Population-based cancer registries allow for data collection on the scale of large populations, outside the limits of a single institution, and facilitate study of rare entities. The SEER database has been used to study more than 7000 cases encompassing a wide variety of relatively rare sinonasal malignant histologies. Clinically useful parameters have been gleaned from these analyses. Important limitations, such as omission of chemotherapy data, surgical approach used, type of radiation administered, and selection and confounding bias, should be considered. Nevertheless, population-based analyses yield readily generalizable and clinically relevant information regarding the management of sinonasal malignancies for the practicing clinician.
Key points
- •
In the United States, population-based analyses of sinonasal malignancies have relied on the Surveillance, Epidemiology, and End Results (SEER) registry developed and maintained by the National Cancer Institute, representing 17 geographic areas in the United States and accounting for 26.2% of the population.
- •
Many clinically useful analyses that include epidemiologic and survival information regarding 13 distinct malignant histologies are made possible through the use of registry data.
- •
Although these analyses are powerful, important limitations, such as selection and confounding bias, omission of chemotherapy data, type of surgical approach used, and timing of radiation treatment, should be considered.
| AJCC | American Joint Committee on Cancer |
| API | Asian/Pacific Islander |
| DLBCL | Diffuse large B-cell lymphoma |
| DSS | Disease-free Survival |
| EBV | Epstein-Barr virus |
| ENB | Esthesioneuroblastoma |
| EMP | Extramedullary plasmacytoma |
| ENKTL | Extranodal natural-killer/T-cell lymphoma |
| NHW | Non-Hispanic white individuals |
| NPC | Nasopharyngeal carcinoma |
| OS | Overall survival |
| RS | Relative survival |
| RT | Radiation Therapy |
| SCC | Squamous cell carcinoma |
| SEER | Surveillance, Epidemiology, and End Results |
| SNAC | Sinonasal adenocarcinoma |
| SN-ACC | Sinonasal adenoid cystic carcinoma |
| SNEC | Sinonasal neuroendocrine tumor |
| SN-MEC | Sinonasal mucoepidermoid carcinoma |
| SN-RMS | Sinonasal rhabdomyosarcoma |
| SNUC | Sinonasal undifferentiated carcinoma |
Introduction to population-based analyses and registry data
Population-based cancer registries allow for collection, classification, and consolidation of cases on the scale of populations, outside the limits of any individual institution. This provides a method for measuring and studying patterns of disease over time across state lines, borders, and geographic locations inclusive of a wide spectrum of demographics and genetic compositions.
In the United States, population-based analyses in surgical oncology have relied in part on the use of the Surveillance, Epidemiology, and End Results (SEER) registry developed and maintained by the National Cancer Institute. This database has allowed for investigations into the course and treatment outcomes of specific malignancies that are more broadly generalizable than studies from individual institutions. The SEER registry began collecting data in 1973 and currently represents 17 geographic areas in the United States accounting for 26.2% of the population. Although the data are considered highly valid, there is a slight overrepresentation of those living below the poverty level, inhabiting urban centers, born in foreign countries, and with education levels below a high-school diploma. Information in the registry includes patient demographics, tumor histology, sites of involvement, extent of disease, use of surgical and/or radiation therapy (RT) modalities of treatment within 4 months of diagnosis, and patient survival, among other parameters. Importantly, however, tumor stage is not always reported, particularly along TNM specifications, and metastases are not specifically defined. Furthermore, preexisting comorbidities, use of chemotherapy, and type of surgical intervention are not reported in the registry.
There are important caveats to consider when assessing applicability of conclusions drawn from population-based analyses as it pertains to the use of registry data. The SEER registry in particular, although clearly a powerful tool, is nonetheless vulnerable to all types of bias, particularly selection and confounding bias. For instance, the decision-making process leading one patient to receive surgical treatment instead of, or in conjunction with, another treatment modality is not captured by the registry. Similarly, observational studies run the risk of drawing invalid inferences with incomplete control of confounding variables, such as comorbidities, which have a tendency to be undercoded.
An important limitation in the SEER registry is the omission of chemotherapy treatment. The contribution of chemotherapy either as a primary or adjuvant therapy on survival cannot be assessed. For instance, the impact of chemotherapy on lymphoma, which is this entity’s treatment of choice, cannot be determined with registry data. Also more subtle comparisons, such as the impact of the BRAF deletion characteristic of mucosal melanoma and its reduced chemo-sensitivity, could not be studied.
Although the use of RT is reported in the registry, 2 important limitations deserve attention. First, the timing of radiation, whether neoadjuvant, adjuvant, or primary, is not consistently reported. Second, the decision making associated with whether to add RT is not included in the registry, resulting in the possibility that radiation treatment is often reserved for refractory or late-stage disease or in patients with difficult surgical approaches. This has the effect of biasing survival analyses, as it would appear that patients treated with RT alone or with combination therapy have worse outcomes, although in reality it is a reflection of more advanced disease that would be expected to have reduced survival.
The conclusions regarding surgical treatments also warrant caution. The SEER registry does not provide information on surgical approaches or extent of surgery, such as gross total, subtotal, debulking, or palliative approaches, making the study of extent of resection difficult. Furthermore, the role of modern day endoscopic resection techniques, popularized over the past 20 years, may be changing the face of overall survival rates. Without this explicitly included in the registry data, conclusions are largely speculative. Finally, but equally important, pathologic data with regard to surgical margin status is not included, making the study of the merits of negative margin resection difficult.
