Keywords
head and neck - esthesioneuroblastoma - Hyams grade - regional failure - elective
neck treatment - radiation - survival
Introduction
Esthesioneuroblastoma (ENB) is a rare malignant neuroendocrine tumor arising from
the olfactory epithelium of the nasal cavity and paranasal sinuses. These rare tumors
are characterized by a propensity for delayed regional recurrence warranting long-term
follow-up.[1]
[2] Recurrences in ENB show a biphasic pattern with a propensity for late locoregional
recurrences, unlike head and neck squamous cell carcinomas. Several studies have reported
regional recurrence rates of up to 40% in patients presenting without involved nodes
at the time of diagnosis.[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
A consensus surrounding nodal management remains elusive given the rarity of this
disease. Some authors have suggested elective treatment of the uninvolved neck(s)
to prevent regional failure,[10]
[11] particularly in patients who are high risk.[12]
[13] However, identifying patients who are likely to recur is challenging.
Prognostic factors associated with locoregional control (LRC) included Hyams grade,
Dulguerov stage, Kadish stage, extent of resection, and margin status.[14]
[15]
[16]
[17] The Hyams histopathologic grading system is based on cellular architecture and pleomorphism,
the presence of neurofibrillary matrix, rosettes, mitoses, necrosis, glands, and calcifications
has been shown to have prognostic significance for assessing outcomes.[17]
[18]
[19] It is often used as a two-tiered grading system, separating tumors into either low
grade (grades I and II) or high grade (grades III and IV).[15]
[17]
[18]
[19]
[20]
[21]
[22] High-grade ENBs have been associated in some studies with unresectability, leptomeningeal
metastasis, frequent lymph node involvement, and poor survival,[15]
[18]
[19]
[22]
[23]
[24] while other studies have failed to demonstrate an association between Hyams grade
and recurrences.[14] Alternatively, both Kadish and Dulguerov staging systems have been shown to correlate
with outcome.[5]
[14]
[17]
[25]
[26]
[27]
The primary aim of this study is to review our institutional patterns of regional
recurrence with ENB and to evaluate whether Hyams grade serves as a prognostic factor
for the development of neck recurrences in patients with and without elective neck
treatment. Furthermore, we explored the prognostic implications of the radiographic
extent of disease in this group of patients.
Methods
Study Population
Study approval was obtained from the University Health Network Research Ethics Board.
Patients with a histopathologic diagnosis of ENB were identified by querying the Princess
Margaret Cancer Center Registry, a prospectively maintained database of patients treated
at the Princess Margaret Cancer Center and University Health Network between 1972
and 2016. Patient eligibility required: (1) a histologically confirmed diagnosis of
ENB, (2) patients treated with curative intent (with surgery and/or radiotherapy,
whether with or without chemotherapy), and (3) the availability of pretreatment primary
site biopsy or surgical specimen for analysis. Patients were excluded if they had
distant metastatic disease and/or noncurative disease at presentation.
Data Collection
A retrospective review of the medical records, review of pathology by two attending
pathologists (I.J.W. and B.P.O.), and imaging review by an attending head and neck
radiologist (E.Y.) was performed. Patient demographics, disease, stage, treatment,
and oncologic outcomes were collected. Hematoxylin and Eosin stained slides of tumors
were retrieved and reviewed, and patients who were not previously classified were
classified into Hyams grade based on predefined criteria.[17] Patients with Hyams grade I and II were then classified as low-grade tumors, and
those with grade III and IV tumors were classified as high-grade tumors. Patients
were further stratified by Kadish stage.[25]
Imaging (either computed tomography [CT] or magnetic resonance imaging [MRI]) was
reviewed by a head and neck radiologist (E.Y.) blinded to tumor histopathology and
operative findings. Imaging studies were evaluated for the extent of disease upon
pretreatment imaging. The anatomic sites included in our study were: olfactory groove,
olfactory nerve, cribriform, dura, periorbita, orbital fat, pterygopalatine fossa,
palate, ethmoid sinus, frontal sinus, sphenoid sinus, and nodal involvement. Radiotherapy
plans were reviewed to determine treatment to the primary site and the neck(s).
Statistical Analysis
Statistical analyses were performed using SPSS, version 21.0 SPSS Inc., Chicago, Illinois,
United States) and Microsoft Excel. All p-values were two-tailed and a value of p ≤ 0.05 was considered statistically significant. Patient variables were compared
using the Fisher's exact test. Our primary outcome measure was regional failure in
patients who had no clinically involved nodes at presentation. Crude regional recurrence
rates were compared by histopathologic grade (low vs. high). Survival analysis was
performed using Kaplan–Meier methods comparing low-grade tumor and high-grade tumors.
