Introduction.
Endodontic surgery may be indicated in cases where posts cannot be removed, following recent prosthodontic treatment, where apical radiolucencies persist and when root canal systems are inaccessible (Sultan & Pitt Ford 1995). Failure of endodontic surgery may be a result of a poor seal with the root-end filling, difficulties encountered during surgical procedures, untreated root canals and unsealed isthmus areas (Weller et al . 1995). To avoid leakage of pathogens and toxins from infected root canals, an apical seal is established by resecting approximately 3 mm of the root-end with no or a small bevel angle (Carr 1994, Gilheany et al . 1994), the preparation of a root-end cavity (Von Arx et al . 1998) and the placement of a root-end filling.
Dye penetration studies have been undertaken to assess leakage of Super-EBA (EBA) (Saunders et al . 1994, Gagliani et al . 1998) and Mineral Trioxide Aggregate (MTA) (Torabinejad et al . 1993, Aqrabawi 2000), but this methodology is also known to have drawbacks (Sultan & Pitt Ford 1995). Air entrapment, use of pressure and the unsuitable molecular size of dyes can contribute to variation of the results (Wu et al . 1994, Kazemi & Spångberg 1995). Furthermore, dye penetration studies are destructive and do not reveal the influence of mastication on leakage of a root-end filling material. Fluid transport models comparing microleakage of MTA and amalgam (Yatsushiro et al . 1998) or EBA, amalgam and MTA (Wu et al . 1998) showed less microleakage when root-end fillings consisted of MTA rather than in EBA and amalgam. However, the application of solutions under pressure does not represent clinical conditions and dye or isotope molecules lack correlation to bacterial leakage. Nevertheless, dye penetration is the most frequently used method to evaluate the sealing ability of root-end filling materials (Torabinejad & Pitt Ford 1996).
Bacterial leakage studies have been recommended to test root-end filling materials in order to overcome the disadvantages related to dyes and radioisotopes (Wong et al . 1994). Nevertheless, these methods also have shortcomings, and are not readily comparable to clinical conditions (Torabinejad & Pitt Ford 1996).
Another approach to assess the sealing ability of a filling material is to determine the quality of marginal adaptation under a scanning electron microscope. This method has been used to evaluate root-end filling materials (Stabholz et al . 1985). The marginal adaptations of amalgam EBA, IRM and MTA root-end fillings have been evaluated using an SEM, revealing that MTA had better adaptation (Torabinejad et al . 1995).
In order to simulate clinical conditions as closely as possible, restorations such as composite fillings have been subjected to chewing cycles in a computer-controlled masticator (Krejci et al . 1990b, Airoldi et al . 1992, Krejci et al . 1993). The aim of this scanning electron microscopic study was to evaluate the marginal adaptation of EBA and MTA root-end filling materials before and after occlusal loading in vitro for a five-year equivalent period in a computer-controlled masticator. The effect of occlusal loading on microcrack formation in the root-ends was also investigated.

Materials and methods.
Twelve maxillary and 12 mandibular human molars were prepared to an apical size 40 using conventional techniques and root-filled with laterally condensed gutta-percha and AH Plus (DeTrey-Dentsply, Konstanz, Germany) as the sealer. The specimens were randomly divided into two equal groups of 12 teeth each and stored in 0.1% thymol solution. The specimens were fixed into custom-made holders so that approximately 3 mm of the root apex could be resected with a handpiece (Kavo Dental AG, Brugg, Switzerland) and a diamond bur (Perioset, no. 575, Intensiv, Viganello, Switzerland) at 10000 r.p.m., with copious amounts of water coolant (Peters et al . 2001).
In maxillary molars, the mesio-buccal and palatal roots were used; in mandibular molars, root-end cavities were prepared in both the mesial and distal roots. Preparation of the root-end cavities was performed using two types of ultrasonic tips (CT-2, EIE Analytic, Glendora, CA, USA and EA 2068 A, EMS, Nyon, Switzerland) with continuous water coolant in an ultrasonic unit (EMS 400, EMS). In each specimen, one root-end cavity was prepared with a CT-2 tip, whilst the other cavity was prepared with an EA tip. The root-end cavities were extended 3 mm into the root canal. All root-end cavities were prepared under a stereomicroscope at 10 to 15 magnification.
After root-end cavity preparation, the cavities were examined under a stereomicroscope at 10 to 15 magnification. Debris was washed away using water and compressed air from a dental unit. The root-end cavities and resection surfaces were dried using compressed air. The teeth were kept moist at all times by keeping them wrapped in gauze saturated with tap water.

