Skip to content

Type 2 diabetes — pathophysiology and subtypes

TL;DR — T2D emerges when β-cells can no longer compensate for insulin resistance; hyperglycaemia then amplifies β-cell dysfunction and metabolic flux abnormalities. The “twin-cycle” model links chronic energy surplus to liver fat, hepatic insulin resistance, hyperinsulinaemia and pancreatic fat, while calorie-restriction studies show that these processes can reverse in a subset (Taylor 2016, PMID 30058916; Taylor 2024, PMID 39038473). Data-driven clustering identifies reproducible phenotypic groups with different complication risks, but simple clinical variables often predict treatment response as well as cluster labels (Ahlqvist 2018, PMID 29503172; Dennis 2019, PMID 31047901). T2D is therefore mechanistically heterogeneous, yet precision treatment based on subtype remains unproven (Tobias 2023, PMID 37794253).

Core physiological defect

Glucose homeostasis depends on insulin secretion, insulin action, glucagon restraint, renal glucose handling, gastrointestinal signals and neural regulation. T2D is not a single-lesion disease: defects across these systems accumulate, while β-cell reserve determines when glycaemia crosses diagnostic thresholds (DeFronzo 2010, PMID 20206731).

Component Direction in established T2D Consequence Evidence boundary
Skeletal muscle Reduced insulin-stimulated glucose uptake Postprandial hyperglycaemia Physiological phenotype, not unique to T2D (DeFronzo 2010, PMID 20206731)
Liver Inadequate suppression of glucose production Fasting hyperglycaemia Closely linked to liver fat in remission studies (Taylor 2024, PMID 39038473)
β-cell Impaired glucose-responsive insulin secretion Failure to compensate for resistance Progressive but partly reversible early (Taylor 2021, PMID 33289165)
α-cell Inappropriate glucagon Increased hepatic glucose output One element of the multi-organ model (DeFronzo 2010, PMID 20206731)
Adipose tissue Increased lipolytic flux/ectopic fat delivery Lipotoxic stress and resistance Distribution matters beyond total BMI (Taylor 2024, PMID 39038473)
Gut Reduced/altered incretin effect Impaired meal-related insulin response Therapeutically targetable (Davies 2022, PMID 36148880)
Kidney Increased glucose reabsorption Maintains hyperglycaemia One element of the multi-organ model and target of SGLT2 inhibitors (DeFronzo 2010, PMID 20206731)
Brain Altered appetite/metabolic regulation Sustains positive energy balance Mechanisms overlap obesity biology (Taylor 2024, PMID 39038473)

β-cell failure is relative, not absolute

Most people with T2D retain measurable insulin secretion. “Failure” means secretion is insufficient for the prevailing insulin resistance and glucose load, not necessarily absence of insulin. Progressive glycaemic deterioration in conventional-treatment cohorts reflects declining functional reserve, but rapid calorie restriction can restore first-phase secretion in some early disease (Taylor 2016, PMID 30058916; Taylor 2021, PMID 33289165).

Hyperglycaemia itself worsens secretion (“glucotoxicity”), so measurements taken during severe decompensation can underestimate recoverable reserve. C-peptide must therefore be interpreted with concurrent glucose, renal function, medication exposure and disease duration.

The twin-cycle model

Proposed step Testable observation Status
Chronic energy surplus raises liver fat Liver fat falls rapidly during severe calorie restriction Supported in experimental-remission programmes (Taylor 2016, PMID 30058916)
Hepatic insulin resistance raises fasting glucose and insulin Fasting glucose can normalise before maximum weight loss Supported, but timing varies (Taylor 2021, PMID 33289165)
Increased VLDL export delivers fat to pancreas Pancreatic fat falls in responders Association supports model; causal cellular pathway remains debated (Al-Mrabeh 2020, PMID 33225228)
β-cell function recovers below an individual fat threshold Remission probability rises steeply with weight loss DiRECT provides clinical support (Lean 2018, PMID 29221645)
Susceptibility differs at the same BMI T2D occurs across BMI ranges and ethnic groups “Personal fat threshold” is plausible but not a routine biomarker (Taylor 2024, PMID 39038473)

The model explains rapid reversibility better than an irreversible β-cell-loss account. It does not explain all T2D: lean phenotypes, lipodystrophy, medication-induced disease, monogenic diabetes and latent autoimmune diabetes require different causal frames.

