The Pediatric Sinus Staging System: A CT-based Approach to Grading Pediatric Sinus Disease

Abstract

Objectives:

Pediatric chronic rhinosinusitis (CRS) is a prevalent disease with few objective measurements available to predict which patients will require surgical intervention. The Lund-Mackay (LM) score for CT scans is one objective data point available for the adult population; however, a dedicated scoring system in the pediatric population has not been popularized. We present a pediatric sinus staging system (PSSS) which considers both opacification and the varying developmental stages of each sinus.

Methods:

This was a retrospective chart review in which we analyzed CT scans of pediatric patients with a diagnosis of CRS. Both LM and PSSS scores were calculated for each scan. Groups were formed based on treatment outcomes and included patients who 1) were treated successfully with medical therapy and/or adenoidectomy (med/adenoid), 2) required FESS and 3) required revision FESS.

Results:

Overall, 76 patient scans were reviewed. PSSS values were significantly less than LM for the control (p=0.001) group and significantly higher for patients with cystic fibrosis (p=0.027) and with CRS with polyps (p=0.001). The ideal cutoff for PSSS to distinguish between med/adenoid and single FESS treatment with a sensitivity 90.6% and specificity of 50.0% was ≥2.

Conclusion:

The PSSS gives a more descriptive score by accounting for the opacification and pneumatization of each sinus. Our current results show similar values between PSSS and LM scores which suggests internal validity. In addition, a PSSS score of ≥2 may help physicians better counsel families on the likelihood of requiring FESS. Further investigation is needed to fully validate the PSSS.

Level of Evidence: 3B

Introduction:

Pediatric chronic rhinosinusitis (CRS) is a diverse and poorly understood disease that leads to reduced quality of life and significant antibiotic exposure (1). Pediatric CRS accounts for 5.6 million outpatient visits per year in the United States alone in comparison to 1.6 million visits per year for acute rhinosinusitis (2). Annually, it is responsible for 2.1% of national ambulatory visits for patients under 20 years old. Pediatric CRS patients have higher rates of unhappiness, nervousness, home behavioral issues, comorbid anxiety, depression, and asthma (3). Pediatric CRS also contributes to the prolonged use of antibiotics, as the first line therapy recommended by the American Academy of Otolaryngology-Head and Neck surgery (AAO-HNS) is 20 days of oral antibiotics (4). It also leads to significant absenteeism from school for patients and work for caregivers. Diagnosing pediatric CRS can be challenging for providers as we must rely on symptomatology reporting from caregivers to meet the AAO-HNS criteria for pediatric CRS (4). In many cases imaging can play a pivotal role in the diagnosis and management of CRS and help guide physicians as to the severity and chronicity of disease (5).

The most commonly used classification for staging sinus disease on computed tomography (CT) is the Lund-Mackay (LM) system. This was originally developed for an adult population but has been widely used in pediatrics (6). There are numerous advantages to this system including its familiarity, ease of use, and high inter-observer reliability (7). However, there are shortcomings, specifically, the oversimplification of the sinus opacification score and the lack of consideration for underdeveloped and absent sinuses. The latter is more important in a pediatric population when grading CT scans to accurately reflect the severity of disease.

Some studies have reported new staging systems which score opacification with increased levels of gradation. These systems, namely the modified Lund-Mackay and Zinreich system, have more classification tiers than the traditional LM system (8, 9). However, they have not shown superiority to the LM system and are not used in mainstream literature (7).

Pediatric paranasal sinuses are vastly different from adult sinuses and even vary significantly amongst pediatric age groups. There are numerous studies detailing the developmental timeline of pediatric sinuses in cadaveric and imaging models (10–13). It is generally accepted that sinuses develop in the following order: ethmoids, maxillary, sphenoid, and lastly, frontal; however, there is significant variability in the timeframe of complete pneumatization of each sinus.

