The study sample consisted of 117 BD patients who were hospitalized between April 2019 and June 2019, as well as 47 healthy volunteers as controls.

The inclusion criteria for BD patients were as follows: (1) diagnosis of BD according to the diagnostic criteria of the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5); (2) age 18–60 years; (3) Han nationality; (4) primary school education or above; and (5) the patient volunteered to participate in this study and signed the informed consent form.

The inclusion criteria for BD patients in the euthymic state were as follows. In addition to meeting the above inclusion criteria for BD patients, the subjects also met the following requirements simultaneously: (1) stable current condition, without obvious clinical symptoms; (2) Hamilton Depression Rating Scale (HAMD) score ≤7; and (3) Young Mania Rating Scale (YMRS) score ≤7.

The inclusion criteria for the healthy control group were as follows: (1) healthy adults without a history of mental illness; (2) HAMD score ≤7; (3) YMRS score ≤7; (4) age 18–60 years; (5) Han nationality; (6) primary school education or above; and (7) the individuals volunteered to participate in this study and signed the informed consent form.

The study exclusion criteria were as follows: (1) any physical diseases that may affect olfactory function, such as nasal polyps, chronic sinusitis, and other nasal or paranasal sinus diseases or surgery, acute upper respiratory tract infection in the past 2 weeks; (2) neuropsychiatric diseases that may affect olfactory function, such as Alzheimer’s disease, Parkinson’s disease, epilepsy, multiple sclerosis, and stroke; (3) moderate or severe cognitive impairment as revealed by Mini-Mental State Examination (MMSE) scores <20 (primary school or below) or <24 (junior high school and above); (4) a history of alcohol or drug abuse or dependence in the past year; and (5) inability to complete the olfactory function test.

All patients or their legal representatives and members of the healthy control group signed informed consent forms before beginning the study. The study followed the principle of voluntary participation and participants were able to withdraw from the study at any time. The study was approved by the Ethics Committee of Shunde Wu Zhongpei Hospital, Foshan City before the experiments began.

The SPSS statistics software (Version 25, IBM Corp) was used for statistical analysis. The measurement data were first tested for normality and homogeneity of variance, and those with a normal distribution were expressed as the mean ± standard deviation (SD). Data with a non-normal distribution were expressed as a median (lower quartile, upper quartile) [M (Q_L, Q_U)]. An independent samples t-test or non-parametric test was used to compare two means. A one-way analysis of variance (ANOVA) was used to compare normally distributed data between multiple groups, and pairwise comparisons were further performed if statistical differences were detected. The Least-Significant Difference (LSD) method was used if the variance between groups was equal, and Tamhane’s T2 method was used if the variance was not equal. A non-parametric test (Kruskal–Wallis H-test) was used to compare the non-normally distributed data between multiple groups. If statistical differences were found, the Mann–Whitney U-test was further performed for pairwise comparisons, and the Bonferroni’s method was used to correct the test level. Categorical variables are given as percentages. The chi-square test was used for comparisons between categorical variables. Correlations were analyzed using Pearson correlation analysis (normally distributed data) or Spearman correlation analysis (non-normally distributed data). The significance level was set at α = 0.05 (two-tailed).

We enrolled 117 patients with BD and 47 healthy volunteers. The groups were not significantly different in terms of age, gender, education level, and smoking status (P > 0.05). Among the 117 patients, 110 (94.0%) used atypical antipsychotics, 115 (98.3%) used mood stabilizers, 13 (11.1%) used antidepressants, and 72 (61.5%) used benzodiazepines. The participants with BD had decreased OS (P = 0.004) and OI (P = 0.005) compared with controls, indicating that they had defects in olfactory function (see Table 1).

We divided the 117 BD patients into a BD I group (n = 86) and a BD II group (n = 31). We found no significant differences in age, gender, education level, and smoking status among the BD I, BD II groups and controls, and no significant differences in the course of disease between the BD I and BD II groups (P > 0.05). There were more patients with psychotic symptoms in the BD I group and more patients using antidepressants in the BD II group (see Table 2).

We found significant differences in OS (χ² = 15.920, P < 0.001) and OI (χ² = 12.185, P = 0.002) among the BD I, BD II groups and controls. We used the Mann–Whitney U-test for pairwise comparison of the OS and OI scores, and the Bonferroni’s method to correct the test level. The adjusted α was 0.017. We found statistically significant differences in OS (P < 0.001) and OI (P = 0.001) between the BD I group and controls, and in OS (P = 0.008) between the BD I and BD II groups (see Supplementary material 1 and Figure 1). These results indicate that impairments in olfactory function may differ according to different subtypes of BD.

