ICANCER RESEARCH 58. 442-447. February I. 1998]
Calcium and Fructose Intake in Relation to Risk of Prostate Cancer1
Edward Giovannucci,2 Eric B. Rimiti, Alicja Wolk, Alberto Ascherio, Meir J. Stampfer, Graham A. Colditz, and
Walter C. Wille«Channinx Uihoratory, Department t>fMedicine. Harvard Medical School and Brighimi ami Women's Hospital, Boximi, Massachusetts 02115 IE. G.. E. B. R.. M. J. S.. W. C. W.¡;
Department of Nutrition, Hunurd School of Public Health. Boston. Massachusetts 02115 ¡E.G., E. B. R., A.A., M. J. S., G. A. C., W.C.W.]: Department of Epidemiology,Han-ard School of Public Health. Boston. Massachusetts 02115 IE. B. R.. A. A., M. J. S., G. A. C., W. C. W.I; and Department of Medical Epidemiology, Kamlinska Inslitulel,
S-I7Ì77 Stockholm, Sweden ¡A.W.¡
ABSTRACT
Laboratory and clinical data indicate an antitumor effect of 1,25(OH)2vitamin I) 11.25i()1h ,l)i on prostate cancer. High calcium intake suppresses formation of l,2>iOlli.l) from 25(OH)D, thereby decreasing the1.25(0111.1) level. Ingestion of fructose reduces plasma phosphate transiently, and hypophosphatemia stimulates 1,25(OH)2D production. Wethus conducted a prospective study among 47,781 men of the HealthProfessionals Follow-Up Study free of cancer in 19X6 to examine whether
calcium and fructose intake influenced risk of prostate cancer. Between1986 and 1994,1369 non-stage Al and 423 advanced (extraprostatic) cases
of prostate cancer were diagnosed. Higher consumption of calcium wasrelated to advanced prostate cancer |multivariate relative risk (RR), 2.97;95% confidence interval (CD, 1.61-5.50 for intakes s=2000 rng/day versus
<500 mg/day; P, trend, 0.002] and metastatic prostate cancer (RR, 4.57;CI, 1.88-11.1; P, trend, <0.001). Calcium from food sources and from
supplements independently increased risk. High fructose intake was related to a lower risk of advanced prostate cancer (multivariate RR, 0.51;CI, 0.33-0.80, for intakes >70 versus S40 g/day; P, trend, 0.007). Fruit
intake was inversely associated with risk of advanced prostate cancer (RR,0.63; 95% CI, 0.43-0.93; for >5 versus SI serving per day), and thisassociation was accounted for by fructose intake. Non-fruit sources of
fructose similarly predicted lower risk of advanced prostate cancer. Amoderate positive association between energy-adjusted fat intake and
advanced prostate cancer was attenuated and no longer statistically significant when controlled for calcium and fructose. Our findings provideindirect evidence for a protective influence of high 1,25(OH)2D levels onprostate cancer and support increased fruit consumption and avoidance ofhigh calcium intake to reduce the risk of advanced prostate cancer.
INTRODUCTION
I,25-(OH)2D3 has a well-established role in calcium and phospho
rus homeostasis, but many cell types possess functional vitamin Dreceptors and are responsive to the actions of 1,25-(OH)2D (1).Typically, higher levels of 1,25-(OH)2D inhibit cellular proliferation
and induce differentiation for normal and neoplastic prostate cells invitro (2-5). Moreover, limited studies in rodents support antitumor
potency of 1,25(OH),D analogues against prostate cancer (6. 7), andolder men who have high circulating 1,25(OH)2D are at reduced riskof poorly differentiated and clinically advanced prostate cancer (8. 9).Additional preliminary evidence of a role of 1,25(OH)2D is thatgenetic polymorphisms of the vitamin D receptor gene, which maycorrelate with activity of the receptor, also predict risk of prostatecancer (10. 11).
If circulating 1,25(OH)2D confers these benefits, adequate intake ofvitamin D or production in the skin through sunlight exposure may
Received 7/10/97; accepted 11/26/97.The costs of publication of this article were defrayed in part by the payment of page
charges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.
' Supported by Grants CA 55075 and HL 35464 from the NIH and Special Institution
(¡rant IS from the American Cancer Society.2 To whom requests for reprints should be addressed, al Channing Laboratory. 181
Longwood Avenue. Boston. MA 02115.'The abbreviations used are: 1.25-(OH),D. I.25(OH)2 vitamin D; RR, relative risk;
CI. confidence interval; BMI. body mass index.
help prevent prostate cancer. Although some geographic evidencesuggests that sunlight may be beneficial (12). ecological, case-control,
and cohort studies consistently find higher intakes of dairy products,the major dietary source of vitamin D, associated with an enhancedrisk of prostate cancer (13). This apparent paradox may be resolved byconsidering that 1.25(OH)2D, rather than the precursor 25(OH)D, isbiologically relevant. As shown in Fig. 1, 1.25(OH)2D is tightlyregulated, primarily through the activity of renal 1-a-hydroxylase.
which is stimulated to hydroxylate 25(OH)D to 1,25(OH)2D when thecirculating concentration of calcium is low (14, 15). Because of itsprecise regulation. 1,25(OH)2D is not correlated with vitamin Dintake or 25(OH)D levels. Instead, because dairy products are themajor source of calcium, their overall effect is to suppress1,25(OH)2D levels.
Another potentially important dietary factor is phosphorus, becausereductions in circulating phosphate increase 1,25(OH)2D levels appreciably ( 16, 17). Because phosphorus is generally abundant in mostdiets and is well absorbed intestinally. dietary-induced hypophos
phatemia is rare. Moreover, phosphate may bind calcium, reducing itsbioavailability. and low phosphate stimulates parathyroid hormone;these properties tend to decrease circulating 1,25(OH)2D, making theoverall impact of dietary phosphorus on circulating phosphate or1,25(OH)2D levels unclear. Dietary fructose can reduce plasma phosphate levels by 30 to 50% for more than 3 h due to the rapid shift ofphosphate from the extracellular to intracellular compartment (18,19). This hypophosphatemia occurs because fructose is very rapidlyphosphorylated in the liver, catalyzed by fructokinase (20), whichby-passes the phosphofructokina.se regulatory step in glycolysis (21).
