Comparison of Patient Characteristics and Olfactory Sensitivity for Trigger Odorants in Parosmia and Phantosmia

This study aimed to determine the characteristics of patients with qualitative olfactory dysfunction (qualOD) and whether individuals with parosmia exhibit increased olfactory sensitivity to previously reported odorous triggers of parosmia.


INTRODUCTION
][9][10] Parosmia is the "distorted perception of an odor in the presence of an odor stimulus," whereas phantosmia (also: olfactory hallucination) is the perception of an odor in the absence of an odor source. 11Despite its generally unpleasant nature, some studies have shown parosmia as an indicator of recovery of OD particularly in cases of post-viral olfactory dysfunction (PVOD), 5,[12][13][14][15] whereas another showed a lower rate of recovery among those with parosmia. 167][18][19][20][21][22][23][24][25][26] Despite the surge in cases, there remains limited understanding of the pathophysiology of qualOD, 8 but it is suspected to arise from a complex interaction between peripheral and central mechanisms. 8,9,20arosmia typically occurs after quantitative OD, among younger women 20 after post-infectious olfactory loss 27 or head trauma, 27 whereas phantosmia may occur in idiopathic OD, with tumors, or neuropsychiatric conditions. 28Parosmia may be "triggered" by specific odorants, 9 and these may affect one's eating habits, nutrition, weight, mental health 18 (also associated with depression 29,30 ), and overall quality of life. 25Previous publications on parosmia and phantosmia have relied greatly on clinical questionnaires, with a limited number of studies attempting to characterize these conditions using psychophysical tests.Landis et al. 31 proposed a 4-item questionnaire to evaluate parosmia; in addition, Hummel et al. 32 proposed a 3-degree scoring system that includes frequency of occurrence, intensity, and social effects.More recently, Liu et al. 33 demonstrated the potential of odor valence (hedonic range and direction) as a means to assess qualOD.However, the test poorly distinguished patients with and without parosmia among PVOD patients. 34At present, there is still no widely used, validated, psychophysical or objective measure for qualOD.
The aims of the present study include: to describe the characteristics of patients with qualOD (including clinical-demographic data, psychophysical olfactory tests, questionnaires related to depression [General Depression Scale -Long Form (in German), ADS-L] and significance of olfaction [Importance of Olfaction Questionnaire, IOQ]); to compare the characteristics of patients with parosmia versus phantosmia to determine which factors, if any, can distinguish one from the other; and to investigate whether parosmia is associated with increased olfactory sensitivity for parosmia "trigger" molecules, specifically: 2-Furfurylthiol (FFT) and 2,6-nonadienal (Nonadienal).

MATERIALS AND METHODS
This study was conducted at the Smell and Taste Clinic, University Hospital Dresden.Data were retrospectively and prospectively collected within the period of May 2020 to May 2021, including patients that consulted from 2018 to 2021.The design was approved by the local ethics committee (EK-78022020).

Participants
Inclusion criteria were adult patients (≥18 years) with olfactory dysfunction who consulted at the Smell and Taste Clinic.Exclusion criteria were patients with both parosmia and phantosmia, as it was difficult to isolate the contributions of each type of qualOD.Those with incomplete data were excluded in specific analyses.
All participants underwent systematic medical history taking (age, gender, body mass index [BMI], smoking history, alcohol consumption, OD cause [PVOD, post-traumatic-PTOD, idiopathic, and others], and OD duration [months]).A subset of parosmia patients (n = 167) were asked whether they considered coffee or cucumber as parosmia triggers.All participants underwent an ENT examination including nasal endoscopy.
Based on the responses of participants to the qualOD interview and the results of the "Sniffin' Sticks Test" (see below), they were divided into three groups: only quantitative OD (composite [Threshold, Differentiation, Identification] TDI score of <30.75 and no report of parosmia/phantosmia), parosmia, or phantosmia (with the latter two classified based on their interview responses only).
Quantitative olfactory testing was done using the "Sniffin' Sticks" (Burghart Messtechnik, Holm, Germany). 35,36Assessment of qualOD was done by interview (AH, TH) where patients were asked multiple times about the presence ("Do odors appear distorted from how you remember them to be?") and the nature (frequency of occurrence, intensity, and specific effects [i.e., social consequences, significant weight loss, significant change of habits; questions were particularly focused on weight changes)] of their distorted perceptions. 32A point each was given for distortions that occurred daily, very intensely, and had significant consequences, for example, resulting in weight changes, with a maximum score of 3 (also qualOD score). 32Other information related to qualOD (duration, intensity, pleasantness, timing of onset, percentage of distorted odors) were also determined.Patients also answered the ADS-L 37 and a modified 21-item version of the IOQ. 38Detailed descriptions of the methodology are included in the section on Supporting Information (Materials and Methods).

