Pharyngoesophageal Reconstruction With the Ulnar Forearm Flap

Objective  To evaluate the use of a generous fasciocutaneous ulnar forearm free flap in pharyngoesophageal reconstruction.

Design  Tertiary referral center.

Setting  Private practice.

Patients  From September 1, 1991, to October 30, 1996, 20 ulnar forearm free flaps were used to reconstruct the pharyngoesophagus in 19 patients after surgery for squamous cell carcinoma. There were 13 circumferential defects and 7 near-circumferential defects (<2 cm of mucosa remaining). The reconstructions were performed primarily (at the time of tumor resection) in 16 cases and secondarily in 4 cases.

Intervention  A generous fasciocutaneous ulnar forearm flap with a minimum dimension of 9×22 cm was harvested to reconstruct the pharyngoesophagus. A 2-layer closure was performed in flap tailoring and proximal (pharynx to flap) and distal (flap to esophagus) anastomoses to minimize the risk of leakage. Also, portions of the flap were used to monitor flap viability, to cover cervical vessels, to obliterate dead space, and as skin coverage when the skin flaps were deficient.

Results  Nineteen of the 20 flaps transferred were successful. Swallow function was restored in 18 cases, and voice was rehabilitated in all the patients using either tracheoesophageal puncture and a voice device or an electrolarynx device. There was 1 perioperative mortality. Three fistulas occurred, all of which healed with nonsurgical treatment. One stricture developed that required intermittent dilatation. The donor site morbidity was minor.

Conclusions  This versatile technique achieves excellent results with a decreased complication rate compared with other methods currently available in pharyngoesophageal reconstruction.

HISTORICALLY, the reconstruction of pharyngoesophageal defect after tumor ablative surgery has been fraught with difficulties. Local and regional flap reconstructions have required multiple procedures.^1,2 The enteric transposition flap³ and the gastric pull-up⁴ have been associated with significant morbidity. Also, these procedures have not consistently achieved satisfactory functional rehabilitation of speech and swallow functions. Although pedicled musculocutaneous flaps, such as the pectoralis major and the latissimus dorsi,^5,6 provide single-stage reconstruction of a pharyngoesophageal defect with less donor site morbidity, the long-term functional results have remained disappointing. Furthermore, fistula and stricture rates can be high, and the limitation of the pedicle length as well as the bulkiness of these flaps can be problematic.

Over the past decade, advances in microsurgical free tissue transfer techniques have improved the outcome of pharyngoesophageal reconstruction.^7,8 The jejunal free flap has been the flap of choice in many major centers, with a reported flap survival rate of 97% and the restoration of swallow function in more than 80% of the patients in recent series.^9,10 The jejunal free flap, however, is not without problems: morbidity associated with the donor site can be significant^11,12; there can be persistent dysphagia from hyperperistalsis of the flap⁸; the speech result of the neopharynx is less than satisfactory¹³; and the incidence of fistula and stricture formation can be as high as 29%¹⁴ and 22%,¹⁵ respectively.

The forearm free flap, based on either the radial¹⁶ or the ulnar artery,¹⁷ has frequently been used in head and neck reconstruction because of its reliability and low donor site morbidity. Harii et al¹⁸ and Anthony et al^8,19 have advocated its use in pharyngoesophageal reconstruction. Although donor site morbidity with the forearm free flap is less than with the jejunum free flap, the incidence of fistula formation can be higher.^8,18,19 Herein, we review our experience with the use of a generous fasciocutaneous ulnar forearm flap in the reconstruction of pharyngoesophageal defect.