We performed multivariate analysis using Cox proportional hazards regression methods
adjusting for the following covariates: age, sex, Hyams grade, Kadish stage, radiotherapy,
chemotherapy, surgical approach, radiographic extent of disease, and margin status.
Results
Patients
A total of 43 patients (mean age = 51.5 years; standard deviation [SD] = 15.1; 52%
male) met our inclusion criteria ([Table 1]). A total of 25 patients (58%) had Hyams low-grade tumor, and 18 patients (42%)
had Hyams high-grade tumor. Two patients (5%) had Kadish stage A, 14 (34%) had Kadish
stage B, 19 (44%) had Kadish stage C, and 9 (21%) had Kadish stage D. Follow-up periods
ranged from 9 to 315 months (median = 84).
Table 1
Patient characteristics
|
Variables
|
Number of patients (%)
|
|
|
Population
n = 43
|
Hyams low grade (n = 25)
|
Hyams high grade (n = 18)
|
p-Value
|
|
Mean age ( ± SD)
|
51.4 ( ± 15.2)
|
50.1 ( ± 14.3)
|
53.2 ( ± 16.6)
|
0.5
|
|
Gender, male
|
22 (51%)
|
14 (56%)
|
8 (44%)
|
0.5
|
|
T category
|
T1- 6 (14%)
|
T1- 3 (12%)
|
T1- 3 (17%)
|
0.4
|
|
T2- 13 (30%)
|
T2- 10 (40%)
|
T2- 3 (17%)
|
|
T3- 15 (35%)
|
T3- 7 (28%)
|
T3- 8 (44%)
|
|
T4- 9 (21%)
|
T4- 5 (20%)
|
T4- 4 (22%)
|
|
N category
|
N0- 34 (79%)
|
N0- 21 (84%)
|
N0- 13 (72%)
|
0.4
|
|
N1- 9 (21%)
|
N1- 4 (16%)
|
N1- 5 (28%)
|
|
Surgical approach, open
|
29 (67%)
|
16 (64%)
|
13 (72%)
|
0.4
|
|
Margin status, positive
|
4 (10%)
|
2 (9%)
|
2 (12%)
|
1
|
|
Radiotherapy (primary and/or neck)
|
37 (86%)
|
22 (88%)
|
15 (83%)
|
0.7
|
|
Neck radiation
|
16 (37%)
|
9 (36%)
|
7 (39%)
|
1
|
Abbreviation: SD, standard deviation.
Treatment
Of the 43 patients included in our study, 34 (79%) had no evidence of regional disease
at presentation ([Fig. 1]). Of these patients with N0 neck at presentation, 8 (24%) had prophylactic neck
radiation, and 26 (76%) did not. Of the eight patients with prophylactic neck radiation,
six were bilateral and two were ipsilateral. One patient had a prophylactic neck dissection.
Of the patients with prophylactic neck radiation, six were Hyams low grade, and two
were high grade. All nine patients who had evidence of regional disease at presentation
underwent neck dissection, seven of which also had neck irradiation.
Fig. 1 Patient flow chart.
A total of 29 patients (67%) underwent open surgery and 14 (33%) underwent endoscopic
surgery. Total 37 patients received radiation to the primary site, and 16 received
radiation to the neck. Four patients received neoadjuvant chemotherapy and none were
given adjuvant chemotherapy.
Regional Recurrence and Survival
A total of 10 (23%) patients developed regional recurrences (median time-to-failure
57 months). Total 6 of 25 patients (24%) with low-grade tumors and 4 of 18 patients
(22%) with high-grade tumors developed neck recurrences. There were no statistically
significant differences in 5-year regional control in the Hyams low-grade tumors and
high-grade groups (78 vs. 89%; p = 0.4; [Fig. 2]). The 5-year locoregional control rates in patients with low-grade versus high-grade
were 73 versus 89% (p = 0.6). The 5-year overall survival rates in patients with low-grade versus high-grade
tumors were 86 versus 63% (p = 0.1). Univariable and multivariable analysis identified surgical margins and surgical
approach as significant predictors of overall survival (surgical margin hazard ratio
[HR]: 17.45, p < 0.001, and endoscopic HR: 1.7; p < 0.001).
Fig. 2 Kaplan–Meier survival curves for Hyams low grade compared with high grade.
