Ethnic-Specific Type 2 Diabetes Risk Factor PAX4 R192H Is Associated with Attention-Specific Cognitive Impairment in Chinese with Type 2 Diabetes

Abstract

Background:

Type 2 diabetes mellitus (T2DM) has been shown to increase the risks of cognitive decline and dementia. Paired box gene 4 (PAX4), a transcription factor for beta cell development and function, has recently been implicated in pathways intersecting Alzheimer’s disease and T2DM.

Objective:

In this report, we evaluated the association of the ethnic-specific PAX4 R192H variant, a T2DM risk factor for East Asians which contributes to earlier diabetes onset, and cognitive function of Chinese T2DM patients.

Methods:

590 Chinese patients aged 45–86 from the SMART2D study were genotyped for PAX4 R192H variation using Illumina OmniExpress-24 Array. The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) which had been validated in the Singapore population was administered to assess five cognitive domains: immediate memory, visuospatial/constructional, language, attention, and delayed memory. Multiple linear regression was used to assess the association of the R192H risk allele and cognitive domains.

Results:

Patients with two PAX4 R192H risk alleles showed significantly lower attention index score (β= –8.46, 95% CI [–13.71, –3.21], p = 0.002) than patients with wild-type alleles after adjusting for age, gender, diabetes onset age, HbA1c, body-mass index, renal function, lipid profiles, systolic blood pressure, metformin usage, smoking history, education level, Geriatric Depression Scale score, and presence of APOE ɛ4 allele.

Conclusion:

Ethnic-specific R192H variation in PAX4 is associated with attention-specific cognitive impairment in Chinese with T2DM. Pending further validation studies, determining PAX4 R192H genotype may be helpful for early risk assessment of early-onset T2DM and cognitive impairment to improve diabetes care.

Keywords

Cognitive impairment ethnic-specific PAX4 type 2 diabetes

INTRODUCTION

Diabetes has been identified as a risk factor for cognitive dysfunction and dementia [1]. Cognitive decrements can present during pre-diabetic stage and progress throughout the disease course, at a rate that exceeds that of normal cognitive aging [2]. Three stages of cognitive deficits in type 2 diabetes mellitus (T2DM) can be broadly defined in order of severity: diabetes-associated cognitive decrements, mild cognitive impairment (MCI), and dementia [3]. A recent study identified Paired box gene 4 (PAX4) as the novel missing link between T2DM and Alzheimer’s disease (AD) [4].

PAX4 is a transcription factor that is essential for beta cell development and differentiation. In experimental studies, Pax4-null mice did not survive beyond 3 days of birth and were found to possess more alpha cells than beta and delta cells [5]. Homozygous PAX4 knockout (KO) rabbits generated using CRISPR/Cas9 system exhibited growth retardation, persistent hyperglycemia, reduced insulin-producing beta cells and increased glucagon-producing alpha cells. In addition, they also developed diabetic nephropathy, hepatopathy, myopathy, and cardiomyopathy [6]. PAX4 is also implicated in regulation of beta cell survival and proliferation in the mature pancreatic islets [7, 8]. While PAX4 has been widely linked to beta cell physiology and implicated in diabetes, there is sparse evidence to date to suggest its involvement in diabetes-associated cognitive impairment.

Clinically, PAX4 R192H (rs2233580) genetic variation was identified in several large-scale genome-wide association analysis to be associated with the risk of T2D at non-trivial odds-ratios ranging from 1.48 to 1.79 [9 –12]. We also observed this PAX4 R192H genetic variant in our prior studies of early-onset T2DM patients with suspected monogenic diabetes [13] and were able to show that this variant contributes to earlier onset of T2DM subsequently in Chinese patients from two other large T2DM cohorts [14]. Functional and structural studies suggest that this variant is pathogenic and the defects were consistent with PAX4’s roles in beta cell regulation [10 , 16].

Neurologically, there has been increasing evidence of PAX4’s involvement in neurological pathways. Volodin et al. demonstrated that PAX4 is critical for activating expression of genes required for myofibril breakdown after denervation [17]. PAX4 was also subsequently identified as one of the transcription factors affecting most of the downstream differentially expressed genes in the Parkinson’s disease regulatory network [18]. More recently, PAX4 was also found to regulate cytoskeletal changes through common effector proteins in both T2DM and AD pathways [4]. Given the link between PAX4 with T2DM and its potential role for mediating neurological functions, we sought to evaluate the association of the ethnic-specific T2DM risk factor, PAX4 R192H genetic variation, with cognitive impairment in Chinese patients with T2DM.

