Abstract
Objective
Sinonasal inverted papillomas (IP) are benign neoplasms with a propensity for local recurrence. Many risk factors are reported with little consistency between studies. This study aimed to comprehensively assess for demographic, imaging, histopathologic, and intraoperative risk factors for recurrence.
Methods
We performed a single-center retrospective cohort study of patients with pathologically diagnosed IP without malignancy who underwent surgical resection between 1997 and 2018. Eligible patients were identified through a database maintained by the Department of Pathology. Logistic regression identified variables associated with recurrence, and conjunctive consolidation was performed to create a predictive model.
Results
Of 76 subjects, 37% (n=28) had recurrence. Median follow-up and time to recurrence were 2.9 (range 0.5-21.1) and 1.7 (range 0.2-13.0) years, respectively. Confirmed negative margins on histology was protective (OR 0.25, 95% CI 0.08-0.85). Frontal sinus involvement (aOR 5.83, 95% CI 1.20-28.37), incomplete resection (aOR 9.67, 95% CI 2.24-41.72), and presence of dysplasia (aOR 4.38, 95% CI 1.01-19.10) were significantly associated with recurrence on multivariable analysis. A three-level composite recurrence risk staging system was created by consolidating the above three variables. The recurrence risks of Composite Stage I, II, and III disease were 20%, 38%, and 100%, respectively. No demographic, imaging, staging, or surgical approach variables were associated with recurrence.
Conclusion
Frontal sinus involvement, incomplete resection, and dysplasia were significant risk factors for IP recurrence while confirmed negative margins were protective. Creation of a composite staging system using the above variables may allow for risk stratification and a patient-specific approach to postoperative IP management.
Keywords: Inverted papilloma, paranasal sinus, recurrence, dysplasia, frontal sinus
Introduction
Sinonasal inverted papillomas (IP), named for their characteristic endophytic growth pattern on histology, are locally aggressive benign neoplasms associated with osseous remodeling and destruction, local recurrence, and malignancy.1–3 They comprise up to 4% of all primary nasal neoplasms.4 Although benign, IPs may be associated with epithelial atypia, dysplasia, carcinoma in situ, and invasive carcinoma.5,6 They have a 7-11% overall malignancy rate, which can be synchronous (discovered at the same time as primary IP but not necessarily in the same lesion) or metachronous (at a site of previous IP resection, indicating malignant transformation).7–9
Sinonasal IP also has a high propensity for local recurrence with cited rates of 14-33%, most of which are due to residual disease at the original resection site.7,9–13 Many risk factors for recurrence are reported; however, there remains controversy regarding which is most clinically important.14 Some conflicting risk factors include those based on patient history, such as smoking status and prior sinonasal surgery.14–17 Other factors include morphologic variables such as tumor location and osseous invasion18,19; tumor stage10,14,20–23; histopathologic measures including atypia, dysplasia, mitotic index, and growth pattern5,6,8,11; and surgical technique such as endoscopic endonasal, limited external, open, and combined approaches.12,14,24,25
No single study has comprehensively examined all of the above demographic, clinical, paraclinical, and surgical approach data and reported their associations with recurrence. As a result, the medical literature is rife with studies identifying contradictory risk factors. Furthermore, no study has examined overall severity of comorbidity in relation to IP outcomes. Thus, we performed this observational cohort study to evaluate the independent prognostic impact of all these parameters and their association with sinonasal IP recurrence. In addition, we designed a 3-category staging system based on surgical and pathological findings to establish a patient’s risk for recurrence.
Materials and Methods
We conducted a single-center retrospective cohort study of consecutive patients with sinonasal IP to assess for recurrence. Institutional review board (IRB) approval was obtained prior to chart review and data collection (“Inverted papilloma: incidence and risk of malignancy,” IRB #201610037). Inclusion criteria included: 1) patients with evidence of sinonasal IP on histopathology and 2) patients who underwent sinonasal surgical resection between March 1998 and July 2018 at our tertiary care hospital. All patients with sinonasal IP were identified using an electronic database maintained by the Department of Pathology. Exclusion criteria included: 1) patients presenting to our institution with recurrent IP, 2) patients with follow-up of less than 6 months, and 3) those with associated carcinoma. Patients who underwent non-oncologic sinonasal surgery for either concurrent or misdiagnosed polyposis and were then referred to our institution for more complete extirpative surgery were still included in the study. Conversely, patients who already underwent extirpative surgery for IP resection at another institution and then presented to our center with recurrence were excluded.