We divided the 117 BD patients into Depressive BD (D-BD, n = 36), Manic BD (M-BD, n = 44), and Euthymic BD (E-BD, n = 37) groups according to episode type. We found no significant differences in age, gender, education level, and smoking status among the D-BD, M-BD, E-BD groups and controls, and no significant differences in the course of disease among the D-BD, M-BD, and E-BD groups (P > 0.05) (see Table 3).

Because the OS of the different subgroups was normally distributed and the OI was non-normally distributed, we used different statistical methods to process these datasets. We found significant differences in OS (F = 6.695, P < 0.001) and OI (F = 7.927, P = 0.048) among the D-BD, M-BD, E-BD groups and controls, respectively. Further LSD tests with multiple comparisons showed statistically significant differences in OS between the D-BD, M-BD groups and controls (P = 0.036, P < 0.001) and between the M-BD and E-BD groups (P = 0.001). We used the Mann–Whitney U-test to conduct pairwise comparisons of OI scores, and the Bonferroni’s method to correct the test level. The adjusted α was 0.0083. We found no statistically significant differences in OI scores among the groups. Nonetheless, our results showed that defects in olfactory function differ among BD patients according to episode type (see Supplementary material 2 and Figure 2).

We used Pearson correlation analysis or Spearman’s correlation to examine the relationship between OS or OI and clinical characteristics in the BD I and BD II groups, respectively. The results revealed that OI was positively correlated with the YMRS score in patients with BD I (r = 0.235, P = 0.030) while OS was negatively correlated with the HAMD score in patients with BD II (r = −0.428, P = 0.016) (see Table 4). Thus, greater mania severity in BD I patients was associated with better olfactory identification, while more severe depressive symptoms in BD II patients were associated with lower olfactory sensitivity.

We similarly analyzed the correlations between OS or OI and clinical characteristics in all BD patients according to episode type. However, we found no significant associations between OS or OI and any of the symptom or social function scales.

Our data indicated that all 117 patients with BD had decreased OS and OI performance compared with the healthy controls. Thus, BD patients may have impaired olfactory function that manifests as a decline in OS and OI ability. Accordingly, olfactory dysfunction may be a specific biological biomarker that could facilitate the early diagnosis or treatment of BD. Although some previous studies supported our results (21, 22, 27), this was not the case for others (23–25, 28–30). This could be explained by differences in sample size and the types of episodes experienced by the enrolled BD patients. For instance, the sample size of most previous studies was relatively small. Regarding the composition of BD patients in different studies, some only had depressive episodes (23), some were only in the euthymic state (29), and some were in a stable state (24). However, some studies mixed patients with BD and depression (30).

Chen et al. (16) found OI and OS deficits in patients with generalized anxiety disorder, compared with healthy controls. And a few studies in other anxiety disorders such as post-traumatic stress disorder (17, 18) and obsessive-compulsive disorder (19, 20) reported the deficits of olfactory function. Given the high prevalence of anxiety disorders (up to 50%) among BD patients (40), this comorbidity might aggravate the impairment of olfactory function. However, previous olfaction studies in panic disorder have yielded conflicting results, which showed higher OS of these patients as well as a greater olfactory awareness (41) or normal OI (42) compared to the healthy controls. Psychotic symptoms are similarly common in BD patients. More than half experience psychotic features in their lifetime (43). van Bergen et al. reported in 2019 that the vast majority (73.8%) of BDI patients had experienced psychotic symptoms in their lifetime (44). Psychotic symptom comorbidities may contribute to the occurrence of impaired olfactory function in BD patients (26).

Whether or not olfactory dysfunction can be used as a biological marker of BD is of great importance for the early diagnosis of BD. Unfortunately, research on this topic is sparse and inconsistent. Only a few studies have reported results suggesting that impairment of OI may be a potential trait marker for BD (21, 22, 26, 27). Cumming et al. (21) examined OI on the UPSIT accuracy between BD and schizophrenia patients and showed that OI deficit was present in both disorders compared with healthy participants, but BD patients were less severely affected than schizophrenic patients. Lahera et al. reported that OI deficits persisted in BD patients undergoing remission (22). Li et al. (27) suggested OI impairment is a trait, but not a disease-specific marker in BD, by comparing olfactory function between all BD patients in different mood episodes and controls, as well as between BD, major depressive disorder, schizophrenia and controls.

