Introduction
SWItch/Sucrose Non-Fermentable-related matrix associated actin-dependent regulator of chromatin subfamily B member 1 (SMARCB1)-deficient sinonasal carcinoma (SDSC) is a rare malignancy of the nasal cavity and paranasal sinuses, accounting for approximately 3–5 per cent of all sinonasal carcinomas.Reference Zhang, Gao, He, Xu and Lu1 Sinonasal carcinomas are a heterogenous group of rare tumours, often with high-grade or undifferentiated morphologies and an aggressive clinical course.Reference Taverna, Agaimy and Franchi2 With increasing molecular testing and profiling, novel molecular subgroups and genetic alterations have been identifiedReference Taverna, Agaimy and Franchi2, leading to the discovery of SDSC by Agaimy et al.Reference Agaimy, Koch, Lell, Semrau, Dudek and Wachter3 and Bishop et al.Reference Bishop, Antonescu and Westra4 in 2014.
SMARCB1, also known as Integrase Interactor 1(INI-1), is a core subunit of the SWItch/Sucrose non-fermentable (SWI/SNF) complex, responsible for regulating cellular differentiation and proliferation by promoting tumour suppressor genes.Reference Agaimy and Bishop5 In the head and neck region, SMARCB1 and SMARCA4 (BRG1) have been identified as the main SWI/SNF components underlying aggressive malignancies. Molecular studies have confirmed the paucity of additional genetic abnormalities in these malignancies, highlighting the central role of SWI/SNF deficiency as the primary genetic driver.Reference Agaimy and Bishop5 Furthermore, patients with SDSC often present at an advanced stage with frequent multi-sinus involvement.Reference Mittal, Nagar, Patil, Rane, Nisarga and Rabade6
However, due to its rarity, diagnostic and treatment challenges persist. In recent years, there have been attempts to systematically review the literature on SDSC and meta-analyses have been performed to determine the clinical profile of SDSC as well as to evaluate treatment options and survival data on patients with SDSC. The recent systematic reviews and meta-analyses performed by Wang et al.Reference Wang, Wang, Fang, Zhong, Hou and Ma7 and Lee et al.Reference Lee, Tsang, AWI, Chan, Chung and Tong8 in 2022 purely analysed the existing information on SDSC, while a more recent individual patient data (IPD) meta-analysis by Vuong et al.Reference Vuong, Le, TTB, Le, El-Rassi and McKinney9 in 2023 investigated sinonasal carcinomas as a whole, with a subset of SDSC included in their analysis. Since then, multiple new large cohort studies on SDSC have been published because of increasing awareness and diagnosis of SDSC. Hence, we performed an updated systematic review with pooled descriptive summaries and an IPD survival analysis on SDSC, to more accurately identify the clinicopathological features of this disease, as well as to determine survival outcomes and treatment effectiveness. The review question and inclusion criteria were defined using the PICO (Population/Problem, Intervention, Comparison and Outcome) framework (Table 1).
Review question and inclusion criteria using PICO framework
Table 1 Long description
The table defines a PICO framework for selecting studies about SMARCB1-deficient sinonasal carcinoma. The population is patients aged 18 years and older with a confirmed diagnosis of this cancer. The intervention includes any therapeutic approach, without restricting treatment type. There is no control or comparison group specified, so studies may be single-arm or descriptive. Outcomes of interest include clinicopathologic characteristics such as tumour histopathology and TNM staging, along with treatment modalities and survival outcomes. Because no comparator is required, the table supports broad inclusion but limits direct treatment-effect comparisons across groups.
PICO = Population/Problem, Intervention, Comparison, and Outcome.
Materials and methods
Protocol and registration
We conducted this systematic review according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement for standard protocols,Reference Liberati, Altman, Tetzlaff, Mulrow, Gotzsche and Ioannidis10 where the Prisma Flowchart can be seen in Figure 1. This review was previously registered on PROSPERO (CRD 42024606903).
PRISMA flow diagram. PRISMA, Preferred Reporting Items for Systematic reviews and Meta-Analyses.
Figure 1 Long description
The flowchart outlines the process of study selection in a systematic review. In the identification stage, 646 studies were sourced from databases like Embase, Scopus, Web of Science and PubMed, along with 3 references from other sources. After removing 402 duplicates, 247 studies were screened. During screening, 117 studies were excluded, leaving 130 studies sought for retrieval. Of these, 11 were not retrieved and 119 were assessed for eligibility. In this stage, 49 studies were excluded for reasons such as being abstract only, repeated data, inadequate data, not in English, no isolated data, wrong publication type, or not on SMARCB1-deficient sinonasal carcinoma. Finally, 70 studies with 370 patients were included, comprising 66 studies with individual data and 4 without. Additionally, 2 patients treated at the institution with complete data were included, resulting in 372 patients in the review.
