Repurposing Antidiabetic Medications for Parkinson’s Disease: Focus on Biomarker Strategies on Disease Modification

1. Introduction

Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterized by loss of dopaminergic neurons within the substantia nigra thought to be predominantly caused by accumulation of α-synuclein (αSyn) pathology. Gradual depletion of dopaminergic neurons results in impaired motor activity resulting in resting tremor, bradykinesia, rigidity, and postural instability [1]. In addition, non-motor symptoms, i.e. mood disorders, dysautonomia, sleep disturbances, cognitive impairment, and GI complications, can occur up to 20 years preceding motor manifestation of the disease [2]. Despite tremendous advances in symptomatic treatment over the last 50 years, there are currently no FDA approved disease-modifying therapies capable of slowing or halting neurodegeneration. PD continues to remain incurable despite significant progress made in its management.

Over the years, there has been a focus on advancing diagnostic testing to better diagnose PD early, accurately, and to initiate appropriate treatment. The landscape has included tests such as dopamine transporter imaging (DaTscan), olfactory tests, polysomnography, gastric emptying assay, αSyn skin biopsy, cerebrospinal fluid (CSF) αSyn seed amplification assay, and genetic testing [3,4]. In conjunction, there has been development in biochemical biomarkers that facilitate the early diagnosis of PD. Repurposing therapies that target these biochemical biomarkers may help not only in earlier diagnosis but also better outcomes for PD patients.

There has been an emerging link between type 2 diabetes mellitus (T2DM) and PD. T2DM is a metabolic disease defined by insulin resistance and hyperglycemia. While distinct in their primary clinical manifestations, increasing evidence suggests that metabolic dysfunction, impaired insulin signaling, oxidative stress, blood-brain barrier dysfunction, and mitochondrial dysregulation contribute to PD pathophysiology, linking central nervous system degeneration to systemic metabolic disturbances [4,5]. Accordingly, antidiabetic medications have shown neuroprotective effects in preclinical and early human studies. Therefore, there has been an increasing interest in translational research looking at how antidiabetic medication alters biomarkers of neurodegeneration in PD.

In this review, we will explore the recent advances in anti-diabetic medications (biguanides, sulfonylureas, thiazolidinediones, dipeptidyl peptidase-4 inhibitors, sodium glucose cotransporter 2 inhibitors and Glucagon-like peptide-1 receptor agonists) that have been linked with disease modification on key PD biomarkers and their proposed outcomes in preclinical and clinical models of PD.

2. Relationship Between T2DM and PD

T2DM has been shown to increase the risk of developing PD and accelerate its progression. In multiple meta-analyses, individuals with T2DM were associated with increased risk for PD with relative risk ranging from 1.19-1.27 [6,7,8]. Cohort studies from Korea have shown that T2DM duration was associated with higher PD risk [9]. Another large cohort study from the UK has shown that there were elevated rates of PD following T2DM (hazard ratio 1.32, 95% confidence interval 1.29-1.35; p < 0.001) and that the increase was greater in individuals with complicated T2DM [10]. Table 1 summarizes the relationship and interplay between these two diseases.

Mechanistically, there is extensive literature that shows pathophysiological similarities between T2DM and PD. First, insulin resistance and dysregulated insulin signaling are central to both diseases. In T2DM, misfolding of amylin may contribute to pathology due to loss of pancreatic beta cells which in turn leads to peripheral insulin resistance, then leads to hyperglycemia and β-cell dysfunction [11]. Amylin has been suggested to be involved in PD by interacting with αSyn, but whether it contributes αSyn aggregation in PD has not been studied [12]. In PD, insulin resistance occurs in the central nervous system and contributes to dopaminergic neuronal degeneration [11,12,13]. Insulin is thought to influence nervous system development, acts as neuroprotector by controlling apoptosis and ischemia [14], and insulin metabolism impairment is thought to lead to cognitive impairment and increase risk of neurodegenerative disorders.

