Difficulties of nasocutaneous fistula repair following lacrimal outflow system malignancy resection

Abstract

Purpose:

To evaluate the incidence of nasocutaneous fistula (NCF) development following en bloc resection of lacrimal outflow system malignancies (LOSM) and describe methods of surgical repair

Methods:

Retrospective review of all patients who underwent resection of LOSM with reconstruction and post-treatment protocol at the University of Miami between 1997 and 2021.

Results:

Of 23 included patients, 10 (43%) developed postoperative NCF. All NCFs developed within one year of surgical resection or completion of radiation therapy. NCF were seen more frequently in patients who underwent adjuvant radiation therapy and those who had reconstruction of the orbital wall with titanium implants. All patients underwent at least one revisional surgery to close the NCF, including local flap transposition (9/10), paramedian forehead flap (5/10), pericranial flap (1/10), nasoseptal flap (2/10) and microvascular free flap (1/10). Local tissue transfer, pericranial, paramedian, and nasoseptal forehead flaps failed in most cases. Two patients had long-term closure; one patient who underwent a paramedian flap and a second who underwent a radial forearm free flap, suggesting that well-vascularized flaps may be the most viable option for repair.

Conclusions:

NCF are a known complication following en bloc resection of lacrimal outflow system malignancies LOSM. Risk factors for formation may include adjuvant radiation therapy and use of titanium implants for reconstruction. Surgeons should consider utilizing robust vascular pedicled flaps or microvascular free flaps for repair of NCF in this clinical scenario.

Introduction

Lacrimal outflow system malignancies (LOSM) are rare malignancies of the nasolacrimal system. This histologically diverse group of malignancies arises from the various cell types of the lacrimal outflow system (LOS), including squamous and columnar epithelial cells, melanocytes, lymphocytes, mesenchymal cells, and goblet cells. Epithelial tumors, including squamous and transitional cell carcinoma, are the most common, followed by mucoepidermoid, adenocarcinoma, and adenoid cystic carcinoma.¹ The symptoms of LOSM are similar to benign disease of the LOS, with the most frequent symptoms being epiphora, a palpable mass, and/or recurrent swelling.¹

Surgical resection plays an important role in the treatment of these malignancies. En bloc resection of the tumor, the nasolacrimal system, and the underlying periosteum is currently the favored surgical treatment method for tumors confined to the LOS.^1–4 In addition to surgical excision, those with tumors with aggressive cellular features or positive margins often undergo adjuvant radiotherapy which may lead to aberrant wound healing.^1,3,5,6

Nasocutaneous fistulas (NCFs) are a known complication following en bloc resection of LOSM.^4,7,8 These fistulas lead to multiple functional and cosmetic problems, including significant pain, wound breakdown, constant discharge, or poor cosmesis.⁹ Although NCF is encountered after LOSM resection, few studies describe the incidence and management of this challenging complication. In this study we evaluate the incidence and risk factors of NCF development and describe attempted methods and outcomes of surgical repair of NCF in patients following resection of LOSM.

Methods

The study protocol was approved by the University of Miami, Miller School of Medicine Institutional Review Board. The study was conducted in accordance with the provisions of the Declaration of Helsinki. Written consent was obtained from selected patients to publish clinical photographs.

A retrospective chart review analyzed the clinical records of the 24 patients with LOSM from 1997 to 2021 treated with surgical excision. The information collected included sex, age at the time of initial surgery, LOSM pathology, resection method, radiation exposure, post-surgical fistula formation, timing of fistula formation, and reconstruction method. One patient was excluded from analysis as the primary tumor was nasal adenoid cystic carcinoma initially treated with endoscopic resection, with subsequent secondary involvement of the lacrimal outflow system.

Results

23 patients with primary LOSM were included in this study. All patients underwent en bloc resection of LOSM under the care of both oculoplastic surgery and head and neck surgery. 14 (61%) of patients were male, and 9 (39%) were female. The average age at the time of initial surgery was 60 years old (range 33-88). The most common indication for surgery was squamous cell carcinoma (57%), followed by transitional cell carcinoma (26%), adenoid cystic carcinoma (9%), mucoepidermoid carcinoma (4%), and oncocytoma (4%). 15 (65%) of the 23 patients underwent adjuvant radiation therapy, with an average dose of 62.3 Gray (SD 5.8 Gray). The average follow-up length for all patients was 62 months (range 3-178 months). 20 (87%) of patients had reconstruction of the medial wall and medial floor of the orbit with a tailored contoured titanium implants to support the globe and eyelid.

