An Anatomic Analysis of Fibula Flap Mandible Reconstructions: Implications for Endosseous Implant Placement

Abstract

Background:

Precise planning and evaluation of the fibula bone is necessary if immediate endosseous implant placement is considered. Limited information is available on the anatomical dimensions or density of fibula used in vivo mandibular reconstructions. This study aims to describe the morphology and dimensions of the fibula used to reconstruct segmental mandibular defects and contrast the findings with the native mandible.

Methods:

A retrospective analysis was performed of patients who underwent segmental mandibulectomy reconstructed with osteocutaneous fibula flaps and had at least 1 post-operative CT scan. Fibula cross sectional dimensions and densities were evaluated with three-dimensional software. Radiographic measurements were obtained from the contralateral mandible medial to the first molar for comparison.

Results:

477 fibula cross sections from 159 segments were evaluated. Cross sectional shapes of oval, quadrilateral, triangular, and pentagonal differed significantly in proportion (p<0.001). Thirty-eight percent of segments (95%CI: 30–46%) had differences in cross section height greater than 1 millimeter (p<0.001). Between segments within the same patient, the median height difference was 1.58 mm (range: 0.14–6 mm). The superior cortex density was significantly higher for the fibula than the native mandible; however, the medullary space density was significantly lower (p<0.001).

Conclusions:

The current study comprises the most comprehensive description of fibula morphology in mandibular reconstructions and highlights the significant variability which exists. The findings provide justification for the added time and cost of CAD-CAM in centers interested in performing immediate dental implant placement as the technology provides the necessary precision and accuracy.

INTRODUCTION

First described by Hidalgo in 1989¹, the osteocutaneous fibula free flap is currently regarded as the standard method for reconstruction of large segmental mandibular defects for benign or malignant disease. This flap is preferred because of its dual blood supply, length of bicortical bone stock, sufficient vascular pedicle, ability to be contoured, and its suitability as a recipient site for osseointegrated endosseous implants.^2,3,4,5 Precise evaluation of the fibula dimensions are needed if immediate or delayed implant placement is being considered. Endosseous implants are intended to be placed entirely within an osseous structure; therefore, implant size is contingent on the morphology and anatomical dimensions of the bone being utilized. Failure to adequately assess the proposed implant site can result in implant-related complications, such as implant instability, implant thread exposure and/or implant failure secondary to the endosseous implant not being completely housed in bone. Moreover, inaccurate measurement of the fibula dimensions can lead to perforation of both cortices by the endosseous implant potentially leading to fracture.

Although cadaver studies have been performed, data on fibula morphology in-vivo following actual mandible reconstruction is limited.^{6,7,8,9,10,11,12} A relevant aspect of dental implant placement not reported in the literature is the relative density of the fibula compared to the native mandible. The current study aims to describe the morphology, dimensions, and density of the fibula bone from a cohort of reconstructed segmental mandibular defects to inform dental implant placement. The hypothesis is that significant variation exists in fibula anatomy both within and between patients.

MATERIALS AND METHODS

Patient Identification

The study was approved by the Institutional Review Board at Memorial Sloan Kettering Cancer Center (#17–271). A retrospective study was completed of all patients 18 years or older who underwent segmental mandibulectomy reconstructed with osteocutaneous fibula free flaps from January 2006 until March 2017 at a tertiary cancer center (N=414) and had at least 1 post-operative CT with or without contrast material (N=94). For patients with multiple CT scans, the CT obtained nearest to the date of reconstruction was used for the analysis. A clinical chart review was completed to gather demographic, tumor, and treatment information.

