The Zurich magnetic resonance imaging protocol for standardized staging and restaging of sinonasal tumours

161 To the Editor: In combination with paranasal sinus computed tomography (CT), cross-sectional imaging with magnetic resonance imaging (MRI) is mandatory for staging and restaging of primary sinonasal malignancies . In the initial staging, MRI defines tumour size, provides information on extension into adjacent compartments of the sinonasal tract (in particular orbit, anterior or middle cranial fossa, leptomeningeal and brain parenchyma) and consecutively helps to determine the clinical T category. Furthermore, MRI delineates tumour from surrounding tissue (e. g. retention of mucus, reactive polyps) and may even identify perineural spread and bone marrow infiltration . The signal intensity of tumours varies depending on their cellularity, mucin content and presence of hemorrhage. However, even state-ofthe-art cross-sectional imaging may fail to correctly identify orbital or skull base infiltration. Thus, both, false-positive and false-negative findings must be considered. Common pitfalls particularly include 1) the discrimination of bony pressure erosion and bony infiltration of the anterior skull base or the medial orbital wall and 2) the discrimination of reactive dural enhancement and dural infiltration by tumour . Based on these difficulties and in analogy to upper aero-digestive tract squamous cell carcinomas, we recently suggested an obligatory exploration of all sinonasal tumours under general anesthesia and targeted biopsy, if necessary . Besides its role in the initial staging (Figure 1), MRI is also important in the restaging setting, where tumour persistence or recurrence and treatment-associated alterations may be challenging


161
To the Editor: In combination with paranasal sinus computed tomography (CT), cross-sectional imaging with magnetic resonance imaging (MRI) is mandatory for staging and restaging of primary sinonasal malignancies (1,2) . In the initial staging, MRI defines tumour size, provides information on extension into adjacent compartments of the sinonasal tract (in particular orbit, anterior or middle cranial fossa, leptomeningeal and brain parenchyma) and consecutively helps to determine the clinical T category.
Furthermore, MRI delineates tumour from surrounding tissue (e. g. retention of mucus, reactive polyps) and may even identify perineural spread and bone marrow infiltration (3,4) . The signal intensity of tumours varies depending on their cellularity, mucin content and presence of hemorrhage. However, even state-of-the-art cross-sectional imaging may fail to correctly identify orbital or skull base infiltration. Thus, both, false-positive and false-negative findings must be considered. Common pitfalls particularly include 1) the discrimination of bony pressure erosion and bony infiltration of the anterior skull base or the medial orbital wall and 2) the discrimination of reactive dural enhancement and dural infiltration by tumour (5,6) . Based on these difficulties and in analogy to upper aero-digestive tract squamous cell carcinomas, we recently suggested an obligatory exploration of all sinonasal tumours under general anesthesia and targeted biopsy, if necessary (7) .
Besides its role in the initial staging (Figure 1), MRI is also important in the restaging setting, where tumour persistence or recurrence and treatment-associated alterations may be challenging   Table 1 provides details on the proposed MR sequences.

Assessment of the bony and dural anterior skull base
In combination with CT imaging, an adequate assessment of the bony and dural skull base is pivotal. Eisen et al. stated, that the presence of pial enhancement, focal dural nodules or dural thickening of more than 5 mm is highly accurate in predicting dural invasion, while linear dural enhancement may also be reactive (5) . McIntyre et al., however, found that the presence of "≥ 2 mm of dural thickening", "loss of hypointense zone on T1-weighted images", and "nodular dural contrast enhancement" were highly predictive for dural invasion (8) . As Schuknecht et al. stated, the bony skull base is best seen on T1-weighted, contrast enhanced, fat-suppressed series with a dark signal of the bone and an adjacent, bright signal of the nasal mucosa (1) . Fat suppression is needed to eliminate any high signal intensity from adjacent fat (i.e., fatty marrow, etc.) that might be confused with or obscure actual enhancement.

Assessment of the orbit
Radiological criteria for determining orbital infiltration in MRI include (a) bony orbit and periorbita, (b) extraconal adipose tissue, (c) extraocular eye muscles and (d) intraconal structures (2,7) . However, as previously shown, one has to account for false positive findings and an overestimation of the true extent of infiltration (7) . The most important plane for extra-and intraconal structures is the coronal plane. As the lamina papyracea is thinnest directly posterior to the nasolacrimal duct, a careful attention in the coronal section must be paid in this distinct area. The coronal plane also allows detection of eye muscle infiltrations, extraconal extension, and infiltration of the optic nerve (9) . Healthy muscles have a low signal intensity in T1-weighted images, in comparison to the high signal intensity of the adipose tissue (9) . Intraconal masses can be visualized on unenhanced T1weighted sequences, because nearly all intraconal pathologies are hypointense in comparison to hyperintense surrounding adipose tissue (9) .

Conclusion
We here present a standardized and easy to reproduce MRI protocol for staging and restaging of sinonasal tumours, which allows a thorough assessment of the orbit and the anterior skull base. In order to achieve a high inter-patient and intrapatient comparability, this protocol should remain unchanged, regardless the timing of examination, tumour entity, treatment algorithms or tumour origin.

Authorship contribution
CMM: designed the study, collected data, designed the figures,