The silent epidemic within the pandemic: pathophysiology and prediction of post-COVID-19 diabetes

Abstract

Background

The coronavirus disease 2019 (COVID-19) pandemic has presented extraordinary challenges to global public health, with impacts reaching beyond acute respiratory manifestations to include long-term metabolic disturbances. Emerging evidence indicates a significant link between SARS-CoV-2 infection and the onset of diabetes mellitus, establishing this condition as a major element of the post-acute sequelae of COVID-19, often referred to as Long COVID.

Main body

This review synthesizes epidemiological findings that demonstrate a elevated incidence of new-onset diabetes following COVID-19, particularly among certain high-risk demographic groups. We examine the molecular mechanisms underpinning this association, such as viral entry into pancreatic β-cells via ACE2 receptors, systemic inflammation leading to insulin resistance, and the potential diabetogenic effects of glucocorticoids used in COVID-19 treatment. Furthermore, this review outlines biomarker profiles that distinguish COVID-19-associated diabetes from traditional type 2 diabetes, underscoring important pathophysiological differences. Additionally, we evaluate advances and ongoing challenges in developing predictive risk models that combine clinical and molecular data to identify individuals at elevated risk for post-COVID diabetes.

Conclusions

By integrating multidisciplinary evidence, this comprehensive narrative review aims to guide future research and shape clinical approaches for early detection, prevention, and management of diabetes following COVID-19, thereby confronting a latent health crisis emerging within the broader pandemic context.

Introduction

The pandemic of coronavirus disease 2019 (COVID-19), caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), represents one of the most significant public health crises of the century, exerting extensive and varied impacts on healthcare systems worldwide [1]. Initially recognized for its acute respiratory manifestations, including pneumonia and acute respiratory distress syndrome (ARDS), COVID-19 is now increasingly acknowledged as a systemic illness that can result in long-term complications affecting multiple organ systems, such as the respiratory, cardiovascular, neurological, renal, and endocrine systems [1–3]. The recognition of these persistent effects, often termed “long COVID” or post-acute sequelae of SARS-CoV-2 infection (PASC), has sparked considerable research interest due to their ramifications for patient health and the distribution of healthcare resources [4, 5].

Among the spectrum of complications following COVID-19, the emergence of new-onset diabetes mellitus (NODM) has become a particularly concerning metabolic sequel. Diabetes, defined as a chronic metabolic disorder marked by elevated blood glucose levels resulting from insufficient insulin secretion, impaired insulin action, or both, poses a significant global health challenge. The bidirectional relationship between COVID-19 and diabetes has gained increasing attention: individuals with pre-existing diabetes face an elevated risk of severe outcomes from COVID-19, while infection with SARS-CoV-2 may trigger the onset of diabetes or exacerbate glycemic control in individuals who previously had normal blood sugar levels [6–8]. This phenomenon raises critical questions regarding the epidemiology, pathophysiology, and clinical management of diabetes as a post-infectious complication resulting from COVID-19.

Epidemiological studies have identified a variety of incidence rates regarding the onset of diabetes following COVID-19 infection, with certain populations showing a notable increase in diabetes diagnoses during and after the pandemic compared to periods prior to the outbreak [9, 10]. For instance, a retrospective cohort study carried out in the United States revealed that individuals who recovered from COVID-19 were at an increased risk of developing diabetes within months after infection, regardless of pre-existing risk factors [6]. Similarly, pediatric groups have exhibited heightened rates of newly diagnosed type 1 (T1DM) and type 2 diabetes mellitus (T2DM) throughout the pandemic, often presenting with more severe clinical symptoms, including diabetic ketoacidosis (DKA) [9, 11]. However, the attributable risk remains unclear due to the variations in research methodologies, demographic factors, viral strains, and vaccination statuses across different studies [6]. Moreover, ongoing research into the longevity of hyperglycemia and diabetes following COVID-19 indicates that while some patients may experience transient dysglycemia, others may go on to develop chronic diabetes that requires long-term management [12].

While numerous meta-analyses have authoritatively quantified the epidemiological association between COVID-19 and new-onset diabetes [13–15], a comprehensive synthesis that integrates the distinct molecular pathogenesis with clinically actionable risk prediction models remains limited. In contrast to these largely statistical aggregations, this review focuses on elucidating the unique ‘etiopathogenetic triad’ and critically evaluates emerging predictive tools, with the overarching aim of translating mechanistic insights into strategies for personalized risk stratification.

The molecular pathways that clarify the connection between COVID-19 and the development of diabetes are complex and multifaceted. SARS-CoV-2 primarily gains entry into host cells via the angiotensin-converting enzyme 2 (ACE2) receptor, which is not only found in the respiratory system but also in pancreatic islet cells, adipose tissue, and various metabolic organs [8, 16]. The virus’s infection of pancreatic β-cells may lead to impaired insulin secretion and β-cell dysfunction, contributing to hyperglycemia [6, 17]. Additionally, the systemic inflammation, cytokine storms, and immune dysregulation typically associated with severe COVID-19 can induce insulin resistance and metabolic imbalances [7, 17]. The impact of stress-induced hyperglycemia, glucocorticoid therapies, and the revelation of previously undiagnosed diabetes further complicates the clinical landscape [7, 18]. Recent studies also indicate that metabolic byproducts, such as methylglyoxal, may be involved in the development of COVID-19-related hyperglycemia, linking altered glycolytic pathways to β-cell damage and systemic insulin resistance [17]. Collectively, these mechanisms suggest both direct and indirect influences of SARS-CoV-2 on glucose metabolism, thereby supporting the classification of COVID-19-associated diabetes as a unique clinical condition within the wider framework of post-COVID-19 syndromes [6].

In light of the potential for COVID-19 to intensify the worldwide diabetes epidemic, there is an urgent necessity for a thorough and integrative methodology to elucidate the epidemiological connections, unravel the underlying molecular mechanisms, and create validated models for risk assessment. Such initiatives would significantly improve the understanding of the lasting metabolic repercussions associated with COVID-19 and provide essential insights for public health policies and clinical management approaches aimed at alleviating the effects of this emerging complication. This review seeks to meticulously assess the correlation between SARS-CoV-2 infection and the onset of diabetes, clarify the molecular pathways involved, and suggest frameworks for risk stratification and predictive modeling. The outcomes of this study will be crucial in directing future research, clinical methodologies, and public health initiatives in tackling the interconnected issues posed by COVID-19 and diabetes.

Epidemiological evidence of COVID-19 and newly diagnosed diabetes

Epidemiological data analysis of new-onset diabetes risk in COVID-19 infected individuals

A multitude of cohort and case-control studies have consistently demonstrated an elevated relative risk for the development of diabetes following COVID-19 infection, highlighting a reciprocal relationship between SARS-CoV-2 and disturbances in glucose metabolism. Epidemiological evidence indicates that individuals who contract COVID-19 exhibit a markedly increased frequency of newly diagnosed diabetes cases, encompassing phenotypes similar to T1DM, T2DM, and ketosis-prone diabetes [14, 19]. Meta-analytical evaluations estimate that around 19.7% of COVID-19 patients also have diabetes, while 25.2% present with hyperglycemia; notably, those with newly diagnosed diabetes show significantly higher mortality rates in comparison to their non-diabetic peers [14]. Longitudinal cohort studies conducted in Naples, Italy, further corroborate a substantial increase in T2DM incidence throughout the pandemic, with rates rising from 4.85 to 12.21 per 1,000 person-years before and during the COVID-19 crisis, respectively, reflecting a more than twofold increase in new diagnoses [20]. Furthermore, pediatric populations have reported a rise in T1DM incidence and occurrences of DKA during the pandemic, with meta-analyses revealing incidence rate ratios (IRR) ranging from 1.14 to 1.27 for T1DM and 1.26 for DKA when juxtaposed with pre-pandemic periods [21]. These findings suggest that SARS-CoV-2 infection may serve as a trigger for the onset of diabetes across diverse age groups.