TeraRecon Software and Study Design

Three-dimensional medical imaging software (TeraRecon, Inc.; Foster City, CA) was utilized to evaluate CT scans of 94 patients who had reconstructed mandibles. Each reconstruction was categorized based on the HCL classification system.¹³ For each reconstruction the number of vertical (i.e. ramus) and horizontal fibula segments were recorded as well as the total number of fibula segments. The length of each horizontal fibula segment within a mandibular reconstruction was measured in millimeters and systematically divided into thirds creating sub-segments (Figure 1). Each sub-segment was evaluated in cross-section to assess fibula shape, anatomical dimensions, and density. If the midpoint of a segment in cross section involved a reconstruction plate or fixation screw, the cross section closest to the midpoint without a radiographic obstruction was used. In many instances streak artifact from metal plates was able to be avoided because we generally use mini-plates for bony fixation as opposed to reconstruction bars. Fibula shape was classified as oval, pentagonal, quadrilateral, or triangular (Figure 2). Fibula height was measured in millimeters from the superior cortex to the inferior cortex. The heights of each cortex and medullary space were recorded and summed to determine height. The lateral cortex, medullary space, and medial cortex were measured in millimeters and summed to determine width (Figure 3). Lastly, the density of each fibula cross-section was measured in the superior cortex, medullary space, and inferior cortex. Bone density was recorded in Hounsfield units (HU) by way of a 1mm×1mm averaged circle (Figure 4).

Figure 1.

Horizontal Fibula Segment length measured in millimeters (yellow), sectioned into thirds (red), and divided in half to measure each cross section (CS) (orange) (Image reconstructed with source data from CT scan for demonstration purposes courtesy of Materialise, USA)

Figure 2.

Fibula Morphology: Oval, Quadrilateral, Pentagonal, and Triangular. (Images reconstructed with source data from CT scans for demonstration purposes courtesy of Materialise, USA)

Figure 3.

Height and width of cross sections (mm) (Images reconstructed with source data from CT scans for demonstration purposes courtesy of Materialise, USA)

Figure 4.

Density of cross sections measured in HU (Images reconstructed with source data from CT scans for demonstration purposes courtesy of Materialise, USA)

Of the 94 patients, 31 had sufficient radiographic data to measure the contralateral native mandible. The area mesial to the 1^st molar was evaluated as this is a common area for implant placement. The same protocol was used to record native mandible anatomical dimensions at this site [height (mm) and density (HU)].

Biostatistics

Measurements were described on the cross-sectional level overall and by shape and visualized with box plots. The box represents 50% of the observations, the interquartile range (IQR) from the 25^th to the 75^th percentile. The line in the middle represents the median, and the circle is the mean. The bars extend to 1.5 times the IQR (75^th-25h percentile), and observations beyond these bars represent outliers. Generalized estimating equations (GEE) with identity link functions, and exchangeable correlation matrices were used to assess the associations between shape and continuous covariates. Correlation within segment and within patient were accounted for. Additionally, the relationship between prior RT and reconstructed measurements was assessed with GEE models. For GEE models, we used the least squares mean (LSmean), from the model as estimates. For the relationship between time since reconstruction and shape, time since reconstruction was log-transformed.

The data for native mandible measurements were only available in a subset of patients. Therefore, the data were checked for informative missingness of the reconstructed values with missingness and patient characteristics with missingness. GEE were used for the former, and Fisher’s Exact and Wilcoxon Rank Sum with the latter. As only one measurement was available for the contralateral native mandible, the average of all segments and cross sections were taken and analyses were done on the patient level data. The relationships between native and reconstructed mandible measurements were assessed with the Wilcoxon Signed Rank test. We estimated the proportion of cases along with adjusted Clopper-Pearson 95% confidence intervals (CI) where the difference in cross sectional measurements for total height was larger than 1 mm on the segment level. We assessed if this proportion was greater than zero using the one sample Rao-Scott chi square test, which accounted for multiple measurements within the same patient. We used the same method to assess if the proportions of each shape differed on the cross-section level.

Two-sided p-values less than 0.05 were considered statistically significant, and all analyses were performed with SAS 9.4 (The SAS Institute, Cary, NC).