Between September 1991 and October 1996, 19 patients underwent 20 free ulnar forearm flap reconstruction of pharyngoesophageal defects after tumor ablation. There were 13 men and 6 women, with an average age of 60 years (range, 41-74 years). All primary lesions were squamous cell carcinoma. Informed consent was obtained before treatment began in all cases. The reconstructions were performed primarily (at the time of tumor resection) in 16 cases and secondarily in 4 cases. Indications for surgery in primary reconstructions were for primary tumor ablation in 7 cases and surgical salvage for recurrence in 9 cases. The primary sites were the hypopharynx in 10 cases, the larynx in 3 cases, and the cervical esophagus in 3 cases. Four cases of secondary reconstruction were performed for either stricture (2 cases) or persistent pharyngocutaneous fistula (2 cases). These 4 patients were unable to tolerate oral feedings or to undergo voice rehabilitation. One patient underwent 2 ulnar forearm free flap reconstructions. The first reconstruction was performed after partial pharyngectomy and cervical esophagectomy. Tumor recurrence required a laryngopharyngectomy and a second ulnar forearm flap 8 months later. Twelve patients underwent preoperative radiotherapy, and 6 patients underwent postoperative radiotherapy.

There were 13 circumferential defects, with the length of the defect ranging from 4 to 12 cm. Seven of the reconstructions were for near-circumferential defects with a maximum of a 2-cm strip of mucosal wall remaining. In the majority of cases, an end-to-side anastomosis from the ulnar artery to the external carotid artery was performed along with an end-to-side basilic vein to internal jugular vein anastomosis. However, 1 case of end-to-end ulnar artery to facial artery anastomosis, 1 case of end-to-end anastomosis of the basilic vein to the cephalic vein at the deltopectoral groove, and 2 cases of venae comitantes anastomoses to the internal jugular vein were treated. The donor site was covered with an unmeshed split-thickness skin graft harvested from the thigh. The skin graft donor site was covered with an occlusive dressing for 10 days. The arm was splinted for 7 days.

Postoperatively, feeding was via either a nasogastric or a gastrostomy tube for 10 days. A barium swallow study was performed or a clear liquid dyed with methaline blue was administered to evaluate the integrity of the reconstruction prior to oral feedings. If the test showed no evidence of leakage, a clear liquid diet was started. The diet was advanced to solid food as tolerated, which generally takes 2 to 3 weeks from the start of the oral feeding.

Both the radial and the ulnar forearm free flaps are thin fasciocutaneous flaps that are well suited for reconstruction of the pharyngoesophagus, although other fasciocutaneous flaps, such as lateral arm and lateral thigh flaps, can also be used. In our experience, the forearm flap is comparatively thinner and more pliable and thus is easily folded to contour the neopharyngoesophagus. We prefer the ulnar forearm flap because of its relative hairlessness compared with the radial forearm flap. Furthermore, the ulnar donor site is more proximally located along the ulnar border, which is less obvious during normal daily activity.¹⁷

An Allen test is routinely performed preoperatively to evaluate the patency of the radial and ulnar arteries as well as the palmar arch. Flap harvest begins with the tourniquet being placed on the upper part of the arm and inflated to 250 mm Hg. Then, a large 9×22-cm (7 cm wide distally, 10 cm wide proximally) flap is sited over the ulnar artery, with the proximal border 1 cm below the antecubital crease and the distal border 1 cm proximal to the volar wrist crease. Starting from the lateral border, the flap is elevated superficial to the deep fascia, which is incised within 1 cm lateral to the ulnar neurovascular bundle, and the elevation is continued to incorporate the anterior medial intermuscular septum, the ulnar artery and venae comitantes, and the septocutaneous perforators. The medial longitudinal incision is made just anterior to the ulna, and the flap medial to the ulnar neurovascular bundle is elevated in a suprafascial manner to include the basilic vein. A curvilinear incision is made over the antecubital fossa, and the basilic vein and its tributaries from the proximal limit of the flap are traced superiorly to obtain an additional 10- to 15-cm length of the venous pedicle. The tourniquet is released and the pedicles are ligated and divided after the adequacy of the circulation to the digits is ensured using the Doppler technique while occluding the ulnar artery.

To reconstruct the pharyngoesophageal defect, the vascular anastomoses are first completed to minimize the ischemia time. Suturing of the flap begins at the superior limit of the defect and continues along the lateral extent inferiorly. In circumferential defects, the portion of the flap that would be required to achieve a tubular closure is estimated, allowing some redundancy to avoid stricture. The region of the flap along the suture line is deepithelialized, and the adjacent vascularized fascia provides additional soft tissue cover for a second layer of closure. Finally, the inferior anastomosis between the flap and the cervical esophagus is completed. Some flap redundancy is purposely created at the proximal and distal anastomoses to prevent stricture.

