Vitamin A-Containing Intranasal Drops Improve Odour Detection Threshold

Abstract

Introduction

Retinoic acid, a metabolite of vitamin A (VA), has been shown in animal work to support olfactory receptor neuron regeneration. A limited amount of clinical research has also demonstrated potential therapeutic benefit in patients with olfactory dysfunction (OD). We therefore performed, to our knowledge, the first prospective study investigating whether the intranasal VA-containing “Coldastop” is beneficial in treating patients with non-sinonasal OD.

Methods

A non-blinded prospective comparative study was performed, in patients with OD due to post-infectious, post-traumatic, and idiopathic aetiologies. Patients were treated with olfactory training (OT) alone, OT plus intranasal VA (Coldastop, OT+VA), or OT plus intranasal steroids (mometasone furoate, OT+S) for 12 weeks. Full psychophysical olfactory testing was performed before and after treatment using the “Sniffin’ Sticks” test battery (with results for odour threshold, discrimination, identification, and composite “TDI”).

Results

A total of 144 patients completed the study: 52 OT, 42 OT+S, 50 OT+VA. The majority of patients had post-infective OD. There were no significant differences between groups in age/sex. There were statistically significant improvements in each subtest (T/D/I) and composite scores (TDI) within each treatment regimen. Across groups, there was a significant interaction between treatment and time (controlling for baseline characteristics) for odour threshold, F(2, 139) = 3.2, p = 0.044, with the largest improvement being seen in the OT+VA group. Further, the greatest proportion of patients reached clinically significant improvement within the OT+VA group (T/D/I/TDI), but this did not reach statistical significance.

Conclusion

Intranasal VA may be of benefit in treating OD. Further large-scale randomised trials are required.

Introduction

Prior to the global COVID-19 pandemic, olfactory dysfunction (OD) was thought to affect approximately 22% of the general adult population [1]. As commonly recognised acute and post-infective sequelae of SARS-CoV-2 infection, this prevalence is set to increase in future years [2]. Through its negative impact on food and nutrition, social communication and environmental hazard avoidance, OD can have significant impact on quality of life, causing symptoms of depression in up to 1/3 of affected patients [3]. Consequently, this form of sensory impairment carries marked personal and societal burden of disease [4].

Common causes of OD include sinonasal inflammatory disease (“sinonasal disease” [SND]), upper respiratory tract infections (“post-infective OD” [PIOD]), head injury (“post-traumatic OD” [PTOD]), neurodegenerative diseases (including Alzheimer’s and Parkinson’s), and ageing [5]. A proportion of OD will not fall into any known categories, and after appropriate workup, can be classified as idiopathic (IOD) [6]. Treatment guidelines for SND-OD are well established – with topical +/− systemic steroids being the mainstay of treatment, but with newer immunomodulatory approaches gaining favour in selected patients with severe disease [6, 7]. Effective treatment options for non-SND-OD have traditionally been more limited, with olfactory training (OT) having the strongest evidence base. However, recent years have seen increased interest and funding, with associated early progress [6].

Several emerging therapies for non-SND-OD aim to augment or support the innate regenerative capacity of the olfactory neuroepithelium. A metabolite of vitamin A (VA), retinoic acid (RA) is a transcription regulator needed for tissue development and regeneration [8]. Correct RA signalling is needed for olfactory system embryogenesis and adult neurogenesis [9]. In a previous retrospective study, we demonstrated that intranasal application of VA, in combination with OT (which is standard of care in our unit), produced greater levels of clinical olfactory improvement compared to OT alone [10]. We aimed to expand on this work by conducting a non-blinded, prospective comparative study in patients with non-SND-OD of mixed cause (PIOD, PTOD, or IOD) receiving OT + VA, compared with two control groups, OT + intranasal steroids or OT alone.

Methods

Ethical Considerations

Ethical approval was obtained from the Hospital Ethics Committee of University Hospital Dresden “Carl Gustav Carus” [EK60032013]. The study was conducted in keeping with the Declaration of Helsinki. Full written informed consent was obtained from participants at time of recruitment. Participants were free to withdraw from the study at any time.

Study Design and Participants

A prospective non-blinded comparative study was performed. Adult patients (≥18 years) attending the multidisciplinary Smell and Taste Clinic, Department of Otolaryngology, University Hospital Carl Gustav Carus Dresden, for the assessment of OD were screened for eligibility. Following a standardised, structured clinical history, patients underwent full ENT examination including nasendoscopy, with diagnosis of underlying aetiology based on these findings. Patients with sinonasal, congenital or neurodegenerative causes of OD were excluded. Patients with PIOD, PTOD, and IOD were included. Those with comorbidities that could affect olfaction (e.g., hypothyroidism or clinical depression) were excluded. Only patients with chronic OD (≥3 months) were included. There was no upper limit for duration of OD.

