Keywords
type 2 diabetes - sleep quality - sleep duration - brain-derived neurotrophic factor
- Ghana
Introduction
Type 2 diabetes (T2DM) is a condition whereby the body tissues are unable to sense
and respond to insulin, leading to metabolic abnormalities in carbohydrate, lipid,
and protein homeostasis. In 2019, it was estimated that about 463 million adults
were living with diabetes and that number may grow to about 700 million people with
diabetes by 2045 [1]. The prevalence of
diabetes in the Ghanaian adult population was estimated to be 6.46% by a recent
meta-analysis study [2]. T2DM negatively
affects sleep quality due to the effect of stress from various diabetic management
regimens, as well as the presence of diabetic complications [3]. There is a bidirectional association
between T2DM and insomnia, with some studies reporting that sleep disturbances
contribute to the pathogenesis of T2DM, while others report sleep disturbances as
a
complication of T2DM [4]
[5]
[6]. Sleep adequacy is usually assessed in terms of sleep duration and
quality of sleep, with the two dimensions usually overlapping in many research
reports [7]
[8]. Few studies have discussed the
relationship between diabetes and sleep disturbances in Africa [9]
[10] and no studies have investigated the underlying mechanisms of poor
sleep in the African population.
Brain-derived neurotrophic factor (BDNF) is the most common neurotrophic growth
factor in medical literature, and it is reported to regulate the survival,
development, and differentiation of neurons [11]. Several studies have reported that the dysregulation of serum BDNF
affects cognitive functions through modulation of neurite outgrowth, neuronal
differentiation, survival, and growth [12]. Furthermore, BDNF is reported to regulate tissue metabolism through its
central and peripheral influence on various enzymes that regulate intermediary
metabolism, with abnormal levels of BDNF resulting in dysglycemia and dyslipidaemia
[13]. BDNF levels are generally
reported to be reduced in patients with T2DM [14] and individuals with poor sleep [15], although some studies have reported contrasting findings [16]. The relationship between diabetes,
mental disorders, and circulating BDNF levels may imply that serum BDNF may be an
important psychophysiological biomarker of metabolic and mental disorders [17]. There is a paucity of data on the
levels of BDNF in patients with T2DM and sleep quality and duration in the
sub-Saharan African population. This study investigated the prevalence of poor sleep
quality and short duration in patients with T2DM in Ghana and their association with
serum BDNF levels. We hypothesize that serum BDNF levels would be low in patients
with T2DM and associated with poor sleep quality and short sleep duration.
Methods
The study was a case-control design, conducted at the Diabetic Clinic, Korle Bu
Teaching Hospital in Accra, Ghana, from December 2022 to June 2023. Patients with
T2DM were selected using a systematic random sampling as every third eligible
patients aged 30 through 65 years was invited to participate in the study.
Thereafter, comparable nondiabetic controls were purposively contacted from the
communities close to the hospital to participate in the study. T2DM status was
determined clinically as patients who were diagnosed with diabetes after age 30 and
were managed initially on lifestyle modification or antidiabetic drugs. The
exclusion criteria were: patients with type 1 diabetes, infectious disease or
terminal illness, a diagnosis of neurological or psychiatric disease, multiple
sclerosis, chronic periodontitis, rheumatoid arthritis, coronary heart disease,
heart failure, or chronic liver or kidney disease, or being treated with
clopidogrel, corticosteroids, antidepressants, statins or aspirin, as well as those
on shifting work schedules.
A structured questionnaire containing elements of sociodemography, lifestyle, and
clinical and medical history was administered to all participants. Blood pressure
was measured in a seated position after 5 min rest using an automated digital blood
pressure monitor (Omron 907XL pro, Healthcare, Inc., Vernon Hills, IL). Body weight
and height were measured with a validated scale (Seca 740 scale) and a stadiometer,
respectively, and the body mass index (BMI) was calculated as: weight (kg)/ height
(m2).