The primary outcome was sinonasal IP recurrence, which was defined as the presence of a gross lesion at a site of prior resection visible on nasal endoscopy that was confirmed pathologically to be IP. Zero time was defined as the date of index surgery. Demographic information, including age, sex, smoking status, history of chronic rhinosinusitis, diabetes mellitus, and cancer at any location, was reviewed and recorded. Clinical information, such as method of referral; initial symptom; iatrotropic stimulus defined as the reason for a patient to pursue medical care including symptoms, signs, or need for routine screening; presence of nasal congestion/obstruction versus facial pain/pressure; previous otolaryngologic evaluation; mistaken polyp diagnosis; and prior sinonasal surgery, was also recorded. Comorbid ailments were identified through chart review, and overall severity was determined using the Adult Comorbidity Evaluation-27 (ACE-27).26
Paraclinical information, defined as morphologic features assessed outside the physical exam, included imaging, histopathologic, and surgical exploration data. All patients underwent pre-operative imaging with computed tomography (CT). A board-certified neuroradiologist re-reviewed all pre-operative CT scans to assess for specific findings such as radiographic Krouse and Cannady stage20,22; underlying osseous changes including sclerosis, thinning, and dehiscence; and skull base involvement. Board-certified pathologists reviewed the histopathologic specimens at the time of tissue resection, and the reports were retrospectively reviewed to assess for dysplasia and pathologic Krouse and Cannady stage. All specimen tissue was entirely submitted for histologic examination to identify any dysplasia or carcinoma. Surgical exploration data, which was assessed from operative notes, included degree of resection, final margin status, drill use at tumor site, tumor base morphology, specific sinus involvement, and skull base involvement. Resection was considered incomplete if non-oncologic surgical resection such as functional endoscopic sinus surgery was performed, residual tumor was left behind due to its proximity to crucial structures, and microscopic tumor was seen on final histopathologic margin analysis. Tumor base morphology was separated into pedunculated and sessile categories. Based on the operative report, tumors were defined as pedunculated if a discrete tumor attachment was identified, and they were considered sessile if they had a broad-based attachment to the underlying sinonasal bones. Lastly, surgical approach was classified into one of 3 categories: 1) endoscopic endonasal, 2) open, and 3) combined, which included Caldwell-Luc approaches.
Follow-up information, which consisted of comprehensive history and otorhinolaryngologic examination including rigid nasal endoscopy, was also obtained from chart review. At our institution, follow-up usually occurred 3-5 times during the first year and 1-3 times every year afterwards depending on surgeon preference. After 3 years, most patients are instructed to follow up annually for indefinite surveillance. Routine CT or magnetic resonance imaging (MRI) supplemented clinical examination at the discretion of the examining physician and whenever there was clinical suspicion of recurrence. Typically, the physician did not order follow-up imaging when he/she was certain that there was no evidence of disease on endoscopy. The decision to obtain CT versus MRI was surgeon- and situation-dependent.
Descriptive statistics were used to explore the distribution of demographic, imaging, surgical, and histopathologic characteristics among patients with IP recurrence and those without recurrence. Univariate logistic regression was used to explore the association of each of the characteristics with IP recurrence, and effect sizes as odds ratios (OR) and 95% confidence intervals (CI) were reported. Variables that were significantly associated with recurrence and clinically important were entered into a multivariable logistic regression model. Adjusted odds ratios (aOR) and 95% CI were then reported. The Hosmer-Lemeshow test was utilized to test the model fit to the data.
To develop a composite clinical and pathological-severity staging system to predict recurrence using significant variables from the logistic regression model, we employed conjunctive consolidation, a process of intuitively grouping clinical data based on statistical isometry and biological coherence. Gradients of recurrence risk were then assessed. Statistical calculations were performed with SPSS 25 (IBM Corp., Armonk, NY) and STATA 15 (College Station, TX).
Results
A total of 76 patients with pathologically confirmed sinonasal IP who underwent surgical resection were investigated. The recurrence rate in the study population was 37% (28/76). Median follow-up was 2.9 (range 0.5 to 21.1) years. In the patients with recurrence, half the patients were diagnosed at 1.7 (range 0.2 to 13.0) years after surgery. No patients with recurrence had secondary malignancies identified during re-resection. Follow-up imaging with either CT or MRI was performed in 36% (27/76) and 45% (34/76) of patients within 1 year and 3 years of surgery, respectively. Follow-up imaging was never performed in 53% (40/76) of patients.
