Salivary cortisol is associated with cognitive changes in patients with fibromyalgia

Fibromyalgia (FM) is a stress-related chronic pain disorder with common cognitive complaints. This study characterized cognitive dysfunction in patients with FM and explored whether these changes are linked to altered cortisol levels. Consecutive 44 patients with FM and 48 healthy controls were enrolled for the assessments of subjective and objective cognitive functions and diurnal levels of salivary cortisol (sampled at awakening, 30 min after awakening, 3 pm, and bedtime). All measurements were compared between the groups and evaluated for clinical correlation. The FM group had more subjective cognitive complaints and performed poorer in objective cognitive testing in memory (delayed recall in Chinese Version Verbal Learning Test and Taylor Complex Figure Test), language (Boston Naming Test), and executive domains (Wisconsin Card Sorting Test) after adjustments for education. The diurnal cortisol levels of patients with FM tended to be lower, especially at 30 min after awakening and bedtime. Moreover, moderate positive correlations existed between the Chinese Version Verbal Learning Test, Boston Naming Test and the morning cortisol levels within the FM group. We suggested the altered cognitive function in FM may be linked to stress maladaptation. Future studies are warranted to elucidate whether stress management improves cognitive performance in patients with FM.


Abbreviations
www.nature.com/scientificreports/ Fibromyalgia (FM) is a chronic pain disorder characterized by widespread body pain and several associated symptoms such as fatigue, cognitive problems, unrefreshed sleep, and depression 1 . Although a complex interaction between genetic disposition and environmental factors may cause FM, its pathophysiology remains undetermined 2 . Many patients with FM identified stress or a stressful event during their lifetime as a pivotal trigger for their chronic pain. In fact, patients with FM reported a higher proportion of early life adversities such as physical and sexual abuse 3 ; moreover, post-traumatic stress syndrome (PTSD) is a well-identified comorbidity of FM 4 . The disabling pain per se and complex comorbidities in FM (psychiatric and several others including migraine, irritable bowel syndrome, bladder hyperactivity, and restless leg syndrome) tremendously affect the quality of life of patients 5 . Therefore, FM may be regarded as a stress disorder 2 , and stress may be a window into a thorough understanding of this elusive disorder. The most studied physiological system of stress is the hypothalamic-pituitary-adrenal axis and its downstream product cortisol 6 . A lower cortisol level has been reported in many stress-related disorders such as PTSD 7 and chronic fatigue syndrome (CFS). In FM, however, the basal cortisol level and their diurnal change tended to be lower but inconsistent across studies [8][9][10] , as discussed in a recent meta-analysis 11 . These inconsistent results of cortisol levels in FM might be because of different methods of measurement and the lack of a gold standard 11 . Similar to FM, an animal study 12 demonstrated chronic stress induced a hypoactive pattern of cortisol through enhanced negative feedback 13 , whereas childhood maltreatment (chronic stress) was associated with higher cortisol levels in patients with FM in the afternoon 14 . Based on the association between hypocortisolism and fatigue experience in CFS, the altered cortisol levels in FM might be considered a maladaptive response to stress 15 . However, some researchers argue that this is a protective mechanism because a low cortisol level may reduce allostatic load or enhance the body's defense system against chronic inflammation 16 .
Cognitive problems, a core symptom of FM, were reported in more than 76% of patients with FM 2 , majorly involving memory, executive, and language domains 17 . In fact, FM is the only pain syndrome that incorporates cognitive changes into its diagnostic criteria. Cognitive changes are also pivotal for the assessment of functional disability in FM, such as that using the revised Fibromyalgia Impact Questionnaire (FIQR) 18 . In a recent review, decreased mental reserve due to pain, fatigue, and depression has been proposed to cause the cognitive symptoms of FM 17 . Nevertheless, few studies have explored the role of stress in relation to cognitive changes in FM. In animal studies, chronic stress can lead to structural and functional changes in some neural correlates of pain perception or pain modulation, notably in the hippocampus, which may concomitantly affect cognitive functions 2,19 . The hippocampus is vulnerable to stress owing to the high expression of glucocorticoid receptors 20 . In human studies, hypercortisolism is hypothesized to be crucial in the pathogenesis of Alzheimer's disease 21 , whereas the connection between lower cortisol levels and cognitive dysfunction has been observed in Addison's disease 22 , PTSD 23 , and the normal population 24 . Although either a low or high cortisol level beyond the physiological range seems detrimental to cognitive function, how cortisol affects cognitive function in FM remains to be investigated.
This study hypothesized that cortisol levels are altered in patients with FM, and this alteration is related to cognitive performance in FM. Accordingly, in the present study, we used comprehensive subjective and objective cognitive assessments to determine whether patients with FM display deficits in different cognitive domains and explored whether these cognitive deficits were associated with the diurnal salivary cortisol levels.

