The Endoscopic Endonasal Approach for Sinonasal and Nasopharyngeal Adenoid Cystic Carcinoma.

Abstract

Objective:

To determine factors affecting outcomes for patients with sinonasal and nasopharyngeal adenoid cystic carcinoma (SNACC) treated using the endoscopic endonasal approach (EEA) with preservation of key structures followed by adjuvant radiotherapy (RT).

Method:

Retrospective case series of 30 patients treated at the University of Pittsburgh between 2000–2014. Hospital records were reviewed for clinical and pathologic data. Outcome measures included overall survival (OS), disease-free survival (DFS), local recurrence-free survival (LRFS) and distant metastasis-free survival (DMFS) rates.

Results:

The majority of patients had T4a and T4b disease (23.3%, and 63.3%). Microscopically positive margins were present in 21 patients (63.6%). Positive margins were present in nine patients (30.0%). The mean and median follow-up were 3.97 and 3.29 years. Five-year OS, DFS, LRFS, and DMFS were 62.66%, 58.45%, 87.54% and 65.26%. High-/intermediate-grade tumors had worse DFS, p=0.023, and LRFS, p=0.026, (HR=4.837, 95%CI=1.181–19.812). No factors were associated with significantly worse DMFS. No patient suffered CSF leak, optic nerve, or internal carotid injury. The mean and median length of hospital stay was 4.1 days and 2.0 days (range: 0 – 32 days).

Conclusion:

Organ-preserving EEA with adjuvant RT for low-grade, SNACC offers 5-year survival similar to that reported by other studies, which include radical, open skull base surgery. Patients with high-grade disease do poorly and may benefit from novel treatment strategies. For low-grade disease, organ-preserving EEA with RT may be the best option, offering a balance of survival, quality of life, and decreased morbidity for patients with this difficult to cure disease.

Introduction

Sinonasal adenoid cystic carcinoma (ACC) accounts for 10–25% of head and neck ACC and arises from the minor salivary glands lining the aerodigestive tract.^1,2 Sinonasal ACC represents only 6.2% of all sinonasal malignancies in a recent SEER-based study.³ Nasopharyngeal ACC is even more rare and accounts for 4.3% of cases.⁴ 1979, Spiro et al. first suggested that sinonasal ACC has a worse prognosis than ACC of other sites.⁵ Since then, studies show that sinonasal and nasopharyngeal ACC (SNACC) are characterized by advanced T-stage at presentation, skull base invasion, frequent perineural invasion (PNI), and positive margins.^2,6–8

The literature to date is limited to small case series of SNACC, offering little data to guide treatment.^4,7,9–11 Previous groups have attempted to identify factors influencing overall survival (OS), disease-free survival (DFS), and local recurrence (LR) with conflicting results. Recently, however, a meta-analysis studying sinonasal ACC showed that positive margins and tumors involving the ethmoid and sphenoid sinuses had worse disease-specific survival (DSS), (p<0.001); while PNI and adjuvant therapy had no impact.¹² Other studies suggest that PNI and adjuvant radiotherapy affect LR.^8,13 The impact of PNI and adjuvant therapy on DSS remains unclear; because, DSS may be driven by local and/or distant recurrence.^14,15

Current treatment paradigms maintain that aggressive surgical resection with negative margins and adjuvant radiotherapy offers the best survival outcomes; however, this treatment philosophy for SNACC requires morbid facial incisions, craniotomy, removal of dura, cavernous sinus resection, and resection of cranial nerves.^13,16 Such aggressive, open skull base approaches are also associated with significant brain retraction, encephalomalacia, aseptic meningitis, regular need of microvascular free flap (MVFF) reconstruction, cerebrospinal fluid (CSF) leak and perioperative CSF diversion.^10,11,16 Orbital exenteration, if required, causes significant psychological effects and functional deficits (monocular vision and lack of depth perception). Even with aggressive, open surgery, negative margins are only achieved 50% of the time, and has yielded little to no improvement in OS or DFS since the 1960s.^{2,5,10,16–18}

Organ-preservation therapy (OPT) became popular for the oropharyngeal, laryngeal, and hypopharyngeal squamous cell carcinomas in the 1990s. Mendenhall et al. studied OPT for ACC, and found that primary surgery with adjuvant radiation (RT) improved cause-specific survival, local control, and distant control, thus establishing the current treatment standard.¹³ In this study, we report our experience with the endoscopic endonasal approach (EEA) followed by RT as organ preservation surgery (OPS) for SNACC and advocate for its use when treating these rare tumors.^5,7,14,17,19

