Cancer Causes & Control

, Volume 18, Issue 1, pp 41–50

Meat and dairy consumption and subsequent risk of prostate cancer in a US cohort study

Authors

  • Sabine Rohrmann
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
    • Division of Clinical EpidemiologyGerman Cancer Research Center
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
    • Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins
    • James Buchanan Brady Urological Institute, Johns Hopkins Medical Institutions
  • Claudine J. Kavanaugh
    • Center for Food Safety and NutritionUnited States Food and Drug Administration
  • Lucy Thuita
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
  • Sandra C. Hoffman
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
    • George W. Comstock Center for Public Health Research and PreventionJohns Hopkins Bloomberg School of Public Health
  • Kathy J. Helzlsouer
    • Department of EpidemiologyJohns Hopkins Bloomberg School of Public Health
    • Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins
    • George W. Comstock Center for Public Health Research and PreventionJohns Hopkins Bloomberg School of Public Health
    • Mercy Medical Center
Original Paper

DOI: 10.1007/s10552-006-0082-y

Cite this article as:
Rohrmann, S., Platz, E.A., Kavanaugh, C.J. et al. Cancer Causes Control (2007) 18: 41. doi:10.1007/s10552-006-0082-y

Abstract

Objective

To evaluate the association of meat and dairy food consumption with subsequent risk of prostate cancer.

Methods

In 1989, 3,892 men 35+ years old, who participated in the CLUE II study of Washington County, MD, completed an abbreviated Block food frequency questionnaire. Intake of meat and dairy foods was calculated using consumption frequency and portion size. Incident prostate cancer cases (n = 199) were ascertained through October 2004. Cox proportional hazards regression was used to calculate hazard ratios (HR) of total and advanced (SEER stages three and four; n = 54) prostate cancer and 95% confidence intervals (CI) adjusted for age, BMI at age 21, and intake of energy, saturated fat, and tomato products.

Results

Intakes of total meat (HR = 0.90, 95% CI 0.60–1.33, comparing highest to lowest tertile) and red meat (HR = 0.87, 95% CI 0.59–1.32) were not statistically significantly associated with prostate cancer. However, processed meat consumption was associated with a non-statistically significant higher risk of total (5+ vs. ≤1 servings/week: HR = 1.53, 95% CI 0.98–2.39) and advanced (HR = 2.24; 95% CI 0.90–5.59) prostate cancer. There was no association across tertiles of dairy or calcium with total prostate cancer, although compared to ≤1 serving/week consumption of 5+ servings/week of dairy foods was associated with an increased risk of prostate cancer (HR = 1.65, 95% CI 1.02–2.66).

Conclusion

Overall, consumption of processed meat, but not total meat or red meat, was associated with a possible increased risk of total prostate cancer in this prospective study. Higher intake of dairy foods but not calcium was positively associated with prostate cancer. Further investigation into the mechanisms by which processed meat and dairy consumption might increase the risk of prostate cancer is suggested.

Keywords

Prostate cancerMeatDairyCohort study

Introduction

Prostate cancer is one of the most commonly diagnosed cancers in the US, but few modifiable risk factors for this disease have been identified. The results of two [1, 2] out of three prospective cohort studies have suggested that higher intakes of processed meat [13] might be associated with a higher risk of this disease. It has also been suggested that high consumption of dairy foods, and, thus, high calcium intake might increase the risk of prostate cancer, especially advanced disease, but the results are still controversial (reviewed in [4]). Several hypotheses have been postulated to explain these associations. Cooking muscle meat at high temperatures, especially barbecuing, grilling, frying, and oven-broiling, leads to the formation of mutagenic heterocyclic aromatic amines [5]. Nitrites in smoked or cured meats can be transformed into carcinogenic N-nitroso compounds by bacteria in the colon [6]. Red meat is high in iron, which can cause oxidative DNA damage [7]. High calcium intake can down-regulate the production of 1,25-dihydroxy vitamin D, which has been shown to inhibit prostate cancer cell growth in vitro [8]. Also, calcium intake and serum concentration of insulin-like growth factor-1 (IGF-1) are positively correlated [9, 10], and a higher IGF-1 concentration has been found to be associated with an increased risk of prostate cancer [11, 12].