Information sources and search strategy
References for this systematic review were identified through the search of five databases: PubMed, Cochrane, Embase, Web of Science and Scopus. We used the search terms “SMARCB1 OR SMARCB-1 OR Swi-Snf-Related Matrix-Associated Actin-Dependent Regulator Of Chromatin Subfamily B Member 1 OR Swi Snf Related Matrix Associated Actin Dependent Regulator Of Chromatin Subfamily B Member 1 OR INI1 OR INI-1 OR Integrase interactor 1 protein”, “sinonasal OR paranasal OR nasal OR pananasal sinus*” and “cancer OR carcinoma OR malignanc* OR tumour* OR tumor OR neoplasia OR neoplasm”. The search date started from the inception of these databases to 24 May 2025. The abstract and full text review were carried out by two authors (O.T.W.H. and B.L.S.Y.) independently. In addition, we included two cases of patients with SMARCB1-deficient sinonasal carcinoma treated at our institutions under the National Healthcare Group in the subsequent analysis. Institutional Review Board approval was not required for this study.
Eligibility criteria
Case reports, case series, observational studies, cross-sectional studies as well as any interventional trials were included; ideas, editorials, conference abstracts and perspectives, records not published as full-length articles in peer reviewed journals and non-English publications without available translations were excluded. Study populations included were individuals with diagnoses of SDSC, while patients whose data were repeated in subsequent publications were excluded.
Data extraction
Extracted information include cohort size, age, gender, clinical presentation, TNM staging, tumour site, microscopic features, immunohistochemical findings, genetic testing, follow-up duration, treatment modalities, survival outcome and status. With the included full texts, we also identified studies with individual patient data for our survival analysis. These studies had to contain information on individual patient demographics, follow-up duration, survival outcome, TNM staging and treatment modality. For studies without these data, attempts were made to obtain the individual data from the respective authors, failing which these studies were not included in the eventual survival analysis.
Statistical analysis
All statistical analyses were performed using using R 4.1.0 (R Core Team, 2021), the rstanarm (v2.13.1; Gabry & Goodrich, 2016) and the report (v0.2.0; Makowski, Patil, & Lüdecke, 2019) packages. A two-sided p-value of less than 0.05 indicated statistical significance. We analysed the overall survival (OS) of patients with individual patient data by Kaplan–Meier methods. OS was defined as the time from diagnosis to death from any cause, with patients censored at the last reported follow-up if alive. Comparison of OS between subgroups was performed via log rank tests. Cox proportional hazard models were used in the univariable and multivariable analyses of prognostics factors of OS.
Risk of bias assessment
Risk of bias assessment was performed using the Joanna Briggs Institute (JBI) critical appraisal tool for case reports and case series.Reference Munn, Barker, Moola, Tufanaru, Stern, McArthur, Stephenson and Aromataris11 These tools evaluate methodological quality across key domains, including clarity of patient demographics and clinical history, validity of diagnostic methods, completeness of intervention and outcome reporting and adequacy of follow-up. Each included study was independently appraised across eight criteria, and items were rated as “Yes”, “No”, “Unclear” or “Not applicable”. Overall risk of bias was determined based on the percentage of “Yes” responses, with a less than 50 per cent of “Yes” responses deemed high risk of bias, 50–67.5 per cent deemed moderate risk of bias and greater than 67.5 per cent deemed low risk of bias. Disagreements were resolved through discussion to ensure consistency in assessment.
Results
The study selection process is summarised in Figure 1. A total of 646 studies were identified from the database, of which 402 duplicates were removed. Of the 130 full texts sought for retrieval, 11 could not be retrieved, and 49 were excluded due to duplicate patient data, incorrect study types, a lack of isolated data or a non-English language. After full text screening, a further 49 studies were excluded; thus, a total of 70 studies from the literature were included in this systematic review.Reference Zhang, Gao, He, Xu and Lu1, Reference Bishop, Antonescu and Westra4, Reference Mittal, Nagar, Patil, Rane, Nisarga and Rabade6–Reference Lee, Tsang, AWI, Chan, Chung and Tong8, Reference Agaimy, Hartmann, Antonescu, Chiosea, El-Mofty and Geddert12–Reference Zhao, Du, Zhang, Yang, Zhou and Xiao76 Of the 70 included articles, four studies did not have individual patient data and were thus not included in the survival analysis.Reference Contrera, Shakibai, Su, Gule-Monroe, Roberts and Brahimaj22, Reference Cruz-Tapia, Cano-Valdez, Meneses-Garcia, Correa-Arzate, Molotla-Fragoso and Villagómez-Olea23, Reference Wang, Wang, Tang, Wang, Li and Song68, Reference Woods, Scholfield, Axiotakis, Fitzgerald, Adilbay and Cracchiolo70 Additionally, we included two patients from our own institution with data on baseline characteristics, treatment details and survival outcomes, making up a total of 372 patients in total for this systematic review. Summary of the included studies as well as the risk of bias assessment can be found in Supplementary Table S1.
Baseline characteristics and clinical presentation
The baseline characteristics of the included patients are shown in Supplementary Table S2. The median age was 54 (range 10–95) of 275 patients with individual age data. There was a male sex preponderance, where 64.8 per cent of the patients were male and 34.7 per cent female.
