Imaging of sino-nasal inverted papilloma: How can we emphasize the usefulness of the “striated pattern” sign?

Inverted papilloma (IP) is an uncommon sino-nasal neoplasm with potentially distinctive MRI and CT features. The MR “striated” imaging pattern has been reported as a valuable MR imaging feature of IP. The purpose of this study was to validate the usefulness of this sign using detailed criteria of the sign itself. All imaging findings were described for 16 patients in descriptive pattern. The nasal involvement showed highest prevalence followed by the maxillary sinus. The bony changes include remodeling and defects (n = 6), remodeling only (n = 4), and None (n = 6). Changes were severe 1/10, moderate in 3/10, and mild 5/10. Focal hyperostosis was seen in six. A striated pattern was “diffuse” in all 16 (100%) of the IPs; it followed a certain direction in all 16 patients: divergent (n = 6), convoluted (n = 2), and parallel (n = 8). Divergent growth from focal hyperostosis was seen (n = 4). Visibility of the sign was superior in T2 images (n = 11), superior in enhanced images (n = 3), and equal in both (n = 2). The striated pattern is a reliable MR imaging feature of sino-nasal IPs. Validity and usefulness of this sign can be emphasized by defining parameters of the signs itself including the following: diffusivity, direction of striations and relation to hyperostosis, and comparing visibility in both T2 and GAD-enhanced images, as well as combining this imaging signature with remolding bony changes < 1.5.


Background
Inverted papillomas (IPs), though being an uncommon benign epithelial neoplasm, have generated considerable interest through literature. This interest is that though they are benign, they are locally aggressive and have a propensity to recur, and are sometimes associated with malignancy [1,2] Recurrent disease and metachronous carcinoma can even develop after a prolonged period of time. The term papilloma means neoplasia with epithelial growth, while the US National Cancer Institute's had defined inverting papilloma as a type of tumor in which surface epithelial cells grow downward into the underlying connective tissue, and thus have been coined the term (inverting = inverted) [3].
Pre-operative imaging of the sinuses is essential before a treatment plan can be proposed to provide additional information for staging and treatment planning of sino-nasal masses including the inverted papillomas [4][5][6]. Both CT and MRI are typically performed for evaluation of sinus tumors and they are complementary modalities in evaluating disease extent, and in distinguishing tumor from infection, retained secretions, and granulation or scar tissue; all are better resolved by MRI [4][5][6].
Though imaging cannot differentiate the sino-nasal tumors, Ojiri et al. [7] discussed some potentially distinctive features of inverted papilloma on MR imaging, of which a convoluted cribriform pattern on T2 and enhanced T1-weighted images may be distinctive in 80% of cases. This sign, the convoluted cerebriform pattern (CCP), can be displayed on MR imaging by the characteristic alternating hypointense and hyperintense bands on T2-weighted and contrast-enhanced T1-weighted images, reflecting underlying histopathology of the lesion. This was further studied and discussed by Branes et al. [8], Maroldi et al. [9], and Jeon et al. [10].
The purpose of this retrospective study was to evaluate the same sign, the CCP, as depicted in the MRI of proven 16 cases to reveal some imaging details of the same sign which could elucidate and highlight the specificity of this imaging signature in making definite pre-operative diagnosis of inverted papilloma.

Patients
The protocol of our retrospective study had approval made by the Institutional Review Board (IRB) of our university hospital. Patient approval or informed consent was not required for the review of images and clinical records. Of all patients referred to our institution between January 2018 and January 2019, sinonasal inverted papillomas were confirmed in pathological specimens after surgery in 25 patients. Only 16 out of 25 patients with pre-operative imaging findings of striated pattern and had both CT and MRI made for pre-operative evaluation. Patients who had other variegated non-striated imaging appearance, as well as those who had single imaging exam, either only CT or MRI had been excluded. The CT and MRI images obtained in these patients were retrospectively reviewed.