Methods
Participants and procedure. Consecutive patients with FM aged 20-60 years were enrolled from the Neurological Institute of Taipei Veterans General Hospital in Taiwan. All patients fulfilled the modified 2010 American College of Rheumatology (ACR) Fibromyalgia Diagnostic Criteria 1 ; however, those with any autoimmune rheumatic disease were excluded. Healthy volunteers who did not have past or family histories of FM and who had not experienced any significant pain during the past year were recruited as controls. All participants denied having any history of systemic or major neuropsychiatric disease and exhibited normal results of physical and neurological examinations and brain magnetic resonance imaging (MRI). Participants who were receiving any medication or hormone therapy on a daily basis were excluded. Shift workers and patients who experienced severe dry mouth that may interfere with diurnal cortisol measurement or saliva sampling were also excluded. The institutional review board of Taipei Veterans General Hospital approved the study protocol, and each participant provided written informed consent. All methods were carried out in accordance with relevant guidelines and regulations (Declaration of Helsinki).
At the first visit, all patients with FM and healthy controls completed a questionnaire on the extent of body pain (Widespread Pain Index, [WPI]; range 0-19) 1 , body pain intensity (0-10 on a numerical rating scale), duration (in years) of body pain, and number and severity of the centralized symptoms of chronic pain (Symptom Severity Scale [SSS]; range 0-12), including fatigue, unrefreshing sleep, cognitive symptoms, headache, lower abdominal pain or cramps, and depression 1 . The FIQR was administered to all participants to assess their FMrelated functional disability 18 . To evaluate depression severity, the Beck Depression Inventory version 1 (BDI-I) was administered to all participants 25 . Each participant also completed a manual tender point survey. Trained research assistants palpated on the 18 specific anatomical positions defined by the 1990 ACR FM classification at a pressure of 4.0 kg/m 2 as measured using a dolorimeter 26  Salivary cortisol level measurement. One day prior to the aforementioned cognitive assessments, all participants were instructed to follow the protocol modified from the work of McLean, et al. 36 to collect their saliva in sample tubes at home at four time points: awakening, 30 min after awakening, 3 pm, and bedtime. For saliva sampling at awakening, participants were asked not to brush their teeth, eat, or drink before sampling. All collected saliva samples were brought to the hospital the next day and preserved in our laboratory refrigerator at − 20 °C until further analysis using commercial enzyme-linked immunosorbent assay kits, AssayMax ELISA Kits (Assaypro, St. Charles, MO, USA). The cortisol awakening response (CAR) was calculated as the difference in cortisol levels between awakening and 30 min after awakening 37 .

Statistical analysis.
Because of the small sample size, Mann-Whitney U tests and chi-square tests were used to examine intergroup differences in demographic and clinical profiles, subjective cognitive complaints, and diurnal cortisol levels. For objective cognitive testing, we used stratification (education years ≤ 12 vs. > 12) to adjust education in both FM and control groups. After stratification for education, fewer participants (both FM and controls) were in the lower education group. Stratification was also applied on the results of subjective cognitive complaints questionnaire to compare the subjective and objective cognitive deficits. Correlation analysis was used to explore the association among demographic and clinical profiles, items of the subjective cognitive complaint questionnaire, objective cognitive testing, and diurnal cortisol levels. For the effect size, φ was for the between group comparisons of the subjective cognitive complaints questionnaire by chi-square test, Cohen's d for the continuous variables such as total score of the subjective cognitive complaints questionnaire and diurnal cortisol levels, and Hedges' g for the objective cognitive tests after stratification because of smaller sample size. SPSS version 22 (IBM Inc., Armonk, NY, USA) was used to perform the aforementioned statistical analyses, and a P value of < 0.05 was considered statistically significant. The effect size was small, medium and large if 0.1, 0.3, 0.5 for φ and 0.2, 0.5, 0.8 for Cohen's d 38 and Hedges' g 39 , respectively.

