composite is an important factor for a successful prediction.
The comparison between in vitro/in vivo is sometimes hampered by the great variability of
e sam
emy
tracer penetration and the marginal analysis of restorations
of Dental Materials. Published by Elsevier Ltd. All rights reserved.
e).
placed in extracted teeth (human or bovine). These methodsare also part of the ISO Technical Specification on the adhesionto tooth structure [1]. The concept of bond strength testing
is based on the assumption that low values correspond withearly retention loss of cervical restorations and poor marginaladaptation. Likewise, poor marginal adaptation or the deeppenetration of tracers in restorations placed in extracted teethis indicative of the poor clinical performance of an adhe-sive system. There is a vast body of published studies thathave applied one of these methods. According to a reviewpublished in 2007, almost 1500 publications are cited in MED-LINE which apply some sort of bond strength test, while in877 publications the evaluation is based on tracer penetration(microleakage), and about 90 publications use the microscopicevaluation of marginal gaps [2]. A systematic review of thecorrelation between bond strength tests and either tracer pen-etration or the microscopic evaluation of the marginal sealrevealed that about 80% of the selected studies showed no cor-relation [2]. Furthermore, a poor or no correlation was foundbetween tracer penetration and SEM analysis of restorativemargins [3–5].
Laboratory testing should not become an end in itself butallow a prognosis of the clinical performance of restorativematerials and/or operative techniques. Therefore, a significantcorrelation between laboratory test methods and the clinicaloutcome is indispensable and should be part of each valida-tion process. In a review which involved a limited number ofcases, the researchers tried to correlate the in vitro data of ClassV fillings published by one dental institute with the outcome ofclinical trials with the same adhesive system [2]. In the study,only clinical trials were taken into consideration which evalu-ated at least two adhesive systems (in the same study), mostlyusing a split-mouth design. It was assumed that the variabil-ity of placement and evaluation of the restorations was lowerwithin the same study than in studies of different researchinstitutes. In 9 of the selected 11 studies, the laboratory datadid not match the clinical data and could therefore not pre-dict the clinical outcome [2]. Contrary to Class II restorations,Class V restorations are especially useful for in vitro/in vivocomparison due to the easy access of the restoration mar-gins for clinical inspection and evaluation. Furthermore, theoperational technique for the placement of Class V restora-tions presents less variability compared to that of Class IIrestorations, taking into account the cavity size, applicationof the adhesive, layering technique, curing protocol, matrixtechnique, removal of excess, etc.
In the same study, the researchers also tried to corre-late tracer penetration with the clinical outcome, but it wasnot possible to select an adequate number of adhesive sys-tems tested at the same research institute with the same testmethod [2]. The broad variability of the tracer penetrationmethodology with regard to test parameters has been con-firmed by a review study that evaluated the reliability of thisspecific test method [6]. But even if the same methodologywas applied in conjunction with the same material, multicen-ter studies yielded different results for Class V fillings [7] andfissure sealants [8].
Likewise, due to the microscopic evaluation of restorativemargins, the methodology differs quite a lot between one
institute and the other. Another influencing factor that con-tributes to the bias of the laboratory method is that the resultsfrom quantitative margin analysis may differ up to 20%, whichhas been confirmed in a study where different operators evalu-
5 ( 2 0 0 9 ) 605–620
ated the same margins [9]. A recently published new approachwhich uses an optical device that automatically measures therestorative margins eliminates the operator bias but can onlybe applied in cylindrical cavities [10,11].
As up to now no comprehensive systematic research hasbeen carried out to correlate the in vitro findings in regard to themarginal adaptation of restorations with the clinical outcome,the goal of the present study was to elucidate the in vitro/in vivorelationship by selecting clinical outcome variables (retention,marginal discoloration, marginal integrity) of prospective clin-ical trials on Class V restorations and relate them to laboratorydata of the same materials in Class V restorations coming fromtwo research institutes (Berlin, Zurich).
The following hypotheses were examined:
1. There is no correlation between in vitro and in vivo findings.2. The correlation is independent of whether the same com-
posite material has been used for the in vivo and in vitrostudy.
3. The correlation is independent of whether the cavity inthe clinical trial has been prepared or not or whether theenamel has been beveled or not.
4. There is no correlation between both laboratory test meth-ods.
2. Materials and methods
2.1. Evaluation of restorative margins in vitro
Two research institutes (University of Berlin, University ofZurich) have established standardized test methods to evalu-ate adhesive systems in extracted teeth in the laboratory. Sincethe late 1980s, adhesive systems are routinely tested at bothinstitutes. Therefore, they possess a multitude of data on thedifferent adhesive systems; most of the data have not beenpublished in dental journals. Both test methods involve thequantitative analysis of the restorative margin of Class V cer-vical fillings using SEM at high magnification. However, bothtest methods differ with regard to the type of tooth, cavitypreparation, artificial ageing of restorations and evaluationcriteria.
The test method developed at the University of Zurich isdescribed in detail elsewhere [12,13]. This method was devel-oped to evaluate in vitro the suitability of restorative materialsto meet the Swiss quality guidelines for restorative den-tistry. In these guidelines, a clinical service time of more than5 years has to be assured (SSO 1999). Hence, an in vitro setupwas calculated, which should simulate this service period. Inthe following, a short description is given. In extracted pre-molars, six Class V cavities are prepared per material group,placing them both at the lingual and buccal side. Since 1998eight restorations are placed per group. One day before toothpreparation, the teeth are connected to a dentinal fluid simu-lation device, which contains horse serum diluted with Ringersolution. Thus, a hydrostatic pressure of about 25 mm Hg
is established within the tooth. Two Class V wedge-shapedcavities are prepared at the cemento-enamel junction undera stereo microscope at 12× magnification, one on the lin-gual and one on the buccal surface. Half of the cavity is
ocated in dentin, the other half in enamel. It features theollowing dimensions: mesial–distal 3–3.5 mm, apical–coronal.5–3 mm, depth 1.5 mm. A 1-mm wide bevel is placed at thenamel margin and rounded with flexible aluminum oxideiscs. The adhesives are applied according to the manufac-urer’s instructions and light-cured. Finally, two layers of theespective composite material are applied and light-cured sep-rately, the first layer covering the dentinal margin with ahickness of 1.3–1.5 mm and the second covering the enamel
argin. Excess is immediately removed using flexible alu-inum oxide discs of decreasing roughness. After having
tored the specimens in distilled water for 7 days at 37 ◦C, allhe teeth are subjected to 3000 thermal cycles (5 ◦C–55 ◦C–5 ◦C,.e. 6000 temperature changes) and 1,200,000 load cycles at9 N with a frequency of 1.7 Hz in a chewing simulator (ther-omechanical loading TML) with integrated dentinal fluid
imulation; the chewing simulator is described elsewhere [14].or the SEM analysis, impressions of the restorations for theeplicas are made before and after thermomechanical loading.he replicas are analyzed with a SEM at 200× magnificationy using a computerized analysis program to determine theercentage of continuous margin of the entire dentinal andnamel margin (modified NIH Image, National Institute ofealth, USA). A continuous margin is defined as having noaps, irregularities or fractures. The dentinal and enamel partf the restorative margin is reported on separately.