Despite these limitations, in the absence of high-quality randomized controlled trials, the SEER database and other registries offer several advantages in the study of sinonasal and ventral skull base malignancies in particular. As sinonasal malignancies are rare, constituting fewer than 3% of all upper aerodigestive tract malignancies, using a population-based resource to study them allows for access to data from thousands of patients, sufficiently powering the statistical analyses in a way not possible with data from a single institution. Certain types of bias are reduced, as there is a wide array of geographic areas and institutions represented lending external validity to the studies. Furthermore, active follow-up enables and improves accuracy of survival analyses. As results are more readily generalizable, practitioners are better poised to apply results to individual patients.
Introduction to population-based analyses and registry data
Population-based cancer registries allow for collection, classification, and consolidation of cases on the scale of populations, outside the limits of any individual institution. This provides a method for measuring and studying patterns of disease over time across state lines, borders, and geographic locations inclusive of a wide spectrum of demographics and genetic compositions.
In the United States, population-based analyses in surgical oncology have relied in part on the use of the Surveillance, Epidemiology, and End Results (SEER) registry developed and maintained by the National Cancer Institute. This database has allowed for investigations into the course and treatment outcomes of specific malignancies that are more broadly generalizable than studies from individual institutions. The SEER registry began collecting data in 1973 and currently represents 17 geographic areas in the United States accounting for 26.2% of the population. Although the data are considered highly valid, there is a slight overrepresentation of those living below the poverty level, inhabiting urban centers, born in foreign countries, and with education levels below a high-school diploma. Information in the registry includes patient demographics, tumor histology, sites of involvement, extent of disease, use of surgical and/or radiation therapy (RT) modalities of treatment within 4 months of diagnosis, and patient survival, among other parameters. Importantly, however, tumor stage is not always reported, particularly along TNM specifications, and metastases are not specifically defined. Furthermore, preexisting comorbidities, use of chemotherapy, and type of surgical intervention are not reported in the registry.
There are important caveats to consider when assessing applicability of conclusions drawn from population-based analyses as it pertains to the use of registry data. The SEER registry in particular, although clearly a powerful tool, is nonetheless vulnerable to all types of bias, particularly selection and confounding bias. For instance, the decision-making process leading one patient to receive surgical treatment instead of, or in conjunction with, another treatment modality is not captured by the registry. Similarly, observational studies run the risk of drawing invalid inferences with incomplete control of confounding variables, such as comorbidities, which have a tendency to be undercoded.
An important limitation in the SEER registry is the omission of chemotherapy treatment. The contribution of chemotherapy either as a primary or adjuvant therapy on survival cannot be assessed. For instance, the impact of chemotherapy on lymphoma, which is this entity’s treatment of choice, cannot be determined with registry data. Also more subtle comparisons, such as the impact of the BRAF deletion characteristic of mucosal melanoma and its reduced chemo-sensitivity, could not be studied.
Although the use of RT is reported in the registry, 2 important limitations deserve attention. First, the timing of radiation, whether neoadjuvant, adjuvant, or primary, is not consistently reported. Second, the decision making associated with whether to add RT is not included in the registry, resulting in the possibility that radiation treatment is often reserved for refractory or late-stage disease or in patients with difficult surgical approaches. This has the effect of biasing survival analyses, as it would appear that patients treated with RT alone or with combination therapy have worse outcomes, although in reality it is a reflection of more advanced disease that would be expected to have reduced survival.
The conclusions regarding surgical treatments also warrant caution. The SEER registry does not provide information on surgical approaches or extent of surgery, such as gross total, subtotal, debulking, or palliative approaches, making the study of extent of resection difficult. Furthermore, the role of modern day endoscopic resection techniques, popularized over the past 20 years, may be changing the face of overall survival rates. Without this explicitly included in the registry data, conclusions are largely speculative. Finally, but equally important, pathologic data with regard to surgical margin status is not included, making the study of the merits of negative margin resection difficult.
Despite these limitations, in the absence of high-quality randomized controlled trials, the SEER database and other registries offer several advantages in the study of sinonasal and ventral skull base malignancies in particular. As sinonasal malignancies are rare, constituting fewer than 3% of all upper aerodigestive tract malignancies, using a population-based resource to study them allows for access to data from thousands of patients, sufficiently powering the statistical analyses in a way not possible with data from a single institution. Certain types of bias are reduced, as there is a wide array of geographic areas and institutions represented lending external validity to the studies. Furthermore, active follow-up enables and improves accuracy of survival analyses. As results are more readily generalizable, practitioners are better poised to apply results to individual patients.
General considerations in sinonasal and ventral skull base population-based results
The SEER registry identified more than 7000 cases of sinonasal malignancies of all types since 1973. Of these patients, 58.3% are of male gender, and 82.2% are white. The tumors are of mainly epithelial origin, including squamous cell carcinoma (SCC) (51.6%), adenocarcinoma (SNAC) (12.6%), esthesioneuroblastoma (ENB) (6.3%), and adenoid cystic carcinoma (SN-ACC) (6.2%), although many others have been studied.
The overall incidence of sinonasal malignancies is 0.556 cases per 100,000, stable over the past 30 years, compared with 1 to 5 per 100,000 for other head and neck sites. There is an overall male-to-female ratio of 1.8:1.0, although incidence among men has trended down over time. With regard to race, incidence among white individuals has been constant, whereas among black individuals and “other” races there has been a decline. In terms of histology, SCC incidence has been declining substantially over time, whereas that of melanoma and ENB has been increasing. Five-year survival for all-comers is 54.5% ± 0.9%, with no significant change over time. Survival is better for white individuals (56.3% ± 1.0%) and “others” (50.3% ± 2.8%) as compared with black individuals (40.0% ± 2.8%). No differences in survival rates over time are noted for gender or for certain pathologies, including SCC, SN-ACC, ENB, melanoma, or sinonasal undifferentiated carcinoma (SNUC). In contrast, 5-year survival for SNAC shows improvement over time and is discussed later in this article.
