The Accuracy of Guided Surgery Via Mucosa-Supported Stereolithographic Surgical Templates

The accuracy of guided surgery via mucosa-supported stereolithographic surgical templates in the hands of surgeons with little experience

In the last few years, the use of stereolithographic surgical templates to guide implant placement has been heavily promoted. These guides are produced by computer-aided design/computer-assisted manufacture (CAD/CAM) technology and intend to transfer the ideal implant position from the computerized planning to the surgical setting (Lal et al. 2006). This planning can ideally be performed using 3D imaging (MSCT/CBCT; BouSerhal et al. 2002). By loading these 3D images in a software program, together with images of the planned prosthetic device, it becomes possible to virtually plan the optimal implant position, taking both anatomical and restorative information into account. As such, a prosthetically driven implant placement becomes feasible, which benefits the cooperation between surgeon and prosthodontist (Lal et al. 2006; Vercruyssen et al. 2008; D'haese et al. 2012a,b; Hultin et al. 2012). Theoretically, this concept may offer a lot of advantages compared to free-handed implant placement (Hultin et al. 2012). A pre-surgical planning minimizes the need for decision at time of implant surgery, which may shorten the duration of the procedure. Moreover, it facilitates prosthetic rehabilitation and might eliminate esthetic and biomechanical risks involved in standard implant surgery (Vercruyssen et al. 2008; Dandekeri et al. 2013; Cassetta et al. 2014). It offers also the opportunity to apply a flapless approach, which on its own may give extra advantages. The main advantage of flapless implant placement is reduction of postoperative discomfort (pain, hematoma, swelling, etc.; Fortin et al. 2006; Hultin et al. 2012). Furthermore, it offers a shortened surgical time, minimizes changes in crestal bone levels, reduces the risk of inflammation, and limits bleeding (Becker et al. 2005). These features could especially be advantages in medically compromised (anticoagulantia, bisphosphonates, etc.) or anxious patients (D'haese et al. 2012a,b).

However, unrealistic clinical expectations concerning the efficacy and ease of computer-aided implantology may arise (Hämmerle et al. 2009). In a recent systematic review (Hultin et al. 2012) examining the advantages of computer-guided implant placement, the authors identified a number of complications that can occur. Some of these complications are specific for the applied technique. They are often related to the limited access and visibility, due to the flapless approach or due to a misfit of the guide.

The literature is not consistent on whether a learning curve is present for guided implant placement. Vasak et al. (2011) reported a clear learning effect over time regarding the accuracy of implants placed with NobelGuide™ templates. They concluded that guided implant surgery only offers a simple aid, without any gain in procedural reliability and security, during the first implantations performed by an inexperienced surgeon. This conclusion was supported by Cassetta et al. (2014) who concluded that computer-aided implant placement is a useful complement to conventional surgery, but it only offers a guide and does not replace surgical experience, skill, and knowledge of anatomy. Three other studies (Valente et al. 2009; Cassetta et al. 2013; Vercruyssen et al. 2014) did not observe a clear learning curve. However, all surgeons who did the implant surgery in those studies were experienced in conventional implant placement and only inexperienced in the use of stereolithographic guides.

The aim of the current study was to investigate the accuracy of implant placement with mucosa-supported stereolithographic guides, executed by inexperienced surgeons (inexperienced in both implant placement and guided surgery) and to compare these data with data of an experienced implant surgeon (experienced in guided surgery), who used the same treatment and accuracy analysis protocol (Vercruyssen et al. 2014) and with the data reported by systematic reviews.

Material and methods

Patient

Sixteen patients (mean age: 59.5 years, 35–78; 10 males, 6 females, and 3 smokers), who were in search of oral implant therapy of at least four implants in a fully edentulous maxilla or mandible, were recruited. To be included, the patients had to fulfill the criteria given in Table 1 and had to sign an informed consent form. All patients were treated between November 2011 and January 2014 by students who followed a 3-year postgraduate training in periodontology at the Catholic University Leuven, in Belgium. The study was approved by the ethical committee of the University Hospital of the Catholic University of Leuven (s55275).