These findings strongly suggest that OI impairment may be a potential marker for BD; however, further study is needed before use in the clinical setting. Given that OI deficits are also seen in other psychiatric disorders, it has relatively low specificity as a potential diagnostic tool for BD. Further studies should be carried out to develop diagnostic tools with high sensitivity and specificity for olfactory function, improve existing olfactory indicators, and even combine them with other biological indicators. Advances stemming from these works should help improve the sensitivity and specificity of early diagnostic tools for BD, promote the standardization of treatment, and improve prognosis in patients.

Few studies have examined olfactory function in BD patients according to subtypes. The results of this study showed decreased OS and OI in BD I patients compared with controls, which was partially consistent with previous studies. In 2018, Kamath et al. examined OI and OS in 42 BD I patients, 45 BD II patients, and 71 controls. They found that the OI in BD I patients was lower than that of the controls, while the OS in 18 BD I patients and 22 BD II patients was not significantly different from that in the controls (26). Possible explanations for this inconsistency with respect to the present study are the sample size and methods used to measure olfactory function.

In this study, we found no significant differences in olfactory function between BD II group and the controls. This is similar to the results of Kamath (26), but contrary to the results reported by Lovdahl et al. (45). Lovdahl et al. used the symptom list questionnaire to assess olfactory impairment in patients with BD II. They found that 14% of patients with BD II (n = 21) had olfactory impairment, while the olfactory impairment rate of the controls (n = 21) was 0%. However, they performed semi-structured interviews, and thus used a subjective method to evaluate olfactory function, rather than an olfactory assessment of OI or OS via the UPSIT or SST.

According to the DSM-5, the key to distinguishing BD I from BD II is the presence of manic episodes in BD I and hypomanic episodes in BD II. The conditions also differ in terms of lifetime prevalence (46, 47), age of onset (48), clinical manifestations (49), comorbidity pattern (50), and the degree of impairment with respect to psychosocial function (51). A study from 2022 examined the genetic overlap and distinction between BD I and BD II via integrative post-GWAS analyses, and found genetic differences with a set of candidate genes distinguishing BD I from BD II (52). Similarly, Liu et al. used resting state functional magnetic resonance imaging (fMRI) and data preprocessing technology to uncover shared and unique neurobiological mechanisms between BD I and BD II (53). They reported that the dynamic amplitudes of low-frequency fluctuation (dALFF) values in BD II patients were significantly higher than those in BD I patients in the right superior temporal gyrus, which indicated that activity in this region could act as a potential biomarker for the differential diagnosis of BD subtypes. Kamath et al. (26) examined olfactory function (including OI, OS, odor discrimination, and odor hedonic processing performance) in patients with BD I, BD II, major depressive disorder, anxiety, and controls, and compared olfactory scores between diagnostic groups and controls. However, they did not conduct pairwise comparisons between diagnostic groups (i.e., BD I vs. BD II). To the best of our knowledge, the current study is the first to examine olfactory function via pairwise comparisons between BD I and BD II patients. In our study, the OS in BD I patients was significantly decreased compared with that in BD II patients. This result is consistent with the views of Hardy et al. (24), who believed that diminished OS could predict social impairment, and Judd et al. (51), who concluded that BD I patients had poorer psychosocial function than BD II patients.

We found that BD I patients had OS impairment compared with BD II patients and controls, suggesting that among BD patients reduced OS may be prominent only in those diagnosed with BD I. Our findings provide evidence for BD I as an independent subtype and suggest that OS can be a potential biomarker for distinguishing BD I from BD II. Considering that the guidance-based medication principles (especially the use of antidepressants) and prognosis of BD I and BD II differ, the differentiation of BD I and BD II is of great clinical significance, and may advance appropriate treatment approaches for these patients.

We found decreased OS in BD patients with manic and depressive episodes compared with controls, and decreased OS in patients with manic episodes compared with euthymic patients. These results are in accordance with Li et al. (27), who reported that only depressive and manic BD patients (but not euthymic BD patients) had poor OS compared with control subjects. However, our findings are inconsistent with the findings of Kazour et al. (28) and Swiecicki et al. (23), who found no significant differences in OS score between depressive BD patients and controls.

We found no statistically significant differences in OI scores among BD groups according to episode type, which is similar to the findings reported by Swiecicki et al. (23) and Kazour et al. (28). However, this is contrary to Li et al. (27), who demonstrated OI deficits in all BD patients (depressed, manic, and euthymic subgroups), indicating that OI may be a trait marker for BD. These inconsistencies are likely caused by the heterogeneity of the patient groups (different sociodemographic and clinical characteristics) as well as different sample sizes.