Recent studies have found the nasal cavity to be the most commonly involved site (43.9%), followed by the maxillary sinuses (35.9%) and ethmoid sinuses (9.5%), although this varies by pathology. The frontal and sphenoid sinuses are rarely involved, totaling 3.3% of reported sinonasal cases. Tumors originating in the nasal cavity have significantly better survival than tumors in other sinonasal locations. Tumors are also more likely to present at an advanced stage with concordant poor survival. Reported 5-year relative survival is approximately 80%, 50%, and 30%, with no metastases, regional, and distant metastases respectively.
With respect to 5-year survival, relatively favorable prognosis is seen for ENB (71% ± 2.6%) and SN-ACC (69.5% ± 2.8%), an intermediate prognosis for SCC (53.1% ± 1.1%) and SNAC (63.0% ± 2.1%), and poor prognosis for melanoma (34.7% ± 3.1%) and SNUC (34.7% ± 4%). By site of origin, 5-year survival is best for nasal cavity (71.0% ± 1.3%), followed by ethmoid sinus (45.2% ± 2.8%), sphenoid sinus (40.6% ± 5.2%), maxillary sinus (39.5% ± 1.5%), and frontal sinus (35.3% ± 7.5%). Among those with localized disease, surgery alone offers the best survival, with combination surgery and RT portending a worse survival. Patients treated with RT alone trend toward inferior survival (49.4% ± 1.1%) as compared with no RT (64.7% ± 1.5%). In patients with distant disease, those treated with RT have improved survival (31.1% ± 2.9% vs 21% ± 4.2%). Among those with regional disease, surgery, or the combination of surgery and RT (hereby referenced as combination therapy) is superior to RT alone. For those with distant disease, a similar trend is seen, although adding RT to surgery is better than surgery alone.
A query of the Netherlands Cancer Registry, encompassing the entire national population of 16.5 million between 1989 and 2009, found 3329 patients with sinonasal malignancies showing similar histologic trends with SCC as the most common histologic type (48%), followed by SNAC (15%), and melanoma (8%). Consistent with other studies, SCC more commonly involves the nasal cavity and maxillary sinus, whereas SNAC more commonly involves the ethmoid sinus. Incidence among men rose until 1995, but has since seen a decline in both SCC and SN-ACC (11/1,000,000 per year), whereas incidence among woman was stable until 2006 and has since seen a significant increase due to SCC. This is coincident with measures to limit occupational exposure to wood dust in the Netherlands, implemented in 2000.
Multiple studies have concluded that surgery is the single most effective modality for sinonasal malignancy. Radiation offers limited survival benefit, particularly for early-stage disease. In more advanced lesions, combination therapy consistently failed to demonstrate a significant benefit, although this may represent a selection bias, as more T3/T4 lesions were treated with combination therapy and would be expected to have a reduced survival regardless of therapy.
Specific sinonasal malignancies
Squamous Cell Carcinoma
As the predominant paranasal sinus solid tumor histologic subtype across all studies, SCC represents approximately half of all cases in the literature. Occupational exposure to nickel, industrial fumes, and textile dust have been identified as risk factors, and SCC is particularly common among men and those of African and Asian (particularly Japanese) descent. Chronic rhinosinusitis, allergies, nasal polyposis, Epstein-Barr virus (EBV), and tobacco are other risk factors. A study by Ansa and colleagues of 2553 patients with SCC in the SEER registry between 1973 and 2009 noted more than half were between ages of 60 to 79, 63.6% were men, 75% were white, and 76% presented with maxillary sinus origin. The overall incidence decreased over time, although incidence was increased among black individuals and other minorities. Of note, presence of “distant” disease decreased from 14.7% to 9.5% over the study period. The more advanced tumors involved the sphenoid and ethmoid sinuses. More than half of all cases were treated surgically with or without RT for all sites except those of the sphenoid sinus that were rarely treated surgically.
The 5-year and 10-year overall survival (OS) is 30.2% and 21.0%, respectively, and noted to be better in men and white individuals than women and black individuals. The benefit of surgery alone versus combination therapy was similar, although both groups were better than RT alone. Factors associated with a higher mortality included advanced-stage disease, older age, and black race, where a 22% increase in mortality was noted with regional and distant disease compared with white individuals, even after control for stage and treatment. No difference in survival by sex or primary tumor site was noted. However, when examined by race, white individuals had the highest long-term survival (30.93% at 20 years), whereas black individuals consistently had the lowest survival rates coupled with the highest incidence rates. Patients with localized disease treated with surgery had the best 5-year survival at 85.71% compared with combination (80.38%) or RT alone (78.47%). Those with regional spread and distant metastases fared significantly worse.