At the University of Berlin, extracted upper central incisorsre used to prepare a Class V cavity on the labial surface. Eightllings are placed per material group. On the apical side, a 90◦
ngle is prepared with the margin in dentin, whereas on theoronal side the enamel is beveled with a finishing diamondur [15,16]. The cavity dimensions are as follows: apical-oronal 4 mm, mesial-distal 3 mm, and depth 1.5 mm. Onealf of the margin length is located in dentin. The adhesiveystem is applied according to the manufacturer’s instruc-ions and light-cured. The composite material is applied inwo layers starting at the cervical margin and light-curedeparately. After completion of the restoration the excess isemoved with flexible discs, the restoration is polished andhe teeth are stored in water at room temperature for 3 weeksrior to thermocycling (2000× 5 ◦C/55 ◦C, i.e. 4000 temperaturehanges). Before and after thermocycling, the restorative mar-in is quantitatively assessed using SEM (200× magnification),hereby four different criteria are allocated to each part of theargin:
Criterion 1. Margin not or hardly visible, no gap.Criterion 2. No gap but severe irregularities such as porositiesor marginal fractures.Criterion 3. Visible gap up to 2 �m.Criterion 4. Severe gap exceeding 2 �m.
For each criterion, the percentage related to the entire mar-in length is calculated with the help of a software program.or the present study, only data after thermocycling were usedtime 1 = T1).
When the University of Berlin started investigating adhe-ive systems that involved etching of the enamel, only theentinal margin was evaluated as the enamel margin waslways close to perfect. However, an inferior quality of the
( 2 0 0 9 ) 605–620 607
enamel margins was observed in conjunction with self-etching adhesive systems and in most material groups, theenamel margin was evaluated separately (for these materials)using the same criteria.
The specimens fabricated with some AS were subjected toprolonged storage in water and additional thermocycling atdifferent time points T2-T5:
• T2: after 1 year water storage;• T3: after a second phase of thermocycling (2000×);• T4: after 2 additional years of water storage (total of 3 years’
water storage);• T5: after a third phase of thermocycling (2000×).
2.2. Selection of clinical trials on Class V restorations
For the selection of clinical trials, prospective clinical stud-ies on Class V restorations were searched in MEDLINE andthe IADR abstract database. Additionally, a hand search of theGerman dental journal “Deutsche Zahnärztliche Zeitschrift”,which is not available in MEDLINE, was carried out. The searchwords were “Class V” and “clinical”, the search period was June2007. The inclusion criteria were as follows:
1. Prospective clinical trial involving at least 1 adhesive sys-tem in cervical Class V cavities.
2. Minimal duration of 18 months.3. The following criteria had to be investigated for: retention,
marginal discoloration, marginal integrity, and secondarycaries. These criteria were defined as “outcome variables”.
Studies that involved different cavity classes (Class V andClass III) were excluded.
For each clinical trial, the following data had been includedin the data file:
• type of composite material per adhesive system;• number of restorations per material at baseline and each
follow-up/recall;• number of subjects per material at baseline and each follow-
up/recall;• preparation of dentin yes/no;• beveling of enamel yes/no;• the outcome variables per material at each recall visit.
If a clinical trial investigated the effect of etching theenamel by comparing the results with those of etch & rinseadhesives, only the data of the etching group were selected.
2.3. Statistical analysis
The following clinical outcome variables were defined and cor-related with the results of the in vitro measurements:
100 − % of retention loss (in what follows 100 − R).100 − % of marginal discoloration (in what follows 100 − MD).
100 − % of detectable margins (in what follows 100 − MI).
In order to summarize the in vivo performance, we com-bined these three clinical outcomes into one single clinical
Table 1 – Adhesives systems and their allocated numberevaluated in 44 clinical trials.
1 A.R.T. Bond2 AdheSE3 Admira Bond4 All Bond 25 Clearfil Liner Bond6 Clearfil Liner Bond 27 Clearfil Protect Bond8 Clearfil SE Bond9 Dyract PSA
10 EBS11 Excite12 F2000 SEP13 Fuji Bond LC14 Fuji II LC15 Futurabond NR16 GC Conditioner17 Gluma 200018 Gluma Solid Bond19 Hybrid Bond20 iBond21 One Coat Bond22 One Step23 OptiBond FL24 OptiBond Solo25 Permaquick26 Prime & Bond 2.127 Prime & Bond NT28 Prisma U Bond 329 Prompt-L-Pop30 Single Bond31 Scotchbond Multipurpose32 Syntac33 Syntac Single Component34 Tenure35 Tripton
This index was calculated at 12, 18, 24 and 36 months,obtaining hence IND12, IND18, IND24 and IND36.
For the Berlin data, dentin, respectively dentin and enamel,were recorded at five or three different time points, respec-tively (T1–T5, see above). in vitro data at T1 on dentin wereavailable for 32 adhesive systems, for which also data of in vivostudies were available; the number of in vivo experiments inrelation to the number of adhesive systems was 76, as manystudies evaluated more than one adhesive systems. As datafor dentin at T2, T3,T4 and T5 were only available for 11 adhe-sive systems, we restricted our attention to dentin on T1 (afterthermocycling). Original data were expressed as a distributionon a four-point ordinal scale (for each experiment, we had thus4 percentages corresponding to the values of 1–4). To combinethis information into a single index, we considered the averageof the
Berlin index
= 1 × % of 1 + 2 × % of 2 + 3 × % of 3 + 4 × % of 4100
.
Thus an index of 1 corresponds to the best possible distri-bution (100% of 1, 0% of 2, 0% of 3, 0% of 4), whereas an index of4 corresponds to the worst possible distribution (0% of 1, 0% of2, 0% of 3, 100% of 4). In the analysis below, we also considered100 − % of 1 as a second indicator of the in vitro performancein Berlin.
