The parahippocampal complex seems to be a critical region in the pathophysiology of type 2 diabetes mellitus (T2DM), mild cognitive impairment (MCI), and dementia (Zilliox et al., 2016; Bello-Chavolla et al., 2019). Memory-related impairments are found in this population (Zilliox et al., 2016), along with evidence of a link between T2DM and MCI and/or dementia (Bello-Chavolla et al., 2019). Nonetheless, a clear understanding of the pathophysiological mechanisms in T2DM and the development of cognitive deterioration and ultimately dementia is still far from understood.

Pathophysiological mechanisms linking T2DM and Alzheimer’s dementia (AD) and brain atrophy have been suggested to involve insulin resistance and hyperglycemia (Bello-Chavolla et al., 2019; Li et al., 2021). However human studies investigating the role of metabolic control in brain atrophy are mainly cross-sectional, preventing a direct test of a temporal link between glycemic control and brain structural integrity (Gold et al., 2007; Moran et al., 2013; Hirabayashi et al., 2016; Ogama et al., 2018). Furthermore, longitudinal studies on brain atrophy so far were mostly dedicated to investigating total brain volume and/or hippocampus alone (Debette et al., 2011; Callisaya et al., 2019). In addition, no associations with hyperglycemia have been particularly explored. Knowing whether impaired metabolic control leads to irreversible brain atrophy in T2DM, especially in memory-related regions, is important for the implementation of an early and timely management of metabolic control.

In this study, we report a longitudinal study over 7 years to investigate whether initial higher levels of glycated hemoglobin (HbA_1C) in T2DM, are associated with a long-term decrease in grey matter volume in three core brain regions involved in memory processing – hippocampus, parahippocampal gyrus, and fusiform gyrus and if these changes revert upon improvement of metabolic control.

We found a negative statistically significant correlation between baseline HbA_1C and the longitudinal changes in grey matter volume in the T2DM group in three ROIs – the right parahippocampal gyrus (rs = −0.665, p = 0.026, n = 12, power = 0.889), the left fusiform gyrus (rs = −0.703, p = 0.016, n = 12, power = 0.937) and the right fusiform gyrus (rs = −0.689, p = 0.019, n = 12, power = 0.921; see Figure 1). We also found a marginal negative correlation between HbA_1C and the changes in grey matter volume in the right hippocampus (rs = −0.580, p = 0.062, n = 12, power = 0.742). No statistically significant correlation was found for the left hippocampus (rs = −0.446, p = 0.149, n = 12) and the left parahippocampal gyrus (rs = −0.495, p = 0.121, n = 12) in the T2DM, group. As expected, we found no statistically significant correlations between the control group grey matter volume and HbA_1C in any of the ROIs. Importantly, no correlation was found in T2DM between the longitudinal changes in HbA_1C and the longitudinal changes in grey matter volume of the regions whose grey matter volume which proved to be initially correlated with HbA1_c, showing that the improvement in metabolic control was not sufficient to revert changes in grey mattern integrity. This result is also corroborated by whole-brain VBM analysis.

We were able to replicate the ROI approach by using whole-brain VBM linear regression analysis. We found for the T2DM, group three different main clusters in the regions under study showing a longitudinal negative correlation between changes in grey matter volume and the HbA_1C at baseline: (a) one cluster, with 1,143 voxels (26, −2, −50), whose peak voxel was in the right inferior temporal gyrus (rITG) and included voxels in right fusiform gyrus and right parahippocampal gyrus; (b) a cluster, composed of 544 voxels (32, −11, −27), with peak voxel on the right parahippocampal gyrus and (c) another cluster, with 598 voxels, on left parahippocampal gyrus (−29, −14, −30). This latter cluster had also voxels in the left fusiform gyrus, being in congruence with our ROI analysis taken from the AAL atlas, which independently showed a correlation between HbA_1C and changes in grey matter volume in left fusiform gyrus (see Figure 2). Importantly, the more anterior part of the right parahippocampal gyrus cluster included voxels in right hippocampus. This was not the case for the left parahippocampal gyrus. This analysis provides statistical support even for the right hippocampus for the ROI analysis taken from the AAL atlas. Additionally, the cluster from the whole-brain VBM analysis revealed voxels in left parahippocampal gyrus. No suprathreshold clusters were found for the control group in the whole-brain linear regression analysis. The description of other regions, obtained for the linear regression analysis for the T2DM group, are depicted in Supplementary Table S1.

