Long-Term Implications of Short-Term Symptom and Radiographic Severity Outcomes After ESS in CRS Patients

Abstract

Chronic rhinosinusitis (CRS) is a syndrome defined by both symptomatic presentation and objective findings, such as those visible on endoscopy and computed tomography (CT). These symptoms, endoscopic and radiographic measures, are important for guiding surgical decision-making and are widely used for post-operative surveillance. The severity of CRS symptoms, CT findings, and endoscopic findings respectively, are quantified using validated tools like the sinonasal outcomes test-22 (SNOT-22), the Lund–Mackay (LM), and the modified Lund–Kennedy (MLK) scores. Studies have shown that preoperative SNOT-22 and LM scores, whether categorized by quartiles or arbitrary ranges, generally show significant improvement following endoscopic sinus surgery (ESS) [1-4]. However, we are not aware of any studies demonstrating pre-ESS SNOT-22 scores predicting post-ESS scores. Although most studies focus on pre-ESS scores, there is limited analysis of the implications of post-operative measures on long-term outcomes. Notably, although higher post-ESS SNOT-22 scores have been found to predict the need for future revision ESS, we were unable to find any studies that studied the predictability of long-term outcomes using post-ESS treatment outcomes despite routine surveillance of these measures [5]. Furthermore, a study demonstrated that endoscopic scores tend to worsen between 6 and 12 months after ESS, raising questions on the long-term stability of both symptom and radiographic findings [6]. Here we aim to investigate post-ESS long-term stability and prognostic value of short-term symptom and radiographic status in predicting long-term symptom outcomes.

A prospective study of adult CRS patients who underwent ESS was conducted at Northwestern Medicine between 2017 and 2023. A total of 113 patients with (n = 48) and without nasal polyps (n = 65) with CT-proven bilateral disease LM≥4 were included. Excluded patients <18 or >85 years, with very minimal evidence of inflammation on CT, recurrent acute sinusitis, on anticoagulants or having coagulation disorder, received solid organ transplant and on immunosuppressants, and history of infectious diseases like Hepatitis B or Hepatitis C. They were followed up at 6–12 (V1, short-term) and 18–60 (V2, long-term) months post-ESS. At each visit, patients completed the SNOT-22 questionnaire and underwent a CT scan and endoscopy, with scores analyzed [2]. These were considered as both continuous and binary categorical variables. The Wilcoxon matched-pair test compared mean SNOT-22 and LM scores between V1 and V2, whereas the Spearman correlation assessed score correlations at both time points. Binary classification of SNOT-22 and LM was defined as <17 and <4, respectively, for symptomatic and radiographic normalization based on studies that found that these were the mean scores of patients without CRS [7, 8]. Contingency tables with normal (SNOT-22<17, LM<4) versus abnormal status (SNOT-22≥17, LM≥4) were utilized. Chi-square analysis evaluated V1 SNOT-22's long-term symptom prediction alone and with V1 LM status. Data were analyzed using Prism 9.

The study involved a total of 113 patients, of which 42.48% had nasal polyps. SNOT-22, LM, and MLK scores were evaluated at two follow-up points: short-term, with a mean duration of 8 months and a long-term, with a mean duration of 34 months. Endoscopic MLK scores had highly significant moderate positive correlation with LM scores cross-sectionally at V1 and V2 (Figure S1).

Mean SNOT-22 scores decreased significantly (p < 0.0001) from pre-ESS (45.97 ± 19.99) (Figure S2a) but showed no significant difference between V1 (19.06 ± 15.95) and V2 (20.67 ± 17.41) (Figure 1a). Similarly, mean LM scores were significantly (p < 0.0001) lower than pre-ESS (12.23 ± 4.93) (Figure S2b), with no significant difference between V1 (4.65 ± 4.7) and V2 (5.12 ± 4.2) (Figure 1b). A moderate significant positive correlation was observed between V1 and V2 SNOT-22 scores (r = 0.59, p < 0.0001) and between V1 and V2 LM scores (r = 0.49, p < 0.0001). However, no significant correlation existed between LM and SNOT-22 scores at or between V1 and V2 (Figure 1c). Correlation with pre-ESS scores is shown in (Figure S2c).

Chi-square 2 × 2 contingency tables assessed the stability and prognostic ability of normal versus abnormal classification at V1 and V2. At V1, 51% patients achieved symptomatic normalization, slightly decreasing to 50% at V2. Radiographic normalization occurred in 53% at V1, dropping to 42% at V2. The prognostic ability of V1 SNOT-22 for V2 SNOT-22 was significant (χ² 14.91, df = 1, p < 0.0001). Patients with symptomatic normalization at V1 had a 67% chance of maintaining it at V2, whereas those with  abnormal symptoms had a 69% chance of persistently elevated symptoms (Table 1).

Combining radiographic and symptomatic normalization at V1 improved predictive power. V1 SNOT-22 and LM status significantly predicted V2 SNOT-22 status (χ² 15.50, df = 3, p = 0.0014). A pairwise chi-square analysis was performed to determine groups driving significance. Here we report one pairwise comparison; the overall 4 × 2 table and remaining three pairwise comparisons are detailed in Table S1. Patients with normalization of both SNOT-22 and LM at V1 had a 69% chance of maintaining symptomatic normalization at V2, slightly higher than using SNOT-22 at V1 alone (67%). Those with abnormal SNOT-22 and LM status at V1 had a 74% chance of remaining symptomatically abnormal at V2, representing a modestly higher negative predictive value than SNOT-22 alone (69%) (Table 1).

This longitudinal study examines stability of long-term symptom and radiographic scores and the predictability of long-term symptoms using short-term symptoms and radiographic measures after ESS. Our findings indicate that short-term and long-term SNOT-22 and LM scores are relatively stable and improved from pre-ESS values. These are similar in magnitude to other studies reporting SNOT-22 and more recently LM scores post-ESS, reinforcing the long-term stability of symptom and radiographic measures [1, 9].

We found no significant correlation between SNOT-22 and LM scores before and after ESS, consistent with meta-regression analysis showing no correlation between objective and subjective measures [10]. However, a significant moderate longitudinal correlation between short- and long-term SNOT-22 scores suggests that short-term SNOT-22 is a reliable predictor of long-term outcomes. Patients with normalized SNOT-22 status shortly after ESS had a 67% chance of maintaining symptomatic normalization long-term.

Few studies have concurrently examined post-surgical patient-reported outcomes and clinically observed measures over time. One study identified that minimal clinically important changes in the SNOT-22 scores within 12 months post-ESS predict the need for revision ESS [5] but the clinical reasoning for these revisions was not detailed. Our group has also shown significant correlations between radiographic opacification and intensity of residual inflammation measured by cytokines post-ESS, suggesting that objective measures better relate to inflammation intensity than patient reported ones [2]. This longitudinal study's mean 3-year follow-up duration with both SNOT-22 and LM scores covers a clinically relevant outcome window. However, we find that the best long-term predictor of long-term SNOT-22 outcomes was the short-term response. The addition of objective radiographic measures to the patient reported ones only improved accuracy of prognostication slightly.

David B. Conley reports personal fees for Intersect ENT and XORAN. Robert C. Kern reports personal fees from Sanofi, Novartis, Lyra Pharmaceutical, and Neurent. Bruce K. Tan reports personal fees from Sanofi Regeneron/Genzyme and GSK. The rest of the authors declare no conflicts of interest.