Based on the evidence supporting antitumor properties of1,25(OH)-,D (1-11). we examined dietary calcium, vitamin D, phos
phorus, and fructose in relation to risk of prostate cancer in theongoing Health Professionals Follow-Up Study.
PATIENTS AND METHODS
Study Population and Follow-up of the Cohort. The Health Professionals Follow-Up Study, an ongoing prospective cohort study (22), consists of
51.529 U.S. male dentists, optometrists, osteopaths, podiatrists, pharmacists,and veterinarians, who were 40-75 years in 1986. Through mailed question
naires, these men provided information on age. marital status, height andweight, ancestry, medications, disease history, physical activity, and diet. Forthis analysis, we excluded men who had a diagnosis of cancer (other thannonmelanoma skin cancer) or who did not adequately complete the dietquestionnaire (3% of total), leaving 47.781 tor follow-up. Follow-up question
naires were sent in 1988. 1990. 1992, and 1994 to ascertain new cases of avariety of diseases, including prostate cancer. The response rate through 1994was 94%. Deaths in the cohort were reported by family members in responseto the follow-up questionnaires or were discovered through the postal system
or the National Death Index, a highly sensitive method (23) used to identitydeaths among nonrespondents. Through these measures, over 98% of the
deaths are ascertained.Dietary Assessment. Diet was assessed with a self-administered semi-
quantitative food frequency questionnaire. Men were asked how often onaverage, over the past year, they consumed each item on a list of 131commonly consumed foods and beverages. In addition, we assessed frequency
442
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
RISK 01 PROSTATE CANCER AND CALCIUM AND FRUCTOSE INTAKE
7-Dehydrocholesterol
skin <UV)
liver
VitaminDr -VITAMIND
i PHOSPHORUSI
l CALCIUM111
Blood Calcium
Fig. I. Proposed interrelationships between dietary calcium, phosphorus, and fructosewith vitamin D homeostasis and prostate cell differentiation and proliferation.
of supplement use and included an open-ended section for commonly eaten
foods not listed. Nutrient intakes were the product of the frequency of intakeand the nutrient composition (24) of the specified portion size. Analyses wereconducted using energy-adjusted nutrient intakes based on residuals from the
regression of nutrient intake on total caloric intake (25).We conducted a validation study of 121 men in the cohort to compare
questionnaire-based intakes to those estimated from two 1-week diet records(26). Correlation coefficients between the energy-adjusted calcium intakesmeasured by diet records (de-attenuated for week-to-week variability) and by
the second questionnaire were 0.61 for calcium with supplements and 0.61 forcalcium without supplements, and 0.63 for phosphorus. Composition forvitamin D was not available from diet records, but correlation coefficientsbetween the questionnaire and the diet records for dairy foods, the majordietary source of vitamin D, was 0.70 (27).
Fructose is ingested as a monosaccharide or as part of sucrose. Becausefructose is readily liberated from sucrose in the small intestine, fructose from
sucrose has similar metabolic properties as monosaccharide fructose (28). Themajor dietary sources of total fructose are fruits, carbonated beverages, sweetbakery products, candy, and added sugar. The median intake of total fructoserepresented 9.5% of total energy intake, with a range from 4.99Õ-in the tenth
percentile to 16.1% in the ninetieth, comparable with national data (28).Although diet record values for fructose and sucrose were not available forcomparison, the major dietary sources of fructose were measured well (27).Correlation between the intakes measured by diet records and by the foodfrequency questionnaire for the four top contributors of monosaccharide fructose intake were 0.78 for orange juice (which contributed 15.99!- of total
monosaccharide fructose). 0.84 for soft drink (cola) beverages (9.6%). 0.59 forraisins (5.2%), and 0.70 for apples (4.6%). Among the four major contributorsof sucrose (with available data), the correlations were 0.80 for bananas (5.2%),0.45 for candy (3.8%). 0.71 for ice cream (3.7%), and 0.70 for apples (3.6%).Data were unavailable for added sugar (7.2% of sucrose), and the within-
person variability from diet records was too high to calculate reliable coeffi
cients for cake (4.7%).Identification of Cases of Prostate Cancer. Whenever a man reported a
diagnosis of prostate cancer, we asked him for permission to obtain hospitalrecords and pathology reports. For deceased men. we contacted next of kin forthis information. A study physician reviewed the medical reports and classifiedthe stage of prostate cancers. Information from any work-up during the initial
diagnosis including staging prostatectomy and bone scans was taken into
account. From 1986 to the end of this study period (January 31. 1994), weidentified 1414 incident cases of prostate cancer. Of these. 1262 cases (89.3%)were confirmed by medical records, and of the remaining 152 cases. 130 men(85%) provided information regarding the basis of diagnosis. When we obtained medical records and pathology reports, we confirmed a diagnosis ofadenocarcinoma of the prostate for 99%.
Data Analysis. Beginning on the month of return of the baseline questionnaire, each of the 47.781 eligible men contributed follow-up time until the
month of diagnosis of prostate cancer for cases, or month of death from othercauses, or the end of the study period for noncases. Our primary outcomesconsisted of the 1369 nonstage A1 prostate cancer cases and 423 extraprostatic
(stage C or D) prostate cancer cases. We excluded the stage Al cancers fromour analysis because they are relatively innocuous and asymptomatic. Theserepresent only 3% of the total. We examined these subgroups of advanced andfatal cancers based on the findings that low I.25(OH),D levels appear to berelated to aggressive behavior in prostate cancer (8).