Root-end fillings.
Following the randomly assigned numbers of the specimens, one of the two root-end cavities in each tooth was filled with fast set EBA (Harry J. Bosworth, Skokie, IL, USA) and the other filled with Pro Root MTA (Dentsply Maillefer, Ballaigues, Switzerland) using hand instruments specially designed for microsurgical endodontics (Micro Retro Modul, Martin, Tuttlingen, Germany). An equal distribution of filling material and root type was obtained by ensuring that an equal number of mesial, distal, palatal and mesio-buccal roots were filled with EBA and MTA.
Twenty-four root-end cavities were filled with each material: 12 mesial and 12 distal roots in mandibular molars and 12 palatal and 12 mesio-buccal roots in maxillary molars. The teeth were kept wrapped in moistened gauze during the entire filling procedure, exposing only the surface of the resected root-end. After the rootend filling was completed, the specimens were returned into the vials containing 0.1% thymol solution. Setting times according to the manufacturers’ instructions were 45–90 s for Super EBA (fast set) and 4 h for MTA. To ensure complete setting of both filling materials, the extracted teeth remained immersed in 0.1% thymol solution for 24 h.
Setting of the filling materials was verified by exerting light pressure on the surface of the fillings with a periodontal probe. Subsequently, the EBA root-end fillings were smoothed using water-cooled diamond burs (Perioset, #540 and #515, Intensiv, Viganello, Switzerland), rotating at 10 000 r.p.m.
Impressions of the root-ends were taken before and after the chewing cycles (President, Coltène, Altstätten, Switzerland) and replicas were made using a transparent resin (Stycast, Emerson & Cuming, Oevel, Belgium). The replicas were gold-sputtered (500 A , Balzers CSD 030, Balzers, Liechtenstein, Liechtenstein) for examination in a scanning electron microscope (Amray 1810 T, Bedford, MA, USA). Photomicrographs were taken at 20 and 50 magnification. Regions of special interest were photographed at 100 magnification.

Crack formation.
Replicas of the root-end surfaces were examined under an SEM at 20 magnification. Occurrence and location of microcracks was noted before and after occlusal loading of the specimens. The type of root in which the microcrack had occurred was noted and the type of microcrack was classified. The microcracks were divided into four types: Incomplete canal cracks extending from the canal lumen into the surrounding dentine, complete canal cracks (through and through cracks), intradentine cracks that were confined to the dentine and cemental cracks originating in the cemental surface, running into the root dentine and stopping short of the canal lumen.

Marginal adaptation.
Replicas of the root-end fillings were examined at 200 magnification before occlusal loading and re-examined after the specimens were subjected to the chewing cycles. Marginal adaptation and continuity of the two filling materials were categorized as:

1. Continuous margin: No gap visible between root-end cavity wall and filling material
   a) Overfilled margin: Filling material covers interface between filling and dentine
   b) Underfilled margin: Surface of filling material below level of surface of resected root-end
2. Non-continuous margin: Gap apparent between rootend cavity wall and filling material
   

The data for margin quality were assessed by utilizing a frame grabber card and image analysis software (NIH 1.61, NIH Bethesda, MD, USA). Percentages of continuous and non-continuous margin were calculated according to the formula (Airoldi et al . 1992): Sum of lengths of visibly open margin [pixels] 100/total length of margin [pixels].

Preparation of the specimens for occlusal loading.
To simulate a periodontal ligament, the root surfaces of the 24 specimens were covered with an approximately 1 mm thick coating of silicon (Dow Corning 734, Dow Corning, Wiesbaden, Germany) using a small brush. The roots of the specimens were then embedded in resin (Paladur, Kulzer, Wehrheim, Germany). The same resin was used to mount specimens onto metal stubs (014001-T, Balzers). The stubs were fastened into a holding device in the testing chambers of the computer-controlled masticator. The antagonist in the chewing act was a palatal cusp of a maxillary molar that had previously been set into the testing chamber. Depending on the size of the access cavity, the contact areas between antagonists were either on enamel or on temporary fillings.

Computer-controlled masticator.
The in vitro occlusal loading (Krejci et al . 1993) was performed in a computer-controlled masticator (Department of Preventive Dentistry, Periodontology & Cariology, Zürich Dental School, University of Zurich). The amount of pressure applied was 49.4 0.7 N. The specimens were subjected to 10 chewing cycles per min. The force was directed axially onto the test specimens which had been secured on a rubber connector on a 20 plane. Occlusal loading was continued for 1.2 million cycles or a five-year equivalent. During loading, the specimens remained immersed in tap water at a temperature of 37 C.

