Abstract
Background/Aim: There is significant variation in post-treatment surveillance imaging for sinonasal malignancies. This study examined the utility of surveillance imaging in detecting recurrence in patients treated for sinonasal malignancies. Patients and Methods: We performed a retrospective review on an IRB-approved dataset of patients with sinonasal malignancies treated at a single institution between 2005 to 2021. Patients were categorized into groups based on the frequency of annual imaging and total number of imaging studies. We compared time-to-recurrence between the groups using log-rank test. A two-sided p-value of <0.05 was considered as the threshold for significance. Results: A total of 93 patients were eligible for this study with a median follow up of 42.3 months and 25.8% (n=24) of patients had documented recurrence. Sensitivity and specificity for recurrence based on computed tomography (CT) scans within one year of treatment completion were 50.0% and 19.5%; positron emission tomography/CT was 90.0% and 19.5%; and magnetic resonance imaging was 60.0% and 61.0%, respectively. Regardless of the type of imaging, symptomatic presentation after treatment had a specificity of 91.0% with a positive likelihood ratio of recurrence of 2.95 (95%CI=1.06-8.22). The frequency of scans was not associated with the risk of recurrence (HR=0.55; 95%CI=0.23-1.29, p=0.17). Similarly, no association was noted between the total number of scans and risk of recurrence (HR=0.64; 95%CI=0.27-1.51, p=0.31). Conclusion: The total number of frequency of scans within the first year after treatment had no association with time to recurrence of sinonasal malignancies. Symptomatic presentation was strongly associated with recurrence and should be investigated with appropriate imaging.
Sinonasal malignancies constitute a rare group of head and neck cancers, representing approximately 1% of all malignancies, and often present at a locally advanced stage with an overall poor prognosis (1). The choice of optimal surveillance imaging modality and frequency are controversial (2-4).
Existing literature reports recurrence rates as high as 56% within the first 2-3 years after treatment completion (5, 6). The National Comprehensive Cancer Network (NCCN) guidelines will often group the diverse subset of sinonasal malignancies with other head and neck cancers, suggesting repeat imaging every one to three months over the first year, two to six months over the second year, and then four to eight months over the third through fifth years after definitive treatment – though no site specific recommendations are provided (7). The guidelines also recommend endoscopic inspection of the paranasal sinuses with each visit.
Few studies have explored the subject of optimal surveillance (endoscopy, imaging modality, and frequency) for sinonasal malignancies (8). This study aimed to examine the utility of surveillance imaging performed in detecting recurrence in patients treated for sinonasal malignancies.
Patients and Methods
We performed a retrospective review on an Institutional Review Board approved dataset of patients with sinonasal malignancies treated at a single institution between 2005 to 2021. We identified 267 patients with sinonasal malignancies who were treated with definitive intent. Ninety-three consecutive patients with at least one post-treatment surveillance imaging were eligible for this study.
Patients with incomplete surveillance imaging or histologic sarcomas, lymphomas, and melanoma were excluded from the study. Information regarding tumor location, staging, pathology, treatment modalities, and surveillance imaging including type, frequency, and concern for recurrence were collected.
Treatment strategies varied among the patient population including definitive surgical resection with and without adjuvant treatment (radiation alone or with chemotherapy) and definitive radiation with and without chemotherapy. At our institution, patients were followed in a multidisciplinary fashion with endoscopic evaluations every 3-4 months for the first two years and were gradually spaced out to every 6 months for years 3-5. Additional computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET)/CT scans, as well as endoscopic evaluation, were obtained based on provider preference and clinical suspicion. Outcomes were assessed from the end of treatment date to the date of recurrence or date of last imaging for patients who did not have a recurrence. Imaging and endoscopic evaluation were scored as negative, likely reactive, suspicious for disease, or positive for disease or new primary (7). Recurrence was defined as biopsy-proven locoregional failure or distant failure. Patients were categorized into groups through cut-point analysis based on the frequency of scans in the first year after treatment (1-2 rounds of scans per year versus >2 rounds of scans per year) and total number of scans in the first year after treatment regardless of frequency (0-4 scans versus >4 scans). Of note, during each round of scans, there could be one or multiple scans (for example, an MRI and a CT or an MRI and a PET/CT, etc.).
Statistical analysis. Descriptive statistics were used to characterize the data and were compared among patients with and without recurrence using non-parametric tests. Receiver operating characteristics (ROC) were determined for each type of imaging. Time to recurrence was estimated using Kaplan–Meier methods and compared among different groups based on frequency of scans and total number of scans in the first year after treatment using log-rank test. A sensitivity analysis was performed by excluding patients with neuroendocrine tumors, as these tumors often behave differently from other sinonasal malignancies. Statistical analysis was performed using R version 4.1.2 with its statistical packages and a two-sided p-value of <0.05 was considered the threshold for significance.