MATERIALS AND METHODS

Study population

Singapore Study of Macro-angiopathy and Micro-vascular Reactivity in Type 2 Diabetes (SMART2D) Phase 2 was a cross-sectional study on adults clinically diagnosed with T2DM between 21–93 years old who attended the Diabetes Centre in the Alexandra Health Private Limited (AHPL) regional hospital or a primary-care polyclinic in Northern Singapore from September 2014 to October 2018. Subjects were excluded if they had one of the following: active inflammation (e.g., systemic lupus erythromatosis or overt untreated cancer, use of non-steroidal anti-inflammatory medications with the same day of assessment, fasting glucose < 4.5 mM after phlebotomy, use of oral steroids equivalent to prednisolone > 7.5 mg/ day, contraindications to bio-impedance analysis (e.g., cardiac pacemaker), existing neuropsychiatric disease which could impair neurocognitive ability, and inability to give written informed consent. These study participants also did not have any cognitive diseases documented in their medical records. Out of a total of 1,547 patients recruited into this cohort during this period, 1,185 patients aged 45 years old and above underwent neurocognitive assessments, and a sub-set of 590 Chinese patients with no missing data or outliers (N = 2 in RBANS total score) in the study variables of interest were analyzed (Fig. 1). Written informed consent was obtained from all study participants. The study was approved by the National Healthcare Group Domain Specific Review Board in Singapore (DSRB 2014/00667). All experiments were performed in accordance with relevant guidelines and regulations.

Fig. 1

Study sub-cohort assembly. Participants’ data was included in the analysis if they fulfilled the criteria of Chinese ethnicity with no missing data and no outliers in study variables of interest. SD, standard deviation.

Demographics and clinical data collection

Trained research nurses collected data from patients on demographics, educational level, medical conditions, and medications. Blood pressure (BP) was measured with a standard sphygmomanometer for the participants in the sitting position after resting for at least 10 min (2 readings were obtained). Blood and urine samples were sent to KTPH Department of Laboratory Medicine with accreditation from the College of the American Pathologists (CAP) for clinical measurements. Clinical parameters measured included: Haemoglobin A1c (HbA1c) with the Tina-quant Haemoglobin A1c Gen.3 (Roche Cobas^® c501, Switzerland); serum low density lipoprotein cholesterol (LDL-C) with enzymatic colorimetric test (Roche Cobas^® c501, Switzerland); serum creatinine with enzymatic colorimetric test (Roche Cobas^® c501, Switzerland); and urinary albumin with immunoturbidimetric assay (Roche Cobas^® c501, Switzerland). The Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation formula was used to estimate glomerular filtration rate (eGFR) [19]. A 15-item Geriatric Depression Scale (GDS) was used to assess for depressive symptoms [20]. Individuals were considered as having depressive symptoms if they scored ≥5 in GDS [21]. Apolipoprotein E (APOE) genotype (presence of APOE ɛ4 allele) and PAX4 R192H genotype were determined using Illumina OmniExpress-24 array (Illumina, California, USA). PAX4 R192H genotype comprises of: CC genotype (0 risk allele), CT genotype (1 risk allele), and TT genotype (2 risk alleles).

Cognitive function assessments

Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) was used to assess cognitive function in 12 subsets (list learning, story memory, figure copy, line orientation, picture naming, semantic fluency, digit span, coding, list recall, list recognition, story recall and figure recall). Five domains of cognition were derived from these subsets and they were 1) attention, 2) language, 3) visuospatial or constructional abilities, 4) immediate memory, and 5) delayed memory. An overall score (total score) was derived from the five sub-domains [22]. RBANS has been widely used for cognitive assessments worldwide and has also been extensively validated and recognized in our Singapore population [23 –25]. The index scores were standardized to a mean of 100 and standard deviation of 15. A cut-off of 1-2 SD is typically used to indicate mild cognitive impairment. In addition, the locally translated and validated version of the Mini-Mental State Examination (MMSE) was also used [26, 27] to assess the global cognitive function of our study participants. MMSE measures cognitive functioning in the domains of memory, attention, language, praxis, and visuospatial ability [28]. The summed scores range from 0–30 and a cut-off of < 24 is typically used to indicate dementia. For mild cognitive impairment, cut-offs ranging from 24 to 29 have been used [27]. Both RBANS and MMSE instruments were administered by trained research nurses.

