Keywords
SNUC - sinonasal undifferentiated carcinoma - survival - treatment paradigms - prognosis
Introduction
Sinonasal undifferentiated carcinoma (SNUC) is a rare, highly aggressive malignancy
that lacks clearly defined treatment protocols and concrete stage-based survival data.
Overall SNUC mortality rates are high, with 5-year survival ranging from 20 to 63%
in the literature.[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8] SNUC is extremely locally destructive,[9] frequently invading and eroding adjacent structures early on in the disease course.[10] Additionally, it most often presents at American Joint Committee on Cancer (AJCC)
stage IV.[1]
[3]
[4]
[5]
[6]
[8]
[11]
[12] The reasons for advanced disease at presentation are multifactorial, but delay in
diagnosis is a principal feature, as symptoms of SNUC are more readily attributed
to benign causes early on.[13]
[14]
[15] Often, a patient with a locally advanced SNUC will present at evaluation for nasal
congestion, sinusitis, or symptoms of facial pressure. The large potential space of
the sinuses also offers an opportunity for unimpeded initial growth, further contributing
to high preliminary T stage of the tumor.[16]
SNUC is a unique neoplasm, distinguished from esthesioneuroblastoma (ENB) by a lack
of periodic Homer-Wright rosettes and intercellular fibrils commonly associated with
ENB. It also differs from Epstein–Barr virus-related nasopharyngeal-type undifferentiated
carcinoma, which is a distinct entity with superior outcomes.[17] There is also an overexpression of p16 in SNUC, related to human papillomavirus
(HPV) infection in many, but not all, cases.[18] SNUC was previously described as an anaplastic malignancy, and it can be epithelial
or nonepithelial in origin.[11]
[19] Treatment for SNUC is generally surgery, radiation, chemotherapy, or some combination
of these modalities; a consensus on timing and sequence of treatment has not yet been
established. Some studies have argued superiority of surgery alone, chemoradiation
alone, or the use of neoadjuvant or adjuvant chemoradiation.[5]
[17] However, no uniform agreement on survival advantage with any particular treatment
modality has been established.[13]
Several case series have been published since SNUC was first identified as an independent
entity in 1986,[11] but the number of patients in each series is quite limited due to the infrequency
of SNUC occurrence. This study, conducted at the University of Pennsylvania hospitals,
is one of the largest single-institution case series to date, comprised of 27 patients
over a 15-year period. It is our hope that this contribution to the literature will
provide useful additional data for discerning optimal management and prognostication
of this disease.
Methods
Patient Selection
Approval was obtained from the University of Pennsylvania Institutional Review Board
for a retrospective cohort study of SNUC patients from January 1992 to December 2017
treated at the Hospital of the University of Pennsylvania or Pennsylvania Hospital.
All patients were identified using pathology criteria from our internal pathology
report database, and all available inpatient and outpatient records were obtained
for each patient. Records contained treatment reports from a multidisciplinary team
at the University of Pennsylvania with an otolaryngologist, medical oncologist, radiation
oncologist, and pathologist. In some cases, a neurosurgeon was also consulted. All
patients that did not undergo surgical or medical management for SNUC with posttreatment
follow-up at the University of Pennsylvania were excluded.
Data Collection
Patient demographic characteristics, including age at diagnosis, gender, smoking and
alcohol histories, and prior radiation treatment, were recorded. If tumor stage was
provided in the clinical record in a radiology, radiation oncology, or otorhinolaryngology
note then this was noted, otherwise tumors were staged using the guidelines provided
by the AJCC based on available radiographic and clinical information. Orbital involvement
and presence of neck metastasis at diagnosis were two additional factors obtained.
When obtainable, tumor size was reported in centimeters in largest dimension.
Each patient in the data set underwent surgical, radiation, and/or chemotherapy treatment
and was followed at the University of Pennsylvania as an outpatient for at least 2
years or until time of death. Surgical factors, including extent of resection, type
of surgical approach, neurosurgical involvement, and intraoperative or postoperative
complications, were recorded. Specific type of chemotherapy and/or radiation dose
(in units Gray) was also recorded. Death was the primary examined outcome, but tumor
recurrence, as assessed by imaging or endoscopy, was a secondary outcome.
Statistical Analysis
Size of the data set precluded the use of robust statistical techniques to examine
all independent variables on primary and secondary outcome measures, but patient-
and disease-specific characteristics were examined to assess their effect on the primary
outcome of disease-specific survival. Patients were censored from the Kaplan–Meier
survival curve if lost to follow-up after 2 years. Statistical comparisons were performed
using Cox proportional hazards modeling for assessing the effect of patient variables
on survival. For outcome variables, statistical comparisons were performed using the
log-rank Mantel–Cox test as appropriate. A value of p < 0.05 was considered statistically significant. Statistical analyses were conducted
using STATA 13 (Stata Corp, College Station, Texas, United States).