We evaluated the association of the intake of total meat, specific types of meat, dairy foods and other rich in calcium, and total dietary and supplemental calcium intake with prostate cancer in the prospective CLUE II cohort of Washington County, Maryland. In addition to the associations with total prostate cancer, we also explored whether the associations differed for low- and high-stage prostate cancer.

Methods

Study participants

In 1989, 25,080 residents of Washington County, Maryland, participated in the CLUE II study named for the study slogan “Give us a clue to cancer and heart disease.” Of the 10,457 male participants, men who were younger than 35 years of age in 1989 (n = 3,150) and who had a diagnosis of cancer prior to the baseline survey (n = 489) were excluded. Follow-up, i.e., knowledge of a participant’s vital status was 93% complete for participants 35–44 years of age at baseline and more than 96% complete for participants 45 years of age and older at baseline.

At baseline, participants were asked to complete an abbreviated version of the Block food frequency questionnaire, which comprised 60 food items [13]. Of the 6,818 eligible men, 5,620 completed and returned the FFQ. Of the respondents, men were excluded from the analysis because more than 32 food items were skipped (n = 193), the number of food items consumed per day was less than three (n = 118), or the calculated energy intake was outside the range of 800–5,000 kcal/day (n = 1,417). A total of 3,892 men were included in the analysis. The 3,892 men included in the analysis had the same mean age at baseline as the 6,818 eligible men (both groups 53.8 years), a similar mean BMI (included: 26.5 kg/m2; eligible: 26.9 kg/m2), and a similar distribution of smoking habits (included: current smokers 17.1%, never smokers 38.3%; eligible: current smokers 18.4%, never smokers 37.1%).

Exposure assessment

On the FFQ, meat consumption was assessed as the frequency of consumption (in nine categories) and the portion size (in three categories: small, medium, and large) of the following food items: hamburger, beef (steaks, roasts etc., including on sandwiches), beef stew, liver, pork (including chops and roasts), fried chicken, turkey, and chicken (roasted, stewed, or boiled, including on sandwiches), hot dogs, ham/bologna/salami/other lunch meats, bacon, and sausages. Also, the consumption frequency and portion size of fried fish, and boiled/baked fish were assessed. Participants were also asked how often they had eaten flame-broiled foods during the past month.

Consumption frequency and portion size of dairy foods (cheese, whole milk, milk with 2% fat, and skim milk, cream, ice cream) also were assessed. Milk consumed with cereal was estimated based on cereal portion size and type of cereal assuming that the same type of milk was used on cereals as was used for drinking. Calcium intake was calculated using the intake from foods and from supplements. In addition, calcium intake was calculated according to source: dairy or vegetable. Daily intake of total energy and total and saturated fat was calculated by summing over all foods the product of the frequency of consumption, the portion size, and the macronutrient content of a food using the DietSys program version 3.0 [13]. Nutrient intake was adjusted for energy intake using residual analysis [14].

Outcome assessment

Incident prostate cancer cases were ascertained by linkage to the Washington County Cancer Registry and since 1992 also to the Maryland Cancer Registry. Maryland death certificates were used to ascertain death from prostate cancer as the underlying cause. Men who were 50 years of age or older and who were considered to be lost to follow-up were also linked to the National Death Index. One hundred ninety-nine men were diagnosed with adenocarcinoma of the prostate, and 20 of these men died of their disease. Information on prostate cancer stage was available for 128 of the 199 cases. We defined low-stage prostate cancer (n = 74) as SEER stages one or two and high-stage (advanced) prostate cancer (n = 54) as SEER stages three or four [15] or death from prostate cancer. Missing information on stage was due to the information not being available in the cancer registries. In the final group of men included in our analysis, 27.1% of the 199 prostate cancer cases were high stage and in the eligible men 21.8% of 330 cases were high stage.

Statistical analysis

We examined the association of the following meat groups with prostate cancer: total meat (hamburgers + beef + beef stew + liver + pork + fried chicken + chicken/turkey + hot dogs + ham/lunch meats + bacon + sausages), beef (beef + beef stew + hamburgers), processed meat (hot dogs + ham/lunch meats + bacon + sausages), red meat (hamburgers + beef + beef stew + pork + hot dogs + ham/lunch meats + bacon + sausages), poultry (fried chicken + chicken/turkey), and fish (fried fish + other fish). We also examined the association of dairy foods (cheese + whole milk + milk with 2% fat + skim milk), cheese, and milk (whole milk + milk with 2% fat + skim milk) with prostate cancer. If the frequency of consumption was missing we assumed that this food item was not consumed. If a man reported consuming a particular food, but did not indicate the portion size, we assumed a medium portion size. Intake in grams per day was used to examine the intake of total meat, red meat, and calcium with prostate cancer. The association of beef, pork, poultry, fish, processed meat, cheese, and milk was evaluated using consumption in portions per month, week, or day. To facilitate comparisons, consumption of dairy foods was examined both in grams and in servings per unit time.