As the initial symptom, 67% (51/76) of the study population presented with nasal congestion/obstruction while 11% (8/76) and 7% (5/76) presented with facial pain/pressure and nasal drainage, respectively. Other initial symptoms included epistaxis and tearing. Additionally, the iatrotropic stimulus was nasal congestion/obstruction in 50% (38/76), facial pain/pressure in 13% (10/76), nasal drainage in 7% (5/76), and mass protruding from the nasal vestibule in 5% (4/76). Of note, 16% (12/76) sought medical attention for causes likely unrelated to sinonasal IP, including hearing loss, chronic cough, sialadenitis, sore throat, neck swelling, nausea, seizure, breast tenderness, tremors, and abdominal aortic aneurysm screening, after which IP was incidentally discovered.
Distribution of baseline characteristics and results from univariate logistic regression separated by presence and absence of recurrence are presented in Tables 1, 2, 3, 4. In univariate analysis, incomplete resection, sessile tumor base, frontal sinus involvement, and presence of dysplasia were significantly associated with recurrence while confirmed negative margins on histopathology was significantly associated with no recurrence (Tables 3–4). Moreover, these variables demonstrated large and somewhat precise point estimates for effect size.
Table 1.
Baseline demographic, clinical, and comorbidity characteristics of the study population with reported odds ratios (OR) with 95% confidence intervals (CI) calculated using univariate logistic regression. Acronyms: SD, standard deviation; ACE, Adult Comorbidity Evaluation; ENT, Ear Nose Throat.
| Variables | No Recurrence (n=48) | Recurrence (n=28) | OR (95% CI) |
|---|---|---|---|
| Age at Diagnosis (years), Mean (SD) | 53.3 (16.1) | 60.5 (13.0) | 1.03 (1.00 to 1.07) |
| <40 | 6 (12.5%) | 6 (21.4%) | Ref |
| 41-55 | 14 (29.2%) | 9 (32.1%) | 0.64 (0.16 to 2.63) |
| 56-70 | 18 (37.5%) | 9 (32.1%) | 0.50 (0.13 to 2.00) |
| 70+ | 10 (20.8%) | 4 (14.3%) | 0.40 (0.08 to 2.02) |
| Sex | |||
| Male | 32 (66.7%) | 15 (53.6%) | Ref |
| Female | 16 (33.3%) | 13 (46.4%) | 1.73 (0.67 to 4.50) |
| Diabetes Mellitus | 5 (10.4%) | 5 (17.9%) | 1.87 (0.49 to 7.13) |
| Chronic Rhinosinusitis | 30 (62.5%) | 21 (75.0%) | 1.80 (0.64 to 5.07) |
| Prior FESS | 20 (41.7%) | 16 (57.1%) | 1.87 (0.73 to 4.80) |
| History of Cancer | 13 (27.1%) | 6 (21.4%) | 0.73 (0.24 to 2.22) |
| Smoking History | |||
| No | 18 (37.5%) | 15 (53.6%) | Ref |
| Yes | 30 (62.5%) | 13 (46.4%) | 0.52 (0.20 to 1.34) |
| Symptomatology* | |||
| Nasal congestion | 24 (50.0%) | 18 (64.3%) | 1.80 (0.69 to 4.69) |
| Facial pain/pressure | 7 (14.6%) | 4 (14.3%) | 0.98 (0.26 to 3.68) |
| Attributable symptom | 40 (83.3%) | 26 (92.9%) | 2.60 (0.51 to 13.22) |
| Referral Pattern | |||
| External | 35 (72.9%) | 25 (89.3%) | Ref |
| Within Institution | 13 (27.1%) | 3 (10.7%) | 0.32 (0.08 to 1.25) |
| Previous ENT Evaluation | 22 (45.8%) | 15 (53.6%) | 1.36 (0.54 to 3.47) |
| ACE-27 Score | |||
| None/Mild | 32 (66.7%) | 19 (67.9%) | Ref |
| Moderate/Severe | 16 (33.3%) | 9 (32.1%) | 0.95 (0.35 to 2.56) |
For Symptomatology, the OR represents the ratio of the odds of recurrence for the presence versus absence of each symptom.
Table 2.
