Background
Olfactory function is an important role in health and behavior because it may be a valid marker of the integrity of the aging brain. The prevalence and severity of olfactory dysfunction increase substantially with aging. Olfactory dysfunction represents an important clinical symptom suggestive of early stage of neurodegenerative disorders, including Parkinson’s disease (PD) [
1].
A series of community-based and hospital-based studies in western population have demonstrated that in older adults, impaired olfactory function was closely associated with the decline of cognitive functions, especially Alzheimer’s disease (AD) in the preclinical stage, although this dysfunction is more likely to be due to problems of olfactory identification (OI) than detection [
2‐
6]. As a transient condition, mild cognitive impairment (MCI) is gaining more attention. It occurs along the progression from normal aging to dementia, so it comprises a broad clinical spectrum of pre-dementia stages [
7]. It was also reported that in MCI, olfactory impairment may herald progression to dementia [
8‐
10].
There was no general standard test for olfactory because of the cultural differences in diverse regions. Modern tests for OI include the Connecticut Chemosensory Clinical Research Center Test (CCCRCT), University of Pennsylvania Smell Identification Test (UPSIT), Cross-Cultural Smell Identification Test (CC-SIT), Pocket Smell Test (PST), Odorant confusion matrix, Biolfa olfactory test, Sniffin’ Sticks (SS), Smell Disketts Test, Scandinavian Odor-identification test (SOIT), and San Diego Odor Identification Test (SDOIT) in western countries [
11]. As in Chinese population, only studies in Hongkong have been conducted to investigate the applicability of the OI and threshold tests across the Chinese culture [
12,
13].
China has one of the fastest ageing in the world. According to the 2014 census, the number of people aged 60 years and older was 212 million, occupying 15.5 % of the population [
14]. So far in mainland China, no large-sampled population-based study was conducted to obtain the epidemiological data related to olfactory function in the elderly, and the previous reports were very limited from just hospital-based studies [
12,
15]. The Shanghai Aging Study was a community-based cohort study for investigating the progression of cognitive decline in Chinese elderly [
16]. With a study design, operational procedures and diagnostic criteria similar to most cohort studies in developed countries, the Shanghai Aging Study is recognized the first epidemiologic study conducted in mainland China. The Sniffin’ Sticks Smell Test -12 (SSST-12) is a rapid, portable, suited for repetitive and inexpensive screening of OI in a population-based study [
17]. By using it, we have a unique opportunity to identify the association between OI and cognitive function in community-dwelling elderly in China.
Discussion
To our knowledge, this is the first epidemiological study that reports the normative data of OI among community-dwelling elderly in China by using the SSST-12. The main finding of this community-based study was that, poor OI was an independent factor associated with the cognitive decline in elderly, after adjusting possible confounders. The advantages of this study include community-based design with large sample size, reliable diagnosis of cognitive function based on comprehensive clinical and neuropsychological exams, and detailed influence factor profiles (including APOE genotype) for each participant.
Some small-sampled case-control studies have reported the relation between poor olfaction and AD/MCI. AD patients were more impaired on OI and recognition tasks than on odor detection thresholds task. They were also more strongly impaired on higher-order olfactory tasks involving specific cognitive processes [
32]. A study in Germany found that, 7/14 AD patients and 4/8 individuals with MCI showed no olfactory event-related potentials (tested by the SS), suggesting hyposmia, while all comparison individuals had clearly discernible responses [
33]. Serby M et al. studied OI in 55 AD patients and 57 controls by using the UPSIT, and found that OI performances were nearly 40 % lower in mild AD patients than in controls [
34]. A Hong Kong study revealed that 12 probable AD cases identified significantly fewer odors (tested by Olfactory Identification Test) and had significantly higher olfactory threshold (tested by Alcohol Sniff Test) than their age- and education-matched normal controls [
12]. OI (tested by lexical-based OI, lexical-based picture identification, picture-based OI, and odor-detection threshold) was reported to be correlated with cognitive functions in individuals with preclinical AD compared with normal control subjects before they showed any changes in daily functioning [
34,
35].
Some prospective studies, mostly in western countries, demonstrated that older persons with impaired OI were more apt to experience cognitive decline than those with relatively preserved OI. The Mayo Clinic Study of Aging followed the 1430 cognitively normal participants over a mean 3.5 years, and observed an association between decreasing olfactory identification with a decreasing number of correct responses in Brief Smell Identification Test (B-SIT) score, and a higher risk of amnestic MCI (aMCI). Compared with the upper B-SIT quartile (quartile [Q] 4, best scores), hazard ratios (HRs) (95 % CI) were 1.12 (0.65–1.92) for Q3 (
P = 0.680); 1.95 (1.25–3.03) for Q2 (
P = 0.003); and 2.18 (1.36–3.51) for Q1 (
P = 0.001) (worst scores; P for trend <0.001) after adjustment for gender and education, with same age. The B-SIT score also predicted progression from aMCI to AD, with a significant dose-response with worsening B-SIT quartiles [
6]. The community-based longitudinal study of memory and aging in the Japanese-American community in King County, WA screened 1985 persons using the Cognitive Abilities Screening Instrument (CASI) and the CC-SIT at the baseline. After 2 years, the authors determined an OR for cognitive decline (defined as loss of ≥5.15 points/100 on the CASI) of 0.90 (95%CI: 0.84–0.97) for an increasing in each correct point on the CC-SIT [
5]. Another population-based study followed up 1920 older participants in the Epidemiology of Hearing Loss Study in Beaver Dam, Wisconsin. Olfactory was measured by the SDOIT, and the incident cognitive impairment was defined as a MMSE of less than 24 or reported diagnosis of dementia. There was significant association between olfactory impairment at baseline and 5-year incidence of cognitive impairment (OR = 3.72, 95 % CI: 2.31–5.99) after adjusting for possible confounders [
2]. The association of OI with decline rate in different cognitive domains was examined in 481 older participants from the Rush Memory and Aging Project. Lower OI score was related to lower function at baseline in each cognitive domain, especially in perceptual speed and episodic memory, after adjustment for age, sex, and education [
3]. One hundred seventy-three independent residents of a continuing care retirement community completed UPSIT and MMSE and the Executive Interview (EXIT25) at baseline and were examined twice over 3 years. It concluded that impaired OI non-demented individuals was related to an AD-like memory impairment and an increased rate of cognitive decline [
36]. Swan GE et al. reported that, a total of 359 individuals were administered the BSIT for OI, and verbal learning, memory, executive control and global function for cognitive function. Impaired olfactory function was associated with a greater 4.5-year decline on several indices of verbal memory after adjustment for baseline cognitive performance, by a multivariate analysis [
37]. Devanand DP et al. examined 90 outpatients with MCI and 45 controls by UPSIT for OI at baseline, and diagnosed incident AD by 2 years. In patients with high MMSE (≥27 of 30), low olfaction with lack of awareness was explored a significant predictor of AD. At follow-up, olfaction scores of 30–35 showed moderate to strong sensitivity and specificity for diagnosis of AD [
8].