Imaging techniques
The MRI examination studies were made on one closed 1.5 Tesla Magnet (Siemens; Avanto, Germany). A dedicated multi-channel head and neck coil was used in all patients. MRI exam included the following sequences: axial T1 turbo spin echo (TSE), axial T2 turbo spin echo with and without fat suppression, and axial and coronal T2 sequence without fat suppression. The scanning parameters were as follows: a slice thickness of 5 mm was used with an inter-slice gap of 3 mm, and a 16 × 16 cm field of view (FOV), with 5 averages used. T1-weighted images (TR, 550 ms; TE, 10 ms; 256 × 256 matrix) and turbo spin echo T2-weighted images (TR, 4400 ms; TE, 105 ms; 250 × 250 matrix) were used. Diffusionweighted imaging was obtained using single-shot spin echo planar imaging in axial plane, with fat suppression made by chemical shift selective fat suppression. Diffusion was obtained by "high" repetition time (TR) of 1700 ms, "short" echo time (TE) of 100 ms, slice numbers = 30, slice thickness = 5 mm, inter-slice gap, and number of averages = 5. Three b-factors are obtained including 0, 500, and 1000 s/ mm 2 in the axial plane. Contrast-enhanced MRI sequence was made by manual injection of Gadolinium (GAD, Gadopentetate Dimeglumine with a dose of 0.1 mmol/kg) followed by 20 ml saline flush. Enhanced images are obtained through the lesion in axial plane and at different time intervals. Enhanced images are then obtained in three orthogonal planes with fat suppression. For CT imaging, non-contrast scans were performed on 128 and 160 multi-slice scanners (Aquillion Toshiba, Canon; Japan). Spiral scanning was performed at 120 kVp and 70 mAs. CTA was made on the same machine with parameters of 120 kV and 100 mAs. The reconstructed axial images were sent to an offline workstation (Vetria 2.2 workstation), where multi-planar reformations have been viewed including the coronal and sagittal planes.

Image analysis on MRI
All of the MR images were retrospectively reviewed by two dedicated senior and junior head-and-neck neuroradiologists who have been practicing in the field for 23 years and 12 years, respectively. The following imaging details are analyzed: (a) Laterality of the lesion: either right or left side.
For a lesion involving a single moiety sphenoid sinus, a midline location is coined. Bilateral disease is coined only if a lesion is certainly crossing through a solid midline structure (e.g., nasal septum). So a nasal lesion crossing midline choana is not bilateral. For definitely bilateral lesion, predominance of a side is reported. (b) Exact location of the lesion: either sinus (maxillary, ethmoidal, etc.) or a nasal cavity. For a lesion occupying more than one region, a description is made by combining the two locations, e.g., naso-ethmoidal. The first component of the complex name indicates the epicenter of the lesion. (c) The size of the mass. This is reported in MRI because it nicely separates the outlines of the solid mass from the signal of obstructed retained sections. The longest diameter of the solid mass is measured in the three orthogonal planes and the longest one is the one reported in any of the three orthogonal planes in contrast-enhanced images. This is reported in centimeters (cm in direct contact with bony sino-nasal wall, the comment on borders would be non-applicable. T1 signal intensity-this is reported in comparison with muscles (of mastication). T1 signal pattern is reported as "striated" or "None". T2 signal intensity-this is reported as compared to visualized brain parenchyma in fat suppressed T2 images. This is reported as "iso" or "hypo" or "hypo to iso". Also, T2 signal pattern is reported as "striated" or "None". The pattern of striations is examined in both axial and coronal images.
For ADC values, a region of interest (ROI) will be drawn on the lesion, being sufficiently drawn at the solid portion of the lesion excluding areas of necrosis or hemorrhages. ADV value is then reported. The enhancement is reported in enhanced T1 images and reported to be either "striated" or "None". Presence or absence of internal cysts are examined and detected first in T2 images and expected to show fluid signal comparable to sinus sections and CSF signal. The enhancement of the cysts is then examined in GADenhanced images (either "enhanced" or "NOT").