Of the 23 patients, 10 (43%) developed postoperative NCFs. All NCF developed within one year of surgical resection or completion of radiation therapy. Most cases of NCF formation (60%) occurred within the first three months postoperatively. Of the 10 patients who developed NCF, one (10%) had undergone surgical resection alone, two (20%) developed NCF before the initiation of radiation therapy, and seven (70%) developed NCF after undergoing adjuvant radiation therapy. The association between fistula formation following adjuvant radiation exposure was not significant (Fisher’s one-sided exact test, p=0.237). Among the patients in the radiation group, the average radiation doses in patients developing fistulas as compared to those who did not develop fistulas did not vary significantly (one-sided t-test p=0.493). 10 of the 20 patients who underwent orbital wall reconstruction with a titanium implant developed NCF, while the three patients who had orbital reconstruction by other means did not develop fistulas, which did not reach statistical significance (Fisher’s one-sided exact test, p=0.161).

All 10 NCF patients underwent at least one revisional surgery to close the fistula. Surgical methods included: undermining or local flap transposition (9/10), paramedian forehead flap (5/10), pericranial flap (1/10), nasoseptal flap (2/10), and microvascular free flap (1/10), with some patients undergoing combined procedures, or multiple attempts at closure. Two patients showed promising results. One patient experienced a failed local transposition flap followed by a failed paramedian forehead flap for a medial canthal NCF. He ultimately underwent a radial forearm free flap, which was successful for two years until tissue breakdown occurred at the nasal side wall. Notably, the original medial canthal NCF remained closed without recurrence. Another patient underwent a failed glabellar flap, followed by a successful paramedian flap, with sustained closure of the NCF 11 years postoperatively.

Discussion

While no individual outcome was found to be significantly associated with fistula formation in the present study, there are many factors that may have contributed to the rate of NCF in this cohort that our study was underpowered to detect. Firstly, most patients who developed NCF underwent adjuvant radiation therapy (9/10), suggesting that post-operative irradiation may have compromised healing of healthy skin and integration of reconstructive flaps. The detrimental effect of radiation on surgical beds is well described, and thus likely contributed to flap breakdown and fistula formation.⁶ Another factor that may contribute to high rates of fistula formation is the use of titanium implants for the orbital wall reconstruction. Half of the patients who underwent reconstruction with titanium implants developed NCF, while the three patients who had orbital reconstruction by other means did not develop fistulas. Surgeons should be cautious when the use of titanium implants is combined with high dose radiation therapy; in animal models, radiation has been shown to reduce osteointegration and implant stability.¹⁰ In fact, the authors’ preferred method of reconstruction in these patients no longer involves placement of titanium implants, and has trended towards use of resorbable implants, especially when adjuvant radiation is planned.

In the current study, NCF development occurred in 43% of patients undergoing LOSM resection, all of whom underwent medial maxillectomy with resection of the nasolacrimal duct, medial canthus, and lacrimal puncta, and frequently the involved skin from prior external dacryocystorhinostomy surgery. The previously reported incidence of NCF development varies between 11% to 15% of all cases of LOSM undergoing primary surgical resection.^8,11 In an extensive series of 99 patients undergoing sinonasal cancer resection by Cianchetti et al., 15% of patients undergoing trans facial resection developed sinonasal-cutaneous fistula.¹¹ In addition to LOSM, this large series also included patients with nasal cavity, ethmoid sinus, sphenoid, and maxillary sinus malignancies; thus, it is difficult to compare rates of fistula formation directly. Notably, the study excluded fistulas from tumor recurrence/progression and patients who developed fistulas before radiation therapy, while our study included all fistulas at any post-operative state. Additionally, it is unclear how fistulas were graded in their study, which included “grade 3 fistulas or higher”, and thus may further underestimate the true incidence of NCF formation. If we apply the identical exclusion criteria, the fistula rate in our series drops to a similar incidence of 23%. Furthermore, despite including various tumors of the maxilla, ethmoid, sphenoid, and nasal cavity, 7/8 fistulas in their study occurred at the medial canthus. This highlights the increased risk in reconstructing this area and suggests that the rate of fistula formation may be higher than 15% for LOSM alone. Finally, their study found that squamous cell carcinoma histology had a 5.7 times higher likelihood of fistula formation than non-squamous cell carcinoma histology. Squamous cell carcinoma represented 27% of their patient population, but represents 57% of our patient population, thus possibly contributing to the higher NCF incidence rate in our study.

