Introduction

The prevalence of olfactory dysfunction in the general population is a matter of debate and frequently underestimated. In a large population-based German survey of 1277 participants aged 25–75 years, a prevalence of 3.6% functionally anosmic and 18% hyposmic subjects was found [1]. While approximately 75,000 consultations per year at German ENT clinics account for smell impairment, the main reason for olfactory loss is found in sino-nasal diseases followed by postinfectious and idiopathic smell loss [2].

Olfactory loss is highly prevalent in patients with sino-nasal disease such as chronic rhinosinusitis (CRS) [3,4,5]. However, the underlying pathology is not fully understood and so far no clinical staging system highly correlated with the olfactory function as to be useful in a daily clinical setting. Based on the two pathophysiological mechanisms of olfactory dysfunction suggested in CRS patients—the conductive (swollen or hypertrophic mucosa) and the sensorineural (affecting directly olfactory sensory neurons) loss [6,7,8]—several studies tried to identify and quantify the inflammation of the sinuses and the olfactory cleft (OC) using different CT scan analysis and to correlate with the olfactory function [9, 10].

Commonly used CT staging systems for CRS like the Lund–Mackay Score focus specifically on the sinuses rather than quantifying the extent of disease in the OC [11]. But recent studies analyzing the OC opacification by means of three-dimensional volumetric measurements show correlations between OC opacification and smell tests to some extent [12, 13]. With the sinus-specific scoring systems showing no such promising correlations, olfactory cleft changes seemed more predictive of olfactory function than sinus opacification.

Nasal endoscopy was used to assess patients’ response to medical therapy, evaluate surgical need or to predict the need for revision surgery [14]. Reports on the correlation between postoperative Lund–Kennedy scores and quality of life have been conflicting [15,16,17]. Correlations with the olfactory function in patients with sino-nasal or any other entities of smell disorder has been weakly described in literature and remains mostly unclear. Only recently a paper by Soler et al. focused on the correlation between endoscopic findings in the OC and olfactory function in patients with chronic rhinosinusitis [18]. However, this elegant study failed to look at controls and forms of olfactory loss beyond CRS. Therefore, the goal of this study was to determine whether the Lund–Kennedy score of the olfactory cleft correlates with the olfactory function in smell-impaired patients and whether this correlation is superior to the conventional sinus-specific Lund–Kennedy staging system. Secondary goals are to investigate correlations in relation to the characteristics of smell impairment (i.e., self-ratings of olfaction and nasal breathing, results from questionnaires related to nasal function) and the OC scoring system.

Materials and methods

Study cohort

A prospective study was conducted at the Smell and Taste Clinic (Department of Otolaryngology) of the “Technische Universität” (TU) Dresden. The study was performed according to the Declaration of Helsinki and approved by the Ethics Committee of the Medical Faculty of the TU Dresden (application number: EK122032011). Only participants above the age of 18 were included in the study and all experiments were undertaken with the understanding and written consent of each participant. The following exclusion criteria were applied: pregnancy, neurological diseases and any systemic disease associated with smell disorders like chronic renal failure or untreated thyroid disorders. If—due to superior–anterior septal deviations for example—the assessment of the entire olfactory cleft was not possible, the participant was further excluded from the study. Participants were instructed to only drink water one hour prior to the experiment and further not to wear any scented products on the day of testing.

According to the cause of smell impairment, participants were categorized in five groups: (0) control, (1) idiopathic, (2) sino-nasal, (3) postinfectious and (4) post traumatic olfactory loss (see position paper on olfactory dysfunction [19]).

While postinfectious olfactory loss was defined as a smell impairment following an upper respiratory tract infection, post traumatic olfactory loss occurs as a result of a head trauma. Sino-nasal olfactory loss, typically fluctuating in nature, arises from inflammatory disease of the nasal mucosa, including smell impairment due to allergic/non-allergic rhinitis and chronic rhinosinusitis with and without nasal polyps. If no underlying pathology for olfactory impairment could be evaluated through well-established diagnostics, participants were diagnosed as having an idiopathic smell loss. And finally, participants with a normal sense of smell, represented in a TDI score (composite threshold–discrimination–identification score) of > 30.5 [20] were classified as the control group.