Results
Demographics and clinical profile. In total, 44 patients with FM and 48 healthy controls were enrolled in this cross-sectional study. Age and sex did not differ between the FM and control groups, whereas the years of education were fewer in the FM group than in the control group (P = 0.002; Table 1). As expected, the FM group demonstrated higher TTP and TTS on palpation and reported higher scores on the WPI. Moreover, compared with controls, a higher proportion of patients with FM reported fatigue, cognitive symptoms, unrefreshed sleep, headache, and depression on the SSS; the total score on the SSS was also higher in the FM group. The scores on the BDI-I and FIQR were also higher in the FM group than in the control group (Table 1).
Subjective cognitive complaint questionnaire. Compared with the control group, the FM group reported more subjective cognitive complaints, mostly with medium effect sizes for each item ( Table 2). In the FM group, 90%, 60%, 33%, and 16% of patients reported at least one symptom in executive, memory, language, and visuospatial domains, respectively. The total score on the subjective cognitive complaint questionnaire was worse in the FM group with a large effect size (FM: 4.1 ± 2.6; controls: 1.3 ± 2.2, P < 0.001, d = 1.171). The total score was highly correlated with the severity of cognitive symptoms in the SSS (0-3) in all participants (FM and controls combined, r = 0.537, P < 0.001) and controls (r = 0.478, P = 0.001) but not in the FM group (r = 0.117, P = 0.485). The total score in the FM group was moderately correlated with functional disability (FIQR, r = 0.335, P = 0.040). After stratification, patients with FM still reported more subjective cognitive complaints than the controls regardless the education level (Supplementary Table S2 Table 3. In the lower education subgroup analysis, objective cognitive performance did not differ between FM and control subjects (Table 4 Table S4) or in all participants (FM and controls combined; Supplementary Table S5). Notably, correlations were observed in language and executive domains but not in the memory and visuospatial domains in the FM group. Also, no correlation was observed between the total score on the subjective cognitive complaint questionnaire and objective cognitive performance.

Diurnal cortisol levels and clinical correlation.
Diurnal cortisol levels in the FM group tended to be lower than those in the control group (Fig. 1). Correlations between the diurnal cortisol levels and objective cognitive performance were noted in several cognitive domains, such as memory, visuospatial, language and executive functions in the FM group (Supplementary Table S6). In contrast, the correlations were fewer and only observed in the executive domain when all participants (both FM and controls) were included in the correlation analysis (Supplementary Table S7). Among the cognitive tests showing differences between FM and controls, moderate positive correlations existed between the CVVLT−10 M and the cortisol level at 30 min after awakening (r = 0.385, P = 0.014) and between the BNT and the awakening cortisol level within the FM group (r = 0.367, P = 0.020). The CAR was not correlated with any objective cognitive tests. www.nature.com/scientificreports/ Depression (the BDI-I score) was not correlated with subjective memory complaints or most of the items of the objective cognitive tests in the FM (Supplementary Table S4), and the aforementioned differences between the groups remained the same after adjustment for the BDI-I score.

