Keywords esthesioneuroblastoma - craniofacial resection - proton beam radiation - sinonasal
malignancy
Introduction
Esthesioneuroblastoma (ENB) is a relatively uncommon malignant tumor of the nasal
vault believed to originate from olfactory stem cells of neural crest origin.[1 ]
[2 ] Despite being historically labeled as a low-grade malignancy,[3 ] ENB demonstrates a propensity for frequent recurrence following conventional treatment
and a relatively high incidence of regional metastasis.[4 ]
[5 ] Since its initial description in 1924, multiple institutions have developed treatment
strategies that use various combinations of surgery, radiation therapy, and chemotherapy.[1 ]
[2 ]
[6 ]
[7 ]
[8 ]
[9 ]
[10 ]
[11 ]
[12 ]
[13 ]
[14 ] Although no definitive regimen has been established, analysis of this collective
experience demonstrates that multimodality therapy, particularly surgery followed
by radiation, achieves the highest cure rates.[1 ]
In 2008, we published our experience treating ENB in 10 patients with craniofacial
resection and adjuvant proton beam radiation with or without chemotherapy.[15 ] Our findings reinforced the benefits of multimodality treatment and suggested that
proton beam radiation might contribute to excellent outcomes while limiting severe
sequelae of radiation toxicity.[15 ] In the 5 years since this initial publication, our patient population has more than
doubled. Such an increase in cohort size, along with an additional 60 months of follow-up
time, allows for a more robust analysis of treatment outcomes and complications, as
well as offering further insight into the natural history of this unique tumor. As
such, this update reports and evaluates the results of our ongoing experience with
a standardized multimodality treatment protocol.
Materials and Methods
After approval from the institutional review boards at Massachusetts Eye and Ear Infirmary
(MEEI) and Massachusetts General Hospital (MGH), all patients diagnosed with ENB from
January 1997 to January 2013 were identified by a review of records. Each patient
had been evaluated by a multidisciplinary team that included a radiation oncologist,
medical oncologist, otolaryngologist, pathologist, and neurosurgeon.
All patients were evaluated with magnetic resonance imaging (MRI) of the brain and
skull base and a fine-cut computed tomography (CT) scan of the skull base. All medical
records including imaging reports, operative reports, and chemotherapy and radiotherapy
records were reviewed retrospectively. Extracted information included patient demographics,
tumor stage and grade, treatment received, treatment complications, tumor recurrence,
and patient survival.
The Kaplan-Meier method was used to estimate overall survival (OS), with an end point
of death from any cause, and disease-free survival (DFS), with an end point of disease
recurrence. Survival trends were compared by surgical margins status and tumor stage.
Associations with p < 0.05 were considered significant. All analyses were performed using Stata v.12
(StataCorp, College Station, TX, USA).
Results
Patient and Tumor Characteristics
We identified 22 patients who were treated for ENB at MEEI and MGH from 1997 to 2013.
There were 11 men and 11 women in the cohort with a median age of 45.5 years (range:
11–77 years). The average follow-up time for this group was 73 months (range: 24–183
months).
All pathologic specimens were reviewed by the Department of Pathology at MGH. The
diagnosis of ENB was confirmed by a combination of histologic findings and positive
immunohistochemical staining for synaptophysin, chromogranin, and S100.[15 ] Tumors were generally negative for cytokeratins, desmin, vimentin, HMB45, and common
leukocytic antigen.[1 ]
[5 ]
Tumors were staged using the Kadish classification[16 ] and the TNM schema proposed by Dulguerov and Calcaterra on the basis of physical
examination and radiographic findings at the time of initial presentation.[12 ] Ten patients (45.5%) presented with Kadish stage B disease; the remaining 12 (54.5%)
had stage C disease. Using the TNM staging system, 14 patients (63.6%) in the cohort
presented with advanced (T3 or T4) tumors. Three patients (14.3%) had regional metastases
to the cervical lymph nodes at the time of presentation.
Treatment
From 1997 to 2000, the first three patients in the cohort were initially managed nonsurgically
with induction chemotherapy followed by proton beam radiation, as described by Bhattacharyya
and colleagues.[2 ]
[15 ] Since there was no associated disease response in any case with induction chemotherapy,
each patient was subsequently managed with craniofacial resection and adjuvant proton
beam radiation. After 2000, the remaining 19 patients were managed with upfront craniofacial
resection followed by adjuvant proton radiation. Concurrent chemotherapy was also
administered to five patients.