Cox proportional hazards regression was used to calculate hazard ratios (HR) of prostate cancer overall and by stage comparing meat, dairy, and calcium consumption across categories. We adjusted for age (continuous), and in the multivariable model, we additionally adjusted for intake of energy, saturated fat, tomato products (tomatoes/tomato juice + pasta with tomato sauce), and BMI at age 21. We also considered race (black [0.8% of men in the analysis] versus white), BMI at baseline, alcohol consumption, cigarette smoking, and total fat intake, but we did not include these variables in the final model because they did not change the HR appreciably. Trend tests were performed by assigning to each subject the median intake of that third or category and modeling this term as a continuous variable, the coefficient for which was evaluated by the Wald test. To explore if the association of meat, dairy, and calcium intake with prostate cancer differed by age at diagnosis we stratified by age using 70 years old as the cutpoint. We tested for multiplicative interaction by including a cross-product term for age and the exposure variable (binary) in the models along with the main effect terms. The statistical significance of the coefficient for the cross-product term was evaluated using the Wald test. Participants were censored at their date of cancer diagnosis, at their date of death, or at the end of the study period in October 2004, whichever came first. We assumed that persons without information on prostate cancer or death were alive and free of prostate cancer through the end of the study period. All analyses were conducted using SAS version 9.1 (SAS Institute, Cary, North Carolina).

Results

The 3,892 men in the analysis contributed an observation time of 50,466 person-years. The mean age at prostate cancer diagnosis was 69 years, ranging from 46 to 90 years. Men who consumed the most meat were younger, they had a higher intake of total energy and fat, were more likely to be current smokers, and consumed more dairy foods per day (Table 1).
Table 1

Baseline characteristics of study participants by thirds of total meat consumption and dairy food consumption, male participants in the CLUE II cohort, Washington County, MD, 1989

 

Total meat (g/day)

Dairy products (servings/day)

<91.9

91.9–146.3

≥146.4

<0.9

0.9–1.8

≥1.9

Number of participants

1,297

1,297

1,298

1,322

1,288

1,282

Number of prostate cancer cases

83

63

53

61

68

70

Number of high-stage prostate cancer cases

24

17

13

14

21

19

Number of low-stage prostate cancer cases

33

23

18

18

30

26

Age, mean (years)

56.3

53.6

51.4

53.8

52.7

54.8

BMIa, mean (kg/m2)

25.7

26.5

27.4

26.2

26.7

26.7

BMI at age 21, mean (kg/m2)

22.5

22.7

23.0

22.6

22.9

22.7

Energy intake, mean (kcal/d)

1,755

1,926

2,369

1,900

1,965

2,193

Total fat intake, mean (g/d)

71.4

83.6

109.2

82.8

87.9

93.8

Saturated fat intake, mean (g/d)

24.1

28.9

38.0

27.4

30.3

33.6

Intake of total meat, mean (g/day)

56.5

113.5

207.8

123.4

126.1

123.3

Intake dairy foods, mean (servings/day)

1.8

1.6

1.7

0.3

1.4

3.3

Intake tomato products, mean (g/day)

52.0

57.7

67.1

57.3

57.3

61.6

Alcohol intake, mean (g/day)

0.7

0.8

0.8

1.0

0.7

0.6

Use of calcium supplements (%)

27.3

25.9

22.4

23.6

26.5

26.9

Race

Caucasian (%)

98.1

99.5

98.7

99.0

99.1

98.2

African–American (%)

1.2

0.4

1.0

0.5

0.6

1.4

Cigarette smoking status

Never (%)

39.7

37.9

37.2

35.0

43.9

42.3

Former (%)

45.3

44.6

44.4

46.9

37.7

42.9

Current (%)

15.0

17.4

18.4

18.0

18.3

14.8

PSA or DREb (%)