Materials and Methods:

Surgical Approach

The EEA has been well described, with advantages including: improved visualization, less neural manipulation, no facial incisions, reduced length of hospital stay (LOHS), and shortened time to initiation of adjuvant therapy. Our treatment paradigm for both sinonasal and nasopharyngeal ACC is the same and attempts gross tumor removal with endoscopi “organ preservation surgery,” that is, preservation of the orbital contents, optic nerves, carotid arteries, and motor cranial nerves. OPS includes removal of the pterygopalatine fossa contents, involved bone of the paranasal sinuses and sphenoid bone, involved sensory nerves (the Vidian, descending palatine, and infraorbital nerves), the nasopharynx, and cartilaginous eustachian tube.

Data Collection

Institutional review board approval was obtained and a retrospective chart review of all SNACC treated primarily using the EEA at the University of Pittsburgh Medical Center between 2000–2016 was conducted. Those patients undergoing, open skull base surgery, definitive nonsurgical therapy, or those presenting with inoperable distant metastases at diagnosis were excluded. T-stage was defined according to the American Joint Committee on Cancer Classification and preoperative imaging was reviewed by authors (R.K. and M.T.) for determination of T-stage. We defined three surgical scenarios: (1) complete resection with negative histologic margins (2) gross total resection (GTR) with microscopically positive margins; and (3) subtotal resection (STR) with grossly positive margins. Complete resection was as defined by negative margins read in the pathology report. Achievement of GTR and STR were determined based on operative report findings. GTR was achieved when there was no apparent residual tumor after removal of mucosa or bone beyond the tumor. STR was achieved when gross tumor was identified invading the cavernous sinus, bone, dura, the optic nerves, or carotids. Final pathology reports were reviewed for adverse features known to impact prognosis including: T-stage, PNI, angiolymphatic invasion (ALI), and grade. “Resectable disease” was defined as T2, T3, or T4a disease, while “unresectable disease” was defined as T4b per guidelines (NCCN Guidelines v2.2018 – Salivary Gland Tumors).²⁰ Subsites abutting the skull base (ethmoid, sphenoid, and nasopharynx) were considered high-risk based on previous studies.^4,7,12 Cases with missing data were censored.

Statistical Analysis

SPSS version 24 (SPSS, Inc, an IBM Company, Chicago, Illinois) was used for statistical analysis. Significance thresholds were set at 0.05. Demographic and pathologic data were summarized with descriptive statistics. Outcome measures included OS, DFS, local-recurrence-free survival (LRFS), and distant-metastasis-free survival (DMFS) rates. The chi-square test and Cox-proportional hazard ratios were used to compare differences between groups and to determine significance of adverse features on survival. OS rates were calculated from the date of surgery until the date of death by other causes or disease using the Kaplan-Meier method and statistical significance determined by log-rank test. Survival was then stratified by “resectability,” margin status (GTR/microscopically positive vs. STR/grossly positive), PNI, ALI, histologic grade, and recurrence. DFS, LRFS, and DMFS were calculated using the competing-risk method with death from other causes as a competing risk. Univariate hazard ratios (HR) and 95% confidence intervals (95% CI) for recurrence were calculated by competing-risks regression method. Multivariate analysis could not be performed given the small number of patients and events.

Results:

Thirty patients fulfilled the inclusion and exclusion criteria. See Figure 1. Five patients (20%) underwent EEA as well as maxillectomy approaches for cancers with hard palate extension. The endonasal approach was supplemented with additional approaches in 5 patients (Caldwell-Luc – 3, Denker – 2). The surgical site was reconstructed in 17 patients. A nasoseptal pedicle flap was used in 8 patients. Abdominal fat, free mucosal grafts, and split thickness skin grafts were used in the remainder of the cases. Two patients underwent OPS as salvage treatment. One very young patient underwent EEA with curative intent after surgical excision of a bone metastasis diagnosed his ACC. He then underwent adjuvant radiation to both the primary and the bone metastasis. The mean and median ages in our patient population at the time of diagnosis were 53.00 years and 54.50 years. The mean and median follow-up times were 3.97 and 3.29 years.

Figure 1:

Flow-chart describing study population.