RESULTS

Postoperative CT images of 94 patients were analyzed after primary fibula free flap reconstruction for tumor ablation or pathologic fracture. The median time between mandibular reconstruction and CT was 5.8 months (range: 0.1–78.6 months). No significant association was found between timing of CT and the reconstructed mandibular shape (p=0.67). There were 56 male (59.6%) and 38 female (40.4%) patients in the study cohort (median BMI 25.4; range 15.0–40.0). Fifty patients (47%) received adjuvant radiotherapy. Detailed tumor characteristics are presented in supplemental materials (Supplemental Table 1). Mandibular reconstructions classified using the HCL system¹³ are presented in Table 1. Sixty-two reconstructions consisted of only horizontal segments while 32 mandibular reconstructions involved a vertical or ramus component. Overall, 159 separate fibula segments were analyzed with a total of 477 cross sections.

Table 1.

Reconstruction characteristics

		N (%)
Reconstruction Site	H	16 (17)
	L	31 (33)
	C	1 (1.1)
	HC	8 (8.5)
	LC	19 (20.2)
	LCL	18 (19.1)
	HCL	1 (1.1)
# Segments	1	47 (50)
	2	29 (30.9)
	3	18 (19.1)
# Horizontal Segments	1	47 (50)
	2	29 (30.9)
	3	18 (19.1)
# Vertical Segments	0	62 (66)
	1	32 (34)
Shape	Oval	155 (32.5)
	Quadrilateral	152 (31.9)
	Pentagonal	55 (11.5)
	Triangular	115 (24.1)

In the cross sections, oval shape was identified most commonly (155/477, 32.5%, 95%CI: 26–40%), followed by quadrilateral (152/477, 31.9%, 95%CI: 26–38%), triangular (115/471, 24.1%, 95%CI: 18–31%), and pentagonal (55/471, 11.5%, 95%CI: 8–16%) [Figure 2]. These proportions were significantly different (p<0.001).

Fibula cross sectional height stratified by anatomic shape is presented in Figure 5. The median cumulative height for the fibula free flap was 13.42mm (range: 8.99–19.38mm). More specifically, the median superior cortex height was 3.20 mm (range: 1.43mm to 8.69mm), the median medullary space height was 6.18mm (range: 1.60mm to 12.40mm), and the median inferior cortex height was 3.45mm (range: 1.47mm to 6.54mm). The distribution of total height measurements was similar for all four shapes.

Figure 5.

Boxplots of Fibula Height Measurements by Anatomic Shape.

The box represents 50% of the observations, the interquartile range (IQR) from the 25^th to the 75^th percentile. The line in the middle represents the median, and the diamond is the mean. The bars extend to 1.5 times the IQR (75^th–25^th percentile), and observations beyond these bars represent outliers.

The fibula cross sectional width measurements are presented Figure 6. The median total width for the fibula was a 9.71 mm (range: 5.90–15.73mm). The median lateral cortex width was 2.45 mm (range: 1.25mm to 5.40mm), the median medullary space width was 4.31 mm (range: 1.06mm to 8.84mm) and the median medial cortex width was 2.80 (range: 1.35mm to 5.31mm). The distribution of widths was similar for the four anatomic shapes.

Figure 6.

Boxplots of Width Measurements by Shape. The box represents 50% of the observations, the interquartile range (IQR) from the 25^th to the 75^th percentile. The line in the middle represents the median, and the diamond is the mean. The bars extend to 1.5 times the IQR (75^th–25^th percentile), and observations beyond these bars represent outliers.

The density of fibula cross sections is presented in Figure 7. The median superior cortex density was 1337 HU (range: 510 HU to 1846 HU), the median medullary space density was 82 HU (range: 1–546 HU), and the median inferior cortex density was 1334 HU (range: 740–1850 HU). The density of the superior and inferior cortices did not differ significantly although both were significantly higher than the medullary space (Parameter Estimate (PE): 1204–1217; p<0.001). The distributions of density measurements were similar across all 4 anatomic shapes.

Figure 7.

Boxplots of Density Measurements by Shape. Thee box represents 50% of the observations, the interquartile range (IQR) from the 25^th to the 75^th percentile. The line in the middle represents the median, and the diamond is the mean. The bars extend to 1.5 times the IQR (75^th–25^th percentile), and observations beyond these bars represent outliers.