To evaluate the mucosal conditions with focus on the olfactory cleft, each participant underwent nasal endoscopy through an experienced ear, nose, and throat specialist. Nasal endoscopy exams were scored semi-quantitatively using the Lund–Kennedy staging system. Olfactory function was quantified by means of the extended Sniffin’ Sticks test and nasal symptoms were assessed using the sino-nasal outcome test—20 (see below). Further, participants were asked to rate their olfactory ability and nasal airflow on Likert-type visual analogue scales ranging from 1 to 10 (1—no olfaction/nasal airflow to 10—very good olfaction/nasal airflow).

Nasal endoscopy and Lund–Kennedy score

Nasal endoscopy was performed in each participant during consultation at the Smell and Taste Clinic. Initially the internal nose was scored semi-quantitatively using the Lund–Kennedy scoring system (classical Lund–Kennedy Score, c-LK-Score) with subsequent focus on the olfactory cleft. Here, a modified Lund–Kennedy score (olfactory cleft Lund–Kennedy score, OC–LK score) was used to describe the mucosa of the olfactory cleft separately as detailed in Table 1. Each side was evaluated separately with a total possible score of 10 on each side and each region (whole nose/olfactory cleft). The olfactory cleft was regarded as a three-dimensional space consisting of following borders: anteriorly defined through the anterior attachment of the middle turbinate, posteriorly the anterior wall of the sphenoid sinus, medially the nasal septum and laterally through the middle and superior turbinate. The superior boundary was the cribriform plate, whereas the inferior limitation was set around 1 cm below the skull base. In line with the literature in which olfactory cleft opacification was measured using CT images [10], we divided the olfactory cleft in an anterior and a posterior portion separated by the anterior end of the superior turbinate. If the superior turbinate was not present or sufficiently visible, the posterior third of the middle turbinate was defined as the border between anterior and posterior olfactory cleft. If only one portion was affected by the polyps—either the anterior or posterior—1 point (< 50%) was added to the OC–LK Score. However, if both portions were affected or polyps crossed the defined border, 2 points (> 50%) were added to the score. Figure 1 pictures different mucosal conditions of the olfactory cleft as used in the modified Lund–Kennedy score of the OC (OC–LK Score).

Table 1 Lund–Kennedy endoscopic scoring system. Endonasal endoscopy with (A) evaluation of the endonasal mucosa (classical Lund–Kennedy Score, c-LK Score) and (B) evaluation of the olfactory cleft mucosa (olfactory cleft Lund–Kennedy Score, OC–LK Score)
Full size table
Fig. 1
figure 1

Pictures different mucosal conditions of the olfactory cleft as used in the modified Lund–Kennedy score of the OC (OC–LK Score). a Polyps left nostril with the thin arrow pointing at the middle turbinate and the thick arrow at the polyps within the OC. Polyps can be seen posterior to the middle turbinate (> 50% OC affection, 2 points in OC–LK score respectively). **Discharge. b Edema of the right OC (arrow) containing thick discharge. Narrowing of the cleft region due to distinct mucosa swelling of the OC (accounting for 2 points in OC–LK score). c Discharge of the right OC. Bulky arrow indicating thick discharge (2 points in OC–LK score respectively) and the slim arrow pointing at the middle turbinate

Full size image

Olfactory testing

Olfactory function was quantified in a total of 288 participants using the Sniffin’ Sticks test which consists of three subtests [21]. During the odor threshold task, subjects had to detect the odorized (phenylethylalcohol, PEA) pen among three samples with the other two pens containing odorless propylene glycol. With the odor discrimination task again a triplet of pens was presented to the subject who had to discriminate one different odor amongst two identical odors. Regarding the odor identification task, subjects had to smell a single odorous pen and choose the correct answer from a list of four descriptors. The sum of the three tests (threshold–discrimination–identification) accounted for the composite TDI-score—demonstrating the final olfactory test result—with a maximum of 48 points (each subtest added a maximum of 16 points) [22]. Normosmia was set at 30.5 points or more in the composite TDI score, whereas hyposmia was diagnosed at scores between 16.5 and 30.5 points and functional anosmia below 16.5 points.

Sino-nasal outcome test-20 (SNOT-20)

Each of the 288 participants completed the SNOT-20 Questionnaire. This questionnaire measures both sino-nasal symptoms and general quality-of-life parameters in a composite fashion poses the sino-nasal outcome test-20 (SNOT-20). It is a modification of the 31-item rhinosinusitis outcome measure and contains 20 questions [23] which are based on a 0–5 scale, where 0 defines no problems and 5 defines maximal problems with the given symptom (Table 2).