The likelihood of developing new-onset diabetes subsequent to COVID-19 infection varies among different demographic and clinical subgroups. Factors such as age and sex emerge as significant modifiers, with older adults and males exhibiting heightened vulnerability and less favorable health outcomes [22, 23]. Comorbidities, including obesity, hypertension, cardiovascular conditions, and pre-existing metabolic disorders, further intensify the risk of experiencing severe COVID-19 and the subsequent development of diabetes [24, 25]. For example, diabetic individuals infected with COVID-19 are at an increased risk of severe complications, such as elevated rates of hospitalization, intensive care unit (ICU) admissions, and mortality [26, 27]. Notably, research from various regions, including the Middle East, North Africa, and South Asia, underscores the heightened susceptibility of diabetic individuals to COVID-19 and the associated risk of new-onset diabetes within these demographics [28, 29]. Pediatric investigations indicate that although there has been an increase in new-onset T1DM cases, the direct causal link with SARS-CoV-2 remains to be fully elucidated, with some evidence indicating no significant difference in the seroprevalence of anti-SARS-CoV-2 antibodies between newly diagnosed T1DM patients and control groups [30].

The severity of COVID-19 infection demonstrates a significant association with the increased probability of newly diagnosed diabetes. Instances of severe COVID-19, characterized by extensive pulmonary involvement, systemic inflammatory responses, and cytokine storms, are particularly likely to instigate metabolic dysregulation and beta-cell dysfunction, ultimately leading to hyperglycemia and the development of diabetes [19, 31]. Retrospective studies reveal that individuals suffering from severe infections or those requiring admission to ICUs exhibit elevated rates of new diabetes diagnoses and incidences of DKA [32, 33]. Investigations employing animal models corroborate this association, indicating heightened pneumonia severity and insulin resistance in diabetic mice infected with SARS-CoV-2, thereby providing a mechanistic basis for the observed epidemiological patterns [34]. Moreover, the presence of both obesity and diabetes not only intensifies the severity of COVID-19 but also undermines the efficacy of vaccination, further complicating disease management [35].

Taken together, the epidemiological data from a range of studies and meta-analyses confirm that COVID-19 infection markedly elevates the risk of newly diagnosed diabetes across various demographic populations, with risk factors including age, gender, pre-existing health conditions, and the severity of the infection playing a significant role (Fig. 1). These findings underscore the necessity for proactive screening and continuous monitoring of glucose metabolism in COVID-19 survivors, especially among those who have endured severe illness or possess predisposing risk factors. Future prospective studies are essential to elucidate the progression of diabetes linked to COVID-19, identify individuals at increased risk, and develop targeted prevention and management strategies [19, 36].

Fig. 1.

Epidemiological evidence and stratification for new-onset diabetes post-COVID-19

Global and regional diabetes disease burden assessment

The assessment of the global and regional ramifications of diabetes, particularly in the context of the COVID-19 pandemic, has leveraged an extensive array of publicly available datasets alongside advanced epidemiological modeling techniques to estimate the incidence, prevalence, and disability-adjusted life years (DALYs). Systematic reviews and meta-analyses, which have incorporated data from millions of COVID-19 patients, have quantified the prevalence of diabetes across varying severity levels of the disease, revealing an overall diabetes prevalence of approximately 14.7% among confirmed COVID-19 cases. Significantly, elevated prevalence rates have been observed in patients experiencing severe disease and those who have died from the illness [37]. These analyses utilize comparative risk assessment models to compute the population attributable fraction (PAF) of diabetes concerning the severity and mortality of COVID-19, indicating that diabetes accounts for nearly 9.5% of severe COVID-19 cases and 16.8% of COVID-19-related deaths globally. These statistics demonstrate variations influenced by factors such as national income levels, the quality of healthcare, and the baseline prevalence of diabetes within different populations [37].

Regional disparities in the burden of diabetes during the pandemic are notably pronounced. For instance, in India, ecological studies have revealed strong positive correlations between the incidence of COVID-19 cases and fatalities at the state level with non-communicable disease (NCD) risk factors, including diabetes and obesity, as well as indicators of urbanization and healthcare access [38]. Similarly, in the United States, racial and ethnic minority groups disproportionately bear the burden of metabolic diseases, including diabetes, which contribute to adverse outcomes in COVID-19 cases [39, 40]. Modeling studies conducted in low- and middle-income countries (LMICs) indicate that undiagnosed diabetes has significantly contributed to COVID-19-related hospitalizations and deaths, accounting for as much as 21.1% of hospital admissions and 30.5% of fatalities in eight examined LMICs, thereby underscoring the hidden burden of diabetes exacerbated by the pandemic [41]. Moreover, the pandemic has been linked to the emergence of new-onset diabetes, as cohort studies reveal increased risks and incidences of diabetes and antihyperglycemic medication use among COVID-19 survivors in the post-acute phase, compared to both contemporary and historical control groups [42]. This bidirectional relationship suggests that COVID-19 not only worsens outcomes for individuals with pre-existing diabetes but may also trigger the onset of diabetes, thereby further intensifying the global diabetes burden.

DALYs function as a comprehensive metric that combines years of life lost (YLL) due to premature mortality with years lived with disability (YLD). According to the Global Burden of Disease (GBD) 2021 report, NCD, particularly diabetes, were responsible for an estimated 1.73 billion DALYs globally in 2021. Furthermore, the age-standardized rates of DALYs associated with diabetes experienced an approximate increase of 14% from 2010 to 2021 [42]. The COVID-19 pandemic disrupted the previously noted decline in all-cause DALYs, establishing COVID-19 as the leading contributor to DALYs worldwide in 2021 [43]. Investigations into the acute and post-acute impacts of COVID-19 in Australia have suggested that long COVID and potential long-term complications, such as diabetes, may substantially increase total DALYs in the post-vaccination era, thereby underscoring the importance of incorporating chronic sequelae into burden evaluations [44]. Cross-national analyses reveal that the burden of diabetes is often correlated with socioeconomic factors and the strength of healthcare systems. For example, in Italy, regional differences in disease burden and mortality rates have been linked to variations in comorbid conditions, including diabetes, with northern regions generally exhibiting better health outcomes despite higher YLD figures [45]. In Scotland, the DALYs attributable to COVID-19 showed pronounced disparities influenced by levels of deprivation, with diabetes significantly contributing to the overall health decline in economically disadvantaged regions [46].

Projections based on GBD data indicate a sustained rise in the prevalence of diabetes and its associated DALYs globally, a trend further intensified by the COVID-19 pandemic. A study utilizing GBD 2021 data predicted that the global prevalence of T2DM would exceed pre-pandemic estimates by 2030, with both mortality rates and DALYs anticipated to accelerate in the post-pandemic context [47]. This situation highlights the urgent need for enhanced public health strategies that emphasize the prevention, early detection, and management of diabetes in the wake of COVID-19. Overall, the evaluation of the global and regional burden of diabetes during the COVID-19 pandemic integrates extensive epidemiological data, modeling of disease prevalence and outcomes, as well as the quantification of DALYs to reflect both mortality and morbidity impacts. The burden of diabetes is notably heterogeneous across regions, influenced by socioeconomic factors, healthcare quality, and the severity of the pandemic. The pandemic has intensified the diabetes burden through increased severity in existing cases and the rise of new diabetes cases, posing long-term challenges for public health systems worldwide. Effective monitoring, prudent allocation of resources, and targeted interventions are essential to address this growing concern (Fig. 2).

Fig. 2.

The bidirectional relationship between the COVID-19 pandemic and the global burden of diabetes

Molecular mechanism analysis of COVID-19 induced diabetes

Mechanisms of β-cell injury: reconciling direct viral entry with microenvironmental dysfunction

The molecular basis for pancreatic β-cell injury following SARS-CoV-2 infection involves a complex interplay of putative direct viral entry and indirect inflammatory mechanisms. Although ACE2 is the major receptor enabling viral attachment and cellular internalization, its expression in human pancreatic β-cells remains incompletely characterized. Transcriptomic and immunohistochemical analyses consistently report low or absent co-expression of ACE2 and the spike protein-priming protease TMPRSS2 in mature β-cells, suggesting limited opportunity for direct viral entry in vivo [48, 49]. Instead, ACE2 protein is predominantly localized to pancreatic microvascular pericytes and ductal cells, implying that viral entry may occur primarily in non-β cell populations [50, 51]. By contrast, studies using human pluripotent stem cell-derived pancreatic endocrine progenitors and fetal pancreas models demonstrate ACE2 expression in insulin-producing β-cell precursors, rendering them permissive to SARS-CoV-2 infection in vitro via ACE2-mediated endocytosis—a mechanism distinct from respiratory tract infection pathways [52]. These discrepancies highlight how cellular developmental stage and differentiation status may critically influence viral susceptibility.