Sleep quality was assessed using the Pittsburgh Sleep Quality Index (PSQI), which
is
a validated questionnaire, used to measure sleep quality over the past month and has
previously been implemented among diabetes patients. The items on the PSQI
instrument are categorized into seven different domains (sleep quality, sleep
latency, sleep duration, sleep efficiency, sleep disturbances, sleep medication, and
daytime dysfunction) with scores ranging from 0 (no difficulty) to 3 (severe
difficulty). The scores for the seven domains were summed up to give a final PSQI.
A
total PSQI score of<5 was used to define poor sleep quality as per guidelines
[9]
[18]
[19].
Participants were asked about their bedtimes and wake-up times on the weekdays and
the weekends. Sleep duration was calculated as the difference between the bedtime
and waking time for weekdays and weekends and weighted as 5/7*(weekday sleep
duration)+2/7*(weekend sleep duration). Short sleep was defined as sleeping
duration<7 h based on the recommendation of the National Sleep Foundation [20].
After 8–12 h of overnight fasting, blood samples were collected into appropriate
vacuum tubes, centrifuged at 4000 G, and the serum/plasma was aliquoted and stored
at − 70°C until further analysis. Levels of fasting plasma glucose, total
cholesterol, high-density (HDL) lipoprotein cholesterol and plasma triglyceride
levels were analyzed using a semi-automated biochemistry analyzer (Contec BC 400,
China) and commercial reagents (Randox Laboratory Reagents, UK).
Serum levels of BDNF were measured by enzyme-linked immunosorbent assay (ELISA)
according to the procedures supplied by the manufacturer (DuoSet, R&D Systems,
Minneapolis, MN, USA). All samples were assayed in triplicate, and the technician
was blind to the group assignment of the samples. The lower detection limit was
5 pg/mL. Concentrations were expressed as ng/mL. The inter- and intra-assay
coefficients of variation were less than 5%.
The study protocol was approved by the Ethics and Protocol Review Committee of the
College of Health Sciences of the University of Ghana (Protocol ID number:
CHS-Et/M24.12/2018–2019), and each participant provided written voluntary informed
consent before being included in the study.
Sample size calculation
The sample size required for this study was calculated based on the pilot data of
BDNF levels from 20 patients with T2DM and 20 nondiabetic controls (23.5±12.1 vs
28.9±11.4 ng/mL). A minimum of 76 participants were required in each group to
achieve a power of 80% at a 95% significance level. We, therefore, recruited 100
patients with T2DM and 80 nondiabetic controls.
Statistical analysis
Data were analyzed using SPSS version 28. Data were presented as mean±standard
deviation for continuous variables and as proportions for categorical variables.
Differences between patients with T2DM and nondiabetic control with regards to
their socio-demographic, clinical, and biochemical variables were analyzed using
a chi-squared (χ2) test to compare categorical variables and
Student's t-test for continuous measures. Three different binary and
multivariable logistics regression models to assess the association between
T2DM, serum BDNF, and sleep status, i. e., poor sleep quality and sleep
duration. In the first logistic regression model, serum BDNF was excluded, and
in the second model, serum BDNF was added to the model, while T2DM status was
excluded. The third model included both T2DM and serum BDNF, as well as their
cross-product term (T2DM+×+BDNF), to assess multiplicative interaction. The
biological interaction between T2DM and serum BDNF levels was estimated using
the relative excess risk due to interaction (RERI), the attributable proportion
due to interaction (AP), and the synergy index (S) as described by Rothman et
al. [21]. RERI or AP=0 means no
interaction or exactly additivity; RERI or AP+>+0 means positive interaction
or more than additivity; RERI or AP<0 means negative interaction or less than
additivity. The level of significance was set at p<0.05.
Results
General characteristics of study participants
In this study, compared to nondiabetic controls, patients with T2DM were likely
to be hypertensive, less educated, and had previously smoked and taken alcohol.
The mean BMI, blood pressure, fasting plasma glucose, triglycerides, total, HDL,
and LDL cholesterol levels were significantly higher in patients with T2DM
compared to the nondiabetic individuals. Serum BDNF levels were lower in
patients with T2DM compared to the nondiabetic controls ([Table 1]). The median duration of
diabetes in patients with T2DM was 8.1 years (range: 0.1–17.6 years), with 46,
37, and 17 patients having a duration of T2DM with<5 years, 5–10 years,
and+>+10 years, respectively. Concerning diabetes treatment, four patients
were on lifestyle management, 61 patients were on oral hypoglycemic agents, and
35 patients were on insulin and oral hypoglycemic agents.