In conjunction with the Zurich method, dentin or dentinand enamel data (both before and after thermomechanicalloading TML) were available for 28 adhesive systems. In vivodata were also available for these systems. The number ofin vivo experiments in relation to the number of adhesivesystems was 70. Two different indices were defined to char-acterize the in vitro performance:
Delta D = dentin before − dentin after TML,
Delta DE = dentin/enamel before−dentin/enamel after TML.
Note that all the indices characterizing the in vivo andin vitro performance were defined in such a way thatlarger values corresponded to a worse performance of theadhesive system. Therefore, positive correlations betweenthese indices would indicate a positive association betweenthem.
One goal of the present study was to explore at whichrecall and with which outcome variable of the in vivo measure-ments the in vitro measurements were the most correlated. Forthis purpose, we calculated a Spearman correlation coefficient
rho between each in vitro index and each in vivo measure-ment/index, and plotted the data using scatter diagrams. Thenumbers in these diagrams refer to the number of the adhesivesystem (see Table 1).
36 Tyrian37 Xeno III
We could not test the statistical significance of thesecorrelations since the experiments were not conducted inde-pendently (many studies involved more than one adhesivesystem which were partly applied in the same subjects). Totest the significance of the relationship between in vivo andin vitro indices, while accounting for this dependence, weused a linear mixed model which included a random studyeffect:
In vivo index = a+b×in vitro index+study effect+random error.
Since each study involved a different number of patients(sample sizes varied from 11 to 81), and thus the results did notshow the same degree of precision, we weighted the differentexperiments by their sample sizes. The p-value for the slope babove was calculated using SPSS (release 13.0) and based on a“pseudo-degree of freedom” (indicator of the real sample sizeavailable).
Predictive values had been calculated based on cut-off val-
ues. For the clinic index IND24, the cut-off value was set at 10%after 2 years. For the Berlin index, a cut-off value of 1.2 and forthe Zurich Delta index a cut-off value of 10% was defined.
2 5
d
p
wiet
d
p
wBot
3
Spmtar
trodcO(at
ttdroFaitomimpib
yttt0
d e n t a l m a t e r i a l s
A predictive value for poor performance of the AS wasefined as follows:
V = a
a + b,
here a = number of experiments with IND24 >10% and Berlinndex >1.2, respectively Zurich index >10%; b = number ofxperiments with IND24 >10% and Berlin index <1.2, respec-ively Zurich index <10%.
A predictive value for good performance of the AS wasefined as follows:
V = d
c + d,
hereby d = number of experiments with IND24 <10% anderlin index <1.2, respectively Zurich index <10%; c = numberf experiments with IND24 <10% and Berlin index >1.2, respec-ively Zurich index >10%.
. Results
ixteen clinical studies were identified with an observationeriod of 18 months, 31 with an observation period of 24onths, 18 with an observation period of 36 months and only
wo studies with an even longer observation period (60 monthsnd 84 months, respectively). Of these studies, 28 also reportedesults after 12 months (Table 2).
As most clinical data were available for 12 and 24 months,he main analysis was limited to these recall intervals. Asegards the in vivo data, 158 out of the 170 percentages of sec-ndary caries (SC) across the different experiments and theifferent time points were equal to 0. This is why we did notonsider SC when characterizing the “in vivo performance”.n the other hand, the average percentages of retention
R), marginal discoloration (MD) and marginal integrity (MI)cross studies and time points were 8%, 13% and 16%, respec-ively.
Clinical data were available for 31 adhesive systems (AS)hat could be compared to the Berlin data, and for 28 AShat could be compared to the Zurich data. For the Berlinata, we could find 19 AS for which the composite mate-ial was the same as in the clinical study, whereas this wasnly the case for 13 AS investigated with the Zurich method.or three AS which had been tested clinically (nos. 14, 16nd 36), no in vitro data were available neither in Berlin norn Zurich. Therefore, these AS do not appear in the scat-er plots below (Figs. 2–5) which thus contain a maximumf 34 AS. As far as the clinical studies are concerned, theajority of them (19 studies) only evaluated one AS, 18 stud-
es evaluated two AS, 5 studies three AS and two studies asany as four AS. In 14 of the 44 studies, the dentin was pre-
ared and in 19 studies the enamel was beveled; however,n only 7 studies the dentin was prepared and the enameleveled.
Fig. 1 shows the clinical index for the respective AS at the 2-ear recall (IND24). One can see a large variability, even among
he studies using the same AS. For instance the results forhe AS no. 30 (Single Bond) varied between 5% and 20% andhose for the AS no. 31 (Scotchbond Multipurpose) between% and 27%. The results were not very different when only
( 2 0 0 9 ) 605–620 609
those studies were considered for which the same compositehad been used.
The scatter diagrams in Figs. 2–5 show the correlationbetween the in vitro indices (the Berlin index in Figs. 2 and 3and the Zurich Delta D index in Figs. 4 and 5) and the in vivomeasurements for the different recall intervals (12, 18, 24,36 months), the different outcome variables (R, MD, MI) andthe summarized clinical index IND. In all the figures, plots ofeach experiment are shown indicating the number of the ASused (see Table 1). Most of the correlations in Figs. 2 and 4were low, with a few exceptions, such as the correlationrho = 0.61 between MD after 18 months and the Zurich Delta Dindex, which could be a chance. Similar low correlations wereobtained using the percentage of 1 in Berlin or the Zurich DeltaDE index.
Interestingly, these correlations increased when restrict-ing our attention to those experiments which used the samecomposite (in vivo and in vitro), especially in conjunction withthe Berlin data where we had several correlations larger than0.5 (Fig. 3), and to some lesser extent in conjunction with theZurich data (Fig. 5). Note that only a few experiments with thesame composite were available for the Zurich index at the 12and 18 months recall, so that the corresponding correlationswere not meaningful.
As indicated in the statistical section, it was not possibleto directly test the significance of these correlations, so thatwe used linear mixed models instead. As it is not a good sta-tistical practice to perform too many tests, we restricted ourattention to the in vivo indices after 12 and 24 months (IND12and IND24), which were summary indices without too manymissing values.
Table 3 contains the estimates of the regression slopes fromthese linear mixed models, together with standard errors,t-statistics, pseudo-degrees of freedom and p-values, for dif-ferent subsets of experiments. Note that one convergenceproblem occurred because of the small sample size available(reported as NA in Table 3). The slope indicates the aver-age increase of the in vivo index when the in vitro indexincreases by one unit. For instance, a slope of 11.8 for theBerlin index means that the increase of 1 point on this invitro scale leads to an average increase of 11.8 on the clinicalscale.