Surgery
All patients underwent endoscopic intranasal biopsies as a diagnostic step. In all
cases, craniofacial resection was performed by a multidisciplinary team composed of
a neurosurgeon and head and neck surgeon. No patient underwent transfacial surgery
alone. Endoscopic resection was initially attempted in one patient, but the procedure
was aborted due to an inability to obtain complete tumor extirpation via a strictly
endoscopic approach. The tumor was subsequently resected en bloc by uncomplicated
craniofacial resection. Positive margins were reported in 9 of 22 patients (40.9%).
These were most commonly present at the periorbita (three patients) or posterior skull
base (six patients). Selective neck dissection was performed in the three patients
who presented with cervical lymph node involvement.
Radiotherapy
All patients were treated with definitive radiotherapy with curative intent. The technique
for planning and delivery of proton beam therapy at MGH has been described previously.[15 ] The median total dose delivered to the primary site was 66.5 cobalt gray equivalent
(CGE) (range: 54–70). On average, patients received 1.85 CGE per fraction over 35
fractions. A total of eight patients (36.4%) received a combination of proton and
photon radiation therapy to the neck bilaterally. In the three patients who presented
with cervical lymph node metastases, the doses of therapeutic proton beam irradiation
of the neck ranged from 60 to 66 Gy. The remaining five patients were treated with
bilateral elective neck irradiation (ENI): 60 CGE of proton beam radiation to the
upper neck and 50 Gy of external beam photons to the lower neck. All patients tolerated
radiotherapy without a treatment break.
Chemotherapy
As mentioned earlier, the first three patients in our series were treated with induction
chemotherapy with no response. All three patients received two cycles of etoposide
and cisplatin every 3 weeks. One of these patients also received an additional two
cycles of carboplatin and etoposide. At the discretion of the medical oncologist,
a total of five patients (22.7%) were managed with concurrent adjuvant chemotherapy
using our standard regimen of cisplatin and etoposide or carboplatin alone.
Outcomes
Local, Regional, and Distant Control
Six patients had disease recurrence an average of 73.4 months after diagnosis with
a range of 13 to 145 months ([Table 1 ]). Patient 1 was treated initially with chemotherapy, followed by craniofacial resection
and proton irradiation. He recurred both regionally and distantly and was managed
with surgical resection of spinal metastases and neck dissection, as well as reirradiation
and chemotherapy. He is currently alive with no evidence of disease (NED) over 4.5
years later. Patient 2 developed distant metastases to the brain and spine that were
managed with palliative chemotherapy. She eventually succumbed to her disease 31 months
after diagnosis. Patient 20 had a recurrence in the scalp and parotid gland 13 months
after diagnosis. He was treated with wide local excision, parotidectomy, and neck
dissection. Twenty-eight months later, he is currently alive with disease. Patient
21 developed recurrent ENB in the neck bilaterally ∼ 2 years after diagnosis. He was
treated with bilateral neck dissections and radiation therapy to the neck and is NED
6 months after treatment.
Table 1
Tumor recurrence, treatment, and outcomes
Patient no.
Location of recurrence
Time to recurrence, mo
Treatment
Final status
1
C-spine; left neck; leptomeninges
82
Rsxn C-spine metastases; left ND; RT + CT
NED
2
Right frontal lobe; thoracic spine
16
Palliative chemotherapy
DOD
3
1. Left parotid
2. Left skull base
61
145
1. Left parotidectomy/ND + RT
2. RT + CT
NED
4
1. Temporal scalp
2. Vertebrae
3. Right neck; ninth rib
4. Sixth rib
72
92
104
124
1. WLE + RT
2. Rsxn + RT + CT
3. Right ND; RT to rib
4. RT to rib
AWD
20
Left scalp; left parotid
13
WLE; left parotidectomy/ND
AWD
21
Bilateral neck
23
Bilateral ND + RT
NED
Abbreviations: AWD, alive with disease; CT, chemotherapy; DOD, died of disease; ND,
neck dissection; NED, no evidence of disease; Rsxn, resection; RT, radiation therapy;
WLE, wide local excision.
Two patients developed multiple recurrences. Patient 3, treated with induction chemotherapy
followed by craniofacial resection and proton irradiation, recurred regionally 61
months after diagnosis and was managed with a parotidectomy and neck dissection and
radiation. At 145 months after diagnosis, she developed a second recurrence located
at the skull base. This was treated with concomitant chemoradiation, and she is currently
NED at 154 months. Patient 4, who had a gross total resection with positive microscopic
margins in the sphenoid sinus, recurred locally at the edge of his frontal craniotomy
bone flap 72 months after diagnosis. The tumor was managed with surgical excision
and reirradiation. Metastatic lesions were subsequently discovered in his vertebrae
and were treated with resection followed by concomitant chemoradiation. At 104 months,
he presented with a third recurrence in his neck and ninth rib. He underwent neck
dissection and received radiation therapy to the rib. Twenty months later an additional
lesion developed in his sixth rib, which was also managed with radiation therapy.
