During the past decade, composites have become the most commonly used restorative materials  and . They have gradually replaced amalgams, which have been the standard restorative material for more than 100 years, not only because of their desirable esthetics, ease of handling and minimally invasive preparation technique, but also because dental amalgams have been associated with environmental pollution  and  and even with alleged negative health effects due to release of mercury  and . In addition, thanks to their bonding potential to the tooth tissues, good mechanical properties and lower cost compared to other indirect restorations, the application of composites has expanded to a wide variety of clinical situations, some of which previously could only be treated with indirect prosthetic restorations. Nowadays, composites are not only used to restore decayed or traumatized teeth, but they are also routinely used both as direct and indirect restoratives to improve esthetic properties of discolored or malpositioned teeth. All in all, the use of these versatile materials is still on the rise.
However, a number of clinical studies have reported shorter longevity and higher failure rates for composite restorations compared to amalgams , , , , , ,  and . One of the main reasons for failure was secondary or recurrent caries , , , , , , ,  and . It is clear that these findings evoked scientific debate, as premature renewal or replacement of restorations is a heavy burden on health care expenditure. In addition, secondary caries (SC) always results in further tooth structure loss and may weaken the remaining tooth. Especially in case of extensive secondary-caries lesions and repetitive restorative interventions, this may eventually lead to premature loss of the tooth.
Various definitions of SC have been proposed in scientific literature and medical dictionaries. First, the terms ‘secondary’ and ‘recurrent’ caries are used interchangeably, with the difference that ‘secondary caries’ is used more commonly in European papers, while ‘recurrent caries’ is typically used in North America . As for the meaning of these terms, it seems that they can signify both ‘new caries around a restoration’ (irrespective of the fact whether the restoration was placed due to a primary caries) as well as ‘residual/remaining caries’ depending on the context . Nevertheless, Mjör and Toffenetti  made a distinction between these terms and defined SC as a ‘new lesion at the margin of existing restorations’. Even though the clinical differentiation between residual and newly developed caries is sometimes impossible, clear distinction is still preferable since residual caries underneath a restoration is nowadays not always considered anymore as failure due to a paradigm shift in caries treatment. For example, carious dentin can be left on purpose when applying conservative operative techniques, such as stepwise excavation, partial caries removal or no-dentinal-caries removal  and . In the remainder of this review we will reserve the term ‘secondary caries’ only for progressive and active caries lesions at the margins or under composite restorations that were placed following complete caries removal. Papers dealing with residual caries were excluded from the review.
A recent Delphi survey on the future of restorative dentistry for the next 20 years identified ‘prevention of secondary caries’ as of the highest importance . Yet, it is currently not known whether SC with composites is a material-based problem. It is often stated by dentists and scientists that SC is a much more frequent problem with composites than with amalgam or other restorative materials (Fig. 1) . The purpose of this narrative review was, therefore, to present the current knowledge about the incidence and prevalence, clinical diagnosis, histopathology and factors involved in the onset and development of SC around composite restorations.
Clinicians often claim that secondary caries is a much more frequent problem with composites than with amalgam or other restorative materials. In this RX-bite wing of a 59 years old patient, a recurrent caries was diagnosed under a composite restoration in the first lower molar. In contrast to the old amalgam restorations in all other teeth, this restoration was placed only a couple of years before. These typical clinical cases strengthen clinician's beliefs that composite restorations are more sensitive to secondary caries.
2. Search strategies
Using Pubmed and Medline, international literature was searched for all papers about different aspects of secondary caries such as clinical diagnosis, histopathology and all factors that were associated with the development of SC around composite restorations, which were discussed in separate paragraphs. Therefore, for each paragraph we performed a different search strategy with appropriate key words. Additional studies were found after checking the reference lists of the included papers. Since this review is narrative in nature, it was practically not possible to strictly define inclusion and exclusion criteria, and various types of papers were selected (in vitro, in situ, clinical studies, different types of review papers, as well as meta-analyses) based on a search with the combination of the following keywords: secondary caries, recurrent caries, dental composite, diagnosis, radiopacity, histopathology, outer and wall lesion, microleakage, interfacial gap, polymerization shrinkage, bacterial adhesion, surface roughness, surface free energy, biodegradation of composites, antibacterial composites etc. Only for the evaluation of the incidence of SC, a systematic search of literature was performed and strict inclusion and exclusion criteria were defined, which have been described in more detail below(§4, Table 1 and Fig. 2)
Inclusion and exclusion criteria used for the selection of clinical studies.