As seen in Supplementary Table S2, symptoms and signs of SDSC were wide-ranging and poorly localised, with facial, dental, oral, nasal, ocular and even neurological presentations. The most common group of symptoms were nasal symptoms (n = 97; 25.5%), with the most common symptom being epistaxis (n = 52; 14.0%), followed by nasal obstruction (n = 36; 9.7%). Ocular symptoms were also common (n = 79; 25.5%), with proptosis being the most common (n = 22; 5.9%), followed by neurological symptoms (n = 36; 9.7%), with headaches being the most common (n = 24; 6.45%).
Tumour characteristics
Tumours tended to be locally aggressive, where most were T4 at initial staging (n = 257; 69.1%). More than half of the patients did not exhibit locoregional or distant metastasis: 211 patients (56.7%) were N0 and 194 were M0 (52.2%). The full breakdown of the various TNM staging categories can be found in Supplementary Table S3. For the N staging, studies which did not specify the exact N staging or where N staging was unknown were categorised as Nx/Unknown. Only studies which specified exact N staging were included in the subsequent analysis of prognostic factors and grouped under N+. The majority of SDSC was found in the nasal cavity/septum (n = 139; 37.4%), followed by the ethmoid sinus (n = 124; 33.3%). Many had extension to the orbits (n = 52; 14.0%), in keeping with the high frequency of orbital symptoms as described above. A significant proportion also had skull base (n = 26; 7.0%) and intracranial (n = 23; 6.2%) involvement as seen in Supplementary Table S3.
Histology and cytomorphology
Detailed histological and cytomorphological features of the tumours were reported in 217 patients (Supplementary Table S5), revealing a large variety of morphological subtypes. Of note, there were two key groups reported, with the most common being the basaloid subtype (n = 102; 27.4%), followed by the plasmacytoid/rhabdoid subtype (n = 58; 15.6 per cent).
Genetic testing and immunohistochemistry
Immunohistochemistry was used to confirm loss of SMARCB1 expression in the reported cases, but several other studies also used additional genetic testing methods to confirm the deletional status of SMARCB1. Fluorescence in-situ hybridisation (FISH) was used in 72 patients,Reference Agaimy, Hartmann, Antonescu, Chiosea, El-Mofty and Geddert12, Reference Allison, Bishop and Ali14, Reference Dogan, Cotzia, Ptashkin, Nanjangud, Xu and Momeni Boroujeni24, Reference Kezlarian, Lin and Dogan38, Reference Laco, Chmelařová, Vošmiková, Sieglová, Bubancová and Dundr40, Reference Li, Han, Xu, Wu, Lin and Cao43, Reference Skálová, Taheri, Bradová, Vanĕček, Franchi and Slouka62 targeted exome sequencing in 19 patients,Reference Dogan, Cotzia, Ptashkin, Nanjangud, Xu and Momeni Boroujeni24, Reference Kezlarian, Lin and Dogan38, Reference Skálová, Taheri, Bradová, Vanĕček, Franchi and Slouka62, Reference Tamaki, Teruya, Hirakawa, Tomita and Wada65 next generation sequencing (NGS) in 15 patientsReference Skálová, Taheri, Bradová, Vanĕček, Franchi and Slouka62, Reference Tamaki, Teruya, Hirakawa, Tomita and Wada65 and multiplex ligation-dependent probe amplification in eight patients.Reference Libera, Ottini, Sahnane, Pettenon, Turri-Zanoni and Lambertoni45
Treatment and survival status
Supplementary Table S4 shows the breakdown of the various treatment regimens in the included studies. Treatment modalities were diverse, with most patients being treated by surgery with concurrent or adjuvant chemotherapy and/or radiotherapy (n = 148). Neoadjuvant chemotherapy in conjunction with surgery and adjuvant therapy was utilised in 27 patients. A total of 60 patients were treated non-surgically with either chemotherapy or radiotherapy or a combination of both. Notably, tazemetostat was suggested as a novel treatment drug in three patients.Reference Qayoumi, Mushtaq, Jansen, O’Reilly and Jamaluddin52, Reference Saleh, Classe, Nguyen, Moya-Plana and Even53 Immune checkpoint inhibitors (ICIs) were also used in six patients, of which four were started on programmed cell death protein 1 (PD-1) and programmed death-ligand 1(PD-L1) inhibitor therapy.Reference Zhang, Gao, He, Xu and Lu1, Reference Kang, Koh, Kim, Han, Won and Kim37, Reference Yang, Zhou, Zhong, Li and Wang73, Reference Zhao, Du, Zhang, Yang, Zhou and Xiao76
Treatment outcomes of patients from our institution
We included two cases of SDSC from the National Healthcare Group Singapore. The first case is a 44-year-old Chinese female, diagnosed with Stage IVC cT4aN0M1 right nasal cavity and ethmoid sinus SDSC. She first presented with epistaxis, and nasoendoscopy showed a mass arising lateral to the right middle meatus. Histology of the mass revealed SDSC, with locoregional and staging scans showing skull base involvement with lymph node and bony metastases. The patient first received chemoradiotherapy but developed tumour recurrence and chemotherapy side effects. She eventually underwent multiple repeat surgeries to resect the tumour and is currently still on follow-up.