We calculated the incident rate of total, advanced, and metastatic prostatecancer for men in a specific category of nutrient or food intake in 1986 bydividing the number of new cases by the number of person-years. The RR for
each category of intake was then computed as the incident rate among men inthat category divided by the incident rate among men in the lowest intakecategory. Categories were established a priori to represent a wide range ofintakes. We used the Mantel-Haenszel summary estimator (29) to adjust torage (across 2.5-year categories) and pooled logistic regression (30) to estimate
relative risks when controlling simultaneously for more than one risk factor.We tested for linear trends by using continuous variables in logistic regressionmodels. All reported />s are two-sided.
We evaluated potential confounding from nondietary factors (tobacco use.physical activity, and vasectomy) and various dietary factors hypothesi/ed tobe related to prostate cancer risk (total energy and fat. types of fat. vitamin E.and lycopene). We also assessed current BM1 and BMI at age 21 years, whichwas strongly inversely related to advanced prostate cancer risk in this population (31).
RESULTS
Calcium. As shown in Table 1, high calcium consumption wasrelated to less tobacco use. higher level of physical activity, higherconsumption of multivitamins. fiber, phosphorus, vitamin E, andlycopene. In age-adjusted and multivariate analyses (Table 2), con
sumption of calcium was related to higher risk of total, advanced, andmetastatic prostate cancer, especially at intakes exceeding 2000 nig/
Table 1 Age-standardi^etl characteristics according tit lu\\- ami /Õ/K/Õintake of totalcalcium. Itital fructose, fruit fructose, and non-fruil fructose in the Health
Professionals Follow-up Study (I9fif>)
TotalcalciumCalcium
(mg)"<7renergy fromfructose''Current
smokers(%)Pastsmokers(%)BMI
(kg/rrr)BMI(kg/m: at age21Physical
activity(MET-hours/week)Multivitamin
use(%)Vasectomy(%)Intakes
(daily)"Total
fat(g)l'!renergy fromfatCarbohydrates
(g)Fiber(g)"Phosphorus(mg)"Vitamin
D(IU)VitaminE(IU)Lycopene
(/xg)"Low4329.814.143.125.522.615.427.419.073.532.8220.619.11133193614650High24748.16.841.225.323.422.27424.566.829.6245.624.018188353715140TotalfructoseLow9156.211.345.225.723.117.641.422.675.033.1214.220.91429378HIS4960HighM216.68.736.125.422.822.542.418.264.829.8267.623.61309330875080Fruit
fructoseLow8180.622.142.825.722.813.436.021.879.935.8200.517.01.340307793957High9658.54.040.025.323.325.746.117.558.326.0282.628.913924191446374Non-FruitfructoseLow')M)2.211.44K.«25.823.319.146.523.073.832.8201.021.214844151274445High75412.713.035.825.422.516.838.917.769.130.3269.819.61207295794525
" Adjusted for total energy intake.'' For fruit and non-fruit fructose, this represents percentage of energy from fruit and
non-fruit fructose, respectively.
443
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
RISK OF PROSTATE CANCER AND CALCIUM AND FRUCTOSE INTAKE
Table 2 KK and 95% CI for total (nonstage A-l). advanced, and metastatic prostatecancer according to total calcium intake in Health Professionals Follow-up Study
(1986-19941
Tolal calcium intake(mg/day)<500Person-Years
34,563Total
(n)Age-adjustedRR
MullivariateRR"95%
CIAdvanced(n)Age-adjusted
RRMultivariateRR"95%
CIMetastatic(n)Age-adjustedRRMulm
.m.ii.-RR"95%
CI1071.0
1.0311.0
1.0131.0
1.0500-999213.4158171.20
1.200.97-1.492331.19
1.190.79-1.781131.37
1.490.81-2.751000-149975,2992921.10
1.080.83-1.401021.33
1.300.81-2.08441.36
1.460.71-2.971500-199919.325881.211.200.86-1.66301.43
1.550.87-2.76161.83
2.180.94-5.08==20009,067651.70
1.711.19-2.46272.42
2.971.61-5.50153.22
4.571.88-11.1P
trend0.360.002<0.001
"Multivariate model includes age, BMI at age 21, intake of total energy, total fat,
fructose, phosphorus, vitamin D, vitamin E. and lycopene.
Table 3 RR" and 95% CI for meiastatic prosiate cancer according to calcium intake
from foods and supplements
(0.0-0.0)Calcium
from foods(mg)<600600-1000>IOOOCalcium01.01.2(0.8-1.8)1.6(0.9-3.0)from
supplements(mg)1-900
>9001.0
3.6(0.4-2.7)(1.5-8.8)0.9
1.8(0.4-1.8)(0.7-1.7)2.6
3.8(1.1-6.0)(1.2-12.1)
" RR adjusted for age. BM1 at age 21, total energy, total fat, vitamin D, phosphorus,
vitamin E. lycopene. and fructose by multiple logistic regression.
day. The RRs for an equivalent increase of calcium from dairy andnondairy sources (including supplements) were similar; the multi-variate RR for metastatic prostate cancer was 1.8 (95% CI, 1.0-3.3;P = 0.05) for a 1000-mg increment of dairy calcium and 1.7 (95% CI,1.1-2.6; P = 0.02) for a 1000-mg increment of nondairy calcium.
Increasing calcium from foods and supplements independently increased the risk of prostate cancer (Table 3).
Total consumption of milk, 83% from skim and low-fat milk, was
associated with higher risk of advanced prostate cancer (RR, 1.6; 95%CI, 1.2-2.1 for >2 versus 0 glasses per day; P, trend = 0.002) andmetastatic prostate cancer (RR, 1.8; 95% CI, 1.2-2.8; P,trend = 0.01). We examined whether other components of dairy
products, including vitamin A, protein, vitamin D, phosphorus, andsaturated fat, could have accounted for the association. After including each of these in multivariate models with calcium, calcium intakepersisted unchanged as a significant risk factor in every model.
