Insulin index
The insulin index (II) is a dietary metric that quantifies the postprandial insulin response elicited by consuming a 1000 kJ (approximately 240 kcal) portion of a food, expressed relative to white bread, which is assigned a reference value of 100.[1] Developed in 1997 by researchers Susanne Holt, Janette Brand Miller, and Peter Petocz, it provides a standardized way to assess how various foods—beyond just carbohydrates—stimulate insulin secretion in healthy individuals.[1] The methodology involves measuring plasma insulin concentrations at regular intervals (typically every 15 minutes) over 120 minutes following food consumption, with the insulin score calculated as the incremental area under the insulin response curve (iAUC) compared to the reference food.[1] In the original study, 38 common foods from six categories (fruits, bakery products, snacks, carbohydrate-rich foods, protein-rich foods, and breakfast cereals) were tested on 11–13 healthy subjects, revealing substantial variability in insulin responses both within and across categories.[1] Unlike the glycemic index (GI), which focuses solely on blood glucose elevation primarily from carbohydrates, the II accounts for the insulinogenic effects of proteins and fats, showing that protein-rich foods (e.g., eggs, fish) and certain bakery items (e.g., croissants) can provoke insulin responses disproportionate to their glycemic impact.[1] The II correlates moderately with the GI (r = 0.70, P < 0.001), but carbohydrate content (r = 0.39, P < 0.05) and sugar (r = 0.36, P < 0.05) positively influence insulin scores, while fat and protein show inverse but non-significant trends.[1] Subsequent research has expanded the II into the food insulin index (FII), an algorithm applicable to mixed meals and broader dietary patterns, enabling calculations of dietary insulin index (DII) and dietary insulin load (DIL) to predict overall insulin demand. High DII and DIL have been associated with increased risks of insulin resistance, metabolic syndrome, type 2 diabetes, and cardiometabolic disturbances in cohort studies from diverse populations, including Iranian adults. For instance, observational data link elevated insulinemic potential to greater inflammation markers and body weight gain, underscoring the II's relevance beyond carbohydrate-focused metrics. In clinical applications, the FII aids in personalized nutrition for diabetes management, outperforming traditional carbohydrate counting for predicting postprandial insulin excursions and optimizing insulin dosing in type 1 diabetes. It supports dietary strategies to mitigate hyperinsulinemia, such as favoring low-II foods like yogurt or lentils over high-II options like potatoes or white rice, potentially reducing long-term complications like cardiovascular disease. Ongoing studies emphasize the need for larger, diverse trials to refine II databases and explore gene-diet interactions, but its integration into glycemic control protocols highlights its growing utility in preventive nutrition.Background
Definition
The insulin index (II) is a metric that quantifies the postprandial insulin response elicited by consuming a food, specifically measuring the area under the curve (AUC) of blood insulin concentration over a 2-hour period following ingestion of an isoenergetic portion equivalent to 1000 kJ (approximately 240 kcal).[2] This value is then expressed as a percentage relative to the insulin response from an equivalent portion of white bread, which is assigned an II of 100 as the reference standard.[2] The primary purpose of the II is to evaluate how various foods stimulate insulin secretion independently of their carbohydrate content, thereby capturing the contributions of proteins, fats, and other macronutrients to overall insulin demand.[2] Unlike the glycemic index, which focuses on blood glucose excursions, the II emphasizes insulinemic effects, revealing that certain protein-rich foods can provoke substantial insulin release with minimal impact on glycemia.[2] This focus on insulin demand makes the II particularly relevant for understanding metabolic responses in conditions such as insulin resistance and hyperinsulinemia, where excessive insulin secretion may contribute to disease progression, and for informing dietary strategies in non-insulin-dependent diabetes mellitus management.[2]History
The insulin index (II) was developed in the 1990s by Susanne Holt, Jennie Brand-Miller, and Peter Petocz at the University of Sydney's Human Nutrition Unit, as a metric to quantify the postprandial insulin response to foods beyond carbohydrate content alone.[1] This work built on earlier glycemic index research to address the insulinogenic effects of proteins and fats. The concept was formally introduced in a seminal 1997 study published in the American Journal of Clinical Nutrition, where the researchers tested isoenergetic 1000-kJ portions of 38 common foods in healthy subjects, establishing II values relative to white bread, which was assigned a reference value of 100.[1] This publication provided the foundational methodology and demonstrated that foods like dairy and proteins elicited unexpectedly high insulin responses, highlighting II's potential for nutritional assessment.[1] In the following decades, the II evolved from single-food evaluations to applications for mixed meals and expanded databases. A key advancement came in 2009 with a study validating the food insulin index for predicting insulin demand from composite meals, showing that II-based calculations accurately estimated responses in real-world eating scenarios better than carbohydrate-focused metrics.[3] During the 2010s, researchers broadened II testing to include more diverse foods, incorporating it into dietary load calculations (e.g., dietary insulin load) to evaluate overall meal and daily insulin demands.