In keeping with the associated literature base, OT is standard of care in our unit. Therefore, all recruited participants underwent a period of 12-week OT, using 4 standard odours (phenyl ethyl alcohol [rose], eucalyptol [eucalyptus], citronella [lemon], and eugenol [closes]), as described in detail elsewhere [11, 12]. In addition to OT, patients were pseudo-randomly allocated (sequential allocation to treatment arm 1–3, with oversight to ensure equal distribution of patient number and demographic features) to receive either no additional treatment or treatment with either intranasal VA or intranasal steroid (steroid, S). Intranasal steroids were used as an experimental arm for all included OD aetiologies, as there is evidence for the role of inflammation in both PIOD (particularly the COVID-19 literature) and PTOD [6, 13]. Though less clear as a treatment in IOD, the low side effect profile also enabled use in this cohort. The VA group received intranasal VA drops (Coldastop, Desitin Arzneimittel, Hamburg, Germany) at a dose of 10,000 IU once daily for 3.75 (SD ± 0.82) months. Participants were trained to self-administer four drops (to each nostril) in the “Kaiteki” position, which has been shown to maximise delivery of medication to the olfactory cleft [14]. The “Mygind” position has previously been shown to result in drop penetration mainly within the middle meatus [15]. However, recent work specifically investigating distribution within the olfactory cleft demonstrated non-statistically significant superiority of the Kaiteki position compared with the Mygind [16]. For this reason, as well as anecdotally superior comfort, the Kaiteki position was used, though the Mygind position could be considered (according to patient preference) for future studies. The steroid group received intranasal mometasone furoate nasal spray (applied with an elongated nebuliser, which aimed to improve delivery of the medication to the olfactory cleft) one spray per nostril twice a day for 12 weeks. Participants were instructed to keep a diary of compliance with OT/medications and of any side effects experienced.

All patients underwent psychophysical testing before and after treatment, using the “Sniffin’ Sticks” test battery (Burghart Messtechnik, Wedel, Germany). This orthonasal tool allows for separate testing of odour threshold (T) and the suprathreshold functions of odour discrimination (D) and identification (I). Individual subtest results have previously been suggested to differentially reflect peripheral or central olfactory function, and composite “TDI” score increases diagnostic sensitivity. For a full description of testing procedure, please see ref [17]. In brief – odour threshold involves a forced choice staircase procedure in which the patient is required to identify the target, odour-containing stimulus from two non-odour-containing “distractors.” Increasing concentrations of odour are presented until the patient can reliably identify target from distractors. Odour discrimination testing also involves a forced choice paradigm in which the patient is presented with three suprathreshold odours, of which two are the same in quality and one is different, and asked to identify the “odd one out.” Finally, odour identification is again a forced choice task in which the patient is asked to identify which of four verbal/visual cues correspond with the presented suprathreshold odour. In each subtest, a maximum score of 16 can be obtained, with a corresponding maximum composite TDI score of 48: higher scores denote better olfactory function. Normosmia is defined as a composite TDI of >30.5, hyposmia >16.5, ≤30.5, anosmia ≤16.5 [18]. Clinical improvement correlates with increases of ≥5.5 for TDI, ≥3 for identification/discrimination, or ≥2.5 for odour threshold.

COVID-19 status was ascertained from patient self-report; further testing was not performed as part of the study. Patients who were re-infected during the study period were excluded.

Statistical Analysis

Statistical analysis was undertaken using SPSS (version 27 for Mac, IBM Corp, Chicago, Il, USA) and GraphPad Prism (version 9, GraphPad Software, La Jolla, CA, USA). Statistical significance was assumed where p < 0.05. Welch’s one way ANOVA or Kruskal-Wallis tests were performed to investigate differences in age, duration of OD, or parosmia/phantosmia scores between groups. A repeated measures ANCOVA was performed [within subject factor – time, 2 levels; between subject factor – treatment type, 3 levels] to investigate for interactions between time and treatment type, controlling for confounders. Given the repeated measures design, age and sex were not used as nuisance covariates. Post hoc unpaired t-tests or Mann-Whitney tests were performed where appropriate. Comparison of proportions of participants with different attributes (e.g., aetiology, achieving clinically significant improvement) was performed using the χ² test. Data from participants that were lost to follow up were not analysed.

