Multivariate rate ratios calculated by restricted cubic spline Cox proportional
hazards model. Rate ratios are adjusted for age (in months), body mass index
(quartiles), educational level (less than high school, high school, university),
total energy intake (quartiles), and energy-adjusted intakes of saturated
fat, dietary fiber, calcium, zinc, beta carotene, folate, and vitamin B6 (all in quartiles). Solid curve represents point estimates and dashed
curves represent 95% confidence intervals.
Larsson SC, Bergkvist L, Wolk A. Magnesium Intake in Relation to Risk of Colorectal Cancer in Women. JAMA. 2005;293(1):86-89. doi:10.1001/jama.293.1.86
Author Affiliations: Division of Nutritional
Epidemiology, The National Institute of Environmental Medicine, Karolinska
Institutet, Stockholm, Sweden (Ms Larsson and Dr Wolk); and Department of
Surgery and Centre for Clinical Research, Central Hospital, Västerås,
Sweden (Dr Bergkvist).
Context Animal studies have suggested that dietary magnesium may play a role
in the prevention of colorectal cancer, but data in humans are lacking.
Objective To evaluate the hypothesis that a high magnesium intake reduces the
risk of colorectal cancer in women.
Design, Setting, and Participants The Swedish Mammography Cohort, a population-based prospective cohort
of 61 433 women aged 40 to 75 years without previous diagnosis of cancer
at baseline from 1987 to 1990.
Main Outcome Measure Incident invasive colorectal cancer.
Results During a mean of 14.8 years (911 042 person-years) of follow-up,
805 incident colorectal cancer cases were diagnosed. After adjustment for
potential confounders, we observed an inverse association of magnesium intake
with the risk of colorectal cancer (P for trend = .006).
Compared with women in the lowest quintile of magnesium intake, the multivariate
rate ratio (RR) was 0.59 (95% confidence interval [CI], 0.40-0.87) for those
in the highest quintile. The inverse association was observed for both colon
(RR, 0.66; 95% CI, 0.41-1.07) and rectal cancer (RR, 0.45; 95% CI, 0.22-0.89).
Conclusion This population-based prospective study suggests that a high magnesium
intake may reduce the occurrence of colorectal cancer in women.
Magnesium is required for a wide range of biological functions. Apart
from being essential for the maintenance of genomic stability and for DNA
repair, magnesium has a crucial role in modulating cell proliferation, cell
cycle progression, and cell differentiation.1 Magnesium
supplementation has been demonstrated to reduce the incidence of experimentally
induced colon cancer in animals,2,3 which
might be related to a decrease in colonic epithelial cell proliferation.3- 6 Magnesium
has an important role in maintaining the antioxidative status of the cell1; animals deficient in magnesium display an increased
susceptibility to oxidative stress.1,7,8
High circulating concentrations of C-peptide, a marker for insulin secretion,
have been associated with increased risk of colorectal cancer in humans9,10; conceivably, dietary factors that
improve insulin sensitivity and lower insulin concentrations may have an impact
on colorectal cancer risk. Magnesium supplementation increased insulin sensitivity
among healthy subjects11 and among patients
with type 2 diabetes.12,13 Furthermore,
recent epidemiologic studies reported an inverse association of magnesium
intake with insulin concentrations.14,15 Despite
evidence that magnesium may be implicated in colorectal carcinogenesis, there
is no epidemiologic study pertaining to the association between magnesium
intake and risk of colorectal cancer. Therefore, we conducted a prospective
analysis of magnesium intake in relation to incidence of colorectal cancer
using data from the Swedish Mammography Cohort, a population-based prospective
cohort of 61 433 women.
Details of the Swedish Mammography Cohort have been described previously.16 In brief, this population-based cohort was established
between 1987 and 1990, when all women aged 40 to 75 years living in Uppsala
and Västmanland counties, central Sweden, received a mailed questionnaire
that elicited information about diet (along with data on weight, height, and
educational level). In total, 66 651 women, representing 74% of the source
population, returned a completed questionnaire. A new questionnaire, sent
to all surviving participants in 1997, was expanded to include data on a family
history of colorectal cancer, cigarette smoking, physical activity, and use
of multivitamin supplements and aspirin. The study was approved by the ethics
committee at the Karolinska Institutet in Stockholm and the Uppsala University
Nutrient intakes were computed by multiplying the consumption frequency
of each food by the nutrient content of age-specific (<53, 53-65, ≥66
years) servings, using composition values from the Swedish National Food Administration
Database.17 In a validation study in a subsample
of 129 women randomly selected from the cohort (A. Wolk, unpublished data
1992), Pearson correlation coefficient between intake of magnesium reported
in the baseline questionnaire and in four 1-week dietary records was 0.44,
indicating reasonable validity of our questionnaire-based assessment of magnesium
For this analysis, we excluded women with an erroneous national registration
number, women with extreme energy intake estimates (ie, 3 SDs from the mean
value for log-transformed energy), and women with previously diagnosed cancer
(other than nonmelanoma skin cancer) at baseline. After exclusions, the study
population comprised 61 433 eligible women who were followed up until
a diagnosis of colorectal cancer, death, or June 30, 2004.
The women were categorized into quintiles according to magnesium intake.