Statistical analysis

Data was analyzed using SPSS 27.0 for Windows (SPSS Inc., Chicago, IL, USA). Continuous variables were presented as median and interquartile range (IQR). The Kruskal-Wallis test was used to compare continuous variables. Categorical variables were reported as frequencies or percentages. The Fisher-Freeman-Halton test was used to compare categorical variables. A two-tailed value of p < 0.05 was considered as statistically significant. PAX4 R192H genotype (CC (n = 418), CT (n = 162), and TT (n = 10)) fulfilled Hardy-Weinberg equilibrium with Chi-squared value of 1.62 and p value of 0.20. Multiple linear regression was used to assess the association of PAX4 R192H with RBANS sub-domain and total scores. Covariates known to affect cognitive function and scoring were adjusted for and they consist of age [29], gender [30], diabetes onset age (< 40, 41–64, ≥65) [31], body mass index (BMI) [32], HbA1c [31], systolic blood pressure (SBP) [33], high-density lipoprotein cholesterol (HDL-C) [34], triglycerides [35], eGFR < 60 and ≥60 ml/min/1.73 m² [36], urinary albumin/creatinine ratio (uACR) < 30, 30–300 > 300 mg/g [37], metformin usage [21], smoking history (non-, ex-, current-) [38], education level (≥7 years) [39], GDS score (≥5) [40], and presence of APOE ɛ4 allele [41].

RESULTS

Clinical characteristics of study participants analyzed in this project (overall as well as stratified according to their PAX4 R192H genotype) are given in Table 1. This sub-cohort of 590 study participants comprised of 54.2% male participants. The median age for this cohort was 63.0 (IQR 57.0–69.0) years old and age of diabetes onset was 48.0 (IQR 40.0–55.0) years old with median duration of 14.5 (IQR 8.0–20.0) years. Patients with 2 risk alleles (TT genotype) were younger in index age (median age: 58.5 years old), diabetes onset age (median age: 42.5 years) and had longer duration of diabetes (median years: 18.0 years) as compared to those with CC or CT genotypes. This is consistent with our reported observation that presence of the minor risk (T) allele contributes to earlier onset of T2DM [14].

Table 1

Clinical characteristics of study participants

Clinical characteristic	% or Median (IQR)
	Overall	PAX4 genotype Chr7(GRCh37):g.127253550
		CC (70.8%)	CT (27.5%)	TT (1.7%)	p
Male gender	54.2%	55.0%	53.1%	40.0%	0.598
Age (y)	63.0 (57.0–69.0)	64.0 (57.0–69.0)	62.5 (56.0–68.0)	58.5 (50.3–73.3)	0.177
Age of onset (y)	48.0 (40.0–55.0)	48.5 (40.0–55.0)	45.0 (39.0–54.0)	42.5 (28.5–52.3)	0.058
Age of onset (y)
≤40	27.3%	26.3%	29.0%	40.0%
41–64	67.6%	69.1%	64.8%	50.0%	0.403
≥65	5.1%	4.5%	6.2%	10.0%
Diabetes duration (y)	14.5 (8.0–20.0)	14.5 (8.0–21.3)	14.5 (8.0–22.3)	18.0 (9.0–29.3)	0.694
BMI (kg/m²)	25.5 (23.0–28.0)	26.9 (24.4–29.9)	26.4 (23.7–29.4)	24.0 (22.4–27.4)	0.610
HbA1c (%); (mmol/mol)	7.4 (6.8–8.2)	7.4 (6.8–8.3)	7.2 (6.8–8.0)	7.6 (7.2–8.6)	0.136
57 (51–66)	57 (51–67)	55 (51–64)	60 (55–70)
Systolic blood pressure (mmHg)	141 (129–152)	141 (129–155)	140 (129–154)	137 (129–147)	0.784
Diastolic blood pressure (mmHg)	78 (73–83)	78 (72–83)	77 (73–83)	80 (71–84)	0.812
Pulse pressure (mmHg)	63 (53–75)	63 (53–77)	62 (53–74)	63 (50–73)	0.937
Total cholesterol (mM)	4.16 (3.66–4.69)	4.17 (3.66–4.70)	4.14 (3.67–4.65)	3.97 (3.49–5.62)	0.959
HDL-C (mM)	1.31 (1.07–1.58)	1.29 (1.06–1.55)	1.40 (1.13–1.63)	1.17 (1.05–1.63)	0.025
LDL-C (mM)	2.48 (2.09–2.98)	2.49 (2.10–3.02)	2.49 (2.09–2.80)	2.10 (1.88–2.66)	0.321
Triglycerides (mM)	1.31 (0.98–1.86)	1.35 (1.03–1.90)	1.23 (0.97–1.64)	1.33 (0.98–3.22)	0.081
uACR (mg/g)	25.0 (7.0–127.3)	26.0 (8.0–135.3)	21.5 (4.8–117.0)	44.5 (6.0–271.3)	0.207
uACR (% < 30 mg/g)	54.4%	53.3%	57.4%	50.0%	0.791
Serum creatinine (μmol)	74.0 (60.0–102.3)	74.0 (60.0–101.3)	74.5 (57.0–105.0)	83.0 (69.0–127.0)	0.485
CKD-Epi-eGFR (ml/min/1.73m²)	87.9 (61.4–98.8)	87.9 (61.7–98.8)	87.9 (61.4–99.1)	61.1 (47.9–104.2)	0.739
CKD-Epi-eGFR (% ≥60 ml/min/1.73m²)	75.9%	76.1%	76.5%	60.0	0.478
MMSE Total score	29.0 (28.0–30.0)	29.0 (28.0–30.0)	29.0 (28.0–30.0)	29.0 (26.8–30.0)	0.102
(N = 579)	(N = 409)	(N = 160)	(N = 10)
RBANS Total score	101.7 (97.5–105.5)	101.5 (97.4–105.4)	102.7 (98.2–106.0)	98.8 (95.0–102.1)	0.086
Domain_immediate memory	103.7 (98.2–108.9)	103.5 (98.0–108.2)	104.5 (98.3–109.7)	104.6 (98.4–110.3)	0.651
Domain_visuospatial	102.3 (97.8–105.6)	102.3 (97.8–105.6)	102.3 (97.8–105.9)	93.4 (84.2–107.8)	0.499
Domain_language	101.9 (95.3–106.5)	100.8 (95.1–106.5)	102.7(96.2–107.8)	102.7 (98.2–104.6)	0.747
Domain_attention	101.9 (94.6–108.9)	101.8 (93.8–108.8)	102.9 (96.3–109.4)	96.1 (87.0–98.2)	0.011
Domain_delayed memory	101.1 (95.1–105.6	100.5 (94.9–105.4)	102.8 (96.4–106.2)	97.5 (91.1–104.2)	0.074
Geriatric Depression Scale	0.0 (0.0–1.0)	1.0 (0.0–2.0)	0.0 (0.0–1.0)	1.0 (0.0–1.3)	0.048
Geriatric Depression Scale (% ≥5)	3.9%	4.3%	3.1%	0.0%	0.758
Education level (% ≥7 y)	63.4%	62.7%	66.0%	50.0%	0.479
APOEɛ4 allele (% present)	16.8%	17.9%	14.2%	10.0%	0.601
Metformin usage (% Yes)	80.8%	80.6%	80.9%	90.0%	0.916
Smoking
Non-smoker	86.3%	85.4%	88.3%	90.0%
Ex-smoker	6.3%	6.5%	5.6%	10.0%	0.840
Current smoker	7.3%	7.9%	6.2%	0.0%