2-Furfurylthiol and Nonadienal Threshold Test
FFT and Nonadienal (products W249300, W337706, respectively; Sigma-Aldrich, Darmstadt, Germany) were chosen for threshold testing because coffee 7,9 and cucumber 7 were among the most frequently reported odor triggers for parosmia. 7,39FFT is found in many coffee compounds providing "roast, coffee-like aroma notes" 40 and Nonadienal is the main volatile molecule found in cucumbers. 41Increased olfactory sensitivity to any of these two molecules was hypothesized to be a possible additional measure to distinguish individuals with qualOD from purely quantitative OD.
For each odorant, eight concentrations (1:10 1 to 1:10 8 ) were prepared in brown glass jars (50 mL volume, 6 cm opening diameter) with 1,2-Propanediol used as solvent (product 134368; Sigma-Aldrich, Darmstadt, Germany).Trigger odor threshold testing was done similar to phenyl ethyl alcohol (PEA) odor threshold testing in the "Sniffin' Sticks" test. 35Three jars were presented to each participant in a randomized order, with two containing the solvent and one containing the odorant.Participants were asked to identify which one contained the odorant.A higher concentration was presented after one incorrect response out of two possible presentations, whereas a lower concentration was presented after two consecutive correct responses, each being regarded as a "reversal."The mean of the last four reversals was the score, with a lower score corresponding to a higher threshold and worse olfactory sensitivity.

Statistical Analyses
Statistical analyses were performed using SPSS Version 28.0 (IBM SPSS Statistics for Windows 28.0; IBM, Armonk, NY, USA).Chi-squared test for association (and Fisher's exact test), t-test, Mann-Whitney U test, one-way and repeated measures analysis of variance (ANOVA) with Bonferroni adjusted post-hoc tests, Pearson's r and Spearman's rho correlations, and binomial logistic regression were done.Power analysis was performed using G*Power 42 (Version 3.1.9.7, https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-und-arbeitspsychologie/ gpower; Germany).A p-value <0.05 was considered significant for variables with n ≥ 100 whereas p < 0.01 considered significant for variables with n < 100.Further details on statistical analyses are found in the section on Supporting Information.

Parosmia or Phantosmia in Contrast to Only Quantitative OD
Compared to patients with quantitative OD, parosmia patients were younger (F 2,326 = 8.33, p < 0.001).Phantosmia patients had lower identification (F 2,318 = 9.73, p < 0.001) and TDI (F 2,314 = 12.49, p < 0.001) scores than those with quantitative OD.ADS-L (F 2,99 = 3.30, p = 0.04) and IOQ (F 2,93 = 3.69, p = 0.03) scores had significant effects for group on one-way ANOVA, but only tendencies for higher ADS-L scores (p = 0.055) and higher IOQ scores (p = 0.053) in parosmia compared to quantitative OD were observed in the post-hoc analysis with Bonferroni correction.BMI, smoking, and alcohol consumption had no significant effects among the groups.
Logistic regression was done to determine the effects of factors on the likelihood of reporting the presence of parosmia versus phantosmia.Excluded factors are listed in Table II and the remaining factors were entered into the model as listed in Table II.Cause (PVOD, idiopathic, PTOD, and others) and qualitative OD timing were recoded as dichotomous variables.The model explained 51.8% of the variance and correctly classified 82.6% of the cases.In the final model, increasing OD duration was associated with an increased likelihood of having phantosmia ( p = 0.03), whereas higher TDI scores ( p < 0.001) and higher qualOD scores (p < 0.001) were associated with an increased likelihood of having parosmia.It also showed that PVOD increases the likelihood of parosmia by 3.5 times ( p = 0.04) (Fig. 3, Table III).