The distribution of tumors based on “resectability,” subsite risk, margin status, PNI, ALI, and grade is presented in Table 1. The majority of cases were either T4a or T4b (86.6%). 44.8% of the tumors were considered “traditionally resectable” (T2, T3, T4a), while the other half was considered “traditionally unresectable” (T4b). The most common tumor subsites were the following: nasal cavity, 11 patients (36.67%); maxillary sinus, nine patients (30.0%); and nasopharynx, five patients (16.67%). With respect to subsite risk, 19 (63.33%) patients were low-risk and 11 (36.67%) were high-risk. 19 patients (63.6%) had GTR/microscopically positive margins, while nine (30.0%) had STR/positive margins. Only two patients had negative margins. Twenty-three patients (76.67%) had PNI, while 2 patients (6.67%) had no PNI. ALI data was missing/not assessed in 18 pathology reports; however, it was present in 60.00% of tumors for which there was data. 22 (73.33%) patients had low-grade disease, while eight (26.67.1%) had intermediate/high-grade disease. The small numbers and few events in patients with negative margins did not yield reliable data to analysis the effect on outcomes in this subgroup.

Table 1:

Demographic and Clinicopathologic Data.

Parameter	Classification	N (Valid %)
Margin Status (N=30)	Positive (grossly)	10 (30)
	Microscopically Positive	19 (63.3)
	Negative	2 (6.7)
Histologic Grade (N=30)	Low	22 (73.3)
	Intermediate/High	8 (26.7)
Subsite (N=29)	Maxillary Sinus	9 (31.0)
	Nasal Cavity	11 (37.9)
	Ethmoid	3 (10.3)
	Sphenoid	1 (3.4)
	Nasopharynx	5 (17.2)
Subsite Risk (N=29)	Low	20 (69.0)
	High	9 (31.0)
T-stage (N=30)	T1	0 (0)
	T2	1 (3.3)
	T3	3 (10.0)
	T4a	7 (23.3)
	T4b	19 (63.3)
Resectability (N=30)	Resectable	11 (36.7)
	Unresectable	19 (63.3)

Mean and median OS in our study population were 3.97 and 3.29 years, which was the same as our follow-up. OS rates at 2, 5, and 10 years were 85.23%, 62.66%, and 54.82%. DFS rates at 2, 5, and 10 years were 73.84%, 58.45%, and 12.63%. Local recurrence rates at 2, 5, and 10 years were 7.70%, 12.46%, and 36.89%. Distant recurrence rates at 2, 5, and 10 years were 18.96%, 34.74%, and 69.53%. Patterns of recurrence are shown in Figure 2. Metastases were most commonly found in the lung only (45.5%), followed by the lung and other sites (27.3%), the liver only (9.1%) and bone only (9.1%). Of those patients developing distant metastases, 72.72% (8/11 metastases) appeared within five years of follow-up. The median time to death in patients dying of local disease was 0.58 years, while the median time to death of those dying of distant disease was 2.36 years.

Figure 2:

Patterns of Treatment Failure.

The effect of relapse on OS was examined using Cox regression analysis trended toward worse OS, p=0.076, (HR 4.545, 95%CI=2.118–7.590). See Figure 3. PNI and ALI significantly affected OS in univariate analysis (p=0.000 and p=0.035). Other adverse features such as “traditionally unresectable” disease, positive margin status, and high-risk subsite did not affect OS.

Figure 3.

Kaplan-Meier curve for overall survival stratified by relapse. Patients with relapse had significantly worse overall survival, (p=0.002, HR 30.705, 95%CI [3.593–262.419]).

Patients with low-grade tumors had significantly better DFS compared to intermediate/high-grade tumors, p=0.023 by log rank test. See Figure 4. PNI, ALI, “resectability,” margin status, and subsite risk did not significant affect DFS. We observed worse LRFS for patients with intermediate/high-grade tumors, p=0.026, (HR=4.837, 95%CI=1.181–19.812). See Figure 5. The 2-, 5-, and 10-year LRFS for low-grade tumors was 100%, 93.23%, and 73.07%; while the 2-, 5-, and 10-year LRFS for intermediate/high-grade tumors were 71.43%, 71.43%, and 50.00%. Of the 66.67% who underwent GTR (microscopically positive margins), the LR rate (LRR) was 20.0% (4/20 patients). Of the 30.0% of patients who underwent STR (positive margins), the LRR was 11.1% (1/9 patients). This difference was not significant (p=0.64, Fisher’s exact test). Competing-risk analysis of DMFS for subgroups with ALI, PNI, unresectable disease, high-risk subsites, and margin status showed equivalent DMFS.

Figure 4:

Kaplan-Meier curve for disease-free survival stratified by grade. High-intermediate grade had worse disease-free survival compared to those with low-grade histology (log rank, p=0.013). Note there are very few censored cases in the low-grade subgroup, which may make findings unreliable.