95.6

95.3

93.6

94.4

95.7

93.9

Vasectomyb (%)

27.9

26.2

23.3

25.3

26.5

25.6

Family history of prostate cancerb (%)

5.7

3.4

3.3

3.8

4.1

4.3

a All results (except age) are directly age standardized

b Assessed in 1996

Consumption of meat and fish

Intake of total meat, red meat (Table 2), and specific types of meat (Table 3) were not consistently associated with an increased risk of total prostate cancer or low-stage disease. Processed meat consumption was related to a non-statistically significantly increased risk of total prostate cancer. Fish and poultry consumption were also not associated with total prostate cancer. However, men who consumed higher amounts of processed meat (Table 3) and pork (Table 4) had a higher risk of high-stage prostate cancer. These associations persisted and were even stronger after adjusting for potentially confounding factors. However, these associations were not linearly increasing.
Table 2

Hazard ratio (HR) and 95% confidence interval (CI) of total prostate cancer, high-stage, and low-stage prostate cancer by thirds of daily consumption of total meat and types of meat (g/day), CLUE II cohort, Washington County, MD, 1989–2004

 

Total prostate cancer

High-stage prostate cancer

Low-stage prostate cancer

T1

T2

T3

p-trendb

T1

T2

T3

p-trendb

T1

T2

T3

p-trendb

Total meata

83/16,448

63/16,932

53/17,086

 

24/16,448

17/16,932

13/17,086

 

33/16,448

23/16,932

18/17,086

 

HRc

1.00

0.89

0.88

0.46

1.00

0.88

0.83

0.59

1.00

0.83

0.77

0.36

HRd; 95% CI

1.00

0.91; 0.65–1.27

0.90; 0.60–1.33

0.58

1.00

0.90; 0.48–1.71

0.90; 0.41–1.93

0.76

1.00

0.81; 0.47–1.40

0.66; 0.34–1.29

0.22

Red meata

82/16,374

66/17,091

51/17,001

 

23/16,374

19/17,091

12/17,001

 

34/16,374

23/17,091

17/17,001

 

HRc

1.00

0.89

0.87

0.41

1.00

0.95

0.81

0.57

1.00

0.76

0.71

0.24

HRd; 95% CI

1.00

0.91; 0.65–1.27

0.87; 0.59–1.32

0.52

1.00

0.97; 0.52–1.83

0.87; 0.39–1.93

0.74

1.00

0.73; 0.42–1.27

0.60; 0.31–1.18

0.13

Dairy foodsa

61/17,056

68/17,005

70/16,405

 

14/17,056

21/17,005

19/16,405

 

18/17,056

30/17,005

26/16,405

 

HRc

1.00

1.20

1.07

0.76

1.00

1.69

1.24

0.68

1.00

1.82

1.34

0.48

HRd; 95% CI

1.00

1.22; 0.86–1.73

1.08; 0.78–1.54

0.72

1.00

1.71; 0.87–3.40

1.28; 0.63–2.59

0.59

1.00

1.87; 1.04–3.38

1.31; 0.71–2.41

0.53

Total calcium intakea,e

58/17,249

65/16,791

76/16,426

 

15/17,249

16/16,791

23/16,426

 

17/17,249

30/16,791

27/16,426

 

HRc

1.00

0.96

0.99

0.97

1.00

0.86

1.03

0.82

1.00

1.47

1.16

0.86

HRd; 95% CI

1.00

0.98; 0.72–1.47

0.99; 0.70–1.41

0.99

1.00

0.89; 0.44–1.81

1.06; 0.55–2.04

0.79

1.00

1.50; 0.82–2.74

1.16; 0.63–2.15

0.86

a The cut points for thirds of intake are as follows: total meat 91.85 g/d, 146.44 g/d; red meat 70.14 g/d, 120.64 g/d; dairy foods 0.9 servings/d, 1.9 servings/d; total calcium intake 685.77 mg/d, 957.58 mg/d

b Trend tests were performed using the median intake of each category

c Adjusted for age

d Adjusted for age, energy intake, consumption of tomato products, BMI at age 21, and intake of saturated fat

e Includes calcium intake from supplements

Table 3

Hazard ratio (HR) and 95% confidence interval (CI) of total prostate cancer, high-stage, and low-stage prostate cancer by categories of weekly consumption of meat and milk, CLUE II cohort, Washington County, MD, 1989–2004