[4] By the 2020s, comprehensive compilations emerged, such as a 2023 collectanea aggregating II values for 629 food and beverage items from 80 studies, facilitating clinical use in personalized nutrition planning.[5] Post-2000, II gained prominence in nutritional research on diabetes and obesity, with studies linking high dietary insulin indices to increased risk of insulin resistance, weight gain, and metabolic disorders.[4] For instance, analyses of dietary insulin load using II data showed associations with biomarkers of obesity and type 2 diabetes in population cohorts.[6] Recent investigations have further explored II variations across protein sources, revealing differences in insulin responses to plant-based versus animal-based proteins, which inform strategies for managing glycemic control and metabolic health; for example, a 2025 clinical trial demonstrated that animal-based proteins result in higher energy expenditure and carbohydrate oxidation compared to plant-based proteins.[7][8]Measurement
Methodology
The methodology for determining the insulin index involves controlled human feeding studies designed to measure postprandial insulin responses to specific foods. In the seminal protocol developed at the University of Sydney, healthy, non-diabetic volunteers—typically 10 to 13 young adults per food category, with normal body mass index (mean 22.7 kg/m²)—undergo testing after a 10-hour overnight fast to ensure baseline insulin levels are standardized.[1] On separate test days, participants consume an isoenergetic portion of the test food equivalent to 1000 kJ (approximately 240 kcal), accompanied by 220 mL of water, while remaining seated to minimize physical activity influences.[1] Blood samples are collected via finger-prick at baseline and at 15-minute intervals up to 120 minutes post-consumption, with plasma insulin concentrations measured using radioimmunoassay techniques, such as the Coat-A-Count kit, which offers low coefficients of variation (within-assay 5%, between-assay 7%).[1] Subsequent studies have adopted similar protocols but may employ enzyme-linked immunosorbent assay (ELISA) for insulin quantification to enhance sensitivity and reduce radioactivity concerns.[9] To ensure reproducibility and accuracy, foods are prepared in bulk to precise energy content based on nutritional databases or manufacturer data, served in standardized portions (e.g., sliced or reheated as needed), and presented under controlled conditions, such as an opaque hood where feasible, to limit anticipatory cephalic-phase insulin release.[1] The reference food, typically white bread, is tested on alternate days in a randomized order across sessions, allowing each participant to serve as their own control within food groups.[1] Pre-testing standardization includes instructions for participants to maintain consistent physical activity, avoid alcohol and legumes the previous evening, and consume similar low-fat meals the night before, with all tests conducted at the same time of day to account for circadian variations.[1] Variability in insulin responses is addressed by averaging individual area-under-the-curve values across multiple subjects and repeating tests as needed for reliability, with statistical analyses like two-way ANOVA used to quantify interindividual differences.[1] These studies are conducted in controlled clinical or laboratory settings, with protocols approved by institutional ethics committees, such as the Human Research Ethics Committee of the University of Sydney, ensuring informed consent and participant safety.[10] Early investigations noted limitations in subject diversity, primarily involving young university students of similar demographics, which may influence generalizability to broader populations.[1]Calculation
The insulin index (II) is computed as a percentage relative to a reference food, quantifying the insulin response elicited by a test food compared to the reference. The formula is: \text{II} = \left( \frac{\text{AUC}_{\text{insulin, test food}}}{\text{AUC}_{\text{insulin, reference food}}} \right) \times 100 where \text{AUC} denotes the incremental area under the 120-minute insulin concentration-time curve above the fasting baseline. This approach normalizes the insulin demand of isocaloric portions (typically 1000 kJ) across foods, with white bread assigned an II of 100 by definition to serve as the standard for comparability. The AUC is estimated using the trapezoidal rule, integrating insulin concentrations measured over time while subtracting the preprandial fasting level to focus on the postprandial increment; any negative excursions below baseline are truncated to zero to avoid underestimation. Individual responses from multiple subjects (typically 11–13 per food) are averaged to yield the mean II, with standard errors reported to indicate variability and account for inter-subject differences in insulin sensitivity. For example, if the AUC for a test food is 50% of that for the white bread reference, the resulting II is 50, signifying a moderate insulinogenic effect compared to the standard.Food Insulin Index Values
Protein-Rich Foods