Procedure

All postgraduate students had a limited surgical experience in implant therapy (30–80 implants placed). The patients in this study were treated by nine different postgraduate students (see distribution in Table 2). During planning and surgery, the same protocol was used as described in (Vercruyssen et al. 2014), with the exception of the pre-surgical scan. In this study, a pre-surgical CBCT (at 85 kV and 6 mA, voxel size 250 μm, Scanora® 3D; Soredex, Tuusula, Finland) was taken instead of the MSCT scan (at 120 kV and 90 mA, 0.6 mm slice thickness, voxel size 330 μm, Somatom Definition Flash®; Siemens, Erlangen Germany). All the postgraduate students had to follow “strictly” a series of steps:

Preparation of scan prosthesis

The scan prosthesis had to be well fitting, stable and had to fulfill the esthetic demands of the patient. If the existing denture fulfilled these criteria and contained all information necessary for the future prosthetic restoration, this denture was transformed into a scan prosthesis by inserting at least eight small glass markers. These glass markers served as radio-opaque markers during the scanning procedure. If the existing denture did not fulfill the criteria, a relining procedure was performed to adapt the prosthesis in the best way to the soft tissues, or a whole new prosthesis was made at the prosthetic department of the University Hospital of Leuven. To secure an optimal fit of the scan prosthesis during the scanning process, a bite index was prepared in putty material (SheraEx-act® 85; Shera GmbH & Co., Lemförde, Germany).

Scanning procedure

A first CBCT scan (at 85 kV and 6 mA, voxel size 250 μm, Scanora® 3D; Soredex) of the patient with the scan prosthesis and the bite index in place was taken. With this scan, the anatomical structures and glass markers were visualized. After this, the scan prosthesis alone was scanned under adapted settings, in a way that not only the glass markers, but also the entire denture became visible.

Pre-surgical planning

The DICOM images of both CBCT scans (patient with prosthesis and scan prosthesis alone) were imported in a software program (Simplant®; DENTSPLY Implants, Mölndal, Sweden) to perform the pre-surgical planning. Via the radio-opaque glass markers, both sets of DICOM images could be matched. The implants were virtually placed in the most optimal position according to bone anatomy and prosthetic design. This pre-surgical planning was checked by both an experienced periodontologist and an experienced prosthodontist. If needed, the planning was adapted. Once the planning was approved, it was sent to the manufacturer for the fabrication of a mucosa-supported stereolithographic drill guide (SIMPLANT SAFE Guide by DENTSPLY Implants).

Surgery

Surgery was performed following the implant manufacturers instructions (Facilitate™; DENTSPLY Implants) at the department of Periodontology of the University Hospital of Leuven, by an inexperienced postgraduate student, who also made the planning. After local anesthesia, the mucosa-supported stereolithographic guide was positioned using a bite index to secure proper seating. The guide was fixed to the jaw by three anchor pins, equally distributed in the jaw. All implants were placed flapless. There was a physical stop during drilling. Seventy-six AstraTechOsseoSpeedTx™ implants (DENTSPLY Implants) were inserted (distribution see Table 2). All surgeries were supervised by an experienced surgeon.

Accuracy analysis

A postoperative CBCT (at 85 kV and 6 mA, voxel size 250 μm, Scanora® 3D; Soredex) was taken to check the final position of the implants. These images were superimposed on the virtual planning via the Mimics® software (Materialise, Leuven, Belgium). An accuracy measurement protocol (DENTSPLY Implants) was used to determine the deviations between planned and placed implants. This process is based on surface registration, which consists of a minimization of distances between the pre-op and post-op jaw model. In this way, an iterative closest point algorithm could be used to match both jaws. Using the same software, coordinates of the planned and placed implants could be calculated, and several inaccuracy parameters could be defined (Fig. 1a). The global deviation is defined as the 3D distance between the coronal/ apical centers of the planned and placed implants. The angular deviation is calculated as the 3D angle between the longitudinal axes of both. Depth deviation is the distance between coronal/apical center of the longitudinal axis of the planned implant and a plane parallel through the coronal/apical center of the placed implant. Moreover, a reference plane was set in bucco-lingual direction by which both the mesio-distal and bucco-lingual deviation could be calculated (Fig. 1b). Also, the inter-implant deviation was calculated, as the difference between the planned and real inter-implant distance.