Flow chart describes all phases of the systematic review of the incidences of secondary caries with composite restorations reported in clinical studies. After the identification of the records in the database based on the specific keywords, and their selection according to predetermined inclusion/exclusion criteria, 241 clinical studies were finally included in the quantitative analysis of cumulative incidences (all studies are listed in Appendix A).
3. Diagnosis of secondary caries next to composites
Diagnosis of SC is always more difficult than that of primary caries due to the sheer presence of the restoration, but some researchers also suggested that there is a lack of standardized diagnostic criteria. They even hinted that a great number of composite restoration replacements due to this condition may in fact be an overtreatment  and .
Traditionally, oral examination, based on visual and tactile inspection combined with radiologic evaluation, is used for diagnosis; however, all these techniques have certain limitations in the detection of SC with composites. With visual and tactile examination it is sometimes very difficult to distinguish SC from certain other conditions such as marginal discoloration and so-called ‘amalgam tattoo’. Brown and black marginal staining, as signs of marginal debonding and consequent microleakage, are more easily detected around tooth-colored restorations, compared to amalgams, and are indeed very often interpreted as a first stage of SC  and . Discolored dentin shining up through enamel often results from uptake of corrosion products from a previous metal restoration and is not necessarily a sign of caries underneath the restoration. Furthermore, during tactile examination a sharp probe can stick even into caries-free marginal gaps or overhangs. Kidd and Beighton  indeed demonstrated the inefficiency of these clinical parameters to reliably predict the presence of the soft infected dentin under tooth-colored restorations.
The most commonly used auxiliary method for the detection of SC around restorations is intraoral radiography. Regarding sensitivity, specificity and accuracy, conventional and digital intraoral radiographs are comparable, but lower radiation doses give great advantage to digital over conventional radiography. For SC diagnosis, however, both techniques are limited by their two-dimensional nature and their diagnostic power depends on the angulation of the beam and superimposition of details in the radiograph. This could be overcome by using cone-beam computed tomography (CBCT)  and , but the higher radiation doses compared to intraoral techniques are not justifiable . Also, the discriminative ability of radiography to detect SC may be influenced by the radiopacity of the restorative material. The radiopacity of most of contemporary composites is slightly higher than that of enamel, which allows more accurate detection of SC next to composites than next to highly radiopaque amalgams ; On the other hand, there are unfortunately still commercial composites with radiopacity lower than enamel, which may hamper radiologic SC diagnosis . Finally, if there is still uncertainty regarding the diagnosis of SC after the use of traditional diagnostic methods, an exploratory preparation into the restorative material adjacent to the suspicious site has been recommended .
Promising alternative detection techniques based on fluorescence such as quantitative light-induced fluorescence (QLF) (Inspector Research Systems, Holland) and DIAGNOdent (Kavo, Biberach an der Riss, Germany) have also been tested to detect SC around composites  and . In vitro studies showed a similar sensitivity, specificity and accuracy of these techniques to bitewing radiographs in combination with visual examination  and , and the results of a recent in vivo study are also promising .
Apart from the detection of SC lesions, it has been suggested that the diagnosis should also include the assessment of the lesion activity. Differentiating between active and arrested lesions is important for treatment planning, since the latter could be treated following a more conservative approach such as repair or refurbishment of the restoration, or by only monitoring of the lesion  and .
4. Incidence of secondary caries with composites
Even though SC has been recognized as an important problem related to composite materials, its incidence and prevalence in the general population have not yet been fully investigated. Surprisingly, there are currently no studies available dealing exclusively with this issue. On the other hand, a lot of information regarding the incidence of SC is available in clinical studies on the performance and longevity of composite restorations and/or dental adhesives.
Using different online databases (Pubmed, Medline), international literature available until August 2014 was searched for articles in English published between 2000 and later that reported on the performance of composites or adhesives. Several papers were found by means of references in other papers. The used keywords were: “adhesive”; “composite”, “performance”, “clinical evaluation”, “longevity”, “survival”, “failure rate”, “posterior/anterior/cervical composite”, and the search filter was “clinical trial”. Inclusion and exclusion criteria are presented in Table 1. The search and the inclusion of articles was performed according to the PRISMA guidelines (Fig. 2)  and . Ultimately, 241 studies could be included (Appendix table). The cumulative incidence of SC, which is the occurrence of SC within the follow-up periods of the study, was calculated. Therefore, the obtained incidence values should be interpreted only with the corresponding follow-up periods, which varied from 1 up to 17 years. In studies reporting data collected at several evaluation moments, we calculated the incidence for each of them, and finally 252 cumulative incidences, could be included in the quantitative analysis. Most of the included studies were prospective, as retrospective studies often did not allow calculation of the caries incidence (uneven evaluation periods for all restorations).