The second case is that of a 79-year-old female who first presented with epistaxis, headaches, eye pain and blurring of vision. Nasoendoscopy showed blood streaks tracking posteriorly from the right middle meatus but no masses were seen. Computed tomography (CT) and MRI scans showed opacification and inflammatory changes of the paranasal sinuses as well as signs suggestive of orbital and intracranial involvement. The patient underwent functional endoscopic sinus surgery (FESS) with an intial diagnosis of chronic sinusitis, but intra-operative biopsies revealed SDSC. Further imaging did not reveal distant metastases, and the patient underwent chemoradiotherapy, but subsequently passed away from the disease. The full details of the two cases of SDSC from our institution can be found in the Supplementary Materials.
Survival analysis
The median follow-up duration across included studies was 14.5 months (range, 0.3-180), of which 32.5 per cent of the patients died. To provide a more accurate estimate accounting for censoring, we also report the median follow-up duration using the reverse Kaplan–Meier method, which was 25.1 months. In the included studies, 196 patients had individual data on survival outcomes and were thus used in our survival analysis. The one-year, two-year and three-year overall survival of the patients was 72.6 per cent, 58.1 per cent and 49.7 per cent, respectively, with a median overall survival of 32 months (95 per cent confidence interval [CI] 26-73) as shown in Figure 2. Females had a longer overall survival of 56.9 months compared to males (Figure 3), while patients with T4 disease had a shorter overall survival of 30 months compared to 43 months for patients with T1-3 disease as seen in Figure 4. The rest of the Kaplan–Meier estimates stratified by gender and T stage of the disease were not statistically significant. Survival outcomes were also stratified based on various treatment groups (Table 2), mainly surgery with radiotherapy and/or chemotherapy, definitive chemoradiotherapy, surgery alone and either chemotherapy or radiotherapy alone. Patients who had definitive chemoradiotherapy had the longest median OS of 56.9 months, followed by surgery with radiotherapy and/or chemotherapy with a median OS of 42 months.
Kaplan–Meier curve showing three-year overall survival.
Figure 2 Long description
3-Year Overall Survival A Kaplan–Meier plot with one stratum labeled All Patients. The x-axis is labeled Time (months), with tick labels at 0, 6, 12, 18, 24, 30 and 36. The y-axis is labeled Survival Probability (percent), with tick labels at 0 percent, 25 percent, 50 percent, 75 percent and 100 percent. A single step curve starts near 100 percent at time 0 months and decreases in small steps across the time range to 36 months. A horizontal dashed line is drawn at 50 percent survival probability. A vertical dashed line is drawn at 32 months, meeting the 50 percent line. Text inside the plot states: Median: 32 month (95 percent CI: 26-73). 12-month OS: 72.6 percent. 24-month OS: 58.1 percent. 36-month OS: 49.7 percent. A risk table below the plot is labeled Number at risk. The row label is All Patients. Values shown at 0, 6, 12, 18, 24, 30 and 36 months are 196, 163, 118, 84, 68, 45 and 33.
Kaplan–Meier curve showing overall survival stratified by gender.
Figure 3 Long description
The x-axis represents Time (months) from 0 to 36. The y-axis represents Survival Probability from 0 percent to 100 percent. A legend at the top shows Gender with two entries: Female and Male. A text label at the top right reads p equals 0.083. Two step lines are plotted. Text inside the plot area reads: Female: 56.9 month (95 percent CI: 28 dash NA) Male: 30 month (95 percent CI: 21 dash 71) A horizontal dashed line is drawn at 50 percent on the y-axis. A vertical dashed line is drawn at 30 on the x-axis. Below the graph is a table titled Number at risk. The left side label reads Strata, with rows Female and Male. The table lists values at Time (months) 0, 6, 12, 18, 24, 30, 36. Female: 64, 54, 43, 33, 27, 17, 14. Male: 129, 106, 73, 50, 40, 28, 19.
Kaplan–Meier curve showing overall survival stratified by T1-3 versus T4 stage disease.
Figure 4 Long description
The x-axis represents Time (months) from 0 to 36. The y-axis represents Survival Probability from 0 percent to 100 percent. A legend at the top shows T Stage with two entries: T1-3 and T4. Text on the right reads p equals 0.424. A horizontal dashed line is drawn at 50 percent on the y-axis. A vertical dashed line is drawn at 30 on the x-axis. Text inside the plot area reads: T1-3: 43 month (95 percent CI: 26 minus NA) T4: 30 month (95 percent CI: 24 minus 81.4) Two step lines with censor marks are shown. A table below the plot is titled Number at risk. The left label reads Strata. Two rows are labeled T1-3 and T4. The x-axis below the table is labeled Time (months) with tick labels 0, 6, 12, 18, 24, 30, 36. Number at risk values: T1-3: 31 at 0, 29 at 6, 22 at 12, 16 at 18, 13 at 24, 10 at 30, 8 at 36. T4: 154 at 0, 125 at 6, 88 at 12, 62 at 18, 51 at 24, 34 at 30, 24 at 36.
