Beta Cell Function and Insulin Resistance: The Evidence Behind the Tool

A 38-year-old software engineer from Bengaluru brings a report to the outpatient clinic. His fasting glucose is 108 mg/dL and his fasting insulin 14 µIU/mL. Beside them, the laboratory has printed “HOMA-IR 3.73, insulin resistance present”. He is not overweight: his BMI is 22.5. But his waist is 94 cm. “Doctor, does this mean I have diabetes?” This is an illustrative case, not a real patient.

It does not, and the number on his report says less than it seems to. By the end of this article you will know what each of these indices measures, why no one can tell you where “resistant” begins, and when the numbers simply lie.

1. Two numbers that set each other

Fasting glucose and fasting insulin are set together. A resistant liver and muscle need more insulin to hold the same glucose; a beta cell that can supply it keeps the glucose normal at the cost of hyperinsulinaemia, and one that cannot lets the glucose rise. The homeostasis model of Matthews and colleagues, 1985, reads the pair against a computer model of that loop and reports two things.

“Two things from two numbers?” Yes: how resistant the body is to insulin, and how well the beta cell is keeping up. Here are the indices, with how each one was checked against the gold standard, the clamp study, in which glucose and insulin are infused and held steady in the laboratory.

For our engineer, worked by hand. HOMA-IR is 14 times 108, divided by 405: 3.73. HOMA-%B is 360 times 14, divided by (108 minus 63): 112 per cent. QUICKI is 1 divided by (log 14 plus log 108), about 0.31, below the reference 0.38, so on the resistant side. The reference point the formulae are built on is a fasting glucose of 81 mg/dL with an insulin of 5 µIU/mL, which gives exactly 1.0 and 100 per cent.

Notice the last column. A correlation of 0.88 is good for a population; a coefficient of variation of about a third means a single patient’s value can move a long way from one morning to the next. And Wallace, Levy and Matthews, one of them an author of the model, wrote in 2004 that reading beta cell function in isolation is a misuse.

2. The hyperbola: sensitivity times secretion

Kahn and colleagues, 1993, measured insulin sensitivity and first-phase insulin response in 93 healthy people, lean to obese, and found the two related as a hyperbola: sensitivity multiplied by secretion is constant. The product is the disposition index, and a person whose beta cell keeps pace with falling sensitivity slides along the curve; one whose beta cell does not falls beneath it.

“What does a hyperbola have to do with my patient?” It describes how a healthy beta cell behaves. When the tissues become more resistant, the beta cell makes more insulin, so the product stays the same and the glucose stays normal. The trouble starts when it can no longer keep pace. The figure in this tool redraws that shape with HOMA-derived axes, sensitivity as 100 divided by HOMA-IR and function as HOMA-%B, and the reference curve through the normal young adult’s point. Kahn used the minimal model and an intravenous glucose load, not HOMA; the transposition is this tool’s.

Here is the point that is easy to miss, and the tool keeps it at the bedside. Divide the %B formula by the IR formula and the insulin cancels out: what is left depends only on the fasting glucose. For our engineer, 145,800 divided by (108 times 45) is 30, and 112 divided by 3.73 is also 30. So a point below the curve says only that his glucose is above 81. It is not a second finding. What the figure does add is where along the curve he sits. With an insulin of 14, he is at the high-insulin end: a beta cell working hard against resistance, rather than a raised glucose with little insulin behind it, which would be deficiency.

3. There is no agreed cut-off

His report said “insulin resistance present”. By whose line? The 2010 insulin standardisation statement of Staten and colleagues says it plainly: there are no criteria by which an individual can be classified as insulin sensitive or resistant. Each population study draws its own line, usually at a centile of its own people. Look at how far apart they fall.

On the Thai line our engineer is resistant; on the French line he is not. A value of 2.5 is the figure most often quoted for adults, and it is also the figure from Indian adolescents in Delhi, which is why the tool names it. Nobody has published an Indian adult threshold made the same way. In young, lean South Indian men studied against the clamp at Vellore, HOMA-IR and QUICKI discriminated poorly. So treat any single threshold with caution, and read the number as a position on a scale, not a diagnosis.

4. When the numbers lie

The common mistake is to compute HOMA in a patient for whom the model was never meant. The insulin assay cannot tell injected insulin from the patient’s own, so HOMA is meaningless on insulin; the C-peptide measures the patient’s own secretion. A sulfonylurea drives the fasting insulin up, so a high %B on glibenclamide is the drug, not the gland. A non-fasting sample, a low glucose on the sample, kidney failure and a different laboratory’s insulin assay all move the numbers. The tool lists each of these; check them before you read the result.

5. The C-peptide is for classification

In insulin-treated diabetes the C-peptide answers a different question: is there enough of the patient’s own insulin left to matter? Jones and Hattersley set approximate thresholds in 2013, which the tool prints, and its main job is to separate type 1 from type 2 or monogenic diabetes three to five years after diagnosis. For how the C-peptide fits with the antibodies and the clinical picture, see the queue’s article Diabetes Classification: The Evidence Behind the Pathway.

6. The thin-fat Indian

Our engineer’s BMI of 22.5 would reassure most of us. It should not, and the reason begins before birth. Thin-fat. Rural Pune newborns were small in every measurement except the subscapular skinfold, which was relatively preserved (Yajnik 2003). At six years, Indian children had a lower BMI than white British children but more body fat and a higher HOMA-IR, by about 0.25 units, after adjusting for that fat (Lakshmi 2012).

That is why the 2009 Indian consensus moved the cut-offs: overweight at a BMI of 23, and abdominal obesity at a waist above 90 cm in men and 80 cm in women. The tool raises a thin-fat flag when the BMI is below 23 and the waist is above its cut-off, or HOMA-IR is 2.5 or above. Our engineer meets both. The pairing is the tool’s own, not a published criterion, and the tool says so.