Figure 5:

Kaplan-Meier curve with competing-risk analysis for local recurrence-free survival in patients with low-grade vs. high- and intermediate-grade disease. Patients with high- and intermediate-grade disease had significantly worse local recurrence-free survival (p=0.026, HR 4.837, 95%CI= [1.181–19.812]).

Operative complications suffered during OPS were limited. Two CSF leaks were identified intraoperatively and successfully treated without CSF diversion. No patient suffered postoperative CSF leak, carotid injury, or optic nerve injury. Our medical complications were seen in one patient who developed a lower extremity venous thrombosis as well as hospital-acquired pneumonia and in a second patient with buprenorphine withdrawal. The mean and median LOHS was 4.1 days and 2.0 days, (range: 0 – 32 days).

We recommended that all patients undergo adjuvant RT, which was given in various forms. External beam radiation was given postoperatively to a dose > 5000 cGy in most patients (83.33%). Proton beam radiation was delivered to two patients (6.67%). Gamma knife or cyberknife boosts were given to two patients (6.67%). One patient could not tolerate the full adjuvant dose. One patient refused radiation. Four were missing data on adjuvant radiotherapy. We did not recommend chemotherapy; however, nine patients (30%) underwent adjuvant chemotherapy (various agents) at the medical oncologist’s discretion.

Discussion:

Surgery with adjuvant radiotherapy for SNACC is widely accepted as the standard of care, due to the presence of positive margins, PNI, local recurrence, and skull base invasion.^{8,13,16,18,21,22} Rhee et al. reported their experience on 35 patients with sinonasal ACC treated with radiation alone or surgery (medial maxillectomy, subtotal maxillectomy, craniofacial resection, or orbital exenteration) with adjuvant radiotherapy.¹¹ Their 5-year OS, DFS, local recurrence and distant recurrence rates were 86%, 51%, 30% and 25%. They also found that 5-year DSS after development of distant metastasis was 17% with median survival of 25 months. Pitman et al. reported on a series of 35 patients treated with open craniofacial resection and adjuvant radiotherapy for treatment of ACC of sinonasal tract and could not identify any factors predicting survival.¹⁰ This cohort had a positive margin rate of 46%, but reported a DFS of 46% at median follow-up of 40 months, as well as local recurrence and distant recurrence rates of 36% and 21%. A recent meta-analysis on outcomes for ACC of the nasal cavity and paranasal sinuses found 5-year OS, DFS, LR and distant recurrence rates of 62%, 43%, 36.6%, and 29.1%.¹² Finally, Ramakrishna et al. reported on the efficacy of aggressive open skull base surgery for treatment of ACC and found that maximal resection with negative margins was associated with improved OS, but 5-year OS, PFS, and LRFS rates were 78%, 46%, and 18%.¹⁶ Margin-negative resection (achieved in only 49% of cases) had no impact on PFS due to distant failures associated with advanced disease. In spite of incremental improvements in local control associated with these approaches, the DFS and DMFS rates have not changed in 20 years.^2,23,24

Recently, EEA has gained popularity and shows equivalent oncologic outcomes when used for the treatment of sinonasal malignancy in appropriately selected patients.^25–27 However, the literature focusing on the endoscopic treatment of sinonasal and nasopharyngeal ACC is limited. Our single-institution study represents the largest series of SNACC treated by the EEA with a mean follow-up of 4 years. Our findings also suggest that SNACC have comparable outcomes when compared to other sites treated with surgery and adjuvant radiotherapy.⁷ Furthermore, our outcomes are similar in terms of OS, DFS, LRFS, and DMFS to previous groups treating SNACC with aggressive, open skull base surgery and adjuvant radiotherapy. Lastly, LRFS was not different between those patients undergoing GTR vs. STR. This supports the use of OPS, which assumes the possibility of positive margins in instances where tumor invades the dura, cavernous sinus, internal carotid arteries, or optic nerves.

Mendenhall et al. studied the optimal treatment for ACC (definitive RT vs. surgery with adjuvant RT) in a cohort of 101 patients with previously untreated disease and found that surgery with adjuvant RT yielded significantly better local control, p=0.010.¹³ In that study, treatment group did not, however, affect OS or cause-specific survival, which was thought to be driven by distant metastasis. Patients with ACC of the skull base (including sinonasal and nasopharyngeal sites) have advanced disease and their survival is dictated by distant metastasis, which often presents within three years.¹⁴ This may be the reason we found no significant difference in OS, DFS, and DMFS associated with microscopically positive margins (GTR) with positive margins (STR). It may also be the reason why our OPS approach achieved OS, DFS, LRFS, and DMFS, comparable to radical surgical approaches for SNACC.^{10–12,16,24,28} One finding of note in our study is that patients with intermediate/high-grade disease had significantly worse DFS and LRFS and so may benefit from more aggressive adjuvant therapies.