 

Total prostate cancer

High-stage prostate cancer

Low-stage prostate cancer

≤1/week

2–4 times/week

5+ times/week

p-trenda

≤1/week

2–4 times/week

5+ times/week

p-trenda

≤1/week

2–4 times/week

5+ times/week

p-trenda

Beef

31/6,291

84/20,203

84/23,972

 

10/6,291

26/20,203

18/23,972

 

9/6,291

30/20,203

35/23,972

 

HRb

1.00

1.00

1.10

0.52

1.00

1.03

0.84

0.53

1.00

1.27

1.65

0.14

HRc; 95% CI

1.00

1.02; 0.67–1.55

1.16; 0.74–1.81

0.40

1.00

1.00; 0.48–2.10

0.83; 0.36–1.92

0.57

1.00

1.31; 0.61–2.78

1.72; 0.79–3.79

0.15

Processed meats

33/9,281

70/16,144

96/25,041

 

7/9,281

20/16,144

27/25,041

 

12/9,281

30/16,144

32/25,041

 

HRb

1.00

1.47

1.38

0.33

1.00

2.09

1.97

0.27

1.00

1.75

1.30

0.98

HRc; 95% CI

1.00

1.55; 1.01–2.38

1.53; 0.98–2.39

0.20

1.00

2.22; 0.92–5.34

2.24; 0.90–5.59

0.22

1.00

1.79; 0.90–3.56

1.30; 0.62–2.74

0.93

Poultry

54/13,816

97/25,463

48/11,187

 

16/13,816

31/25,463

7/11,187

 

18/13,816

33/25,463

23/11,187

 

HRb

1.00

1.04

1.13

0.54

1.00

1.19

0.57

0.18

1.00

1.08

1.62

0.09

HRc; 95% CI

1.00

1.05; 0.75–1.47

1.14; 0.77–1.70

0.51

1.00

1.22; 0.66–2.23

0.60; 0.24–1.49

0.23

1.00

1.07; 0.60–1.91

1.60; 0.85–3.02

0.10

Fish

159/40,600

30/7,620

10/2,246

 

44/40,600

7/7,620

3/2,246

 

62/40,600

9/7,620

3/2,246

 

HRb

1.00

1.12

0.86

0.83

1.00

1.00

0.86

0.81

1.00

0.88

0.65

0.42

HRc; 95% CI

1.00

1.12; 0.75–1.67

0.86; 0.44–1.67

0.81

1.00

1.05; 0.47–2.38

0.92; 0.27–3.21

0.94

1.00

0.85; 0.41–1.72

0.62; 0.19–2.06

0.39

Cheese

81/21,997

47/11,052

71/17,417

 

22/21,997

11/11,052

21/17,417

 

35/21,997

18/11,052

21/17,417

 

HRb

1.00

1.22

1.32

0.10

1.00

1.07

1.58

0.13

1.00

1.09

0.94

0.79

HRc; 95% CI

1.00

1.26; 0.88–1.82

1.43; 1.01–2.03

0.05

1.00

1.11; 0.53–2.30

1.71; 0.88–3.32

0.10

1.00

1.11; 0.62–1.97

0.93; 0.51–1.67

0.78

Milk

49/14,329

14/3,050

136/33,087

 

12/14,329

6/3,050

36/33,087

 

15/14,329

4/3,050

55/33,087

 

HRb

1.00

2.00

1.24

0.35

1.00

4.08

1.37

0.65

1.00

1.97

1.65

0.11

HRc; 95% CI

1.00

2.03; 1.12–3.70

1.26; 0.91–1.74

0.32

1.00

4.20; 1.56–11.3

1.41; 0.73–2.72

0.59

1.00

2.01; 0.66–6.10

1.66; 0.93–2.93

0.11

a Trend tests were performed using the median intake of each category

b Adjusted for age

c Adjusted for age, energy intake, intake of saturated fat, consumption of tomato products, BMI at age 21

Table 4

Hazard ratio (HR) and 95% confidence interval (CI) of total prostate cancer, high-stage, and low-stage prostate cancer by weekly consumption of pork and processed pork foods, CLUE II cohort, Washington County, MD, 1989–2004

 