Results
Patient demographics, treatment characteristics, and their association with disease recurrence are summarized in Table I. Median follow up was 42.3 months (range=32.2-68.1 months). The majority of patients were male (57.3%, n=51) with Eastern Cooperative Oncology Group (ECOG) ≤1. Overall, 61% (n=54) had their primary site within the nasal cavity. The most common histology was squamous cell carcinoma (44%; n=35) followed by esthesioneuroblastoma (25%; n=20). Forty-four patients (47%) had advanced disease at presentation (Stage III/IV or Kadish C). Most patients (88.8%) were treated with some form of radiation therapy. The median radiation dose delivered was 60 Gy (range=60-70 Gy). Overall, only performance status was associated with treatment failure (p=0.006). A median of 3.5 scans per year (range=1-10) and a median frequency of 3 rounds of imaging per year (range=1-6) were obtained in follow up.
Patient and treatment characteristics.
Twenty-four patients (25%) developed a recurrence. Of those, a total of 19 patients (79.2%) experienced locoregional recurrence and 7 experienced distant recurrence. On univariate analysis, only performance status was associated with treatment failure (p=0.006).
On ROC analysis (Table II), CT scan had a sensitivity of 0.500 (0.190-0.810) and a specificity of 0.195 (0.106-0.283) within the first year after treatment completion. MRI had a sensitivity of 0.600 (0.296-0.904) and a specificity of 0.610 (0.501-0.719). PET/CT had a sensitivity of 0.900 (0.714-1.086) and a specificity of 0.195 (0.106-0.0.283). On sensitivity analysis, after removing neuroendocrine tumors, CT scan had a sensitivity and specificity of 0.556 (0.231-0.880) and 0.213 (0.096-0.330), respectively. MRI had a sensitivity of 0.556 (0.231-0.880) and specificity of 0.234 (0.113-0.355). PET/CT had a sensitivity and specificity of 0.889 (0.684-1.094) and 0.234 (0.113-0.355), respectively. The presence of symptoms had a sensitivity of 0.600 (0.260-0.880) and a specificity of 0.900 (0.810-0.950) within the first year of treatment.
Sensitivity and specificity of detection method in detecting recurrence within the first year after treatment completion, with neuroendocrine histologies excluded.
Among the 24 patients who developed a recurrence, an average of four endoscopic evaluations were performed within the first year after treatment completion. Endoscopy was able to discover recurrence in 33.3% (n=8) of patients. One endoscopic evaluation (0.8%) was positive for recurrence, whereas 17.9% (n=21) were concerning for disease recurrence and prompted additional workup.
Freedom from recurrence at 5 years for those who had 1-2 scans/year was 82.5% versus 74.4% for those who had more than 2 scans/year. There was no association of hazard of recurrence with frequency of scans in the first year after diagnosis (HR=0.94, 95%CI=0.38-2.33, p=0.16) (Figure 1). Freedom from recurrence at 5 years for those who had 0-4 scans/year was 79.5% versus 63.6% for those who had >4 scans/year (95%CI=0.468-0.862, p=0.3) (Figure 2). Freedom from recurrence for patients who experienced symptoms was 35.9% versus 85.8% for those without symptoms at 5 years (HR=4.196, 95%CI=1.618-10.880, p=0.0015) (Figure 3).
Kaplan–Meier survival plot demonstrating the time to recurrence for patients with frequency of scanning at 0-2 times per year versus >2 times per year within the first year after treatment completion.
Kaplan–Meier Survival plot demonstrating the time to recurrence in relation to the total number of surveillance scans, 0-4 versus >4, within the first year after treatment completion.
Survival plot examining time to recurrence and presence of symptoms within the first year after treatment completion. Kaplan–Meier Survival plot demonstrating the time to recurrence for patients who reported symptoms within the first year after treatment completion.
Discussion
This study explored the utility of surveillance imaging in patients treated for sinonasal malignancies. Overall, the frequency of imaging and total number of scans were not found to be clinically significant in identifying recurrence in patients that were not experiencing symptoms. In patients that were symptomatic, we found a high likelihood of recurrence. As such, we are led to believe that clinicians should rely on patients’ reported symptoms rather than the scheduled imaging modalities. This is especially true given that physicians may not regularly adhere to recommended guidelines and may be obtaining imaging unnecessarily as has been seen in other head and neck malignancies (9-11).
