The Role of Imaging in Predicting Response to Chemotherapy and Radiotherapy in Sinonasal Malignancies: A Clinico-Pathological Correlative Study

Background: Over the past two decades, nonsurgical therapy has been increasingly utilized in advanced sinonasal malignancies with varying response rates. A complete response is associated with improved survival, and patients may be able to avoid surgical intervention. However, the ability of imaging to predict response to nonsurgical therapy is unclear. We aim to determine the positive predictive value (PPV) and negative predictive value (NPV) of MRI, CT scan, and PET in predicting response to non-surgical therapy by comparing the radiological report to the pathology report of the surgical resection.

Methods: We performed a retrospective chart review of patients treated with systemic induction therapy and/or radiotherapy, followed by surgery, for sinonasal malignancies at MD Anderson Cancer Center between 1992 and 2020.

Results: A total of 99 patients with sinonasal malignancies were treated with systemic induction therapy and/or radiotherapy, followed by surgery. Main histologic types were SCC (62%), sinonasal undifferentiated carcinoma (25%), adenoid cystic carcinoma (4%), esthesioneuroblastoma (3%), and adenocarcinoma (2%). Preoperative imaging modalities utilized to assess response to non-surgical therapy were MRI (n = 69, 70%), PET (n = 11, 11%) and CT scan (n = 38, 39%). MRI had a positive predictive value (PPV) for viable residual tumor of 88.7%, negative predictive value (NPV) of 71.4%, sensitivity of 88.7%, and specificity of 41.7%. They were statistically significant when correlated with the pathology findings using the two-tailed Fisher exact test (p = 0.00122). PET scan results gave a PPV of 80.0%, NPV of 100%, sensitivity of 100%, and specificity of 33.3% (p = 0.27273) (p = 0.27). CT scan reports had PPV of 86.7%, NPV of 25.0%, sensitivity of 81.3%, and specificity of 33.3% (p = 0.58704).

Conclusion: Treatment selection for sinonasal malignancies is complex owing to the often non-specific imaging findings and anatomic changes during posttreatment restaging. MRI is more sensitive to detecting viable residual tumors than a complete response, whereas PET (although a small group in our cohort) provides a reliable high negative predictive value meaning that a negative PET has a strong correlation with complete pathologic response. Further studies are needed to confirm the performance of PET imaging in this setting. Further understanding of imaging characteristics that better predict response to therapy would be an important next step in our ability to formulate optimal treatment algorithms.

No conflict of interest has been declared by the author(s).

Publication History

Article published online:
01 February 2023