Parosmia Trigger Molecule Thresholds
Due to technical issues related to the later investigation of trigger odors (based on Parker et al. 7 ), the testing for trigger thresholds was explored only in a subset of the sample (quantitative OD: n = 44, parosmia: n = 44, phantosmia: n = 8).Given that the scale for the threshold scores for the three orders varied, scores were standardized based on the formula: A power analysis was done after data collection for repeated measures ANOVA factoring in effects of standardized odor threshold scores (PEA, FFT, and Nonadienal) between groups (parosmia, phantosmia, and qualitative OD only).Given an effect size of 1.34 (from partial η 2 = 0.643), a sample size of 96, and α = 0.05, the power achieved for the main effect of the type of odor threshold test was >0.99.However, for the interaction effect between the type of odor threshold test and group, the power achieved was 0.53 (partial η 2 = 0.012, effect size = 0.11).
There were no significant differences between FFT or Nonadienal threshold scores among those with parosmia, phantosmia, or quantitative OD.However, standardized threshold scores for FFT and Nonadienal were higher than PEA (F 2,186 = 167.18,p < 0.001) for all groups ( p < 0.001, Fig. 4).Higher standardized threshold scores corresponded to lower concentrations.It was determined that the molar concentration for the mean thresholds for FFT and Nonadienal was less than that of PEA, revealing that there is increased olfactory sensitivity to the trigger odors compared to PEA (Table IV).
FFT threshold scores were positively correlated with PEA threshold scores (r 96 = 0.21, p = 0.04).However, on subgroup analysis, this correlation was only observed in patients with parosmia (r 44 = 0.35, p = 0.02), but not in phantosmia or quantitative OD.Similarly, TDI scores were positively correlated with FFT (r 96 = 0.32, p = 0.001), and Nonadienal (r 96 = 0.27, p = 0.008) threshold scores.Subgroup analysis revealed that TDI scores were positively correlated with FFT (r 44 = 0.50, p < 0.001) and trended toward positive correlation with Nonadienal (r 44 = 0.34, p = 0.03) thresholds only in those with parosmia.Those with parosmia who reported having coffee as a trigger (n = 80, 47.9%) had a tendency for increased mean FFT threshold scores (t 42 = 1.95, p = 0.058), although the sample was comparatively lower (n = 44) compared to the total number of participants in the study.An exploratory t-test was done taking 30% of parosmia patients with coffee as a trigger who scored the highest on FFT thresholds, as well as the 30% of parosmia patients without   coffee as a trigger who scored lowest on FFT thresholds (n = 12).Those with parosmia who reported coffee as a trigger (n = 6) had higher FFT (t 5 = 7.23, p < 0.001) threshold scores and had a tendency for higher Nonadienal (t 10 = 2.39, p = 0.04) and PEA threshold scores (t 10 = 2.17, p = 0.055) compared to those who did not, but further investigation on this is required.Only seven (4.2%) individuals reported cucumber as a parosmia trigger.

DISCUSSION
This study highlights that there are differences in the characteristics of patients reporting parosmia and phantosmia and that there were no significant differences between FFT or Nonadienal threshold scores among those with parosmia, phantosmia, or quantitative OD; but all patients had significantly increased olfactory sensitivity for FFT and Nonadienal compared to PEA.
Previous studies have found that parosmia was more likely to be reported in young 20,25 women 20,43 with PVOD, 12,13,20,44 whereas phantosmia was more likely in middle-aged individuals after head trauma 20 or idiopathic OD. 12 Women and younger individuals typically exhibit better olfactory function 35 and parosmia may more likely occur in an olfactory system that is not too severely damaged, from the OD cause or prior insults.A study by Yamagishi et al. 45 stated that parosmia or phantosmia was observed in patients with mild to moderate olfactory mucosal remodeling and not in those with completely metaplastic mucosa.Although the present findings support this, parosmia seems to reflect better olfactory function compared to both phantosmia and quantitative OD in this sample.Phantosmia was reported more in men and this may be secondary to the frequency of men having the underlying etiologies associated with phantosmia (i.e., PTOD).There is evidence supporting parosmia as an indicator of olfactory recovery 12,13,46 and regenerative processes. 13However, there are also studies suggesting that parosmia does not indicate such an effect. 13,16In line with previous findings, parosmia 12,15,43 and its association with a significantly higher residual olfactory function may relate to having better clinical outcomes 15 compared to phantosmia.
In the present study, both parosmia and phantosmia were most frequently observed after an initial quantitative OD.Having parosmia at the initial visit was associated with clinically significant recovery in odor discrimination and identification in post-infectious OD patients (which includes PVOD) receiving olfactory   Molar concentration was determined by unit conversion, where 1 mL = 1 cm 3 with the formula: %v=v 100 mL Â Density g ð Þ training 12 and was less likely to result in poor olfactory function as a final outcome. 15Furthermore, parosmia is most frequently observed in PVOD.8][49][50] Qualitative olfactory loss, especially phantosmia, with a later onset ($>12 months) after the initial OD may suggest poorer prognosis.
It was hypothesized that a greater olfactory sensitivity to reported trigger odors may distinguish parosmia from phantosmia but this was not observed in this study.Instead, all groups had increased olfactory sensitivity for trigger odorants compared to PEA.This may indicate preservation or rapid recovery of perception for odorants that signal danger/threats to safety compared to other odors that do not have such characteristics.FFT, although related to coffee, is associated with roasted or burned aromas 51 and is also known to be detectable at very low thresholds 52,53 ; whereas Nonadienal has been associated with algal contamination of drinking water. 54erhaps, the salience of specific trigger odors in parosmia is related to the combination of odorants that make up an odor (the more odorants, the increased likelihood of developing parosmia).Furthermore, the preservation of perception for unpleasant odorants that make up certain odors may dominate over less perceived more pleasant components.This results in an overall unpleasant perception of an odor as experienced in parosmia.
Although parosmia is known to adversely affect mental health and quality of life, 3,9 results of this study only reveal a tendency for increased depression scores (ADS-L) and increased significance of olfaction scores in parosmia compared to quantitative OD.We hypothesized that when greater importance is placed on olfaction, this may result in increased reporting of olfactory distortions.However, further studies are required, possibly with a larger sample.Moreover, parosmia results in more bothersome effects as evidenced by higher qualOD scores and distortions that occur more frequently compared to phantosmia.Increased bothersome effects in parosmia may again be explained by the finding that these individuals have better olfactory function than in phantosmia and quantitative OD and are able to perceive odors more, even including distorted ones.
QualOD was assessed in this study through a series of questions involving experiences of distortion in how known odors are perceived, along with frequency, intensity, and significant consequences due to these distortions. 32In the present study, the focus was on verifying the nature of participants' odor distortions through an interview, giving a more certain appraisal of qualOD.Although previous studies 12,16,20 often used questionnaires or surveys for the determination of qualOD, these methods often relied on single yes or no questions for diagnosis and may be open to misinterpretation when questions are unclear (e.g., "Did your sense of smell change?"[can mean both quantitative and qualitative], "Did you notice that the odors of certain things were different?"[also ambiguous], etc.).Reliable determination of the presence of qualOD depends on asking clear questions and ensuring that the respondents understood the qualitative changes in the sense of smell that were asked.
Although our study included patients who underwent psychophysical olfactory testing and clinical assessment of qualOD, a major challenge to the generalizability of these results is due to the sampling of participants.Because mostly patients who consulted in a tertiary smell and taste clinic were included, it was expected that the distribution of patients in this study would be similar to the frequency of presentations from various OD etiologies and some groups ended up having uneven distribution (OD causes, main group comparisons [parosmia/ phantosmia/quantitative OD]).There were also a smaller proportion of patients (n = 96) who completed all three odor threshold tests (PEA, FFT, and Nonadienal).Future studies should explore thresholds of various odors (trigger vs. non-trigger) in patients with parosmia, as well as investigate the relationship of variables with tendencies for significant differences (ADS-L, IOQ) in larger samples with more balanced groups.

