Glycemic Load, Not GI, and Ovarian Cancer Risk

Glycemic index may be the more famous number, but glycemic load may be the more relevant one for ovarian cancer research. A 2024 meta-analysis found no association between dietary glycemic index (GI) and ovarian cancer risk, while a sensitivity analysis found that higher glycemic load (GL) tracked with a modest increase in risk. That is an association, not proof that carbohydrate intake causes ovarian cancer—and the strongest result depended on excluding heterogeneous studies.

The distinction matters because GI and GL answer different questions. GI estimates how quickly a standardized amount of carbohydrate raises blood glucose. GL incorporates both that rate and the amount of carbohydrate consumed. A food can have a high GI but contribute little total carbohydrate to a meal; a large serving of a moderate-GI food can create a much larger total carbohydrate load. If insulin exposure is the relevant biological pathway, the total load may matter more than a food's ranking.

GI was null; GL produced a conditional signal

Zhu et al.'s 2024 systematic review and meta-analysis pooled nine studies with approximately 4,716 ovarian cancer cases and 119,960 controls. It evaluated GI and GL as separate dietary exposures rather than treating them as interchangeable.

GI was essentially null: the pooled relative risk was 1.02 (95% confidence interval, 0.83 to 1.26). A relative risk of 1.00 means no difference in risk, and this interval includes both small decreases and small increases. The result does not support GI as an independent ovarian cancer risk factor in this evidence base.

The initial GL result was also statistically non-significant: relative risk 1.11 (95% CI, 0.84 to 1.47). That estimate pointed upward, but its confidence interval crossed 1.00, so it cannot rule out no association. The more attention-grabbing result came from dose-response sensitivity analyses that removed studies contributing most to heterogeneity. In that reduced analysis, every 50 GL-unit-per-day increase was associated with a relative risk of 1.11 (95% CI, 1.05 to 1.17). At a 20-unit-per-day increment, the relative risk was 1.04 (95% CI, 1.02 to 1.07).

Those are statistically significant associations within the sensitivity analysis, not precise estimates of a causal dose effect. The difference between the full pooled result and the reduced analysis is the main reason to stay cautious. Removing truly incompatible studies can clarify a signal; removing inconvenient results can also make one appear cleaner than it is. Meta-analysis cannot fully settle that judgment call.

Why glycemic load could matter biologically

A high-GL dietary pattern can produce repeated, larger insulin demands across meals. The proposed mechanism is that chronically elevated insulin lowers hepatic production of insulin-like growth factor binding protein-3, leaving more biologically available IGF-1. IGF-1 signaling through its receptor can promote cell proliferation and inhibit apoptosis. Long et al.'s 2022 prospective-cohort meta-analysis describes this insulin-to-IGF-1 pathway as one plausible route linking GL to cancer development.

Plausible does not mean demonstrated in people eating high-GL diets. A 2026 study of IGF-1-axis genetic variants found that variants in this pathway influenced surgical outcomes and prognosis among people who already had ovarian cancer. It supports the idea that IGF-1 signaling matters in ovarian cancer biology. It did not test diet, glycemic load, insulin responses, or the chance of developing ovarian cancer.

The prospective evidence is less decisive

Long et al. restricted their analysis to prospective cohort studies, where researchers measure diet before a cancer diagnosis. For ovarian cancer, the GL estimate was 1.19 (95% CI, 0.85 to 1.68). The direction matched Zhu et al.'s result, but it was not statistically significant and the interval was wide.

That distinction carries real weight. Case-control studies ask people to recall diet after diagnosis, which can introduce recall bias. A person confronting cancer may remember prior eating differently, or report it differently, than someone without cancer. Prospective studies reduce that problem, though they usually have fewer cases and therefore less precise estimates. The prospective-only result does not disprove a GL association; it does mean the evidence does not yet establish one.

One relevant prospective contribution was Mongiovi et al.'s 2022 analysis in the PLCO screening cohort, which examined GI and GL in relation to incident ovarian cancer while adjusting for established ovarian cancer risk factors. Its prospective design avoids the retrospective dietary reporting inherent in case-control research. It still relies on food-frequency questionnaires, which estimate usual diet imperfectly.

Several apparent confirmations share the same cohort

It is tempting to point to GL research across cancer sites as independent confirmation. That would overstate the case. The ovarian analysis by Mongiovi et al., a 2021 PLCO bladder-cancer study, and a 2025 PLCO lung-cancer study draw on overlapping participants and dietary-assessment infrastructure from the same Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial.

Shared participants, food-frequency questionnaires, and measurement error mean these analyses are not separate replications. A systematic problem in estimating GL within PLCO could affect several cancer analyses at once. The bladder-cancer study itself reported no statistically significant association for either GI or GL: for GL, the highest versus lowest quintile had a hazard ratio of 0.92 (95% CI, 0.65 to 1.30) after a median 12.5 years of follow-up.

What the result can—and cannot—change

Ovarian cancer's established risk factors include reproductive history, BRCA1/2 mutations, and family history. Dietary GL is not part of standard clinical risk assessment. No randomized dietary trial has tested whether lowering GL prevents ovarian cancer, and no causal genetic analysis of the GL-to-ovarian-cancer pathway appears in this evidence base.

GL also travels with other variables: total energy intake, body weight, physical activity, and overall diet quality. Researchers adjust for these factors, but observational studies cannot remove all residual confounding. Food-frequency questionnaires add another layer of uncertainty because studies use different questionnaires and glycemic-reference databases.

Still, a practical dietary conclusion does not require pretending this is settled cancer prevention science. Replacing refined carbohydrate-heavy meals with legumes, fiber-rich whole foods, and non-starchy vegetables can lower dietary GL while supporting metabolic health. Focus on the carbohydrate amount in the whole meal and day, not only the GI printed beside one food. GI remains useful for glycemic management, especially in diabetes; it simply did not predict ovarian cancer risk in Zhu et al.'s pooled analysis.

The current evidence supports dietary prudence, not ovarian-cancer alarm. Higher GL has a biologically plausible, modest observational association with risk, but the overall pooled estimate was non-significant and the prospective-only evidence remains imprecise.