CT imaging evaluation and analysis
For bony changes related to the margin of the sinonasal inverted papilloma at sino-nasal walls, after visualizing the contour of the lesion in T2 and enhanced MRI images, the bony changes are examined in CT in direct contact with the main mass. The following have been analyzed: (a) The changes are reported as either "Remodeling" which means smooth expansion and compression on bone, versus "Permeative" which means aggressive pattern of destruction. If no bony changes are noted, this is reported as "None". The presence or absence of focal hyperostosis/ sclerosis is reported if it is seen in direct contact with a solid portion of the tumor. This is considered when finding a "cone-shaped" or "plaquelike" focal bony thickening, even if this is the thickening of bony struts, like wall of the turbinate.
The changes involving a drainage channel are reported separately from the sino-nasal wall. The changes of sino-nasal walls are applicable to all walls except for the ethmoidal walls since these thin walls could be rarefied by effect of simple chronic inflammation and cannot be evaluated for certain change The detailed evaluation of striated pattern in both coronal and axial T2-and T1-enhanced images include diffuse versus non-diffuse; diffuse is considered only if striated pattern covers more than 90% of the longest diameter. Then direction is documented as parallel, divergent, convoluted, or other. Then this sign is evaluated in relation to focal sclerosis, if present. Finally, visibility of the sign is compared in T2 versus contrast-enhanced images. The "diffusivity" of the "striated pattern sign" is not interrupted by smooth small intra-tumoral retention cysts.

Results
The study included 16 patients (9 males, 7 females; with age range from 29 to 85 years with mean age of nearly 57 years). Only two cases (n = 2) had recurrent disease and the remaining 14 cases were primary presentation The predominant clinical complaint among patients was nasal obstruction, reported in 10 patients, all presenting primarily (Figs. 1, 2, 3, and 4). The two recurrent cases had no specific complaints and only came for routine surveillance. Two cases had epistaxis as the initial presentation, while a case was asymptomatic with incidental finding of sino-nasal mass detected in MRI exam of brain.
For the size and location preference, the lesion sizes ranged from 1.7 to 10.5 cm (mean, 6.2 cm), and the shape was lobulated in most of the patients (n = 10), smoothly oval in a single case (n = 1) and elongated in 4 cases (n = 1), while it was nonapplicable in a single case filling the sinus cavity and not surrounded by secretions, so lesion margins could not be evaluated. Unilateral disease was predominant on the right side (n = 7) and on the left side (n = 4). Midline disease was reported in three cases (n = 3) with single-moiety sphenoid sinus disease (refer to Table 1).
Only two cases (n = 2) had bilateral disease, both had nasal disease. The nasal involvement had the highest prevalence, being noted in ten cases (n = 10), with different combination of nasal and sinus involvement in eight patients and only two strictly unilateral nasal disease; one is showing a nasal mass between the inferior and middle turbinates and another one is epicentered at the classic site of lateral nasal wall, seen opposite the middle meatus. A case of bilateral bi-nasal disease had predominant large nasal lesions with limited frontal sinus extension. Two cases had combined maxillary sinus and nasal disease. Three cases (n = 3) had nasal and ethmoidal involvement with limited frontal sinus extension in two of them and maxillary extension in one. Two cases (n = 2) had combined nasal and maxillary sinus involvement. Three cases (n = 3) had isolated maxillary sinus involvement. Two patients had spheno-ethmoidal sinus involvement.
Regarding bony changes reported on CT, no definite bony changes could be noted at any sinus wall in six cases (n = 6); these cases showed a lesion nearly conforming to the configuration of the sino-nasal cavity. Four patients (n = 4) had "remodeling only" changes at sino-nasal walls. The affected walls are detailed in Table 2. Six cases (n = 6) had remodeling associated with smooth defects, apparently created by pressure erosion and geographic pattern of bony destruction. Only a single case had permeative destruction noted at the nasal septum with trans-sepal crossing; a case proven to have squamous cell carcinoma on top of papilloma. The same patient had defects created by remodeling only. The four patients with "only remodeling" showed limited bony changes < 1 cm. The six cases with "remodeling and defects" showed a single patient with marked extent Only a single case had intra-lesional calcifications. Six cases had focal sclerosis, two show linear smooth and plaque-like sclerosis of the osseous wall of the inferior turbinate (IT) in two nasal lesions. A third case showed focal sclerosis at the postero-lateral wall of the maxillary sinus in direct contact with the base of the lesion. A fourth case had mild linear sclerosis along incomplete inter-sphenoid septum in a sphenoid sinus lesion, a fifth case had smooth minimal linear sclerosis at the base of the uncinate process close to a middle-meatus-epicentered lesion, nearly bisected by the uncinate process. A sixth case show frond-like-branching focal sclerosis at the base of a recurrent inverted papilloma, quite distinguishable from diffuse sclerosis of the operated sinus, mostly representing operative-induced osteogenesis.