Median overall survival and time-specific Kaplan–Meier survival estimates (6–36 months) by treatment group
Table 2 Long description
The table reports Kaplan–Meier overall survival outcomes by treatment group, including numbers at risk at 6 to 36 months, median overall survival with ranges, and survival probabilities with 95 percent confidence intervals. Median overall survival is highest with definitive CRT at 56.9 months, followed by surgery plus RT with or without CT at 42.0 months, then surgery alone at 21.0 months, and lowest with RT or CT alone at 8.0 months. At 6 months, survival probability is about 0.90 for surgery plus RT with or without CT, 0.85 for definitive CRT, 0.70 for surgery alone, and 0.54 for RT or CT alone. By 12 months, survival probability declines to about 0.78 for surgery plus RT with or without CT, 0.76 for definitive CRT, 0.56 for surgery alone, and 0.43 for RT or CT alone. From 18 to 36 months, surgery plus RT with or without CT decreases from about 0.66 to 0.50, while definitive CRT is about 0.66 at 18 months and about 0.50 at 30 and 36 months. Surgery alone is about 0.56 at 12 and 18 months and about 0.48 at 24 to 36 months. For RT or CT alone, survival probabilities are only provided through 18 months, and numbers at risk drop to 2 at 12 months and 1 at 18 months, with later time points listed as not available. Interpretation should consider the small at-risk counts in some groups and the missing later estimates for RT or CT alone, which can make comparisons less stable.
Univariable analysis was performed for some of the prognostic factors of SDSC, which showed that N+ disease (hazard ratio [HR] = 1.95; 95% CI 1.03–3.72; p = 0.04) was a statistically significant poor prognostic factor for overall survival, while surgical treatment (HR = 0.50; 95% CI 0.26–0.96; p = 0.04) was a favourable prognostic factor. Multivariable analysis was also performed adjusting for age, gender as well as the variables that were statistically significant in the univariable analysis. Treatment with either chemotherapy or radiotherapy alone was also shown to be a significant poor prognostic factor for overall survival on both univariable (HR = 5.84; 95% CI 2.27–15.04; p < 0.001) and multivariable (HR = 6.06; 95% CI 2.28–16.08; p < 0.001) analyses. Surgery and treatment group could not be adjusted in the multivariate Cox regression model for treatment related variables due to collinearity, despite showing statistical significance in the univariable analysis. Furthermore, for non-treatment related variables, only treatment group was adjusted in the multivariable analysis. Including both the composite variable (treatment group) and individual treatment modalities (surgery) in the same model would introduce multicollinearity, which inflates variance and undermines the stability and interpretability of the hazard regression coefficients. For treatment-related variables (including surgery), multivariate analysis only adjusted for age, gender and N staging, which was the only other variable statistically significant in the univariable analysis. For non-treatment related variables, multivariate analysis adjusted for age, gender, N staging as well as treatment group (p < 0.001). The sample sizes for T1 category (n = 3) and frontal sinus primary tumour category (n = 3) were too small for regression and, thus, were omitted in the analysis. Multivariable analysis showed that patients with SDSC with the maxillary sinus as a primary site was had significantly poorer prognosis (HR = 3.29; 95% CI 1.11–9.73; p = 0.03). The full details of the analyses can be found in Table 3.
Univariable and multivariable analysis of variables prognostic for overall survival
Table 3 Long description
Hazard ratios for overall survival are reported for multiple covariates using univariable and adjusted multivariable models, with confidence intervals and p-values. The strongest and most consistent association is treatment: RT or CT alone shows about a sixfold higher hazard than surgery plus RT with or without CT in both analyses, with a highly significant p-value. N-positive disease (N1 to N3 versus N0) is linked to worse survival in univariable analysis but is not significant after adjustment. Primary tumor site is mostly not significant, but maxillary site shows higher hazard after adjustment (adjusted hazard ratio about 3.3) with a significant p-value. Age and sex show small to moderate hazard increases but do not reach conventional significance in either model. Surgery appears protective in univariable analysis but is not significant after adjustment, and other therapy-related variables (adjuvant or neoadjuvant) are not significant. Some categories have very small sample sizes and were not suitable for regression, and the adjusted model includes only selected covariates, so results should be interpreted with those limitations in mind.
* Sample size for T-category (T1) (n = 3) and frontal sinus (n = 3) too small for regression
** Multivariate analysis was adjusted for age, sex, and variables that were significant in the univariable analysis (p < 0.05), including N category and treatment group (only for non-treatment related variables). Surgery was excluded from multivariate analysis due to collinearity with treatment group.
Risk of bias assessment
The full breakdown of the risk of bias assessment done for the included studies using the JBI critical appraisal tool can be found in Supplementary Table S1. Among the included studies, six studies had a moderate risk of bias (8.6%), while 64 studies had a low risk of bias (91.4%). None of the included studies had a high risk of bias. The main domain driving risk of bias was that of whether the post-intervention clinical outcome was clearly described, with only 38 (54.3%) “Yes” responses, as well as whether the diagnostic tests or assessment methods and results clearly described, with 43 (61.4%) “Yes” responses.