When stratifying SCC by histologic subtype, including conventional SCC, verrucous (V-SCC), papillary (P-SCC), basaloid (B-SCC), sarcomatoid (SCSC), and adenosquamous (ASC), interesting differences are noted. B-SCC is found to have a statistically significant, albeit slightly lower age of diagnosis at 61.8 ± 15.6 years as compared with 66.0 ± 13.3 years for conventional SCC. SCSC is less likely than conventional SCC to arise in the nasal cavity (28.6% vs 46.3%) and possibly more likely to arise in the maxillary sinus (61.2% vs 41.5%, P = .05). There is a trend for less maxillary sinus involvement for P-SCC and more maxillary sinus involvement for SCSC relative to conventional SCC. V-SCC, a low-grade variant associated with human papilloma virus regarded as having a favorable prognosis, is more likely to be treated with surgery alone (70.7% vs 31.7%) and less likely to be treated with combination therapy (25.9% vs 40.4%) as compared with conventional SCC. No cases of V-SCC were treated with RT alone. Most cases of SCSC were treated with combination therapy (65.6%) and significantly fewer were treated with RT alone (3.1% vs 20.3%). These patterns are likely predicated on perceptions of each type of tumor’s relative aggressiveness.
V-SCC and P-SCC are found to have significantly better 5-year survival as compared with conventional SCC (69.7% and 61.87% vs 45.0%, respectively) possibly related to their predilection for exophytic growth and early detection. Sinonasal ASC has a markedly reduced 5-year survival compared with conventional SCC (15%). Although sinonasal V-SCC and P-SCC are found to have favorable prognoses, SCSC and ASC histologies appear to be poor prognostic indicators approaching statistical significance. With respect to advanced-stage disease, V-SCC, P-SCC, and B-SCC all are associated with improved prognoses as compared with conventional SCC. This is particularly interesting as B-SCC is considered by many to be the most aggressive subtype. Among those with early-stage disease, B-SCC and ASC, considered the most deadly subtypes, show significantly worse prognoses, although the confidence interval was wide in this study. Although nodal metastasis is an important prognostic indicator in ASC, scant data limited analysis in this regard.
Sinonasal Adenocarcinoma
Sinonasal adenocarcinoma (SNAC) is the most common primary malignancy of the ethmoid sinuses and associated with wood dust and other occupational exposures thought to afflict men more commonly than women. The SEER registry identified 1270 cases split almost evenly between men and women potentially explained by the decreased role of occupational exposures and by the exclusion of states with the highest concentration of woodworkers reporting to the SEER registry. The cohort is mostly older than 55 (59%), and white (78.0%).
With respect to Disease-free Survival (DSS), survival is 65.2%, 50.9%, 40.9%, and 36.5% at 5, 10, 15, and 20 years, with women having a slightly better, although statistically nonsignificant, survival rate compared with men. Black individuals demonstrate a worse survival than other racial groups, with a DSS of 53.6%, 38.5%, 30.0%, and 30.0% at 5, 10, 15, and 20 years. The best survival is seen among the “other” subgroup, which includes Asian, Native American, and Hispanic individuals, and unknown and mixed races. Survival is reduced with regional and distant spread and 5-year survival drops from 96.0% in patients with stage I disease to 79.6%, 86.2%, and 61.7% in stage II, III, and IV, respectively. The EUROCARE registry, made up of 67 population-based cancer registries in 22 European countries, identified 204 cases of ethmoid sinus adenocarcinoma afflicting mainly men (86%), except Scotland in which men made up a minority of cases (45%). Five-year net survival overall is 46% in this cohort, and is significantly better in those aged 55 to 64 years (60%) and worst among those older than 65 years (33%).
Of note, data on histologic subtype, namely intestinal and nonintestinal, are not available in the SEER registry, so its influence on survival could not be studied. However, nonintestinal SNAC is not thought to be associated with occupational exposures, tends to be localized, and is considered to have a favorable prognosis, although these could not be studied using available population-based registry data.
Esthesioneuroblastoma
Esthesioneuroblastoma (ENB), also known as olfactory neuroblastoma, thought to arise from the olfactory neuroepithelium, is known to have equal sex distribution with wide age distribution, and may affect the cervical lymph nodes. Jethanamest and colleagues used the SEER registry to identify 311 patients, although only 261 could be staged with available information. The mean age of diagnosis is 53 years with most patients between the ages of 40 and 70 years, favoring a unimodal distribution. Most are men (55%), white (83%), and most commonly presented with Kadish B (49.8%) followed by Kadish D (29.1%) disease. The most common site of involvement is the nasal cavity (77%), followed by the ethmoid sinuses (15.6%). The OS is 62.1 at 5 years and 45.6% at 10 years. Overall mean survival time drops from just over 12 years for Kadish A to 4.48 years for Kadish D. OS and DSS drops from 83.4% and 90.0% for Kadish A to 13.3% and 35.6% for Kadish D. Regional nodal involvement is seen in 12.3% of cases and this is associated with a significantly reduced DSS consistent with the existing literature. In line with current practice, combination therapy is associated with the longest duration of mean survival, followed by surgery alone, although the only statistically significant difference was between RT alone and multimodal therapy.
Platek and colleagues sought to assess the optimal treatment modality with regard to OS. Among 511 cases, patients received surgery only in 22% of cases, followed by radiation only in 11%, combination in 61%, and no treatment in 6%, with statistically significant difference in 5-year OS among the 4 groups, best for combination (73%) followed by surgery only (68%), radiation only (35%), and no treatment (26%), consistent with currently published protocols. With regard to pairwise comparisons among the treatment groups, the only significant differences were between combination therapy versus RT only and between surgery only versus RT only. When Cox-regression analysis was performed, the difference between surgery only and RT only was no longer significant, although all the differences were no longer statistically significant at 10 years. Importantly, comparisons by stage were not performed, and surgical treatment details are not available. Selection bias with respect to treatment planning is an obvious limitation, as the rationale behind treatment decisions are not captured by the information contained in the SEER database.