Between segments within the same patient, variability was wide; the median height difference within the same patient was 1.58 mm (range: 0.14–6 mm). Of the 159 segments evaluated in this cohort, 38% (95%CI: 30–46%) had differences in cross section total height measurement greater than 1 millimeter. This proportion was significantly different than zero (p<0.001). Further, 36% of patients (95%CI: 23–51, N=17/47) had no overlap in height measurements between any of the cross sections. The intra-segment variability was also wide; the median within segment height difference was 0.73 mm with 25% of patients having a difference between 1.48 and 3.72 mm.

In the 31 patients with available data, all native mandible measurements differed significantly from fibula flap measurements (Table 2). The superior cortex density (median: 1374.3 HU) was significantly higher for the fibula than the native mandible (median: 869 HU; p<0.001). In contrast, the inferior cortex density (median: 1370.50 HU) and medullary space density (median: 88.00) were significantly lower for the fibula compared to the native mandible inferior cortex (median: 1562 HU) and medullary space densities (median: 509 HU), respectively (p<0.001). The superior cortex height was higher in the fibula (median: 3.05 mm) compared to the native mandible (median: 2.12 mm; p<0.001). The medullary space height was significantly lower in the fibula (median: 6.33 mm) compared to the native mandible (median: 21.40 mm; p<0.001). Overall, the total height was lower in the fibula (median: 12.73 mm) than in the native mandible (median: 28.70; p<0.001).

Table 2.

Relationship between fibula free flaps and native mandible

	Median(Range)
	Fibula	Native	Difference	p-value
IC Density	1370.5 (848.33–1698.7)	1562.0 (1242.0–1775.0)	−167.7 (−437.3–456.67)	<.001
M Density	88.00 (7.67–284.67)	509.00 (320.00–821.00)	−398.0 (−758.0–−197.0)	<.001
SC Density	1374.3 (771.33–1709.3)	869.00 (532.00–1103.0)	493.33 (−97.67–1111.3)	<.001
IC Height	3.64 (2.25–5.36)	4.62 (3.43–6.84)	−1.21 (−3.52–1.32)	<.001
M Height	6.33 (3.75–9.30)	21.40 (8.76–26.90)	−15.15 (−20.99–−2.56)	<.001
SC Height	3.05 (2.04–6.13)	2.12 (1.37–3.62)	0.84 (−1.30–3.45)	<.001

The inferior cortex of the native mandible density (IC) and height (IC) are without clinical importance as these regions would not be used for conventional dental implant placement.

DISCUSSION

Fibula free flaps utilized for mandibular reconstruction have been incompletely characterized with regard to functional dental planning. Accurate and precise endosseous implant placement requires nuanced understanding of the anatomical geometry as well as the density of fibula segments utilized. Previous studies on fibula flap mandible reconstruction have concentrated on bone shape of the neo-mandible parabola with aesthetic outcomes as the primary focus^1,2 with limited description of fibula morphology as it relates to endosteal implant planning. Moreover, existing cadaveric studies that describe fibula morphology have not focused attention on the clinically relevant distal third of the fibula, the portion most commonly used in mandible reconstruction.^12,14 With this background in mind, the current study comprises the most complete description of fibula morphology and dimensions in mandibular reconstructions of oncologic patients.

Clinical intraoperative experience with immediate endosteal implant placement at the time of fibula transfer has demonstrated the importance of anatomical geometry. For example, depending upon the shape of the bone, ideal endosseous implant trajectory can vary. Intraoperative modifications to the fibula are often required, including use of a bur to flatten prominent projections of the bone to enable a level surface and avoidance of drill slippage. Four recurring shapes were identified upon review of all cross sections of reconstructed neo-mandible. These shapes in order of decreasing frequency included: oval, quadrilateral, triangular, pentagonal, and differed from previous morphological classifications.^8,13 Pre-operative imaging of individual patients fibula allows the clinician to be prepared for these anatomical variations and to anticipate reduction of undesirable osseous structures, such as thin or angulated bone, as these areas impede implant placement. Importantly, the 4 shapes identified did not demonstrate significant differences in either horizontal or vertical dimensions or density. However, striking variation in both fibula height and width was demonstrated as shown in the boxplot distributions (Figures 5 and 6). The clinical importance is that endosseus implant placement is not one size fits all requiring patient specific length adjustments to ensure the implant is long enough to make adequate contact with the inferior luminal cortex, but without perforation.