After the pharyngoesophagus is reconstructed, the remaining distal portion of the flap is deepithelialized, and the remainder of the ulnar forearm fascia is used to obliterate any dead space and/or to cover the cervical vessels. In patients with deficient skin flaps, the forearm flap can be used as skin cover to avoid tight wound closure that may lead to compression of the vascular pedicles or ischemia of the skin flaps. In addition to the standard Doppler technique, a small patch of the skin on the forearm flap is routinely used to monitor flap viability (Figure 1, Figure 2, Figure 3, Figure 4, Figure 5, Figure 6, Figure 7, Figure 8, and Figure 9).

All reconstructions were performed by the senior author (A.H.S.). The mean hospital stay was 12 days (excluding the perioperative death), with a range of 8 to 24 days. Mean follow-up was 19 months (range, 3-63 months).

Of the 20 flaps transferred, 19 were successful. The failed flap occurred early in our series. It was a secondary reconstruction for long-standing stricture after laryngopharyngectomy, bilateral neck dissections, and radiotherapy for a T4 N2 M0 squamous cell carcinoma of the piriform sinus. The defect was near circumferential, with 1 cm of the posterior mucosal wall remaining. Although the flap was successfully transferred, venous thrombosis occurred 18 hours after surgery, and emergent flap salvage was unsuccessful. Consequently, the failed ulnar flap was removed and a pectoralis major myocutaneous flap was used to reconstruct the resultant defect. A fistula developed, but healed spontaneously without problems. Another case of venous thrombosis occurred 36 hours after surgery. This flap was successfully salvaged, without sequelae.

There was 1 perioperative death secondary to pulmonary embolism on the 6th postoperative day. Fistula occurred in 3 cases (including the failed flap with subsequent pectoralis flap reconstruction), 2 of which were primary reconstructions. The fistulas healed spontaneously without surgical intervention; however, a delayed stricture located at the distal suture line between the flap and the cervical esophagus developed in 1 of the 3 cases. Intermittent dilatation was required to maintain a soft-solid/pureed diet.

Donor site complications were few. Forearm tendon exposure was not encountered. A small amount (15%-20%) of skin graft loss over the donor site occurred in 2 cases (10%). These wounds were treated with local care, and epithelialization took place within 2 months. Functional morbidity of the arm was minimal, except for some hypesthesia over the donor site. The donor site scars have been acceptable and not obvious during normal daily activity owing to their proximal location along the ulnar border. The skin graft donor sites on the thigh all healed within 1 to 2 months after surgery.

Three patients underwent the creation of a controlled pharyngostome at the time of flap reconstruction early in our series owing to concerns of fistulization. The pharyngostomes were closed with the patient under local anesthesia in an office setting within 2 to 3 months after surgery. Neither fistula nor stricture developed in these 3 patients. This practice was discontinued as we gained confidence in our reconstructions.

The swallow function was restored in all patients except one who had previously undergone a tongue base resection and partial mandibulectomy. This patient has continued to receive gastrostomy tube feedings. Of the patients whose swallow function was restored, 13 tolerate a regular diet, 3 tolerate a soft/pureed diet, and 2 tolerate a liquids-only diet. Speech rehabilitation with excellent voice quality was achieved with tracheoesophageal puncture and voice prosthesis in 2 patients. The rest of the patients communicate with the aide of an electrolarynx device without difficulty.

The forearm fasciocutaneous flap has high-quality, large-caliber vessels and long vascular pedicles. It is highly vascular and can be harvested rapidly with minimal donor site morbidity. Also, it is thin and pliable, allowing it to be folded and contoured to fit virtually any defect. The free forearm flap is frequently used in head and neck reconstruction, with a success rate exceeding 95% in most large series.^20-23

Harii et al¹⁸ first reported the use of a radial forearm free flap in the reconstruction of pharyngoesophageal defects. In the reconstruction of a circumferential defect, they prefabricated the flap by tubing it on itself at the arm before transferring it to the neck and using a single-layer closure. Although 9 of the 12 patients were able to tolerate a solid diet, stricture developed in 2 patients, and fistula occurred in 5 patients, 2 of whom required surgical closure. To minimize the development of strictures, Harii and colleagues recommend making the suture line wavy and large in diameter.