APOE genotype was identified as an important factor interacting with olfaction. Murphy C et al. have indicated that APOE-ε4 allele is related to OI deficits (tested by SDOIT) in 27 non-demented older persons [
38]. Olfactory dysfunction (tested by CC-SIT) in the presence of one or more APOE-ε4 alleles was related to a 4.9 times the risk of cognitive decline in the longitudinal study of memory and aging in the Japanese-American community in King County [
5]. A study in China used CC-SIT to assess OI performance in 28 patients with MCI and the 30 controls. They found that individuals with APOE-ε4 allele were able to identify less odors than that of the subjects without APOE-ε4 allele (
P < 0.010). It suggested that the decreased OI in MCI may be an indicator for the early diagnosis of AD, and APOE genotype may be a portion of the basis of OI decline [
15]. Our study did not find the significant contribution of APOE-ε4 allele to cognitive decline through the multivariate logistic regression model. The independent association between the poor OI and cognitive decline were explored, after adjusting the possible confounders including APOE genotype (Table
3).
There were several mechanisms or explanations for the association between olfactory dysfunction and cognition impairment. OI is more strictly involved in cognitive functions since it is processed in the mesial structures of the temporal lobe. The olfactory threshold, however, is mostly influenced by peripheral deficits of the smell sense. It has been shown that typical lesions of AD (i.e., amyloid core and neurofibrillar tangles) were already detectable in the preclinical stages in central olfactory pathways [
39]. With recent reports in pathological and brain imaging analyses, it has been found that neurodegenerative changes, including the appearance of Lewy bodies, were commonly observed in regions of the brain which responsible for olfactory perception, such as the amygdala, hippocampus, and orbitofrontal cortex. These changes are apparent from the earliest stages of dementia [
40,
41]. Kjelvik G et al studied 12 patients with aMCI, 6 with early AD, and 30 controls. They reported that the aMCI/AD group with reduced OI ability (measured by B-SIT and SSIT), had significantly smaller hippocampal volume than that of the patient group with OI scores > 50 %. There was a significant association between OI scores and hippocampal volume only in the patient group. The results suggested that the decrease in the hippocampus size in connection with early AD was related to more with loss of OI ability rather than loss of memory, thus demonstrating that impaired OI is an early indicator of medial temporal lobe degeneration [
42].
Our study also explored an interesting finding that, some odors, such as coffee, fish, banana, rose, leather, and liquorice were identified more significantly different between individuals with MCI and cognitive normal, than odors of lemon and cloves. It suggests that some odors might be the sensitive indicators for the early stage of cognitive decline, while some might be not. The transformational mechanism of stimulation by different odors may need more attention.
Olfactory function includes threshold, discrimination, identification and olfactory recognition [
11]. In the current study, we only studied the OI in our study participants. It was reported that odor discrimination and identification performance correlated more significantly than detection thresholds with performance on neuropsychological tests [
9]. Thus other domains of olfaction were also important, especially discrimination, which could also be useful as the early predictor of AD. The second limitation is the lack of assessment for parkinsonism and PD, because the olfactory dysfunction occurs in early stages of clinical PD and in asymptomatic relative of PD patient with a prevalence of approximately 90 % [
43]. Other limitations include the cross-sectional study design, which cannot determine the causal association, so it is possible that the poor OI was the early symptom of MCI rather than a risk factor; and the potentially unmeasured confounding factors, which may influence the results.
Acknowledgments
The authors thank Dr. Minhua Shao for her technical assistance in the APOE genotype assays, Zhaolan Ding, Meihua Jin, Meirong Chen, Zeya Wang, Meizheng Shi, Jingping Ye, Meiping He, Lanfang Yu, Deping Chen, Fusheng Gong, Meili Shi, Wenying Zhou, Shumin Chen, Xiudi Xu, Meiling Huang, Linghua Ding, Wenfan Zhu, Zhi Zhou, Xiaoying Liu, FuqinGao, Peng Gong, Lin Lu, Meng Wang, Ting Zhang, Yaru Guo, Xiaoli Jin, Shiqi Li, Qiongyi Xu, and Yiping Wang for their efforts to the study, and all the participants for their cooperation.