There was no difference in survival between patients that presented with nodal disease
versus N0 patients on univariate or multivariate analyses for local, regional, or
overall survival (p > 0.05). Patient with N0 disease at presentation were stratified to determine the
impact of prophylactic neck radiation on oncologic control ([Table 2]). As shown in [Fig. 3], there was no statistically significant difference in 5-year regional control between
those with and without prophylactic neck radiation (88 vs. 84%, p = 0.3). In patients with low grade. tumors, the 5-year regional control rate in those
with and without prophylactic neck radiation was 100 versus 86% (p = 0.5), and 50 versus 90% in those with high-grade tumors (p = 0.01).
Fig. 3 Kaplan–Meier curves for regional control for patients with and without prophylactic
neck radiation (N0 patients). (A) Hyams low and high grade. (B) Hyams low grade. (C) Hyams high grade.
Table 2
Impact of prophylactic neck radiotherapy on regional control (N0 patients)
|
Population
n = 35[a]
|
p-Value
|
|
5-year regional control
|
Prophylactic neck RT
n = 7
|
No prophylactic neck RT
n = 28
|
0.3
|
|
88%
|
84%
|
|
Hyams low grade
n = 21
|
|
5-year regional control
|
Prophylactic neck RT
n = 6
|
No prophylactic neck RT
n = 15
|
0.5
|
|
100%
|
86%
|
|
Hyams high grade
n = 13
|
|
5-year regional control
|
Prophylactic neck RT
n = 1
|
No prophylactic neck RT
n = 12
|
0.01
|
|
50%
|
90%
|
Abbreviation: RT, radiotherapy.
a 1 Hyams grade unknown.
Disease Extent: Radiographic Analysis
A total of 27 patients had imaging available for review (24 MRI scans and 3 CT scans).
Of the anatomic sites included in our analysis, olfactory groove, olfactory nerve,
dura, and periorbital involvement were statistically associated with decreased 5-year
overall survival ([Table 3]). All other subsite involvements at presentation were not statistically significant
for 5-year regional or LRC (p > 0.05). As shown in [Table 3], orbital fat invasion was associated with lower 1-year regional control, suggesting
early regional failures (33 vs. 91%, p = 0.03).
Table 3
Impact of radiographic evidence of site involvement
|
Site
|
5-year locoregional control
|
p-Value
|
5- year regional control
|
p-Value
|
5-year overall survival
|
p-Value
|
|
Olfactory groove
|
63 vs. 83%
|
0.4
|
63 vs. 83%
|
0.4
|
49 vs. 91%
|
0.04
|
|
Olfactory nerve
|
63 vs. 83%
|
0.4
|
63 vs. 83%
|
0.4
|
49 vs. 91%
|
0.04
|
|
Cribriform
|
68 vs. 81%
|
0.6
|
68 vs. 81%
|
0.6
|
53 vs. 90%
|
0.08
|
|
Dura
|
77 vs. 74%
|
0.8
|
77 vs. 74%
|
0.8
|
44 vs. 92%
|
0.02
|
|
Periorbita
|
62 vs. 80%
|
0.2
|
62 vs. 80%
|
0.2
|
44 vs. 80%
|
0.04
|
|
Orbital fat
|
33 vs. 91%[a]
|
0.03
|
33 vs. 91%[a]
|
0.03
|
100 vs. 95%[a]
|
<0.001
|
|
Pterygopalatine fossa
|
50 vs. 88%[a]
|
0.3
|
50 vs. 88%[a]
|
0.3
|
100 vs. 96%[a]
|
0.09
|
|
Hard palate
|
100 vs. 73%
|
0.6
|
100 vs. 73%
|
0.6
|
67 vs. 100%
|
0.5
|
|
Ethmoid sinus
|
75 vs. 79%
|
0.6
|
75 vs. 79%
|
0.6
|
64 vs. 83%
|
0.5
|
|
Frontal sinus
|
75 vs. 74%
|
0.8
|
75 vs. 74%
|
0.8
|
67 vs. 69%
|
0.9
|
|
Sphenoid sinus
|
100 vs. 83%[a]
|
0.5
|
100 vs. 83%[a]
|
0.5
|
50 vs. 80%[a]
|
0.3
|
|
Nodal involvement
|
57 vs. 80%
|
0.5
|
57 vs. 80%
|
0.5
|
51 vs. 75%
|
0.2
|
a 1-year survival rates as 5-year not applicable.
Discussion
In the present study, we observed that in patients with ENB, regional failures are
not uncommon with a failure rate of 23% in the present study consistent with other
similar studies in the literature.[5]
[7]
[8] Hyams high-grade tumors had similar regional control rates to patients with low-grade
tumors. These findings do not reflect what has been reported in larger series. Goshtasbi
et al reported higher regional metastases rates in patients with high-grade tumors
(18.2%) compared with those with low-grade (7.9%) tumors in a meta-analysis of 525
patients with an odds ratio of 2.08 (95% confidence interval [CI]: 1.099–3.99, p = 0.03).[28] Similarly, the authors reported lower 5-year overall survival rates in high-grade
(54.2%) versus low-grade tumors (80.2%). Despite Hyams histopathologic grading system
being shown to have prognostic significance in some studies, our study likely fails
to demonstrate a difference in regional control.