Current evidence regarding the direct infection of β-cells by SARS-CoV-2 remains debated. While earlier studies suggested ACE2-mediated entry, recent single-cell RNA sequencing data indicate that ACE2 expression is low in β-cells but abundant in pancreatic ductal cells and the microvasculature. Therefore, we propose a dual-mechanism model:

• Direct Cytotoxicity: Although rare, direct viral entry may occur under conditions of upregulated ACE2 expression during inflammatory stress.

• Indirect/Paracrine Effects: The infection of adjacent endothelial and ductal cells triggers local inflammation and microvascular dysfunction/thrombosis, leading to ischemic or inflammatory injury to β-cells (the ‘bystander effect’). This distinction is crucial for developing targeted therapies.

Following viral entry, SARS-CoV-2 infection elicits intracellular responses that culminate in β-cell dysfunction and apoptosis. These include reduced insulin synthesis and secretion, alongside activation of caspase-mediated apoptotic pathways resembling those seen in type 1 diabetes [53]. Phosphoproteomic studies confirm the induction of apoptosis in infected β-cells, indicating cell-autonomous cytotoxicity independent of systemic inflammation [54]. Furthermore, viral RNA and proteins have been detected in β-cells from patient autopsy samples, though the clinical significance of these findings requires further validation. Notably, fibroblast growth factor 7 (FGF7) has been identified as an enhancer of ACE2 expression in β-cells, increasing viral susceptibility and exacerbating dysfunction—suggesting that inhibition of FGF receptor signaling might offer a therapeutic strategy [55].

Experimental models have been instrumental in deciphering viral–host interactions. Infection of isolated human islets and β-cell lines in vitro confirms their susceptibility to SARS-CoV-2, resulting in impaired function [56]. Human embryonic stem cell-derived islet organoids have further illuminated ACE2 regulation and viral entry mechanisms [55]. Although animal models of pancreatic SARS-CoV-2 infection remain limited, they consistently exhibit pancreatic inflammation and β-cell damage, supporting the translational relevance of in vitro findings. The scarce co-expression of ACE2 and TMPRSS2 in mature β-cells in vivo, however, suggests that indirect mechanisms—such as microvascular dysfunction, pericyte activation, and localized inflammation—likely contribute substantially to β-cell injury [51, 57]. Other entry mediators, including neuropilin-1 (NRP1), cathepsin L, and CD147, may also facilitate infection in pancreatic cells, potentially compensating for low ACE2 expression and expanding viral tropism [57] (Fig. 3).

Fig. 3.

Putative mechanisms of SARS-CoV-2-induced pancreatic β-cell damage

Briefly, SARS-CoV-2-induced β-cell injury involves a spectrum of direct and indirect pathways. While in vitro and developmental models support the possibility of ACE2-mediated infection, human tissue data emphasize the importance of alternative receptors and non-cell-autonomous mechanisms such as microvascular compromise and localized inflammation. Refining these mechanistic insights is essential for understanding COVID-19-associated diabetes and developing strategies to preserve β-cell function during infection.

Systemic inflammatory response and cytokine storm-induced insulin resistance

The systemic inflammatory response and cytokine storm induced by COVID-19 infection play a pivotal role in the development of insulin resistance, which is a key factor in the disruption of glucose metabolism and the onset of newly diagnosed diabetes in those affected. A fundamental aspect of this process is the overproduction of pro-inflammatory cytokines, notably interleukin-6 (IL-6) and tumor necrosis factor-alpha (TNF-α), which are markedly increased in severe cases of SARS-CoV-2 infection. These cytokines interfere with insulin signaling pathways by activating serine kinases that phosphorylate insulin receptor substrates, thereby obstructing the downstream signaling pathways essential for proper glucose uptake and metabolism [58, 59]. In addition, IL-6 and TNF-α contribute to a state of chronic low-grade inflammation that exacerbates insulin resistance, particularly in peripheral tissues such as skeletal muscle and liver, resulting in hyperglycemia and metabolic disruptions [60, 61]. Moreover, the dysregulation among T helper cell subsets, especially concerning Th1, Th2, and Th17 cells, further complicates metabolic homeostasis by steering immune responses towards a pro-inflammatory condition. For instance, increased Th17 activity is associated with elevated IL-17 production, which heightens inflammation and insulin resistance, whereas the Th1/Th2 imbalance modifies cytokine profiles that are vital for metabolic regulation [62]. The cytokine storm not only promotes insulin resistance but also incites dysfunction and apoptosis of pancreatic β-cells, either via direct viral invasion or indirectly through immune-mediated injury, further impairing insulin secretion [63, 64].

The inflammatory milieu is intensified by oxidative stress and endothelial dysfunction, conditions that are commonly observed in both COVID-19 and metabolic disorders. This situation fosters a harmful cycle marked by metabolic disturbances and increased immune activation [59, 65]. Investigations employing mathematical modeling have elucidated the complex interactions among cytokines and signaling pathways, highlighting the significance of autocrine loops and positive feedback systems that sustain excessive immune responses and insulin resistance during SARS-CoV-2 infection [59]. Furthermore, systemic inflammation in COVID-19 patients who have pre-existing metabolic issues, such as obesity and T2DM, exacerbates the cytokine storm, thereby worsening insulin resistance and glycemic control [58, 66]. Therapeutic agents like metformin and sitagliptin have shown potential in modulating inflammation and improving insulin sensitivity, partially by reducing levels of IL-6 and TNF-α while reinstating insulin signaling pathways [67, 68]. Moreover, the aryl hydrocarbon receptor (AHR) pathway has emerged as a significant mediator that links SARS-CoV-2-induced immune suppression, systemic inflammation, and metabolic anomalies, particularly insulin resistance in individuals suffering from obesity and diabetes [69] (Fig. 4). In summary, the systemic inflammatory response and cytokine storm associated with COVID-19 disrupt insulin signaling through elevated pro-inflammatory cytokines and dysregulation of immune cells, leading to insulin resistance and irregularities in glucose metabolism that may initiate or exacerbate diabetes in affected individuals. A thorough understanding of these mechanisms is essential for developing targeted interventions aimed at mitigating metabolic complications associated with COVID-19.

Fig. 4.

The inflammatory axis: how COVID-19 cytokine storm drives insulin resistance

Effects of glucocorticoid therapy on glucose metabolism

Glucocorticoids (GCs) have been widely employed in the therapeutic management of COVID-19 due to their substantial anti-inflammatory and immunomodulatory properties, which play a crucial role in mitigating the severe inflammatory responses associated with the disease. The clinical use of synthetic GCs, including dexamethasone and prednisone, has become a pivotal approach in treating moderate to severe COVID-19 cases, particularly in those complicated by ARDS. However, despite their therapeutic benefits, GCs significantly influence glucose metabolism, often resulting in hyperglycemia and increasing the likelihood of newly developed DM in susceptible individuals. This dual action necessitates a careful balance between their anti-inflammatory efficacy and metabolic adverse effects in the context of COVID-19 management [70, 71].