Discussion
Before the 4th edition of the World Health Organisation (WHO) Classification of Head and Neck Tumours, SDSC was not considered a distinct subtype, and many cases were merely labelled as either sinonasal undifferentiated carcinomas (SNUC) or poorly differentiated carcinomas (PDCA).Reference Lee, Tsang, AWI, Chan, Chung and Tong8, Reference Vuong, Le, TTB, Le, El-Rassi and McKinney9, Reference Stelow and Bishop77 However, in recent years, with increasing awareness of SDSC as a separate subtype of sinonasal carcinomas, more studies with larger cohorts have been published by many authors around the world.
Compared to other types of sinonasal carcinomas such as SNUC, treatment strategies and prognosis may differ. The study by Chitguppi et al. reported that patients with SDSC had poorer OS and disease-free survival as well as higher recurrence and mortality rates compared to patients with SNUC with retained SMARCB1 expression.Reference Chitguppi, Rabinowitz, Johnson, Bar-Ad, Fastenberg and Molligan20 A recent meta-analysis by Burggraf et al. reported that the five-year OS for 192 cases of SNUC was 50.5 per cent, which was already slightly higher than the three-year OS of SDSC cases in our study of 49.7 per cent.Reference Burggraf, Schiele, Thölken, López, Elawany and Zenk78 The study also revealed that patients who underwent induction chemotherapy had significantly better five-year OS compared to those that did not. Of interest, they also found that surgical treatment had no effect on OS and that patients who underwent surgery had a significantly higher risk of death compared to those who underwent induction chemotherapy with chemoradiotherapy.Reference Burggraf, Schiele, Thölken, López, Elawany and Zenk78 As such, patients with SDSC may have poorer prognosis compared to other forms of SNUC, and surgery may confer a more favourable prognosis in patients with SDSC compared to SNUC where non-surgical treatment may be preferred. This highlights the need for prompt and accurate diagnosis and distinction of SDSC from other types of SNUC to ensure that these patients receive appropriate treatment for SDSC expediently. Moreover, identification of SMARCB1 deficiency has growing therapeutic relevance, as these tumours may be candidates for targeted therapies, which may not be considered for other forms of SNUC. Thus, due to the differences in prognoses between SDSC and other types of SNUC as well as the variations in treatment modalities, there is a need to differentiate SDSC from other types of sinonasal carcinoma to ensure that patients with SDSC receive prompt initiation of treatment regimens that have been shown to be effective in SDSC.
From our review, patients with SDSC most presented with nasal symptoms such as epistaxis and nasal obstruction. The most common sites where SDSC developed, in more than 70 per cent of the patients, were in either the nasal cavity or ethmoid sinuses. Furthermore, the orbits were the most common site of extension of the tumour, possibly explaining the high rate of ocular complications reported in the included studies. Compared to the previous systematic reviews, our study decided to analyse whether the primary tumour site of SDSC would affect prognosis. Based on our multivariable analysis, patients with SDSC originating from the maxillary sinus had significantly poorer prognosis (HR = 3.14; 95% CI 1.08–9.14; p = 0.04) compared to tumours arising from other sinuses or the nasal cavity. We postulate that this might be due to the concealed location of the maxillary sinus which makes diagnosis in the early stage difficult due to the lack of signs and symptoms the patient may have.Reference Nishio, Fujimoto, Hiramatsu, Maruo, Tsuzuki and Mukoyama79 Furthermore, tumours in the maxillary sinus often invade into the orbit, base of skull and other important surrounding structures in the later stages, making complete resection challenging and thus conferring a poorer prognosis.Reference Shi, Wang, Hou, Huang, Ma and Zhang80 However, this result could be confounded by other important variables such as orbital/skull base involvement and primary tumour resectability, which unfortunately were not available among the included studies for further sensitivity analyses; as such, this finding cannot be readily proven.
SMARCB1 is a tumour suppressor gene located on chromosome 22q11.2. Its deactivation has been found to be the culprit of carcinogenesis in multiple neoplasms including rhabdoid tumours and epithelioid sarcomas.Reference Stelow and Bishop77 The SMARCB1 gene acts as a tumour suppressor, and its absence alters SWI/SNF complex function, leading to increased enhancer of zeste homolog 2 (EZH2) activity and upregulation of oncogenic pathways and suppression of tumour suppressor gene transcription.Reference Chung, Kenee, Mitton and Halderman21 Diagnosis of SDSC based on histology has proven to be challenging not only due to its rarity, but also because it can mimic numerous other types of sinonasal carcinomas.Reference Mittal, Nagar, Patil, Rane, Nisarga and Rabade6 In our review, the most common histological and cytomorphological feature was that of the basaloid subtype, but there were many other different subtypes reported. Therefore, there is a need for other diagnostic methods other than routine histology and cytomorphology to differentiate SDSC from other sinonasal carcinomas. Thus, immunohistochemical staining has been recommended as the routine diagnostic method for SDSC as it is able to detect the biallelic inactivation of the gene resulting in loss of SMARCB1 protein expression.Reference Agaimy and Bishop5 In recent years, other diagnostic methods have been used to diagnose SDSC including FISH and NGS, although studies have reported that these techniques would be more useful to identify suitable targets for tailored therapy.Reference Skálová, Taheri, Bradová, Vanĕček, Franchi and Slouka62 In our review, we also noted that p40, p63 and cytokeratin AE1/AE3 often stained positive. Agaimy et al. previously reported that the basaloid subset of SDSC could show p63 positivity on immunohistochemistry, which may falsely point to a diagnosis of NUT midline carcinoma or non-keratinising/basaloid squamous cell carcinoma (SCC), as the presence of p63 would support squamous differentiation.Reference Agaimy, Hartmann, Antonescu, Chiosea, El-Mofty and Geddert12 Thus, even in the presence of other positive immunohistochemical markers and basaloid features on histology, immunohistochemistry to look for loss of nuclear SMARCB1 expression should still be performed to exclude a diagnosis of SDSC.