Histologic grade of ENB has been reliably reported in the SEER database only over the past 20 years, with grades I and II representing low-grade ENB and grades III and IV as high-grade ENB, although these do not necessarily correlate with a true Hyams grade. Tajudeen and colleagues analyzed a total of 281 patients in whom both grade and stage were reported. Low-grade tumors made up 48% of the cohort, with 52% having high-grade tumors. Advanced age, tumor grade, and modified Kadish stage are independent negative prognostic indicators of OS, whereas female sex independently predicts better OS. With respect to DSS, advanced tumor grade and modified Kadish stage are negative prognostic indicators, whereas radiation predicts better DSS. When looking at low-grade tumors, only surgery is predictive of DSS (hazard ratio [HR] 0.0135, P <.004) with no effect by RT on DSS or OS. With respect to high-grade tumors, neck disease, RT, surgery, and Kadish grade predict DSS. Multivariate analysis reveals modified Kadish stage (HR 2.025, P <.001), and RT (HR 0.433, P <.01) as independent predictors of DSS. Advanced tumor stage and modified Kadish stage are associated with worse DSS, whereas RT independently predicts better DSS with no influence of age or sex on DSS. These results suggest that for high-grade tumors, combination therapy is warranted, whereas for low-grade tumors, surgery with negative margins may be sufficient treatment. However, the role and timing of RT for local control requires further study.
Sinonasal Adenoid Cystic Carcinoma
Adenoid cystic carcinoma, known for its long clinical course and perineural invasion, is of particular clinical interest in the sinonasal tract given the potential for skull base invasion along neural pathways. A SEER registry study involving 412 cases of SN-ACC found both a female predilection and higher incidence among women, accounting for 57.52% of cases, in contrast to most other sinonasal malignancies, although this is consistent with ACC at other locations. SN-ACC occurs predominantly in patients older than 55 years and most are white (77.43%), followed by black individuals (11.41%), and primarily involves the paranasal sinuses (63.35%), followed by the nasal cavity and nasopharynx (19.66% and 16.99%, respectively). There has been a significant decline in incidence over the past 30 years among women. Incidence rates are significantly decreasing among all races, particularly among black individuals and “others” as compared with white individuals, which may reflect changing socioeconomic standards among minorities. OS at 20 years is 22.39%, with a trend for higher survival among women, although this is not statistically significant at 5, 10, and 20 years (69.29%, 50.43%, and 28.84% vs 67.87%, 44.97%, and 28.14% for men). Survival rates at 10 years are significantly lower in this study than prior population-based analyses that reported 10-year survival of 79.88%. White individuals have a significantly better survival rate as compared with black individuals, which is a trend previously reported for ACC at other head and neck sites. Unfortunately, data on histologic subtype, type of treatment, and surgical margin status are not available, limiting a population-based analysis of various treatment paradigms.
Sinonasal Mucoepidermoid Carcinoma
Mucoepidermoid carcinoma, the second most common sinonasal malignancy of salivary gland origin after adenoid cystic carcinoma, is still very rare, accounting for fewer than 0.1% of sinonasal neoplasms. A comparative analysis between sinonasal mucoepidermoid carcinoma (SN-MEC) and common salivary gland MEC (SG-MEC) using the SEER registry identified 4383 cases, of whom 149 were SN-MEC. Both SN-MEC and SG-MEC are similar with respect to sex ratio and racial distribution. Of the SN-MEC, 51% are men, mean age of diagnosis is 58.6 years, and mostly affects whites (78.5%). The most common sites of involvement are the maxillary sinus (51.7%), followed by nasal cavity (32.9%) and ethmoid sinus (11.4%). With respect to histology, SN-MEC is more likely to present with high histologic grade (42.3%) as compared with SG-MEC (25.5%), possibly reflecting a delay in diagnosis or differences in tumor biology. Better 5-year survival is seen with SG-MEC (84.1% vs 61.7%). Survival for high-grade SN-MEC is significantly lower than for low-grade histology (43.6% vs 78%). There is also a significantly higher survival rate in SG-MEC for low-grade and high-grade histology compared with similarly graded SN-MEC tumors. With respect to treatment, survival is best for SN-MEC treated with surgery alone (81.3%), followed by combination therapy (64.4%) and RT alone (25.6%). Although this trend is seen for SG-MEC also, survival was significantly better among the SG-MEC as compared with SN-MEC for each treatment modality. Only MEC in the submandibular gland fails to show a significantly higher survival as compared with SN-MEC. High-grade lesions involving the parotid gland demonstrate a significantly better survival compared with high-grade SN-MEC (60.7 vs 43.6, P = .026), consistent with previously reported prognostic value of tumor grade. Parotid and sublingual gland MEC demonstrated better survival for equivalent treatment as compared with SN-MEC.
Female gender has a lower hazard of death from SN-MEC compared with men, similar to SNAC and SN-ACC, whereas older age, involvement of the ethmoid sinus, and high-grade histology are associated with a greater hazard of death. Primary RT treatment and tumor size larger than 4 cm are associated with markedly increased hazards of death (HR 6.65 and 8.36, respectively) and corroborate surgery as the treatment of choice for this entity.