Compared with controls, we found no significant differences in OS or OI in BD patients in the euthymic stage. This result is in line with the views of Negoias et al. (29). However, Lahera et al. reported that BD patients showed a significant deficit in OI (measured via UPSIT) compared with healthy controls (22). This inconsistency with respect to the present study may be explained by the different patient clinical characteristics and the methods of measuring olfactory function. In the study of Lahera et al., the patients were older, the course of disease was longer, and the proportion of smoking was higher. These factors may have had adverse effects on olfactory function. At present, smoking is generally considered to be negatively associated with olfactory ability in a dose-related manner in smokers (54–56). In addition, Krüger et al. reported that euthymic BD patients with event-triggered episodes had increased OS compared with those without such episodes, although their study had no healthy control group and the sample size was very small (31).

Our data indicate that BD patients who experience manic or depressive episodes might have deficits in olfactory function, mainly related to OS. In contrast, the OS and OI in euthymic patients did not significantly differ from those in controls. Our data indicate that OS may be a promising biomarker of BD state that could be used to differentiate the acute phase of BD from the euthymic state. This could be helpful when monitoring the therapeutic effects of medication. But in contrast to longitudinal studies, this cross-sectional comparison of olfactory function between the acute phase (mania or depression) and euthymic state of BD has limitations and increased bias. A more robust evaluation would have been a prospective comparison of the same patients in the depressive or manic phase and after remission. Further prospective and longitudinal follow-up studies are needed to explore the dynamic changes in olfactory function according to medication use in BD patients.

In this study, we also explored how olfactory function related to emotional symptoms and social function. Our results showed that OS in BD II patients was negatively correlated with depressive symptoms. That is, more severe depressive symptoms were related to a lower OS. This is similar to the views of Li et al., who found that OS was negatively correlated with HAMD scores (27). This is also consistent with the classic symptoms of depressive syndrome (including depressed mood, loss of interest or pleasure, slow thinking, decreased activity, decreased appetite, hypoesthesia, suicidal ideation, or even suicide attempts). For instance, Parker et al. observed that almost all participants with BD reported smells as weaker (e.g., dull, degraded, less strong) when depressed (57). However, this is inconsistent with the view of Hardy et al., who reported that depressive symptoms were related to increased OS (24). The reason for this inconsistent result may be the heterogeneity of the research methods (all participants were clinically stable) or the comparatively small study sample sizes (20 patients) in the study of Hardy et al.

Our data also indicated that OI in BD I patients was positively correlated with manic symptoms, that is, the more severe the manic symptoms in BD patients, the better the OI ability. This is to some extent consistent with the classic symptoms of hypomanic or manic syndrome (including elevated mood and increased energy, attention, or goal-directed activity). Parker et al. reported that almost all participants with BD described enhanced olfaction, using terms like “sharper,” “clearer,” “stronger,” and “more intense” during hypomanic/manic states (57, 58). Indeed, the DSM-5 states that during manic episodes, some BD patients may feel a “sharper sense of smell” (59). This is inconsistent with the views of Hardy et al. (24) and Li et al. (27), who both reported that smell identification was unrelated to the clinical features of BD.

Our findings showed significant associations between olfactory function and emotional symptoms and confirmed the correlation between olfaction and emotion. The common brain regions shared by BD and olfaction might account for some of the olfactory alterations described in BD patients. Currently, there are only a few studies on the relationship between BD and olfactory impairment, and the findings have been inconsistent. Further study combined with new imaging techniques should provide new evidences for clarifying the pathophysiological mechanisms of BD.

Previous studies have shown that olfactory function in patients with BD is related to psychosocial cognition and social function. For instance, Hardy et al. found that BD patients with higher OS performance had lower levels of social fear and social avoidance, along with higher independence and better employment status (24). Furthermore, Cumming et al. found a positive correlation between OI and social competence in patients with BD (21). Although we also compared the relationship between olfactory function and social function in the present study, no statistically significant results were obtained. It may be that the GAF scale utilized to evaluate social function in this study was subjective and not sufficiently sensitive.

We did not find a correlation between OS or OI and the clinical characteristics of BD according to BD diagnosis or different episode types, respectively. That is, there were no significant associations between OS or OI and any of the symptom or social function scales. As previous studies have produced mixed results, further studies are needed.