Phosphorus. Intake of phosphorus was unrelated to risk of totalprostate cancer (multivariate RR for >1800 versus <1200 mg/dayphosphate, 0.94; 95% CI, 0.71-1.26; P, trend = 0.31). For advancedprostate cancer, the age-adjusted model suggested a slightly elevated
risk of 1.10 (not significant) at high intakes, but in the multivariatemodel, an inverse association was suggested (RR, 0.67; 95% CI,0.40-1.13). The difference between the age-adjusted and the multi
variate RRs was due primarily to confounding from calcium intake.Vitamin D. Total vitamin D intake was not associated with risk of
total prostate cancer (multivariate RR, 1.21; 95% CI, 0.92-1.58 foraSOO versus <150 lU/day; P. trend = 0.87) or advanced prostatecancer (RR, 1.48; 95% CI, 0.91-2.39; P, trend = 0.64). Also, vitamin
D intake was not related to risk of metastatic prostate cancer (P,trend = 0.28). No association was observed for vitamin D intake from
foods only (P, trend = 0.30, 0.41, and 0.25, for total, advanced, and
metastatic prostate cancer, respectively) or from supplements (P,trend = 0.84, 0.70, and 0.84).
Fructose. As seen in Table 1, high consumers of fructose wereconsiderably less likely to be or have been smokers, tended to be morephysically active and slightly leaner, and had lower intake of fat andhigher consumption of carbohydrates, fiber, and lycopene. High consumption of energy-adjusted fructose was related to lower age-
adjusted risk of total prostate cancer (Table 4). Additional models(data not shown), which included nondietary risk factors includingcurrent BMI, smoking status, vasectomy, and physical activity level,yielded similar results as the age-adjusted models. Controlling for
various dietary factors, including dietary fat. yielded almost identicalresults as the age-adjusted analysis. High fructose consumption had an
even stronger inverse association with advanced and metastatic prostate cancer (Table 4). Both free fructose (P, trend = 0.04) and fructosefrom sucrose (P, trend = 0.006) had similar inverse associations with
advanced prostate cancer in multivariate models.Total carbohydrates, which had a correlation of 0.74 with total
fructose, also was inversely associated with advanced forms of prostate cancer (RR, 0.74 for a 100-g increment; P = 0.02), but the
relationship with total carbohydrates was attenuated when we controlled for fructose (RR, 0.92; P = 0.66).
Although controlling for known and suspected risk factors forprostate cancer, including fat intake, did not alter our results forfructose, the possibility remained that unknown risk factors associatedwith fructose intake accounted for this relationship. High consumersof total fructose were more likely to have generally "healthy lifestyles," but fructose from fruit and non-fruit sources (predominantly
sweets, carbonated soft drinks, and added sugar) had divergent lifestyle and dietary patterns (see Table 1). We examined fructose fromfruit and non-fruit sources simultaneously as separate predictors of
advanced prostate cancer in a multivariate logistic regression modelthat included age and total energy. Men in the top decile of consumption were at lower risk of advanced prostate cancer relative to menwith the lowest intakes for fruit fructose (RR, 0.55; 95% CI, 0.33-0.92; P = 0.04) and for non-fruit fructose (RR, 0.54; 95% CI,0.36-0.84; P = 0.02)). Modeling both sources as continuous variables yielded similar results for fruit fructose (for a 30-g increment infructose intake: RR, 0.78; 95% CI, 0.61-0.99) and non-fruit fructose(RR, 0.78; 95% CI, 0.63-0.96).
Total fruit intake was inversely associated with risk of advanced
Table 4 RR and 95% CI for total (nonstage A-l I, advanced, and metastatic prostatecancer according to total fructose intake in Health Professionals Follow-up Srud\
(1986-1994)
Total fructose intake(g/day)"%
energy '
Person-YearsTotal
(n)Age-adjustedRR
MultivariateRR'95%
CIAdvanced(n)Age-adjusted
RRMultivariateRR'95%
CIMetastatic(n)Age-adjusted
RRMultivariateRR'95%
CIS40(6.2)
109,2934481.0
1.01451.0
1.0641.0
1.040.1-50(9.0)
89.3873530.94
0.940.82-1.09IK)0.910.930.72-1.19671.25
1.290.91-1.8250.1-60(10.9)
71.0712870.92
0.920.78-1.07940.93
0.960.73-1.26350.77
0.820.53-1.2560.1-70(12.9)
42,1141430.760.760.62-0.93470.78
0.830.59-1.18220.82
0.900.54-1.49>70(16.6)
39,0051380.760.770.62-0.95270.47
0.510.33-0.80130.49
0.590.31-1.12/•trend0.0040.0070.07
" Includes fructose (energy-adjusted) as a monosaccharide plus fructose from glucose.h Mean percentage of energy from fructose in category.' Multivariate model includes age, BM1 at age 21, intake of total energy, total fat,
calcium, phosphorus, vitamin D, vitamin E. and lycopene.
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
RISK OF PROSTATE CANCER AND CALCIUM AND FRUCTOSE INTAKE
Table 5 KR and 95% CI of prostate cancer for a 33 g (15% of energy) increment ttftotal fat, a 50-g (10ck nj total energy) increment of fructose, and a I000-m¡>increment
of calcium
Total fai Total fructose Total calcium
Total prostate cancer (n = 1363)RR (95% CD" 1.09 (0.94-1.25)RR(95%CI)'1.00(0.86-1.17)Advanced
prostate cancer (n = 411)RR(95VrCD"1.34(1.04-1.73)''RR
(95% CI)'1.17(0.89-1.54)Metastatic
prostate cancer (ij —¿�201)RR (95% CI)" 1.50(1.04-2.18)''RR (95% CI)' 1.32 (0.88-1.97)0.75
(0.62-0.90)"0.74(0.61-0.91)''0.59
(0.42-0.83)*0.60(0.42-0.87)'0.57
(0.35-0.93)''
0.61 (0.36-1.04)1.07(0.91-1.26)
1.08(0.91-1.27)1.48(1.14-1.92)'1.52(1.17-1.97)'1.84
(1.28-2.62)''1.88(1.32-2.66)*
" RR from a multivariale model based on modeling the indicated variable (fat. fructose,
or calcium las a continuous variable) controlling forage. BMI ut age 21. intakes of energy,phosphorus, vitamin D. vitamin E. and lycopene. The increments for fat, fructose, andcalcium represent approximately the difference between the 90th and 10th percentiles ofintake for these nutrients.