From a mechanistic standpoint, GCs primarily modulate glucose homeostasis by promoting hepatic gluconeogenesis, reducing peripheral glucose uptake, and inducing insulin resistance. The classic glucocorticoid receptor (GRα) mediates these effects by activating the transcription of genes linked to gluconeogenic pathways in the liver, leading to an increase in endogenous glucose production. In addition, GCs impede insulin signaling in peripheral tissues, such as skeletal muscle and adipose tissue, exacerbating hyperglycemia. For example, prolonged exposure to GCs inhibits insulin-activated Akt signaling in skeletal muscle, thereby contributing to insulin resistance and muscle atrophy [72]. In adipose tissue, GCs affect receptor activation and metabolic pathways, which influence glucose uptake and lipid metabolism, collectively impacting systemic glucose regulation [73, 74]. Notably, the alternative GR isoform GRβ exhibits insulin-mimetic properties in hepatocytes, suggesting a complex receptor-mediated regulation of glucose metabolism [75].

The timing of glucocorticoid administration is also critical in influencing their metabolic outcomes. Studies conducted in murine models have demonstrated that administering synthetic GCs, such as betamethasone, during periods of low endogenous glucocorticoid levels leads to more pronounced disturbances in glucose metabolism, including hyperinsulinemia and reduced insulin sensitivity, compared to administration that aligns with circadian rhythms. This chronopharmacological phenomenon underscores the importance of incorporating circadian biology considerations into glucocorticoid therapy to alleviate metabolic side effects [76, 77].

The clinical manifestation of hyperglycemia resulting from glucocorticoid administration is well-documented, with numerous studies revealing a significant prevalence of glucose metabolism abnormalities in patients undergoing glucocorticoid therapy, even at low doses and in the absence of traditional diabetes risk factors. For instance, patients receiving long-term, low-dose GCs for inflammatory rheumatic diseases exhibited impaired glucose tolerance, which could only be detected through oral glucose tolerance tests (OGTT), despite having normal fasting glucose levels [78]. Similarly, pediatric patients with refractory nephrotic syndrome who were treated with GCs showed persistent glucose metabolism disorders, highlighting the critical need for vigilant monitoring across diverse patient populations [79]. In the context of COVID-19, the administration of GCs may exacerbate hyperglycemia and insulin resistance, potentially resulting in the development of new diabetes cases or the worsening of existing metabolic disturbances [71, 80].

Innovative treatment strategies are being explored to mitigate the metabolic complications associated with GCs. Metformin, a well-established antidiabetic medication, has been shown to be effective in reducing glucocorticoid-induced hyperglycemia by inhibiting key mitochondrial enzymes and maintaining insulin sensitivity during glucocorticoid treatment [81]. In addition, sodium-glucose cotransporter 2 (SGLT2) inhibitors are currently being evaluated as potential alternatives to insulin therapy for managing glucocorticoid-induced diabetes, with ongoing clinical trials investigating their safety and effectiveness [82]. Furthermore, selective GR modulators and tissue-targeted disruption of glucocorticoid signaling offer promising avenues for dissociating the anti-inflammatory benefits of GCs from their metabolic side effects [70] (Fig. 5).

Fig. 5.

Glucocorticoid therapy in COVID-19 between desired anti-inflammation and undesired metabolic effects

While GCs play a crucial role in the management of COVID-19 due to their anti-inflammatory effects, their impact on glucose metabolism is both significant and intricate. The development of hyperglycemia and insulin resistance is driven by complex molecular mechanisms, including hepatic gluconeogenesis, impaired insulin signaling in peripheral tissues, and circadian fluctuations in receptor activity. Clinically, glucocorticoid therapy is associated with an increased risk of new-onset diabetes, underscoring the necessity for careful metabolic monitoring and the exploration of therapeutic strategies to mitigate these adverse effects. A thorough understanding of these interactions is essential for optimizing COVID-19 treatment protocols and improving patient outcomes.

Similarities and differences between COVID-19-related diabetes and traditional T2DM

Clinical presentation and metabolic characteristics comparison

The clinical and metabolic features of diabetes linked to COVID-19 present several notable distinctions when contrasted with those observed in traditional T2DM. In particular, diabetes associated with COVID-19 often exhibits more severe impairments in glucose homeostasis, characterized by persistent hyperglycemia and impaired insulin secretion. These abnormalities may stem from the effects of the virus on pancreatic beta cells, as well as a systemic metabolic alteration induced by SARS-CoV-2 infection. Histopathological studies have demonstrated that SARS-CoV-2 is capable of infecting pancreatic beta cells in vivo, leading to cellular stress characterized by mitochondrial dysfunction, endoplasmic reticulum dilation, and a metabolic shift towards glycolysis, as evidenced by increased levels of free NADH. These changes suggest a beta cell dysfunction that may persist beyond the acute phase of infection, potentially resulting in the emergence of new diabetes or the exacerbation of pre-existing diabetic conditions.

Regarding glycemic control, individuals with diabetes who contract COVID-19 frequently experience more unpredictable and severe hyperglycemia during the acute phase of the illness, which correlates with systemic inflammation and a cytokine storm. Clinical observations reveal that hyperglycemia in the context of COVID-19 tends to be more persistent and less variable, alongside elevated inflammatory markers such as C-reactive protein (CRP), creatinine, and fibrinogen, as well as prolonged hypercoagulability, in contrast to diabetic patients not infected with COVID-19 [83]. This situation sharply contrasts with the more stable glycemic patterns typically seen in conventional T2DM patients receiving standard treatment. Furthermore, acute metabolic crises, including DKA, hyperglycemic hyperosmolar state (HHS), and euglycemic DKA, are reported with increased frequency and severity among COVID-19 patients, often associated with higher mortality rates. The coexistence of DKA/HHS with acute kidney injury (AKI) significantly worsens patient outcomes, highlighting a distinct acute metabolic phenotype linked to diabetes in the context of COVID-19 [84, 85].

The insulin secretion and resistance profiles present significant distinctions between diabetes linked to COVID-19 and conventional T2DM. Infection with SARS-CoV-2 instigates a metabolic reconfiguration in host cells, resulting in alterations in glucose and glutamine metabolism. These changes may obstruct insulin secretion and promote insulin resistance. The viral-induced enhancement of pathways such as hypoxia-inducible factor-1 alpha (HIF-1α) and the mammalian target of rapamycin complex 1 (mTORC1) is pivotal in these metabolic disturbances, consequently worsening hyperglycemia and inflammatory responses [86, 87]. Moreover, hyperglycemia can increase the expression of SARS-CoV-2 entry factors, such as ACE2 and TMPRSS2, in hepatic cells, potentially creating a harmful cycle of infection and metabolic dysfunction [88]. These mechanisms are in stark contrast to the insulin resistance primarily driven by obesity and chronic inflammation observed in typical T2DM.

The acute phase of diabetes associated with COVID-19 is marked by a multifaceted interplay of direct viral cytopathic effects, systemic inflammation, and metabolic dysregulation, leading to a swift deterioration in metabolic control. Conversely, traditional T2DM progresses gradually over several years, characterized by a gradual decline in beta cell function and escalating peripheral insulin resistance. The long-term mechanisms also differ; COVID-19 may cause enduring damage to beta cells and metabolic memory effects, reminiscent of those seen in gestational DM, where such metabolic memory increases vulnerability to SARS-CoV-2 infection in the postpartum period [89]. This suggests that diabetes linked to COVID-19 may result in distinctive chronic consequences, including persistent beta cell dysfunction and altered metabolic profiles, warranting further longitudinal studies.

Metabolomic analyses of patients with COVID-19 have uncovered unique metabolic signatures that set them apart from those with conventional T2DM. Significant alterations in amino acid metabolism, lipid profiles, and energy production pathways have been observed. Patients experiencing severe COVID-19 exhibit marked dysregulation in lipoproteins, hexosylceramides, and phosphoethanolamines, which correlate with disease severity and mortality [90]. These findings contrast with the more stable metabolic disturbances typically associated with T2DM and underscore the profound impact of SARS-CoV-2 infection on host metabolism (Fig. 6). Overall, diabetes associated with COVID-19 reveals distinct clinical features and metabolic characteristics when compared to traditional T2DM. The acute phase is characterized by significant hyperglycemia, stress on beta cells, impaired insulin secretion, and systemic metabolic reprogramming driven by viral infection and inflammatory responses. The long-term consequences may encompass persistent beta cell dysfunction and unique metabolic memory effects.

Fig. 6.