Statistical analysis.
Means ( SD) were tested statistically using repeatedmeasures anova (Stat View II, Abacus, Berkley, CA, USA). Comparisons of microcrack formation numbers were completed using chi-squared tests. The level of significance was set at 0.05.

Marginal adaptation.
Both EBA and MTA yielded excellent results before occlusal loading (Figs 1a, 2a) and demonstrated 99.4 0.5% and 99.2 0.3% continuous margin, respectively. After the specimens had been subject to 1.2 million chewing cycles (Figs 1b, 2b), the amount of continuous margin of EBA and MTA root-end fillings decreased to 93.1 1.3% and 98.9 0.7%, respectively. For EBA, the decrease of continuous margin due to occlusal loading was highly significant ( P < 0.001, Fig. 3).

Figure 1. Example of an EBA root-end filling. Original magnification 50. (a) Before occlusal loading; (b) After occlusal loading.


Figure 2. Example of an MTA root-end filling. Original magnification 50. (a) Before occlusal loading; (b) After occlusal loading.

Additionally, cavity margins with or without apparent gap formation were found to be either overfilled, underfilled or flush filled. Before the specimens were subjected to simulated chewing cycles, 71.3 7.2% of the EBA and 7.6 2.4% of the MTA root-end fillings were overfilled, with surplus filling-material concealing the interface between root-end cavity margin and root-end filling. After loading, 39.5 5.5% ( P < 0.01) of the EBA and 9.6 3.4% ( P > 0.05) of the MTA root-end fillings were overfilled (Figs 4, 5).

Before the loading procedure, 0.1 0.1% of the EBA and 36.3 5.2% of the MTA root-end fillings were underfilled, so that the level of the root-end filling was below the level of the resected root-end surface. After occlusal loading, 5.2 1.4% ( P < 0.01) of the EBA and 52.6 5.9% ( P < 0.05) of the MTA root-end fillings were underfilled (Figs 4, 6).

Figure 3. Changes in percentage of continuous margin of EBA (white columns) and MTA (dark columns) root-end fillings before and after occlusal loading.The decrease of continuous margin in EBA fillings after occlusal loading was highly significant (P > 0.001, anova).

Microcrack formation.

Of the resected root-end surfaces, 12.8% (6/47) showed microcracking before occlusal loading and was 25.5% (12/47, Table 1) after loading.

Figure 4. Mean percentage of overfilled and underfilled margins of EBA (white columns) and MTA (dark columns) root-end fillings before and after occlusal loading.

Figure 5. Overfilled EBA root-end filling in a mesio-buccal root after loading. Marginal adaptation is partially non-continuous (arrow). Original magnification x50.

Figure 6. Underfilled MTA root-end filling in a mesio-buccal root after loading. Marginal adaptation is partially non-continuous (arrow). Original magnification x50.

Microcracks formed in 58.3% (7/12) of the mesiobuccal roots, 16.7% (2/12) of the palatal roots, 9.0% (1/11) of the mesial and 16.7% (2/12) of the distal roots (Table 1). There was a significant difference in the frequency of microcrack occurrence in mesio-buccal roots in comparison with other root types ( P < 0.05, Table 2).

Table 1. Pre-loading and post-loading occurrence of microcracks in different root types.

One-third of the detected microcracks were complete canal cracks (through and through cracks, Fig. 7), onethird of the cracks were incomplete canal cracks, 25% were intradentine cracks (Fig. 8) and one specimen displayed a cemental crack (Table 1). 

Table 2. Numbers and types of microcracks detected after occlusal loading.

Little has been published on how root-end filling materials are affected by loading during chewing action (Blum et al . 1997). A computer-controlled masticator (Krejci et al . 1990b) allows occlusal loading of tooth restorations in vitro in a manner that resembles clinical conditions and therefore delivers clinically relevant results (Krejci et al . 1999). Natural human enamel cusps are used as antagonists in a chewing simulator which is able to simulate physical forces, wear mechanisms and temperature changes that can occur in the oral environment (Krejci et al . 1990b).
For evaluation of the wear of restorative materials, it is important to gain not only short-term data but also long-term results concerning durability and marginal continuity. Thus, the time frame that is aimed at is an observation period of 5 years or a five-year equivalent (Krejci et al . 1990a). It has been repeatedly demonstrated that this method is valid and is not compromised by artefacts. For instance, the amount of margin obscured by bubbles or other artefacts is typically small and not included in the calculation.
The results obtained in the current study show that the root-end filling materials EBA and MTA both yielded excellent pre-loading results with less that 1% gap formation between the surrounding dentine wall of the root-end cavity and the root-end filling. The amount of continuous margin after loading for both root-end filling materials decreased slightly but still displayed highly satisfactory findings. After the specimens were stressed by 1.2 million chewing cycles or a five-year equivalent, EBA still showed 93.1 1.3% and MTA demonstrated 98.9 0.7% continuous margin. These findings confirm that under in vitro conditions both materials were reliable and stable over time. MTA requires a 3–4 h setting time (Torabinejad & Chivian 1999) in comparison to the shorter setting time of EBA (0.75–6 min). This, added to the fact that MTA contains 5% calcium sulphate dihydrate (gypsum, Dentsply-Tulsa 1999) that expands during setting may contribute to the superior marginal adaptation of MTA filling material after loading.


