Magnetic resonance angiography (MRA) of patients with dominant posterior tibial arteries. Preoperative lower extremity 1.5-T MRA (A) and 3-T MRA (B) illustrating dominant posterior tibial arteries with normal-appearing perforators for 2 patients. These patients were found intraoperatively to have skin perforators originating from the posterior tibial vessels that were not apparent on MRA. A normal lower extremity MRA (C) is shown for comparison.
Intraoperative vs magnetic resonance angiography (MRA) detection of skin perforators. A graphical representation of correlation between the number of perforators found on MRA and at surgery. The dashed line indicates data distribution that would occur with 100% agreement between MRA and intraoperative findings.
Miller ME, Moriarty JM, Blackwell KE, Finn JP, Yiee JH, Nabili V. Preoperative Magnetic Resonance Angiography Detection of Septocutaneous Perforators in Fibula Free Flap Transfer. Arch Facial Plast Surg. 2011;13(1):36-40. doi:10.1001/archfacial.2010.110
To investigate whether preoperative magnetic resonance angiography (MRA) is predictive of surgical findings in fibula free flap surgery for head and neck reconstruction.
Retrospective review (April 2004 until September 2009) of 123 patients who underwent preoperative MRA as part of surgical planning for fibula free flap tissue transfer for head and neck reconstruction. Each MRA was reviewed by a board-certified radiologist masked to the intraoperative findings and to the number of septocutaneous perforators documented. Operative notes were reviewed and the number of septocutaneous perforators found during the operation was recorded. A κ interrater agreement statistic was calculated to compare these values.
Two vascular anomalies found during the operation were undetected by MRA. Analysis of the entire cohort demonstrated that agreement between the number of perforators documented on MRA and the number found intraoperatively approached zero (unweighted κ = −0.088, P = .04). The agreement between the 2 values was 17.9% and the average percentage correctly classified was 10.9%.
Contrary to previous reports, preoperative MRA does not accurately predict the presence and/or number of skin perforators found intraoperatively for a fibula free flap operation. The surgeon should not be dissuaded from planning a fibula free flap operation if skin perforators appear unfavorable on preoperative MRA because intraoperative observation is definitive. The surgeon should prepare for anomalous cases in which perforators may arise from the posterior tibial system. Further investigation is needed to achieve more accurate imaging modalities for evaluating septocutaneous perforators prior to free fibula flap transfer.
Preoperative magnetic resonance angiography (MRA) is used in head and neck reconstructive procedures for operative planning for free flap reconstruction. Fibula free flaps are a common reconstructive technique for oral cancers involving the mandible, mandibular osteoradionecrosis, and other benign, destructive, or traumatic mandibular processes.1 As a robust, reproducible, noninvasive imaging modality with no ionizing radiation penalty, MRA is currently the standard preoperative test used for surgical planning of fibula free flaps in the Division of Head and Neck Surgery at UCLA. This modality has decreased morbidity and higher cost-effectiveness compared with traditional angiography.2 Magnetic resonance angiography is used to assess the patency of lower-extremity vessels and to rule out vascular anomalies that could lead to ischemia. Most recently, MRA also has been proposed for evaluation of septocutaneous perforators in a small series.3
The most feared complications of fibula free flap surgery occur if vascular anomalies, which can cause pedal ischemia if the peroneal artery is used, are not identified preoperatively.4 Peronea arteria magna occurs when the peroneal artery is the dominant vessel in the lower extremity; Rosson and Singh5 reported a range of prevalence from 0.2% to 8.3% in the general population. Use of the peroneal artery in this setting requires extensive and costly measures to prevent limb loss, and so Rosson and Singh5 advocated use of preoperative MRA for patient safety as well as cost-effectiveness. Although some authors have shown series of fibula free flaps without preoperative imaging that do not develop complications from vascular anomalies,6,7 others have demonstrated that clinical examination is insufficient to determine whether a patient is at risk for limb ischemia postoperatively.8
Most surgeons now agree that some form of preoperative vessel imaging is prudent to determine vessel patency and rule out vascular anomalies,9 and MRA is the modality of choice in many centers. However, the accuracy of MRA in preoperative examination of septocutaneous perforators remains unclear. To determine whether preoperative MRA provides a useful guide to septocutaneous perforators found during an operation, we retrospectively reviewed our series of 123 fibula free flaps, performed at a single institution, recording preoperative MRA and operative findings.
This study was approved by the institutional review board at UCLA. Patients who had undergone preoperative MRA and fibula free flap surgery from April 2004 to September 2009 at UCLA Medical Center were identified. The diagnoses leading to need for surgery and imaging acquisition parameters, including strength of magnetic resonance scanner field (1.5 T vs 3 T), were recorded. Operative notes were reviewed for the number of septocutaneous perforators found during the operation and any vascular anomalies encountered.
Each MRA was reviewed by a board-certified radiologist (J.M.M.) masked to intraoperative findings and to the number of septocutaneous perforators documented. Assessment of the lower-limb arterial vasculature was performed using contrast-enhanced MRA. Studies were performed in a single center with either a 1.5-T Siemens Avanto Scanner or a 3.0-T Siemens Trio Scanner (Siemens Medical Systems, Erlangen, Germany). At 1.5 T, voxel resolution was 1.1 × 0.9 × 1.1 mm. At 3.0 T, voxel resolution was 1.0 × 0.8 × 0.8 mm.