Baseline imaging characteristics of the study population with reported odds ratios (OR) with 95% confidence intervals (CI) calculated using univariate logistic regression. Acronyms: CT, computed tomography.
| Variables | No Recurrence (n=48) | Recurrence (n=28) | OR (95% CI) | |
|---|---|---|---|---|
| Radiologic Krouse Stage | ||||
| I/II | 10 (20.8%) | 4 (14.3%) | Ref | |
| III/IV | 38 (79.2%) | 24 (85.7%) | 1.58 (0.45 to 5.61) | |
| Radiologic Cannady Stage | ||||
| A | 10 (20.8%) | 4 (14.3%) | Ref | |
| B | 34 (70.8%) | 20 (71.4%) | 1.47 (0.41 to 5.31) | |
| C | 4 (8.3%) | 4 (14.3%) | 2.50 (0.41 to 15.23) | |
| Osseous Involvement on CT | 35 (72.9%) | 19 (67.9%) | 0.78 (0.28 to 2.17) | |
| Skull Base Involvement on CT | 6 (12.5%) | 8 (28.6%) | 2.80 (0.86 to 9.16) | |
| Skull Base Dehiscence on CT | 3 (6.3%) | 4 (14.3%) | 2.50 (0.52 to 12.10) |
Table 3.
Baseline surgical exploration characteristics of the study population with reported odds ratios (OR) with 95% confidence intervals (CI) calculated using univariate logistic regression. Acronyms: FESS, functional endoscopic sinus surgery.
| Variables | No Recurrence (n=48) | Recurrence (n=28) | OR (95% CI) | |
|---|---|---|---|---|
| Surgical Approach | ||||
| Endonasal | 43 (89.6%) | 20 (71.4%) | Ref | |
| Open | 1 (2.1%) | 2 (7.2%) | 4.30 (0.37 to 50.25) | |
| Combined | 4 (8.3%) | 6 (21.4%) | 3.23 (0.82 to 12.72) | |
| Incomplete Resection | 9 (18.7%) | 12 (42.9%) | 3.25 (1.15 to 9.21) | |
| Confirmed Negative Margins | 19 (39.6%) | 4 (14.3%) | 0.25 (0.08 to 0.85) | |
| Unsuspected IP Diagnosis | 22 (45.8%) | 17 (60.7%) | 1.83 (0.71 to 4.71) | |
| Tumor Base | ||||
| Pedunculated | 27 (56.3%) | 7 (25.0%) | Ref | |
| Sessile | 16 (33.3%) | 19 (67.9%) | 4.58 (1.58 to 13.28) | |
| Missing Data | 5 (10.4%) | 2 (7.1%) | ||
| Drill Use at Tumor Site | 17 (35.4%) | 7 (25.0%) | 0.61 (0.22 to 1.72) | |
| Sinus Involvement* | ||||
| Maxillary | 35 (72.9%) | 20 (71.4%) | 0.93 (0.33 to 2.62) | |
| Ethmoid | 26 (54.2%) | 19 (67.9%) | 1.79 (0.67 to 4.74) | |
| Sphenoid | 6 (12.5%) | 4 (14.3%) | 1.17 (0.30 to 4.55) | |
| Frontal | 5 (10.4%) | 9 (32.1%) | 4.07 (1.20 to 13.79) | |
| 2+ | 23 (47.9%) | 19 (67.9%) | 2.30 (0.87 to 6.08) | |
| 3+ | 3 (6.3%) | 6 (21.4%) | 4.09 (0.93 to 17.91) | |
| Skull Base Involvement | 8 (16.7%) | 9 (32.1%) | 2.37 (0.79 to 7.10) | |
For Sinus Involvement, the OR represents the ratio of the odds of recurrence for the sinus being involved with inverted papilloma versus being uninvolved.
Table 4:
Baseline histopathologic characteristics of the study population with reported odds ratios (OR) with 95% confidence intervals (CI) calculated using univariate logistic regression.
| Variables | No Recurrence (n=48) | Recurrence (n=28) | OR (95% CI) | |
|---|---|---|---|---|
| Pathologic Krouse Stage | ||||
| I/II | 13 (27.1%) | 5 (17.9%) | Ref | |
| III/IV | 35 (72.9%) | 23 (82.1%) | 1.71 (0.54 to 5.44) | |
| Pathologic Cannady Stage | ||||
| A | 13 (27.1%) | 5 (17.9%) | Ref | |
| B | 31 (64.6%) | 19 (67.9%) | 1.59 (0.49 to 5.18) | |
| C | 4 (8.3%) | 4 (14.3%) | 2.60 (0.46 to 14.63) | |
| Dysplasia | ||||
| Absence | 42 (87.5%) | 17 (60.7%) | Ref | |
| Presence | 6 (12.5%) | 11 (39.3%) | 4.53 (1.44 to 14.21) |
Descriptive analysis of the frontal sinus subgroup revealed that 64% (9/14) demonstrated recurrence, 5 of which occurred at the frontal sinus while the other 4 occurred at contiguous sites including the ethmoid sinus and skull base. Furthermore, 86% (12/14) were resected via endoscopic endonasal approach only while 14% (2/14) were resected via combined approaches, which included a craniofacial orbitotomy approach and frontal osteoplastic flap. In patients who underwent an endoscopic endonasal approach only, 7/12 had recurrence. In the combined approach cohort, 2/2 had recurrence, but neither involved the frontal sinus.