*PS 0.001.c From multivariate model as above (a) including total fat, fructose, and calcium
simultaneously.''O.Ol < P< 0.05.'0.001 < P £0.01.
prostate cancer (age- and energy-adjusted RR. 0.63; 95% CI, 0.43-0.93, for >5 relative to < 1 serving per day), but this inverse associ
ation did not persist when total fructose was included in this model(RR, 0.87; 95% CI, 0.55-1.36). This finding suggests that the asso
ciation seen for fruit is due to the fructose content of this food.Fructose from carbonated beverages alone (a component of non-fruit
fructose) yielded a similar inverse association (RR, 0.72; 95% CI,0.51-1.03).
The inverse association between fructose intake and risk of totaland advanced prostate cancer did not vary appreciably across levels ofcalcium, phosphorus, total fat. and lycopene intake, and no statistically significant interactions were noted between fructose and any ofthese variables.
Fat. Energy-adjusted fat intake was inversely correlated(r = —¿�0.41)with energy-adjusted fructose intake. As reported in an
earlier follow-up period of this cohort (1986-1990; Ref. 22). energy-
adjusted total fat was significantly related to advanced and metastaticprostate cancer, although in a multivariate analysis (without fructoseand calcium intake) the magnitude of the association was less thanthat observed during earlier follow-up periods (Table 5). Simultane
ous control for fat, fructose, and calcium yielded little change of theresults for calcium and fructose, but the positive association betweentotal fat intake and advanced prostate cancer became attenuated andwas no longer statistically significant (Table 5). These results weresimilar when we substituted animal fat and saturated fat for total fat inthe multivariate model (data not shown). These findings for calciumand fructose also persisted when we substituted total fat in the modelswith a-linolenic acid, for which we had previously reported strong
associations with prostate cancer (22, 32). However, associations fora-linolenic acid persisted but were attenuated (multivariate RR forhigh versus low quintile of a-linolenic acid was 1.33; 95% CI,0.97-1.81; P, trend = 0.08) for advanced cancer and RR = 1.55 (95%CI, 0.97-2.49; P, trend = 0.07) for metastatic prostate cancer).
DISCUSSION
The hypothesis that higher levels of circulating 1,25(OH)2D reduces the risk of prostate cancer is unproven but supported by anincreasing body of in vitro, animal, epidemiológica!, and geneticevidence (1-11), as summarized in the introduction. We thus exam
ined whether dietary factors that may influence 1,25(OH)2D risk arerelated to prostate cancer risk in a large cohort study.
The most obvious dietary factor to influence circulating
1,25(OH)2D is calcium. The predicted plasma levels of 1,25(OH),Dbased on published data (plasma 1,25(OH)2D pmol/l = 99-40 X cal
cium intake mmol/kg/day; Refs. 3 and 33) for our extreme categoriesof intake (medians. 445 and 2346 mg/day) were 39.1 and 28.6 pg/ml.In the two studies that showed lower risks of advanced prostate cancerassociated with higher 1.25(OH)2D levels (8. 17), the cutoff values forthe extreme quartiles were >39 and <27 pg/ml (8) and 40.2 and 28.6(17). Thus, the range of calcium intake in our cohort predicts amagnitude of difference in plasma levels of 1,25(OH),D, which havebeen directly correlated with prostate cancer risk. In this study, wefound dietary and supplementary calcium each associated with higherrisk of extraprostatic, metastatic. and fatal prostate cancer. The association with supplementary sources indicated that calcium, and notanother factor in dairy products, accounted for the increased risk.
Another major determinant of 1.25(OH),D level is circulatingphosphate. A reduction in phosphate levels stimulates renal 1-a-hydroxylase to convert 25(OH)D to 1,25(OH),D, thus raising circulating 1.25(OH)2D (34). Low blood phosphate levels induced experimentally by atypically low phosphorus intakes (500 mg), coupledwith agents that decreased gut absorption of phosphorus, have beenshown to sharply increase 1,25(OH),D blood levels by 80% (34). Inthe same subjects, increasing baseline phosphorus intake from 1500 to3000 mg/day decreased 1.25(OH)2D by 29%. Because of the limitedrange of phosphorus intake in our population (about 1200 to 1800mg/day) and the potentially competing effects of dietary phosphoruson 1.25(OH)2D by binding intestinal calcium and by suppressingparathyroid hormone, the overall impact of phosphorus on1,25(OH)2D are unclear ( 14). Perhaps these complexities may explainwhy we did not observe an association between dietary phosphorusand prostate cancer risk.
Dietary phosphorus deficiency severe enough to lower phosphoruslevels is probably rare, but significant transient reductions of 1-2
mg/dl in plasma phosphate (which averages about 3.5 mg/dl) occurdaily. These fluctuations are likely due primarily to fructose consumption because the high rate of fructose phosphorylation catalyzed byfructokinase ( 14, 15) leads to a sequestration of inorganic phosphateinto the liver (35). Indeed, i.v. or oral consumption of relatively highlevels of fructose (e.g.. 50 g) in controlled situations is known toinduce a profound hypophosphatemia. Several lines of evidence illustrate that the hypophosphatemic effect of fructose occurs in free-
living individuals consuming typical dietary intakes of fructose. Forexample, one of the well-established consequences of reduced phos
phate level is activation of AMP deaminase and nucleotidase, whichinduces increased uric acid production and hyperuricemia (21).Among normal individuals with typical diets, higher intakes of fructose correlate with increases in serum uric acid, strongly suggestingthat fructose intakes in mixed diets appreciably influences phosphatelevels (18. 19. 36).