Results
A total of 27 patients with SNUC were treated and followed at the Hospital of the
University of Pennsylvania or Pennsylvania Hospital between 1992 and 2017. Demographic
data for these patients, as well as tumor characteristics, are listed in [Table 1]. Examples of SNUC histopathology in this case series, including a positive cytokeratin
stain and negative S100 protein and synaptophysin stains, can be seen in [Fig. 1]. A majority of patients (63%) were male, and a majority also had a history of smoking
or regular alcohol consumption. The average age at diagnosis was 55 years, with an
age range of 29 to 80 years. Notably, a preponderance of patients presented at stage
IVA or greater, representing 85% of the cohort, while no patients presented with stage
I disease. Approximately half of patients had orbital involvement at the time of diagnosis,
and a third had documented nodal neck disease. The most common presenting symptoms
were nasal obstruction and sinusitis, followed by epistaxis. Tumor size data was heterogeneous
in nature and not reported in several cases, but in those with available information,
the tumors were an average of 3.9 cm in greatest dimension.
Fig. 1 SNUC histopathology of two patients in the case series. (A1/2) Hematoxylin and eosin
stain. (B1/2) Positive cytokeratin stain. (C1/2) Negative S100 stain. (D1/2) Negative
synaptophysin stain.
Table 1
Patient and tumor characteristics
|
Patient characteristics
|
|
|
Number of patients
|
27
|
|
Age at diagnosis (mean)
|
54.6 ± 12.2
|
|
Male gender
|
17 (63%)
|
|
Smoking history
|
17 (63%)
|
|
Alcohol history
|
16 (59%)
|
|
Prior radiation treatment
|
1 (4%)
|
|
Tumor characteristics
|
|
|
TNM staging
|
|
|
Stage I
|
0 (0%)
|
|
Stage II
|
1 (4%)
|
|
Stage III
|
3 (11%)
|
|
Stage IVA
|
9 (33%)
|
|
Stage IVB
|
11 (41%)
|
|
Stage IVC
|
3 (11%)
|
|
Orbital involvement
|
13 (48%)
|
|
Neck metastasis
|
9 (33%)
|
Abbreviation: TNM, tumor/node/metastasis.
Eighty-five percent of patients underwent primary surgical management ([Table 2]), and half of the nonoperative patients had unresectable stage IVC disease. [Fig. 2] shows a schematic diagram of patient treatment, from primary management to adjuvant
therapies. A large majority of patients had an initial biopsy at an outside hospital,
and were then referred to the University of Pennsylvania for further treatment. Cisplatin
and carboplatin were by far the most commonly used chemotherapy agents. For radiation
therapy, patients received a median dose of 63 Gy to the tumor site, with a range
of 50 to 70 Gy for those completing the full course of radiation.
Fig. 2 Flow diagram of treatment for the SNUC patients in the overall cohort.
Table 2
Treatment and outcomes
|
Treatment
|
|
|
Surgery
|
23 (85%)
|
|
Gross total resection
|
13 (57%)
|
|
Endoscopic
|
13 (57%)
|
|
Comb. neurosurgical
|
10 (43%)
|
|
Complications
|
4 (17%)
|
|
Radiation
|
26 (96%)
|
|
Chemotherapy
|
26 (96%)
|
|
Outcomes
|
|
|
Recurrence
|
14 (52%)
|
|
Distant metastasis
|
8 (30%)
|
Within the surgical cohort, slightly over half of all patients had a gross total resection
(GTR), and a similar proportion also had an exclusively endoscopic approach ([Table 2]). Neurosurgery was involved in 43% of cases, most often due to dural involvement
of the tumor. Surgical or postsurgical complications were present in four patients,
comprised of two cerebrospinal fluid leaks, a necrotic free flap, and severe epistaxis.
Four patients in the overall cohort required orbital exenteration either during or
after the primary surgery.
Disease-specific survival of both surgical and nonsurgical patients was 66% at 2 years
and 46% at 5 years ([Fig. 3]). Notably, disease-specific survival and overall survival were equivalent in our
cohort. When examining seven patient and tumor characteristics for effects on survival
([Table 3]), neck disease at presentation was associated with increased mortality (p < 0.05, hazard ratio, 4.5, 95% confidence interval, 1.04–19.2, [Fig. 4]); median survival of patients with neck disease was 20 months, compared with 126
months for individuals without initially positive neck nodes. The secondary outcome
of disease recurrence was also independently associated with decreased survival (p < 0.01, log-rank Mantel–Cox test). Metastasis before or following treatment was associated
with an especially poor median survival, with mortality in half of these patients
in the 13 months following presentation.
Fig. 3 Kaplan-Meier overall survival analysis.
Fig. 4 Kaplan-Meier survival analysis for all patients, broken into two groups by nodal
neck disease at presentation.