No significant result could be found when considering allexperiments. The only clearly significant results (p = 0.005 orsmaller) could be found for the Berlin index (as well as for thepercentage of 1 in Berlin) when considering only those exper-iments which used the same composite, with the significancebeing even higher after 24 months than after 12 months. Forthe Zurich index of dentinal margins, the corresponding resultwas almost significant after 24 months (p = 0.061), but not atall significant after 12 months. When taking the whole margin(dentin and enamel) into account, results were not at all sig-nificant, the slope being even negative after 12 months. Whenconsidering only clinical studies for which the dentin had beenprepared or for which the enamel had been beveled, no signif-icant positive result could be found, neither in conjunction
with the Berlin data nor in conjunction with the Zurich data.For the Zurich Delta DE index at 12 months, a slightly signifi-cant negative slope could be found when considering beveledenamel only (p = 0.035).
610 d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620
Table 2 – Selected clinical studies on Class V restorations with the adhesive system AS (for number see Table 1) that wasevaluated in relation to the data from Berlin and Zurich and the respective AS (with and without the same composite inthe clinical and the in vitro study).
Study number Study Adhesive system Berlin AS Berlin samecomposite
Cut-off values for characterizing the good and poor perfor-mance of an AS can be defined, and it can be checked whetherthe in vitro performance correctly predicts in vivo performance.If the cut-off value for the clinical index after 2 years was setat 10%, the cut-off value for the Berlin index at 1.2 and thecut-off value for the Zurich Delta D index at 10%, 30–40% ofthe AS were predicted to perform poorly (using either in vitroindex) when indeed the clinical studies showed the contrary(Table 4). The result was similar when the same composite was
used in the in vitro and in vivo experiments.
Fig. 6 illustrates the correlation between the Berlin andZurich in vitro indices. This correlation was rather weak(rho = 0.12) and slightly increased (rho = 0.36) when consider-
31 31
ing only experiments conducted with the Berlin and Zurichmethod using the same composite.
4. Discussion
This is the first study which systematically evaluated the rela-tionship between laboratory data on marginal integrity andthe clinical outcome in Class V restorations. As laboratory
methods are claimed to have clinical significance withoutproviding scientific proof, carrying out a systematic studyof this type seemed overdue. In the laboratory, the develop-ment of marginal discoloration or secondary caries (caries
d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620 611
F (1–3o r oth
aentctm
ig. 1 – Clinical results in relation to the adhesive system ASf the AS no clinical result was available after 2 years but fo
djacent to restorations) cannot be simulated. Therefore, thestablished laboratory methods use surrogate parameters,amely the percentage of continuous margin in relation to
he entire margin, to predict the clinical performance of aertain adhesive system or a combination of adhesive sys-em and overlying composite. A low percentage of continuous
argin should be correlated with a high percentage of dis-
Table 3 – Estimates of regression slopes and associated standarp-values from linear mixed model with random study-effects toand 24 months (IND24) (below) from two in vitro indices where
7) at the 2-year recall (clinical index IND24). Note: For someer recall intervals.
colored margin, secondary caries and also retention loss inClass V restorations. The hypothesis is that even smallmarginal defects or gaps of 1–3 �m width allow the penetra-
tion of cariogenic bacteria and/or the retention of pigmentsthat lead to marginal discoloration. An increased influx orpercolation of liquids and saliva eventually leads to debond-ing and retention loss in Class V fillings. A recently published
d errors, t-statistics, pseudo-degrees of freedom andpredict an in vivo index for 12 months (IND12) (above)
612 d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620
Fig. 2 – Correlations with Spearman correlation coefficient rho between in vivo measurements and the Berlin in vitro indextion,stem
at three recall intervals (12, 18, 24, and 36 months). R = retenIND = clinical index. The numbers represent the adhesive sy
study tried to simulate the clinical findings of retention loss by
means of restorations placed in cervical defects of extractedpremolars without prior preparation using a one-step self-etching adhesive system, which showed high retention lossin clinical trials (>20% after 2 years). The specimens were sub-
MD = marginal discoloration, MI = marginal integrity,(see Table 1).
jected to 1,200,000 cycles of thermomechanical centric loading
followed by 1,200,000 cycles of eccentric loading on the lingualcusp [17]. No single retention loss was observed through-out the whole in vitro procedure, which indicates that factorsother than gaps and water influx contribute to debonding
d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620 613
Fig. 3 – Correlations with Spearman correlation coefficient rho between in vivo measurements and the Berlin in vitro indexw , 18,d e nu
ammap
hen using the same composite at three recall intervals (12iscoloration, MI = marginal integrity, IND = clinical index. Th
nd loss of retention. Even water storage of the restorations
ade with eight self-etching adhesive systems for up to 12onths, multiple thermocycling intervals (up to 30,000 cycles)
nd mechanical loading did not result in any retention loss asreliminary results showed (unpublished data). Another indi-
24, and 36 months). R = retention, MD = marginalmbers represent the adhesive system (see Table 1).
cation of the fact that laboratory tests do not simulate what
happens intraorally is that no loss of restoration was observedwith any of the tested adhesive systems neither in Berlin norin Zurich, even after 1,200,000 cycles of thermomechanicalloading (Zurich) or prolonged water storage (Berlin; [18]). In
614 d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620
Fig. 4 – Correlations with Spearman correlation coefficient rho between in vivo measurements and the Zurich in vitro indexretenstem
(�D) at three recall intervals (12, 18, 24, and 36 months). R =IND = clinical index. The numbers represent the adhesive sy
contrast, the clinical data showed that 20–50% of the restora-
tions were lost within a period of only 2–3 years when thesame materials were used. In addition to thermomechanicalloading, the Zurich method involves the simulation of denti-nal fluid flow during the application of the adhesive system
as well as during the chewing simulation. It has been proven
that the simulation of dentinal fluid flow further decreasesthe amount of continuous margins in conjunction with someadhesive systems, but does not lead to any retention loss[12].
d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620 615
Fig. 5 – Correlations with Spearman correlation coefficient rho between in vivo measurements and the Zurich in vitro index( s (12d e nu
Tifa
�D) when using the same composite at three recall intervaliscoloration, MI = marginal integrity, IND = clinical index. Th
However, the present study has several shortcomings. (1)
he cavity design and preparation technique used for the clin-
cal trials differ from those of the laboratory study and alsorom clinical study to clinical study. (2) The parameters thatre measured both in the clinical trial and with the laboratory
, 18, 24, and 36 months). R = retention, MD = marginalmbers represent the adhesive system (see Table 1).
method are different. (3) The artificial ageing of the specimens
(thermocycling, mechanical loading) do not necessarily reflectthe clinical situation. (4) The clinical trials show a variabilityfor the same materials with regard to the outcome variables,thus making comparisons difficult to interpret.