Protein-rich foods generate notable insulin responses despite containing minimal carbohydrates, a phenomenon captured by the insulin index (II), which quantifies the postprandial insulin secretion relative to an equal-energy portion of white bread. In the foundational 1997 study by Holt et al., protein sources were shown to stimulate insulin primarily through specific amino acids, such as leucine and other branched-chain amino acids, which directly promote beta-cell secretion in the pancreas independent of blood glucose elevation.[1] This mechanism explains why the mean II for protein-rich foods was 61, higher than anticipated based on glycemic effects alone.[11] Representative II values from this study illustrate the variability among protein sources. Beef elicited an II of 51, white fish 59, eggs 31, and cheese 45, demonstrating moderate to substantial insulin demand even without significant carbohydrate content. Dairy proteins exhibited the highest responses in the cohort, with yogurt reaching an II of 115, attributed to its rapid digestion and amino acid profile.[1] These values underscore how proteins can drive insulin secretion comparably to some carbohydrate-rich foods when portioned by energy.[11] Comparisons between animal and plant proteins reveal distinct patterns, with animal sources often producing stronger insulinogenic effects due to higher concentrations of branched-chain amino acids. A 2023 review of postprandial responses confirmed that animal proteins, particularly dairy-derived ones like whey, yield higher insulin excursions than plant counterparts such as soy or pea protein.[12] For example, soy-based products have been measured with lower II values, reflecting their differing amino acid composition and slower absorption. This disparity persists across low-carbohydrate contexts, where protein-induced insulin elevation remains prominent.[12] The following table compiles II values for selected protein-rich foods, drawn primarily from the 1997 Holt study and supplemented by subsequent measurements for broader representation:| Food | Insulin Index (II) | Source |
|---|---|---|
| Yogurt | 115 | Holt et al. (1997)[1] |
| Beef | 51 | Holt et al. (1997)[1] |
| White Fish | 59 | Holt et al. (1997)[1] |
| Lentils | 58 | Holt et al. (1997)[1] |
| Cheese | 45 | Holt et al. (1997)[1] |
| Eggs | 31 | Holt et al. (1997)[1] |
Carbohydrate-Rich Foods
Carbohydrate-rich foods exhibit a wide range of insulin index (II) values, reflecting variations in starch structure, fiber content, and processing that influence postprandial insulin secretion beyond simple carbohydrate content. In the seminal study by Holt et al., isoenergetic 1000-kJ portions of 38 common foods were tested, establishing white bread as the reference with an II of 100. Among carbohydrates, starchy foods like potatoes elicited a notably high II of 121 ± 11, exceeding the reference due to rapid digestion and absorption, while pasta showed a surprisingly low II of 40 ± 5 for both white and brown varieties, attributed to slower gastric emptying and lower glycemic impact.[1] Patterns in II for carbohydrate-rich foods highlight that highly processed or low-fiber starches often provoke stronger insulin responses, sometimes surpassing expectations from glycemic index alone. For instance, boiled potatoes' high II contrasts with the moderate responses from fruits such as bananas (II = 81 ± 5) and oranges (II = 60 ± 3), where natural sugars and fiber moderate insulin demand. Legumes like lentils (II = 58 ± 12) further demonstrate lower II values, influenced by high fiber and protein content that slows carbohydrate breakdown. These deviations underscore how II integrates insulinogenic effects not fully captured by glucose excursions, providing a broader view of metabolic impact.[1]| Food Category | Representative Foods | Insulin Index (II) | Key Influence |
|---|---|---|---|
| Breads and Grains | White bread (reference) | 100 | Baseline for refined carbs |
| Whole-meal bread | 96 ± 12 | Slight fiber moderation | |
| White rice | 79 ± 12 | Moderate starch digestibility | |
| Brown rice | 62 ± 11 | Higher fiber reduces response | |
| Starchy Vegetables | Potatoes (boiled) | 121 ± 11 | Rapid absorption elevates II |
| French fries (oven-baked) | 74 ± 12 | Fat and processing temper response | |
| Pasta | White pasta | 40 ± 5 | Slow digestion lowers II |
| Brown pasta | 40 ± 5 | Similar to white despite fiber | |
| Fruits | Bananas | 81 ± 5 | Fructose and fiber balance |
| Apples | 59 ± 4 | High fiber dampens insulin | |
| Oranges | 60 ± 3 | Citrus acids and pectin moderate | |
| Grapes | 82 ± 6 | Higher sugar content increases | |
| Legumes | Lentils (boiled) | 58 ± 12 | Fiber and co-nutrients lower II |
Comparisons
With Glycemic Index
The insulin index (II) and glycemic index (GI) share methodological similarities, both assessing postprandial responses over a 2-hour period by calculating the area under the curve (AUC) relative to a reference food scaled to 100 (white bread in the original II study and typically glucose or white bread for GI).[1] Portions for II are normalized to equal energy content (1000 kJ), whereas GI uses equal available carbohydrate amounts (usually 50 g), yet both aim to quantify physiological impacts of foods on blood responses.[1] High-GI foods, such as potatoes, often elicit high II values, reflecting their shared sensitivity to rapidly digestible carbohydrates.[1] Studies indicate a moderate positive correlation between II and GI (r = 0.70, P < 0.001).[1] This relationship is particularly evident for carbohydrate-dominant foods like starchy and sugary items. Key discrepancies arise because II accounts for insulin secretion triggered by proteins and fats, independent of carbohydrates, whereas GI focuses solely on blood glucose rises from carbs and assigns near-zero values to non-carbohydrate foods.[1] Consequently, protein-rich foods like eggs exhibit low GI (due to negligible carbs) but a moderate II from amino acid stimulation, and baked beans show low-to-moderate GI yet substantially higher II owing to their protein and fat content amplifying insulin demand.[1] The following table illustrates these alignments and divergences using data from the seminal 1997 study, with published GI values for comparison (note: values can vary slightly by preparation and testing conditions):| Food | Glycemic Index (GI) | Insulin Index (II) |
|---|---|---|
| Potatoes (boiled) | 56–82 | 121 |
| Eggs | 0 | 31 |
| Baked beans | 40 | 120 |