Heterogeneity and classification

The Swedish ANDIS analysis clustered adult-onset diabetes using age, BMI, HbA1c, GAD antibodies, HOMA2 insulin resistance and HOMA2 β-cell function (Ahlqvist 2018, PMID 29503172).

Cluster label Dominant phenotype Reported risk signal Main limitation
SAID Autoimmune, insulin-deficient Resembles adult autoimmune diabetes Not T2D by mechanism (Ahlqvist 2018, PMID 29503172)
SIDD Severe insulin deficiency, high HbA1c Retinopathy signal Cluster membership changes with measurement and treatment
SIRD Severe insulin resistance Kidney and fatty-liver signals HOMA estimates are context-dependent
MOD Mild obesity-related diabetes Slower progression on average Label can obscure within-group heterogeneity
MARD Older onset, relatively mild Lower short-term metabolic severity Age dominates assignment

Replication studies support heterogeneity, but clustering converts continuous variables into categories. Trial-data analysis found that simple models using age, sex, BMI, HbA1c and renal function could outperform cluster assignment for predicting glycaemic progression and treatment response (Dennis 2019, PMID 31047901).

Genetics and mechanistic pathways

Large multi-ancestry genetic analyses partition T2D susceptibility into pathways related to insulin secretion, insulin action, adiposity and lipid metabolism (Suzuki 2024, PMID 38374256). Genetic architecture demonstrates biological heterogeneity but does not yet yield a routine drug-selection test.

Precision layer Potential use Current constraint
Monogenic diagnosis Changes treatment in selected atypical cases Requires correct pretest selection and genetic interpretation
Autoantibodies Detect autoimmune diabetes misclassified as T2D Antibody-negative autoimmune disease and titre decay remain issues
C-peptide Measures endogenous secretion Glucose and renal function confound interpretation
Polygenic score Risk stratification before disease Incremental utility over age, BMI, family history and glycaemia is modest in many settings
Phenotypic clusters Prognosis and trial enrichment A live search through 2026-08-30 found validation/prediction studies but no prospective cluster-guided outcome strategy trial (Tobias 2023, PMID 37794253)
Multi-omics Mechanistic discovery Cost, reproducibility and population transferability (Carrasco-Zanini 2024, PMID 37889320)

Disease stage and reversibility

Stage Dominant process Reversible component
Prediabetes/high risk Resistance with compensatory secretion Weight, activity, diet and some medication effects (Knowler 2002, PMID 11832527)
Early diagnosed T2D Resistance plus recoverable β-cell dysfunction Substantial weight loss can induce remission (Lean 2018, PMID 29221645)
Longer-duration T2D Reduced β-cell reserve plus complications Glycaemia improves with weight loss, but remission probability falls
Decompensated T2D Glucotoxicity, dehydration, relative insulin deficiency Acute metabolic correction can restore apparent reserve

Remission is not cure. Weight regain can re-establish ectopic-fat flux and hyperglycaemia, and vascular risk surveillance continues even after HbA1c falls below diagnostic thresholds (Riddle 2021, PMID 34462270; Lean 2024, PMID 38423026).

Competing and complementary models

  • The multi-organ “ominous octet” is a functional inventory and maps drug targets (DeFronzo 2010, PMID 20206731).
  • The twin-cycle model is a causal energy-storage account with experimental-remission predictions (Taylor 2024, PMID 39038473).
  • Cluster models describe heterogeneity statistically but do not necessarily identify causal subtypes (Ahlqvist 2018, PMID 29503172).
  • Genetic pathway models identify lifelong susceptibility but incompletely capture environment and disease stage (Suzuki 2024, PMID 38374256).

These models are not mutually exclusive. A patient may carry secretion-related genetic risk, develop ectopic fat at a comparatively low BMI, and occupy a severe-insulin-deficient statistical cluster.

Experimental readouts

Process Human readout Constraint
Insulin secretion C-peptide response Depends on glucose and renal clearance
Insulin resistance Clamp, tolerance test, HOMA Precision and feasibility trade-off
Liver fat MRI spectroscopy/imaging Cost and threshold uncertainty
Pancreatic fat MRI-based quantification Small organ and technical variability
Glucose flux Tracer studies Research-intensive
β-cell identity Islet tissue/single-cell data Human sampling and disease-stage bias

Feedback loops

  1. Insulin resistance raises secretory demand.
  2. Compensatory hyperinsulinaemia can sustain normoglycaemia temporarily.
  3. β-cell dysfunction raises glucose.
  4. Hyperglycaemia and excess lipid flux further impair secretion/action.
  5. Decompensation increases dehydration and counter-regulatory hormones.
  6. Weight loss or acute insulin can interrupt parts of the loop without erasing susceptibility.