616 d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620
x: (le
Fig. 6 – Correlations between Berlin and Zurich Delta D indecomposite was used.
Especially the inconsistent clinical results obtained withthe same materials and material combinations are worthmentioning and further complicate the correlation between invitro and in vivo data. This inconsistency can be attributed tovarious reasons and is subject to speculation. No systematicstudies have been carried out to elucidate the possible rea-sons. Factors that may play a role are different operators (nocalibration), different operative techniques (no standardiza-tion), different patients (no stratified randomization), samplessize (low number of subjects), different evaluators (no calibra-tion), different outcome variables or other factors. Therefore,a standardization of clinical trial design is required. Recently,a group of research workers came up with recommendationsfor the design of studies and criteria to evaluate restorations[19].
As far as the selection of studies is concerned, also studies(8 of 44) were included for which only an IADR abstract wasavailable. This was done to back up the comparative analy-sis by the broadest possible range of data and to also includenewer AS for which only IADR abstracts are available so far.IADR abstracts have also been included in other systematicstudies like the one conducted by Peumans et al., which eval-uated the retention loss rate of contemporary AS [20]. Somestudies selected for the present study were first published asIADR abstract before being published in peer-reviewed jour-nals, e.g. [21–23].
The rationale for creating a clinical index is that a betterstatistical analysis can be carried out by correlating one in vivoand one in vitro index. The weighing of the three outcome vari-ables attributing retention loss a fourfold weighing, marginaldiscoloration a twofold weighing and marginal integrity a one-fold weighing was based on the following considerations: (1)the retention loss in a Class V cavity is the most obvious signof failure of an adhesive system and is also the most reliablediagnostic evaluation criterion with little variability betweendifferent evaluators. (2) Marginal discoloration and especially
marginal integrity are outcome variables which may show agreater variability between different evaluators.
For the in vitro results, an index was created too. Forthe Berlin index, all four marginal adaptation criteria were
ft) all cases; (right) only those for which the same
included in one index, which provided a better correlationwith the clinical index than when only criterion 1 (percent-age of continuous margin) was used. In regard to the Zurichindex, better correlation with the clinical index was achievedwhen using the difference between continuous margin afterthermomechanical loading and before than when only usingthe percentage of continuous margin after loading, which isnormally done in publications applying this methodology. Theresults of the latter criterion are not shown in the presentpublication.
One may argue that the observation period of 2–3 years istoo short to detect significant differences between the materi-als, especially when one assumes that, e.g. the Zurich in vitrodata should simulate 5 years in vivo [24]. However, most ofthe studies on Class V restorations have a duration of only3 years. The reason for the short observation period may bethat the American Dental Association’s acceptance period foran adhesive system to be evaluated in Class V cavities is 18months only. There were only two studies which followed upthe restorations for 5 and 7 years, respectively [21,25]. Recently,a study was published showing the performance of sevenadhesive systems over a period of 13 years [26]; but only reten-tion loss was reported. This study has shown that Class Vrestorations placed with adhesive systems that demonstratea retention loss of around 10–20% within the first 5 years ofservice may exhibit a sharp increase in retention loss in theyears thereafter (up to 50–60%).
The correlation between clinical and laboratory data onadhesive systems was weak and only present if studies weretaken into account that evaluated the same composite mate-rial. It came as a surprise that the composite material playeda crucial role in the comparative analysis, as the general viewhas always been that in Class V cervical fillings the adhesivesystem is more decisive than the overlying composite. Per-haps the influence of the composite resin on the longevity ofClass V fillings has been underestimated in the past. How-
ever, no systematic analysis was performed that used thesame adhesive system but various composite resins in thesame patient. No correlation between in vitro and in vivo datawas found between clinical studies in which beveling of the
d e n t a l m a t e r i a l s 2 5 ( 2 0 0 9 ) 605–620 617
Table 4 – Prognosis of the good and bad performance of an AS according to in vitro test results based on defined cut-offvalues (see text).
Method in vitro Prognosis for good performance of AS Prognosis for bad performance of AS
Berlin—all AS 60% 72%
emmlfcTibt
aldlltomaicbdf
vdff2bdfgdmw
umlmamrhTtec
Berlin—same composite 78%Zurich—all AS 67%Zurich—same composite 100%
namel and preparation of the dentin was performed. Oneight assume a better correlation, as beveling the enamelargin and preparation of the dentin was carried out in the
aboratory study as well. As the number of cases was too smallor the group “preparation + beveling + same composite”, noorrelation coefficient was calculated for this combination.he results of the present study suggest, that the compos-
te material plays a more important role for the correlationetween laboratory and clinical indices than dentin prepara-ion or beveling of the enamel.
Time could be another important factor. The artificialgeing procedure including thermocycling and mechanicaloading may not reflect what is happening clinically as theegradation of the adhesive/tooth substance interface is a
ong-term process. There are no Class V restorations that areost as early as during the first month after placement. In ordero obtain meaningful bond strength data, 3–6 months’ storagef the specimens in water prior to testing them is recom-ended [1,27]. The data obtained for some adhesive systems
fter prolonged storage in water using the Berlin methodndicated a rapid deterioration of the marginal adaptation inonjunction with some adhesive systems [18]. But as the num-er of adhesive systems for which prolonged water storageata were available was very limited, calculating correlationsor these groups was not deemed reasonable.
When defining cut-off values for good or bad in vivo and initro performance as it was done in the present study, the pre-ictive value for bad and good performance was 67% and 78%or the Berlin index and 60% and 100% for the Zurich index;or the latter, however, only three cases were included. So, in0–40% of the cases a false prediction may be given, mostlyecause a bad performance is ascribed to an adhesive systemue to the laboratory test results, when indeed its clinical per-ormance may be equal to that of an adhesive system withood in vitro results. However, this discrepancy may partly beue to the simulated clinical service in vitro of 5 years, which isuch longer than the relatively short-term clinical data whiche have available for comparison.