As seen in our analysis, 69.1 per cent of the patients presented with T4 stage disease at initial presentation. Furthermore, 9.9 per cent of the patients presented with nodal disease on initial diagnosis, while 12.9 per cent of the patients were found to have distant metastases on staging scans. However, the number of patients with undetermined TNM staging from the included studies were of a substantial amount; thus, it is possible that the percentage of patients with locoregional and distant metastases may even be higher. It is interesting to note that, despite the low incidence of nodal disease on initial diagnosis among the included studies, our analysis showed that N+ disease was a significant poor prognostic factor for overall survival. Clinically, this would mean that thorough assessment for nodal disease at time of diagnosis is crucial, as evidence of nodal disease would confer poorer prognosis and consequently affect subsequent management. At present, the low incidence of nodal disease at time of diagnosis may not provide sufficient grounds for elective neck dissection, especially given the high morbidity associated with this procedure. However, given the marked adverse prognostic impact of N+ disease, there may be a case for elective neck management especially in selected high-risk patients, such as those with advanced T staging. Nevertheless, further studies with larger cohorts and multivariate analyses are needed to better define risk factors for regional metastasis and to clarify the role of elective neck management in patients with SDSC.
In terms of treatment, multiple studies have suggested that multimodality treatment should be the mainstay for patients with SDSC.Reference Lee, Tsang, AWI, Chan, Chung and Tong8 In our study, we compared different types of unimodality and multimodality treatment strategies to determine which ones would have a significant impact on overall survival. We suggest that surgery should be considered as the first-line treatment for lower T stage tumours. Our analysis revealed that surgical treatment conferred significantly better prognosis in patients with SDSC compared to non-surgical treatment, which differs from the previous systematic reviews. Previous studies have also recommended upfront resection of SDSC due to its aggressive nature, despite it being technically difficult due to the tumour’s advanced presentation and extensive tumour invasion.Reference Al-Mamgani, van Rooij, Mehilal, Tans and Levendag81 Compared to the previous systematic reviews, our study is the first to provide evidence that patients with SDSC who only underwent radiotherapy or chemotherapy alone had significantly poorer prognosis compared to surgery with chemotherapy and/or radiotherapy, and this was consistent in both univariable and multivariable analyses. Our analyses also suggest that a treatment strategy of surgery with chemotherapy and/or radiotherapy conferred a more favourable prognosis compared with definitive combined chemoradiotherapy or surgery alone, although these were not statistically significant. As such, we suggest that, in the treatment of SDSC, isolated chemotherapy or radiotherapy alone should not be pursued and that surgery combined with chemotherapy and/or radiotherapy may be considered as a first-line therapy. More studies will need to be done to compare the efficacy of surgery alone or combined chemoradiotherapy compared to surgery with chemotherapy and/or radiotherapy to determine which treatment modality confers the best prognosis.
For locally advanced or unresectable disease at presentation, neoadjuvant therapy may be considered to determine chemosensitivity before deciding on surgery versus definitive chemoradiotherapy. Lee et al. suggests offering definitive chemoradiation to patients who had greater than 50 per cent of tumour shrinkage after induction chemotherapy as it implies greater chemosensitivity, while reserving definitive surgery with post-operative adjuvant treatment for those with less than 50 per cent tumour shrinkage.Reference Lee, Tsang, AWI, Chan, Chung and Tong8 Although statistically insignificant, in our study patients who underwent neoadjuvant therapy appeared to have better prognosis (HR = 0.66; 95% CI 0.26–1.68; p = 0.39) than those who did not, and those patients who received definitive chemotherapy and/or radiotherapy after neoadjuvant therapy seemed to have poorer prognosis compared to those who received definitive surgery instead (HR = 1.94; 95% CI 0.59–6.43; p = 0.28). The study by Contrera et al. reported that 10 out of 19 patients in their study who received induction chemotherapy were 76 per cent less likely to die from SDSC (Disease Specific Survival HR 0.241; 95% CI 0.058–1.00; p = 0.047).Reference Contrera, Shakibai, Su, Gule-Monroe, Roberts and Brahimaj22 This suggests that neoadjuvant or induction therapy may be beneficial and increase overall survival in patients with SDSC, although more data would need to be analysed in the future to prove a significant relationship. Clinicians will therefore need to balance patient specific comorbidities and the tolerance for chemoradiotherapy related complications before deciding on definitive treatment.