Clinical implications without overclaiming subtype

  • Suspected insulin deficiency changes urgency and makes ketosis safety central.
  • CKD/HF/ASCVD phenotype guides organ-protective therapy even without mechanistic cluster assignment.
  • Large sustained weight loss tests reversibility but does not guarantee remission.
  • Atypical presentation should reopen classification rather than force every case into T2D.
  • Continuous traits should remain continuous in analysis unless categories improve decisions.

Phenotypic and genetic classification reviews emphasise that causal mechanism, statistical phenotype and billing diagnosis are different layers (Deutsch 2022, PMID 35953726).

Cross-domain evidence crosswalk

These adjacent studies constrain interpretation of this page and make explicit where its conclusions depend on prevention, organ-outcome, remission, burden or implementation evidence.

Verified evidence anchor Connection
(Diabetes 2015, PMID 26377054) Diabetes Prevention Program Research Group. Long-term effects of lifestyle intervention or metformin on diabetes development and microvascular complications over 15-year follow-up: the DPP Outcomes Study. Lancet Diabetes Endocrinol. 2015
(TODAY 2021, PMID 34320286) TODAY Study Group. Long-Term Complications in Youth-Onset Type 2 Diabetes. N Engl J Med. 2021
(TODAY 2012, PMID 22540912) TODAY Study Group. A clinical trial to maintain glycemic control in youth with type 2 diabetes. N Engl J Med. 2012
(GRADE 2022, PMID 36129997) GRADE Study Research Group. Glycemia Reduction in Type 2 Diabetes. N Engl J Med. 2022
(Adler 2024, PMID 38772405) Adler AI, et al. UKPDS 91: 24-year post-trial monitoring. Lancet. 2024;404:145-155
(Palmer 2021, PMID 33441402) Palmer SC, et al. SGLT2 inhibitors and GLP-1RA network meta-analysis. BMJ. 2021
(Speight 2024, PMID 38128969) Speight J, et al. Bringing an end to diabetes stigma and discrimination: an international consensus statement on evidence and recommendations. Lancet Diabetes Endocrinol. 2024
(GBD 2023, PMID 37356446) GBD 2021 Diabetes Collaborators. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050. Lancet. 2023;402:203-234

Dynamic natural history: compensation, acceleration and failure

The central transition is failure of β-cell compensation for insulin resistance, not insulin resistance alone. β-cells initially expand secretion, but glucolipotoxicity, endoplasmic-reticulum stress, oxidative stress, islet inflammation, amyloid and altered cell identity can progressively reduce functional mass (Alejandro 2015, PMID 25542976; Halban 2014, PMID 24712577). DeFronzo's broader “ominous octet” adds dysregulated glucagon, incretin signalling, renal glucose reabsorption, adipose lipolysis and central appetite/metabolic signalling to muscle, liver and β-cell defects (DeFronzo 2009, PMID 19336687).

Whitehall II shows that progression is nonlinear. Among 505 incident cases, fasting glucose rose from 5.79 to 7.40 mmol/L and 2-hour glucose from 7.60 to 11.90 mmol/L during an acceleration beginning about three years before diagnosis; HOMA β-cell function rose from 85.0% to 92.6% and then fell to 62.4% at diagnosis (Tabák 2009, PMID 19515410). South Asian participants had a faster fasting-glucose increase (difference 0.22 mmol/L per decade, 95% CI 0.02–0.42) and more rapid loss of insulin sensitivity than White participants, showing that one trajectory cannot be assumed across ancestry (Hulman 2017, PMID 28409212).

Reversibility as a mechanistic experiment

Counterpoint tested 11 people with T2D on 600 kcal/day. Fasting glucose fell from 9.2 to 5.9 mmol/L after one week; hepatic triglyceride fell from 12.8% to 2.9% by week eight, pancreatic triglyceride from 8.0% to 6.2%, and first-phase insulin response rose from 0.19 to 0.46 nmol·min⁻¹·m⁻² (Lim 2011, PMID 21656330). This small, uncontrolled physiology study supports the twin-cycle model but does not establish universal causation or long-term durability (Taylor 2013, PMID 23075228).