Image interpretation was performed on a dedicated workstation (GE Healthcare, Centricity) at UCLA with additional multiplanar reconstruction performed in line as required. Interpretation followed guidelines set out in the literature,10 namely, review of axial, sagittal, coronal, and 3-dimensional reconstructions and cine MRA sequences. Angiographic images were evaluated for anomalies of the tibio-peroneal structure as well as the presence, absence, and number of septocutaneous perforators arising from the peroneal artery. Peroneal branches that entered or sent branches to the calf musculature were discounted.
A κ interrater agreement statistic was calculated using Stata 11 (College Station, Texas). Unweighted κ statistics were first computed. These were compared with weighted κ statistics. Weighted κ statistics were appropriate given the clinical relevance of the subject matter. In unweighted κ statistics, no credit is given to an incorrect answer, and partial credit can be given to an incorrect answer by taking into account its clinical relevance. In this study, an MRA prediction of 3 arteries and an intraoperative finding of 2 arteries are similar enough for clinical use and thus given partial credit. Further analysis was performed to evaluate the relative accuracy of MRA at 1.5 T and 3 T.
One hundred twenty-three patients who underwent preoperative MRA as part of surgical planning for fibula free flap tissue transfer for head and neck reconstruction from April 2004 to September 2009 were included. Patient characteristics are documented in Table 1. The most common diagnoses leading to the need for the fibula free flap operation were oropharyngeal squamous cell carcinoma and/or osteoradionecrosis of the mandible. Most patients underwent 3-T rather than 1.5-T MRA (74.0% vs 26.0%).
Two of 123 patients were found to have a single perforator originating from the posterior tibial artery instead of the peroneal system. The first of these patients had a preoperative 1.5-T MRA that showed widely patent trifurcation arteries with dominant posterior tibial arteries in both lower extremities (Figure 1A). On postoperative review of the preoperative MRA, the radiologist involved in this study identified the presence of 1 perforator. In the operating room, however, the surgical team had encountered 2 skin perforators originating from the posterior tibial system. A proximal branch of the peroneal system to the soleus muscle was identified and preserved, and the lateral half of the soleus muscle was harvested with the free flap.
The second patient had a preoperative 3-T MRA that indicated widely patent and symmetric 3-vessel calf runoff bilaterally and dominant posterior tibial arteries (Figure 1B). As with the previous patient, on review of the MRA, the radiologist recorded 1 perforator. During the operation, the patient was found to have a single large perforator to the skin coming off the posterior tibial artery rather than the peroneal system. The skin paddle could not be used in this case, so fibular bone was used along with a left pectoralis major myocutaneous regional flap for reconstruction of a left buccal mucosa defect.
Including these 2 patients with lower limb vascular variants, 123 patients were included in the statistical analysis. The percentages of patients with 0 to 5 perforators found during the operation were 0%, 17.0%, 54.5%, 22.0%, 4.9%, and 1.6%, respectively. The percentages of patients with 0 to 5 perforators found on MRA review by the radiologist were 19.5%, 44.0%, 27.6%, 8.9%, 0%, and 0%, respectively.
The agreement between the number of perforators documented on MRA and the number found intraoperatively was nearly zero (unweighted κ = −0.088, P = .04) (Table 2). Interrater reliability is considered poor for values less than zero.11 The observed agreement between the 2 values was 17.9%, and the average percentage correctly classified was 10.9% (Figure 2).
Taking into account clinical relevance, weighted κ values were also calculated. Even in an extreme weighted scheme, where all nonzero artery findings were given a full weight of 1 if the other observer also found a nonzero artery, the κ statistic was −0.023.
The 3-T MRA scans were compared with the 1.5-T scans to determine whether high-field strength MRA would give different results. The 3-T κ (−0.075) was not significantly different from the 1.5-T κ (−0.116), and neither was better than the combined value.
Preoperative planning is necessary in fibula free flap surgery to ensure that the perforating vessels will be sufficient to complete the tissue transfer.12 Angiography is the traditional imaging modality used prior to fibula free flap surgery,13 but it is associated with significant patient morbidity.14 Modalities used other than MRA include color flow Doppler ultrasound and computed tomographic angiography (CTA).
Preoperative color flow Doppler evaluation of the lower extremity has been described,15 and it is inexpensive and carries low risk to patients. However, this modality may be limited by inferior anatomic resolution for the deep peroneal system owing to proximity to the fibula and has been criticized for being operator-dependent.16 Although ultrasound imaging may not be a reliable technique by all operators, many surgeons have found that it is useful for mapping perforators preoperatively and intraoperatively, thus affecting skin paddle design.17
Computed tomographic angiography has been used preoperatively for fibula free flaps and is accurate in demonstrating vascular and atherosclerotic disease.14 Computed tomographic angiography also has been shown to identify vascular anomalies of the lower extremity preoperatively.18 Moreover, some authors have suggested that CTA can accurately demonstrate the size, course, and penetration pattern of all perforators greater than 0.3 mm in diameter.19 To our knowledge, no direct comparison of CTA and MRA of the lower extremity has been completed. In 2009, Rozen et al20 found CTA to be superior to MRA for mapping of epigastric vessels, but this study did not examine the lower extremity. Furthermore, the necessity of using a nephrotoxic iodinated contrast medium and the use of ionizing radiation with CTA has led to the adoption of MRA as the preoperative imaging modality of choice at our institution.