Results

One hundred ninety-nine participants were initially recruited. Fifty-five were lost to follow-up leaving 144. The mean duration between testing sessions was 3.7 months. There was no significant difference in proportion of female patients across groups, χ²(2) = 1.3, p = 0.52 or in age between groups W(2.0, 93.7) = 2.5, p = 0.09. The majority of patients in each group had PIOD. However, there were no statistically significant differences in proportions of different aetiologies of OD across the different treatment groups. The duration of OD at baseline was longer in the OT + S group than in the OT or OT+VA group (statistically significant for each comparison, see Table 1). Full demographics for each group are given in Table 1. No significant side effects were reported and were similar to those previously described [10].

Table 1.

Patient demographics

​	OT	OT+S	OT+VA	Between group comparisons
				OT vs. OT+S	OT vs. OT+VA	OT+S vs. OT+VA
N	52	42	50	​	​	​
Age (mean, range), years	47 (20–83)	48 (20–70)	43 (20–70)	F = 1.95, p = 0.15
Sex (F, M)	23, 29	28, 14	38, 12	χ2 = 1.3, p = 0.52
Aetiology, N, %	​			​
PIOD	45, 86.5	40, 95.2	43, 86	χ2 = 2.43, p = 0.3
IOD	5, 9.6	2, 4.8	4, 8	χ2 = 0.79, p = 0.67
PTOD	2, 3.9	0, –	3, 6	χ2 = 2.49, p = 0.29
Duration of OD (at first visit), months	13.3±24	17.4±51.7	11.6±18.2	H = 11.18, p = 0.004
				p = 0.003	p = 0.47	p = 0.004
Parosmia scorea at baseline (mean±SD)	0.6±1.4	0.9±1.4	0.4±1.5	H = 2.63, p = 0.27
Phantosmia scorea at baseline (mean±SD)	−0.8±0.6	−0.4±1.1	−0.9±0.5	H = 11.01, p = 0.004
​	​	​	​	p = 0.02	p = 0.65	p = 0.003
COVID-19 +ve	43	38	35	χ2 = 6.35, p = 0.042
				p = 0.28	p = 0.13	p = 0.02

Results of ANOVA, Kruskal-Wallis, or chi-square tests and associated post hoc p values shown under “between group comparisons”.

^aParosmia/phantosmia score: from −1 (no parosmia) to 3 (most significant degree of parosmia/phantosmia).

Pre- and post-treatment, as well as mean change in psychophysical test scores (Δ score = score_{post treatment} – score_{pre treatment}) are shown in Table 2. For each treatment arm, all within group comparisons (pre- vs. post-training) showed statistically significant improvements in T, D, I and TDI (all p < 0.05).

Table 2.

Pre- and post-training psychophysical test scores

​	Psychophysical test score
	threshold			discrimination			identification			TDI
	OT	OT+S	OT+VA	OT	OT+S	OT+VA	OT	OT+S	OT+VA	OT	OT+S	OT+VA
Pre-training	4.1±2.9	3.7±2.6	2.8±2.3	10.6±3.0	9.9±3.4	9.6±2.7	9.5±3.6	9.5±3.3	9.0±3.4	24.1±7.7	23.0±7.7	21.5±6.7
Post-training	5.2±2.9	4.6±2.5	4.8±3.2	11.6±3.0	11.0±2.7	11.1±3.1	10.4±2.8	10.3±3.0	9.9±3.3	27.2±7.3	25.9±7.1	25.8±7.9
Δ score	1.1±2.2 p = 0.0008	0.9±2.4 p = 0.017	2.0±2.8 p < 0.0001	1.1±3.0 p = 0.015	1.1±3.4 p = 0.043	1.4±2.5 p = 0.002	0.9±2.4 p = 0.0084	0.8±2.1 p = 0.016	0.9±2.5 p = 0.018	3.1±5.0 p < 0.0001	2.8±5.1 p = 0.0008	4.3±4.7 p < 0.0001
Clinically significant improvement, N, %	14, 27	14, 33	20, 40	18, 35	11, 26	18, 36	13, 25	7, 17	14, 28	16, 31	10, 24	21, 42

Pre- and post-training psychophysical test scores (mean ± SD), mean change in score (mean ± SD), and proportion of patients with clinically significant improvement (N, %). p values shown for paired t-tests.

The mean change in score was greatest in the OT+VA group for T, D, and TDI. However, the interaction between the factors time (pre vs. post) and treatment group was only statistically significant for T score, when controlling for duration of OD, cause of OD, and baseline parosmia score, F(2, 139) = 3.2, p = 0.044. Parosmia and duration of OD were included as they have previously been demonstrated to be predictors of recovery in olfactory function [6]. Figure 1 shows unadjusted change in T score across the two time points (A), and adjusted, estimated marginal means for T score, across the two time points, for the three treatment regimens.