Statistical analysis

The outcome variables were the global deviation at the entry and apical point of the implant, the angular deviation, the depth deviation at the entry and apical point, and the mesio-distal and bucco-lingual deviations at the coronal point. A linear mixed model with side (left/right) as fixed factor and patient as random factor was fit for the comparison between sides (left/right). The latter was corrected for simultaneous hypothesis testing according to Sidak (1967). An alternative comparison was made between the data of the inexperienced group and data previously presented by Vercruyssen et al. (2014) and Vercruyssen et al. (submitted, depth and lateral deviations). She can be considered an experienced surgeon in both guided and non-guided implant placement and used a similar protocol, both for the surgery as well as for the inaccuracy evaluation. Again a linear mixed model was applied, with the experience as the only fixed factor.

Results

In the inexperienced group, 17 edentulous jaws in 16 patients were included. No patients were lost before the second scan was taken. One implant (upper jaw) in one patient showed an early failure, without obvious reasons. This implant was excluded for accuracy analysis. All other implants could be loaded after 3 months healing time.

Seventy-five implants were analyzed for accuracy. The deviations between the planned and placed implants are summarized in Table 3. The mean global deviation coronally was 0.87 mm (SD 0.50) and at the apex 1.10 mm (SD 0.53). In depth, the deviation was around 0.5 mm (0.0–2.4 mm). The deviation in angulation was 2.8° ranging from 0.2 to 7.0°. At the implant shoulder, the bucco-lingual inaccuracy was 0.39 mm (ranging from 0.0 to 1.5 mm) and in the mesio-distal direction 0.42 mm (ranging from 0.0 to 1.7 mm). These observations were also compared with the data for the experienced clinician (with 12 jaws and 52 implants). Only for deviation in angulation, the inexperienced group scored inferior to the experienced clinician (Table 4).

These data are further visualized in detail in a boxplot in Figs 2 and 3 showing the accuracy data from the inexperienced group together with the data of a recent systematic review about accuracy of guided surgery (Van Assche et al. 2012) and with the data of the experienced group. The deviations in depth, mesio-distal and bucco-lingual direction of the inexperienced group, together with the median deviation of the experienced surgeon are shown in Fig. 4. In Fig. 5 a boxplot is showing the inter-implant distance.

The implants placed by right-handed surgeons (inexperienced group) were placed more accurate at the right side of the patient (quadrant 1–4) than at the left side (quadrant 2–3). The opposite finding could be made for the left-handed surgeon, whose implants placed at the left side of the patient were more accurate. Statistically significant differences between left and right could be found for following parameters: coronal depth (P = 0.02), apical depth (P = 0.01), and mesio-distal (P = 0.002). The coronal inter-implant deviation (0.32 mm, SD 0.52) was clearly smaller than the global coronal deviation, pointing to the fact that part of the inaccuracy was primarily due to incorrect positioning of the guide.