Table 3
Survival by patient and tumor characteristics
|
Variable
|
Hazard ratio (95% CI)
|
p-Value
|
|
Gender
|
1.80 (0.47–6.88)
|
0.39
|
|
Age
|
1.01 (0.95–1.07)
|
0.84
|
|
Smoking history
|
0.83 (0.24–2.86)
|
0.77
|
|
Alcohol history
|
0.68 (0.20–2.28)
|
0.53
|
|
Neck metastasis
|
4.47 (1.04–19.20)
|
0.04
[a]
|
|
Distant metastasis
|
3.82 (0.76–19.06)
|
0.10
|
Abbreviation: CI, confidence interval.
a
p < 0.05.
Discussion
This study is an overview and analysis of a 27-patient SNUC cohort, and is one of
the largest single-institution case series published to date. SNUC is a highly aggressive
and uncommon neoplasm, with incompletely defined prognostic indicators and treatment
strategies. Most patients in this cohort presented with clinically advanced disease,
and almost all were managed with a multimodality approach of surgical resection, chemotherapy,
and radiation. Five-year overall and disease-specific survival was 46%, and neck disease
at presentation and disease recurrence following treatment were associated with increased
mortality.
To understand the data in our patient population, it is best presented in the context
of other similar case series that have been published in the past 15 years. [Table 4] presents a list of all United States SNUC case series in the literature that are
comprised of greater than 10 patients, with data extracted and compared with this
study. Several notable trends are apparent. Patient age is very similar among all
cohorts, with presentation in the 5th decade of life. Percentage of patients presenting
with T4 disease varies from two-thirds to almost all patients, corroborating what
is known about the SNUC disease course.[1]
[3]
[4]
[5]
[6]
[8]
[11]
[12] Interestingly, our cohort had significantly higher levels of nodal disease at presentation
than those reported at other United States institutions. The reasons for this discrepancy
are unclear, but may partially account for the relatively high proportion of patients
undergoing multimodality therapy with surgery, chemotherapy, and radiation at our
institution. Furthermore, both our study and previous reports have identified nodal
disease at presentation as a very poor prognostic indicator,[24] which has important implications for survival.[1]
[3]
[5]
[6]
[10] Despite this, 2- and 5-year survival in our subjects is comparable to the average
of what has previously been reported.
Table 4
Published SNUC case series in the United States
|
Study
|
Penn
|
Mayo
|
UCSF
|
UVA
|
UM
|
Harvard
|
Einstein
|
UCD
|
MD And.
|
UF
|
UC
|
|
Author
|
Workman et al
|
Gamez et al
|
Chen et al
|
Musy et al
|
Lin et al
|
Gray et al
|
Mourad et al
|
Yoshida et al
|
Rosenthal
|
Tanzler et al
|
Miyamoto et al
|
|
Year
|
Current
|
2017
|
2008
|
2002
|
2010
|
2015
|
2013
|
2013
|
2004
|
2008
|
2000
|
|
Patients (#)
|
27
|
40
|
21
|
20
|
19
|
19
|
18
|
16
|
16
|
15
|
14
|
|
Age
|
55
|
57
|
47
|
58
|
51
|
52
|
52
|
50
|
48
|
57
|
54
|
|
F/u (mo)
|
47
|
83
|
58
|
80
|
21
|
|
26
|
|
81
|
30
|
|
|
T4 (%)
|
85
|
80
|
81
|
73
|
84
|
100
|
67
|
94
|
69
|
100
|
63
|
|
Node + (%)
|
33
|
8
|
10
|
13
|
21
|
16
|
22
|
25
|
0
|
13
|
|
|
Surgery (%)
|
85
|
83
|
90
|
55
|
53
|
63
|
83
|
63
|
63
|
66
|
64
|
|
XRT (%)
|
96
|
100
|
100
|
95
|
100
|
100
|
83
|
63
|
100
|
93
|
86
|
|
Chemo (%)
|
96
|
68
|
62
|
80
|
84
|
100
|
83
|
63
|
|
47
|
43
|
|
2 y OS (%)
|
66
|
|
|
47
|
61
|
|
75
|
|
|
|
45
|
|
5 y OS (%)
|
46
|
44
|
43
|
20
|
22
|
45
|
48
|
33
|
63
|
67
|
36
|
Abbreviations: Einstein, Yeshiva University and Albert Einstein College of Medicine;
f/u, median follow-up; Mayo, Mayo Clinic in Rochester, MN; MD And, MD Anderson Cancer
Center; OS, overall survival; Penn, University of Pennsylvania; SNUC, sinonasal undifferentiated
carcinoma; UC, University of Cincinnati;[3]
[4]
[6]
[7]
[8]
[12]
[20]
[21]
[22]
[23] UCD, University of California, Davis; UCSF, University of California, San Francisco;
UF, University of Florida; UM, University of Michigan; UVA, University of Virginia;
XRT, radiation therapy.