Continuous variables were presented as median and interquartile range (IQR). The Kruskal-Wallis test was used to compare continuous variables. Categorical variables were reported as percentages. The Fisher-Freeman-Halton exact test was used to compare categorical variables. A two-tailed value of p < 0.05 was considered as statistically significant (bold). BMI, body mass index; HbA1c, glycated hemoglobin; HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; uACR, urinary albumin-creatinine ratio; eGFR, estimated glomerular filtration rate; APOE, Apolipoprotein E.

To determine the association of PAX4 R192H genotype with cognitive impairment, multiple linear regression was performed with PAX4 R192H genotype as independent variable and RBANS total score and sub-domain scores as dependent variable while adjusting for age, gender, diabetes onset age, BMI, HbA1c, SBP, HDL-C, triglycerides, eGFR, uACR, metformin usage, smoking history, education level, GDS score, and presence of APOE ɛ4 allele. We observed that the presence of 2 risk alleles (TT genotype) was negatively associated with RBANS total score (β= –3.539, 95% CI [–6.932, –0.146], p = 0.041) as well as the attention index domain score (β= –8.457, 95% CI [–13.705, –3.209], p = 0.002) (Table 2). Comparable associations of the TT genotype with individual component test scores of the attention index domain, namely digit span (β= –9.564, 95% CI [–18.447, –0.682] p = 0.035) and coding (β= –7.350, 95% CI [–11.683, –3.018], p = 0.001) were also observed. MMSE total score (N = 579), however, did not show any significant association with CT genotype (β= 0.195, 95% CI [–0.145, 0.536], p = 0.261) or TT genotype (β= 0.071, 95% CI [–1.092, 1.234], p = 0.905) after full adjustment.