Total prostate cancer

High-stage prostate cancer

Low-stage prostate cancer

Never

<1/week

≥1/week

p-trenda

Never

<1/week

≥1/week

p-trenda

Never

<1/week

≥1/week

p-trenda

Pork

58/13,918

102/26,092

39/10,456

 

12/13,918

30/26,092

12/10,456

 

28/13,918

31/26,092

15/10,456

 

HRb

1.00

1.10

1.14

0.63

1.00

1.68

1.83

0.28

1.00

0.71

0.92

0.90

HRc; 95% CI

1.00

1.11; 0.80–1.54

1.17; 0.77–1.78

0.55

1.00

1.69; 0.86–3.33

1.98; 0.87–4.53

0.21

1.00

0.70; 0.42–1.17

0.88; 0.46–1.70

0.99

Sausages

75/18,257

81/20,898

43/11,311

 

14/18,257

23/20,898

17/11,311

 

35/18,257

25/20,898

14/11,311

 

HRb

1.00

1.07

1.11

0.63

1.00

1.68

2.48

0.02

1.00

0.72

0.79

0.61

HRc; 95% CI

1.00

1.09; 0.79–1.49

1.16; 0.79–1.73

0.52

1.00

1.76; 0.90–3.44

2.83; 1.34–5.99

0.01

1.00

0.71; 0.42–1.19

0.75; 0.39–1.44

0.54

Bacon

54/14,605

71/16,320

74/19,541

 

11/14,605

21/16,320

22/19,541

 

22/14,605

21/16,320

31/19,541

 

HRb

1.00

1.35

1.25

0.50

1.00

2.03

1.93

0.24

1.00

0.99

1.31

0.26

HRc; 95% CI

1.00

1.39; 0.97–1.99

1.32; 0.91–1.93

0.40

1.00

2.09; 1.00–4.38

2.10; 0.97–4.53

0.20

1.00

1.02; 0.55–1.87

1.35; 0.74–2.44

0.26

Ham/lunch meat

31/8,434

54/13,702

115/28,330

 

7/8,434

17/13,702

30/28,330

 

9/8,434

22/13,702

43/28,330

 

HRb

1.00

1.20

1.44

0.07

1.00

1.80

1.83

0.33

1.00

1.74

1.92

0.17

HRc; 95% CI

1.00

1.24; 0.79–1.93

1.54; 1.01–2.33

0.04

1.00

1.81; 0.75–4.40

1.94; 0.82–4.56

0.28

1.00

1.80; 0.83–3.94

2.00; 0.94–4.25

0.17

Hot dogs

46/11,878

108/25,078

45/13,510

 

9/11,878

32/25,078

13/13,510

 

21/11,878

35/25,078

18/13,510

 

HRb

1.00

1.29

1.08

0.78

1.00

2.05

1.65

0.74

1.00

0.94

0.96

0.96

HRc; 95% CI

1.00

1.32; 0.93–1.87

1.12; 0.73–1.73

0.85

1.00

2.06; 0.97–4.36

1.71; 0.70–4.14

0.73

1.00

0.96; 0.55–1.67

0.96; 0.50–1.88

0.95

a Trend tests were performed using the median intake of each category

b Adjusted for age

c Adjusted for age, energy intake, saturated fat intake, consumption of tomato products, and BMI at age 21

To evaluate further the associations for processed meat and pork, we examined the association of consumption of sausages, hot dogs, bacon, and ham/lunch meats, all processed meat products predominantly made of pork, with total, low-stage, and high-stage prostate cancer. Men who consumed bacon, ham, or hot dogs tended to have a slightly increased risk of total prostate cancer compared with non-consumers but the associations were not linearly increasing (Table 4). In addition, men who consumed sausages or bacon once a week or more were more likely to have high-stage prostate cancer compared with non-consumers. The consumption of pork, ham/lunch meat, and hot dogs was also associated with a non-statistically significantly increased risk of high-stage prostate cancer. No statistically significantly positive associations were observed for low-stage prostate cancer, although ham/lunch meat were possibly directly associated.

The association between processed meat consumption (p-interaction = 0.36) and pork consumption (p-interaction = 0.28) and high-stage prostate cancer did not vary by age at diagnosis. Also, when restricting the analysis to men who had a PSA test in the past (including 22 cases in 1,001 men), the findings were similar to the overall findings for total prostate cancer, but there were too few cases to examine the association with high-stage disease.