Fig. 1.

a Violin plots showing unadjusted average T scores for three treatment regimens (internal lines = interquartile range, median). b Mean T score at baseline and follow-up for the three treatment arms, controlled for baseline characteristics (duration of OD, cause of OD, and baseline parosmia). *Statistically significant interaction between time and group in repeated measures ANCOVA.

The proportion of patients who achieved clinically significant improvement was greatest in the OT+VA group for T, D, I and TDI. However, these differences did not reach statistical significance.

Discussion

To our knowledge, this is the first prospective study to investigate the role of topical VA (delivered as part of the intranasal solution “Coldastop”) as an adjunct to OT, compared to OT alone or OT + intranasal steroids. We demonstrated statistically significant improvements in psychophysical test scores (individual and composite) for each treatment regimen. Compared with OT or OT + S, the OT+VA group demonstrated the greatest improvement in mean T, D and TDI score – though statistical significance was only demonstrated for T score, when controlled for baseline participant characteristics. Again, the proportion of patients who reached clinically significant improvement was greatest in the OT+VA group, though this did not reach statistical significance.

The olfactory system can be divided into two subcomponents – the peripheral olfactory apparatus, consisting of the olfactory neuroepithelium (OE) within the nose – and the central olfactory system, comprising all intracranial structures from the OB up. The specialised bipolar neurons within the OE (olfactory receptor neurons [ORN]) are in simultaneous contact with the external environment, where they make contact via their dendritic cilia with odours in the nasal mucus layer, as well as the central nervous system, where their axons synapse within the olfactory bulb. Due to this exposed position, ORN are prone to damage but, under normal conditions, are continually replaced from the neuroepithelium’s resident stem cell population, through a process of neurogenesis [5, 19, 20]. In certain conditions, such as PIOD, this regenerative potential is thought to become exhausted, resulting in patchy interspersion with non-sensory respiratory or metaplastic squamous epithelium [21, 22].

The role of RA signalling in ORN neurogenesis is supported by extensive animal work. Disruption of such signalling during the mid-gestation period has been shown to cause improper olfactory embryogenesis in mice [23]. Moreover, adult murine OE both contains RA receptors and is capable of synthesising RA [24]. In vitro, RA appears to support axonogenesis, deficiency leads to reduced levels of olfactory marker protein (a marker for mature ORN) mRNA, and RA receptor inhibition can cause ORN death [25–27]. Indeed, administration of systemic RA can enhance recovery following transection of the olfactory nerve in mice [28].

This theoretical basis prompted early work into the potential utility of VA in the treatment of human OD. Accordingly, Duncan and Briggs demonstrated improved subjective olfactory function following administration of systemic VA in their 1962 case series of 56 patients with OD of mixed cause [29]. However, this study was not systematic in nature, with varying high levels of VA given. A methodologically more robust RCT performed by Reden and colleagues failed to demonstrated any significant treatment effect, following 12 weeks of systemic use at a dose of 10,000 IU/day [30]. More recently, we performed a retrospective analysis of patients who had undergone OT + administration of intranasal VA, which theoretically allows application of greater concentrations of medication at the OE than are achieved through systemic therapy, in which maximum doses must be observed due to risk of side effects. In this study, the addition of VA to OT resulted in a significantly greater proportion of patients achieving clinical improvement than OT alone [10]. Though we did not find a statistically significant increase in the proportion of patients who achieved clinical improvement in our current study, our findings regarding odour threshold extend on these early results. Odour threshold has previously been suggested to better reflect the peripheral olfactory apparatus (i.e., the OE) [17, 31]. It is therefore possible that our current findings are in keeping with RA-mediated augmentation of ORN neurogenesis, as is seen in the animal literature. However, this remains speculative, and future studies, ideally including histological work, are required.

Study Limitations

We were only able to demonstrate statistically significant improvement in odour detection threshold, not composite TDI, nor the other subcomponents of odour discrimination or identification, in the OT+VA group, compared to the OT alone or OT+S group. In keeping with our above discussion, this may be due to potential peripheral action of VA and the better reflection of peripheral olfactory function by odour detection threshold than odour discrimination or identification [6, 17]. While improvement in overall TDI would have strengthened our conclusions, the interrogation of individual “Sniffin’ Sticks” subtests is supported both statistically and clinically through previous publication of normative values [18, 32], as well as minimum clinically important differences for T, D, and I [33]. Moreover, subcomponent analysis is frequently undertaken in the olfactory literature due to the above discussed differential reflection of peripheral vs. central olfactory function [34–37].

