The application of computer-aided design and computer-aided manufacturing (CAD/CAM) technology in primary and secondary mandibular reconstruction with the free fibula flap (FFF) following ablative surgery is considered to be state of the art nowadays. Several studies have proven its benefits and superiority in terms of operating time, ischemic time, symmetry, bony consolidation, and function (1–4), which may result in a positive cost–benefit balance sheet (5, 6) as well as better functional and aesthetic results (7–9).

After increasing standardization of the surgical processes and the integration of virtual planning processes in the last decade, the trend continues toward cost reduction, as patient-specific cutting guides and osteosynthesis plates are usually offered and produced by various osteosynthesis manufacturers and may even not have led to overall cost reduction (10). On the one hand, this necessitates a functioning infrastructure with nationwide coverage by these companies, and on the other hand, the effort expended entails additional costs for the surgical department and ultimately the healthcare system (10). In addition, the dependence on the industry reduces flexibility of planning timing and, depending on the complexity of the case, requires a lead time of at least seven to ten working days, during which one to three web meetings are held to discuss the planning and its implementation. In this context, two developments can be observed in the daily routine and more recent literature: first, the establishment of low-cost solutions for the in-house production of cutting guides using open-source software and in-house printers (11–14) and, second, the use of partially adjustable resection aids such as the ReconGuide (KLS Martin Group; Gebrüder Martin GmbH & Co. KG; Tuttlingen, Germany) and the MUC-Jig (15, 16).

The purpose of this study was to evaluate our workflow and results of in-house-planned mandibular reconstructions with the FFF and to describe potential pitfalls and solutions for a wider application of this versatile opportunity.

This study is one of the few studies on in-house CAD/CAM solutions that presents a critical contemporary three-dimensional evaluation of virtually planned mandibular reconstructions with FFF. Tarsitano et al. described that the use of CAD/CAM-assisted mandibular reconstruction is economically viable, as the money saved by the reduction of operation duration offsets the associated costs of approximately EUR 3,450 (5). Rommel et al. calculated the increased costs for planning and the use of patient-specific cutting guides offered and manufactured by osteosynthesis manufacturers to be EUR 2,250 per case (3). According to reviews and meta-analyses others reported the cost of virtual surgical planning to range from USD 3,000–8,200 (6, 10). These higher costs can be explained by the fact that the studies included in the reviews/meta-analyses were published between 2013 and 2016, at a time when 3D printing, for example, was even more expensive than it is today, and usually included the production of patient-specific reconstruction plates. Costs resulting from cost-effective in-house design and printing solutions are reported to be significantly lower. Bosc et al. calculated their cost per case for in-house design and printing at EUR 989 (13). In contrast, our reported median costs per case were EUR 14.30 (4.50–25.30). This represents an excellent cost–benefit ratio, We have deliberately omitted recent non-negligible cost items (acquisition and maintenance costs, sterilization time, and planning time of approximately five to six hours, salary, etc.) and did also not calculated overall potential cost savings for simplicity, as have many other authors, and focused exclusively on material costs. More recently, Moe et al. reported their cost per case to be USD 3.87 for in-house design and printing. Dell’Aversana Orabona et al. reported a cost per case of EUR 3 (27). At EUR 14.30, the calculated costs per case are comparable with data from the literature and are completely negligible in relation to the surgical costs. Our slightly higher costs can be explained by two factors: we created and printed repositioning aids for each osteotomy site, and the neomandibula segment was also printed in order to bend the miniplates preoperatively.

The total lead time (planning, designing, printing, post-processing, final preparation with drill sleeves, and sterilization) is two to three days, which agrees very well with the times reported in the literature (13, 14) – and can be expedited further with growing experience and optimization of the processes. Another essential aspect raised by Numajiri et al. is that this form of cost-effective planning is done in the surgeon’s time and is not outsourced to the osteosynthesis manufacturers and their clinical engineers as is usually the case (28). Geusens et al. reported that they have employed a full-time clinical engineer, which could be another solution to the altruistic behavior of interested and motivated surgeons (21). According to Tang et al., the economic benefits and limitations associated with the application of virtual surgical planning must be weighed against patient outcomes (4). The introduction of in-house solutions may represent the ideal approach in this respect, as costs can be saved and treatment optimized.