A comparative view of COVID-19-associated diabetes and traditional T2DM

Discovery and significance of unique biomarkers

Recent studies have identified a variety of unique molecular biomarkers associated with diabetes in the context of COVID-19, which offer valuable prospects for the early identification and monitoring of disease progression. Among these, immune-related markers, particularly specific human microRNAs (miRNAs), have emerged as crucial elements. In particular, miRNAs such as hsa-miR-298, hsa-miR-3925-5p, hsa-miR-4691-3p, and hsa-miR-5196-5p have shown a tendency to preferentially target the genomes of SARS-CoV-2 rather than the diabetes-related mRNAs of the host in pancreatic cells. This finding suggests a mechanistic link between viral infection and the disruption of glucose metabolism, which may contribute to the development of new diabetes cases or the exacerbation of pre-existing diabetes [91]. Additionally, circulating miRNAs are recognized for their roles in modulating immune and inflammatory responses, and their altered expression profiles in COVID-19 patients with comorbidities, such as type 2 diabetes, could act as markers of disease severity and organ dysfunction, highlighting their potential as diagnostic and prognostic indicators [92]. In addition to miRNAs, extensive research has focused on protein biomarkers related to inflammation and tissue damage. Elevated levels of inflammatory cytokines, such as IL-6, CRP, and chemokines like CXCL-10 in both saliva and serum, have been correlated with the severity of COVID-19 and are significantly increased in individuals with diabetes. This correlation indicates their potential utility in monitoring disease progression and evaluating treatment responses [93].

Cardiovascular biomarkers, specifically high-sensitivity troponins and NT-proBNP, exhibit significant elevations in severe cases of COVID-19 and serve as independent indicators of mortality risk. This highlights the critical involvement of cardiac function within the pathophysiological framework of COVID-19, particularly among patients with diabetes, who are predisposed to cardiovascular issues [94, 95]. Moreover, novel biomarkers such as immature granulocytes (IGs) have emerged as promising indicators of the severity of SARS-CoV-2 infection; elevated IG levels correlate with negative clinical outcomes, prolonged hospitalization durations, and increased mortality rates, thereby suggesting their potential utility in early risk stratification [96]. In addition, genomic biomarkers, notably the ATP6V1B2 and IFI27 genes, have exhibited remarkable precision in forecasting SARS-CoV-2 infection status, offering new molecular targets for advancements in diagnostic methodologies [97]. Another noteworthy biomarker, the fibrinogen-to-platelet ratio (FPR), has demonstrated enhanced predictive capability for COVID-19 mortality compared to conventional markers such as the neutrophil-to-lymphocyte ratio (NLR), particularly in non-diabetic patients, thereby indicating its potential relevance for diabetic populations as well [98].

Furthermore, the soluble form of ACE2, which is released from the membrane during the initial stages of infection, has been linked to disease severity and coagulation irregularities, establishing it as a promising biomarker for predicting outcomes in COVID-19 patients, irrespective of their diabetes status [99]. The presence of dimeric IgA in plasma serves as a specific marker for recent SARS-CoV-2 infection, facilitating the distinction between recent and past infections, which is essential for timely diagnosis and epidemiological surveillance [100].

In addition to molecular markers, biochemical indicators such as carbohydrate antigen 15–3 (CA 15–3), associated with lung injury and fibrosis, have been correlated with poorer prognoses in SARS-CoV-2 pneumonia, providing valuable insights into pulmonary complications that may worsen diabetic conditions [101]. Significantly, metabolic biomarkers, including fasting plasma glucose (FPG) levels at the time of admission, have been identified as independent predictors of prolonged viral shedding and adverse clinical outcomes, emphasizing the interconnected relationship between glucose metabolism and the severity of COVID-19 [102]. The integration of these biomarkers into clinical practice holds significant promise for the early detection of diabetes associated with COVID-19, facilitates precise monitoring of disease progression, and enables the development of personalized treatment approaches. However, the specificity of many biomarkers is frequently limited due to their overlap with a range of inflammatory or infectious conditions, underscoring the necessity for further validation in large, well-characterized cohorts. Future research should focus on the incorporation of multi-omics approaches, which include transcriptomic, proteomic, and metabolomic data, to refine biomarker panels that accurately reflect the complex interactions between SARS-CoV-2 infection and the onset of diabetes (Fig. 7). Advancements in this area will be essential for creating dependable risk prediction models and improving patient outcomes in this vulnerable population.

Fig. 7.

A panorama of biomarkers in diabetes associated with SARS-CoV-2 infection

Taxonomy of COVID-19 associated diabetes: a distinct entity or a heterogeneous spectrum?”

Defining the nosological status of COVID-19-associated diabetes within established WHO or ADA frameworks remains a subject of active debate. Our synthesis of the current literature suggests it likely represents aheterogeneous clinical and pathophysiological spectrum rather than a single, distinct entity. This spectrum may encompass several overlapping phenotypes:

• Unmasked pre-existing diabetes: Severe systemic inflammation and insulin resistance during acute COVID-19 can precipitate overt hyperglycemia in individuals with pre-existing, undiagnosed dysmetabolism.

• Acute stress hyperglycemia: A transient state of hyperglycemia driven by counter-regulatory hormones and cytokines, which may resolve following viral clearance and recovery.

• New-onset, insulin-deficient diabetes: Cases presenting with acute hyperglycemia and ketosis, mimicking autoimmune T1DM but often lacking classical islet autoantibodies, suggesting a potential direct or immune-mediated β-cell injury.

A key argument against simply subsuming these cases under pre-existing T2DM is the emerging evidence for distinct metabolic signatures. For instance, specific perturbations in lipid species such as hexosylceramides appear to differentiate COVID-19-associated hyperglycemia from traditional T2DM trajectories, pointing towards unique pathophysiological underpinnings. Furthermore, the frequent presentation with an acute, profound β-cell secretory defect differs from the typical gradual decline seen in T2DM.

Therefore, we propose that ‘COVID-19-associated diabetes’ is best conceptualized provisionally as a form of ‘hybrid’ or ‘infection-related diabetes’. It should be categorized under the broader heading of ‘specific types of diabetes due to other causes’ until long-term, prospective phenotyping studies determine whether it constitutes a persistent, standalone subtype or resolves into one of the classical categories.

Immunological mechanisms and autoimmune dysregulation in newly diagnosed diabetes

Role of autoimmune response in COVID-19-induced diabetes

Viral infections have long been linked to the initiation of autoimmune mechanisms that lead to the development of T1DM through the destruction of pancreatic β-cells. Recently, the attention has shifted towards SARS-CoV-2, the causative agent of COVID-19, as a potential trigger for autoimmune diabetes, which includes both T1DM and latent autoimmune diabetes in adults (LADA). This virus can infiltrate pancreatic islet cells by binding to the ACE2 receptor, which is highly expressed in β-cells. Such interaction results in direct cytopathic damage and activates inflammatory pathways that jeopardize these insulin-secreting cells. A plethora of case reports and observational studies have documented occurrences of newly diagnosed autoimmune diabetes following COVID-19 infection or vaccination, often presenting as DKA alongside the presence of islet autoantibodies, such as glutamic acid decarboxylase (GAD) antibodies, indicative of an autoimmune etiology [103–106]. These findings suggest that SARS-CoV-2 infection could either trigger or accelerate the autoimmune destruction of β-cells in genetically predisposed individuals, potentially through mechanisms such as molecular mimicry, bystander activation, or persistent viral infection [107, 108].