Table 2 Sino-nasal outcome test (SNOT-20)
Full size table

Statistical analysis

Using SPSS 23.0 (SPSS Inc. Chicago, Ill., USA) data were statistically analyzed by means of t-tests for independent samples to investigate differences between normosmic controls and smell impaired patients. To evaluate differences by group (control 0, patients 1–4) multivariate analysis of variance (ANOVA) was applied. Pearson’s correlation coefficients were used to evaluate correlations between study variables. The level of significance was defined as p < 0.05.

Results

Study cohort

The overall cohort included 288 participants out of which 48 (29 women, 19 men, ∅ age 55, range 21–85, SD 15.7) were controls with normal Sniffin’ Sticks test results and 240 (137 women, 103 men, ∅ age 59, range 18–89, SD 14.5) with an impaired sense of smell. According to the cause of smell impairment patients were categorized in four different groups. Group 1 consisted of 116 patients with an idiopathic smell disorder. Group 2 containing patients with sino-nasal smell impairment included 33 patients and group 3 involved 59 patients with postinfectious olfactory loss. Group 4 comprising patients with post traumatic olfactory loss included 32 patients. Demographics, duration of disease, number of smokers and the occurrence of parosmia within each group are summarized in Table 3.

Table 3 Demographics, duration of disease, number of smokers and the occurrence of parosmia by group
Full size table

There were no statistically significant differences in sex distribution (Chi2 [4] = 7.31, p = 0.12) or the proportion of smokers (Chi2 [4] = 4.65, p = 0.33) by group. As expected, significant differences appeared in the occurrence of parosmia between groups (Chi2 [4] = 29.23, p < 0.001). Using univariate analysis of variance (ANOVA), no significant differences emerged in age distribution by group [F(4, 281) = 2.2, p = 0.74]. However, significant differences could be found in the duration of disease [F(3, 230) = 3.84, p = 0.01]. Regarding the duration, the most relevant difference within the diseased groups was shown between the groups with sino-nasal and postinfectious olfactory loss (p = 0.04) with a longer duration for the sino-nasal group.

Olfactory testing, self ratings of olfaction and nasal airflow

Olfactory test results from the Sniffin’ Sticks test and self ratings of nasal breathing and smell ability by group are shown in Table 4. ANOVA revealed highly significant differences in threshold [F(4, 282) = 24.90, p < 0.001], discrimination [F(4, 282) = 16.13, p < 0.001], identification [F(4, 283) = 31.47, p < 0.001] and TDI scores [F(4, 282) = 35.93, p < 0.001] between the groups. As expected, post-hoc analysis presented main differences between control group and each of the smell impairment groups (ps < 0.001). Further, significant differences appeared between group 3 and 4 in threshold (p = 0.008), identification (p < 0.001) and TDI (p < 0.001) score and between group 1 and 4 in identification (p < 0.001) and TDI (p = 0.002) score. Regarding the self ratings significant differences were demonstrated by group [olfaction: F(4, 257] = 85.61, p < 0.001; breathing: F(4,256) = 4.20, p = 0.003). The post-hoc-test showed significant differences between group 0 and group 2 (p = 0.001) and between group 2 and 4 (p = 0.017) in nasal breathing ratings. In terms of olfactory ratings, significant differences emerged between the control group and each of the other groups 1–4 (ps < 0.001) but also between groups 1 and 4 (p = 0.013) and groups 2 and 4 (p = 0.011).

Table 4 Sniffin’ Sticks test scores and self ratings of nasal breathing and smell by group
Full size table