CONCLUSION
Parosmia and phantosmia are reported in patient populations with distinct characteristics and can provide clinicians with a better understanding of potential olfactory outcomes in patients with qualitative olfactory loss.There were no significant differences between FFT and Nonadienal threshold scores among patient groups.However, all patient groups had significantly increased olfactory sensitivity for FFT and Nonadienal compared to PEA which may be related to the preservation of sensation for odors that signal threats/danger in patients with olfactory loss.This finding prompts a new hypothesis for the pathophysiology of qualOD and raises more interesting points and questions about parosmia triggers, odors in the context of warning for danger, and the pathophysiology of qualOD that should be explored in future studies.

Fig. 1 .
Fig. 1.Gender differences in qualitative olfactory dysfunction.[Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

Fig. 2 .
Fig. 2. Frequency of qualitative olfactory dysfunction score.[Color figure can be viewed in the online issue, which is available at www. laryngoscope.com.]

Fig. 3 .
Fig. 3. Characteristics of qualitative olfactory dysfunction patients.OD = olfactory dysfunction; PVOD = post-viral olfactory dysfunction; PTOD = post-traumatic olfactory dysfunction; TDI = "Sniffin' Sticks" composite threshold, discrimination, and identification score; VAS = visual analogue scale rating.Significant predictors in the final model include: OD duration-increased OD duration is associated with increased likelihood for phantosmia; PVOD-increases the likelihood of parosmia by 3.5 times; TDI-higher composite TDI is associated with increased likelihood of parosmia; Qualitative OD score-higher qualitative OD score is associated with increased likelihood of parosmia.Figure created with Biorender.com.[Color figure can be viewed in the online issue, which is available at www.laryngoscope.com.]

R 2 =
proportion of the variance accounted for by the model; B = unstandardized regression coefficient; SE = standard error of the unstandardized regression coefficient; Wald = values indicate which variables are significant from the set of predictors; df = degrees of freedom; Sig = statistical significance; Exp(B) = indication of change in the predicted odds for each unit change in the predictor variable; OD = olfactory dysfunction; PVOD = post-viral olfactory dysfunction; PTOD = post-traumatic olfactory dysfunction; TDI = composite "Sniffin' Sticks" odor threshold, discrimination, identification score.

TABLE I .
Summary of Clinical and Demographic Patient Data.

TABLE III .
Summary of Logistic Regression Results.

TABLE IV .
Molar Concentrations Corresponding to Mean Odor Thresholds for Phenyl Ethyl Alcohol, 2-Furfurylthiol, and Nonadienal.