Chen et al²⁴ reported 6 cases of secondary noncircumferential pharyngoesophageal reconstruction for strictures following colonic transposition. The reconstruction was achieved using a patch radial forearm graft with the peripheral 1 cm of the flap deepithelialized to enable a 2-layer closure. A minor leakage that closed spontaneously within 2 months developed in only 1 patient. Dysphagia occurred in 1 patient owing to flap redundancy that required surgical revision. Necrosis of the neck skin flap led to a secondary chest flap for coverage in 1 patient. All 6 patients were able to tolerate oral feedings postoperatively.

In a report of 12 pharyngoesophageal reconstructions, 7 of which were circumferential defects, Kelly et al²⁵ achieved excellent results with the use of radial forearm flap. They prefabricated the flap by tubing it on itself at the arm before transferring it to the neck, as described by Harii et al.¹⁸ However, a 2-layer closure was used to minimize leakage. All the flaps were successfully transferred, and donor site morbidity was minor. Ten of the 12 patients have had successful restoration of both swallowing ability and voice. Leakage occurred in 8 patients, but was closed with nonsurgical treatment in 5. Only 1 case of distal stricture developed. It was successfully revised with anastomotic Z-plasty. Owing to the high incidence of leakage, Kelly et al recommend additional coverage over the forearm flap with a nonepithelialized myocutaneous flap in patients who are at high risk for fistulization. They comment that although additional soft tissue coverage will create difficulty in monitoring the viability of the forearm flap, the lack of transcutaneous Doppler monitoring in selected cases should be insignificant owing to the reliability of the forearm flap.

In 2 follow-up reports from the same institution, Anthony et al^8,19 observed that a regular diet was tolerated by 88% of the patients with an intact tongue base and that speech rehabilitation was easily achieved with a tracheoesophageal puncture and a voice prosthesis. Moreover, the quality of speech was nearly equivalent to that of patients who had undergone a total laryngectomy and primary closure of the pharynx after partial-circumferential resection. Although the fistula rate has decreased with increasing experience, it has remained high at 50%. Anthony et al advocate a 2-layer closure to minimize the potential for leakage and a Z-plasty–type closure at the anastomotic site to prevent stricture. They also discuss covering the forearm flap with a pectoralis major muscle flap in patients with a high likelihood of leakage.

It is evident that pharyngoesophageal reconstruction using the forearm flap is equivalent, if not superior, to all other reconstructive methods, including free jejunum transfer. However, the forearm flap has been associated with a high rate of fistula formation. Although a majority of the fistulas present as a transient postoperative salivary leak that resolves with nonsurgical management, they still complicate postoperative management, prolong hospital stay, increase the risk of infection and stricture, and delay postoperative adjuvant therapy and the rehabilitation of swallowing ability and voice. Although the use of a pectoralis major myocutaneous flap as a second-layer cover, as advocated by Anthony et al,^8,19 may diminish fistula formation, it compromises forearm flap monitoring. Also, this approach requires an additional donor site and increases surgical morbidity.

The pectoralis major myocutaneous flap has been the workhorse flap for single-stage pharyngoesophageal reconstruction over the past 20 years. However, donor site morbidity can be significant, especially in female patients. Also, the long-term results remain disappointing. Because of the bulkiness of the flap, tailoring and insetting can be difficult, especially in a circumferential defect that requires the tubalization of the flap. Even if the flap is folded into a U shape by suturing it to a skin-grafted prevertebral fascia or to an intact mucosal strip in a partial-circumferential defect, the anastomosis is still difficult to perform. Furthermore, the skin paddle of the flap can be unreliable.²⁶ The pectoralis major myocutaneous flap reconstruction is associated with a high rate of stricture (34%) and fistula (54%) formation,²⁷ as well as poor swallow function.²⁸

The lateral arm²⁹ and thigh flaps³⁰ are both fasciocutaneous free flaps that offer abundant pliable tissue that is suitable for pharyngoesophageal reconstruction. Also, these flaps are associated with minimal donor site morbidity. However, they take longer to harvest than the forearm flaps and are technically more difficult to work with.^29,30 They can also be quite thick, especially in obese patients,^29,30 and thus may be less ideal than the forearm flap in some patients.