We did not include a no-treatment control arm. Therefore, it is possible that a proportion of patients in each treatment arm may have experienced spontaneous recovery. Future trials should aim to mitigate this risk by including a no-treatment arm. Furthermore, patients in our study were not formally randomised across treatment arms. While there were no significant differences in basic demographics (sex/age) between groups, and though we used other baseline characteristics as nuisance covariates in our statistical analysis, in order to fully control for unknown confounders, future work should employ formal randomisation.

The majority of recruited participants had PIOD, meaning that subgroup analysis of PTOD or IOD was not possible due to small resultant subgroup numbers. Subgroup analysis of PIOD patients was additionally not performed as this would reduce the overall patient number, and consequently the associated power of the study. Whilst the distribution of underlying aetiology did not significantly vary between treatment groups, use of “mixed-aetiology” is a limitation of the current study. Given that PIOD is more likely to involve pathobiological processes at the level of the peripheral olfactory system than PTOD or IOD [6, 21], it is possible that topical VA is more or only effective in the former of these conditions. Accordingly, the inclusion of PTOD and IOD may have confounded the observed results. Future studies should aim to investigate the effect of VA on single aetiology patient cohorts.

While patients were trained to deliver 4 drops per nostril during study enrolment, it is possible that slightly less or more of the medication was delivered through administration error. While this well reflects “real-world” variability, it is a potential limitation in our study design. Future studies should aim to produce dosing devices that allow application of a precise amount of VA. Furthermore, Coldastop contains several ingredients beyond VA. These include odorant molecules such as linalool, citral, citronellol, DL-limonene, and terpineol. While, to our knowledge, the effect of regular application of topical odorants on olfactory function has not yet been investigated, it is theoretically possible that this could have acted as a form of “topical OT.” However, previous work has demonstrated OT to improve odour identification function more so than odour detection threshold [38], which was not in keeping with our results. In addition, terpineol and limonene have been shown previously to have anti-inflammatory effects [39]. Together, these are major limitations of the current study. To ensure that “topical OT” or independent anti-inflammatory effects of Coldastop additives did not confound the effects of VA, further studies should be carried out in which a “pure” form of VA was used. It may also be of interest to separately investigate the utility of OT with “topical” odours, as well as the use of topical anti-inflammatory products such as terpineol, independent of VA.

Finally, information regarding comorbidities was collected as part of inclusion/exclusion screening. Patients with SND-related OD were excluded – including those with active or recent sinonasal symptoms. Accordingly, patients with active or recent allergic rhinitis or those with chronic rhinosinusitis (with or without polyps) were excluded. While patients underwent nasendoscopy to support correct diagnosis, grading of any incidental septal deviation in otherwise eligible patients was not undertaken, and this information was not available for analysis. Furthermore, smoking status was not recorded. While the impact of smoking on olfactory function is controversial (please see ref [6] for discussion), future studies should aim to record this, as well as other factors that could affect drop penetration to the olfactory cleft including septal abnormalities and previous nasal surgery.

Conclusion

We demonstrated significant improvement in psychophysical tests scores following OT, OT+S, and OT+VA. The VA-containing Coldastop appeared to confer additional benefit in odour detection threshold, when controlling for baseline participant characteristics. Further work is needed to confirm these findings using pure VA in single aetiology patient cohorts, as well as to fully delineate the long-term clinical impact of these results.

Statement of Ethics

Ethical approval was obtained from the Hospital Ethics Committee of University Hospital Dresden “Carl Gustav Carus” [EK60032013]. The study was conducted in keeping with the Declaration of Helsinki. Full written informed consent was obtained from participants at time of recruitment. Participants were free to withdraw from the study at any time.

Conflict of Interest Statement

Thomas Hummel was a member of the journal’s Editorial Board at the time of submission.

Funding Sources

Thomas Hummel was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG HU 040.1179). K.L.W., M.E.D., and A.H. were unfunded.

Author Contributions

K.L.W. – formal analysis, wiring – original draft, and review and editing. M.E.D.: methodology, investigation, and writing – review and editing. A.H.: methodology and writing – review and editing. T.H.: conceptualisation, methodology, resources, supervision, project administration, and writing – review and editing.

Funding Statement

Thomas Hummel was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG HU 040.1179). K.L.W., M.E.D., and A.H. were unfunded.

Data Availability Statement

All data generated or analysed during this study are included in this article. Further enquiries can be directed to the corresponding author.