Due to the fact that laboratory studies can be conductednder standardized, optimal conditions, clinical observationsay show a different picture and not reproduce data from
aboratory studies. Restorations with microscopically soundargins cannot be produced under clinical conditions. Instudy, in which Class II restorations were placed in pre-olars and extracted after 4–6 months due to orthodontic
easons, the SEM analysis showed that 43% of the cavitiesad been overfilled and 25% had been insufficiently filled [28].
his result was achieved independent of the adhesive system,
he layering technique and the matrix system used, or thexperience of the operator. In a prospective study on Class IIomposite restorations conducted over a period of 10 years, in
67%64%60%
which the marginal adaptation was examined clinically andon replicas by means of SEM every year, showed that alreadyat the 1-year recall more than 90% of the restorations hadmargins which were not perfect on more than two-thirds oftheir length, as SEM analysis of the replicas revealed [29,30].Clinically, less than 5% of the restorations had detectable mar-gins at a length of less than one-third. At the 3-year recallstained margins were observed in about 25% of the restora-tions, whereas more than 90% of them had imperfect marginson the replicas. The marginal imperfections identified, eitherthose detected with SEM on the replicas or those observed clin-ically, did not influence the survival rate of the restorations orthe development of secondary caries.
In a study on Class I molar restorations with five differentadhesive systems, where the in vitro results were comparedto the clinical results by the same researchers, the percentageof gap-free margins was similar in vitro and in vivo. In con-trast to the present study, a high degree of standardization forplacement and evaluation was given [31]. However, the studydid not answer the question about what are the clinical impli-cations of open margins. In another article it was assumedthat poor marginal quality might decrease clinical longevitydue to the misdiagnosis of secondary caries, which leads toovertreatment as the restoration is replaced prematurely [32].
As there is only a weak correlation between the presence ofgaps and the clinical outcome, one may assume that not thepercentage of margins with gaps, but rather the width anddepth of the marginal gap could be a more significant factor.However, no clinical studies exist, which have systematicallyexamined the relationship between the width of the marginalgap and the occurrence of secondary caries and studied thefactors of caries activity, location in the mouth and the type ofrestorative material used. Nevertheless, there are indicationsthat the width of the marginal gap per se is not a prognos-tic indicator for the occurrence of secondary caries. In onestudy on amalgam restorations, parts of the occlusal marginof composite restorations with imperfections, discolorationand marginal gaps were completely removed (tooth struc-ture + restoration) and histologically examined [33]. Marginalgaps were considered to be closely linked to secondary cariesonly if a periodontal probe whose tip had a diameter of 400 �mwas able to penetrate into the gap. In areas such as the prox-imal gingival floor of Class II restorations, which are moredifficult to reach with oral hygiene measures and self-cleaningmechanisms, the marginal gap width in association with sec-ondary caries is most probably smaller. As systematic studiesare lacking, this can only be assumed. The proximal margin
of Class II restorations, however, is the site where 80% of thesecondary carious lesions occur [34].
In summary, the systematic analysis of the correlationbetween laboratory data of marginal adaptation and the
l s 2
r
618 d e n t a l m a t e r i a
outcome of clinical trials of Class V restorations revealedthat the correlation was weak and only present if studieswere compared which used the same composite for the invitro and in vivo evaluation. Therefore, hypothesis 1 and 2had to be partly rejected and hypothesis 3 was accepted.However, one laboratory method (Berlin) showed a moreconsistent and more significant correlation with the in vivofindings than the other method (Zurich), where the speci-mens were stressed more heavily and for a longer periodof time. No correlation was found between both in vitromethods. The variable outcome of different clinical studiesof Class V restorations involving the same materials maybe caused by the lack of a standardized study design andevaluation criteria as well as calibrated operators and eval-uators.
e f e r e n c e s
[1] ISO. Dental materials—testing of adhesion to toothstructure. Technical Specification; 2003. No. 11405.
[2] Heintze SD. Systematic reviews: (1) The correlation betweenlaboratory tests on marginal quality and bond strength. (2)The correlation between marginal quality and clinicaloutcome. J Adhes Dent 2007;9(Suppl 1):77–106.
[3] Schuckar M, Geurtsen W. Proximo-cervical adaptation ofClass II-composite restorations after thermocycling: aquantitative and qualitative study. J Oral Rehabil1997;24:766–75.
[4] Dietrich T, Krämer M, Lösche GM, Wernecke KD, Roulet JF.Influence of dentin conditioning and contamination on themarginal integrity of sandwich Class II restorations. OperDent 2000;25:401–10.
[5] Heintze SD, Forjanic M, Cavalleri A. Microleakage of Class IIrestorations with different tracers—comparison with SEMquantitative analysis. J Adhes Dent 2008;10:259–67.
[6] Raskin A, D’Hoore W, Gonthier S, Degrange M, Dejou J.Reliability of in vitro microleakage tests: a literature review.J Adhes Dent 2001;3:295–308.
[7] Raskin A, Tassery H, D’Hoore W, Gonthier S, Vreven J,Degrange M, et al. Influence of the number of sections onreliability of in vitro microleakage evaluations. Am J Dent2003;16:207–10.
[8] Raskin A, Courson F, Nanc J, Droz D, Muller-Boll M, Bouter D,et al. In vitro microleakage multicentered study of twosealants. J Dent Res 2005;84(Spec Issue B):Abstract No. 213(Continental European and Scandinavian Divisions)(www.dentalresearch.org).
[9] Henisch G. Ein computergestütztes System zur Erfassungund Verwaltung von Messdaten im Rahmen derquantitativen Randanalyse am Rasterelektronenmikroskop.Berlin: Free University; 1989.
[10] Heintze SD, Forjanic M, Jakob G. Automatic gap detection atrestoration margins with an optical sensor in vitro. J AdhesDent 2005;7:95–105.
[11] Heintze SD, Forjanic M, Roulet JF. Automated marginanalysis of contemporary adhesive systems invitro—evaluation of discriminatory variables. J Adhes Dent2007;9:359–69.
[12] Krejci I, Kuster M, Lutz F. Influence of dentinal fluid and
stress on marginal adaptation of resin composites. J DentRes 1993;72:490–4.
[13] Peters O, Göhring TN, Lutz F. Effect of eugenol-containingsealer on marginal adaptation of dentine-bonded resinfillings. Int Endod J 2000;33:53–9.
