Professional Documents
Culture Documents
Original Article
1 Department of Otolaryngology-Head and Neck Surgery, Washington Address for correspondence Jake J. Lee, MD, Department of
University School of Medicine, St. Louis, Missouri, United States Otolaryngology-Head and Neck Surgery, Washington University
2 Department of Radiology, Washington University School of School of Medicine, 660 S Euclid Avenue, Campus Box 8115, St. Louis,
Medicine, St. Louis, Missouri, United States MO 63110, United States (e-mail: jakejlee@wustl.edu).
3 Department of Otolaryngology-Head and Neck Surgery, Emory
University School of Medicine, Atlanta, Georgia, United States
4 Department of Pathology, Washington University School of
Medicine, St. Louis, Missouri, United States
5 Department of Otolaryngology-Head and Neck Surgery, San
Francisco School of Medicine, University of California, San Francisco,
California, United States
Fig. 1 (A) Coronal computed tomographic (CT) image through the sinuses in bone windows demonstrates focal hyperostosis of the left lateral
Fig. 2 (A) Coronal computed tomographic (CT) bone window of a patient with focal hyperostosis of the right fovea ethmoidalis and osteitic changes of the
anterior ethmoid partitions. (B) Sagittal CT bone window of the same patient demonstrating focus of hyperostosis at the anterior ethmoid partitions, which
extends to the skull base. Since the focus was in the anterior ethmoid but also extended to the skull base, this patient was deemed to have sinonasal IP of
ethmoid origin with skull base involvement. (C) Sagittal CT bone window of a different patient with hyperostosis of the inferior sphenoid sinus bone and
thinning of the clivus adjacent to the bulk of opacification. Dehiscence of the planum sphenoidale adjacent to opacification is also seen. This patient was
deemed to have inverted papilloma of sphenoid origin with skull base involvement.
Table 1 Distribution of demographic and clinical characteristics all sites of origin except for nasal cavity (75% for rater 1 and
of the study population 80% for rater 2).
Within the total cohort, 24% (21/86) had tumor involving
Characteristic Total sample the skull base, of which 62% (13/21) and 57% (12/21) were
(n ¼ 86) correctly identified by raters 1 and 2, respectively (►Table 4).
Age at diagnosis, median (min–max) 61 (12–87) y The strength of agreement was only moderate between the
Sex two neuroradiologists (κ ¼ 0.43, 95% CI: 0.20–0.65). For rater
1, CT had 62% sensitivity (95% CI: 39–81%), 86% specificity
Male 54 (63%)
(95% CI: 75–93%), and 80% accuracy for diagnosing skull base
Female 32 (37%)
involvement. For rater 2, CT had 57% sensitivity (95% CI:
Race 34–77%), 98% specificity (95% CI: 91–100%), and 88% accura-
White 55 (64%) cy for detecting skull base involvement.
Black 24 (28%) Next, using the Krouse classification system for tumor
staging, radiographic staging was found to be concordant
Other 7 (8%)
with pathologic staging in 70% (60/86) and 84% (72/86) of
Associated carcinoma 6 (7%) patients for raters 1 and 2, respectively (►Table 5). Interrater
Surgical approach agreement was substantial (weighted κ ¼ 0.62, 95% CI:
Endoscopic endonasal only 76 (88%) 0.49–0.76), and there was moderate-to-substantial agree-
ment between radiographic and pathologic Krouse stages
Open 2 (3%)
(weighted κ ¼ 0.57, 95% CI: 0.43–0.70 for rater 1; weighted
Table 3 Diagnostic test characteristics of computed tomography Table 5 Distribution of radiographic and pathologic Krouse
versus the gold standard of surgical exploration stage for each sinonasal inverted papilloma
Sphenoid sinus origin Abbreviations: pTx, pathologic T stage; rTx, radiographic T stage.
opacification and osteitic changes of adjacent sinuses such as This study does have several limitations. Due to the study’s
the ethmoid sinuses for the four lesions attached to the lateral retrospective nature, the research team had to rely on in-office
lamella and fovea ethmoidalis and the sphenoid sinus for the clinician notes, operative summaries, and pathology reports
lesions originating at the planum sphenoidale and temporal when determining tumor origin, skull base involvement, and
fossa superior to the lateral sphenoid recess. stage. While the sample size of 86 is one of the largest for a
Our study is the largest to assess the diagnostic ability of radiographic study on sinonasal IP, it still resulted in an
CT for primary sinonasal IP to date. Prior studies have demon- imprecision of estimates for our outcomes as demonstrated
strated superior prediction rates of sinonasal IP origin for CT by the wide 95% CI. Another limitation is that 72% (62/86) of the
ranging between 89 and 100%.16,18,20,21 However, these studies study population had complex stages 3 and 4 tumors, likely
were limited by their small sample sizes and may have been related to referral bias within our tertiary care institution.
prone to selection bias by only examining CT images of IP Furthermore, some patients underwent CT imaging after biop-
lesions that demonstrated osteitis. The most recent study of sy or partial debulking of the IP lesion before the pathology was
143 patients assessed CT scans of IP lesions including those that known, which may affect radiographic interpretation. Finally,
did not clearly demonstrate osteitic changes on imaging, which while previous studies have listed various radiographic signs as
revealed that CT accurately determined tumor attachment site markers for site of sinonasal IP origin, they have not been
in only 50% of patients; however, more than half the cohort validated and standardized into established criteria.16–18,20,21
comprised patients with recurrent IP lesions, which may In sum, this study is unique when compared with existing
represent a dissimilar population compared with patients studies in the literature by comprehensively assessing the
with primary IP.17 Our analysis attempted to similarly avoid interrater reliability and diagnostic performance of CT for
selection bias and enhance generalizability by including all CT detecting multiple sinonasal IP features, including origin, skull