Nasopharyngeal Carcinoma
Nasopharyngeal carcinoma (NPC), although rare in nonendemic areas and linked to various food preservatives and EBV, often presents at an advanced stage due to subclinical growth. A SEER analysis of 1234 cases of nonkeratinizing (NK-NPC) and keratinizing (K-NPC) subtypes between 2004 and 2009 reveals a male predominance that is more pronounced for NK-NPC. A significantly lower age of diagnosis is seen for NK-NPC (54.2 years vs 60.8 years). Affected persons are mostly white individuals in K-NPC (72.8%), whereas Asian/Pacific Islander (API) individuals make up most K-NPC cases (48.7%). APIs constitute only 15.6% of the K-NPC cases. Incidence is noted to be increasing over the noted period for NK-NPC but not for K-NPC, possibly representing immigration trends. NK-NPC is more likely to present with grade III (poorly differentiated) histology as compared with K-NPC (62.7% vs 28.5%), with most K-NPC cases either well (12.1%) or moderately (35.4%) differentiated. Both subtypes present at an advanced stage (stage III and IV) and most patients have “regional” disease. The 5-year DSS for NK-NPC was significantly higher at 60.2% compared with 46.0% for K-NPC, which has a nearly twofold increased hazard of death. Female gender is associated with a higher 5-year relative survival (RS) for NK-NPC and significantly lower 5-year RS for K-NPC, whereas API is associated with a higher 5-year RS for both subtypes. However, the influence of EBV status, a known positive prognostic indicator, could not be determined in this study. Interestingly, survival difference by histologic grade is noted only for K-NPC, with grades III and IV demonstrating improved RS compared with grade I. For both cohorts, stage IV disease is a significantly poor prognostic indicator with markedly increased hazards of death. Consistent with prior studies, NK-NPC proved to be more radiosensitive and RT is associated with increased 5-year RS, although in both groups omission of RT is associated with worse outcomes. Importantly, the role of chemotherapy could not be determined, although consensus exists for treating advanced disease with combination therapy.
Sinonasal Undifferentiated Carcinoma
SNUC is a rare and aggressive malignancy of the small round blue-cell variety. A study of 318 cases from the SEER database shows a male preponderance at 62%, afflicting mainly whites (82.7%), with mean age of diagnosis at almost 58 years, although distribution between the ages of 20 to 80 is fairly even. Most patients are treated with radiation (79.6%) with 44.7% treated with combination therapy. Median survival for the entire cohort is 22.1 months, with 3 to 5 and 10- year survival at 44.3%, 34.9%, 31.3%, which has increased significantly over the study period. Those diagnosed after age 60 have significantly worse survival than all other age groups, with a 2.67 greater rate of mortality by age 80. Although survival improves after treatment with both RT and surgery alone, combination therapy is associated with a 42% reduction in mortality compared with those with single modality or no therapy. Importantly, both extent of surgery and treatment with chemotherapy are not assessed and SNUC is deemed sensitive to both induction and concurrent chemotherapy regimens.
Kuan and colleagues extrapolated staging by the modified Kadish system from the SEER registry and is the first population-based analysis to detect a difference in survival by tumor stage. A total of 328 cases were identified, with the most common subsites being the nasal cavity (29.3%), and maxillary (27.4%) and ethmoid sinuses (21.0%). Most present with modified Kadish stage C (51.2%) with regional and/or distant metastases identified in 18.9% of cases. The greatest number are treated with combination therapy (45.7%) and 42.7% are treated with either surgery or radiation. Median DSS is 2.9 years with 2-year, 5-year, and 10-year OS of 43%, 30%, and 25%, respectively. Older age, advanced Kadish stage, and larger tumor size are associated with worse OS and DSS. Treatment with combination therapy is associated with improved OS and DSS achieving a fourfold increase in survival.
In a subset analyses, improved DSS and OS are noted among Asian individuals as compared with white individuals and black individuals. There is also improved OS for tumors in the nasal cavity as compared with the maxillary sinus. Both OS and DSS were reduced with advanced tumor stages. Kadish A is not significantly different from other stages with respect to survival, although this is probably due to the low number of cases. Younger age, RT (and not surgery), and lower Kadish stage are independent positive prognostic indicators with respect to OS. This study, in contrast to prior case series, did not detect an association between nodal disease and survival, although this is possibly due to underreporting of nodal status in the SEER database (an overall incidence of 9.7% is reported).
Sinonasal Neuroendocrine Tumor
Sinonasal neuroendocrine carcinoma (SNEC) is a rare sinonasal malignancy accounting for fewer than 5% of tumors at this site. A total of 201 cases are identified using the SEER registry with a mean age at diagnosis of 55.8 years, mostly men (59.2%), and white (83.6%), followed by black individuals (8.0%). The most common location involved is the nasal cavity (40.8%), followed by the ethmoid sinuses (20.4%), maxillary sinuses (18.4%), and sphenoid sinuses (12.9%), consistent with prior studies. The middle turbinate is commonly cited as the most common site of origin, although at least 1 study notes the ethmoid sinus as the most common location. Most cases are grade III and IV (poorly differentiated, undifferentiated and anaplastic types), accounting for 62.7% of cases, although grade is missing in approximately 30% of cases. Staging information is available for 30% of cases and, of these, 70% were stage IV at diagnosis. The most common treatment is combination therapy (45.3%) followed by RT alone (14.9%) and surgery alone (14.9%).
There is a trend toward both increased incidence and survival rate over time, although these are not statistically significant. Overall 5-year survival was 50.8% with no difference with respect to sex or race. Involvement of the sphenoid sinus is associated with highest DSS at 5 years of 80.7% for unclear reasons, followed by the nasal cavity (59.2%), maxillary sinus (34.5%), and ethmoid sinus (33.3%). RS demonstrates the same trend being the highest for the sphenoid sinus (71.7%) followed by the nasal cavity (61.8%). Involvement of the maxillary and ethmoid sinus is associated with a greater hazard of death compared with nasal cavity involvement. The frontal sinus is not included in the analysis due to low numbers. As expected, survival is best for low-grade SNEC (5-year DSS 92.3%) compared with high-grade (5-year DSS 37%), although this was no longer significant after Bonferroni correction. Interestingly, those treated with surgery alone had the highest DSS at 69.0% compared with combination therapy (59.4%) and RT alone (39.9%).