In this prospective study, we found that high fructose consumption,from either fruit or non-fruit sources, was associated with a reduced
risk of prostate cancer, particularly with advanced disease. Beyondcontrolling for suspected risk factors, we further considered the possibility that fructose was acting as an indirect marker of an actualunconsidered risk factor. This putative risk factor would have to bestrongly and similarly correlated with fructose from fruit and non-fruit
sources because the relationship with prostate cancer risk was virtually identical for these two sources of fructose. Compared to thecorresponding low consumers, the high consumers of fructose fromfruits were much more physically active, unlikely to smoke, and hadquite high fiber intakes, whereas the high consumers of fructose fromnon-fruit sources were less physically active, more likely to smoke,
and were relatively inactive. The very divergent lifestyle patternsbetween high fruit and non-fruit fructose consumers reduces the
445
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
RISK 01 I'KOSTATI. CANCI.R AND CAI.CIl'M AM) IRUTOSI. ISTAK1.
likelihood that a strong unrecognized determinant of prostate cancerwould he equally correlated with fruit and non-fruit fructose, to
produce confounding, while being uncorrelated with numerous otherbehaviors and dietary patterns.
It is important to emphasize that we only examined the link between dietary fructose and prostate cancer risk here and did not studythe proposed intervening steps. We do not know whether any influence of fructose on 1,25(OH)2D levels through hypophosphatemiawould be of sufficient magnitude to influence prostate cancer risk.Also, we cannot rule out alternative mechanisms. Nonetheless, thecharacteristic metabolic, physiological, and clinical consequences offructose taken either p.o. or i.v. are clearly related to the very highactivity of hepatic fructokinase and the resulting sequestration ofphosphate (21 ). Given the importance of hypophosphatemia in determining renal l-a-hydroxylase activity, this proposed mechanism
clearly deserves consideration, although other potential mechanismsshould not be excluded.
We did not observe any relationship between prostate cancer andvitamin D from foods or supplements at total intakes ranging from<150 to >8(X) ILJ/day. This is not surprising because dietary vitaminD influences 25(OH)D levels but not 1,25(OH),D, except perhaps atextremes.
Previous studies of prostate cancer have not examined calcium, butmany have assessed the major calcium sources, milk and other dairyproducts. Per capita milk consumption correlated more highly withnational rates of prostate cancer mortality (;•= 0.66) than other foods
rich in animal fats and cholesterol (meats, r = 0.39; eggs, r = 0.01;
Ref. 37). A correlational study of prostate cancer among regions inItaly found the strongest independent dietary correlations for milk(r = 0.75) and cheese consumption (/•= 0.69; Ref. 38). The most
consistent dietary predictor of elevated prostate cancer risk from sixcase-control and three prospective studies is high milk or dairy con
sumption (13). Magnitudes of RRs between high and low consumption for milk have been approximately 2-fold (13). similar to ourstudy. However, when we considered the additional intake of non-
dairy and supplementary calcium, more striking RRs emerged, presumably because of the greater contrast.
To our knowledge, the hypothesis that high intakes of fructosereduces risk of prostate cancer had not been tested previously, andthere is sparse data regarding fruit intake. One study reported asignificant positive association between fruit intake and prostate cancer risk (39), and another reported a nonsignificant positive association (40). Two studies reported no relationship (41, 42), but a nonsignificant inverse association was observed with dried fruits in thelatter study. Unlike our study, these studies tended to rely on reportedaggregate intakes than of individual items. If fructose is the relevantfactor, it is unclear how well fruit intake in these studies correlatedwith overall fructose consumption.
Although an association with fat intake was noted with advancedprostate cancer, fructose but not fat intake was the independentpredictor of risk in multivuriute analysis. Moreover, fat intake wasstrongly inversely correlated with fruit fructose (Pearson r = —¿�0.44)but only weakly related to non-fruit fructose (r = -0.11); thus, the
similar inverse associations with prostate cancer for both fruit andnon-fruit fructose argues strongly against fat intake directly account
ing for the fructose association. We hypothesi/e that diets high indairy products and meat and low in fruits are associated with increased risk of prostate cancer because these diets tend to be high incalcium and phosphorus and low in fructose, and thus associated withlower circulating 1,25(OH)2D levels. In previous studies (13), including in the Health Professionals Follow-Up Study, fat intake may have
represented an indirect marker of a dietary pattern high in animalproducts and low in fructose. However, we cannot definitively ex
clude an independent role for a-linolenic acid or some close correlate
of this fatty acid.Our findings, if confirmed, have several implications. They provide
indirect support for an influence of dietary determinants of1.25(OH),D levels on prostate cancer carcinogenesis. They also suggest that diets high in animal fat increase the risk of prostate cancerindirectly through their low fructose and high calcium content. Mostimportantly, these results indicate an additional health benefit ofincreased fruit consumption and suggest a potential adverse effect ofhigh calcium intakes specifically among middle-aged and older men.
Given the morbidity and mortality caused by prostate cancer, thesehypotheses warrant further investigation.
REFERENCES
1. Reichel. H.. Koeffler. P.. and Norman. A. W. The role of the vilamin D endocrinesystem in health and disease. N. Engl. J. Med.. 320: 980-991. 1989.
2. Peehl. D. M., Skowronski. R. J., Leung. G. K.. Wong. S. T.. Slamey. T. A., andFeldman. I). Antiproliferative effects of 1,25-dihydroxyvitamin Dt on primary cultures of human prostatic cells. Cancer Res.. 54: 805-810. 1994.