We did, however, demonstrate that orbital invasion may predispose patients to early
regional failures. To the best of our knowledge, this is the first study to demonstrate
this finding. While it has been previously shown that orbital invasion is an adverse
predictor of outcome[29] and also that regional failures are associated with poor outcome, the association
between orbital invasion and neck metastases has not been shown. The biologic rationale
for this is not entirely clear but may have to do with lymphatic drainage of the orbit,
which typically drains to either the preauricular lymph nodes or lymph nodes in the
upper neck.
Patients without nodal involvement typically have a better prognosis than those presenting
with regional nodal involvement. Mortality from disease in patients with lymph node
recurrence is significant when compared with those who never develop neck disease.[27] Among those with regional failures, the overall mortality was 60 versus 32% for
patients without neck recurrence. However, the morbidity of unnecessary overtreatment
may also be significant, particularly when the majority of patients are unlikely to
develop regional recurrence over the course of their lifetime.
Treatment of the clinically negative neck varies by institution and surgeon with elective
neck dissection and radiation playing an important role in many centers.[5]
[7]
[30]
[31]
[32]
[33]
[34] Elective surgical treatment is associated with well-known morbidity for the patient
including wound infection, bleeding, neck and shoulder dysfunction, pain, marginal
mandibular nerve injury, and cosmetic deformity. Elective neck radiation similarly
can be associated with toxicities such as dry mouth, fibrosis, and lymphedema. Despite
the toxicities associated with elective neck treatment, several studies have found
conflicting arguments on the need for elective nodal treatment in patients with clinically
node negative necks. Many studies suggest a need for elective neck treatment based
on institutional experiences of regional nodal metastases ranging from 19 to 44%.[5]
[7]
[30] Elective neck radiation was found to have significant associations with regional
control.[30]
[31]
[32]
[33]
[34] Conversely, some authors recommend against elective neck treatment.[16]
[35] In a review, patients who underwent elective neck radiation or dissection had a
rate of subsequent regional metastases up to 75%.[10] In our cohort, 24% (8/34 patients) received elective neck treatment (radiation and/or
dissection). This was a clinical decision made with variability dependent on the treatment
provider. Factors relating to decision to treat elective included the grade of tumor,
extent of disease, and factors related to the understanding of disease and regional
recurrence patterns which varied over time. Of note, we were more likely to electively
treat the neck in more recent years of treatment as aforementioned literature has
emerged reporting significant rates of delayed nodal recurrence of ENB patients.
Our study did not corroborate these findings. We found no difference in the development
of regional failure from prophylactic neck radiation. Limited sample size may in part
explain our inability to identify a difference in regional control in patients who
did or did not receive radiation to the neck. We were unable to make comparisons between
those who underwent elective neck dissections and those who did not as only three
patients in our cohort underwent elective neck dissections, one (33%) of which had
pathological evidence of nodal disease. Determining which patients are likely to develop
recurrences is important to help identify those who may benefit from elective neck
treatment to prevent the development of regional metastasis and associated morbidity/mortality.
Our study failed to demonstrate prognostic factors for failures, although the small
sample size limits our conclusions. In the meantime, physicians must balance the risk
of recurrence with the added morbidity of radiation therapy and/or surgery to the
neck. Furthermore, long-term follow-up is crucial given the high risk of delayed regional
failure. The identification of some predictive biomarker for regional failure may
help better tailor therapy to those who need it.[36]
[37]
This study has several limitations. Our relatively small sample size limits the statistical
power. The retrospective nature of the study is another limitation. Given the protracted
course of regional failures in this disease, some patients may have been lost to follow
up or may have sought care for recurrent disease elsewhere and as such not been counted
in the analysis. Histopathologic misclassification of ENB is another possibility although
in this particular study, pathologic specimens were classified by head and neck pathologist(s).
Conclusion
ENBs may recur in the neck either in an early or delayed fashion. Patients who recur
early either locoregionally or regionally have a poorer prognosis. The present study
fails to demonstrate an association between Hyams grade and regional recurrences.
However, we did demonstrate the association between olfactory groove, olfactory nerve,
dura, and periorbital involvement and poor survival. Future larger scale multicenter
collaborative efforts might help confirm or refute potential markers for disease recurrence.