Table 1 General characteristics of the study
participants.
|
T2DM patients (n=100)
|
Nondiabetic controls (n=80)
|
p
|
|
Demographical parameters
|
|
Gender, n (%)
|
|
|
0.404
|
|
Male
|
42 (42)
|
35 (43.8)
|
|
|
Female
|
58 (58)
|
45 (56.2)
|
|
|
Age, yr
|
56±8.4
|
53.6±10.6
|
<0.001
|
|
Age decades, n (%)
|
|
|
0.001
|
|
<40
|
12 (12)
|
12 (15)
|
|
|
40–49
|
23 (23)
|
24 (30)
|
|
|
50–59
|
45 (45)
|
32 (40)
|
|
|
60+
|
20 (20)
|
12 (15)
|
|
|
Married
|
68 (68)
|
50 (62.5)
|
0.363
|
|
Educational levels, n (%)
|
|
|
0.005
|
|
None
|
16 (16)
|
9 (11.3)
|
|
|
Junior high school
|
34 (34)
|
28 (35)
|
|
|
Senior high school
|
32 (32)
|
34 (42.5)
|
|
|
Tertiary
|
18 (18)
|
9 (11.2)
|
|
|
Employment, n (%)
|
|
|
0.021
|
|
Formal
|
37 (37)
|
37 (45.6)
|
|
|
Self-employed
|
44 (44)
|
33 (41.3)
|
|
|
Unemployed
|
19 (19)
|
10 (12.5)
|
|
|
Clinical parameters
|
|
Alcohol intake, n (%)
|
31 (31)
|
21 (26.3)
|
0.005
|
|
Previous smoker, n (%)
|
9 (9)
|
5 (6.3)
|
<0.001
|
|
Hypertension, n (%)
|
69 (69)
|
17 (21.3)
|
<0.001
|
|
BMI, kg/m2
|
30.8±7.1
|
25.9±5.9
|
<0.001
|
|
Systolic BP, mmHg
|
147±22
|
138±17
|
0.003
|
|
Diastolic BP, mmHg
|
89±14
|
82±18
|
0.005
|
|
Mean BP, mmHg
|
75±15
|
69±11
|
0.002
|
|
Pulse BP, mmHg
|
61±12
|
52±11
|
<0.001
|
|
Heart rate, beats/min
|
72±16
|
67±8
|
<0.001
|
|
Biochemical parameters
|
|
Fasting plasma glucose mmol/L
|
7.8±3.5
|
5.4±0.8
|
<0.001
|
|
Total cholesterol, mmol/L
|
6.1±2.3
|
5.2±1.4
|
0.002
|
|
Triglycerides, mmol/L
|
2.8±1.1
|
2.2±0.9
|
<0.001
|
|
HDL cholesterol, mmol/L
|
1.2±0.4
|
1.6±0.5
|
<0.001
|
|
LDL cholesterol, mmol/L
|
3.2±1.1
|
2.4±1.2
|
0.004
|
|
BDNF, ng/mL
|
22.1±8.4
|
26.1±10.2
|
0.005
|
SHS, senior high school; BMI, body mass index; BP, blood pressure; PHQ,
Patient’s Health Questionnaire; BDNF, brain-derived neurotrophic factor;
HDL, high-density lipoprotein; LDL, low-density lipoprotein; T2DM, type
2 diabetes mellitus.
Sleep deficits and serum brain-derived neurotrophic factor levels
Patients with T2DM had higher global PSQI scores and a higher prevalence of poor
sleep quality than the nondiabetic controls. Patients with T2DM had lower mean
self-reported sleep duration than the nondiabetic controls and participants who
reported short sleep duration (<7 h sleep) were mostly patients with T2DM
([Table 2]). In T2DM patients,
those with poor sleep quality had lower serum BDNF levels compared to patients
with good sleep quality. In nondiabetic controls, serum BDNF levels were similar
between those with good and poor sleep quality ([Fig. 1]). In patients with T2DM and
nondiabetic controls, those with short sleep duration had lower BDNF levels
compared with participants with normal sleep duration ([Fig. 2]).