The careful review of CT imaging is crucial when considering surgery for sinus disease. The sinuses are assessed for underdevelopment, opacification, and anatomic variants, which factor into the overall decision to offer surgery and the extent of surgery to be performed. For example, a completely opacified frontal bud is not given the same weight as an opacified, fully developed frontal sinus. However, there have not yet been any studies that attempt to turn this qualitative process into a concrete scoring system with quantitative data. Absent sinuses on imaging have been addressed in previous studies, which take the absent sinus out of the equation and apply a correction factor to make scores comparable across age groups (14–16).

The goal of this study was to create a universal sinus staging system for the pediatric population based on CT imaging that addresses opacification and sinus development. Our proposed system, the Pediatric Sinus Staging System (PSSS), incorporates not only a new opacification scale but also addresses the underdeveloped and absent sinus with the goal of allowing accurate comparison of different age groups across a clinical and research setting while preserving ease of use.

Materials and Methods:

An institutional review board approved (IRB PRO17060492) retrospective chart review of patients seen at a tertiary care pediatric hospital in a specialized otolaryngology rhinology clinic was performed. Children with CT maxillofacial scans at age 1 – 18 years were included in this review. Patients with facial fractures, silent sinus syndrome, choanal atresia, acute sinus infections at time of scan (as reported by accompanying documentation), skull base tumors, patients with underlying syndromes, and those who received functional endoscopic sinus surgery (FESS) at an outside hospital or prior to 2010 were excluded. Patients with CRS, nasal obstruction, cystic fibrosis, and immunodeficiency were included. A total of 76 patients were included.

The diagnosis of pediatric CRS was based on the AAO-HNS 2014 consensus statement and defined by three months of two or more of the following symptoms: purulent rhinorrhea, facial pressure/pain, nasal obstruction, cough, and either endoscopic or CT changes suggestive of sinus disease (4). The completion of medical therapy was also defined by the guidelines and included twenty consecutive days of antibiotics, topical nasal steroid spray and daily nasal irrigations. In our cohort, patients were first divided into CRS and controls. The CRS group was then further stratified into patients who were treated successfully with medical therapy and/or adenoidectomy (med/adenoid), patients who underwent FESS, and patients who underwent repeat FESS. The control group of patients had a CT maxillofacial scan for reasons other than CRS.

Patient factors were then collected including demographic profile, related comorbidities such as allergic rhinitis (AR) and cystic fibrosis (CF), extent of surgery, traditional LM score and our proposed PSSS score. The CT scan prior and closest to surgical intervention (if applicable) was reviewed and the LM score was calculated. The PSSS score was calculated by grading the degree of opacification and subsequently looking at the development of each sinus. The grading of all CT scans was completed by two research team members (an Otolaryngology resident and fellow). For the opacification score, similar to the LM system, each sinus (which included the maxillary, anterior ethmoids, posterior ethmoids, sphenoid and frontal sinus) was graded on a scale of 0–2. However, the PSSS score graded a 0 for 0% - 25% opacified sinus, 1 for 26% - 75%, and 2 for 76% −100% opacification (Figure 1). The OMC was given a score of 0 if not occluded or 2 if occluded.

Figure 1.

Left maxillary sinus (†) with varying degree of opacification and corresponding Lund-Mackay (LM) and Pediatric Sinus Staging System (PSSS) scores. 1.1. Minimal opacification. LM score = 1; PSSS score = 0; 1.2. Moderate opacification. LM score = 1; PSSS score = 1; 1.3. Near-total opacification. LM score = 1; PSSS score = 2.

The developmental pneumatization of each sinus was then scored as either absent (A), underdeveloped (U), or developed (D). The grading of underdeveloped versus developed was based on previously published criteria by Shah et al (13). The landmarks used to define our criteria of a developed sinus are illustrated in Figure 2. If one of these landmarks was not met, the sinus was classified as underdeveloped.

Figure 2.