Sinonasal Mucosal Melanoma
Sinonasal mucosal melanoma, comprising 4% of all head and neck melanomas, has frequently been combined with oral cavity melanoma in previously published case reports and single-institution retrospective studies to increase numbers at the expense of potentially obscuring important differences in survival. Gal and colleagues identified 304 patients using the SEER registry between 2000 and 2007, mostly women (56.3%) and whites (90.8%). The mean age of diagnosis was 71.2 years and the nasal cavity was the most commonly affected site (65.5%), followed by the maxillary sinus (15.1%). The primary treatment modality was surgery (81.6%) with or without postoperative RT. Overall 5-year survival was 24.2% with median survival 18 months for both sixth or seventh American Joint Committee on Cancer (AJCC) staging systems. The sixth and seventh editions upgraded mucosal melanoma T staging automatically to T3, eliminating T1 and T2 classifications, to emphasize depth of invasion as an important prognostic indicator as opposed to invasion of a specific anatomic site, as with other histologies. Each additional 10 years of age is associated with a 20% increased risk of death. With respect to treatment, there is a significant increase in survival with combination treatment and surgery alone as compared with radiation alone. However, no significant differences are detected between combination treatment and surgery alone, supporting the idea of mucosal melanoma as a surgical disease. Nodal disease is uncommon in this cohort (only 2 had neck disease), although this information was missing in almost 28% of patients. With respect to the AJCC staging changes, the seventh edition staging system did detect a slightly improved survival profile for stage IVA as compared with IVB or IVC that may represent a more accurate disease profile compared with the sixth edition staging.
Survival based on location reveals a higher DSS and RS for nasal cavity tumors at both 1 year (81.06%) and 5 years (36.66%). The lowest survival is seen in patients with multiple sinus involvement (1 year DSS 54.55%). Relative to nasal cavity tumors, there is a greater hazard of death for maxillary sinus (HR 1.34), ethmoid sinus (HR 1.60), and overlapping sinuses (HR 2.30), explained in part by tumors extending above “Ohngren’s line,” which pose a challenge to complete surgical resection. The sample sizes for frontal and sphenoid sinus disease are small and not included in the analysis, although all these patients died within a year of diagnosis. There are significantly more T4 tumors (44.13% vs 80.56%, P <.001) and TNM stage IV malignancies (51.05% vs 82.86%, P <.05) in the maxillary sinus compared with the nasal cavity, suggesting late diagnosis relative to nasal cavity tumors. With respect to treatment, those with maxillary sinus disease are more likely to receive RT. In patients with nasal cavity disease, patients treated with surgery alone and combination therapy have similar 5-year survival, although those treated with RT alone had significantly reduced 5-year survival. Interestingly, patients with maxillary sinus disease witness the best 5-year survival with surgical resection alone, and reduced survival with combination therapy (46.75% vs 19.63%), although this may be due to advanced disease, positive margins, and/or difficult surgical approaches in those who receive RT. These findings are consistent with current data supporting surgical resection with negative margins for optimal survival, with no established benefit of RT on survival.
Diffuse Large B-Cell Lymphoma
Diffuse large B-cell lymphoma (DLBCL) is one of the most common sinonasal lymphomas, presenting with nonspecific signs and symptoms across a wide age range. A SEER analysis of 852 patients between 1973 to 2009 shows a mean age at diagnosis of 65.8 years, slightly more men than women (54.5%), mostly white individuals (80.9%), and no significant change in incidence during the time period of study. The most common site of tumor involvement is the maxillary sinus (36.9%), consistent with prior studies, followed by the nasal cavity (34.0%), ethmoid sinus (8.7%), sphenoid sinus (4.1%), and frontal sinus (2.3%), with overlapping lesions in 4.3% of cases. Overall 1-year and 5-year survival is 83.7% and 68.7%, with no significant differences by age or sex. There are no statistically significant differences in survival between DLBCL of the sinonasal tract as compared with other head and neck sites. Only overlapping sinus involvement shows a significantly worse survival compared with tumors in the nasal cavity (HR 1.8, P <.05), although the literature suggests a worse prognosis for nasal cavity tumors, not seen in the SEER study. RT is associated with increased DSS and is noted to be a positive prognostic factor. The role of chemotherapy, considered a mainstay of treatment, on survival could not be assessed.
Extranodal Natural-Killer/T-cell Lymphoma
Extranodal natural-killer (NK)/T-cell lymphoma (ENKTL) nasal type is a rare form of non-Hodgkin’s lymphoma associated with EBV, more common in Asia and Latin America with less known about the clinical behavior of this entity in Western, non-Asian, and non-Hispanic populations. The California Cancer Registry identified 213 non-Hispanic white individuals (NHW) and API patients between 2001 and 2008. ENKTL has a higher incidence among APIs than NHWs but is still lower than rates seen in Asian countries. There is a significant increase in incidence among Hispanic men, possibly due to immigration patterns or environmental and lifestyle exposures unique to this population, such as pesticides or solvents. OS is uniformly poor for all racial groups with no association with age, sex, socioeconomic status, or human immunodeficiency virus status.