Lund–Kennedy score analysis and correlations

Table 5 summarizes the classical Lund–Kennedy scores of each nostril (c-LK score left/right) and the modified LK score of the olfactory cleft (OC–LK score left/right) for each side within each group. Comparing normosmic controls (group 0) with all smell-impaired patients (group 1), a significant difference emerged in the OC–LK score on both sides with higher scores seen in patients (left: t [124.5] = − 3.89, p < 0.001, group 0: M = 0.3, SD = 0.6, group 1: M = 0.8, SD = 1.1; right: t [144] = − 4.01, p < 0.001, group 0: M = 0.4, SD = 0.5, group (1): M = 0.8, SD = 1.1). Figure 2 demonstrates the overall OC–LK score of smell-impaired patients compared to normosmic controls. Similarly, a significant difference was shown between normosmic and smell-impaired patients for the left c-LK score with higher scores seen in patients (left: t [89.9] = − 2.15, p = 0.034, group 0: M = 0.6, SD = 1.1, group 1: M = 1.0, SD = 1.6). A weak but significant negative correlation was shown between the OC–LK scores on both sides and the discrimination score (OC–LK left: r = − 0.15, p = 0.01, OC–LK right: r = − 0.2, p = 0.001), and also between the right OC–LK and the total TDI score (r = − 0.13, p = 0.034). Similar negative correlations could be demonstrated between the c-LK score on both sides and the discrimination score (c-LK left: r = − 0.16, p = 0.07, c-LK right: r = − 0.15, p = 0.010) but not with the total TDI score (ps > 0.05). Similar to the correlations of the c-LK score with the SNOT 20 score (left: r = 0.21, p = 0.001; right: r = 0.18, p = 0.002) a weak but significant positive correlation was found with the OC–LK score of both sides (left: r = 0.14, p = 0.024; right: r = 0.16, p = 0.006). However, significant negative correlations with a moderate effect emerged between the OC–LK scores of both sides and the nasal airflow rating (left: r = − 0.28, p < 0.001; right: r = − 0.33, p < 0.001). Similarly, OC–LK scores of both sides demonstrated a weak but significant correlation with the subjective rating of olfaction (left: r = − 0.15, p = 0.02; right: r = 0.14, p = 0.028). The self ratings of olfaction showed significant correlations with the TDI (r = 0.61, p < 0.001), the threshold (r = 0.54, p < 0.001), discrimination (r = 0.46, p < 0.001) and identification (r = 0.55, p < 0.001).

Table 5 Classical Lund–Kennedy scores (c-LK score left/right) and modified LK score of the olfactory cleft (OC–LK score left/right) by nostril and group
Full size table
Fig. 2
figure 2

Olfactory cleft scores of smell impaired patients compared to normosmic controls

Full size image

Assessing the results by group using the analysis of variance (ANOVA), significant differences between the groups emerged for the c-LK scores (left: F(4, 281) = 49.52, p < 0.001; right: F(4, 283) = 50.64, p < 0.001) and the OC–LK scores (left: F(4, 279) = 44.67, p < 0.001; right: F(4, 279) = 46.76, p < 0.001). The post hoc analysis revealed significant differences between group 2 (sino-nasal) and all other groups in the c-LK scores and OC–LK scores of both sides (p < 0.001). When analyzing correlations separately by group, the following correlations emerged in group 2: negative correlations could be demonstrated between the right OC–LK score and the discrimination score (r = − 0.50, p = 0.019), the identification score (r = − 0.47, p = 0.027) and the TDI score (r = − 0.50, p = 0.019). OC–LK score of the left side showed significant negative correlations with the identification score (r = − 0.47, p = 0.028) and the TDI score (r = − 0.43, p = 0.047) and a tendency with the discrimination score (r = − 0.40, p = 0.068). Figure 3 shows the negative correlations between the TDI score and the left/right OC–LK scores. Even though a significant positive correlation with a strong effect could be demonstrated between the OC–LK score and c-LK score (right: r = 0.52, p = 0.013; left: r = 0.51, p = 0.016), correlations with quantitative smell function (Sniffin’ Sticks) could only be shown with the OC–LK score as reported above. A moderate but negative correlation appeared between the right OC–LK score and the nasal airflow rating (r = − 0.46, p = 0.030) and the left OC–LK score with the olfactory function rating (r = − 0.44, p = 0.042).

Fig. 3
figure 3

Correlations between the TDI score and the OC–LK score of both sides

Full size image

Discussion

The current study provides the following major results: (1) significantly higher OC–LK scores on both sides in smell-impaired patients as compared to normosmic controls. (2) A significant negative correlation between the OC–LK scores on both sides and the discrimination score and (3) between the right OC–LK score and the TDI. Similar results could be shown for the c-LK score with (4) significantly higher c-LK scores in olfactory loss patients and (5) a significant negative correlation with the discrimination score. Analyzing smell-impaired patients by group, (6) significant differences of the OC–LK and the c-LK score could be demonstrated between the sino-nasal and all other groups. (7) Significant negative correlations with strong effects were shown in the sino-nasal group between the right OC–LK score and the discrimination, identification and the TDI score and also between the left OC–LK score and the identification and TDI score. However, (8) no such correlation of the c-LK score with the smell function was shown by groups in our study.