Recently, new targeted therapies have been trialled on patients with SDSC. One such example would be tazemetostat, a selective inhibitor of EZH2, which has previously been shown to have clinical activity in patients with advanced epithelioid sarcoma with loss of SMARCB1.Reference Gounder, Schöffski, Jones, Agulnik, Cote and Villalobos82 Furthermore, it was shown to have a good safety profile with a low incidence of grade greater than or equal to 3 adverse events such as anaemia, weight loss and anorexia.Reference Saleh, Classe, Nguyen, Moya-Plana and Even53 With the inactivation of the SMARCB1 gene, there is a loss of activity of the SWI/SNF complex which normally inhibits expression of the oncogenic EZH2 gene.Reference Saleh, Classe, Nguyen, Moya-Plana and Even53 As such, this provides a target for tazemetostat to act on and inhibit this oncogenic gene. Saleh et al. reported a case series of three patients with SDSC who were treated with tazemetostat, of which two of them had prolonged clinical benefit of eight and 13 months after treatment.Reference Saleh, Classe, Nguyen, Moya-Plana and Even53 None of the patients in the study experienced any adverse outcomes or events related to tazemetostat. However, due to the lack of evidence of this novel drug on SDSC, its safety profile and efficacy cannot be determined at this point in time. Another novel therapy being tested would be ICIs, such as PD-1 and PD-L1 inhibitors. Previous studies have shown that SMARCB1 deficient tumours expressed positive PD-L1 expression, which could be a target for certain drugs.Reference Zhao, Du, Zhang, Yang, Zhou and Xiao76 As such, Zhao et al. attempted to give PD-1/PD-L1 inhibitors to three patients in his study, to each of the neoadjuvant, adjuvant and late first-line treatment phases.Reference Zhao, Du, Zhang, Yang, Zhou and Xiao76 All three patients survived, with two of them showing no obvious signs of any recurrence or progression of SDSC, with a disease free survival of more than 19 and 38 months, respectively. Even in the patient who suffered recurrence and metastases of the disease, the overall survival was still more than 27 months as reported in the study.Reference Zhao, Du, Zhang, Yang, Zhou and Xiao76 However, similar to tazemetostat, the available evidence on the effectiveness of these ICIs are scarce in the literature; thus, these novel therapies have not been widely adopted worldwide as of yet. More studies would need to be conducted to fully determine the safety profile and efficacy of these drugs on patients with SDSC.
The advantages of this study would be that this is the largest systematic review on SDSC and its clinicopathological features, treatment options and survival outcomes to date. The most recent systematic reviews performed were more than three years ago with less than half of the included patients in our current study. Our study is the first to prove a statistically significant positive relationship between multimodality treatment and overall survival and the first to show that surgical treatment significantly confers better prognosis to patients with SDSC. Furthermore, this study is also the first to analyse data on patients who underwent neoadjuvant or induction treatment and their effect on overall survival.
Nevertheless, this study should be interpreted in the context of known and potential limitations. The retrospective design of this review coupled with the risk of publication bias may hinder the validity of our findings, given that many of the included studies were case reports and case series. The predominance of case reports and case series in our included studies may also lower the certainty of evidence in this paper due to the absence of control groups, randomisation as well as limited sample sizes. This restricts the ability to infer causality or estimate effect sizes with confidence. Furthermore, many of the included studies did not have complete individual patient data despite our best efforts to contact the individual authors for more data. There was also insufficient data on disease-free and relapse-free survival to conduct any meaningful analysis. In our univariable and multivariable analysis, specific categories such as T1 tumours and frontal sinus primary tumours had too small a sample size to permit robust inclusion in the multivariable models. As a result, these groups were either excluded or collapsed with adjacent categories, potentially diluting granularity in the analysis. Furthermore, due to the inherent diversity in treatment modalities employed by the different authors, as well as the slight nuances in type of surgery performed or chemotherapy drug given, it may be difficult to determine the true impact of the different treatment modalities on survival outcomes. Across included studies, treatment timing and sequencing were frequently not reported in sufficient detail to permit time-dependent modelling, introducing potential immortal time bias in analyses of multimodality treatment. Consequently, observed associations may reflect survivorship and treatment eligibility rather than true causal treatment effects and, therefore, should be interpreted cautiously. In addition, as only 196 of the 372 patients had individual data for survival analysis, this could potentially create a substantial risk of selection bias. Comparison of baseline demographics and clinical characteristics between the patients included and excluded in survival analysis showed that there was minimal difference in terms of age and gender proportion, however there was a substantially higher proportion of T4 stage tumours in the patients included in the survival analysis. As such, the survival outcomes mentioned in our study should be interpreted cautiously given the risk of selection bias.
Conclusion
In conclusion, this is thus far the largest systematic review and IPD survival analysis on SDSC. Our analysis suggests that N+ disease is a significant poor prognostic factor for SDSC, while patients who undergo surgical treatment and multimodality treatment have significantly better prognosis overall. Neoadjuvant therapy may help to improve overall survival for patients with locally advanced or non-resectable disease at presentation. Targeted therapies appear promising and may improve outcomes of patients with SDSC, although more studies and trials will need to be done in the future to confirm its efficacy and safety profile.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0022215126104605.
Competing interests
The authors declare that there were no conflicts of interest involved in this work.
Funding
The authors declare that they did not receive any funding for this work. This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