In a comparative study of sinonasal (SN-ENKTL) and extranasal (EN-ENKTL) disease, the investigators identified 241 cases of SN-ENKTL and 287 cases of EN-ENKTL in the SEER registry. There is a significantly greater male preponderance in cases of EN-ENKTL (72.47% vs 59.75%) with a slightly younger age of diagnosis for SN-ENKTL (49.9 years vs 53.8 years). White individuals make up the most among both types (more than 70%), followed by others (20%–25%) and black individuals (4%–6%). With respect to location, SN-ENKTL commonly involves the nasal cavity (80.50%), followed by the maxillary sinus (6.22%). B symptoms are more common at diagnosis among those patients with EN-ENKTL (38.41% vs 22.86%). SN-ENKTL more commonly presents at stage IE as compared with EN-ENKTL, which presents at stage IIIE/IV. EN-ENKTL has a lower DSS at 1 year and 5 years (41.72% and 28.33% vs 62.34% and 39.13%), possibly due to late diagnosis, and there is no association with gender, race, or histology. With respect to stage, SN-ENKTL has better 1-year and 5-year DSS with stage IE disease (76.59% and 53.99% vs 51.79% and 33.33%) and patients with advanced EN-ENKTL fare worse at 1 year and 5 years (26.48% and 14.63% vs 32.45% and 24.34%). Of note, patients with SN-ENKTL presenting at stage IIE have a nearly threefold increase in mortality than those with stage IE disease, with 5-year DSS dropping to 27.79%. Presence of B-symptoms impacts survival only in the EN-ENKTL group and is associated with a greater hazard of death (HR 1.66, P <.05). Both groups see improved survival with RT, although regression analysis finds withholding radiation only impacts prognosis in the SN-ENKTL group. However, when looking at early-stage disease, both groups have improved prognosis with RT.
Extramedullary Plasmacytoma
Extramedullary plasmacytomas (EMP) are monoclonal B-cell neoplasms that may arise in the head and neck in up to 80% of cases, with nearly two-thirds in the sinonasal tract. A recent SEER registry analysis compared sinonasal (SN-EMP) with other head and neck (HN-EMP) sites between 1973 and 2011. A total of 778 cases were identified, with 411 cases of SN-EMP. SN-EMP presents at a slightly later age than HN-EMP (61.3 years compared with 59.9 years), although there is a male predilection for both locations, which is stronger for SN-EMP (3.65:1 vs 1.87:1). Whites are more commonly afflicted in both types, comprising more than 80% of cases. The B-cell precursor type is the most common histologic type for both groups, followed by differentiated type and most are localized at time of diagnosis. For SN-EMP, the nasal cavity is the most commonly affected region (40.6%), followed by nasopharynx (31.6%) and maxillary sinus (15.5%). Most patients in both groups are most commonly treated with surgery and adjuvant RT, and next most commonly by RT alone. Survival outcomes are excellent in both groups, demonstrating comparable RS and DSS (88.2% at 5 years, 83.3% at 10 years for SN-EMP). Maxillary sinus involvement is associated with worse RS and DSS compared with nasal cavity and nasopharyngeal tumors. Of note, all 3 treatment modalities (surgery alone, RT alone, and combination therapy) have excellent and comparable 10-year DSS consistent with prior studies.
Sinonasal Sarcoma
As a whole, sinonasal sarcomas make up fewer than 10% to 15% of tumors in this location and historically have portended a poor prognosis and high mortality. The standard of treatment in other locations is wide resection, a luxury not often afforded in the sinonasal region. Between 1973 and 2008, a total of 352 patients were identified using the SEER database, mostly men (55%) with a mean age at diagnosis of 44 years, although there was even distribution among all age groups. Patients were predominately white (78%) with disease located in the maxillary sinus (58%). Sinonasal rhabdomyosarcoma (SN-RMS) is the most common histology (34%). Median survival time is 50 months, with 1-year, 5-year, and 10-year RS of 78.8%, 47.4%, and 38.1%, respectively. Men have significantly lower survival relative to women (42.5% vs 53.1% 5-year survival), with women having 30% lower hazard of death. Of the histologic subtypes, chondrosarcoma is associated with the lowest mortality rate with a 5-year survival of 64.4%. Relative to this group, the hazard of death from rhabdomyosarcoma and Kaposi sarcoma is more than 3 and 5 times greater, respectively. Sphenoid sinus involvement has the lowest mortality rate, whereas frontal sinus involvement has the highest. Fibrosarcoma, incredibly rare in the head and neck, and not explicitly studied in the previously referenced analysis, has an overall higher survival rate compared with other histologies at 5 years (OS 71.7%), with worse outcomes in patients receiving RT only. Sinonasal carcinosarcoma, also exceedingly rare, has a female predilection and poor 5-year OS of 48.5%, which is significantly lower than carcinosarcoma at other head and neck sites, excluding the salivary glands.
With regard to SN-RMS, common in the young pediatric population, more than 55% of cases are female, and involve the nasopharynx and nasal cavity most commonly. Interestingly, incidence among men and white individuals has increased over the past 20 years, whereas that of women and black individuals has decreased over this period. Survival, however, is consistently higher at 5 and 10 years in women at 54.88% and 52.83%, respectively, compared with men, at 36.9% and 36.9%, although this is markedly lower than RMS at other head and neck sites. White individuals also have higher survival rates compared with black individuals. Embryonal SN-RMS demonstrated the highest survival rates up to 20 years as compared with other histologic subtypes, although this was not statically significant.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