Secondly, chronic inflammation and activation of the nuclear factor kappa B (NF-κB)/ nod-like receptor pyrin 3 (NLRP3) inflammasome axis are implicated in both T2DM, where hyperglycemia-induced oxidative stress triggers systemic inflammation, and in PD where neuroinflammation is enhanced leading to increased αSyn aggregation and further driving dopamine neuron loss [15,16,17]. NF-κB is a DNA binding protein required for transcription of many pro-inflammatory cytokines and chemokines. NF-κB is activated by reactive oxygen species (ROS) and pro-inflammatory cytokines, thus chronic low-grade inflammatory disorders such as T2DM result in higher levels of activated NF-κB [18]. In PD, NF-κB is thought to cause pathogenesis through induction of inflammation-mediated degeneration of dopaminergic neurons, and this has been demonstrated in a mouse model when inhibiting NF-κB prevented degeneration of dopaminergic neurons [19]. αSyn can also trigger activation of NF-κB and further release pro-inflammatory cytokines which further damages dopaminergic neurons [20]. NLRP3 inflammasome is involved with activation of caspase-1 and maturation of IL-1β and IL-18, which trigger and coordinate release of cytokines. In T2DM, NLRP3 inflammasome functions as a sensor for metabolic stress [21], can reduce glucose tolerance through modulation of gut microbiota [22], and may induce pancreatic β-cell pyroptosis [23]. In PD, NLRP3 inflammasome can also cause degeneration of dopaminergic neurons by induction of pyroptosis, can cause αSyn aggregation [24] but also further inflammatory responses by being triggered by αSyn aggregation [25,26]. Connecting T2DM to PD, chronic hyperglycemia is shown to increase αSyn deposition in pancreatic β cells, and trigger degeneration of dopaminergic neurons in PD [27].

Thirdly, both T2DM and PD have shown mitochondrial dysfunction, ferroptosis, and oxidative stress especially in β-cells and dopaminergic neurons which lead to cell death and disease progression [16,17,28]. Mitochondrial dysfunction in T2DM is thought to be a result of decreased ATP levels likely secondary to insulin dysfunction and increased vulnerability to oxidative stress [29]. PD is a disease that affects the aging brain, and aging results in mitochondrial dysfunction [30]. Specifically, loss of the mitochondrial complex I causes cell death in dopaminergic neurons, and this has been observed in mouse models in T2DM [31,32].

Fourthly, both T2DM and PD show similar patterns of impaired protein handling and aggregation. T2DM is characterized by islet amyloid polypeptide (IAPP) aggregation, while PD features αSyn aggregation. IAPP is a 37-amino acid peptide produced by pancreatic β-cells in response to hyperglycemia. When IAPP aggregate and form fibrils it disrupts β-cell function. Studies have shown that αSyn promotes IAPP aggregation [33]. On the contrary, evidence in animal models, including mice and cynomolgus monkeys, shows that IAPP can act as a trigger for αSyn pathology and accelerates its accumulation and neurotoxicity in dopaminergic neurons [34,35,36]. Speculation is that both proteins interact with one another in the pancreas as well as within the CNS after blood-brain barrier breakdown in elderly populations.

3. Biochemical Biomarkers in PD

Vijiaratnam et al. define a biomarker as “an objectively measured and evaluated characteristic from any substance, structure or process that can be measured in the body or its products as an indicator of normal biological or pathogenic processes, or pharmacologic responses to a therapeutic intervention” [37]. Therefore, an ideal biomarker should be reliable across the population and accessible through clinical samples. Reliable biomarkers become useful when assessing disease progression and response to interventions.

In PD, the landscape of biomarker analysis is vast and includes clinical, imaging, genetic and biochemical markers. For the purpose of this review, we will focus on the biochemical biomarkers that have been involved in the pathogenesis of PD. Biochemical biomarkers are collected from body fluids or tissue biopsies. Four classes of potential biochemical biomarkers that have been studied in PD include αSyn, markers of neuroinflammation, markers of insulin resistance, and markers of synaptic degeneration.

In PD, one of the earliest biomarkers that was evaluated was αSyn, the pathological protein that has been implicated in the substantia nigra resulting in dopaminergic neurodegeneration. Tests, evaluating total, phosphorylated and oligomeric αSyn in CSF, plasma, saliva, and skin samples have been developed [38]. CSF αSyn seeding amplification assay is highly accurate in differentiating PD from healthy controls when studied in a large and well characterized cohort sample but showed variability in LRRK2 variant population [39]. Commonly used in neurology clinics, the skin biopsy detection assay detected cutaneous phosphorylated αSyn with high accuracy in patients with PD and PD plus syndromes making this a minimally invasive technique [40]. Moreover, phosphorylated αSyn, specifically Ser-129p-αSyn, appears to be most specific for disease presence and severity. Unfortunately, none of the αSyn samples have been able to predict disease progression [37,41,42].