3. Miller. G. J.. Stapleton. G. E.. Hedlund. T. E.. and Moffatt. K. A. Vitamin D receptorexpression. 24-hydroxylase activity, and inhibition of growth by l-a,25-dihydroxyvi-tamin D 3 in seven human prostatic carcinoma cell lines. Clin. Cancer Res., I:997-1003. 1995.
4. Hsieh, T-C., Ng, C-Y.. Mallouh, C., Tazaki. H.. and Wu. J. M. Regulation of growth.PSA/PAP and androgen receptor expression by lu.25-dihydroxyvitamin D, in an-drogen-dependent LNCaP cells. Biochem. Biophys. Res. Commun.. 22.ÃŽ:141-146.
1996.5. Esquenel. M., Swinnen. J. V.. Heyns. W.. and Verhoeven. G. Control of LNCaP
proliferation and differentiation: actions and interactions of androgens. l«.25-dihy-droxycholecalciferol. all-/ra/i.v retinoic acid. 9-c/'.v retinoic acid, and phenylacetate.
Prostate, 2«:182-194. 1996.6. Schwartz. G. G.. Hill. C. C.. Oeler. T. A.. Becich, M. J.. and Bahnson, R. R.
l.25-Dihydroxy-16-ene-23-yne-vitamin D, and prostate cancer cell proliferation inrim. Urology, 46: 365-369. 1995.
7. Lucia. M. S.. Anzano. M. A.. Slayler. M. V., Anver. M. R.. Green. D. M.. Shrader.M. W., Logsdon, D. L., Driver. C. L., Brown, C. C., Peer, C., Roberts, A. B.. andSporn. M. B. Chemopreventive activity of tamoxifen. AM4-hydroxyphenybreti-namide. and the vitamin D analogue Ro24-5531 for androgen-promoted carcinomasof the rat seminal vesicle and prostate. Cancer Res.. 55: 5621-5627. 1995.
8. Corder, E. H., Guess. H. A.. Hulka. B. S.. Friedman. G. D.. Sadler. M.. Vollmer.R. T.. Lobaugh, B. Dre/ner. M. K.. Vogelman. J. H., and Orentreich. N. Vitamin Dand prostate cancer: a prediagnostic study with stored sera. Cancer Epidemiol.Biomark. Prev.. 2: 467-472, 1993.
9. Gann. P. H.. Ma. J., Hennekens. C. H., Hollis. B. W., Haddad. J. G., and Stampfer.M. J. Circulating vitamin D metabolites in relation to subsequent development ofprostate cancer. Cancer Epidemiol. Biomark. Prev.. 5: 121-126. 1996.
10. Taylor. J. A.. Hirvonen. A.. Watson. M.. Pittman. G.. Monier. J. L.. and Bell. D. A.Association of prostate cancer with vitamin I) receptor gene polymorphism. CancerRes., 56:4108-4110. 1996.
11. Ingles, S. A., Ross. R. K.. Yu. M. C.. Irvine. R. A.. LaPera, G.. Haile. R. W., andCoet/.ee. G. A. Association of prostate cancer risk with genetic polymorphisms invitamin D receptor and androgen receptor. J. Nati. Cancer Inst.. X9: 166-170. 1997.
12. Schwartz. G. G.. and Hulka. B. S. Is vilamin D deficiency a risk factor for prostatecancer? (Hypothesis) Anticancer Res.. 10: 1307-1311. 1990.
13. Giovannucci. E. Epidemiologie characteristics of prostate cancer. Cancer (Phila.), 75:1766-1777. 1995.
14. Hollick. M. F., Krane. S. M., and Potts. J. T. Calcium, phosphorus, and bonemetabolism: calcium regulating hormones. In: J. D. Wilson. E. Braunwald. K. J.Isselbacher. R. G. Peiersdorf. J. B. Martin. A. S. Fauci, and R. K. Root (eds.),Harrison's Principles of Internal Medicine. Ed. 12. pp. 1888-1902. New York:
McGraw-Hill. Inc.. 1991.
15. Avioli. L. V. Calcium and phosphorus. In: M. E. Shils and V. R. Young (eds.).Modern Nutrition in Health and Disease. Ed. 7, pp. 142-158. Philadelphia: Lea &Febiger. 1988.
16. Portale. A. A.. Halloran. B. P.. and Morris. R. C. Physiologic regulation of the serumconcentration of 1,25-dihydroxyvitamin D by phosphorus in normal men. J. Clin.Invest.. «.?:1494-1499, 1989.
17. Turner. R. T. Mammalian 25-hydroxyvitamin u-la-hydroxylase: measurement andregulation. In: R. Kumar (ed.). Vilamin D, pp. 175-196. Hingham. MA: MartinusNijhoff Publishing. 1984.
18. Hallfrisch. J., Ellwood. K.. Michaelis. O. E.. Reiser, S.. and Pralher. K. S. Plasmafructose, uric acid, and inorganic phosphorus responses of hyperinsulinemic men fedfructose. J. Am. Coll. Nutr.. 5: 61-68. 1986.
19. Israel. K. D., Michaelis. O. E.. Reiser. S.. and Keeney. M. Serum uric acid, inorganicphosphorus, and glutamic-oxalacetic transaminase and blood pressure in carbohydrate-sensitive adults consuming three different levels of sucrose. Ann. Nutr. Metab..27: 425-435. 1983.
20. Woods. H. F.. Eggleslon. L. V.. and Krebs. H. A. The cause of hepatic accumulationof fructose-1-phosphate on fructose loading. Biochem. J.. 119: 501-510, 1970.