He currently is alive with disease 10 months later.
Overall, of the nine patients in our series with positive surgical margins, five recurred.
Also noteworthy was that two of the three patients who recurred in the neck had not
had ENI with their initial treatment. Additionally, all three patients who were initially
managed with the induction chemotherapy protocol recurred.
Survival Rates
Kaplan-Meier survival curves were constructed for 5-year OS and DFS rates. The calculated
5-year OS and DFS rates were 95.2% (95% confidence interval [CI] 70.7–99.3) and 86.4%
(CI 63.4–95.4), respectively ([Figs. 1 ] and [2 ]). There was a significant improvement in 5-year DFS for patients with negative surgical
margins: 92.3% (CI 56.6–98.9) versus 77.8% (CI 36.5–93.9) for positive margins (log-rank
p = 0.015) ([Fig. 3 ]). There was also a trend toward improved OS with negative margins (100% versus 88.9%
with positive margins), although this was not significant (log-rank p = 0.269).
Fig. 1 The 5-year overall survival rate was 95.2% by Kaplan-Meier analysis.
Fig. 2 The 5-year disease free survival rate was 86.4% by Kaplan-Meier analysis.
Fig. 3 The 5-year disease-free survival (DFS) Kaplan-Meier curves by surgical margin status.
Five-year DFS rates were 92.3% with negative surgical margins versus 77.8% with positive
margins; log-rank p = 0.015.
OS and DFS rates were also evaluated by tumor stage. There was no significant difference
in 5-year OS or DFS rates between Kadish stage B and C tumors, 100% versus 91.3% (CI
52.4–98.7) (p = 0.366) and 90% (CI 47.3–98.5) versus 83.3% (CI 48.2–95.6) (p = 0.705), respectively. Results were similar when comparing tumors based on the Dulguerov
staging system. The 5-year OS rate was 100% for stage III tumors versus 88.2% (CI
41.1–98.3) for stage IV tumors (p = 0.387). There was also no significant difference in the 5-year DFS rate, 100% versus
77.8% (CI 36.5–93.9) for stage III and IV tumors, respectively (p = 0.786).
Complications
Treatment toxicity was evaluated based on the Common Terminology Criteria for Adverse
Events (v.4.0) of the National Cancer Institute. Thirteen of our patients (59%) experienced
a total of 25 complications from all modalities of therapy ([Table 2 ]).
Table 2
Treatment complications based on Common Terminology Criteria for Adverse Events (v.4.0)
of the National Cancer Institute
System organ class
Grade
Eye and optical axis (no. of patients)
Central nervous system (no. of patients)
Wound healing and infectious (no. of patients)
1
2
• Epiphora (4)
• Diplopia (2)
• Globe ptosis (1)
• Cranial nerve VI palsy (1)
• Retinopathy (2)
• Radiation brain injury (2)
• Pneumocephalus (1)
• Wound infection (1)
• Orbital cellulitis (1)
3
• CSF leak (1)
• Pneumocephalus (1)
• Osteomyelitis of bone flap (1)
• ORN of skull base (1)
• Sinocutaneous fistula (4)
• Frontal sinus abscess (1)
4
• Unilateral blindness (1)
Total, %
36
9
36
Abbreviations: CSF, cerebrospinal fluid; ORN, osteoradionecrosis.
Eye and Optical Axis
Eight patients (36.4%) had a total of 11 ocular complications. Most of these were
mild to moderate in severity (grade 2) with epiphora the most common (four patients).
A single case of cranial nerve VI palsy resolved spontaneously. Two cases of persistent
diplopia were self-limiting and produced minimal functional limitations for the patient.
A single case of severe late-radiation toxicity occurred in a patient who experienced
blindness in the ipsilateral eye as a result of radiation-induced optic neuritis (grade
4) ([Table 2 ]).
Central Nervous System (CNS)
All five reported CNS complications were experienced by the first two patients (9%)
in our cohort. Radiation brain injury was diagnosed radiographically and defined as
any enhancement seen in delayed hypersensitivity on T1-weighted MRI. One patient with
bifrontal lobe radiation injury experienced recurrent seizures, which are currently
controlled with medication. One patient experienced a postoperative cerebrospinal
fluid (CSF) leak and symptomatic pneumocephalus requiring intubation for airway diversion
and an endoscopic CSF leak repair, which was successful. The second patient developed
asymptomatic postoperative pneumocephalus that resolved with clamping of the lumbar
drain ([Table 2 ]).