Markers of neuroinflammation (GFAP, complement levels, TNF, CRP, interleukin levels), including mitochondrial markers (DJ-1, PPAR-γ), and lysosomal dysfunction (glucocerebrosidase activity, β hexosaminidase, cathepsin D) have shown mixed results and have not been consistent across differentiating from atypical PD syndromes and measuring progression of PD [37]. In addition, inflammatory cytokine levels are elevated in a wide range of conditions.

Due to the increased risk of PD in T2DM patients, markers of insulin resistance have been considered [43]. Specifically, elevated insulin-receptor substrate-1 (IRS-1) phosphorylation, representing attenuated insulin signaling, has been found in PD patients with worsening tremor [44]. Another marker of insulin resistance is increasing hemoglobin A1c levels. Elevated hemoglobin A1c levels were shown to be associated with motor and cognitive severity and axonal damage in PD [45,46]. As discussed earlier, plasma IAPP levels can be an indirect measure for IAPP deposition withing the CNS which can implicate the burden of αSyn aggregation and disease burden in PD [47].

Markers of synaptic degeneration (SNAP25, βsynuclein, GAP43, Contactin-1) and axonal damage (NfL) have shown moderate predictability of severity of motor and cognitive progression when combined with clinical and disease specific fluid biomarkers [37,48]. Specifically, NfL shows promise for discriminating between PD and PD plus syndromes [48,49].

Several of these biochemical biomarkers associated with PD are affected by antidiabetic medications [38].

4. Antidiabetic Medications and Their Interactions on PD Biomarkers

Biguanides

Metformin, a drug from the biguanide class, is a relatively low-cost medication with minimal side effects, which makes it the preferred first-line diabetic treatment worldwide. Metformin stabilizes an individual’s glycemia by inhibiting hepatic gluconeogenesis, promoting peripheral glucose uptake, and increasing insulin sensitivity. Multiple studies from cellular and animal PD models have shown that metformin improves lifespan, reduces dopaminergic neuron loss, and improves motor deficit recovery [50].

Preclinically, these neuroprotective effects are thought to be predominantly produced by activation of autophagy via AMPK signaling, which promotes degradation of misfolded proteins and damaged organelles [50,51,52]. This in turn can inhibit αSyn phosphorylation at Ser129 and reduces accumulation by directly interacting with αSyn monomers [50,53,54]. In MPTP mouse models, metformin use has shown reductions in oxidative stress and mitochondrial dysfunction thus improving neuronal survival and function [50,55,56]. Metformin use has also been shown to upregulate neurotrophic factors such as brain derived neurotropic factor (BDNF) and glial cell-line derived neurotrophic factor (GDNF) which supports dopaminergic neuron health [53,57,58], while suppressing elevated levels of neuroinflammatory markers and glial activation in PD [56,59]. However, metformin appears to also have neuroprotective effects independent of AMPK activation as shown in a study with MPTP-induced AMPK knock out mice treated with metformin [59].

Clinical outcomes focusing on improvement in motor and cognitive impairment showed mixed benefits [60,61,62]. However, a recent randomized pilot study found that there was no significant difference in motor outcomes between metformin and control groups after treatment, though metformin significantly reduced neuroinflammatory and pathological biomarkers [57]. Future clinical trials will need to address different doses of metformin and its effect on αSyn detected through CSF or through skin biopsy assays in the PD population as well as motor or non-motor outcomes to draw direct correlations between disease pathology, clinical status and medication use.

However, metformin may also negatively affect PD neuropathology. In vivo and in vitro studies have found that chronic metformin use can reduce microglial activation and increase dopaminergic damage in response to MPTP [63] but also worsen cognition due to decrease in neurotrophic factors [64]. Clinically, chronic metformin use has shown no neuroprotective effects [61].

Sulfonylureas

Sulfonylureas (glipizide, glimepiride, glyburide) are another widely prescribed antidiabetic medication typically prescribed after metformin. They act by primarily blocking ATP-sensitive potassium channels in the pancreatic beta cells which stimulate insulin secretion thus lowering blood glucose.