In the multivariable model, incomplete resection (aOR 9.67, 95% CI 2.24 to 41.72), frontal sinus involvement (aOR 5.83, 95% CI 1.20 to 28.37), and presence of dysplasia (aOR 4.38, 95% CI 1.01 to 19.10) were significantly associated with recurrence and demonstrated large, clinically meaningful point estimates for effect size (Table 5). Sessile tumor base had the smallest effect size and was not significantly associated with recurrence in the multivariable model. Confirmed negative margin resection was not included in the multivariable analysis due to its overlap with complete resection status and the number of patients that did not undergo routine margin analysis via frozen histopathology (50/76, 66%).
Table 5.
Multivariable logistic regression model. Acronyms: aOR, adjusted odds ratio; CI, confidence interval.
| Variables | aOR (95% CI) |
|---|---|
| Sessile base | 2.86 (0.82 to 10.01) |
| Presence of dysplasia | 4.38 (1.01 to 19.10) |
| Incomplete resection | 9.67 (2.24 to 41.72) |
| Frontal sinus involvement | 5.83 (1.20 to 28.37) |
Conjunctive consolidation was then performed by first examining the rates of recurrence within the four conjoined categories of resection status (complete vs. incomplete) and frontal sinus involvement (uninvolved vs. involved) (Table 6). There was a “double gradient” or “gradient-within-gradient” relationship, where gradients for rates of recurrence between the two levels of resection status and between the two levels of frontal sinus involvement existed. Next, we consolidated patients with different combinations of resection status and frontal sinus involvement that had similar rates of recurrence to create a new three-category “Intraoperative Staging System,” which is aptly named since a clinician would be able to stage the likelihood of recurrence for any sinonasal IP patient during surgical exploration using this system.
Table 6.
Conjunction of Resection status and Frontal Sinus involvement.
| Resection | |||||
|---|---|---|---|---|---|
| Frontal Sinus | Complete | Incomplete | Total | Intraoperative Stage | |
| Uninvolved | 10/44 (23%) | 9/18 (50%) | 19/62 (31%) | A: 10/44 (23%) | |
| Involved | 6/11 (55%) | 3/3 (100%) | 9/14 (64%) | B: 15/29 (52%) | |
| Total | 16/55 (29%) | 12/21 (57%) | 28/76 (37%) | C: 3/3 (100%) | |
Subsequently, the three stages of the Intraoperative Staging System were conjoined with dysplasia (presence vs. absence), resulting in six conjoined categories (Table 7). Once again, there was a “double gradient” relationship within each stage of the Intraoperative Staging System and within the two levels of dysplasia. We consolidated patients with different combinations of Intraoperative Stage and dysplasia that had similar rates of recurrence to create the final three-category “Composite Staging System.” There was a gradient across the three Composite Stages with Stage I, II, and III having recurrence rates of 20%, 38%, and 100%, respectively. The absolute difference in rate of recurrence for patients with Composite Stage III sinonasal IP and those with Composite Stage I sinonasal IP was 80%. A summary of the composite staging criteria using the three variables is provided (Fig. 1).
Table 7.
Conjunction of Intraoperative Stage and Dysplasia.
| Intraoperative Stage | ||||||
|---|---|---|---|---|---|---|
| Dysplasia | A | B | C | Total | Composite Stage | |
| Absence | 7/35 (20%) | 9/23 (39%) | 1/1 (100%) | 17/59 (29%) | I: 7/35 (20%) | |
| Presence | 3/9 (33%) | 6/6 (100%) | 2/2 (100%) | 11/17 (65%) | II: 12/32 (38%) | |
| Total | 10/44 (23%) | 15/29 (52%) | 3/3 (100%) | 28/76 (37%) | III: 9/9 (100%) | |
Figure 1.
Flow diagram summarizing the staging criteria for the 3 composite groups (I, II, and III).
