Skip to content
PaperFren

Is higher blood sugar linked to getting or dying of cancer?

Open paper intelligence

People with higher blood sugar were more likely to develop cancer and especially to die from it, and this held even within the upper-normal range in women.

Source

Blood glucose and risk of incident and fatal cancer in the metabolic syndrome and cancer project (me-can): analysis of six prospective cohorts

Stocks T, Rapp K, Bjørge T, et al. · PLoS medicine · 2009

doi.org/10.1371/journal.pmed.1000201Read the full paper ↗193 citationscc by

Study at a glance

Design
Cohort — Pooled analysis of six population-based cohorts from Norway, Austria and Sweden linked to national cancer and death registers; Cox regression with correction for measurement error using repeated glucose tests.
N
N=549944 · 549,944 people (274,126 men and 275,818 women) in the final dataset; the regression dilution correction used repeat measurements from 133,820 individuals.
Population
Adults attending health examinations in Norwegian, Austrian and Swedish population cohorts, mean age about 45, without cancer at baseline.
Outcome
Incident cancer and fatal cancer overall and by site, plus all-cause mortality, over roughly 10 years of follow-up.

Structured fields used in claim comparison tables when every cited study has a complete layer.

What they did

The researchers pooled health-check data from 549,944 adults in six cohorts in Norway, Austria and Sweden and followed them through national registers for new cancers and cancer deaths. They related blood glucose at the first health examination to cancer risk, adjusting for age, body mass index and smoking and starting follow-up one year after the blood test. Because a single glucose test is noisy, they used repeated measurements in 133,820 people to correct the risk estimates for this random error.

What they found

Each 1 mmol/l increase in glucose was linked to a 5% higher risk of new cancer and 15% higher risk of cancer death in men, and 11% and 21% higher risks in women. Women in the top tenth of glucose had about twice the risk of fatal cancer compared with those in the lowest 40%, and men in the top tenth had a similar rise (1.84 times). Associations were seen for several specific cancers, such as liver, gallbladder and respiratory tract in men and pancreas and stomach in women, but not prostate cancer. The relationship looked roughly linear, and correcting for measurement error made the fatal-cancer excess risk about four times larger than the uncorrected estimate.

The limits

What it doesn't show

This is observational, so glucose may be a marker for other factors such as alcohol, diet or physical activity that were not measured; only BMI and smoking were adjusted for. Glucose was measured differently across cohorts, with varying fasting times, so absolute glucose cut-offs cannot be applied directly. The stronger link with cancer death than diagnosis could reflect later diagnosis or coding differences rather than biology. Many site-specific tests were run, so some individual findings, such as lower thyroid cancer risk in women, may be chance.

Key terms

Incident vs fatal cancer
Incident cancer means a new diagnosis; fatal cancer means death with cancer as the underlying cause.
Regression dilution bias
Weakening of an observed association because a single measurement of the exposure contains random error; corrected here using repeated measurements.
Relative risk (hazard ratio)
How many times higher the rate of an outcome is in one group compared with a reference group, estimated here with Cox regression.
Impaired fasting glucose
Fasting blood glucose above normal but below the diabetic threshold, a pre-diabetic state.
Reverse causation
When the outcome influences the exposure, for example an undiagnosed cancer raising blood glucose; reduced here by excluding the first year of follow-up.

Flashcards

1 / 11

0 of 11 answers reviewed

Research intelligence for this paper

See its role on concept claims, tensions it is part of, placement history, and related discoveries.

Open paper intelligence

Quiz yourself

1 / 6

What type of study was this?

Common questions

Why did they ignore cancers diagnosed in the first year?

An undiagnosed cancer might itself raise blood glucose. Starting follow-up a year after the blood test reduces the chance that cancer caused the high glucose rather than the reverse.

Why does correcting for measurement error make the association stronger?

Random error in a single glucose test makes people's true levels look more similar than they are, which dilutes any real link. Using repeat tests to estimate that error lets the analysis recover the stronger underlying association.

How might high glucose promote cancer?

High glucose raises insulin and insulin-like growth factor, which stimulate tumour growth, and tumour cells use glucose as fuel, though the direct role of blood glucose is uncertain.

More on Endocrinology