In general, the caries incidence in included studies was very variable, ranging between 0% and 44%, with most studies even without occurrence of SC (175 out of 252 incidences). The overall incidences of SC are shown in Fig. 3, as retrieved from studies included in the Appendix. The high variation in caries incidences is most striking but not surprising considering the high variation in materials used, type, location and size of restoration, number and type of patients and operators, study setting (academic versus private practice) and follow-up period. As it was previously speculated in the literature , the study setting seemed to affect the incidence, as SC occurred more often in practice-based studies (median = 0.83% vs 0% in university-based studies) ( Fig. 3a). This difference is most probably due to the fact that majority of university studies included only low-caries risk patients. However, it may also point to the high technique-sensitivity of the composite placement technique, since in university-based studies operators were usually calibrated and trained for the operative procedure.
The mean incidences of secondary caries (SC) next to composites as retrieved from the studies included in Appendix Table. Due to the skewed distribution of the incidences, they are expressed as box plots with median (black line) with 25% and 75% quartile. The outliers are shown as dots and the whiskers represent the lowest value still within 1.5 interquartile range (IQR) of the lower quartile, and the highest value still within 1.5 IQR of the upper quartile. The included studies were subdivided in short-term (evaluation period <3 years), medium-term (evaluation period between 3 and 5 years) and long-term (evaluation period >5 years) studies . (a) There was a marked difference in incidence depending on the study setting. More secondary caries was observed in practice-based studies. A plausible explanation for this observation may be the technique-sensitive adhesive procedure, but also differences in caries risk profile between university and practice recruited patients. (b) The type or restoration was also very determining. Whereas class V restorations seldom exhibited secondary caries (c), the median incidence after long-term periods was 4% for anterior restorations (d) and 1.7% for posterior restorations (e). (f) Class II restorations were more prone to secondary caries than class I restorations.
As expected, also the location of the restoration played a major role in the occurrence of SC (Fig. 3b). Cervical composite restorations (class V) were the least affected by SC, which obviously may be related to the fact that many class V studies were set up to evaluate the clinical effectiveness of adhesives in non-carious cervical lesions in patients with low caries risk and good oral hygiene. The highest overall incidences were found in the posterior region. Also anterior restorations (classes III and IV) seem to be vulnerable to SC, but these results must be interpreted with care as only few studies included anterior restorations (n = 12). Naturally, the length of the observation period is also very important while interpreting these data. Therefore, we subdivided the included studies in short-term (evaluation period <3 years), medium-term (evaluation period between 3 and 5 years) and long-term (evaluation period >5 years) studies (Fig. 3c–e). Only for class V restorations, the incidence of SC did not increase over time (Fig. 3c), which again may be attributed to the fact that the majority of studies involved non-carious lesions. Nevertheless, there were some studies that did report the presence of some SC next to cervical restorations. This finding might point to a restoration-based cause (but not per se material-based) for SC. In contrast, SC incidence significantly increased over time in the other types of restorations, with the highest median incidences recorded after 5 years, more specifically 4% for anterior and 1.7% for posterior restorations ( Fig. 3d and e). When the incidence of the posterior restorations was further evaluated in detail by subdividing them in class I and II restorations (Fig. 3f), it was clear that class II are more prone to SC than class I restorations. In his practice-based survey, Mjör reported the highest percentage of SC lesions located gingivally, irrespective of the restorative material used . Unfortunately, in these studies the exact location of the recurrent caries with regard to the restoration (i.e. gingival, approximal or incisal/occlusal) was often not described in detail and could thus not be retrieved. Notably, even though caries incidence was generally very low in the first years of the follow-up period, there were studies indicating that SC may already occur after 1 or 2 years ,  and .
Even more interesting is comparing the incidence of SC with composites to that with other types of restorations. In Table 2, prospective clinical studies that compared the performance and longevity of composites with amalgam are listed. Most of these studies were practice-based. The incidence of SC around amalgams varied between 0% and 4.9%. In contrast, composites restorations tended to exhibit markedly more SC, with incidences varying between 0% and 12.7%.
Incidence of secondary caries around composites versus amalgams in prospective clinical studies.