Table 2

Association between PAX4 genotype and RBANS total and sub-domain index scores in fully adjusted multiple linear regression model

PAX4 R192H genotype	Coefficient (95% CI), p
	Total score	Immediate memory	Visuospatial	Language	Attention	Delayed memory
CC	Reference	Reference	Reference	Reference	Reference	Reference
CT	0.102	0.115	–0.597	–0.075	0.518	0.551
	(–0.883, 1.088),	(–1.260, 1.490),	(–2.430, 1.235),	(–1.592, 1.442),	(–1.006, 2.043),	(–0.998, 2.100),
	p = 0.839	p = 0.870	p = 0.523	p = 0.923	p = 0.505	p = 0.486
TT	–3.539	–0.398	–5.687	–0.129	–8.457	–3.023
	(–6.932, –0.146),	(–5.130, 4.334),	(–11.993, 0.619),	(–5.349, 5.091),	(–13.705, –3.209),	(–8.355, 2.309),
	p = 0.041	p = 0.869	p = 0.077	p = 0.961	p = 0.002	p = 0.266

To further evaluate whether the association of PAX4 R192H genotype with reduced RBANS total score and attention index score is independent of diabetes onset age, we compared the results from 3 different models of multiple linear regression with different combinations of PAX4 R192H genotype and diabetes onset age (Table 3). As shown, adding diabetes onset age (Model 1) to the model containing PAX4 R192H genotype (Model 2) did not result in significant changes to the association in terms of coefficients and p-values. Diabetes onset age only (Model 3) did not result in significant association with the scores although the trend is consistently negative. Interaction terms of onset age and PAX4 R192H genotype added into the regression models did not give statistically significant results, suggesting they are independent of each other. In addition, stratified analysis was performed to determine the association of PAX4 R192H genotype with the cognitive scores in each of the onset age groups with adjustment for all relevant covariates. As shown in Table 4, significant associations were observed in the middle-aged (≤40) and elderly-aged (≥65) onset groups, with the greatest magnitude observed in the elderly-aged onset group with the lowest median diabetes duration of 5 years old. Collectively, these observations suggest that the association of PAX4 R192H genotype with total score and attention index score is independent of diabetes onset age.

Table 3

Relationship between PAX4 R192H TT genotype and diabetes onset age on RBANS total score and attention index score

Model	Coefficient (95% CI), p
	Total score	Attention
1. PAX4 R192H+onset age	–3.539 (–6.932, –0.146), p = 0.041	–8.457 (–13.705, –3.209), p = 0.002
2. PAX4 R192H only	–3.529 (–6.920,–0.139), p = 0.041	–8.515 (–13.752, –3.278), p = 0.001
3. Onset age only	–1.477 (–3.874, 0.920), p = 0.227	–0.590 (–4.319, 3.139), p = 0.757

Multiple linear regression was used to assess the association of PAX4 R192H genotype and/or diabetes onset age with RBANS total score and attention index score, adjusting for Model 1: full set of covariates including both PAX4 R192H genotype and diabetes onset age, Model 2: full set of covariates including PAX4 R192H genotype but excluding diabetes onset age, Model 3: full set of covariates including diabetes onset age but excluding PAX4 R192H genotype. ^*Coefficient shown is for diabetes onset age≤40 years old against reference of ≥65 years old.

Table 4

Association of PAX4 R192H TT genotype with RBANS total score and attention index score stratified by diabetes onset age

Diabetes onset age (y)	Coefficient (95% CI), p
	Total score	Attention
≤40 (N = 161)	–2.840 (–7.902, 2.222), p = 0.271	–7.267 (–15.380, 0.847), p = 0.079
41–64 (N = 399)	–4.287 (–9.134, –0.559), p = 0.083	–8.759 (–16.162, –1.355), p = 0.020
≥65 (N = 30)	–9.910 (–17.966, –1.854), p = 0.016	–13.403 (–26.116, –0.689), p = 0.039

DISCUSSION

In this paper, we report a novel observation that PAX4 R192H genotype is associated with reduced cognitive function in a Singaporean cohort of Chinese patients with T2DM. For patients with TT genotype, the RBANS total score and attention index score were reduced by 3.5 and 8.5 points (0.2 SD and 0.6 SD) respectively. This suggests that the PAX4 R192H genetic variation is associated with cognitive decrements specifically in areas related to executive functions for T2DM patients. As there is no existing evidence to explain the selective association of PAX4 R192H genetic variation with attention, it would be useful to corroborate this observation using brain imaging to further elucidate the specific pathophysiological pathways underlying such cognitive decrements. No significant associations between PAX4 R192H genetic variation and MMSE were observed. This is not unexpected as MMSE has been reported to be less sensitive towards mild cognitive impairment [27, 42].

