Inverted papilloma (IP) is a benign but locally aggressive epithelial tumor that originates from the Schneiderian mucosa of the nasal cavity and paranasal sinuses [1]. Despite its benign histology, IP is clinically significant because of its high recurrence rate and its potential for malignant transformation into squamous cell carcinoma (SCC), which occurs in approximately 5%–15% of cases [2]. Early and accurate differentiation between IP and SCC arising from IP (IP+SCC) is essential, given the substantial differences in treatment strategies and prognostic implications [3–5].

Computed tomography (CT) is commonly used as the first-line imaging modality for evaluating sinonasal tumors, owing to its superior ability to delineate bony structures. Bone destruction is a key radiologic feature suggestive of malignancy [6]. However, a subset of patients with IP+SCC may not demonstrate overt bony erosion on CT, which complicates radiologic differentiation from benign IP. In such cases, magnetic resonance imaging (MRI), with its excellent soft tissue resolution, may provide valuable complementary information to aid in distinguishing IP from IP+SCC [7].

Previous studies have identified several CT and MRI characteristics that may help distinguish these entities, including the presence of the convoluted cerebriform pattern (CCP), lesion heterogeneity, and contrast enhancement characteristics [6,8]. Nevertheless, the diagnostic challenge remains substantial when bone destruction is absent, and comprehensive analyses focusing on combined CT and MRI findings in this specific context have not been reported.

The purpose of this study was to evaluate CT and MRI imaging features that may facilitate the differential diagnosis between IP and IP+SCC involving the nasal cavity and paranasal sinuses, specifically in cases without bone destruction on CT.

IP represents one of the most common benign sinonasal tumors, with an incidence of 0.2 to 0.6 per 100,000 population per year and accounting for approximately 0.4% to 4.7% of all nasal tumors [1,9]. IP has been reported to occur approximately three times more frequently in males than in females and may be diagnosed at any age, with peak incidence in the fifth and sixth decades of life [10]. Although most cases are unilateral with no side predilection, bilateral involvement of IP has been reported in 4.9% of cases [11]. The incidence of malignant transformation into SCC has been reported to range from 5% to 15%, occurring either synchronously or metachronously [2,12]. IP+SCC often develops insidiously and may lack distinctly aggressive radiologic features in the early stages, particularly in the absence of bone destruction. This subset of cases presents a diagnostic challenge, as bony erosion on CT is commonly regarded as a hallmark of malignancy [6]. In this study, we evaluated combined CT and MRI findings to identify imaging features that may aid in differentiating IP+SCC from benign IP arising from the nasal cavity and paranasal sinuses, specifically in cases without radiologic evidence of bone destruction.

CT plays a crucial role in the initial evaluation of patients with IP. It provides detailed visualization of the bony anatomy of the nasal cavity and paranasal sinuses, enabling assessment of lesion extent, identification of bony remodeling or thinning, and surgical planning [13]. Although IP is typically benign, CT is essential for detecting features that may suggest malignant transformation, such as bone destruction [6]. While areas of bony thickening on CT have been reported to predict the actual tumor attachment site [14], CT may overestimate disease extent because of its limited ability to differentiate tumor from adjacent inflammatory mucosa or retained secretions. Furthermore, our findings reaffirm that the absence of bone destruction does not exclude malignancy. Therefore, MRI serves as a complementary imaging modality to CT in the differential diagnosis of IP and IP+SCC.