After determining that the data conformed to the proportional hazards assumptions,
we used Cox proportional hazards modeling18 with
age in months as the underlying time variable to estimate rate ratios (RRs)
with 95% confidence intervals (CIs). All multivariate models were also simultaneously
adjusted for body mass index (BMI), educational level, and intakes of total
energy, saturated fat, dietary fiber, calcium, zinc, beta carotene, folate,
and vitamin B6. Intakes of nutrients were adjusted for total energy
intake with the residual method.19 To calculate
the P value for trend, participants were assigned
the median value of their quintile of magnesium intake, and this variable
was used as a continuous variable.20 We used
restricted cubic spline regression with 5 knots to flexibly model the association
between magnesium intake and colorectal cancer risk.21 Analyses
were conducted using SAS software (version 8.2, SAS Institute Inc, Cary, NC).
All P values were 2-tailed; P <.05
was considered statistically significant.
The age-standardized baseline characteristics of the study population
by quintiles of magnesium intake are shown in Table 1. Compared with women with a low intake of magnesium, those
with higher intakes generally had lower intakes of energy and saturated fat
and higher intakes of dietary fiber, calcium, zinc, beta carotene, folate,
and vitamin B6. Women with greater magnesium intake also were more
likely to have a postsecondary education.
Over an average follow-up of 14.8 years (911 042 person-years),
805 women were diagnosed with colorectal cancer (547 colon cancer, 252 rectal
cancer, and 6 cases with both colon and rectal cancer). We observed a statistically
significant inverse association between magnesium intake and risk of colorectal
cancer in both the age- and multivariate-adjusted models (Table 2). Compared with women in the lowest quintile of magnesium
intake, the multivariate RR of colorectal cancer for those in the highest
quintile was 0.59 (95% CI, 0.40-0.87; P for trend = .006).
Further control for consumption of red meat, fruits, vegetables, and whole
grain foods yielded virtually the same results (RR, 0.61; 95% CI, 0.41-0.91).
In addition, the inverse association with magnesium intake persisted when
we added 1 at a time to a multivariate model intake of vitamins A, C, D, and
E, and (in place of total dietary fiber) cereal fiber, vegetable fiber, and
fruit fiber (data not shown). The RR was only slightly attenuated when all
these nutrients were included simultaneously in a multivariate model (RR,
0.67; 95% CI, 0.45-1.00). Using data from the 1997 questionnaire, the results
remained essentially unchanged after adjustment for a family history of colorectal
cancer, cigarette smoking, physical activity, and use of multivitamin supplements
and aspirin (RR, 0.60; 95% CI, 0.40-0.88). Excluding cases of colorectal cancer
that occurred within the first 3 years of follow-up did not appreciably alter
the results (multivariate RR comparing extreme quintiles, 0.62; 95% CI, 0.41-0.93).
Intake of magnesium was inversely associated with both colon and rectal cancer
(Table 2); the inverse
association was similar for proximal colon (RR comparing extreme quintiles,
0.56; 95% CI, 0.27-1.16) and distal colon cancer (RR, 0.63; 95% CI, 0.27-1.47).
Because the inverse association of magnesium intake with colorectal
cancer risk appeared to be linear (Figure),
we analyzed magnesium intake as a continuous variable. The multivariate RR
of colorectal cancer for a 50-mg/d increment of magnesium—approximately
equivalent to the magnesium content in 1 small serving of spinach per day,
1 large banana per day, 1 serving of cooked oatmeal per day, 2 slices of whole
grain bread per day, or a half serving of beans per day—was 0.78 (95%
This large population-based prospective cohort study is, to the best
of our knowledge, the first to examine and observe a significant inverse dose-response
relationship between magnesium intake and risk of colorectal cancer.
Major strengths of our study include its large size, population-based
and prospective design, the large number of colorectal cancer cases, and the
completeness of case ascertainment through the Swedish Cancer Registry System.22 These features of the study increase the generalizability
of our results and eliminate potential recall and selection biases. Our study
also has several potential limitations. Because magnesium intake was assessed
through a self-administered food-frequency questionnaire, and our analysis
was based on a single baseline measurement of dietary intake, some misclassification
of magnesium intake is inevitable, which would potentially attenuate any true
relationship. Although we adjusted our estimates for a wide range of potential
confounders, we cannot rule out the possibility that our findings may be biased
by unmeasured confounders or by residual confounding. However, multivariate
analyses yielded results similar to those from age-adjusted analyses, suggesting
that residual confounding is unlikely to have affected our results materially.
In conclusion, this population-based cohort study of women suggests
that a high magnesium intake may reduce the risk of colorectal cancer. While
our findings require confirmation by other large well-designed studies, they
support potential benefits of increasing consumption of major foods contributing
to magnesium intake, including fruits and vegetables, whole grain foods, and
beans, in reducing colorectal cancer incidence. However, the efficiency and
safety of magnesium supplementation for the prevention of colorectal cancer
needs to be specifically addressed in a randomized trial.
Corresponding Author: Susanna C. Larsson,
MSc, Division of Nutritional Epidemiology, The National Institute of Environmental
Medicine, Karolinska Institutet, PO Box 210, SE-171 77 Stockholm, Sweden (firstname.lastname@example.org).
Author Contributions: Ms Larsson had full access
to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Larsson, Bergkvist,
Acquisition of data: Wolk.
Analysis and interpretation of data: Larsson,
Drafting of the manuscript: Larsson.
Critical revision of the manuscript for important
intellectual content: Larsson, Bergkvist, Wolk.
Statistical analysis: Larsson.
Obtained funding: Bergkvist, Wolk.
Administrative, technical, or material support:
Study supervision: Bergkvist, Wolk.
Funding/Support: This work was supported by
research grants from the Swedish Cancer Foundation and the Swedish Research
Role of the Sponsors: Funding sources had no
role in the design and conduct of the study; collection, management, analysis,
and interpretation of the data; and preparation, review, or approval of the