Mechanistic proposition Supporting evidence Counterweight
Ectopic fat is causal and removable Rapid hepatic-insulin-sensitivity and secretion recovery during energy restriction (Lim 2011, PMID 21656330) n=11; selected, short-duration disease; weight loss changes many pathways simultaneously
β-cell failure is progressive Human longitudinal trajectories and cellular stress mechanisms (Tabák 2009, PMID 19515410; Halban 2014, PMID 24712577) Functional recovery in remission shows that “failure” is partly reversible
Discrete clusters represent diseases Cluster replications associate groups with different complication risks in European and Chinese cohorts (Xing 2021, PMID 34276555; Wang 2023, PMID 37683311) Variables are continuous, treatment- and time-dependent; boundaries shift
Genetics can specify treatment 1.4-million-person analysis identified 318 additional loci and complication-linked polygenic signals (Vujkovic 2020, PMID 32541925) Association and target discovery are not evidence that genotype-guided prescribing improves outcomes

Pharmacogenetic research has produced credible drug–gene associations, but clinical utility requires a strategy trial showing that genotyping improves outcomes beyond ordinary phenotype and history (Florez 2017, PMID 28283684). Reviews of T2D genetics similarly distinguish locus discovery from transportable prediction, especially where discovery cohorts underrepresent ancestries bearing much of the global burden (Laakso 2022, PMID 35956377).

Subtype controversy: clusters or continuous traits?

Data-driven clusters reproduce differences in insulin deficiency, resistance, obesity and complication incidence, but a cluster label can lose information contained in the underlying continuous variables. The relevant comparison is therefore not “cluster versus no precision,” but cluster assignment versus a parsimonious model of age at diagnosis, BMI, HbA1c, C-peptide/insulin-resistance proxies, kidney function and comorbidity. The audit's live search through 2026-08-30 retrieved cluster validation and response-prediction studies, but no randomised cluster-use-versus-clinical-feature treatment strategy; clusters are therefore useful research summaries rather than validated treatment rules.

Open questions

  • Can prospective assignment to a mechanistic subtype improve hard outcomes compared with treatment based on clinical features alone? Current clustering evidence is retrospective (Ahlqvist 2018, PMID 29503172; Dennis 2019, PMID 31047901).
  • Which β-cell changes distinguish durable remission from temporary glycaemic normalisation? Five-year relapse remains frequent (Lean 2024, PMID 38423026).
  • Can multi-ancestry genetic pathways identify differential response to incretin, SGLT2 or insulin therapy? Translation remains a stated precision-medicine gap (Suzuki 2024, PMID 38374256; Tobias 2023, PMID 37794253).
  • What is the best operational measure of the “personal fat threshold” across ancestries and body compositions? BMI is an inadequate proxy (Taylor 2024, PMID 39038473).