Fifteen out of sixteen patients had involvement and extension through a drainage channel: Ten patients (n = 10) had extension through smoothly widened OMU unilaterally either in predominant nasal or maxillary sinus component; a case showed bilateral OMU widening and obstruction; and four patients showed smoothly involved spheno-ethmoidal recess. Two patients had involvement of the frontal sinus, one is isolated, while the other is associated with OMU involvement.
Regarding magnetic resonance imaging findings ( Table 3), visualization of T1-weighted images revealed diffusely T1 hyper-intense signal as compared to the masticatory muscles; four of the six cases exhibited slightly hyper-intense signal. The remaining ten cases revealed an iso-intense signal. All lesions revealed homogeneity of signal apart from a single case revealing iso-intense signal and focus of hyper-intense signal. Only three cases (n = 3) revealed a striated pattern, while all remaining (n = 13) showed nonspecific pattern.
Regarding the T2 signal, five cases (n = 5) revealed diffusely iso-intense signal, two (n = 2) had hypointense signal, eight cases (n = 8) had diffusely hypoto-iso-intense signal, while a single case revealed mixed areas of hypo and iso-intense signals. The striated pattern was seen in all cases (n = 16): Only two patients had convoluted pattern as visualized in the axial and coronal T1 + GAD and T2 STIR images. Four out of the 16 patients revealed a cyst within the confines of the mass, adequately differentiated from retained secretions. They revealed homogenous T2 signal of fluid with thin wall. Only a single cyst showed enhancement and three (n = 3) cysts did not enhance.
Regarding mass effect, extension and obstructive sinusitis, thirteen patients (n = 13) revealed obstructive sinusitis and mass effect: Ten (n = 10) showed obstructive maxillary sinusitis, being mixed with ethmoidal sinusitis in two cases, and bilateral in one case. Orbital extension was seen in two cases, one of them had evidence of SCC with permeative pattern of destruction and lytic erosions at the lamina papyracea. Two patients had obstructive sphenoiditis in two sphenoidal-located lesion. Also, frontal sinus extension was noted in two cases, creeping through the fronto-ethmoidal recess. Detailed evaluation of "striated pattern" revealed that all cases revealed the striated pattern (n = 16, Table 4); six patients (n = 6) showed divergent direction of the striations, four out of the six patients are divergent to the focal "sclerosis", and remaining two cases (n = 2) revealed "convoluted" pattern. Eight cases (n = 8) showed parallel direction, and two patients revealed convoluted pattern of the striations. The subjective visibility of the "striated pattern" was made by two experienced senior and junior head and neck radiologists, with 23 and 12 years of head and neck radiology experience, respectively. This evaluation revealed: Higher T2 visibility in 11 cases, higher visibility at T1 + GAD-enhanced images in 3 cases, and equal visibility at both T2 and T1 + GAD in two cases.