Varying developmental classifications based on anatomic landmarks previously published by Shah et al. (9). To meet the criteria for a fully developed sinus, the fontal sinus has to extend superiorly to the vertical portion of the frontal bone (2.1–2). The maxillary sinus should extend inferiorly below the nasal floor and laterally to the zygomatic process (2.3). The ethmoids should extend superiorly to the cribriform plate (2.3). Lastly, the sphenoid should extend superiorly and laterally to the optic strut (2.4).

An underdeveloped sinus was given a multiplier of 0.5 to its corresponding opacification score; therefore, an underdeveloped categorization downgraded the opacification score of that particular sinus by half. Absent sinuses were taken out of the overall calculation which changed the maximum scoring potential for that CT. To compensate for this, the total score for each CT with an absent sinus was multiplied by a correction factor. The correction factor was calculated by dividing 24 by the maximum possible score for that CT scan. This process allowed the scores to be comparable, ensuring that both LM and PSSS had maximum scores of 24.

For an example calculation, a CT with bilateral, underdeveloped and completely opacified maxillary sinuses would get a score of ‘2U’. The ‘U’ represents the 0.5 multiplier, giving each maxillary a final score of 1. If a patient has bilateral absent frontal sinuses, a score of ‘A’ is given for both sinuses, and these sinuses are taken out of the overall equation. If these were the only absent sinuses, the maximum total score this CT could obtain is 20. After all individual sinus and OMC scores are calculated, the correction factor is calculated. In this case, it is 24 divided by 20, which gives us 1.2. This is ultimately multiplied by the final score. A sample calculation is illustrated in Figure 3. Of note, anatomical variations of the nose and paranasal sinuses were not included in data collection as this has previously been shown to have no predictive value on the disease extent and severity (17).

Figure 3.

Sample calculation for PSSS. Right OMC: 2, Left OMC: 2, right maxillary: 2U=1, left maxillary: 2U=1, right anterior ethmoid: 2D=2, right posterior ethmoid: 2D=2, left anterior ethmoid: 2D=2, left posterior ethmoid: 2D=2, right sphenoid: 2U=1, left sphenoid 2U=1, right frontal=A, left frontal=A. Total is 16 out of 20. With correction factor of 1.2 applied, 19.2/24 is obtained. The PSSS score for this scan is 19.2.

Categorical patient characteristics were summarized as n (%). Binomial exact calculation was used to determine 95% confidence intervals for proportions. For the continuous variables, normality was determined using Shapiro-Wilk tests. All variables were not normally distributed and were therefore summarized as median (range). Interrater agreement was evaluated using weighted kappa statistics. Discrepancies in sinus development designations were resolved by a third rater (Otolaryngology attending), who independently reviewed CT scans blinded to the original two reviewers’ ratings.

For LM and PSSS scores, the agreement between the two original raters was evaluated using Spearman rank correlation. Scores of the two raters were averaged, and comparisons of LM or PSSS scores between patients with different characteristics, comorbidities, and levels of disease severity were conducted using Wilcoxon rank-sum or Kruskal-Wallis tests. Wilcoxon signed-rank tests were employed for comparisons between LM and PSSS scores within the same groups of patients. Spearman rank correlation coefficients were calculated to describe the relationships between age at CT and LM or PSSS for all patients, and then separately for each treatment group. Finally, receiver operating characteristic (ROC) curves were constructed to compare the abilities of the LM and PSSS scores and establish cutoffs to distinguish between markers of disease severity. All comparisons were conducted using Stata/SE 16.0 (StataCorp, College Station, TX).

Results:

Demographics and Patient Stratification

Demographic details are listed in Table 1. Median age at CT scan was 10.4 years (range 0.9–17.9). Comorbidities included AR (27.6%), CF (15.8%), immunodeficiency (9.2%), cleft palate (1.3%), and craniosynostosis (1.3%). Controls, med/adenoid group, FESS and multiple FESS surgeries comprised 18.4%, 23.7%, 42.1% and 15.8% patients, respectively. Surgical interventions included adenoidectomy (60.5%) and FESS (57.9%). Inferior turbinate reduction (ITR) was performed in 44.4% of med/adenoid group; 35.5% of the FESS group; and 36.4% of multiple FESS group.