“His glucose is only 108. Why worry now?” The decline is the disease. In UKPDS the fasting glucose rose on every therapy over six years and the rise tracked a progressive loss of beta cell function. A %B that looks acceptable today is a snapshot of a moving picture.

Back to our patient

He does not have diabetes. His fasting glucose of 108 is raised and his insulin is high: the tool reads that as a beta cell working hard, and losing. The index on his report is not a diagnosis, and its “resistance” depends on which population drew the line. Weight, waist and activity are the levers, and a value near a threshold is repeated rather than trusted. Nothing in this tool changes a drug on its own.

Try this

A 52-year-old woman with type 2 diabetes on glibenclamide has a fasting glucose of 140 mg/dL and a HOMA-%B of 180 per cent. Her family doctor reads this as “a strong pancreas”. Is it? Illustrative case.

Answer. No. A sulfonylurea drives the fasting insulin up, so a high %B on glibenclamide is the drug, not the gland. The number cannot be read as her own beta cell reserve, and the tool lists the sulfonylurea among the reasons the indices lie.

Take-home points

• HOMA-IR and HOMA-%B come from one fasting glucose and insulin; each varies by about a third from day to day.
• There is no agreed cut-off for insulin resistance; 2.5 is the figure most often quoted and an Indian adolescent figure.
• A point below the hyperbola only says the fasting glucose is above 81 mg/dL.
• Do not read HOMA on insulin, on a sulfonylurea, on a non-fasting sample or with a low glucose.
• In Indians a normal BMI does not exclude central adiposity: measure the waist.

For setting the HbA1c a treated patient should aim for, see the companion article The HbA1c Target: The Evidence Behind the Tool.

Declaration: the author built and maintains the tool. It is free to use. The figures in this article are those the tool carried in its Evidence tab on 5 November 2026, each from the source the tool names. The illustrative cases are worked by hand from the formulae the tool prints; check the tool’s result for a real patient.

References

1. Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28(7):412-419.
2. Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care. 2004;27(6):1487-1495.
3. Katz A, Nambi SS, Mather K, et al. Quantitative insulin sensitivity check index: a simple, accurate method for assessing insulin sensitivity in humans. J Clin Endocrinol Metab. 2000;85(7):2402-2410.
4. Kahn SE, Prigeon RL, McCulloch DK, et al. Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function. Diabetes. 1993;42(11):1663-1672.
5. Staten MA, Stern MP, Miller WG, Steffes MW, Campbell SE; Insulin Standardization Workgroup. Insulin assay standardization: leading to measures of insulin sensitivity and secretion for practical clinical care. Diabetes Care. 2010;33(1):205-206.
6. Gayoso-Diz P, Otero-González A, Rodriguez-Alvarez MX, et al. Insulin resistance (HOMA-IR) cut-off values and the metabolic syndrome in a general adult population: effect of gender and age: EPIRCE cross-sectional study. BMC Endocr Disord. 2013;13:47.
7. Jones AG, Hattersley AT. The clinical utility of C-peptide measurement in the care of patients with diabetes. Diabet Med. 2013;30(7):803-817.
8. Li X, Zhou ZG, Qi HY, Chen XY, Huang G. Replacement of insulin by fasting C-peptide in modified homeostasis model assessment to evaluate insulin resistance and islet beta cell function. Zhong Nan Da Xue Xue Bao Yi Xue Ban. 2004;29(4):419-423. Chinese.
9. Singh Y, Garg MK, Tandon N, Marwaha RK. A study of insulin resistance by HOMA-IR and its cut-off value to identify metabolic syndrome in urban Indian adolescents. J Clin Res Pediatr Endocrinol. 2013;5(4):245-251.
10. Sreekumar S, Eagappan S, Raghavan KS, Sridhar S. Single point insulin sensitivity estimator as a marker of metabolic syndrome in a South Indian cohort. Indian J Med Res. 2026;163(5):593-597.
11. Jebasingh FK, Anoop S, Dasgupta R, et al. Understanding the predictive accuracy of the InsuTAG index over other surrogate indices in normoglycaemic, non-obese males from Southern India. Sci Rep. 2023;13(1):19497.
12. Simental-Mendía LE, Rodríguez-Morán M, Guerrero-Romero F. The product of fasting glucose and triglycerides as surrogate for identifying insulin resistance in apparently healthy subjects. Metab Syndr Relat Disord. 2008;6(4):299-304.
13. Misra A, Chowbey P, Makkar BM, et al. Consensus statement for diagnosis of obesity, abdominal obesity and the metabolic syndrome for Asian Indians and recommendations for physical activity, medical and surgical management. J Assoc Physicians India. 2009;57:163-170.
14. Yajnik CS, Fall CH, Coyaji KJ, et al. Neonatal anthropometry: the thin-fat Indian baby. The Pune Maternal Nutrition Study. Int J Obes Relat Metab Disord. 2003;27(2):173-180.
15. Lakshmi S, Metcalf B, Joglekar C, Yajnik CS, Fall CH, Wilkin TJ. Differences in body composition and metabolic status between white U.K. and Asian Indian children (EarlyBird 24 and the Pune Maternal Nutrition Study). Pediatr Obes. 2012;7(5):347-354.
16. U.K. Prospective Diabetes Study Group. U.K. prospective diabetes study 16. Overview of 6 years’ therapy of type II diabetes: a progressive disease. Diabetes. 1995;44(11):1249-1258.
17. Abdesselam A, Zidoum H, Zadjali F, et al. Estimate of the HOMA-IR cut-off value for identifying subjects at risk of insulin resistance using a machine learning approach. Sultan Qaboos Univ Med J. 2021;21(4):604-612.