Discussion
In this study, we characterized changes in the cognitive function of patients with FM and investigated their relationship with the clinical profile of FM and diurnal cortisol levels. Patients with FM had subjective cognitive complaints regarding executive, language, and memory domains. In objective cognitive testing, patients with FM demonstrated deficits mainly in memory, followed by executive, and language functions. In addition, patients with FM displayed a lower diurnal pattern of cortisol levels, particularly at 30 min after awakening with a large effect size and at bedtime with a moderate effect size. Moderate positive correlations existed between the CVVLT−10 M and the cortisol level at 30 min after awakening and between the BNT and awakening cortisol level within the FM group. Concerning subjective cognitive complaints, our study demonstrated that patients with FM faced difficulties in memory, executive, and language domains. In agreement with our findings, subjective complaints in memory 40,41 , executive 42 , and language domains 43 have been reported in patients with FM. A recent review summarizing 52 studies on cognitive function in FM also concluded that memory, attention (executive), and word-finding difficulty (language and semantic memory) are the principal subjective and objective cognitive deficits in these patients 17 . Consistent with this review, the present study also demonstrated deficits in memory (delayed recall in CVVLT and TY−CFT), executive (WCST) and language domains (BNT) in patients with FM.
Previous neuroimaging studies have provided a biological basis for cognitive dysfunction in patients with FM. Specifically, poor performance on WCST in our patients with FM may represent disrupted problem-solving, an ineffective hypothesis-testing strategy 35 , increased distractibility, or rule detection deficit 44 . A pertinent MRI study in FM revealed trophic changes in the anterior cingulated cortex, a brain area related to executive function 45 . The episodic memory deficits in delayed recall of CVVLT and TY−CFT were supported by another MRI study demonstrating decreased bilateral hippocampal volumes in FM 46 . On the other hand, though the BNT is commonly categorized as a language test, the worse BNT in our patients with FM may reflect poor access to semantic memory 47 . Reilly, et al. 48 demonstrated the utility of the BNT to assess semantic deficits in patients with Alzheimer's disease and semantic dementia, and several studies also have demonstrated semantic access deficits in patients with FM 49 . Moreover, the semantic access also relies on the function of the hippocampus 50 . Sawrie, et al. 51 demonstrated that the performance on the BNT is associated with the hippocampus in patients with intractable temporal lobe epilepsy on quantitative 1 H magnetic resonance spectroscopy. Thus, both episodic and semantic memory deficits may be linked to hippocampal dysfunction.
In addition to neuroimaging, an association between subjective cognitive complaints and FIQR was demonstrated in the present study and the study conducted by Gelonch, et al. 42 . As reported by Williams, et al. 43 , this association may imply that the subjective cognitive dysfunction in patients with FM may additionally result from pain severity, fatigue, mood, sleep, and other insidious clinical factors that interactively affect the overall perception of functional disability. Given the close relationship between these factors and stress, we assumed that stress is an omnibus factor that directly or indirectly affects the cognitive function of patients with FM 2 . In agreement, our data demonstrated that stress, as assessed by salivary cortisol levels, was associated with TTP in all participants and with the performance of patients with FM on the CVVLT-10 M and BNT.
The relationship between the altered cortisol levels and cognitive functions has been observed in several diseases. Studies on PTSD 52 , CFS 17 , Cushing syndrome 53 , and Addison's disease 22 have reported deficits mostly in memory, executive function, and attention, in line with our results. However, few studies have investigated the relationship between cortisol and cognition in specific functional domains in patients with FM. Sephton,et al. 54 demonstrated that the mean diurnal levels of salivary cortisol were positively correlated with the performance on the visual reproduction task. Barcelo-Martinez, et al. 55 demonstrated a negative correlation of the diurnal difference (the cortisol level at 8 am to 9 am-cortisol level at 4 pm to 5 pm) in serum cortisol levels with verbal memory tested using the Rey Auditory Verbal Learning Test and cognitive flexibility tested using the WCST, suggesting that a more fluctuated diurnal cortisol level may impair the cognitive function. However, in the study by Barcelo-Martinez et al., the cortisol level in the FM group was comparable to but tended to be higher than that in the control group, different from our patient group with lower cortisol levels. This difference might be because of the difference in the sampling method. Note that the stress of venepuncture in the study by Barcelo-Martinez et al. may have also biased the subsequent data on that day 56 .
Our study demonstrated that hypocortisolism is associated with cognitive change. The correlations between WMS-LM, TMT-B and diurnal cortisol levels may imply that attenuated diurnal cortisol levels account for the cognitive deficits. Especially, in the two cognitive tests that showed groups differences, moderate positive correlations were noted between the BNT and the awakening cortisol level, and between the CVVLT-10 M and the cortisol level at 30 min after awakening. Therefore, we suggested a generally inadequate diurnal cortisol level may be harmful to cognitive function in patients with FM. However, not all correlations between cortisol and objective cognitive tests in the FM group showed consistent results. For example, the digital backward showed a negative correlation with the awakening cortisol level. Further studies are needed to clarify the inconsistent cortisol effect upon different cognitive tests.