In another study of 14 patients with lacrimal sac malignancies, nine underwent en bloc resection and reconstruction, while the remaining five underwent exenteration or radiation therapy alone.⁸ One of the nine patients developed an NCF. Notably, in four of the nine patients, the surgical technique included a free flap to primarily reconstruct the site, differing from the repair method in our cohort.⁴ The placement of a highly vascularized free flap to repair a myocutaneous defect instead of local advancement flaps may explain the lower rates of NCF development and further supports the success of our patient who underwent free flap NCF repair.

While previous studies have touched on the incidence of NCF in this patient population, few studies describe the means of repair once the complication has occurred.⁴ In our study, all 10 fistula patients underwent secondary repair utilizing various surgical techniques. Attempted methods for repair included myocutaneous advancement, glabellar flaps, pericranial flaps, paramedian flaps, nasoseptal flaps, and a radial forearm free flap. All attempts at reconstruction failed within one year, except in two patients, which yielded longer-term success in our cohort. In the patient who underwent a radial forearm free flap, the medial canthal fistula remained successfully closed. However, two years postoperatively, a new cutaneous fistula developed at the nasal edge of the defect. In another patient who initially underwent failed glabellar flap closure of an NCF, a subsequent paramedian flap remained successful 11 years after closure. Our findings are similar to the Cianchetti et al. study, where the authors used various techniques to repair sino-nasal fistulas, including four-layer closure using paramedian forehead flaps, pedicled glabellar flaps, rectus abdominus flaps, radial forearm flaps, and skin grafts, with varying degrees of success.¹¹ The repair method with the highest success in the Cianchetti cohort was free flap repair, representing three of their four successful closures. Only one of the three glabellar flaps in their cohort demonstrated successful closure, which is consistent with our observations of recurrent NCF following local tissue flaps. Nelson et al. describe four cases of successful closure of NCF in their patients with a combined paramedian with extended galealfrontalis/pericranial flap.¹² In their series, one patient had the NCF develop in the setting of LOSM malignancy with four months of postoperative follow-up recorded; the other three NCFs were in patients with sinonasal carcinomas in the ethmoid and frontal sinuses. Their technique differs from our paramedian forehead flaps in that an extended galealfrontalis and pericranial flap is excised and secured to the subcutaneous tissue. The base is tunneled subperiosteally without the need for secondary pedicle transection. In our series, only one of five paramedian forehead flaps resulted in the successful closure of the NCF. In all five of our cases, the pedicle was secondarily transected and inset into the defect. The Nelson et al. technique creates and maintains a robust vascular supply to the NCF and should be considered in cases where a microvascular free flap is inappropriate.

This series highlights our department’s experience with NCF in patients undergoing surgical resection of LOSM. Given the need for high-dose adjuvant chemoradiation in many of these patients, both primary and NCF-closure reconstruction techniques should focus on ensuring a robust vascular supply to prevent postoperative tissue breakdown. Though the optimal method of fistula closure remains unclear, surgeons should be wary of closure using advancement or transpositional flaps, with consideration for repairs utilizing robust vascular pedicled flaps or microvascular free flaps.

Figure 1.

A) Contrast-enhanced computerized tomography (CT) scan in patient with small right sided nasocutaneous fistula following reconstruction of medial maxillectomy defect with titanium mesh. B) Contrast-enhanced CT scan in patient with large left sided nasocutaneous fistula following reconstruction of medial and inferior orbital wall and medial canthus with titanium plate. Attention is drawn in both cases to the relative lack of soft tissue overlying and underlying the titanium plate, placing the patient at risk of fistula expansion or fistula-closure failure if local advancement flaps are performed.

Table 1.

Baseline characteristics of patients, tumors, and rates of fistula formation.

Baseline Characteristics
Sex (N (%))
Female	9 (39%)
Male	14 (61%)
Age (years, mean (SD))	60 (15)
Histopathologic Diagnosis (N (%))
Squamous Cell Carcinoma	13 (57%)
Transitional Cell Carcinoma	6 (26%)
Adenoid Cystic Carcinoma	2 (9%)
Mucoepidermoid Carcinoma	1 (4%)
Oncocytoma	1 (4%)
Fistula Formation	Fistula (N=10)	No fistula (N=13)
Radiation
No radiation (N=8)	1	7
Radiation (N=15)	9	6
Titanium Mesh
No titanium mesh (N=3)	0	3
Titanium mesh (N=20)	10	10