From an immunological standpoint, COVID-19 is characterized by an aberrant immune response, typified by a cytokine storm and altered T-cell populations, which may contribute to the breakdown of self-tolerance and the development of autoimmunity. Regulatory T cells (Tregs), which usually play a role in suppressing autoreactive T-cell responses, may become depleted or lose their functional capacity during SARS-CoV-2 infection, leading to uncontrolled autoimmune reactions aimed at pancreatic islets [108]. Furthermore, changes in immune cell composition, including an increase in activated CD8+ T cells and modifications in B-cell subsets, have been observed in patients with autoimmune disorders and COVID-19, thereby strengthening the hypothesis that immune dysregulation is a significant factor in the pathogenesis of the disease [109]. Epidemiological data indicate an increase in the occurrence of T1DM alongside various autoimmune diseases, such as autoimmune thyroiditis, during the COVID-19 pandemic. Some studies suggest that newly diagnosed T1DM patients exhibit a higher prevalence of SARS-CoV-2 antibodies compared to their healthy counterparts [110, 111]. In contrast, other research has not established a definitive link, highlighting the need for longitudinal studies to clarify causal relationships and temporal patterns [30, 112]. It is important to note that cases of autoimmune diabetes following COVID-19 vaccination have been reported, although these occurrences are rare. This suggests that immune activation provoked by viral antigens or adjuvants may unmask pre-existing autoimmunity in genetically predisposed individuals [106, 113–115] (Fig. 8).

Fig. 8.

Proposed mechanisms of COVID-19-induced autoimmune diabetes

Thus, the autoimmune response plays a crucial role in the context of diabetes associated with COVID-19. This involves the direct infection of pancreatic β-cells by the virus, the immune-mediated destruction of these cells, and the dysregulation of immune cell populations that disrupt self-tolerance. The interplay between SARS-CoV-2 infection, the host’s genetic predisposition, and changes in the immune system contributes to the development or worsening of autoimmune diabetes. It is vital for healthcare providers to remain vigilant regarding autoimmune diabetes in patients presenting with hyperglycemia or DKA following COVID-19 infection or vaccination. Utilizing immune profiling methods, such as autoantibody testing and C-peptide evaluations, can aid in accurate diagnosis and management [103–105]. Further mechanistic studies and epidemiological surveillance are essential to elucidate the underlying mechanisms and to formulate preventive and therapeutic strategies targeting the autoimmune effects of COVID-19.

Synergistic role of inflammation and oxidative stress in islet injury

The interplay between systemic inflammation and oxidative stress plays a crucial role in exacerbating damage to pancreatic islet cells, particularly in the context of SARS-CoV-2 infection and its association with newly diagnosed diabetes. The systemic inflammatory response triggered by COVID-19 initiates a cytokine storm characterized by elevated levels of pro-inflammatory mediators such as interleukins, TNF-α, and interferons. These inflammatory cytokines not only have direct cytotoxic effects on pancreatic β-cells but also promote the infiltration of immune cells into islet tissues, thereby intensifying local inflammation and cellular injury. Concurrently, oxidative stress arises from an imbalance between the generation of reactive oxygen species (ROS) and the antioxidant defenses present within islet cells. The excessive production of ROS, which is catalyzed by mitochondrial dysfunction and the activation of NADPH oxidases during viral infection and hyperglycemia, leads to lipid peroxidation, DNA damage, and protein oxidation, further jeopardizing β-cell function and viability. This synergistic interaction between inflammation and oxidative stress cultivates a harmful microenvironment that accelerates β-cell apoptosis and dysfunction, ultimately resulting in impaired insulin secretion and disrupted glucose homeostasis [116].

At the molecular level, several key signaling pathways mediate the interaction between inflammation and oxidative stress in islet injury. The renin-angiotensin-aldosterone system (RAAS), particularly through angiotensin II, is upregulated in the settings of COVID-19 and diabetes, promoting the activation of NADPH oxidase and subsequent ROS production. This cascade activates nuclear factor kappa B (NF-κB), a transcription factor that enhances the expression of inflammatory cytokines and adhesion molecules, thereby sustaining inflammation and oxidative damage. Additionally, the mitogen-activated protein kinase (MAPK) pathways are implicated in cellular stress responses, including apoptosis and fibrosis within islet tissue. The accumulation of advanced glycation end products (AGEs) during hyperglycemic states further exacerbates oxidative stress by binding to their receptors (RAGE), thereby triggering downstream inflammatory pathways. Moreover, oxidative stress-induced mitochondrial dysfunction reduces ATP production, which is crucial for insulin secretion, thus worsening β-cell impairment. The convergence of these pathways results in a detrimental cycle of inflammation and oxidative stress that culminates in islet fibrosis, amyloid deposition, and β-cell loss, all of which are characteristic features associated with the pathogenesis of diabetes in the context of COVID-19 [65, 116].

Understanding the regulatory mechanisms that govern these biological pathways is essential for the development of targeted therapeutic strategies. In diseases such as COVID-19 and diabetes, the body’s antioxidant defenses—comprising superoxide dismutase (SOD), catalase, and glutathione peroxidase—often become insufficient. This inadequacy underscores the potential benefits of interventions aimed at restoring redox balance. Pharmacological methods that inhibit key elements of the RAAS or the nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) signaling pathway have shown promise in reducing inflammation and oxidative stress in preclinical investigations, suggesting their possible relevance in clinical settings. Furthermore, targeting oxidative stress pathways to modulate mitochondrial function and prevent pancreatic β-cell apoptosis may contribute to preserving islet integrity and functionality. Consequently, a combined approach that addresses both inflammation and oxidative stress may be crucial for mitigating pancreatic islet damage and the subsequent progression or worsening of diabetes in patients affected by COVID-19 (Fig. 9). This holistic viewpoint emphasizes the need for multifaceted management strategies that tackle both systemic and cellular stressors to improve outcomes for this vulnerable population [65, 116].

Fig. 9.

A vicious cycle of inflammation and oxidative stress in COVID-19-associated islet injury

Predictive modeling at the nexus of COVID-19 and diabetes: from clinical metrics to mechanistic precision

Transcending conventional variables: the integration of machine intelligence and molecular pathophysiology

The construction of risk prediction models for new-onset diabetes following SARS-CoV-2 infection has traditionally relied on a bedrock of static clinical variables—age, BMI, and inflammatory markers like CRP [117, 118]. However, the field is currently pivoting away from this rudimentary framework, recognizing that these macroscopic variables are often surrogates for deeper metabolic dysregulation rather than direct mechanistic drivers. While studies confirm that pre-existing hypertension and hyperglycemia are potent predictors of viral severity and subsequent metabolic collapse [117, 118], reliance on these broad clinical phenotypes limits the granularity required for precision medicine.

To bridge this gap, the next generation of predictive architectures is integrating multi-omics data to decipher the specific endotypes of post-COVID diabetes. The inclusion of polygenic risk scores and proteomic profiles—strategies successfully deployed in predicting pancreatic cancer within diabetic cohorts [119]—offers a template for SARS-CoV-2 research. Emerging longitudinal analyses suggest that subtle shifts in metabolite profiles precede overt hyperglycemia, offering a temporal window for intervention that standard laboratory metrics miss [120].

Technologically, the shift involves moving beyond linear assumptions. While traditional LASSO and Cox proportional hazards models effectively minimize overfitting in smaller datasets—such as predicting vaccine response in transplant recipients [121, 122]—they often fail to capture the non-linear, high-dimensional interactions inherent in viral-host dynamics. Advanced machine learning paradigms, including extreme gradient boosting (XGBoost) and deep neural networks, have demonstrated superior capacity in disentangling these complex relationships, accurately predicting viral shedding and antibody kinetics by synthesizing clinical data with immunological features [117, 123, 124]. Crucially, the “black box” nature of these algorithms is being addressed through explainable AI (XAI) frameworks like SHapley Additive exPlanations (SHAP), which dismantle the model’s decision-making process to validate that predictive power stems from pathophysiologically relevant signals rather than statistical artifacts [125] (Fig. 10).

Fig. 10.

A paradigm shift in predicting post-COVID diabetes by integrating multi-omics and explainable AI

The generalizability paradox: critical limitations of current architectures

Despite the proliferation of models boasting high internal validity—with AUC values ranging from 0.73 to 0.95 for severe outcomes [126–128] and 0.77 to 0.85 for diabetic complications [129, 130]—a critical evaluation reveals a pervasive “generalizability paradox.” Most existing algorithms are trained on homogenous, single-center cohorts, rendering them fragile when deployed across diverse genetic and environmental landscapes.