Wound Healing and Infectious
Eight patients (36.4%) had a total of nine wound complications. Four patients developed
chronic sinocutaneous fistulae in the medial canthal region (grade 3). One patient
was successfully repaired with a paramedian forehead flap and prosthesis placement.
A second patient underwent repair with a radial forearm free (RFF) flap. She refistulized
a year later and a glabellar rotational flap was used to successfully close the resulting
defect. A RFF flap was used for definitive repair of a sinocutaneous fistula in the
setting of osteoradionecrosis of the skull base in a third patient. The final patient
succumbed to progressive disease prior to fistula repair.
Several infectious complications were also reported. A postoperative wound infection
(grade 2) and an episode of orbital cellulitis (grade 2) were successfully addressed
with intravenous antibiotics. One patient developed osteomyelitis of the frontal bone
flap (grade 3) 2 years postoperatively, but this also resolved with intravenous antibiotics.
Finally, antibiotics and surgical debridement were required to address a frontal sinus
abscess (grade 3) that developed in another patient.
Discussion
With an average follow-up > 6 years, our 5-year DFS rates (86.4%) and OS rates (95.2%)
compare extremely favorably with the literature, which reports DFS rates ranging from
33% to 87% and OS rates from 60% to 76%.[1 ]
[7 ]
[8 ]
[10 ]
[13 ]
[17 ]
[18 ] This is undoubtedly most attributable to the use of a consistent multimodality treatment
algorithm, which is supported in the literature. Organizing the treatment, however,
remains somewhat controversial. In > 40 years of experience treating ENB with preoperative
radiation therapy with or without chemotherapy followed by craniofacial resection,
the group at the University of Virginia has achieved an impressive 15-year DFS rate
of 82.6%.[7 ]
[8 ] Nevertheless, the meta-analysis by Dulguerov and colleagues demonstrates that surgery
followed by adjuvant radiation is the most frequently used approach and also produces
the highest OS rates: 65% at 5 years.[1 ] Our series reinforces these findings and shows excellent outcomes with upfront craniofacial
resection followed by adjuvant radiation therapy with or without chemotherapy. Our
experience also demonstrates the importance of negative surgical margins in local
control and highlights the benefits of craniofacial resection, which permits en bloc
extirpation of the tumor as well as better assessment of margin status and intracranial
extension.[1 ] This, too, is consistent with the reported experience of several institutions that
have demonstrated superior local control and improved OS with craniofacial resection
over transfacial approaches.[7 ]
[11 ]
[12 ]
[17 ]
Locoregional Disease and Recurrence
A collective review of the contemporary ENB literature reveals a high rate of cervical
nodal metastases, ranging from 17% to 33%.[4 ]
[5 ]
[14 ]
[19 ]
[20 ]
[21 ]
[22 ] Furthermore, cervical lymph node involvement has been established as a negative
prognostic indicator. Koka and colleagues reported a 0% 2-year survival rate for patients
with regional metastases.[19 ] Dulguerov's meta-analysis found a significant difference in survival, noting that
only 29% of node-positive patients were treated successfully, compared with 64% of
N0 patients.[1 ] In their series of 27 patients, Resto et al described cervical lymph node involvement
in a total of 9 patients (33%).[14 ] Statistical analysis showed nodal metastasis to be the only significant predictor
of survival. Seven patients without evidence of cervical nodal metastases at the time
of definitive treatment received elective radiation therapy to the ipsilateral neck.
In this group, only one patient (who also had positive surgical margins) went on to
develop cervical nodal disease 3 years after treatment.[14 ]
In our series, a total of 6 patients (27%) developed metastatic disease in the cervical
lymph nodes: three at presentation and three with regional recurrence. Notably, in
the three patients with regional recurrence, two had not received elective radiation
to the neck. One patient subsequently developed additional distant metastases and
is currently alive with disease. The remaining five patients are alive with no evidence
of disease. The reason for this is undoubtedly multifactorial but likely due, at least
in part, to the aggressive identification and management of the disease process. Given
the high rate of documented regional involvement, as well as the prognostic significance
described in the literature, we now routinely perform bilateral ENI in all ENB patients.
Future analysis of this growing subset of patients will provide a more definitive
assessment on outcomes using this approach.