References

  1. DeFronzo RA. Current issues in the treatment of type 2 diabetes. Am J Med. 2010;123:S38-S48. PMID 20206731
  2. Taylor R. Calorie restriction and reversal of type 2 diabetes. Expert Rev Endocrinol Metab. 2016. PMID 30058916
  3. Taylor R. Type 2 diabetes and remission: practical management guided by pathophysiology. J Intern Med. 2021;289:754-770. PMID 33289165
  4. Taylor R. Understanding the cause of type 2 diabetes. Lancet Diabetes Endocrinol. 2024;12:664-673. PMID 39038473
  5. Al-Mrabeh A. Pathogenesis and remission of type 2 diabetes: what has the twin cycle hypothesis taught us? Cardiovasc Endocrinol Metab. 2020;9:132-142. PMID 33225228
  6. Lean MEJ, et al. Primary care-led weight management for remission of type 2 diabetes. Lancet. 2018;391:541-551. PMID 29221645
  7. Lean MEJ, et al. 5-year follow-up of the Diabetes Remission Clinical Trial. Lancet Diabetes Endocrinol. 2024. PMID 38423026
  8. Riddle MC, et al. Definition and Interpretation of Remission in Type 2 Diabetes. Diabetes Care. 2021. PMID 34462270
  9. Ahlqvist E, et al. Novel subgroups of adult-onset diabetes. Lancet Diabetes Endocrinol. 2018;6:361-369. PMID 29503172
  10. Dennis JM, et al. Disease progression and treatment response in data-driven subgroups of type 2 diabetes. Lancet Diabetes Endocrinol. 2019;7:442-451. PMID 31047901
  11. Deutsch AJ, et al. Phenotypic and genetic classification of diabetes. Diabetologia. 2022;65:1758-1769. PMID 35953726
  12. Suzuki K, et al. Genetic drivers of heterogeneity in type 2 diabetes pathophysiology. Nature. 2024;627:347-357. PMID 38374256
  13. Tobias DK, et al. Second international consensus report on precision diabetes medicine. Nat Med. 2023. PMID 37794253
  14. Carrasco-Zanini J, et al. Multi-omic prediction of incident type 2 diabetes. Diabetologia. 2024;67:102-112. PMID 37889320
  15. Knowler WC, et al. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med. 2002;346:393-403. PMID 11832527
  16. Davies MJ, et al. Management of Hyperglycemia in Type 2 Diabetes, 2022. Diabetes Care. 2022. PMID 36148880
  17. Diabetes Prevention Program Research Group. Long-term effects of lifestyle intervention or metformin on diabetes development and microvascular complications over 15-year follow-up: the DPP Outcomes Study. Lancet Diabetes Endocrinol. 2015. PMID 26377054
  18. TODAY Study Group. Long-Term Complications in Youth-Onset Type 2 Diabetes. N Engl J Med. 2021. PMID 34320286
  19. TODAY Study Group. A clinical trial to maintain glycemic control in youth with type 2 diabetes. N Engl J Med. 2012. PMID 22540912
  20. GRADE Study Research Group. Glycemia Reduction in Type 2 Diabetes. N Engl J Med. 2022. PMID 36129997
  21. Adler AI, et al. UKPDS 91: 24-year post-trial monitoring. Lancet. 2024;404:145-155. PMID 38772405
  22. Palmer SC, et al. SGLT2 inhibitors and GLP-1RA network meta-analysis. BMJ. 2021. PMID 33441402
  23. Speight J, et al. Bringing an end to diabetes stigma and discrimination: an international consensus statement on evidence and recommendations. Lancet Diabetes Endocrinol. 2024. PMID 38128969
  24. GBD 2021 Diabetes Collaborators. Global, regional, and national burden of diabetes from 1990 to 2021, with projections of prevalence to 2050. Lancet. 2023;402:203-234. PMID 37356446
  25. DeFronzo RA. From the triumvirate to the ominous octet. Diabetes. 2009;58:773-795. PMID 19336687
  26. Alejandro EU, et al. Natural history of β-cell adaptation and failure in type 2 diabetes. Mol Aspects Med. 2015;42:19-41. PMID 25542976
  27. Halban PA, et al. β-cell failure in type 2 diabetes: postulated mechanisms and prospects. J Clin Endocrinol Metab. 2014;99:1983-1992. PMID 24712577
  28. Tabák AG, et al. Trajectories of glycaemia, insulin sensitivity, and insulin secretion before diagnosis of type 2 diabetes. Lancet. 2009;373:2215-2221. PMID 19515410
  29. Hulman A, et al. Glycaemic trajectories in South Asian and White individuals before type 2 diabetes. Diabetologia. 2017;60:1252-1260. PMID 28409212
  30. Lim EL, et al. Reversal of type 2 diabetes: normalisation of beta-cell function with decreased pancreas and liver triacylglycerol. Diabetologia. 2011;54:2506-2514. PMID 21656330
  31. Taylor R. Reversing the twin cycles of type 2 diabetes. Diabet Med. 2013;30:267-275. PMID 23075228
  32. Xing L, et al. Clinical Characteristics and Risk of Diabetic Complications in Data-Driven Clusters. Front Endocrinol. 2021;12:617628. PMID 34276555
  33. Wang W, et al. Data-driven subgroups of newly diagnosed type 2 diabetes in Chinese adults. Diabetes Metab Syndr. 2023;17:102850. PMID 37683311
  34. Florez JC. Pharmacogenetics in type 2 diabetes: precision medicine or discovery tool? Diabetologia. 2017;60:800-807. PMID 28283684
  35. Laakso M, Fernandes Silva L. Genetics of Type 2 Diabetes: Past, Present, and Future. Nutrients. 2022;14. PMID 35956377
  36. Vujkovic M, et al. Discovery of 318 new risk loci for type 2 diabetes and related vascular outcomes. Nat Genet. 2020;52:680-691. PMID 32541925