Figure 7. Palatal root with MTA root-end filling displaying through and through microcrack (arrows) after occlusal loading. Original magnification x100.


Figure 8. Mesio-buccal root revealing intradentine microcrack (arrows) after occlusal loading of a specimen with an EBA rootend filling. Original magnification x100.

The majority of EBA root-end fillings were overfilled initially, even though these fillings had been carefully finished using a fine diamond-coated bur under the operating microscope. The amount of surplus filling material decreased significantly during occlusal loading and was probably washed away in the test chamber by surrounding fluid.
In contrast, a substantial number of the MTA root-end fillings were underfilled before loading, the amount of underfilled margin increased significantly after occlusal loading. Other authors have discovered that the outer layer of MTA dissolves in a solution of phosphate-buffered saline, resulting in a loss of half a millimetre of the surface of root-end fillings (Yatsushiro et al . 1998). This finding corresponds with the results of the present study, with 36.3 5.2% of the MTA filling margins being underfilled (Fig. 6) before the loading procedure had taken place. After occlusal loading, the amount of underfilling had increased significantly, but there was no apparent influence on the marginal integrity of the MTA root-end fillings.
Under in vivo conditions, this surface disintegration of the root-end filling material might not take place to such an extent or might be self-limiting. Studies have shown an inductive effect of MTA on cementoblasts, resulting in growth of a complete layer of cementum-like material over the surface of MTA root-end fillings in monkeys as early as 5 months (Torabinejad et al . 1997). Investigations have also shown cell growth of osteoblasts on MTA (Mitchell et al . 1999). This creation of an intimate connection between the filling material MTA and the periodontium may prevent or limit possible loss of material.
The second topic of investigation in the current study was crack formation before and after occlusal loading of the specimens. Other authors have divided microcracks into incomplete or complete canal cracks, intradentine cracks and cemental cracks (Layton et al . 1996). In the current study, only one cemental crack was observed. The distribution of microcracks into complete canal cracks (through and through cracks, Fig. 7), incomplete canal cracks and intradentine cracks (Fig. 8) did not differ significantly.
Mesio-buccal roots were shown to be significantly more susceptible to crack formation than the other root types. A possible explanation is the cross-sectional shape of this root with a deeply fluted indent between the two main canals. As a result, the biconcave middle portion of a mesio-buccal root is potentially overprepared or weakened during root-end preparation and is predisposed to microcracking (Abedi et al . 1995, Frank et al . 1996). With respect to different types of cracks, only four cracks were complete. However, it is not known whether or not the other types of microcracks could eventually become complete cracks or root fractures.

References.