There has not been promising data to suggest that sulfonylureas are effective at disease modification in PD [65]. One study has shown that it may even risk of PD [66]. No clinical trials of sulfonylureas use in PD were found.

Thiazolidinediones (TZDs)

Thiazolidinediones (pioglitazone and rosiglitazone) are a class of peroxisome proliferator-activated receptor gamma (PPAR-γ) agonists that improve insulin sensitivity and decrease blood glucose levels and are thus used for treatment in diabetes.

Preclinical data have shown potential modification of inflammatory effects through inhibition of microglia and astrocytes as well as production of proinflammatory cytokines and nitric oxide [67]. Several in vivo PD models showed neuroprotective benefits of TZDs, specifically their role in facilitating hippocampal neurogenesis, improving dopamine survivability, and improving motor performance [68,69,70].

Clinically, retrospective cohort and meta-analysis studies have shown a potential neuroprotective effect of TZD in PD and suggest that it may even reduce the incidence and risk of PD progression [71,72,73]. A Taiwan based large population database analysis of diabetic patients revealed that those taking TZDs had significantly lower risk of developing PD compared to the non-TZD group in a dose dependent manner [74]. However, in a phase 2 double blind, randomized trial, Pioglitazone at 15mg/day and 45mg/day could not modify progression of disease in early PD [75]. Disappointingly, there are limited studies that address the mechanism behind the benefits of TZDs in PD risk. The discrepancy between the preclinical data, meta analyses and clinical data have not indicated yet whether this class of diabetes medication truly holds promise in the landscape of more novel diabetic therapies.

Dipeptidyl Peptidase-4 Inhibitors (DPP4i)

DPP4i (sitagliptin, linagliptin, saxagliptin, alogliptin) are hypoglycemic agents associated with a low risk of hypoglycemia and weight gain in patients with diabetes. They improve glucose metabolism by enhancing the bioavailability of active glucagon-like peptide-1 (GLP-1) and by inhibiting its degradation.

In vivo model studies show that DPP4i 1) upregulate neuroprotective pathways such as PI3K/AKT and Nrf2 [76], 2) increase levels of neurotropic factors such as BDNF and C-reactive element binding protein (CREB) [77,78], 3) reduce oxidative stress through indirect measurements of glutathione and malondialdehyde [78], 4) downregulation of proinflammatory markers (TNF-α, IL6, Iba-1, and GFAP), NF-κB, and reduce dopaminergic degeneration [77,79], and 5) decreases αSyn aggregation [80].

In the clinical setting, drug naïve, de novo PD patients with diabetes that were treated with DPP4i had significantly higher baseline DAT availability in the anterior, posterior, and ventral putamen compared to diabetic PD patients not on DPP4i [81]. Furthermore, this study showed that the DDP4i group had a slower increase in levodopa-equivalent dose over time and a lower rate of levodopa-induced dyskinesia.

While in vivo studies draw strong links between DPP4i use and αSyn aggregation, clinical studies are yet to explore whether these antidiabetic agents affect αSyn physiology.

Sodium-Glucose Cotransporter-2 Inhibitors (SGLT2i)

SGLT2i (dapagliflozin, empagliflozin, canagliflozin, ertugliflozin, ipragliflozin, tofogliflozin, remogliflozin) are a class of antidiabetic medications that inhibit the reabsorption of glucose via the SGLT2 in the proximal tubules of the nephron and therefore lower blood sugar by enhancing glycosuria. Owing to their pleiotropic effects, they have also shown great efficacy for cardiovascular health, renal diseases, weight loss, and stabilizing blood pressure [82]. One of the predominant mechanisms by which SGLT2i produces these additional benefits is through reduction of reactive oxidative species and protecting the integrity of the mitochondria [83]. Therefore, studies indicate that SGLT2i improves mitochondrial function by reducing oxidative stress, enhancing mitochondrial biogenesis, and restoring autophagic-lysosomal balance [84,85,86].