This study provides evidence of a negative association between grey matter volume loss along ~7 years in medial temporal lobe memory encoding regions (right hippocampus, both parahippocampal gyrus and both FGs), and initial levels of HbA_1C in a T2DM, sample without MCI. This suggests that impaired glycemic control (higher levels of HbA_1C) might lead to long term atrophy in memory related regions. Importantly, longitudinal changes in HbA_1C did not impact longitudinal modifications in grey volume suggesting that the atrophy found in these regions might be irreversible. This demonstrates the importance of early and timely glycemic control to prevent such irreversible changes. The ideia that irreversible changes in brain morphometrical properties might occur in T2DM in relation with hyperglycemia is corroborated by studies that focused on the complex deleterious effect of hyperglycemia in the brain. Possible mechanisms behind brain atrophy may include effects in addition to direct vascular impairments (Williams et al., 1998; Biessels et al., 2006; Moulton et al., 2015). We suggest that subtle alterations in neurovascular coupling is indeed suggested for people with T2DM (Duarte et al., 2015, 2023) or even direct damage to neurons that can also occur prior to vascular damage (Reis et al., 2014) may play a role. The study by Reis et al. (2014) indeed found that both people with type 1 diabetes, with and without blood-retinal barrier’s dysfuntion, sustained significant differences in visual functions suggesting that neuronal changes in retina occur prior to vasculopathy. Furthermore, augmented oxidative stress, causing inflammation is also another mechanisms behind brain atrophy. Dysfunctional insulin signaling and elevated blood glucose levels lead to heightened expression and activation of NF-κB, a pivotal participant in the inflammatory cascade and a regulator of apoptosis (Sima et al., 2009; Moulton et al., 2015). Moreover, advanced glycation end products, also leads to inflammation, as a consequence of chronic activation of microglia, leading to neuronal and glial apoptosis (Moulton et al., 2015; Sun et al., 2020; Yu et al., 2023). We favour the view that neurodegenerative processes underlying T2DM’s pathophysiology might also leads to the presence of brain atrophy, where toxic glucose levels have a harmful impact on various biochemical processes associated with tau integrity, promoting tau cleavage and phosphorylation (Kubis-Kubiak et al., 2019; Huang et al., 2020). Additionally, elevated glucose levels influence the production and degradation of Aβ (Macauley et al., 2015; Kubis-Kubiak et al., 2019). There is evidence indicating that certain Aβ products undergo predominant degradation through the proteasome pathway, which is significantly destabilized by increased glucose levels (Kubis-Kubiak et al., 2019; Patil et al., 2020). Nonetheless, different regions seem to have specific vulnerabilities and different mechanisms might underly atrophy in particular regions.

Human studies investigating the role of metabolic control in brain structural dysfunction are overall cross-sectional. In Ogama et al. (2018), postprandial hyperglycemia was associated with lower brain volume. Another study found that 2-h postload glucose levels were associated with lower total brain volume and hippocampus volume (Hirabayashi et al., 2016). Studies with animal models have found an association between hyperglycemia and memory loss in relation with signs of neuroinflammation (Wanrooy et al., 2018; Rom et al., 2019) namely in the hippocampus (Wanrooy et al., 2018). Several studies focused on the association between hyperglycemia, oxidative stress, and brain atrophy (Ogama et al., 2018), and others investigated the effect of hyperglycemia in microvascular structures (Stehouwer, 2018). Indeed, brain structures are fairly vulnerable to microvascular disturbances since brain microvessels have low resistance and sensitive auto-regulatory processes, being more vulnerable to phenomena like augmented pulsatile load that occurs with elevated glucose levels (Stehouwer, 2018). Nonetheless, a recent study suggested that brain atrophy in hippocampal subfields are associated with cognitive impairment in this disease rather than microstructural lesions, where HbA_1c levels were negatively correlated with hippocampal atrophy but not with the burden of white matter hyperintensities (Zhang et al., 2021). Furthermore, previous randomized clinical trial testing the effects of intensive glycemic control in T2DM participants versus a standard glycemic control on the longitudinal changes in brain atrophy showed that intensive glycemic treatment decelerated the volume loss in regions like bilateral superior temporal gyrus, left pre and postcentral gyri, in bilateral fronto-orbital gyrus and bilateral frontal cingulate regions (Erus et al., 2015). Importantly, no significant changes were found in medial temporal gyrus and in particular in hippocampus and parahippocampal regions (Erus et al., 2015) which is in agreement with our study.

In overall, several possible mechanisms might explain the association between hyperglycemia and an augmented loss of brain volume inT2DM, from microvascular complications to augmented toxicity and neuroinflamation, leading to increased dysfunction. This is the first longitudinal study to provide evidence for a possible irreversible long term loss in a T2DM, sample without MCI. The association between long term brain atrophy and early HbA_1C changes in core regions of the inferior and medial temporal lobes, such hippocampus and parahippocampal gyrus, is of particular interest given the hypothesized association between T2DM, and AD (Bello-Chavolla et al., 2019). Evidence that initially impaired metabolic control in T2DM, patients with no MCI may be associated with irremediable long term brain atrophy in key structures in the medial temporal lobe involved in memory processes observed within 7 years, consolidates the idea that pathological events precede cognitive deterioration. Furthermore, sex differences have been suggested to underlie cognitive deterioration in T2DM, with female having higher risk for accelerated cognitive deterioration (Verhagen et al., 2022). This variable was matched across our groups.

The present study provides evidence for a possible irreversible deleterious effect of poor glycemic control and the loss of grey matter volume in participants with T2DM, without mild cognitive impairment. Since participants neither fulfilled criteria for aMCI nor showed different cognitive profile relative to controls in any cognitive measures we suggest that structural atrophy might proceded cognitive deterioration. Further monitoring of the cognitive status of these participants is needed to explore if the consequences of poor glycemic control on the longitudinal changes in structural integrity in memory-related regions have a deleterius effect in cognition. The results highlight the importance of an early intensive control of glycemia and overall intensive management of T2DM.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving humans were approved by Comissão de Ética da Faculdade de Medicina da Universidade de Coimbra. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

NC: Conceptualization, Formal analysis, Investigation, Writing – original draft. JC: Investigation, Writing – review & editing. JD: Investigation, Methodology, Writing – review & editing. CM: Investigation, Validation, Writing – review & editing. HQ: Investigation, Writing – review & editing. LG: Investigation, Validation, Writing – review & editing. FO: Investigation, Supervision, Writing – review & editing. MC-B: Conceptualization, Data curation, Funding acquisition, Investigation, Project administration, Resources, Supervision, Validation, Writing – review & editing.