Three fundamental flaws currently impede the clinical utility of these models:

• Lack of mechanistic specificity: Many models rely heavily on generic inflammatory markers (e.g., LDH, CRP) that rise in any severe infection [126, 131]. Consequently, these tools often predict “severity” rather than specific pancreatic β-cell dysfunction, leading to high false-positive rates for diabetes-specific risk in the general sepsis population.

• Temporal and viral obsolescence: The predictive weights of variables often degrade as the virus evolves. Models calibrated on ancestral SARS-CoV-2 strains or unvaccinated populations frequently drift in accuracy when applied to Omicron-era cohorts or breakthrough infections, as the interplay between viral virulence and host immunity shifts [132, 133]. Static models fail to account for this dynamic evolutionary landscape.

• Demographic and ethnic bias: There is a stark divergence in risk factors across populations; Western models emphasize BMI and socioeconomic deprivation, whereas models derived from Asian cohorts highlight distinct genetic metabolic traits [134, 135]. The application of Eurocentric models to indigenous or minority groups, who often face structural healthcare disparities, risks exacerbating inequity rather than mitigating it [136, 137].

Optimization requires a paradigm shift from static to dynamic modeling. This involves the continuous ingestion of real-world data to update algorithms in near real-time, accounting for new variants and vaccination statuses [132, 133]. Furthermore, the integration of federated learning allows institutions to collaboratively train models on decentralized data, enhancing diversity and robustness without compromising patient privacy [130]. Future calibration must also prioritize distinct biomarkers of insulin resistance over generic inflammation to specifically identify the “diabetogenic” footprint of the virus [131, 138] (Fig. 11).

Fig. 11.

Overcoming the generalizability paradox: a framework for robust and equitable predictive models

Translational challenges: from algorithmic theory to clinical reality

The transition of predictive models from in silico validation to bedside application is fraught with systemic and ethical bottlenecks. While epidemiological data clearly define a 1.5- to 3-fold increase in diabetes risk post-infection [139, 140], and models utilizing glucose and oxygenation metrics show promise in forecasting DKA and hyperglycemia [141, 142], implementation is stalled by the “data quality chasm.”

Real-world clinical data is notoriously noisy. Reliance on retrospective Electronic Health Records (EHR) introduces significant bias due to inconsistent coding, missing data not at random, and variable testing frequencies [143, 144]. A model is only as robust as its training data; algorithms fed with incomplete phenotypic records often learn to predict healthcare utilization patterns rather than physiological risk. Furthermore, the distinction between Type 1, Type 2, and “stress hyperglycemia” remains blurred in many datasets, confusing the training labels and diluting predictive power for specific therapeutic interventions [145, 146].

Ethically, the deployment of these tools demands rigorous scrutiny regarding algorithmic fairness. If models are trained primarily on data from well-resourced academic centers, they may systematically underperform for racial minorities and lower-income populations who are disproportionately affected by both COVID-19 and DKA [144, 147]. Transparency in model design is therefore not merely a technical requirement but a bioethical imperative to prevent the automation of existing health disparities.

Ultimately, the successful clinical integration of these models depends on “actionability.” Clinicians face alert fatigue; thus, predictions must be coupled with clear, standardized intervention protocols—such as thresholds for endocrinology referral or specific glycemic monitoring schedules [148, 149]. Moving forward, the field must embrace a “living model” framework-one that is iteratively validated, ethnically inclusive, and mechanistically anchored—to transform high-dimensional data into tangible improvements in patient outcomes (Fig. 12).

Fig. 12.

Bridging the translational chasm: from predictive algorithms to clinically actionable insights

Future research directions and public health strategy recommendations

Necessity of multicenter large-scale cohort studies

The importance of conducting multicenter, large-scale cohort studies to explore the association between SARS-CoV-2 infection and the emergence of new-onset diabetes is highlighted by the intricate and varied clinical presentations of COVID-19, as well as its long-term consequences across different populations. Such extensive cohort studies with prolonged follow-up periods are instrumental in gathering comprehensive epidemiological data that encompass a wide array of demographic, clinical, and molecular factors affecting disease outcomes. For example, multicenter cohorts established in various geographic and healthcare environments have demonstrated differing incidence rates of diabetes following COVID-19, underscoring the necessity for large sample sizes and population diversity to derive generalizable and robust conclusions [150, 151]. These investigations allow for stratification based on age, sex, comorbidities, and viral variants, which is essential since the risk of developing diabetes post-infection varies according to these parameters [132, 152]. Additionally, long-term follow-up is critical for differentiating transient hyperglycemia associated with acute illness from chronic diabetes, as evidenced by studies monitoring glycemic outcomes several months post-infection [153]. In the absence of extended observation, the temporal patterns of diabetes onset and progression remain ambiguous, hindering the comprehension of causative factors and underlying pathophysiology.

Moreover, multicenter frameworks enhance statistical power and facilitate the amalgamation of diverse data sources, which include clinical parameters, laboratory biomarkers, and molecular profiles. This synthesis is crucial for uncovering the molecular mechanisms that contribute to beta-cell dysfunction or autoimmunity induced by SARS-CoV-2, aspects that may be insufficiently explored in smaller or single-center studies [154]. The inclusion of varied populations also enables the evaluation of the impact of genetic, environmental, and socioeconomic factors on disease susceptibility and outcomes, a consideration that is vital for effective risk prediction and the development of personalized medicine strategies. For instance, research has indicated that immunocompromised individuals or those with hematological malignancies display unique antibody responses and varying risks for reinfection, highlighting the necessity for customized risk assessment models [155].

Importantly, multicenter cohort studies highlight the critical need to combine clinical, epidemiological, and molecular biological data to construct comprehensive frameworks for risk prediction. These frameworks can encompass demographic factors, comorbidities, inflammatory markers, and viral genomic data to evaluate the probability of developing diabetes following SARS-CoV-2 infection [156, 157]. For example, the application of machine learning techniques in multicenter ICU cohorts has identified key predictors of COVID-19 outcomes, which could be utilized for diabetes risk assessment [156]. Additionally, the amalgamation of serological information with clinical findings enhances our understanding of vertical transmission and immune responses in particular populations, such as pregnant women, thereby emphasizing the significance of data integration [158] (Fig. 13). Overall, large-scale multicenter cohort studies with extended follow-up durations are crucial for capturing the intricate dynamics of SARS-CoV-2 infection and its sequelae, including the development of diabetes. These studies provide the necessary statistical rigor, population diversity, and data integration to clarify epidemiological patterns and molecular mechanisms, as well as to develop accurate risk prediction models. Such comprehensive approaches are essential for guiding clinical practices, shaping public health strategies, and steering future research directions in the post-pandemic context.

Fig. 13.

A large-scale, multicentre cohort study exploring the core mechanisms linking diabetes and SARS-CoV-2 infection

In-depth analysis of molecular mechanisms and novel therapeutic targets

The complex molecular pathways that enable SARS-CoV-2 infection and their association with the onset of DM highlight the importance of employing advanced technologies such as single-cell sequencing and spatial omics. These cutting-edge approaches are vital for clarifying the intricate pathological mechanisms occurring at both cellular and tissue levels. Single-cell sequencing facilitates an in-depth examination of diverse cell populations, revealing particular cell types and states affected by the viral infection, including pancreatic β-cells, immune cells, and endothelial cells, all of which are pivotal in regulating metabolism and inflammation. Concurrently, spatial omics adds a critical dimension of spatial context, allowing for the accurate localization of molecular changes within the microenvironments of impacted tissues, such as the pancreas and lungs. This enhances our comprehension of cell-cell interactions and localized immune responses that are essential to the disease’s progression [159, 160]. These sophisticated methodologies have already illuminated the dysregulation of immune-inflammatory pathways, endothelial dysfunction, and fibroblast activation, which may play a role in the pathogenesis of SARS-CoV-2 and the complications related to diabetes, thereby emphasizing the potential molecular interactions between viral infection and metabolic disturbances.