Discussion
This study of 76 patients with pathologically confirmed sinonasal IP who underwent surgical resection attempted to comprehensively assess multiple demographic, clinical, paraclinical, and comorbidity risk factors for recurrence. We were able to identify morphologic, intraoperative, and histologic risk factors for recurrence in our sample. In multivariable analysis, frontal sinus involvement, incomplete resection, and presence of dysplasia were associated with recurrence. Sessile tumor base morphology was not significantly associated with recurrence in the multivariable model but likely has a clinically meaningful effect given the effect size and upper bound of the 95% CI. A three-level composite staging system was then designed using conjunctive consolidation where patients with stage I, II, and III tumors had recurrence rates of 20%, 38%, and 100%, respectively.
Sinonasal IP has up to a 30% recurrence rate based on the literature.10 While the literature cites many risk factors for recurrence, the results are conflicting, and many are still debated.14 These cited risk factors can be separated into domains using a clinical epidemiological approach popularized by Feinstein et al., which are demographic, clinical, paraclinical, and comorbidity.27,28 Clinical risk factors are based on patient symptomatology, physical examination findings, and referral patterns. Paraclinical risk factors include laboratory values and morphologic variables assessed outside the physical examination with modalities such as endoscopy, imaging, histopathology, and surgical exploration. Demographic risk factors cited in the literature include smoking history and prior sinonasal surgery.14–17 Cited paraclinical risk factors include tumor location assessed via nasal endoscopy and/or CT scan, tumor stage, dysplasia, human papilloma virus status, and surgical technique.5,6,8,10–12,14,20–25 Assessment of clinical and comorbidity risk factors, however, has not been reported in the literature.
For example although many staging systems exist for IP based on location and extension such as the Krouse and Cannady systems, there is conflicting evidence on whether increased tumor stage is associated with recurrence. Some observational studies have shown significant differences in recurrence rate among Krouse and Cannady stages while other studies have found no such association between any classification system and recurrence.14,21 Furthermore, a meta-analysis revealed an increased risk of recurrence between Krouse T2 and T3 stages (OR 1.51, 95% CI 1.09 to 2.09) but no difference between T1 and T2 and between T3 and T4.23 One reason for a lack of association is the heterogeneity of certain stages, including Krouse T3/Cannady T2 and Krouse T4. For example, a Krouse T3 IP involving the sphenoid sinus overlying the opticocarotid recess is the same stage as IP involving the posterior wall of the maxillary sinus despite the discrepant difficulty of extirpative surgery between the two lesions. Furthermore, the extent of the tumor periphery, which is measured by the staging systems, may be less clinically meaningful than the extent of tumor attachment in regards to ability to achieve complete resection.
Other controversial morphologic risk factors are frontal sinus location of tumor and dysplasia. For instance, two studies have described an association between frontal sinus location and recurrence; however, other studies found no such association.14,17–19 Frontal sinus involvement may be a risk factor for recurrence due to the physical and visual limitations of endoscopic endonasal frontal surgery as a single modality, and clinicians should consider Draf III or combined approaches as part of their armamentarium for extensive frontal sinus disease. Another conflicting risk factor is dysplasia; some studies have reported an association between dysplasia and recurrence while others found no association.5,6,8,29
Our model supports the generally accepted paradigm that complete resection should attempt to be achieved if possible. In this sample, resection was usually incomplete because IP was mistaken for polyps and more complete extirpative surgery was deferred due to patient preference, morbidity concerns, or comorbid conditions. In other instances, patients had microscopic residual disease on permanent pathology. These patients are at a higher risk of recurrence and may benefit from repeat surgery for complete extirpation of tumor. The last scenario was seen in patients whose resection was incomplete due to the tumor’s proximity to vital structures such as the skull base, dura, infraorbital nerve, internal carotid artery, or optic nerve. These patients may benefit from closer and more frequent surveillance and proper counseling.
By assessing broader risk factors for sinonasal IP recurrence, we have created a predictive model for risk stratification, which is not present in the existing literature. While still requiring validation in larger cohorts, the model may have clinical applications. For instance, we found that confirmed negative margins on frozen sections were protective, suggesting that routine frozen sections to confirm complete resection of IP should be used for all cases.9 On the other hand, frontal sinus involvement and presence of dysplasia cannot be altered by clinician behavior, but incorporating these two variables into the model is necessary for stratifying risk. Higher risk patients, such as those with composite stage III, may benefit from more frequent follow-up visits with nasal endoscopy and routine imaging even when years removed from surgery.