Abedi H, Van Mierlo B, Wilder-Smith P, Torabinejad M (1995) Effects of ultrasonic root end cavity preparation. Oral Surgery, Oral Medicine and Oral Pathology 80 , 207-13.
Airoldi RL, Krejci I, Lutz F (1992) In vitro evaluation of dentinal bonding agents in mixed Class V cavity preparations. Quintessence International 23, 355-62.
Aqrabawi J (2000) Sealing ability of amalgam, super EBA cement, and MTA when used as retrograde filling materials. British Dental Journal 11, 266-8.
Blum JY, Esber S, Parahy E, Franquin JC (1997) Effect of masticatory cycles on tooth compression and resultant leakage of amalgam retrofills. Journal of Endodontics 23 , 605-9.
Carr G (1994) Surgical endodontics. In: Cohen S, Burns R., eds. Pathways of the Pulp, 4th edn. St. Louis, MO, USA: CV Mosby Co., 546-52.
Dentsply-Tulsa (1999) ProRoot MTA Root Canal Repair Material.Product Information. Tulsa, OK, USA: Dentsply-Tulsa.
Frank RJ, Antrim DD, Bakland LK (1996) Effect of retrograde cavity preparations on root apexes. Endodontics and Dental Traumatology 12, 100-3.
Gagliani M, Taschieri S, Molinari R (1998) Ultrasonic root-end preparation: influence of cutting angle on the apical seal. Journal of Endodontics 25, 726-30.
Gilheany PA, Figdor D, Tyas MJ (1994) Apical dentin permeability and microleakage associated with root end resection and retrograde filling. Journal of Endodontics 20, 22-6.
Kazemi RB, Spångberg LS (1995) Effect of reduced air pressure on dye penetration in standardised voids. Oral Surgery, Oral Medicine, Oral Pathology, Oral Radiology and Endodontics 80, 720-5.
Krejci I, Albert P, Lutz F (1999) The influence of antagonist standardization on wear. Journal of Dental Research 78, 713-9.
Krejci I, Heinzmann JL, Lutz F (1990a) Verschleiss von Schmelz, Amalgam und ihrer Schmelzantagonisten im computergesteuerten Kausimulator. Schweizerische Monatsschrift der Zahnmedizin 100, 1285-91.
Krejci I, Kuster M, Lutz F (1993) Influence of dentinal fluid and stress on marginal adaptation of resin composites. Journal of Dental Research 72, 490-4.
Krejci I, Reich T, Lutz F, Albertoni M (1990b) An in vitro test procedure for evaluating dental restoration systems. 1. A computer-controlled mastication simulator. Schweizerische Monatsschrift für Zahnmedizin 100, 953-60.
Layton C, Marshall J, Morgan L, Baumgartner J (1996) Evaluation of cracks associated with ultrasonic root-end preparation. Journal of Endodontics
22 , 157-60.
Mitchell PJ, Pitt Ford TR, Torabinejad M, McDonald F (1999) Osteoblast biocompatibility of mineral trioxide aggregate. Biomaterials 20, 167-73.
Peters CI, Peters OA, Barbakow F (2001) An in vitro study comparing root-end cavities prepared by diamond-coated and stainless steel ultrasonic retrotips. International Endodontic Journal 34, 142-8.
Saunders WP, Saunders EM, Gutmann JL (1994) Ultrasonic root-end preparation Part 2. Microleakage of EBA root end fillings.International Endodontic Journal 27, 325-9.
Stabholz A, Friedman S, Abed J (1985) Marginal adaptation of retrograde fillings and its correlation with sealability. Journal of Endodontics 11, 218-23.
Sultan M, Pitt Ford TR (1995) Ultrasonic preparation and obturation of root-end cavities. International Endodontic Journal 28, 231-8.
Torabinejad M, Chivian N (1999) Clinical applications of Mineral Trioxide Aggregate. Journal of Endodontics 25, 197-205.
Torabinejad M, Pitt Ford TR (1996) Root end filling materials: a review. Endodontics and Dental Traumatolology 12, 161-78.
Torabinejad M, Pitt Ford TR, McKendry DJ, Abedi HR, Miller DA, Kariyawasam SP (1997) Histologic assessment of mineral trioxide aggregate as a root-end filling in monkeys. Journal of Endodontics 23, 225-8.
Torabinejad M, Smith PW, Kettering JD, Pitt Ford TR (1995) Comparative investigation of marginal adaptation of mineral trioxide aggregate and other commonly used root-end filling materials. Journal of Endodontics 21, 295-9.
Torabinejad M, Watson TF, Pitt Ford TR (1993) Sealing ability of a mineral trioxide aggregate when used as a root end filling material. Journal of Endodontics 19, 591-5.
Von Arx T, Kurt B, Ilgenstein B, Hardt N (1998) Preliminary results and analysis of a new set of sonic instruments for root-end cavity preparation. International Endodontic Journal 31, 32-8.
Weller R, Niemczyk S, Kim S (1995) Incidence and position of the canal isthmus. Journal of Endodontics 21, 380-3.
Wong WS, Rosenberg PA, Boylan RJ, Schulman A (1994) A comparison of the apical seals achieved using retrograde amalgam fillings and the Nd: YAG laser. Journal of Endodontics 20, 595-7.
Wu MK, De Gee AJ, Wesselink PR (1994) Fluid transport and dye penetration along root canal fillings. International Endodontic Journal 27, 233-8.
Wu MK, Kontakiotis EG, Wesselink PR (1998) Long-term seal provided by some root-end filling materials. Journal of Endodontics 24, 557-60.
Yatsushiro JD, Baumgartner JC, Tinkle JS (1998) Longitudinal study of the microleakage of two root-end filling materials using a fluid conductive system. Journal of Endodontics 24, 716-9.