In this framework, concentrating on the modulation of inflammation and the preservation of β-cell function emerges as vital strategies for developing therapeutic interventions. The SARS-CoV-2 infection triggers a cytokine storm characterized by the excessive production of pro-inflammatory cytokines, including IL-6, TNF-α, and interferons. This exacerbates insulin resistance and harms β-cells, consequently hastening the onset of hyperglycemia and either triggering diabetes or worsening existing conditions [107, 161]. Novel pharmacological agents aimed at modulating these inflammatory pathways, such as p38 MAPKα inhibitors (e.g., pamapimod) and compounds that exhibit both anti-inflammatory and metabolic regulatory properties (e.g., pioglitazone), have shown synergistic antiviral effects against SARS-CoV-2 and its variants. This indicates their dual potential to inhibit viral replication while mitigating metabolic inflammation [162]. Additionally, targeting elements of the RAAS, particularly ACE2 and associated pathways like the apelin-ACE2 axis, offers a promising therapeutic strategy by reducing cardiorenal injuries and restoring metabolic equilibrium disrupted by SARS-CoV-2 [163]

The identification of molecular targets within both viral and host systems has markedly propelled the development of novel antiviral therapies. Notably, compounds that inhibit the viral main protease (Mpro), such as nirmatrelvir and ensitrelvir, have been structurally characterized to address resistance issues and maintain efficacy against newly emerging variants, thus laying the groundwork for future therapeutic strategies [164, 165]. Furthermore, antisense oligonucleotides (ASOs) designed to disrupt conserved viral RNA structures demonstrate potent inhibition of viral replication, representing an innovative approach to nucleic acid-based therapies [166, 167]. The exploration of natural compounds, including madecassic acid and flavonoids like naringin and hesperidin, through computational and molecular docking studies, illustrates their capability to interact with various viral entry factors and proteases, underscoring the potential of multi-targeted phytochemicals in combating SARS-CoV-2 [168, 169].

Importantly, the interaction between SARS-CoV-2 and host metabolic regulators, such as fatty acid-binding protein 4 (FABP4) and ATP citrate lyase (ACLY), has been associated with viral replication and pathogenesis, suggesting that metabolic enzymes could represent promising therapeutic targets aimed at the host [170, 171]. These findings underscore the necessity for a holistic therapeutic approach that combines antiviral agents with modulators of host metabolism and immune responses to effectively address the multifaceted nature of SARS-CoV-2 infection and its metabolic repercussions, including the onset of new diabetes cases. Briefly, the application of single-cell and spatial omics technologies to investigate the molecular pathology of SARS-CoV-2 infection has unveiled critical inflammatory and metabolic pathways amenable to therapeutic intervention. The development of new pharmacological agents targeting viral proteases, host entry factors, inflammatory signaling pathways, and metabolic regulators holds promise for mitigating both viral transmission and the onset or exacerbation of diabetes. The continued integration of molecular insights with advanced technological platforms will be essential for identifying and validating effective therapeutic targets, ultimately improving clinical outcomes for individuals affected by COVID-19 and associated metabolic disorders (Fig. 14).

Fig. 14.

Molecular pathways from SARS-CoV-2 infection to diabetes and therapeutic avenues

Public health interventions and diabetes prevention strategies

The COVID-19 pandemic has underscored the critical interplay between infectious diseases and NCDs, with a particular focus on diabetes, thereby highlighting the urgent need for comprehensive public health strategies. Individuals diagnosed with diabetes are known to possess an increased susceptibility to severe COVID-19 outcomes, and recent findings suggest a potential correlation between SARS-CoV-2 infection and the subsequent development of diabetes. Consequently, it is imperative to enhance diabetes screening and management protocols in the context of COVID-19 recovery. Evidence reveals that those with diabetes are at a heightened risk of contracting COVID-19, experiencing severe disease progression, and facing increased mortality rates, which accentuates the necessity for early identification and effective glycemic control to mitigate adverse health outcomes [172–174]. Therefore, public health initiatives should prioritize the enhancement of diabetes screening measures for individuals recovering from COVID-19, especially considering the potential for SARS-CoV-2 to provoke dysglycemia and initiate new-onset diabetes [175, 176].

Furthermore, it is essential for policymakers to recognize the underlying threat posed by the hidden diabetes epidemic, which may be exacerbated by the pandemic’s direct and indirect effects, including lifestyle disruptions, healthcare access limitations, and socioeconomic disparities. The pandemic has illustrated that the prevalence of diabetes and suboptimal metabolic health significantly affects the severity of COVID-19, with populations experiencing health inequities being disproportionately impacted [28, 177]. Therefore, holistic strategies for diabetes prevention and management should be incorporated into the frameworks for pandemic preparedness and response. These strategies ought to include public education campaigns aimed at increasing awareness of diabetes risk factors, encouraging healthy lifestyle choices such as proper nutrition and physical activity, and addressing the social determinants that influence diabetes incidence and management [178, 179]. In addition, the adoption of digital health solutions and telemedicine has demonstrated promise in improving diabetes self-management throughout the pandemic, thereby ensuring continuity of care in the context of social distancing measures [180, 181]. Efforts to boost vaccination rates among diabetic patients also require targeted strategies, as vaccine hesitancy and concerns regarding safety may impede immunization coverage within this at-risk group [173, 182]. Tailored communication approaches, which incorporate culturally sensitive messaging and engagement with trusted healthcare professionals, can enhance vaccine acceptance and adherence to preventive measures [183, 184].

Finally, it is imperative that policy development incorporates surveillance frameworks aimed at monitoring the emergence of new diabetes cases post-COVID-19, as well as evaluating the efficacy of the interventions that have been put in place. Promoting collaboration across various sectors, including healthcare providers, public health organizations, and community groups, is essential for the formulation and execution of evidence-based policies that effectively tackle both the immediate and long-term public health issues resulting from the interplay between COVID-19 and diabetes [185, 186] (Fig. 15). By recognizing the interdependent relationship between these two conditions and implementing thorough screening, management, and prevention initiatives, public health systems can mitigate the compounded challenges posed by these dual pandemics and improve health outcomes for the broader population.

Fig. 15.

A comprehensive public health framework to mitigate the concurrent epidemics of COVID-19 and diabetes

Conclusion

This review provides a critical synthesis of the distinct pathogenesis of COVID-19-associated diabetes, highlighting its unique mechanistic and clinical features that set it apart from traditional type 2 diabetes. Unlike previous reviews, we emphasize the novel etiopathogenetic triad: viral-mediated pancreatic injury (encompassing both potential direct and predominant indirect microvascular effects), systemic inflammation triggering insulin resistance, and glucocorticoid-induced hyperglycemia. Importantly, COVID-19-related diabetes demonstrates divergent molecular and immunological profiles, necessitating a reappraisal of conventional classification and management paradigms. Our analysis underscores the critical need for specialized diagnostic and therapeutic frameworks addressing this emerging condition. We propose that COVID-19-associated diabetes represents a distinct diabetes subtype requiring tailored interventions and long-term surveillance strategies. The review further highlights the urgent necessity of developing integrated prediction models combining clinical and omics-based biomarkers for early risk stratification. The broader significance of this work lies in its redefinition of the interplay between viral infections and metabolic diseases. By elucidating the unique pathophysiology of COVID-19-associated diabetes, this review provides a foundation for future research into targeted therapies and preventive measures. It calls for multidisciplinary collaboration among virologists, immunologists, endocrinologists, and public health experts to address the growing burden of post-COVID metabolic disorders. Ultimately, this synthesis advocates for a paradigm shift in how we perceive, diagnose, and manage diabetes in the context of pandemic infections, with important implications for clinical practice and health policy development.

Author contributions

HF prepared the initial draft of the review. QW and HF assisted with the drafting and editing process. QW and HF revised and compiled the final version. All authors have read and approved the final version of the manuscript.

Funding

This work was supported by grants from the Natural Science Foundation of Beijing Municipality (7232047 and 7242192) and the Beijing Municipal Administration of Hospitals Incubating Program (PX2023019).

Data availability

Data sharing is not applicable to this article as no data were created or analyzed in this study.

Declarations

Ethics and consent to participate

Not applicable.

Conflicts of interest

The authors declare no conflict of interest.