Discussion

To our knowledge, this is the first study to assess the accuracy of dental implants placed in vivo by inexperienced surgeons with a stereolithographic guide. The impact of experience on the chance of osseointegration for oral implants has, however, been explored extensively (PreiskelTsolka 1995; Lambert et al. 1997; Kohavi et al. 2004; Melo et al. 2006; Zoghbi et al. 2011). Some of these studies comparing implants placed by inexperienced and experienced surgeons and concluded that the rate of osseointegration was significantly higher for experienced surgeons (PreiskelTsolka 1995; Lambert et al. 1997; Zoghbi et al. 2011). A possible explanation for the poor outcome of implants placed by inexperienced surgeons is that with less experience the frequency of problems such as excessive heat during drilling, non-stabilization of the implant, lack of adequate planning may increase. This finding could not be confirmed by others (Kohavi et al. 2004; Melo et al. 2006). In the study of Melo et al. (2006), the inexperienced surgeons were supervised by experienced surgeons, which can explain the better outcomes. In the present study, all the different steps necessary in planning and implant placement were strictly supervised by an experienced surgeon. This may be a reason for the comparable outcomes in accuracy between both groups of experience.

In a recent in vitro study, the impact of operator experience on the accuracy of implant placement with stereolithographic bone-supported guides was examined (CushenTurkyilmaz 2013). Implant placement was performed with the Facilitate™ system (DENTSPLY Implants). Two operators had some experience (over 100 implants placed), and 2 had little experience (less than 10 implants placed). Statistical analysis showed significant difference between the experienced and inexperienced group for angular and horizontal error at implant apex and at entry point, with more accuracy for the experienced clinicians. The same conclusion could be drawn out of an in vitro study on the accuracy of micro-implant placement performed by experienced and inexperienced dentists (Cho et al. 2010).

In the current study, an initial power analysis was not performed as we included all the patients that visited us in the time span of the study, and this may be a shortcoming of the study.

As can be seen in the boxplots (Figs 2 and 3), the accuracy of the implants placed by inexperienced surgeons is comparable with the experienced surgeon (Vercruyssen) and with the results of a recent systematic review (Van Assche et al. 2012). Of this systematic review, only the data of in vivo studies, using a mucosa-supported drill guide, were taken into account. The data of that systematic review are based on seven studies that included 462 implants.

The 'Material and methods' of the present study and the study of Vercruyssen et al. (2014) were exactly the same, except the pre-operative radiography. In this study, we decided to take a CBCT instead of the CT image, because of the lower radiation dose. Vercruyssen and co-workers already mentioned in the discussion that this is a good option, because the accuracy between both imaging techniques is similar (Arisan et al. 2013; Poeschl et al. 2013).

In the present study, one of the inexperienced surgeons was left-handed. For the statistical analysis of the accuracy comparison of the implants placed at the patients left (quadrant 2–3) and right (quadrant 1–4) side, we decided to mirror the data for the left-handed person. Implants in quadrant 1–4 placed by this surgeon were considered to be placed at the left side (quadrant 2–3) and vice versa. The implants placed at the right side of the patient by right-handed surgeons are significantly more accurate than the accuracy at the left side. A possible explanation can be the tolerance within the guide. The tolerance between the drill – drill guide and drill guide – sleeve can be responsible for deviation in several directions (Van Assche & Quirynen 2010; Koop et al. 2013). The authors advise the clinician to move the drill several times in and out to feel the smoothest drill position. So, it is obvious that the protocol of computer-aided implantology involves a sequence of many steps including production of a radiographic template, scanning procedure, planning, and surgery, and mistakes can arise at all different stages. The final (in)accuracy of the implant placement is a summation of all steps. Therefore, it is crucial to understand the importance of each step and to realize the magnitude of the cumulated inaccuracy.

In this study, the mean inter-implant distance between the planned and placed implants is 0.32 mm, and thus much lower than the global coronal inaccuracy. Based on these findings, one can conclude that part of the observed inaccuracy is caused by the malpositioning of the guide during the surgery (D'haese et al. 2012a,b). The chance for malpositioning is more likely to increase when the mucosa is thicker (Vasak et al. 2011; D'haese & De Bruyn 2013; Ochi et al. 2013). In the present study, we tried to minimize this effect using a bite index and fixation pins (Horwitz et al. 2009; Van Assche et al. 2012; Cassetta et al. 2013).