5 ( 2 0 0 9 ) 605–620
[14] Krejci I, Reich T, Lutz F, Albertoni M. In- Vitro- Testverfahrenzur Evaluation Dentaler Restaurationssysteme1. Computergesteuerter Kausimulator. Schweiz MonatsschrZahnmed 1990;100:953–60.
[15] Blunck U, Roulet JF. In vitro marginal quality ofdentin-bonded composite resins in Class V cavities.Quintessence Int 1989;20:407–12.
[16] Blunck U, Roulet JF. Marginal adaptation of compomerClass V restorations in vitro. J Adhes Dent 1999;1:143–51.
[17] Heintze SD, Cavalleri A. Retention of restorations placed innon-carious cervical lesions after centric and eccentricocclusal loading in a chewing simulator—a pilot study.J Adhes Dent 2006;8:169–74.
[18] Blunck U, Zaslansky P. Effectiveness of all-in-one adhesivesystems tested by thermocycling following short andlong-term water storage. J Adhes Dent 2007;9(Suppl 2):231–40.
[19] Hickel R, Roulet JF, Bayne S, Heintze SD, Mjör IA, Peters M, etal. Recommendations for conducting controlled clinicalstudies of dental restorative materials. Clin Oral Investig2007;11:5–33.
[20] Peumans M, Kanumilli P, De Munck J, Van Landuyt K,Lambrechts P, Van Meerbeek B. Clinical effectiveness ofcontemporary adhesives: A systematic review of currentclinical trials. Dent Mater 2005;21:864–81.
[21] Kubo S, Kawasaki K, Yokota H, Hayashi Y. Five-year clinicalevaluation of two adhesive systems in non-carious cervicallesions. J Dent 2006;34:97–105.
[22] Peumans M, Munck J, Van Landuyt K, Lambrechts P, VanMeerbeek B. Three-year clinical effectiveness of a two-stepself-etch adhesive in cervical lesions. Eur J Oral Sci2005;113:512–8.
[23] Perdigão J, Carmo AR, Geraldeli S. Eighteen-month clinicalevaluation of two dentin adhesives applied on dry vs moistdentin. J Adhes Dent 2005;7:253–8.
[24] Krejci I, Lutz F. In-Vitro-Testverfahren zur Evaluationdentaler Restaurationssysteme. Korrelation mitIn-vivo-Resultaten. Schweiz Monatsschr Zahnmed1990;100:1445–9.
[25] Peumans M, De Munck J, Van Landuyt K, Lambrechts P, VanMeerbeek B. Seven-year clinical effectiveness of two 3-stepetch-and-rinse adhesives. J Dent Res 2005;84(Spec IssueB):Abstract No. 147 (Continental European and ScandinavianDivisions) (www.dentalresearch.org).
[26] Van Dijken JW, Sunnegardh-Gronberg K, Lindberg A. Clinicallong-term retention of etch-and-rinse and self-etchadhesive systems in non-carious cervical lesions—A 13years evaluation. Dent Mater 2007;23:1101–7.
[27] De Munck J, Van Landuyt K, Peumans M, Poitevin A,Lambrechts P, Braem M, et al. A critical review of thedurability of adhesion to tooth tissue: methods and results.J Dent Res 2005;84:118–32.
[28] Opdam NJ, Roeters FJ, Feilzer AJ, Smale I. A radiographic andscanning electron microscopic study of approximal marginsof Class II resin composite restorations placed in vivo. J Dent1998;26:319–27.
[29] Gaengler P, Hoyer I, Montag R. Clinical evaluation ofposterior composite restorations: the 10-year report. J AdhesDent 2001;3:185–94.
[30] Gaengler P, Hoyer I, Montag R, Gaebler P. Micromorphologicalevaluation of posterior composite restorations—a 10-yearreport. J Oral Rehabil 2004;31:991–1000.
[31] Frankenberger R, Krämer N, Lohbauer U, Nikolaenko SA,Reich SM. Marginal integrity: is the clinical performance ofbonded restorations predictable in vitro? J Adhes Dent
2007;9(Suppl 1):107–16.
[32] Sarrett DC. Prediction of clinical outcomes of a restorationbased on in vivo marginal quality evaluation. J Adhes Dent2007;9(Suppl 1):117–20.
[33] Kidd EA, Beighton D. Prediction of secondary caries aroundtooth-colored restorations: a clinical and microbiologicalstudy. J Dent Res 1996;75:1942–6.
[34] Mjör IA. The location of clinically diagnosed secondarycaries. Quintessence Int 1998;29:313–7.
[35] Alhadainy HA, Abdalla AI. 2-Year clinical evaluation ofdentin bonding systems. Am J Dent 1996;9:77–9.
[36] Van Meerbeek B, Peumans M, Gladys S, Braem M,Lambrechts P, Vanherle G. Three-year clinical effectivenessof four total-etch dentinal adhesive systems in cervicallesions. Quintessence Int 1996;27:775–84.
[37] Aw TC, Lepe X, Johnson GH, Mancl LA. A three-year clinicalevaluation of two-bottle versus one-bottle dentin adhesives.J Am Dent Assoc 2005;136:311–22.
[38] Van Dijken JW. Durability of three simplified adhesivesystems in Class V non-carious cervical dentin lesions. Am JDent 2004;17:27–32.
[39] Abdalla AI, Garcia-Godoy F. Clinical evaluation of self-etchadhesives in Class V non-carious lesions. Am J Dent2006;19:289–92.
[40] Dalton Bittencourt D, Ezecelevski IG, Reis A, Van Dijken JW,Loguercio AD. An 18-months’ evaluation of self-etch andetch & rinse adhesive in non-carious cervical lesions. ActaOdontol Scand 2005;63:173–8.
[41] Türkün LS. Clinical performance of a new antibacterialadhesive system at 18-months. J Dent Res 2004;86(SpecIssue A):Abstract No. 0226 (www.dentalresearch.org).
[42] Ritter AV, Heymann HO, Swift Jr EJ, Sturdevant JR, Wilder JrAD. Clinical evaluation of an all-in-one adhesive innon-carious cervical lesions with different degrees of dentinsclerosis. Oper Dent 2008;33:370–8.
[43] Schattenberg A, Werling U, Willershausen B, Ernst CP.Two-year clinical performance of two one-step self-etchingadhesives in the restoration of cervical lesions. Clin OralInvestig 2008;12:225–32.
[44] Gallo JR, Burgess JO, Ripps AH, Walker RS, Ireland EJ,Mercante DE, et al. Three-year clinical evaluation of acompomer and a resin composite as Class V fillingmaterials. Oper Dent 2005;30:275–81.