Fig. 1 Comparison of serum BDNF levels in study participants by
their quality of sleep status.
Fig. 2 Comparison of serum BDNF levels in study participants by
the duration of sleep.
Table 2 Comparison of PSQI scores and sleep duration among
study participants.
|
T2DM patients
|
Non-diabetes controls
|
p
|
|
PSQI score
|
8.8±3.7
|
5.9±2.8
|
<0.001
|
|
PSQI score+>+5
|
99 (61.9)
|
22 (27.5)
|
<0.001
|
|
Sleep duration, h
|
6.1±2.2
|
6.9±1.1
|
0.003
|
|
Sleep duration classification
|
|
0.01
|
|
<7 h
|
60 (60)
|
34 (42.5)
|
|
|
+≥+7 h
|
40 (40)
|
46 (57.5)
|
|
PSQI, Pittsburgh Sleep Quality Instrument; T2DM, type 2 diabetes
mellitus.
Relationship between type 2 diabetes mellitus, brain-derived neurotrophic
factor, and sleep status
In the logistic regression models, T2DM status was associated with increased odds
of poor sleep quality, while an increase in serum BDNF level was associated with
decreased odds of poor sleep quality in unadjusted models. In the adjusted
model, having T2DM was associated with increased odds of poor sleep quality
compared to nondiabetic controls. In the interactive model, the multiplicative
interaction between T2DM and serum BDNF levels significantly increased the odds
of having poor sleep quality in the unadjusted model but was non-significant in
the adjusted models. There was a negative biological interaction between T2DM
and serum BDNF levels, as indicated by RERI. In addition, the interaction
between T2DM and serum BDNF was associated with 18% and 12% higher risk (AP
score) in unadjusted and adjusted models, respectively ([Table 3A]).
Table 3 The interactive effects of T2DM and serum BDNF on
sleep quality and duration from logistic regression
models.
|
Unadjusted model
|
Adjusted model*
|
|
|
|
|
|
OR (95% CI)
|
p
|
OR (95% CI)
|
p
|
|
T2DM
|
4.28 (2.38–7.68)
|
<0.001
|
2.06 (1.07–6.43)
|
0.039
|
|
BDNF
|
0.84 (0.62–0.97)
|
0.043
|
0.93 (0.47–1.06)
|
0.094
|
|
Interactive model
|
|
|
|
|
|
T2DM
|
3.17 (1.93–8.55)
|
0.002
|
2.62 (1.11–8.2)
|
0.029
|
|
BDNF
|
0.69 (0.41–0.93)
|
0.009
|
0.86 (0.58–1.12)
|
0.103
|
|
T2DM+×+BDNF
|
2.43 (1.07–5.09)
|
0.03
|
2.21 (1.03–4.82)
|
0.047
|
|
Measures of biological interaction
|
|
|
|
|
RERI
|
− 0.44 (-0.19–− 0.75)
|
|
− 0.28 (− 0.09–− 0.88)
|
|
|
AP
|
− 0.18 (-0.05–− 0.46)
|
|
− 0.12 (− 0.03–− 0.67)
|
|
|
S
|
0.77 (0.39–0.98)
|
|
0.34–1.06)
|
|
|
|
|
T2DM
|
2.31 (1.12–7.69)
|
0.01
|
1.63 (1.03–3.79)
|
0.028
|
|
BDNF
|
0.78 (0.41–0.95)
|
0.017
|
0.81 (0.59–1.01)
|
0.061
|
|
Interactive model
|
|
|
|
|
|
T2DM
|
2.68 (1.14–4.99)
|
0.003
|
1.43 (1.06–4.56)
|
0.014
|
|
BDNF
|
0.74 (0.52–1.04)
|
0.087
|
0.68 (0.42–0.98)
|
0.008
|
|
T2DM+×+BDNF
|
2.03 (1.02–6.87)
|
0.034
|
1.29 (0.81–4.66)
|
0.201
|
|
Measures of biological interaction
|
|
|
|
|
|
RERI
|
− 0.39 (− 0.08–− 0.81)
|
|
− 0.18 (− 0.01–− 1.01)
|
|
|
AP
|
− 0.19 (− 0.02–− 0.53)
|
|
− 0.14 (− 0.01–− 1.06)
|
|
|
S
|
0.78 (0.35–0.92)
|
|
2.64 (0.96–5.82)
|
|
*Adjusted for age, sex, BMI, diabetes medication, smoking,
alcohol, hypertension,. educational level and employment status. T2DM,
type 2 diabetes mellitus; BDNF, brain-derived neurotrophic factor; PSQI,
Pittsburgh Sleep Quality Instrument; RERI, relative excess risk due to
interaction; AP, attributable proportion due to interaction; S, synergy
index.