The study design and methodical analyses of the mandibular reconstruction is generally heterogenous among the available studies (4), which makes a direct comparison difficult. In a recent systematic review and meta-analysis on virtual surgical planning in mandibular reconstruction Barr et al. were not able to define a valuable parameter and result for “accuracy”, even though many authors highlight the introduction of CAD/CAM algorithms as beneficial (6). The reason is again the heterogeneity of parameters. The majority of studies compared the preoperative virtual plan to the postoperative situation by measuring the fibula segment lengths (21, 29–31), point-to-point distances and angles (3, 13, 14), intercondylar distance (32), or intercoronoid distance (21) or by comparing interfragmentary gap distances (33). Numajiri et al. analyzed an algorithm for the production of low-cost cutting guides designed and printed in-house (28). They reported an error deviation of 2.4 mm in 12 studied in-vitro model surgeries. For this laboratory setting under ideal conditions, an error deviation of 2.4 mm seems to be quite high. Four years later, Numajiri et al. published a clinical study comparing freehand and in-house-planned CAD/CAM FFF cases (14). Based on a point-to-point analysis, they describe their in-house Blender-based solution as accurate. Interestingly, the results of the more recent study are more precise than the results of their in vitro studies, which can probably be attributed to the research group’s growing experience with CAD/CAM-based surgical planning and the improvement of the processes. Yu et al. aligned the virtual plan and the postoperative result. But the evaluation was again limited only to distance measurements between corresponding points (8). Bosc et al. analyzed their postoperative results of in-house (Meshmixer or Blender) planned and printed cutting guides after surface alignment using CloudCompare, another potent open-source software solution for this purpose (13). However, the authors only used surface matching to get the maximum points of convergence for better correspondence of analyzed distances and angles. In all studies mentioned, the deviations between planning and operative results were 1–3 mm.

The disadvantage of all “conventional” measurement methods, which essentially involve only the analysis of distances (point-to-point or bone segment lengths) and angles, is that the effect of a small deviation on the entire three-dimensional geometry is neglected. An involuntary deviation from the planned osteotomy angle, fibula segment length, or fibula segment position will result in the derotation of the remaining mandible, and consequently in changes to the condyle angle.

For this reason and unlike most studies that have performed analyses of CAD/CAM-assisted mandibular reconstructions with the FFF, our focus has been on a three-dimensional analysis that encompasses the entire mandibular geometry. For example, Wallner et al. and van Eijnatten et al. applied three-dimensional parameters to compare the accuracy and comparability of open-source software solutions and influence of threshold setting for DICOM data set segmentation (23, 24). This approach seems to be a contemporary and objective way to compare two similar and rather complex objects but has not yet been used routinely in analyses of mandibular reconstructions (11, 12, 27). Each of our examined three-dimensional parameters – RMSE, MSD, and HD – had both excellent intra- and interobserver reliability, making the measured results valid for further analyses and discussion. In contrast, the conventional measurement of condylar angles in particular showed poorer intraobserver reliability for both examiners, which was confirmed by comparable interobserver reliability ( Supplementary Tables 1 , 2 ). Horizontal distances also revealed very good intra- and interobserver reliabilities. Dell’Aversana Orabona et al. used the open-source software InVesalius for in-house planning in four consecutive cases (27). They reported a distance between three-dimensional pre- and postoperative mesh points of 1.63 mm and a standard deviation of 5.45 mm in a volume overlay analysis. Recently, Moe et al. published results about their in-house workflow and described a mean surface overlay difference of 1.90 mm with an RMSE of 3.72 mm for 29 virtually planned cases, including 24 FFF (12). These results coincide with ours (virt-post MSD -0.1 (-1.0–5.6); virt-post RMSE 2.3 (1.3–10.7)). A novel aspect of our study is the comparison of preoperative situation and virtual model (pre-virt), which reflects the expected deviation from the “ground truth” model. Consequently, the difference between pre-post and pre-virt also reflects the accuracy when the value reaches 0. For this new parameter, we revealed excellent results for RMSE, MSD, and HD ( Table 2 ).

In addition to potential cost savings and increased accuracy, we believe that this in-house planning has a great advantage especially in the training of junior surgeons. Dealing with virtual planning leads to the trainees intensively dealing with the clinical case, and potentially developing a higher demand for the final surgical result. Therefore, in our department, junior surgeons are introduced to virtual in-house fibula planning and learn to independently plan and create resection and repositioning guides. Unlike clinical engineers, they also perform flap elevation and mandibular reconstruction under the guidance of experienced specialists, so they will be involved in all steps of these complex reconstruction cases at an early stage in their careers. But this aspect (e.g. evaluation of junior surgeon´s satisfaction, confidence or virtual planning) was no not evaluated this study.

After a certain learning period, open-source software facilitates cost-effective and precise in-house virtual planning of mandibular reconstructions with a short lead time and without the need for external companies. Even highly complex reconstructions are thus possible with favorable results. In addition, the open-source software offers an excellent possibility to illustrate the surgical procedure. This might enhance the understanding for younger colleagues and increase their likelihood and frequency of an ideal surgical result.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by The institutional ethics committee of the Technical University of Munich, Klinikum rechts der Isar (Approval number: 326/21 S-EB). Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

LR: Study design/conduction, operations, data interpretation, and major contribution to manuscript writing and revision. PK: Surface matching and analysis, data acquisition and interpretation, contribution to revision. FG: Surface matching and analysis, data acquisition and interpretation, statistical analysis and contribution to revision. MS: Creation of figures, statistical analysis, contribution to revision. JW: Study design, data interpretation, contribution to revision. MN: Data interpretation, contribution to revision. K-DW: Data interpretation, contribution to revision. AF: Study design/conduction, operations, data interpretation, and major contribution to manuscript writing. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.