There are several limitations to this study. As with any retrospective study, we relied on clinician notes, operative summaries, and pathology reports when assessing for risk factors, which introduces potential bias due to missing information and post hoc interpretation. In addition, the sample size of 76, while relatively large for a study on a disease with low prevalence, is limited, resulting in imprecision in the estimates for recurrence and low number of patients in some combination categories during conjunctive consolidation. Another limitation is that the recurrence rate in the study population was higher than most published data, indicating potential referral bias where complex sinonasal IP cases are overrepresented at our tertiary care institution. Also while patients with follow-up duration of less than 6 months are excluded, 6 months is still a relatively short time interval for assessing recurrence, and some of those patients with no evidence of disease at 6-12 months may recur in the future. Lastly, we are limited by our detection methods, and there may be false negatives where patients with recurrence have not been properly diagnosed due to sampling error of margins, limited visualization on endoscopic exam, and subpar sensitivity of follow-up imaging.
Nevertheless, the strengths of this study, including assessment of many demographic, clinical, paraclinical, and comorbidity variables and generation of a new predictive model, separate it from the existing studies in the literature. We hope these results will add to the literature and promote further studies across multiple institutions. Future directions include utilizing a large retrospective or prospective cohort study to validate the Composite Staging System; a large multi-center prospective cohort study to confirm demographic, clinical, paraclinical, and comorbidity risk factors; and survival analysis to investigate the association of certain risk factors with 5-year disease-free survival.
Conclusion
Sinonasal IP is a challenging disease to treat due to its high rate of recurrence even years after surgery. Frontal sinus involvement, incomplete resection, and presence of dysplasia were significant risk factors for local recurrence while confirmed negative margins on histology were protective. The creation of a clinical predictive model incorporating frontal sinus involvement, resection status, and dysplasia may allow for further risk stratification, thereby fostering a patient-specific approach to postoperative IP management.
Acknowledgments
Funding: Research reported in this publication was supported by the National Institutes of Health (NIH) under Award Number 5T32DC000022-30 and by the National Center for Advancing Translational Sciences of the NIH under Award Number UL1TR002345. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Footnotes
Conflicts of Interest: none
Presented at the 2019 Triological Society Combined Sections Meeting
Coronado, CA, USA
January 25, 2019
References
- 1.Chiu AG, Jackman AH, Antunes MB, Feldman MD, Palmer JN. Radiographic and histologic analysis of the bone underlying inverted papillomas. Laryngoscope 2006; 116(9):1617–20. [DOI] [PubMed] [Google Scholar]
- 2.Hyams VJ. Papillomas of the nasal cavity and paranasal sinuses. A clinicopathological study of 315 cases. Ann Otol Rhinol Laryngol 1971; 80(2):192–206. [DOI] [PubMed] [Google Scholar]
- 3.Ringertz N Pathology of malignant tumors arising in the nasal and paranasal cavities and maxilla. Stockholm, 1938. [Google Scholar]
- 4.Vrabec DP. The inverted Schneiderian papilloma: a 25-year study. Laryngoscope 1994; 104(5 Pt 1):582–605. [DOI] [PubMed] [Google Scholar]
- 5.Snyder RN, Perzin KH. Papillomatosis of nasal cavity and paranasal sinuses (inverted papilloma, squamous papilloma). A clinicopathologic study. Cancer 1972; 30(3):668–90. [DOI] [PubMed] [Google Scholar]
- 6.Suh KW, Facer GW, Devine KD, Weiland LH, Zujko RD. Inverting papilloma of the nose and paranasal sinuses. Laryngoscope 1977; 87(1):35–46. [DOI] [PubMed] [Google Scholar]
- 7.Lawson W, Patel ZM. The evolution of management for inverted papilloma: an analysis of 200 cases. Otolaryngol Head Neck Surg 2009; 140(3):330–5. [DOI] [PubMed] [Google Scholar]
- 8.Mirza S, Bradley PJ, Acharya A, Stacey M, Jones NS. Sinonasal inverted papillomas: recurrence, and synchronous and metachronous malignancy. J Laryngol Otol 2007; 121(9):857–64. [DOI] [PubMed] [Google Scholar]