Discussion
Inverted papilloma is a relatively uncommon neoplasm of the nasal cavity, constituting 0.5% to 4% of the whole primary nasal tumors and yet the commonest benign sino-nasal tumor [11]. The peak of incidence of IPs is in the fifth and sixth decades of life Fig. 4 A 52-year-old male presented with unilateral nasal obstruction and nasal inverted papilloma. Axial T2-weighted image reveals T2 linear divergent diffuse striations with T2 hypo-intensity with T2 'black-out' (b) reflected by compact cellularity and low apparent diffusion co-efficient value measuring 1.1 × 10 -3 cm 2 /s as seen in apparent diffusion co-efficient map (c). Axial post-gadolinium-enhanced image (d) reveals the striations to be more conspicuous than in T2 image. Axial computed tomography (e) reveals rarefied and widened post-meatal accessory ostium and smooth remodeling of medial maxillary sinus wall. Coronal computed tomography image shows sino-maxillary opacification continuous through smoothly widened ostio-meatal unit with smooth sclerosis of the middle turbinate wall Eid and Eissa Egyptian Journal of Radiology and Nuclear Medicine (2020) 51:29     and still it has been reported in all age groups [12,13]. Obviously, men are affected more often than women with a nearly 3-to-1 male-to-female predominance ratio [12,13]. Our study revealed no sex preference and nearly equal proportions were reported, while age incidence was of old adults. IP most commonly arises from the lateral wall of the nose, with epicenter at the middle meatus. It can affect any sinus, the maxillary sinus being most common (Fig. 5), while isolated sphenoid sinus localization was reported as the lowest incidence (Figs. 6 and 7) [14][15][16]. Bilateral IP and multifocal involvement have been reported in the literature but still are an uncommon with uncommon finding of intracranial, skull base or orbital extension in IP without SCC [17,18]. Our study revealed nasal involvement to have the highest prevalence. CT is usually the initial investigation and shows red flag signs of sino-nasal mass. It is an excellent method for visualization of associated bony changes unlike a malignant tumor; IP causes bone remodeling (thinning and bowing) and resorption rather than osseous destruction [19]. Bone remodeling is reported to be most commonly at the medial wall of the maxillary sinus, followed by the lamina papyracea. The nasal septum is preserved until late in the disease course [20]. Our study revealed similar involvement of medial maxilla (= lateral nasal wall) and lamina papyracea. All patients had resorption pattern created by remodeling, and permeative erosion was seen in the single case with IP plus SCC (Table 5).
In IP, two patterns of focal hyperostosis have been described: the cone-shaped and the plaque-like hyperostosis; the former corresponds to tumor origin whereas the plaque-like hyperostosis may or may not be an indicator of tumor origin; identification of which has surgical importance, as its complete resection is required to avoid recurrence [21]. Six cases had focal hyperostosis: Only two had the coneshaped appearance and both involved a maxillary sinus wall, one of them was post-operative recurrence. Plaque-like thin sclerosis was seen at the IT (n = 2), the uncinate process (n = 1) and intersphenoid septum (n = 1). The focal hyperostosis (cone-shaped) was suggested by several authors as a predictor of the IP implantation site, with a positive predictive value (PPV) of 89-95%, depending on reports. The underlying histopathology was suggested to be some form of chronic osteitis and hypervascularity, which may induce neo-osteogenesis or in other words, new bone formation. Still, our study did not correlate with post-operative specimens between sclerosis and origin of the tumor. Also, intra-lesional calcification is rare and non-specific; only one case presented with spotty calcified focus [21][22][23][24][25].