MRI allows excellent delineation of the tumor from surrounding soft tissues, inflammation, and retained secretions. T2-weighted MRI and contrast-enhanced T1-weighted MRI are used in conjunction to further evaluate surrounding soft tissue and to define the extent of orbital or intracranial involvement [15]. Our results demonstrated that signal intensities on T2-weighted and contrast-enhanced T1-weighted MRI differed significantly between IP and IP+SCC. Although IP typically exhibited high signal intensity on contrast-enhanced T1-weighted images, IP+SCC more frequently demonstrated intermediate enhancement. This pattern may reflect increased cellular density, necrotic components, or altered vascularity within malignant regions. Benign IP is commonly composed of papillomatous epithelium with fibrovascular and edematous stroma, which can show prominent and relatively homogeneous gadolinium uptake and therefore be graded as “high.” In contrast, SCC arising in IP frequently results in a heterogeneous lesion containing mixed components, including residual IP and invasive SCC. The SCC component may exhibit altered microvascular perfusion and more heterogeneous enhancement because of increased cellularity, keratinizing or collagenized stroma, and microscopic necrosis, which can reduce overall enhancement and shift lesions toward the “intermediate” category on routine postcontrast sequences. Notably, prior work has suggested that necrosis within a mass exhibiting an IP-like cerebriform pattern strongly favors coexistent carcinoma [16]. Similarly, intermediate signal intensity on T2-weighted images in IP+SCC may be attributable to higher cellularity and reduced free water content compared with benign IP, which typically maintains high T2 signal intensity because of its edematous stroma. However, the “low” category was rare in our study, and the three-tier (low/intermediate/high) analysis therefore had limited statistical power and should be interpreted cautiously.

In our study, benign IP most frequently involved the ethmoid sinus, whereas IP-associated SCC most commonly involved the maxillary sinus; however, this distribution did not reach statistical significance. Furthermore, the viable portions of tumors with malignant transformation tended to exhibit significantly lower ADC values than benign IP, with a median diffusion coefficient of 0.99×10⁻³ mm²/s versus 1.20×10⁻³ mm²/s. Consistent with previous reports [6,17], malignant lesions demonstrated significantly lower ADC values than benign counterparts. This finding is likely attributable to restricted diffusion associated with increased tumor cellularity in SCC [18]. Although some overlap in ADC ranges was observed, the statistically significant difference underscores the potential role of DWI as a noninvasive biomarker for differentiating IP+SCC from IP, particularly when bone destruction is not evident.

However, there was no statistically significant difference in tumor volume between the two groups, suggesting that lesion size alone may not be a reliable parameter for differentiating IP+SCC from IP. Similarly, the degree of enhancement on CT did not differ significantly, indicating that CT contrast behavior is insufficient as an isolated diagnostic criterion. Interestingly, although the CCP has historically been regarded as a hallmark of IP [6,14], our study demonstrated no significant difference between groups. This finding suggests that CCP alone may not reliably differentiate malignant from benign lesions in the absence of bony changes. Although loss of CCP has been proposed as a potential indicator of malignant transformation, its low sensitivity in our cohort limits its utility as a standalone criterion. Nevertheless, given the small sample size and limited statistical power, these findings should be interpreted with caution and warrant validation in larger cohorts.

There are several limitations to our study. First, the sample size was relatively small, which may limit statistical power and generalizability. The small number of cases in the “low” enhancement or signal categories resulted in sparse cells, limiting the robustness of conclusions based on three-tier categorization. Second, the retrospective design inherently introduces selection bias. Third, interobserver variability was not assessed, although image interpretation was performed in consensus by experienced reviewers. Lastly, histopathologic correlation with specific imaging features was not performed, which could have provided additional insight into the mechanisms underlying the observed imaging differences. Future studies should incorporate one-to-one radiologic–pathologic correlation by coregistering the lowest-ADC regions and intermediate T2-signal areas on MRI with corresponding histopathologic findings, such as high cellularity, invasive foci, and necrosis, to clarify the biological basis of these imaging signatures and strengthen diagnostic interpretability.

In summary, differentiating IP+SCC from IP in the absence of bone destruction remains difficult using CT alone. MRI findings, including intermediate signal intensity on contrast-enhanced T1-weighted and T2-weighted images and lower ADC values on DWI, are more suggestive of IP+SCC. In contrast, tumor volume, CT enhancement, and the presence of the CCP did not differ significantly between the two groups.