Similarly, T2DM status was associated with increased odds of short sleep
duration, while an increase in serum BDNF level was associated with decreased
odds of short sleep duration in unadjusted models. In the adjusted model, having
T2DM was associated with increased odds of short sleep duration compared to
nondiabetic controls. In the interactive model, the multiplicative interaction
between T2DM and serum BDNF levels significantly increased the odds of short
sleep duration in the unadjusted model, but there was no association in the
adjusted model. There was a negative biological interaction between T2DM and
serum BDNF levels in the unadjusted and adjusted models as indicated by RERI
values. In addition, the interaction between T2DM and serum BDNF was associated
with 19% and 14% excess risk (AP score) in unadjusted and adjusted models,
respectively ([Table 3B]).
Discussion
The main findings of this study were that, compared to nondiabetic controls, patients
with T2DM had low levels of BDNF and a high prevalence of self-reported poor sleep
quality and shorter sleep duration. Increased serum BDNF was associated with
decreased odds of poor sleep quality in T2DM patients, and decreased odds of short
sleep duration in nondiabetic controls.
We have previously reported a high burden of poor sleep in patients with T2DM
compared to nondiabetic controls, and this was associated with reduced HDL
cholesterol levels and increased triglyceride levels [22]. We found that patients with T2DM had
short sleep duration compared to nondiabetic controls, similar to what has been
reported in other studies. In the Taiwanese population, those with short sleep
duration (+≤+5 h) had twice the odds of diabetes compared to those with 7 h or more
sleep duration [18].
We found that serum BDNF levels were significantly lower in patients with T2DM
compared to their nondiabetes counterparts. This is consistent with a study by
Krabbe et al., who reported reduced plasma BDNF levels in patients with diabetes
compared to nondiabetic controls, and even in healthy individuals, hyperglycemia
reduces circulating BDNF [14].
Furthermore, Chinese patients with T2DM had lower serum levels of BDNF compared to
nondiabetic controls [23]. Contrary to our
findings, some studies have reported high BDNF levels in patients with T2DM compared
to nondiabetic individuals [24]
[25]. These conflicting data about the
association between BDNF and diabetes may be due to, at least in part, ethnic
differences, duration, and severity of diabetes [26]. BDNF has been shown to have an anti-diabetic effect by increasing
insulin secretion and sensitivity in peripheral tissues, and decreasing blood
glucose through insulin-independent mechanisms [27]. For instance, intraventricular administration of BDNF in diabetic
mice was reported to mitigate hyperglycemia by reducing hepatic glucose output
through the normalization of glucagon secretion and hepatic expression of
gluconeogenic enzyme synthesis, without affecting insulin secretion or sensitivity
[28].