The radial and ulnar forearm flaps are technically easy to harvest, and both offer superior quality vessels with thin, pliable skin that is well suited for pharyngoesophageal reconstruction. The main disadvantage of the radial forearm flap is donor site morbidity. Skin grafting over the exposed flexor tendons is frequently unsuccessful, resulting in prolonged wound healing problems. The reported rate of this complication ranges from 19% to 33%.^31-33 Although the incidence of tendon exposure can be reduced by suturing the adjacent muscles over the exposed tendons or by preserving the paratenon over the tendon, the failure of the skin graft, with resulting tendon exposure, remains a problem. The ulnar forearm donor site can also have associated tendon exposure problems, especially when a very large flap is required, as seen in Figure 2. However, when a smaller ulnar flap is needed, the flap can be limited proximally toward the ulna without exposing the flexor carpi radialis and brachioradialis tendons, thus reducing the potential of wound healing problems. Indeed, the repair of a radial forearm defect with the use of an ulnar forearm flap has been advocated.³⁴

The vascularity of the hand is dependent on the deep palmar arch arising from the radial artery as well as on the superficial palmar arch arising from the ulnar artery. Communication between the deep and superficial systems is crucial in maintaining the vascularity of the hand following sacrifice of either the radial or the ulnar artery. Therefore, a positive Allen test result is crucial prior to flap harvest. It has been suggested that one of the disadvantages of the radial forearm flap is the sacrifice of the radial artery, which is the major vascular supply to the hand.¹⁷ Ischemic complication following radial forearm flap harvest despite a positive Allen test result, although extremely rare, has been reported.³⁵ Therefore, the ulnar forearm flap may be advantageous in minimizing the potential vascular compromise of the hand. Nevertheless, a preoperative Allen test ensuring the communication between the deep and the superficial systems, as well as intraoperative assessment of the adequacy of the circulation to the digits using Doppler technique while occluding the ulnar artery, should be mandatory to avoid the devastating complication of hand ischemia from either the radial or the ulnar forearm harvest.

The appearance of the donor site of the radial forearm flap, although perhaps not a major concern of male patients, is an important issue for female patients.³⁶ The ulnar flap donor site is more esthetically acceptable, since it is more proximal along the ulnar border and therefore less noticeable during regular daily activity.

In our series, 19 (95%) of 20 ulnar forearm flaps were successfully transferred. Speech rehabilitation was achieved in all patients with tracheoesophageal puncture or esophageal speech or with the aide of an electrolarynx device. The swallow function was restored in all 18 patients who have an intact tongue base. There was 1 perioperative mortality. The donor site morbidity was minor: a small area of skin graft loss occurred in 2 patients (10%). No tendon exposure was encountered. The fistula rate was low (15%), and stricture developed in only 1 case.

We believe that the use of a generous forearm flap with an extra-long venous pedicle offers several advantages in the reconstruction of the pharyngoesophagus. It allows customized resurfacing of virtually any defect and provides abundant vascularized fascia to achieve a double-layer closure to minimize leakage. The redundant portion of the flap can be deepithelialized and used to fill dead spaces and cover cervical vessels. It can be used as skin cover when skin flaps are deficient. It can also be used to monitor flap viability. Although this technique may have increased donor site morbidity owing to additional surgical dissection with a large donor site defect, we did not encounter any significant problem in our series. The cosmetic appearance has also been quite acceptable.

Accepted for publication April 23, 1998.

Presented at the 39th Annual Meeting of the American Society for Head and Neck Surgery, Scottsdale, Ariz, May 13, 1997.

Corresponding author: Kasey Li, DDS, MD, 750 Welch Rd, Suite 317, Palo Alto, CA 94304.