Limitations of our study include: the small sample size, retrospective nature, short follow-up, and inability to determine the effect of negative margins and adjuvant treatment on outcomes. With 30 patients, this is the largest series of patients undergoing the EEA for treatment of SN ACC. It also represents one of the largest series of SNACC in the literature. With respect to our limited follow-up, most studies offer only five-year survival rates; and, most patients who recur will do so within three to five years, according to the literature.^10,14,16 In our study, the mean follow-up was 3.97 years and 72.72% (8/11 metastases) of distant metastases appeared within five years. For this reason, we believe this study yields meaningful results. Another limitation of our study and approach is that negative margins were only achieved in two patients, and so the effect of negative margins could not be studied. However, our approach assumes that traditional margin-negative resection for SNAAC will rarely be achieved. Finally, this study cannot clarify the role of adjuvant radiation for SNACC, because all patients were deemed high-risk and received radiotherapy as a standard of care. The role of chemotherapy in the treatment of SNACC, given in 30% of our patients, remains undefined.^12,16

Previous authors have argued for the use of maximally safe, conservation surgery to treat this disease with the least morbidity.^16,19 Open approaches have been associated with postoperative pneumocephalus, intracranial hematoma, pituitary insufficiency, facial nerve paralysis, CSF leak, dystopia, enophthalmos, oroantral fistula, and malocclusion.^15,16,18 Our major surgical complications were limited to two CSF leaks, which were identified and treated without postoperative CSF diversion. Five patients developed chronic unilateral serous otitis media with effusion; however, this is a common side effect in the treatment of SNACC near the Eustachian tube. The average LOHS for aggressive open surgery is a week either do to neurological or MVFF postoperative care.¹⁶ Our mean LOHS was 4.1 days, which was skewed by the one patient undergoing radical maxillectomy with MVFF reconstruction. Our median LOHS, however, was 2.0 days.

OPS with EEA provides an alternative to open approaches in the treatment of SNACC that are associated with increased morbidity, facial deformity, encephalomalacia, and increased LOHS. We believe that the current treatment algorithm for SNACC should include OPS as a mainstay of treatment until better adjuvant therapies are developed for the control of distant disease. We have shown that patients with low-grade disease do well. Patients with intermediate/high-grade disease do poorly and would benefit from participation in clinical trials. For all of the previously discussed reasons, we propose a new treatment algorithm for SNACC in order to minimize morbidity and to preserve quality of life when treating this indolent, difficult to cure disease whose outcomes are drive by distant metastases and the rare ability to achieve negative margins.^14,16 See Figure 6.

Figure 6:

Proposed Treatment Algorithm for ACC of the Skull Base. OPS = Organ Preservation Surgery.

Currently, many groups are looking at the role of systemic adjuvant therapies for the treatment of advanced or metastatic adenoid cystic carcinoma. RTOG 1008 is examining the role of adjuvant chemotherapy in high-risk minor salivary gland tumors ( NCT01220583).²⁹ Outside of a trial, cisplatin and paclitaxel are not recommended given their lack of impact and toxicity profiles.³⁰ There are also groups looking for molecular targets and agents in the treatment of advanced-stage, progressive, or metastatic ACC.^31,32 A recently completed trial is examining the role of sunitinib for the treatment of metastatic ACC ( NCT00886132),³³ while patients with EGFR/erb-2- and c-kit-positive tumors may be candidates for treatment with dasatinib and lapatinib.^31,34 Ongoing trials looking at other novel agents, such pembrolizumab are also underway ( NCT03087019).³⁵ It is our opinion that selecting appropriate patients for these trials should be prioritized by physicians treating this disease.

Conclusion:

EEA and adjuvant RT may be the best treatment option with the least morbidity for patients with SNACC. For low-grade SNACC, EEA with adjuvant RT offers similar survival to that reported by groups using open approaches. Patients with high-grade disease do poorly and may benefit from aggressive open approaches, novel adjuvant therapies and participation in clinical trials. For these reasons, we propose a new treatment algorithm for SNACC.