446
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
RISK OF PROSTATE CANCER AND CALCIUM AND FR1XTOSI-. INIAKI
21. Mayes. P. A. Intermediary metabolism of fructose. Am. J. Clin. Nutr.. 58 (Suppl.).-
754S-65S. 1993.22. Giovannucci. E.. Rimm. E. B.. Colditz. G. A.. Slampfer. M. J.. Ascherio, A.. Chute.
C. C. and Wille».W. C. A prospective study of dietary fat and risk of prostate cancer.J. Nati. Cancer Inst., SS: 1571-I579, 1993.
23. Stampfer. M. J.. Wille«.W. C.. Speizer, F. E.. Dysert, D. C.. Lipnick. R.. Rosner. B.,and Hennekens. C. H. Test of the National Death Index. Am. J. Epidemiol.. /19:837-839. 1984.
24. Uniled States Department of Agriculture. Agricultural Handbook No. 8. Compositionof Foods, Raw. Processed and Prepared. 1963-1988. Washington. DC: United Stales
Government Printing Office. 1989.25. Wille».W. C. Nutritional epidemiology. New York: Oxford University Press. 1989.26. Rimm. E. B.. Giovannucci. E. L.. Stampfer. M. J.. Colditz. G. A.. Litin. L. B.. and
Willen. W. C. Reproducibility and validity of an expanded self-administered foodfrequency questionnaire among male health professionals. Am. J. Epidemiol.. /.?5.-114-126. 1992.
27. Feskanich. D., Rimm. E. B., Giovannucci, E. L.. Colditz. G. A.. Stampfer. M. J.. Litin.L. B., and Willett, W. C. Reproducibility and validity of food intake measurementsfrom a semiquantitative food frequency questionnaire. J. Am. Diet Assoc.. 93:790-796. 1993.
28. Park. Y. K.. and Yetley. E. A. Intakes and food sources of fructose in the UnitedStates. Am. J. Clin. Nulr.. 5«(Suppl.): 737S-747S, 1993.
29. Rothman. K. J. Modern Epidemiology. Boston: Little. Brown. 1986.30. Cupples. L. A.. D'Agostino. R.. Anderson. K., and Kännel. W. Comparison of
baseline and repeated measure covariate techniques in the Framingham Heart Study.Slat. Med.. 7: 205-218. 1988.
31. Giovannucci. E.. Rimm. E. B.. Slampfer. M. J.. Colditz. G. A., and Willett. W. C.Height, body weight, and risk of prostate cancer. Cancer Epidemiol. Biomark. Prev..6: 557-563. 1997.
32. Gann. P. H.. Hennekens. C. H.. Sacks. F. M.. Grodstein. F.. Giovannucci. E.. and
Stampfer. M. J. A prospective study of plasma fatty acids and risk of prosiate cancer.J. Nail. Cancer Inst., 86: 281-286, 1994.
33. Gray. R. W.. Lemann. J.. and Adams. N. D. The regulation of I.25-(OH),-Dconcentrations in healthy adults. In: A. W. Norman. K. Schaefer. D. V. Herralh. H-G.Grigoleit, J. W. Coburn. H. F. De Luca, E. B. Mawer, and T. Suda (eds.). Viiamin D:Basic Research and Its Clinical Application, pp. 545-551. New York: Walter deGruyter & Co., 1979
34. Portale, A. A.. Halloren, B. P.. Murphy. M. M., and Morris, R. C.. Jr. Oral ¡makeofphosphorus can determine the serum concentration of l.25-dihydroxyvitamin D bydetermining its production rate in humans. J. Clin. Invest.. 77: 7-12. 1986.
35. Vrana. A., and Fabry. P. Metabolic effects of high sucrose or fructose intake. WorldRev. Nutr. Diet. 42: 56-101. 1983.
36. Narins. R. G.. Weisberg. J. S.. and Myers. A. R. Effects of carbohydrates on uric acid.Metabolism. 2): 455-465. 1974.
37. Rose. D. P.. Boyar. A. P.. and Wynder. E. L. International comparisons of mortalityrates for cancer of the breast, ovary, prostate, and colon, and per capita foodconsumption. Cancer (Phila.), 58: 2363-2371. 1986.
38. Mezzanotte, G.. Cislaghi. C, Decarli. A., and La Vecchia. C. Cancer mortality inbroad Italian geographical areas. 1975-1977. Tumori. 72: 145-152, 1986.
39. Talamini. R.. Franceschi. S.. La Vecchia. C.. Serraino. D.. Barra. S.. and Negri. E.Diet and prostatic cancer: a case-control study in northern Italy. Nutr. Cancer. 18:277-286. 1992.
40. Severson, R. K.. Nomura. A. M. Y., Grove, J. S., and Stcmmermann, G. N. Aprospective study of demographics, diet, and prostate cancer among men of Japaneseancestry in Hawaii. Cancer Res.. 49: 1857-1860. 1989.
41. Snowdon, D. A.. Phillips. R. L.. and Choi. W. Diet, obesity, and risk of fatal prostatecancer. Am. J. Epidemiol., 120: 244-250. 1984.
42. Mills. P. K.. Beeson, W. L.. Phillips, R. L.. and Fraser. G. E. Cohort study of diet,lifestyle and prostate cancer in Advemist men. Cancer (Phila.). 64: 598-604.1989.
447
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
1998;58:442-447. Cancer Res Edward Giovannucci, Eric B. Rimm, Alicja Wolk, et al. CancerCalcium and Fructose Intake in Relation to Risk of Prostate
Updated version
http://cancerres.aacrjournals.org/content/58/3/442
Access the most recent version of this article at:
E-mail alerts related to this article or journal.Sign up to receive free email-alerts
Subscriptions
Reprints and
To order reprints of this article or to subscribe to the journal, contact the AACR Publications
Permissions
Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)
.http://cancerres.aacrjournals.org/content/58/3/442To request permission to re-use all or part of this article, use this link
Research. on November 2, 2020. © 1998 American Association for Cancercancerres.aacrjournals.org Downloaded from
Top Related