Table 1.

Patient Characteristics Impacting Sinus Scoring

			Lund-Mackay (LM)			Pediatric Sinus Staging System (PSSS)			LM vs PSS
Characteristic	n	%	median	range	p†	median	range	p†	p‡
Sex					0.522			0.407
Male	36	47.4	10.5	0–24.0		10.0	0–24.0		0.898
Female	40	52.6	9.3	0–22.5		5.9	0–24.0		0.987
Allergic Rhinitis					0.343			0.360
Without	55	72.4	10.3	0–23.5		8.6	0–24.0		0.986
With	21	27.6	7.0	0–24.0		6.0	0–24.0		0.847
Cystic Fibrosis					<0.001			<0.001
Without	64	84.2	6.3	0–24.0		5.4	0–24.0		0.178
With	12	15.8	17.3	12.5–23.5		19.7	15.0–23.5		0.027
CRS (n=62)					<0.001			<0.001
Without Polyp	39	62.9	7.0	0–17.0		5.8	0–20.7		0.055
With Polyp	23	37.1	16.5	7.0–24.0		18.5	6.0–24.0		0.001
Age Group					0.067			0.042
0–6 years	27	35.5	12.0	0–24.0		14.7	0–24.0		0.110
7–11 years	24	31.6	5.0	0–23.5		3.1	0–24.0		0.149
12–17 years	25	32.9	6.0	0–21.0		5.0	0–23.5		0.443
Age at CT	n	%	r		p§	r		p§
All Patients	76	100.0	−0.257		0.025	−0.279		0.015
Control	14	18.4	0.147		0.617	0.080		0.786
Med/Adenoid	18	23.7	−0.651		0.003	−0.719		0.001
FESS	32	42.1	−0.108		0.558	−0.106		0.565
Multi FESS	12	15.8	−0.081		0.803	0.042		0.897

CRS: Chronic Rhinosinusitis; FESS: Functional Endoscopic Sinus Surgery; Med/Adenoid: Medical Therapy +/− Adenoidectomy. Bold indicates p<0.05.

^†Scores were compared between groups of patients using Wilcoxon rank-sum or Kruskal-Wallis tests.

^‡LM was compared with PSSS using Wilcoxon signed-rank tests.

^§The Spearman rank correlation between age at CT and LM or PSSS was calculated for all patients and then separately for each treatment group.

Sinus Development

One important aspect in the development of the PSSS was determining the trajectory of sinus development in our pediatric cohort. Patients were divided into three age groups based on previous studies to capture development and disease progression (0–6, 7–11 and 12–17 years). The proportions of patients with underdeveloped or absent sinuses in each age group are shown in Figure 4. All patients had fully developed ethmoid sinuses. The two reviewers demonstrated 83.6% agreement in ratings of maxillary sinus development (weighted κ=0.527, SE=0.100, p<0.001), 87.8% agreement regarding sphenoid sinus development (weighted κ=0.517, SE=0.085, p<0.001), and 87.5% agreement regarding frontal sinus developmental (weighted κ=0.728, SE=0.088, p<0.001). As expected, the 12–17-year-old group demonstrated the smallest proportion of absent or underdeveloped sinuses. The median ages of patients with fully developed sinuses were 11.2 years (range 2.6–17.9) for maxillary, 11.7 years (range 5.2–17.9) for sphenoid and 13.5 years (range 3.4–17.9) for frontal.

Figure 4.

Proportions of patients with absent or underdeveloped sinuses in each age group. Error bars indicate 95% confidence intervals. All patients had fully developed ethmoid sinuses.