One interesting finding of our study was the significantly negative biological
interaction between T2DM and serum levels of BDNF concerning poor sleep quality;
however, the multivariable model revealed that the interaction between T2DM and BDNF
did not contribute to poor sleep duration. Therefore, it is reasonable to infer that
the quality of sleep, rather than quantity, may be associated with circulating BDNF
levels. In patients with T2DM, unlike their nondiabetic counterparts, there are a
lot of sleep problems potentially contributing to poor sleep quality [6], and this may be responsible for the
reduction of BDNF. The relationship between the duration of sleep and serum BDNF
levels may follow a U-shaped pattern, as reported in adolescents [29]. Hence, the linear models applied to
our analysis of the data from this study might have masked possible biological
interaction. We could not examine the U-shaped relationship because of too small
number of participants with excess sleep duration (>9 h) to be analyzed
separately. Further studies may be required to assess whether this observation may
be due to metabolic abnormalities due to insulin dysfunction or the presence of
other comorbidities in patients with T2DM. On the other hand, short sleep duration,
rather than poor sleep quality, was relevant in maintaining serum BDNF levels in
nondiabetic controls. This is consistent with previous studies that reported low
levels of BDNF in patients with insomnia [30]. Preclinical studies have also demonstrated the association between
short sleep duration and BDNF [31]. In
interventional studies, the reversal of sleep deficits with pharmacological agents
[32] or non-pharmacological such as
exercise and repetitive transcranial magnetic stimulation [33] were able to increase circulating BDNF
levels. In contrast to our findings, studies conducted in the Japanese population
reported no association between subjective sleep quality and serum BDNF levels [34]. Likewise, Mokoteit et al. reported an
association between serum BDNF and rapid-eye-movement sleep, but did not find a
correlation with objective sleep quality through polysomnography [35]. The underlying mechanism of reduction
of BDNF levels in T2DM and insufficient sleep may be related to stress [5]
[36]. Both diabetes and insufficient sleep hypertactivate the dual stress
loop, involving the hypothalamic-pituitary-adrenal and sympatho-adrenomedullary axes
[5]
[37]. This leads to high stress levels in
patients, which has been shown to reduce the synthesis of BDNF mRNA in the brain
[36].
Limitations of study
The interpretation of the findings of this study has some limitations. The data
were collected cross-sectionally in a single facility, limiting the inference of
causality and generalization to the entire Ghanaian population. Quality and
duration of sleep in this study were self-reported, which is prone to recall
bias. Furthermore, we measured circulating levels of BDNF in the serum, which
may differ from plasma and cerebrospinal BDNF levels [12]. The concentration of BDNF in serum
has been reported to be 50 times higher than that of plasma. This is due to the
capacity of platelets to absorb BDNF produced by the brain and release them into
serum during the coagulation process [38]. This may explain the observed moderate correlation between
plasma BDNF and hippocampal BDNF in a previous study [11]. In our methodology, we reduced the
impact of diurnal variability and storage effect on BDNF levels by taking
fasting blood samples early in the morning before 9 am and measuring serum
levels within 6 months of storage at − 80°C. Indeed, 12 months of storage of
samples at that temperature has been shown to have no significant effect on BDNF
levels in a healthy population [39].
However, the Elisa method we used to test serum levels of BDNF is reported to
capture both mature BDNF and proBDNF forms [40], which could have introduced some errors in our analysis. We,
however, expect the effect of this error to be negligible with respect to our
sample size and the use of nondiabetic controls.
Conclusion
In our study population, we found a high burden of self-reported poor sleep quality
and short duration in patients with T2DM compared to nondiabetic controls. There was
a negative interaction between T2DM and serum BDNF, causing sleep deficits. These
findings emphasize the importance of sleep screening and management as part of
diabetes care to minimize the impact of diabetes on factors that regulate the
functioning of the nervous system.
Authors’ contributions
KY conceptualized the study, analysed the data and drafted the manuscript. JAA
analyzed the data and made scientific contributions to the manuscript. All authors
approved the content of the manuscript.
Availability of data
The dataset supporting the conclusions of this paper is available and can be
requested from the corresponding author.
Ethics approval and consent to participate
Ethics approval and consent to participate
All procedures performed in this study involving human participants were conducted
in
conformity with the Helsinki Declaration on Human Experimentation, 1964, with
subsequent revisions, latest Seoul, October 2008. Ethical approval was obtained from
the Ethics and Protocol Review Committee of the College of Health Sciences of the
University of Ghana (Protocol ID number: CHS-Et/M24.12/20182019) and each patient
provided written voluntary informed consent after the rationale and procedure of the
study were thoroughly explained.