Although no direct evidence supports the link between altered cortisol levels and both episodic and semantic memory deficits, some indirect evidence might suggest hippocampus as a possible neural substrate for this link. The hippocampus is a limbic structure that is closely related to stress processing 20 and both episodic and semantic memory function 50 , as tested by the CVVLT and BNT in the present study. Earlier studies in patients with PTSD reported a correlation between hippocampal atrophy and the accumulative effect of chronic stress (disease duration) 57 and also a correlation between hippocampal atrophy and verbal memory deficits 58 . Taken together, we propose that the memory deficit in patients with FM may be associated with hippocampal dysfunction secondary to chronic stress. www.nature.com/scientificreports/ Intriguingly, not all functional domains involved in subjective cognitive complaints from patients with FM demonstrated objective deficits on cognitive function testing. The subjective cognitive complaints were only correlated with the objective cognitive performances in the language and executive domains but not in the memory and visuospatial domains. In the lower education subgroup analysis, the FM group had more subjective cognitive complaints than the controls but comparable objective cognitive performances in memory and executive domains. In fact, such a "mismatch" in cognitive function has been observed in CFS 17 , a common comorbidity of FM, and other clinical pain conditions such as migraine 59 . Biologically, the mismatch may be explicable by cognitive compensation amid objective function testing. In functional MRI studies, subjective illness experience and objective performance revealed separate neural substrates 60 . Another study revealed that patients with FM recruited additional neural networks to achieve the same performance on the Go/NO-go test 61 , and the resultant mental overloading and exhaustion may aggravate subjective cognitive impairment. Psychologically, several potential factors may lead to greater subjective experience of cognitive deficits (compared with objective cognitive testing). In addition to the aforementioned perceived functional disability and mental exhaustion related to brain processing overloading, memory perfectionism, heightened self-monitoring 17 , and altered interoceptive awareness 62 may contribute to subjective cognitive dysfunction. Studies using personality inventories have been attempting to establish a "pain personality," including the previous conversion model and the newer vicious cycle model, which refers to the role of problematic stress coping because of the trait of harm avoidance and low self-directedness 63 .
This study has some limitations. First, women comprised 91% of our FM group, and the generalizability of our data is thus limited. Second, the manual tender point survey was not objective, and interpretation of result may be biased by the location of the survey sites, patient and examiner positioning, order of examination, and pressure application technique 64 . However, our purpose was to confirm that these patients with FM also fulfilled the 1990 ACR criteria and provided an additional assessment of tenderness following the work of Hsiao, et al. 65 . Third, our subjective cognitive complaint questionnaire has not been validated. In fact, validated tools for evaluating subjective cognitive decline are still under development 66 . We attempted to use a comprehensive questionnaire as an exploratory tool to measure subjective cognitive complaints, correlations to the cognitive problems of SSS, and objective cognitive performance (Supplementary Tables S4 and S5) in our study, which may support the validity of our subjective cognitive complaint questionnaire. Fourth, we did not assess the effort required for performing objective cognitive tests. Because body pain may affect motivation, Lockhart and Satya-Murti 67 emphasized the role of inadequate effort, especially in a clinical scenario involving patients with FM, where the base rate of inadequate effort is up to 35%. However, Pidal-Miranda, et al. 41 suggested that this effect might be small in clinical studies because these study volunteers usually have higher motivation and less compensationseeking tendencies. Future studies may incorporate performance-validating tests 68 to explore effort as a significant contributing factor for subjective or objective cognitive functions in FM. Fifth, the cortisol level of participants was tested only for 1 day, and testing for more days would provide more reliable results, considering day-to-day differences in levels 69 . Finally, our exploratory cross-sectional study demonstrated the relation between stress and cognitive change; a longitudinal study is warranted to confirm the causality.

Conclusion
Our study demonstrated that patients with FM experienced both subjective and objective cognitive changes in memory, executive and language functions. In objective cognitive testing, both episodic and semantic memory functions in patients with FM were correlated with decreased salivary cortisol levels. The link between cognitive change and cortisol suggests stress maladaptation may play some role in the cognitive dysfunction associated with FM. Additional studies must elucidate whether stress management improves cognitive performance in patients with FM.

Data availability
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. www.nature.com/scientificreports/ Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.