Finally, the incidence of SC naturally depended on the caries risk of the patients included in the study. In the study by van Dijken et al. , the calculated incidence of SC was 6.5%, but 63% of the SC lesions were found in patients with high caries risk. Similarly, in the study by Opdam et al. , the incidence of SC in the low caries risk group was much lower than in the overall population (1.8% versus 6.6%, respectively). Interestingly, this study found a comparable incidence of SC between amalgams and composites in low and medium caries-risk patients, while in high caries-risk patients SC incidence was significantly higher around composites, suggesting that the latter perform worse under cariogenic challenge .
There is definitely need for more epidemiologic research with regard to SC, as there is currently a lot of information missing on the exact location of the caries, the size at diagnosis, the type of cavity (fresh cavity, or one that had been restored for several times).
5. Histopathology of secondary caries around composites
Histologically, SC has been found to be no different from primary caries, which has been described as a localized process of both demineralization of enamel and dentin and enzymatic and bacterial degradation of dentin  and .
Theoretically, secondary-caries lesions may consist of two parts: (i) the outer lesion, developing from the tooth outer surface next to the restoration margin, and (ii) the cavity wall lesion, developing along the tooth-restoration interface  (Fig. 4). In extensive lesions, an outer lesion usually also involves the cavity wall, so that it is difficult to distinguish between the two. It was described that the shape of an outer lesion and its relation to the cavity-wall lesion depend on the direction of the enamel prisms at the cavity margin (Fig. 4) , , , ,  and . In contrast, a cavity-wall lesion, usually present as a narrow zone of enamel or dentin demineralization at the cavity wall, seems to progress independently of the angle between prisms and the cavity wall ,  and . Only wall lesions reaching the dentin–enamel junction tends to spread laterally, thus affecting dentin on a wider front . In cavity-wall lesions, demineralization develops perpendicularly to the cavity wall, which implies that hydrogen ions first have to diffuse into the microspace between tooth and restoration. Therefore, it has been concluded that the wall lesion develops exclusively as a consequence of microleakage, even in the absence of an outer lesion , , , ,  and , whereas outer lesions may be regarded as primary caries lesions that happen to occur near an existing restoration . In other words, the histopathological appearance of SC lesions may thus give some insights in the etiology of SC.
Histopathology of secondary caries lesions next to composite. The location and shape of the lesion may give an indication of the underlying cause. (a) Secondary caries may present as an outer lesion, a cavity-wall lesion or as a lesion consisting of both an outer and cavity-wall lesion. (b) According to the theory by Hals et al. (1971) the shape of an outer lesion and its relation to a wall lesion depend on the direction of the enamel prisms at the cavity wall: in case of unsupported prisms at the cavity wall, an outer lesion develops from the surface toward the cavity, forming a funnel-shaped connection with the wall lesion, while in case of the supported prisms, the outer lesion is triangular and develops away from the cavity. According to this theory, the direction of the prisms adjacent to the restoration, will determine the shape of the secondary caries . (c) A cavity-wall lesion is thought to be the result of microleakage. Some researchers doubt whether secondary caries next to composite may present as wall lesions in clinic, but there is some evidence from in vitro research that the presence of a gap may result in a wall lesion. (d) Most often in clinic, secondary caries presents as an outer lesion which involves the interface with the composite restoration. In such case, it is unclear whether microleakage played a role in the development of the caries.
In the past, several researchers tried to investigate the histopathology of SC around composites restorations by inducing artificial caries lesions in vitro and in vivo. They concluded that, compared to lesions around amalgam, those around composite exhibited significantly fewer wall lesions ,  and , especially after etching of the cavity walls ,  and . This led to the conclusion that composites have much better initial sealing capacity than amalgams, thereby preventing microleakage and the formation of a cavity-wall lesion . This conclusion is remarkable as some of these studies are almost 40 years old, and the composites in these studies were used without adhesive. Unfortunately, there are only few recent clinical studies on this topic so that it is unclear whether outer or cavity-wall lesions are more frequent with contemporary composites in clinical situations. Thomas et al.  found in an in situ study that SC next to composite appeared merely as an outer lesion that seemed to progress as primary caries. Some other researchers even suggested that wall lesions next to composites do not really occur as a separate entity, but that they are just an aspect of an outer lesion developing toward and reaching the cavity wall, or a consequence of extended etching of the cavity wall surface with enamel prisms perpendicular to the cavity wall , ,  and . Nevertheless, there is some evidence from in vitro investigations suggesting that wall lesions can exist as own entity, but only in the presence of gaps at the tooth-restoration interface ,  and . This would only be the case in composites that could not properly seal the cavity walls. However, achieving a good seal over a long time under clinical conditions, as will be shown later, seems to be quite challenging.