As PAX4 R192H genetic variation is also known to associate with earlier onset of T2DM in Chinese, it is of interest to determine if this association with cognitive impairment was indirectly caused by the earlier T2DM onset resulting in longer disease duration and increased likelihood of sub-optimal glycemic control resulting in secondary complications such as cognitive impairment [43]. Our analyses support the independent association of PAX4 R192H genotype with cognitive function. Interestingly, our stratified analyses revealed that the association was most prominent in the elderly-onset group with the shortest diabetes duration. This calls to attention the possibility of multiple roles of PAX4 in both T2DM and also cognitive function. While PAX4 has been implicated in beta cell function, there has been no prior reports to suggest a direct role in cognitive function. Nonetheless, our observations contribute to growing evidence pointing to PAX4’s potential involvement in the neurocognitive domain [17, 18]. Given that PAX4 regulates expression of Grb2 and NOX4 which are the effector proteins shared by T2DM and AD pathways, further functional studies to decipher the relationships between R192H variation and these 2 key players would be important to establish the mechanistic link between PAX4 R192H genotype with T2DM and cognitive function. In addition, it would be of interest to explore the interaction of PAX4 R192H T allele with APOE ɛ4 allele in a larger cohort sufficiently powered to determine their contribution to the risk of cognitive impairment. Our stratified analyses have demonstrated that the association of this variant with cognitive impairment is independent of its association with earlier onset of T2DM, suggesting its potential dual roles in T2DM and cognitive impairment. Therefore, it would be interesting to determine if indeed the association of this PAX4 variant with cognitive impairment is also present in the healthy population in future studies.

Our study has several limitations. First, since this is a cross-sectional study, we are unable to establish the causal contribution of PAX4 R192H genotype to cognitive function. Secondly, we were unable to corroborate our findings with brain imaging data as advanced magnetic resonance imaging and positron emission tomography techniques [44 –46] required to visualize mild cognitive impairments were beyond the accessibility of this study. Thirdly, concurrent clinical diagnosis for mild cognitive impairment was not available for our cohort. Finally, our findings are preliminary and would benefit from validation in an alternative cohort to further establish the association.

Conclusion

In summary, we have shown that PAX4 R192H TT genotype is associated with increased cognitive impairment specifically in the attention domain. This novel but preliminary finding suggests the potential use of PAX4 R192H genotype for screening and risk assessment of Chinese T2DM patients for early intervention and management of diabetes-associated cognitive dysfunction.

Footnotes

ACKNOWLEDGMENTS

We would like to thank the patients for their participation and doctors for their referrals without which this study would not have been possible.

Su Chi Lim is supported by the Singapore Ministry of Health’s National Medical Research Council under its Clinician Scientist Award (MOH-000714-01). The SMART2D cohort is supported by the Singapore Ministry of Health’s National Medical Research Council under its CS-IRG (MOH-000066).

Authors’ disclosures available online ().

References

Biessels

, Staekenborg

, Brunner

, Brayne

, Scheltens

(2006) Risk of dementia in diabetes mellitus: A systematic review. Lancet Neurol 5, 64–74.

Biessels

, Strachan

, Visseren

, Kappelle

, Whitmer

(2014) Dementia and cognitive decline in type 2 diabetes and prediabetic stages: Towards targeted interventions. Lancet Diabetes Endocrinol 2, 246–255.

Biessels

, Despa

(2018) Cognitive decline and dementia in diabetes mellitus: Mechanisms and clinical implications. Nat Rev Endocrinol 14, 591–604.

Majumder

, Chanda

, Das

, Singh

, Chakrabarti

, Jana

, Mukhopadhyay

(2021) A nexus of miR-1271, PAX4 and ALK/RYK influences the cytoskeletal architectures in Alzheimer’s disease and type 2 diabetes. Biochem J 478, 3297–3317.

Sosa-Pineda

(2004) The gene Pax4 is an essential regulator of pancreatic beta-cell development. Mol Cells 18, 289–294.

, Wang

, Song

, Deng

, Chen

, Ouyang

, Lai

, Li

(2018) Generation and phenotype identification of PAX4 gene knockout rabbit by CRISPR/Cas9 system. G3 (Bethesda) 8, 2833–2840.

Brun

, Gauthier

(2008) A focus on the role of Pax4 in mature pancreatic islet beta-cell expansion and survival in health and disease. J Mol Endocrinol 40, 37–45.

Panneerselvam

, Kannan

, Mariajoseph-Antony

, Prahalathan

(2019) PAX proteins and their role in pancreas. Diabetes Res Clin Pract 155, 107792.

Fuchsberger

, Flannick

, Teslovich

, Mahajan

, Agarwala

, Gaulton

, Ma

, Fontanillas

, Moutsianas

, McCarthy

, et al. (2016) The genetic architecture of type 2 diabetes. Nature 536, 41–47.

10.