Chang and colleagues already demonstrated the importance of an olfactory cleft-specific grading system by showing that olfactory cleft CT scores (Lund–Mackay score of the OC) correlated better with the olfactory function than the sinus specific scoring [9]. Likewise, Saito et al. showed strong correlations of the olfactory cleft CT score of CRS patients and the T&T recognition threshold [24]. Soler et al. who used a quantitative volumetric method to determine OC opacification, similarly illustrated correlations between OC opacification and olfactory function in CRS patients with nasal polyps [12].

The role of an endoscopic scoring system of the nasal cavity in evaluating olfactory function was shown in a few studies [4, 25, 26]. But as with the sinus-specific CT scoring no consistent correlations of the classical Lund–Kennedy score with the olfactory function was described. A modified LK scoring (eliminating scarring and crusting score) was shown to correlate with patient-reported outcome measure in CRS patients, however, olfactory function was not investigated [17]. Our data did not show any correlations of the c-LK with the olfactory function when analyzed by group. Hence, we introduced a modified Lund–Kennedy scoring system to evaluate the morphological changes at the OC in smell-impaired patients and to assess any predictability of the olfactory function from this grading system. Studies showing an improvement of olfaction in patients with nasal polyps after functional endoscopic sinus surgery (FESS) [3, 27,28,29]—and especially in those with olfactory cleft surgery [30, 31]—but also a study from Vandenhende-Szymanski et al. demonstrating the predictability of postoperative smell function from preoperative OC opacification (CT score) [32] further demonstrated the importance of evaluating the olfactory cleft separately. In line with this demand, we could show a significant negative correlation with strong effect between OC–LK score and the olfactory function in the sino-nasal group, indicting the ‘worse’ the mucosal changes in the olfactory cleft the lower the smell performance. This finding highlights the association of a morphological correlate (i.e., structural changes of OC) with the smell function indicating smell impairment in sino-nasal olfactory loss to originate from OC mucosa changes. The validity of this correlation is further underlined by the fact that in no other group—such as the post traumatic, postinfectious or idiopathic smell loss group—correlations between OC–LK score and smell function emerged, as in these groups smell loss is known to originate from other than OC changes.

Interestingly, besides correlations with olfactory function, a moderate but significant negative correlation appeared between the right OC–LK score and the subjective nasal airflow rating. As olfactory dysfunction and nasal obstruction [33] pose the most commonly reported and apparently most bothersome symptom in CRS patients leading to surgical interventions such as FESS, OC–LK scores could further be helpful in preoperative evaluation of CRS patients refractory to medical treatment.

Furthermore, the addition of an innocuous and easily available morphological criterion (nasal endoscopy in an outpatients setting, no radiation) to the psychophysical test results meets a desire in olfactory diagnostics. The classical Lund–Kennedy Score, meaning the overall mucosal condition of the nasal cavity so far did not allow us to reliably predict olfactory function. However, the degree of disease of the OC, reflected in a reliable OC endoscopic scoring system, may reasonably complement the functional tests in olfactory diagnosis of patients with sino-nasal smell disorder. Whether the modified endoscopic scoring system (OC–LK score) could be used as a predictor of olfactory improvement in patients receiving FESS remains an interesting question and should be subject to further investigations.

As mentioned above, we demonstrated correlations of the OC–LK score with odor discrimination, identification and the TDI score but not with the threshold score as one might have expected indirect evidence for threshold scores to represent peripheral olfaction (OSN) [34]. As the threshold test is a subthreshold test and scores especially in smell-impaired patients are rather low (and vary less) as compared to the other two subtests, correlations with the rather gross Lund–Kennedy scoring system are presumably less frequently found.

Conclusion

Quantitative evaluation of olfactory cleft opacification using a modified Lund–Kennedy score correlates with the olfactory function in patients with sino-nasal smell disorder. This finding could reflect the underlying pathophysiological mechanism of smell loss in sino-nasal disease. Therefore, the endoscopic scoring system of the olfactory cleft may complement olfactory diagnostics in patients with sino-nasal smell disorder.