Surgical planning for fibula free flaps is altered when MRA reveals proximal vessel occlusion in the setting of normal dorsalis pedis pulses.2 Surgeons also use findings from MRA when deciding to change the side of the planned fibula free flap in patients with unilateral vascular anomalies and to rule out the surgical procedure for those with bilateral anomalies. Thus, MRA remains a critical preoperative step to evaluate large-vessel anatomy and has been reported to detect significant lower-extremity vessel stenosis with a sensitivity of 77% to 100% and specificity of 87.6% to 99.7%.21
Despite reports that MRA can detect vascular anomalies preoperatively, as described by Rosson and Singh,5 it failed to identify 2 of 123 patients in the present study who had a single perforator originating from the posterior tibial artery instead of the peroneal system. As discussed, the planned skin paddle in both of these patients could not be used and alternative soft-tissue reconstruction was performed. Although MRA may provide enough anatomic resolution to identify extreme vascular anomalies, such as peronea arteria magna,5 it may not possess sufficient spatial resolution to reveal important vascular variants, such as atypical skin perforators, that affect surgical planning and outcome.
Furthermore, although MRA has been shown to be useful in evaluation of larger vessels, we found that it has very poor correlation with operative findings for skin perforator anatomy. In fact, our data suggest that the correlation is not significantly different from zero. Even when an extreme weighted scheme is used to determine whether MRA and surgical findings agree when separated into zero (no perforators) and nonzero (at least 1 perforator) categories, the correlation is still very low.
Other studies that have examined the use of preoperative MRA in fibula free flap design are limited. Fukaya et al3 found that MRA demonstrated the septocutaneous perforators from the peroneal artery in fibula free flaps with 100% accuracy; however, this study had a limited sample size of 7 patients, none of whom had abnormal vessel anatomy. Our results, in the largest series to date (N = 123), indicate the opposite finding and exemplify the power and implications of a larger sample size.
Although the present study is larger than several studies discussed, it has several limitations. Masking the radiologist who reviewed the MRA studies to operative findings eliminated a bias; however, MRA interpretation may have some interrater variation. Moreover, MRA technique varies among institutions. In fact, results between 1.5-T and 3-T scans showed some variability even within this study, although the difference was not significant.
Although preoperative MRA has been shown to be useful in demonstrating lower-extremity large-vessel anatomy and has been used to identify vascular anomalies, our results indicate that this modality is far from ideal for preoperative identification of septocutaneous perforators. It not only failed to recognize vascular variations in 2 patients, leading to intraoperative alterations in the surgical plan, but the correlation between number of skin perforators shown on MRA and those found during the operation was very poor. Despite these limitations, preoperative vascular imaging with MRA continues to be successful as an integral part of planning whether a fibula free flap is safe to perform. Although no major vascular anomalies were discovered in MRA analysis of this patient cohort, it is clear from the literature that MRA is essential prior to fibula free flap operations to prevent catastrophic ischemia and possible lower-limb loss.
The study reported here demonstrates that the surgeon should approach MRA findings with caution when determining preoperative skin paddle candidacy. Moreover, the surgeon should proceed with harvesting a fibula free flap in the setting of an MRA showing main pedicle patency, regardless of the presence of skin perforators on imaging. For now, only intraoperative identification of skin perforators and their vessel of origin will determine skin paddle adjustments or the need to seek additional soft-tissue reconstructive options. Improved imaging techniques are needed for more reliable preoperative detail of skin perforator anatomy in free fibula transfer.
Accepted for Publication: February 27, 2010.
Correspondence: Mia E. Miller, MD, David Geffen School of Medicine, University of California, Los Angeles, Building CHS 62-237, 10833 Le Conte Ave, Los Angeles, CA 90095.
Author Contributions: Dr Miller had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Moriarty, Blackwell, Finn, and Nabili. Acquisition of data: Miller, Moriarty, Blackwell, Finn, and Nabili. Analysis and interpretation of data: Miller, Moriarty, Yiee, and Nabili. Drafting of the manuscript: Miller, Moriarty, Finn, and Nabili. Critical revision of the manuscript for important intellectual content: Miller, Blackwell, Finn, Yiee, and Nabili. Statistical analysis: Yiee and Nabili. Obtained funding: Nabili. Administrative, technical, and material support: Finn and Nabili. Study supervision: Moriarty, Blackwell, and Nabili.
Financial Disclosure: None reported.
Previous Presentations: This study was presented and awarded Best Clinical Research at the American College of Surgeons Southern California Chapter Head and Neck Section; January 24, 2010; Santa Barbara, California.