Consistent with our data, previous studies have shown PET/CT to have high negative predictive value and specificity in assessing asymptomatic patients without suspicious endoscopy or suspicious imaging during the first 3 years following definitive treatment (12). However, only 6.67% of patients in that study were found to have disease recurrence on PET/CT over the 3-year period post definitive treatment. This needs to be weighed against the significantly higher expense of PET/CT and the likelihood of more frequent imaging leading to additional tests and invasive procedures (13). Furthermore, it has been demonstrated that additional PET/CT scans after an initial negative PET/CT at 3 months post treatment is low yield in the absence of clinical findings or concerning imaging (14, 15). Other studies have also shown that most locoregional disease recurrences/secondary cancers were discovered because of patient-reported symptoms rather than by routine testing in asymptomatic patients (16-18). Our study further supports these findings.
In regards to surveillance of distant recurrence, Ng et al. reported that 60% of distant recurrences were found on imaging in asymptomatic patients with head and neck squamous cell cancer (9). However, surveillance imaging in asymptomatic patients with head and neck cancers beyond 2 years was found to have a low yield and a high cost. This would seem to suggest that annual surveillance up to two years in asymptomatic patients is ideal, but beyond this time frame may be of low yield.
Neither frequency nor total number of scans within the first year were statistically associated with recurrence-free survival (RFS). However, patients who developed symptoms had worse RFS (p=0.0028). This leads us to believe that the presence of patient-reported symptoms warrants further imaging.
Most patients in our study developed symptoms that would prompt endoscopic evaluation following definitive treatment. Almost all patients experience chronic crusting/sinusitis and thus were considered “symptomatic”. These patients were therefore scheduled for more frequent visits for nasal debridement and thus symptoms could not be used as a gauge to warrant endoscopic evaluation. Therefore, we only examined endoscopic data on patients who were positive for recurrence and found endoscopy was accurately able to warrant additional workup in 17.9% of patients. When pairing this finding with their ability to manage post-treatment side effects, endoscopy may be a cost-effective tool for surveillance for these patients to warrant additional imaging.
This study has limitations in addition to its retrospective nature. We did not account for differences in expected disease prognosis based on histopathologic findings, which is a pivotal factor driving outcomes in these malignancies (19). For example, the prognosis of esthesioneuroblastoma is better than for other histologic subtypes (20). Specifically, it was found that the five-year actuarial survival was 71% in patients with esthesioneuroblastoma, 65% in squamous cell carcinomas, 31% in adenocarcinomas, and 17% in undifferentiated carcinomas requiring a craniofacial resection. Other important limitations include the small sample size, limited number of overall events due to short follow-up duration, patient adherence, and inter-provider heterogeneity in the practice of post-treatment surveillance including variation in the frequency of scanning and nature of imaging studies used. Typically, patients at a higher risk of recurrence are often subjected to more frequent imaging and therefore, could have resulted in selection bias in our study. Moreover, this study failed to account for broader factors that assess the utility of routine surveillance. This means quantifying the actual value of imaging when it comes to balance the cost of regular scans with patient anxiety with each negative scan.
Future studies should focus on predictors of locoregional and distant failure in order to stratify patients into low risk and high risk of recurrence. Routine surveillance may be warranted for those with a high risk of recurrence.
In conclusion, with a median follow-up of 3.5 years, total number, frequency, or type of surveillance imaging within the first year after treatment was not associated with time to recurrence for patients treated for sinonasal malignancies. Symptomatic presentation was strongly associated with recurrence and should prompt further workup, which includes directed imaging. Careful attention to symptoms combined with physical examination and endoscopy during follow up should guide imaging and further evaluation for a suspected recurrence, thus reducing costs and allowing for an improved care delivery model that optimizes patient outcomes.
Footnotes
Authors’ Contributions
Conception and design were conducted by Drs. K. King, R. Rauch, and N. Joshi; data collection by Drs. K. King and R. Rauch; data analysis Drs. K. King, R. Rauch, S. Roy, and N. Joshi; manuscript written by Drs. K. King and R. Rauch; manuscript editing by Drs. K. King, R. Rauch, S. Roy, O. Menyok, K. Tatebe, B. Tajudeen, P. Papagiannopoulos, P.S. Batra, M. Bhayani, S. Al-Khudari, K. Stenson, M.J. Jelinek, M.J. Fidler, and N. Joshi; with a final approval by Dr. N. Joshi.
Conflicts of Interest
The Authors have no conflicts of interest regarding this study.
- Received September 14, 2022.
- Revision received September 21, 2022.
- Accepted September 22, 2022.
- Copyright © 2022 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.