For MR imaging, multiple distinctive imaging features have been illustrated through literature. The lesion shows non-specific T1 iso-to-hyper-intense signal and T2 signal, nearly hypo-to-iso-intense to brain parenchyma or skeletal muscles with fat suppression  [26][27][28][29]. Though Yousem et al. [30] stated that no signature pattern of MR imaging is specific of IP, later studies by Barnes et al. [8] and Ojiri et al. [7] all showed a distinctive gross mucosal morphology of inverted papilloma called a convoluted cerebriform pattern. Different terms have been used as "convoluted", "cerebriform", "striated", "gyriform", and "columnar"; all are just synonyms for the same sign. We preferred the term "striated" in our study since it appears more general. By histopathological examination, the IP is made of a compact cellular metaplastic epithelium and a subepithelial loose connective tissue. This pattern contributes to the striated pattern in the MRI: the compact epithelium contributes to linear hypo-intense signal and less enhanced line in enhanced images, while the sub-epithelial loose connective tissue contributes to more T2 hyper-intense signal line and accumulation of gadolinium in variable concentrations according to its architecture. Barnes et al. [8] and Ojiri et al. [7] suggested that this convoluted pattern was specific in disagreement with the study by Yousem et al. [30] and suggested that a focal loss of this signature may suggest a focus of SCC. A study made by Maroldi et al. [9] made observation of this sign in 16 patients (69.5%) on T2-weighted images and 23 patients (100%) on contrast-enhanced T1-weighted images and was identified in only 1 (4.3%) of 23 other malignant sino-nasal tumors on both T2-weighted and contrast-enhanced T1-weighted images. They tried further to specify the sign by combining it with bony changes and concluded that this "striated columnar pattern" is a reliable MRI indicator of IP and reflects its histological architecture (positive predictive value of 95.8%), and that the combination of this finding with the absence of extended bone erosion allows for the confident discrimination of IPs from malignant tumors with high diagnostic accuracy up to 97.8%. The study by Jeon et al. [10], a significant statistical difference in the prevalence of a CCP between IP (30 of 30, 100%) and other malignant sino-nasal tumors (17 of 128, 13%) with the overall sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of CCP for the diagnosis of IP at 100%, 87%, 64%, 100%, and 89%, respectively. In our study, we tried to emphasize this sign by showing details of the same sign: we meant to describe diffusivity of this lesion by a defining "diffuse" only when striated pattern covers > 90% of the longest tumoral diameter and this was seen in all striated lesions (100%). The study by Jeon et al. [10] stated that their study could have achieved more specificity of this sign if they defined the term striated only when it is diffusely involving the lesion. Further, the direction of the striated pattern was described in a certain pattern. Since the pathologic architecture follows a certain pattern of growth made by in-folding epithelium, a certain smooth direction of striations is expected. The 16 cases showed certain direction along most of the lesions like "linear", "convoluted", or "divergent". This would not be expected in lesions with disorganized matrix. Furthermore, we considered that presence of intralesional cyst would not interrupt this sign. These cysts represent retention sub-epithelial cyst lined by glands and contains secretion. This would reveal smooth imperceptible line and T2 fluid signal, while enhancement could be only marginal or central and smooth, reflecting either retained secretions or lining epithelial glands, respectively. Relation of the striations to the focal hyperostosis was elucidated in the 6 patients with this finding: it showed divergent growth in four (n = 4) and was bisected by the ostosis in one patient (n = 1), unrelated and away from the focal sclerosis in the six patients. The literature included reported cases of multiple bony attachments with higher post-operative recurrences. However, our retrospective survey revealed only single attachments [31].
Although prior imaging by Jeon et al. [10] showed that visibility of the sign was superior in post-GAD images, our study showed that T2 images are superior than enhanced images in 11 out of 16 cases; this may attributed to the fact that differential enhancement of the epithelium and sub-epithelial tissues is limited by the fact that an intrinsic variant exists; the subepithelial tissues are variable in cellularity and stroma which would lead to variable accumulation of contrast in these tissues that could be equal to the epithelium. An extrinsic variable is variability of timing acquisition and mode of dynamic contrast enhancement, which could limit the differential GAD enhancement. Obviously, our study has limitations. First, it is a retrospective study selectively retrieving the archived cases with the targeted imaging pattern. Secondly, a small number was examined (16 cases is still larger than sample size of 10 cases in the study made by Ojiri et al. [7]). This limited size rendered statistical analysis inappropriate. A third limitation is that our study did not include the pattern associated with metachronous SCC.

Conclusion
The inverted papillomas have potentially distinctive cross-sectional imaging; the most distinctive one is the "striated" pattern. This sign could be more specific when it is diffusely involving the lesion (> 90% of the longest diameter), related to or divergent from focal hyperostosis (if present), considered not interrupted by smooth retention cyst, and showing better visibility in postgadolinium contrast-enhanced images (as observed in our study). Also, presence of remodeling bony changes and smooth non-erosive resorptions are adding to the diagnostic yield of preoperative imaging study.