Differences between LM and PSSS

In order to help validate the PSSS, analysis was performed to describe any differences between the ability of the LM and PSSS scores to distinguish between treatment groups. Spearman rank correlation revealed very strong positive correlation between the two raters for both the LM (ρ=0.947, p<0.001) and PSSS (ρ=0.953, p<0.001) scores. When the scores from the two raters were averaged, median PSSS scores were 0.3 (range 0.0–5.0) in the control group, 1.6 (range 0.0–20.7) in the med/adenoid group, 12.9 (range 0.0–24.0) in the single FESS group, and 16.8 (range 7.5–24.0) in the multiple FESS group. Median LM scores were 2.0 (range 0.0–6.0) in the control group, 3.0 (range 0.0–16.0) in the med/adenoid group, 12.3 (range 0.5–22.5) in the single FESS group, and 14.0 (range 6.5–24.0) in the multiple FESS group. PSSS values were significantly less than LM for the control (p=0.001) (Figure 5). In addition, ROC analysis showed no significant differences between LM and PSSS between the treatment groups (Figure 6). ROC analysis also showed that the ideal cutoffs (based on the greatest percent of patients correctly classified) to distinguish between med/adenoid and single FESS were ≥7 for LM (sensitivity: 81.3%, specificity: 66.7%, positive likelihood ratio: 2.44, negative likelihood ratio: 0.28; area under the curve (AUC): 0.791) and ≥2 for PSSS (sensitivity: 90.6%, specificity: 50.0%, positive likelihood ratio: 1.81, negative likelihood ratio: 0.19; AUC: 0.756). In other words, of the 43 patients with LM≥7, 26 (60.5%) failed management with medication and/or adenoidectomy, and of the 38 with PSSS≥2, 29 (76.3%) failed.

Figure 5.

Lund-Mackay (LM) and Pediatric Sinus Staging System (PSSS) scores for each treatment group. Treatment group had a significant impact on both LM and PSSS scores (p<0.001 for both systems). When comparing the two systems, PSSS values were significantly less than LM for the control group, **p=0.001 for comparison of LM vs PSSS using Wilcoxon signed-rank tests.

Figure 6.

Receiver operating characteristic (ROC) curves showing no significant difference in area under the curve for the Lund-Mackay (LM) and Pediatric Sinus Staging System (PSSS) in discriminating between patients in the medical therapy +/− adenoidectomy and single functional endoscopic sinus surgery groups.

Impact of Patient Characteristics on Sinus Scores

A secondary goal of this study was to determine which patient characteristics were associated with greater LM and/or PSSS scores. The impact of patient characteristics on PSSS and LM scores are shown in Table 1. Both LM and PSSS scores were significantly greater in patients with CF and in patients with CRS with polyps compared to those without. However, PSSS scores were significantly greater than LM scores for patients with CF (p=0.027) and patients with CRS with polyps (p=0.001). AR did not have a significant impact on scores using either system.

Discussion:

Pediatric CRS is a complex and poorly understood disease. CT imaging can be an essential tool for the pediatric otolaryngologist to determine disease severity and specific sinus involvement. The LM score is commonly used to stage sinus disease on CT; however, this has several limitations, namely the lack of gradation of opacification as well as the lack of consideration of underdeveloped/ absent sinuses. To date, there has been no published standardized grading system specific to the pediatric sinus population.

Like the LM score, the PSSS was scored on a scale of 0–2, with a maximum score of 24. Our opacification percentages differ in that a sinus score of 0 encompasses not only the absence of any disease, but also disease that is minimal and clinically insignificant. In the same vein, our score of 2 encompasses complete opacification along with disease that would be considered severe.