[45] Van Meerbeek B, Kanumilli PV, De Munck J, Van Landuyt K,Lambrechts P, Peumans M. A randomized, controlled trialevaluating the three-year clinical effectiveness of two etch &rinse adhesives in cervical lesions. Oper Dent2004;29:376–85.
[46] Swift Jr EJ, Perdigão J, Wilder Jr AD, Heymann HO, SturdevantJR, Bayne SC. Clinical evaluation of two one-bottle dentinadhesives at three years. J Am Dent Assoc 2001;132:1117–23.
[47] Van Dijken JW. Clinical evaluation of three adhesive systemsin Class V non-carious lesions. Dent Mater 2000;16:285–91.
[48] Tyas MJ. Clinical performance of two dentine adhesives:2-year results. Aust Dent J 1996;41:324–7.
[49] Türkün SL. Clinical evaluation of a self-etching and aone-bottle adhesive system at two years. J Dent2003;31:527–34.
[50] Van Meerbeek B, Braem M, Lambrechts P, Vanherle G.Two-year clinical evaluation of two dentine-adhesivesystems in cervical lesions. J Dent 1993;21:195–202.
[51] Brunton PA, Cowan AJ, Wilson MA, Wilson NH. A three-yearevaluation of restorations placed with asmear-layer-mediated dentin bonding agent in non-cariouscervical lesions. J Adhes Dent 1999;1:333–41.
[52] Baratieri LN, Canabarro S, Lopes GC, Ritter AV. Effect of resinviscosity and enamel beveling on the clinical performance
of Class V composite restorations: three-year results. OperDent 2003;28:482–7.
[53] Hennig AC, Helbig EB, Haufe E, Richter G, Klimm HW. DieVersorgung von Klasse-V-Kavitäten mit dem
[54] Peumans M, Van Meerbeek B, Lambrechts P, Vanherle G.Two-year clinical effectiveness of a resin-modifiedglass-ionomer adhesive. Am J Dent 2003;16:363–8.
[55] Browning WD, Brackett WW, Gilpatrick RO. Two-year clinicalcomparison of a microfilled and a hybrid resin-basedcomposite in non-carious Class V lesions. Oper Dent2000;25:46–50.
[56] Merte K, Frohlich M, Hafer M, Hirsch E, Schneider H, WinklerM. Two-year clinical performance of two primer adhesiveson Class V restorations. J Biomed Mater Res 2000;53:93–9.
[57] Brackett WW, Brackett MG, Dib A, Franco G, Estudillo H.Eighteen-month clinical performance of a self-etchingprimer in unprepared Class V resin restorations. Oper Dent2005;30:424–9.
[58] Hörsted-Bindslev P, Knudsen J, Baelum V. 3-Year clinicalevaluation of modified Gluma adhesive systems in cervicalabrasion/erosion lesions. Am J Dent 1996;9:22–6.
[59] Blunck U. Klinische Prüfung von AdheSE und Excite zurVersorgung von Zahnhalsdefekten - Ergebnisse nach 18 und36 Monaten; Bericht an Ivoclar Vivadent 2006.
[61] Smith CD, Dickinson GL, Morris CF, Cliett BA. Two-yearclinical evaluation of non-retentive Class V restorations.J Dent Res 2000;75(Spec Issue A):Abstract No. 3031, 396.
[62] Sturdevant J, Bayne S, Heymann H, Wilder A, Roberson TM.3-Year clinical evaluation of LC resin-modified glassionomer restoration. J Dent Res 2000;75(Spec Issue A):Abstract No. 1223, 170.
[63] Brackett WW, Browning WD, Ross JA, Brackett MG. Two-yearclinical performance of a polyacid-modified resin compositeand a resin-modified glass-ionomer restorative material.Oper Dent 2001;26:12–6.
[64] Friedl KH, Hiller K-A, Jung H, Schlittenbauer T, Lichtblau J,Schmalz G. Clinical evaluation of composite restorationsusing different adhesive systems 2004;83(Spec Issue A):Abstract No. 535 (www.dentalresearch.org).
[65] Folwaczny M, Loher C, Mehl A, Kunzelmann KH, Hinkel R.Tooth-colored filling materials for the restoration of cervicallesions: a 24-month follow-up study. Oper Dent2000;25:251–8.
[66] Latta MA, Barkmeier WW, Triolo PT, Cavel WT. Three-yearclinical evaluation of the Clearfil Liner Bond 2 System. J DentRes 2000;75(Spec Issue A):Abstract No. 1030, 272.
[67] Latta MA, Barkmeier WW, Shaddy RS, Cavel WT, BlankenauRJ. Clinical evaluation of the Prime and Bond 2.1 adhesivesystem. J Dent Res 2000;75(Spec Issue A):Abstract No. 1027,272.
[68] Pollington S, van Noort R. A clinical evaluation of a resincomposite and a compomer in non-carious Class V lesions.A 3-year follow-up. Am J Dent 2008;21:49–52.
[69] Brackett WW, Dib A, Brackett MG, Reyes AA, Estrada BE.Two-year clinical performance of Class V resin-modifiedglass-lonomer and resin composite restorations. Oper Dent2003;28:477–81.
[70] Ermis RB. Two-year clinical evaluation of fourpolyacid-modified resin composites and a resin-modifiedglass-ionomer cement in Class V lesions. Quintessence Int2002;33:542–8.
[71] Schwartz RS, Haveman CW, Conn LJ, Summitt JB, RobbinsJW. Clinical evaluation of a one bottle adhesive: 18 month
results. J Dent Res 2000;75(Spec Issue A):Abstract No. 1534.
[72] Papathanasiou A, Kugel G, Kastali S. Clinical evaluation of anew adhesive-composite system. J Dent Res 2004;84(SpecIssue A):Abstract No. 0538 (www.dentalresearch.org).
[73] Abdalla AI, Garcia-Godoy F. Clinical performance of aself-etch adhesive in Class V restorations made with andwithout acid etching. J Dent 2007;35:558–63.
[74] Özgünaltay G, Önen A. Three-year clinical evaluation of aresin modified glass-ionomer cement and a composite resin
5 ( 2 0 0 9 ) 605–620
in non-carious Class V lesions. J Oral Rehabil2002;29:1037–41.
[75] Matis BA, Cochran MJ, Carlson TJ, Guba C, Eckert GJ. Athree-year clinical evaluation of two dentin bonding agents.J Am Dent Assoc 2004;135:451–7.