When compared to native mandibular bone, significant differences were seen both in the vertical dimension of measurement and in the density of the fibula bone (Table 2). The principle finding of relevance is that the density of the medullary space is much greater for the native mandible than the fibula. These radiographic findings translate to clinical differences when placing endosseous implants. For example, when endosteal implants are placed in the fibula, because there is minimal density to the medullary space, the dental implant must engage both the superior and a portion of the inferior cortex to be stable without a cantilever effect. In contrast, endosseous implants only need to be placed in the medullary space of the native mandible to obtain adequate stability. Lastly, relative differences in density between the native mandible and fibula may require adjustments in drill speed to avoid thermal damage and osteocyte necrosis. Although outside the scope of this study it is hypothesis generating for future investigations.

Another observation from this study is the variability of fibula height within a segment and between fibula segments from the same individual. For example, the distribution of vertical height within each fibula segment varied by more than 1mm 38% of the time. Therefore, it is unsafe to assume that the same height dental implant can be used anywhere along the fibula bone. Clinically this can translate into under-preparation of the endosseous implant osteotomy site resulting in instability of the implant and/or perforation of both fibula cortices during implant placement. Preoperative planning which ensures use of the intended fibula segment for reconstruction and implant placement should be closely followed to mitigate risk of such outcomes.

The findings of the current study are particularly important as utilization of CAD/CAM technology for head and neck reconstruction has increased in recent years. The accuracy and precision of the neo-mandible reconstruction compared to the native mandible can be reliably assured with this technology; therefore, the opportunity for successful immediate dental implant placement is maximized. The cutting jigs applied to the fibula not only facilitate the bony osteotomies but can simultaneously serve as drilling guides for the dental implants. Knowledge about fibula shape and size variation both within and between individuals obtained in the current study is useful both during the virtual surgical planning process, as well as intraoperatively when adapting the jig to the fibula. For example, if the jig is not applied to the area of the fibula intended in the virtual plan, it will not adapt to the bone properly creating inaccurate dental implant position and trajectory. To improve the accuracy of fibula jig placement two clinical reference points are now routinely used intraoperatively. First, the fibula cutting guide position is referenced to a predetermined distance proximal to the lateral malleolar prominence to ensure conformance with virtual surgical plan. Second, the fibula jig has a notch annotating the position of the skin island perforator identified on the scan and used in the virtual planning session. In short, we believe the current manuscript provides justification for the added time and cost of CAD-CAM in centers interested in performing immediate dental implant placement as the traditional free hand method does not provide sufficient information to assess the many variations in fibula morphology.

This study has several limitations which should be acknowledged. First, since this was a retrospective cross-sectional study, CTs were not taken at a consistent time points post-operatively. Additionally, the CTs were of variable slice thickness which can impact image resolution. Although the slice thicknesses were adequate for measurement, standardized CT slice thickness would ensure consistency. Assessment of fibula morphology and measurements were obtained by only one evaluator, so no information is available on measurement correlation between different raters. Additional studies on this topic should also evaluate the longitudinal impact of endosseous implant on fibular flap bone height and morphology as this is not well understood.

In summary, the fibula bone used in neo-mandible reconstruction has significantly different characteristics from that of the native mandible and should be approached with this in mind prior to proceeding with endosseous implantation. The study provides anatomical dimensions of neo-mandible reconstructions and highlights the variability present both within and between individuals’ fibulas. Pre-surgical planning technology can be leveraged to facilitate immediate dental implant placement; however, the virtual plan should be closely followed intraoperatively to maximize successful osseointegration and minimize complications.