The truly unique aspect of our study is the inclusion of underdeveloped sinuses in our scoring system. The division of each underdeveloped sinus opacification score in half was based on acknowledging that a surgeon was less likely to operate on an underdeveloped sinus and also wanting to keep the scoring process easy and straight forward. Scoring systems are best utilized when they are concise, clear, and easy to perform. Although a more precise gradation scale could be used, this is unlikely to make a significant difference in overall scoring. In addition, the calculation would be burdensome and thereby decrease the likelihood of use. In our cohort, the overwhelming majority of patients in the 0–6 years age group had underdeveloped or absent sinuses. The largest percentage was the frontal sinus which reflects current understanding from sinus development studies (10–13). Taking absent sinuses out of the overall total and multiplying by a correction factor is not novel to our project (14–16). This seems logical, as it does not make sense to grade a sinus that is not present. When considering surgery and the extent, most surgeons will take development into consideration and an appropriate scoring system should as well.

The second part of our system was the change in opacification scoring. There have been other proposed scoring systems which increase gradation of disease with subsequent change of overall score to greater than 24. This is problematic as there is substantial familiarity with a grading system out of 24, and in addition, this number is commonly used within sinus research to communicate findings. Other papers have proposed that CT volumetric analysis provides a more global analysis of overall sinus disease and have suggested that this may correlate better with sinus disease in an adult population (18); however, despite this, volumetric analysis has not been extensively used.

One of the unique aspects of the population studied in this report is that over half (57.9%) received surgical intervention with FESS. This is not representative of most pediatric otolaryngology clinics and may be due to absence of imaging in patients managed with medical therapy +/− adenoidectomy. As expected, CF patients (15.8%) had significantly higher scores (both LM and PSSS), and repeat surgery was common in this group (41.7%).

Our study design included a control group and three different treatment groups. Each treatment group was used as a marker of increasing disease severity (mild, moderate, severe disease). In our current data, both the LM and PSSS were significantly different between groups; however, there was no difference between LM and PSSS in distinguishing between treatment groups. This suggests that we were unable to demonstrate superiority of either scoring system in distinguishing between these markers of disease severity. Interestingly, the PSSS was significantly lower than the LM score for the control group. This is likely secondary to the opacification gradation, as those with minimal disease are downgraded to a 0 in the PSSS as opposed to a LM score of 1.

Our analysis showed ideal cutoffs to distinguish between single FESS and adenoid/med were ≥7 for LM (sensitivity: 81.3%, specificity: 66.7%) and ≥2 for PSSS (sensitivity 90.6%, specificity 50%). This has the potential to impact clinical practice and guide patient/parent counseling. One study showed around 50% patients with CRS fail adenoidectomy and suggested that predictors of failure include: age <7 years, presence of asthma (19). Similar to these risk factors, the PSSS can be used in the pre-operative workup to better counsel families regarding the need for FESS in the future. By using the PSSS instead of the LM, clinicians can more accurately grade their patient’s CT scans and advise parents that if they have a PSSS ≥ 2 there is an approximately 75% chance they will fail medical/adenoidectomy only. This, in conjunction with other important clinical information (age, co-morbidities, previous antibiotic use), will help clinicians to more accurately diagnose and manage pediatric CRS patients.

There are several limitations of our study, including the retrospective design. The diagnosis of CRS was obtained from electronic medical records where it was not possible to explore symptom severity. In addition, our population is skewed towards those more likely to undergo surgery as this is the group that most frequently undergoes imaging.

Conclusion:

The PSSS offers a new way of looking at CT sinus disease in a pediatric population which includes a change in the grading of opacification, the incorporation of a multiplier for underdeveloped sinuses, and a correction score for absent sinuses. We found that a PSSS score of ≥2 distinguishes between children with sinus disease who are more likely to require FESS as opposed to medical therapy/ adenoidectomy alone with a sensitivity of 90.6% and specificity of 50.0%. This finding may help physicians better counsel patients presenting with sinus disease on the likelihood of requiring surgery, regardless of age and sinus development. Overall, the PSSS was not found to be significantly different than the LM score in distinguishing between treatment groups. Further investigation is required to fully validate the PSSS.