Interestingly, SGLT2 are also found in the hippocampus, cerebellum, and at the blood-brain barrier endothelial cells [87]. This makes the SGLT2i particularly intriguing CNS penetrating drugs. Preclinical studies have provided promising solutions for PD. In a rat model, rotenone was used to mimic PD like motor deficits and empagliflozin, SGLT2i, treated groups showed improvements in motor function, decreases in αSyn accumulation, decreased inflammation, which was suggestive of its neuroprotective effect [88]. Similarly in a rotenone induced PD zebrafish model, empagliflozin treated groups showed improved DOPA/DA ratio, improved performance on Y-maze task, increased markers related to inflammation and autophagy [89]. Another rotenone induced PD rat model study has shown attenuation of cognitive dysfunction, improvement of dopamine secretion, and decreased dopaminergic neuronal loss [90]. Multiple similar studies in toxin induced PD models have shown improvements after SGLT2i therapy through improvements in autophagy, mitochondrial function, and inflammation [90,91,92,93]. In addition, SGLT2i have shown to downregulate Snca gene expression in a diabetic mouse model [94].

Clinically, SGLT2i use has also shown reductions in risk of PD dementia [95] and lower risk of incidence of PD in T2DM populations [96,97]. In a large head to head comparison study between SGLT2i and metformin using 20 year dataset of almost 900,000 patients, SGLT2i was associated with a 28% lower PD risk than metformin suggestive of its superiority in neuroprotection [98]. Another study has shown that SGLT2i were superior to DPP4i in significantly lowering risk of PD after using a retrospective cohort analysis of 89,000 Medicare beneficiaries [96]. However, this meta-analysis looking at dapagliflozin found no significant association between the medication use and risk of dementia in PD [99]. Taken together, there is promising in vivo data regarding the use of SGLT2i therapy in PD models, however, there have not been any randomized controlled trials studying this medication in PD populations.

Glucagon-like Peptide-1 Receptor Agonists (GLP-1 RA)

GLP-1 RA (exenatide, liraglutide, semaglutide, dulaglutide, tirzepatide, lixisenatide) are incretin hormones that enhance glucose-induced insulin secretion and inhibit both gastric emptying and glucagon secretion [100]. Like the SGLT2i class, they have multiple pleiotropic effects including reducing obesity, improving cardiovascular health, and improving bone health. They act via both peripheral and central mechanisms, thus there has been a lot of interest in the GLP-1 RA and their role in PD.

In preclinical models, GLP-1/GIP dual agonist DH3-CH alone and liraglutide alone (GLP-1 RA) were both effective at reducing motor decline in the MPTP mouse model of PD. This study also found decreased expression of αSyn and activation of Wnt/β-catenin signaling pathway, regulatory molecule in cell proliferation, differentiation, apoptosis, neurogenesis and stem cell maintenance [101]. Similarly, another study looking at another GLP-1/GIP dual receptor agonist, DA5-CH, in a MPTP mouse model showed reversal of MPTP induced decrease of BDNF and GDNF where as it reduced αSyn, levels of inflammation, and proinflammatory markers (IL6, Il-Iβ) [102]. Another study in the MPTP PD mouse model found that GLP-1 RA activated PGC-1α, which regulates autophagy and cell apoptosis, and induced a neuroprotective effect [103]. In MitoPark mice, GLP-1 RA was found to reduce tyrosine hydroxylase expression by lowering reactive oxidative species and thus decreasing dopaminergic denervation [104], and another GLP-1 RA, PT320, was shown to improve L-DOPA-induced dyskinesias [105].

Clinical trials with exenatide have shown modest improvements in motor and cognitive symptoms [106,107,108]. Lixisenatide therapy resulted in modest effects in terms of progression of motor disability in early PD [109]. Liraglutide treatment was effective at improving both motor and nonmotor symptoms in PD, with effects sustaining while off medication [110]. In a population-based study, new users of GLP-1 RA have shown to have a 23% lower risk of PD than new users of DPP4i use [111]. However, a double blind randomized, placebo-controlled trial looking at exenatide over 96 weeks did not find any evidence of it acting as a disease-modifying treatment in PD patients when using MDS-UPDRS part III scores [112]. Interestingly, a recent metanalysis of studies involving GLP-1 RA demonstrated a detrimental effect on motor outcomes compared to placebo [113]. Overall, this class of antidiabetic medication shows the most promise in PD disease modification and multiple clinical trials are under way to draw clearer conclusions on their effects in PD.