Proton Beam Radiation, Outcomes, and Late Toxicity
Standard radiotherapy for tumors of the skull base typically involves external-beam
photons and a three-field technique, with doses ranging from 55 to 65 Gy.[1 ]
[15 ] The associated radiation toxicity and resulting complications, especially along
the optical axis, can be a significant problem. Indeed, the literature reports severe
ocular radiation injury leading to a poor or nonfunctioning eye at rates ranging from
8% to 24%.[10 ]
[12 ]
[15 ]
[23 ] Unlike photons, which deliver their highest radiation dose at the skin followed
by a continuously decreasing dose with increasing depth of penetration, protons deposit
nearly all of their energy at the point of greatest penetration. This phenomenon is
known as the Bragg peak. The exact depth to which protons penetrate depends on the
energy of the proton beam, which can be precisely controlled to place the Bragg peak
within the targeted tissues.[24 ]
[25 ] Furthermore, because most protons are absorbed at this point without an exit dose,
normal tissues beyond the target receive very little or no radiation.[24 ]
Theoretically, these properties should allow for maximal radiation doses to be focused
on the tumor, thus improving local control while sparing adjacent uninvolved tissues,
thereby decreasing the incidence and severity of acute and late toxic effects.[15 ]
[24 ]
[25 ] With sinonasal malignancies, this offers the additional advantage of sparing critical
adjacent structures, such as the globe, optic nerves, optic chiasm, and brain.[15 ] In spite of this, the use of proton beam therapy as a component of multimodality
treatment for ENB has not been examined extensively in the literature. Only a handful
of studies have documented their experience using proton beam radiation in the treatment
of sinonasal and skull base malignancies, including ENB and neuroendocrine carcinoma,
as well as adenoid cystic carcinoma.[2 ]
[9 ]
[15 ]
[26 ]
[27 ]
In general, the authors reported excellent local control and survival rates while
toxic effects were minimized. In their series of nine patients treated with induction
chemotherapy followed proton beam radiation, Bhattacharyya et al noted only a single
ocular complication, which resolved spontaneously.[2 ] However, the median follow-up was only 20.5 months. With a longer follow-up of 40
months, Nishimura and colleagues reported no toxic effects greater than grade 2.[9 ] Fitzek et al experienced no significant ocular complications, although four patients
did experience grade 2 or 3 radiation injury to the frontal lobes.[26 ] Using proton beam radiation as monotherapy or as adjuvant treatment following surgical
resection, Pommier et al achieved an encouraging 5-year local control rate of 93%
for patients with adenoid cystic carcinoma of the skull base.[27 ] The median dose delivered to the primary site was 76.4 CGE. Subsequent ocular toxicities
were not insignificant; one patient developed chronic grade 4 retinopathy and three
patients had chronic grade 3 adverse effects requiring surgical intervention.[27 ] Furthermore, 10 patients developed grade 3 radiation-induced brain injury manifested
by seizures or memory loss, and one patient died from the toxic effects of radiation-induced
brain injury (grade 5) at 61 months after radiation treatment. Another patient succumbed
to infectious meningitis secondary to a persistent CSF leak at 9 months following
radiation therapy.[27 ]
All patients in our series were treated with proton beam radiation to the primary
site. Even though the median radiation dose to the tumor was near maximal at 66.5
CGE, there were no breaks in treatment as a result of acute toxicities. However, with
an average follow-up of 73 months, 13 patients (59%) did experience several mild to
severe complications due to late-radiation toxicity. This included a case of ipsilateral
blindness resulting from radiation-induced optic neuritis. Because this patient presented
with intracranial extension of her tumor, had positive surgical margins at the skull
base, and was treated with 70 CGE plus concomitant cisplatin and etoposide, she was
undoubtedly at increased risk for treatment morbidities. In several other cases, surgical
intervention was required to resolve adverse radiation effects, including four persistent
sinocutaneous fistulas and three infectious complications at the anterior skull base.
Thus although the use of proton beam radiation may have minimized acute radiation
toxicity in our patients, extended follow-up reveals the development of several ocular
and wound complications resulting from late-radiation toxicity. Nevertheless, many
of these were mild to moderate in severity and have either resolved or been corrected
with appropriate treatment. Furthermore, despite an unfortunate case of blindness
(4.5%), our rate of severe ocular injury remains significantly less than those resulting
from standard radiation protocols.
Conclusion
ENB is a unique tumor of the skull base with a predilection for late recurrence, as
well as regional and distant metastasis. Over the past 16 years, we have used a combined
treatment protocol consisting of craniofacial resection followed by proton beam radiation
and demonstrated excellent survival outcomes. When placed in this context, and considering
that severe complications from radiation toxicity are less than those reported for
standard radiotherapy, proton beam radiation deserves consideration as a worthwhile
alternative to more traditional radiation treatment regimens for ENB. Additional experience
with more patients and longer follow-up should provide more compelling information
regarding its proper use.