Two meta-analyses and two national database reports have been published for SNUC in
the past 5 years, and these provide important information that affects the evaluation
and treatment of SNUC.[1]
[25]
[26]
[27] A meta-analysis by Reiersen et al in 2012 found that while only 20% of all patients
treated for SNUC received surgery, radiation, and chemotherapy, a combination of any
of these therapies was strictly superior in offering improved survival when compared
with any single modality. Among the treatment modalities, surgery had the largest
effect, while adjuvant therapy was most beneficial in patients with advanced local
or neck disease. In the context of our current case series, 81% of our patients received
all three therapies, and the available evidence points to aggressive multimodal management
as the recommended approach. Surgery, especially when GTR is achieved, is highly predictive
of local disease control.[28]
[29] This is further corroborated by reports showing that 70% of patients with SNUC will
still have viable tumor at surgical resection following chemoradiation, highlighting
the insufficiency of nonsurgical management.[4]
While surgery is clearly a critical component of SNUC treatment, adjuvant therapy
plays a demonstrable role. SNUC neoplasms tend to be microscopically invasive, and
even negative margins should be further treated with radiation and chemotherapy. Cisplatin,
5-fluorouracil, and paclitaxel have an overall response rate of 80% in SNUC,[30] and these are the most frequently used therapies in practice.[16] At our institution, cisplatin or carboplatin were the first line agents of choice.
Mean radiation dose across the majority of studies is 63.9 Gy, comparable to the median
value of 63 Gy in our cohort. Most radiation treatments given in our series were postsurgical,
and radiation's role in downstaging patients prior to surgical treatment is still
unclear. However, even in patients without demonstrated nodal disease at presentation,
prophylactic neck irradiation has proven effective for preventing development of regional
disease.[12] Once cervical lymphadenopathy is present on physical exam, greater than 90% of patients
will have regional or distant metastases.[1] In considering SNUC management, clinicians must assess the need for elective prophylactic
surgical or radiation treatment of the neck to improve prognosis.
A meta-analysis by Chambers et al, a study in the Surveillance, Epidemiology, and
End Results database, and a study in the National Cancer Database provide other important
demographic and survival considerations for SNUC.[25]
[26]
[27] There is a 2:1 predilection to SNUC in males, which is also borne out in our cohort
(63%). Older age, advanced stage, metastatic disease, and Asian race were all significantly
associated with decreased survival, while combination surgery with adjuvant therapy
was again more effective than single-modality treatment. Notably, a large majority
of patients who died of their disease did so within 20 months of treatment,[25] suggesting that there is a time during which surveillance is most critical.[31]
[32] Like many other sinonasal cancers, a proportion of SNUC is related to HPV infection,
although p16 is often overexpressed in SNUC even in the absence of HPV.[18] Both HPV positive and p16 positive SNUC patients showed improved survival in a preliminary
study.[20]
The long timespan covered by this case series makes for a diagnostic challenge. The
updated 2017 World Health Organization criteria for histopathological diagnosis of
head and neck tumors describes SNUC as a diagnosis of exclusion, without squamous
or glandular differentiation present.[33] Furthermore, distinction of SNUC from large cell neuroendocrine carcinoma is difficult.
Diagnoses of exclusion can be susceptible to sampling error and a definitive classification
cannot always be ensured; our case series is not immune to this potential pitfall.
Specific stains, such as the integrase interactor 1 protein stain and the nuclear
protein in testis stain were not performed on some patients in the case series, as
these stains have only recently been commonly utilized by pathologists for SNUC diagnosis.
Other limitations of our study are related to sample size; a total of 27 patients
precludes robust regression analyses or power to detect differences in all variables
examined in our cohort. However, the rarity of SNUC ensures that no single institution
will have a large enough group for adequate statistical considerations, and it is
unlikely that a randomized controlled trial evaluating SNUC therapies would be possible
even with a multi-institutional effort. The infrequency of SNUC cases makes case series
valuable, both for future meta-analyses and appraisal of homogenous, single-institution
data.
Overall, this study assesses demographics, tumor characteristics, and clinical management
of a relatively large cohort of SNUC patients over a 15-year period. Long-term survival
of SNUC patients is universally poor, as almost all patients present with very advanced
disease. Neck metastases were present in a high proportion of our patient population,
and were associated with decreased survival time. There was a heavy emphasis on a
multimodal approach of surgical resection with adjuvant chemoradiotherapy in almost
all patients, regardless of disease extent, and the literature is largely supportive
of this methodology. Persistence of disease following treatment was a very poor prognostic
sign and highly predictive of mortality. Identification of optimal strategies to prevent
disease progression is a high priority for developing improved treatments for SNUC.