Sujjitjoon

, Kooptiwut

, Chongjaroen

, Semprasert

, Hanchang

, Chanprasert

, Tangjittipokin

, Yenchitsomanus

, Plengvidhya

(2016) PAX4 R192H and P321H polymorphisms in type 2 diabetes and their functional defects. J Hum Genet 61, 943–949.

11.

Cheung

, Tang

, Xu

, Lee

, Au

, Xu

, Fong

, Kwok

, Chow

, Woo

, Yuen

, Hai

, Jin

, Cheung

, Tan

, Cherny

, Zhu

, Thomas

, Cheng

, Jiang

, Lam

, Tse

, Sham

, Lam

(2017) Exome-chip association analysis reveals an Asian-specific missense variant in PAX4 associated with type 2 diabetes in Chinese individuals. Diabetologia 60, 107–115.

12.

Cho

, Jang

, Chung

, Kim

(2021) Exome chip analysis of 14,026 Koreans reveals known and newly discovered genetic loci associated with type 2 diabetes mellitus. Diabetes Metab J 45, 231–240.

13.

Ang

, Lim

, Tan

, Fong

, Kon

, Lian

, Subramanium

, Sum

(2016) A preliminary study to evaluate the strategy of combining clinical criteria and next generation sequencing (NGS) for the identification of monogenic diabetes among multi-ethnic Asians. Diabetes Res Clin Pract 119, 13–22.

14.

Ang

, Tan

CSH

, Wang

, Dorajoo

, Fong

JCW

, Kon

WYC

, Lian

, Ang

, Rahim

, Jeevith

, Lee

SBM

, Tang

, Subramanium

, Sum

, Liu

, Lim

(2019) PAX4 R192H is associated with younger onset of Type 2 diabetes in East Asians in Singapore. J Diabetes Complications 33, 53–58.

15.

Kooptiwut

, Plengvidhya

, Chukijrungroat

, Sujjitjoon

, Semprasert

, Furuta

, Yenchitsomanus

(2012) Defective PAX4 R192H transcriptional repressor activities associated with maturity onset diabetes of the young and early onset-age of type 2 diabetes. J Diabetes Complications 26, 343–347.

16.

Hosoe

, Suzuki

, Miya

, Kato

, Tsunoda

, Okada

, Horikoshi

, Shojima

, Yamauchi

, Kadowaki

(2021) Structural basis of ethnic-specific variants of PAX4 associated with type 2 diabetes. Hum Genome Var 8, 25.

17.

Volodin

, Kosti

, Goldberg

, Cohen

(2017) Myofibril breakdown during atrophy is a delayed response requiring the transcription factor PAX4 and desmin depolymerization. Proc Natl Acad Sci U S A 114, E1375–E1384.

18.

Kong

, Lei

, Zhang

, Li

, Zhao

, Zhang

(2018) Integrated microarray analysis provided a new insight of the pathogenesis of Parkinson’s disease. Neurosci Lett 662, 51–58.

19.

Levey

, Stevens

, Schmid

, Zhang

, Castro

, Feldman

, Kusek

, Eggers

, Van Lente

, Greene

, Coresh

(2009) A new equation to estimate glomerular filtration rate. Ann Intern Med 150, 604–612.

20.

Sheikh

, Yesavage

(1986) Geriatric Depression Scale (GDS): Recent Evidence and Development of a Shorter Version. In Clinical Gerontology: A Guide to Assessment and Intervention, Brink

, ed. Haworth Press, Binghamton, pp. 165–173.

21.

, Feng

, Yap

, Lee

, Tan

, Winblad

(2014) Long-term metformin usage and cognitive function among older adults with diabetes. J Alzheimers Dis 41, 61–68.

22.

Randolph

, Tierney

, Mohr

, Chase

(1998) The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): Preliminary clinical validity. J Clin Exp Neuropsychol 20, 310–319.

23.

Lim

, Collinson

, Feng

, Ng

(2010) Cross-cultural application of the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): Performances of elderly Chinese Singaporeans. Clin Neuropsychol 24, 811–826.

24.

Collinson

, Fang

, Lim

, Feng

, Ng

(2014) Normative data for the repeatable battery for the assessment of neuropsychological status in elderly Chinese. Arch Clin Neuropsychol 29, 442–455.

25.

, Nyunt

MSZ

, Shuvo

, Eng

, Yap

, Hee

, Chan

, Scherer

(2018) The neighborhood built environment and cognitive function of older persons: Results from the Singapore Longitudinal Ageing Study. Gerontology 64, 149–156.

26.

Folstein

, Folstein

, McHugh

(1975) “Mini-mental state”: A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12, 189–198.

27.