- 9.Nygren A, Kiss K, von Buchwald C, Bilde A. Rate of recurrence and malignant transformation in 88 cases with inverted papilloma between 1998–2008. Acta Otolaryngol 2016; 136(3):333–6. [DOI] [PubMed] [Google Scholar]
- 10.Bugter O, Monserez DA, van Zijl F, Baatenburg de Jong RJ, Hardillo JA. Surgical management of inverted papilloma; a single-center analysis of 247 patients with long follow-up. J Otolaryngol Head Neck Surg 2017; 46(1):67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Christensen WN, Smith RR. Schneiderian papillomas: a clinicopathologic study of 67 cases. Hum Pathol 1986; 17(4):393–400. [DOI] [PubMed] [Google Scholar]
- 12.Busquets JM, Hwang PH. Endoscopic resection of sinonasal inverted papilloma: a meta-analysis. Otolaryngol Head Neck Surg 2006; 134(3):476–82. [DOI] [PubMed] [Google Scholar]
- 13.Sham CL, Woo JK, van Hasselt CA, Tong MC. Treatment results of sinonasal inverted papilloma: an 18-year study. Am J Rhinol Allergy 2009; 23(2):203–11. [DOI] [PubMed] [Google Scholar]
- 14.Lisan Q, Laccourreye O, Bonfils P. Sinonasal Inverted Papilloma: Risk Factors for Local Recurrence After Surgical Resection. Ann Otol Rhinol Laryngol 2017; 126(6):498–504. [DOI] [PubMed] [Google Scholar]
- 15.Moon IJ, Lee DY, Suh MW et al. Cigarette smoking increases risk of recurrence for sinonasal inverted papilloma. Am J Rhinol Allergy 2010; 24(5):325–9. [DOI] [PubMed] [Google Scholar]
- 16.Roh HJ, Mun SJ, Cho KS, Hong SL. Smoking, not human papilloma virus infection, is a risk factor for recurrence of sinonasal inverted papilloma. Am J Rhinol Allergy 2016; 30(2):79–82. [DOI] [PubMed] [Google Scholar]
- 17.Sciarretta V, Fernandez IJ, Farneti P, Pasquini E. Endoscopic and combined external-transnasal endoscopic approach for the treatment of inverted papilloma: analysis of 110 cases. Eur Arch Otorhinolaryngol 2014; 271(7):1953–9. [DOI] [PubMed] [Google Scholar]
- 18.Katori H, Nozawa A, Tsukuda M. Histopathological parameters of recurrence and malignant transformation in sinonasal inverted papilloma. Acta Otolaryngol 2006; 126(2):214–8. [DOI] [PubMed] [Google Scholar]
- 19.Kim DY, Hong SL, Lee CH et al. Inverted papilloma of the nasal cavity and paranasal sinuses: a Korean multicenter study. Laryngoscope 2012; 122(3):487–94. [DOI] [PubMed] [Google Scholar]
- 20.Cannady SB, Batra PS, Sautter NB, Roh HJ, Citardi MJ. New staging system for sinonasal inverted papilloma in the endoscopic era. Laryngoscope 2007; 117(7):1283–7. [DOI] [PubMed] [Google Scholar]
- 21.Gras-Cabrerizo JR, Montserrat-Gili JR, Massegur-Solench H, Leon-Vintro X, De Juan J, Fabra-Llopis JM. Management of sinonasal inverted papillomas and comparison of classification staging systems. Am J Rhinol Allergy 2010; 24(1):66–9. [DOI] [PubMed] [Google Scholar]
- 22.Krouse JH. Development of a staging system for inverted papilloma. Laryngoscope 2000; 110(6):965–8. [DOI] [PubMed] [Google Scholar]
- 23.Lisan Q, Moya-Plana A, Bonfils P. Association of Krouse Classification for Sinonasal Inverted Papilloma With Recurrence: A Systematic Review and Meta-analysis. JAMA Otolaryngol Head Neck Surg 2017; 143(11):1104–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kim JS, Kwon SH. Recurrence of sinonasal inverted papilloma following surgical approach: A meta-analysis. Laryngoscope 2017; 127(1):52–8. [DOI] [PubMed] [Google Scholar]
- 25.Walgama E, Ahn C, Batra PS. Surgical management of frontal sinus inverted papilloma: a systematic review. Laryngoscope 2012; 122(6):1205–9. [DOI] [PubMed] [Google Scholar]
- 26.Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL Jr. Prognostic importance of comorbidity in a hospital-based cancer registry. JAMA 2004; 291(20):2441–7. [DOI] [PubMed] [Google Scholar]
- 27.Feinstein AR, Pritchett JA, Schimpff CR. The epidemiology of cancer therapy. II. The clinical course: data, decisions, and temporal demarcations. Arch Intern Med 1969; 123(3):323–44. [DOI] [PubMed] [Google Scholar]
- 28.Feinstein AR, Spitz H. The epidemiology of cancer therapy. I. Clinical problems of statistical surveys. Arch Intern Med 1969; 123(2):171–86. [PubMed] [Google Scholar]
- 29.Suh JD, Ramakrishnan VR, Thompson CF et al. Inverted papilloma of the sphenoid sinus: risk factors for disease recurrence. Laryngoscope 2015; 125(3):544–8. [DOI] [PubMed] [Google Scholar]