Feng

, Chong

, Lim

, Ng

(2012) The Modified Mini-Mental State Examination test: Normative data for Singapore Chinese older adults and its performance in detecting early cognitive impairment. Singapore Med J 53, 458–462.

28.

, Niti

, Chiam

, Kua

(2007) Ethnic and educational differences in cognitive test performance on mini-mental state examination in Asians. Am J Geriatr Psychiatry 15, 130–139.

29.

Levy

(1994) Aging-associated cognitive decline. Working Party of the International Psychogeriatric Association in collaboration with the World Health Organization. Int Psychogeriatr 6, 63–68.

30.

Jancke

(2018) Sex/gender differences in cognition, neurophysiology, and neuroanatomy. F1000Res 7, F1000FacultyRev-805.

31.

Roberts

, Geda

, Knopman

, Christianson

, Pankratz

, Boeve

, Vella

, Rocca

, Petersen

(2008) Association of duration and severity of diabetes mellitus with mild cognitive impairment. Arch Neurol 65, 1066–1073.

32.

Michaud

, Siahpush

, Farazi

, Kim

, Yu

, Su

, Murman

(2018) The association between body mass index, and cognitive, functional, and behavioral declines for incident dementia. J Alzheimers Dis 66, 1507–1517.

33.

, Tan

, Shen

, Xu

, Hou

, Dong

, Tan

, Yu

(2020) Blood pressure and risks of cognitive impairment and dementia: A systematic review and meta-analysis of 209 prospective studies. Hypertension 76, 217–225.

34.

Hottman

, Chernick

, Cheng

, Wang

, Li

(2014) HDL and cognition in neurodegenerative disorders. Neurobiol Dis 72 Pt A, 22–36.

35.

Parthasarathy

, Frazier

, Bettcher

, Jastrzab

, Chao

, Reed

, Mungas

, Weiner

, DeCarli

, Chui

, Kramer

(2017) Triglycerides are negatively correlated with cognitive function in nondemented aging adults. Neuropsychology 31, 682–688.

36.

Wang

, Zhang

, Liu

, Wang

(2010) Level of kidney function correlates with cognitive decline. Am J Nephrol 32, 117–121.

37.

Takae

, Hata

, Ohara

, Yoshida

, Shibata

, Mukai

, Hirakawa

, Kishimoto

, Tsuruya

, Kitazono

, Kiyohara

, Ninomiya

(2018) Albuminuria increases the risks for both Alzheimer disease and vascular dementia in community-dwelling Japanese elderly: The Hisayama Study. J Am Heart Assoc 7, e006693.

38.

Sabia

, Elbaz

, Dugravot

, Head

, Shipley

, Hagger-Johnson

, Kivimaki

, Singh-Manoux

(2012) Impact of smoking on cognitive decline in early old age: The Whitehall II cohort study. Arch Gen Psychiatry 69, 627–635.

39.

Lovden

, Fratiglioni

, Glymour

, Lindenberger

, Tucker-Drob

(2020) Education and cognitive functioning across the life span. Psychol Sci Public Interest 21, 6–41.

40.

Thompson

, Jones

(2020) Depression screening in cognitively normal older adults: Measurement bias according to subjective memory decline, brain amyloid burden, cognitive function, and sex. Alzheimers Dement (Amst) 12, e12107.

41.

Liu

, Liu

, Kanekiyo

, Xu

, Bu

(2013) Apolipoprotein E and Alzheimer disease: Risk, mechanisms and therapy. Nat Rev Neurol 9, 106–118.

42.

Baek

, Kim

, Park

, Kim

(2016) The validity and reliability of the Mini-Mental State Examination-2 for detecting mild cognitive impairment and Alzheimer’s disease in a Korean population. PLoS One 11, e0163792.

43.

Reijmer

, van den Berg

, Ruis

, Kappelle

, Biessels

(2010) Cognitive dysfunction in patients with type 2 diabetes. Diabetes Metab Res Rev 26, 507–519.

44.

, Sun

, Yan

, Hu

, Nan

, Yang

, Liu

, Wang

, Cui

(2016) Multimodal MRI for early diabetic mild cognitive impairment: Study protocol of a prospective diagnostic trial. BMC Med Imaging 16, 50.

45.

Biessels

, Reijmer

(2014) Brain changes underlying cognitive dysfunction in diabetes: What can we learn from MRI? Diabetes 63, 2244–2252.

46.

Nugent

, Potvin

, Cunnane

, Chen

, Duchesne

(2020) Associating type 2 diabetes risk factor genes and FDG-PET brain metabolism in normal aging and Alzheimer’s disease. Front Aging Neurosci 12, 580633.