Men who consumed flame-broiled foods during the last four weeks prior to the baseline dietary assessment did not have a higher risk of prostate cancer (age-adjusted HR = 1.06; 95% CI 0.79–1.42) or high-stage prostate cancer (age-adjusted HR = 0.95; 95% CI 0.55–1.65) compared to men who did not consume flame-broiled foods.

Consumption of dairy foods and calcium intake

There was no association across tertiles of daily grams of dairy foods consumed with risk of total prostate cancer or high-stage prostate cancer (Table 2). A statistically significantly increased risk of low-stage prostate cancer was observed for those men with a medium but not for those with a high intake of dairy foods (Table 2). Risk of total, high-, and low-stage prostate cancer tended to be higher in the middle category of milk consumption than in the high category. Men who consumed cheese at least five times/week had an increased risk of total and high-stage prostate cancer, but not low-stage prostate cancer (Table 3). When we instead expressed intake of dairy foods on the same servings scale as cheese and milk [≤1/week (19 cases), 2–4/week (14 cases), and 5+/week (168 cases)], compared to those who consumed dairy foods ≤1/week, the multivariate RRs of prostate cancer for 2–4/week and 5+/week were 1.41 (95% CI 0.71–2.83) and 1.65 (95% CI 1.02–2.66).

Median calcium intake from diet and supplements was 814 g/day. Only 15.3% of men had an intake of 1,200 mg/day or higher and only 2.3% had a daily intake of at least 2,000 mg. A high total intake of calcium was not associated with total, low-stage or high-stage prostate cancer (Table 2). Men who consumed 1,200 mg calcium per day or more also did not have a higher risk of total prostate cancer (age-adjusted HR = 0.92, 95% CI 0.63–1.34), low-stage (age-adjusted HR = 0.87, 95% CI 0.47–1.62) or high-stage disease (age-adjusted HR = 1.22, 95% CI 0.64–2.34). The median intake of calcium from supplements was 130 mg/day; only 76 men (2.0%) reported an intake of 500 mg calcium or more per day from supplements. The use of calcium supplements was not associated with a higher risk of total prostate cancer (age-adjusted HR = 0.86, 95% CI 0.62–1.19), low-stage (age-adjusted HR = 1.02, 95% CI 0.56–1.85), or high-stage disease (age-adjusted HR = 1.01, 95% CI 0.60–1.69). We also considered calcium intake from dairy foods separately from vegetable sources (which is about 10% of the total calcium intake in this cohort of men). The HR of total (age-adjusted HR = 1.08, 95% CI 0.76–1.54), low-stage (age-adjusted HR = 1.50, 95% CI 0.82–2.72), or high-stage prostate cancer (age-adjusted HR = 1.10, 95% CI 0.57–2.11) were not statistically significantly higher in men with a high calcium intake from dairy foods compared with men in the lowest tertile of intake. Also, there was no association of calcium intake from vegetable sources with total prostate cancer (age-adjusted HR = 0.81, 95% CI 0.57–1.14), low-stage (age-adjusted HR = 1.01, 95% CI 0.57–1.81), or high-stage disease (age-adjusted HR = 0.71, 95% CI 0.46–1.47). No significant effect modification by age (all p-interactions > 0.05) or varying effects in men who had undergone PSA screening was observed.

Discussion

In this prospective US cohort study, no positive association of total meat, red meat, fish, and poultry intake with total, low-stage, or high-stage prostate cancer was observed. However, we observed possible positive associations of processed meat, pork, sausages, bacon, ham/lunch meat, and hot dogs with high-stage prostate cancer. Also, men who ate processed meat had a non-statistically significantly higher risk of total prostate cancer. Higher intake of dairy foods but not calcium was positively associated with prostate cancer and we could not rule out positive, but non-linear associations for milk and cheese with prostate cancer.

Associations between the consumption of processed meat and prostate cancer, especially advanced prostate cancer, have been reported previously in two out of three prospective cohort studies [13]. A higher risk of metastatic prostate cancer in men with a higher consumption of processed meat and bacon was observed in US men [1]. Also, in a Dutch study cured meat consumption was positively associated with prostate cancer [2]. It has been hypothesized that the fat content, especially saturated fat, might underlie the association between meat consumption and prostate cancer risk by influencing the production of sex steroid hormones [16]. However, the results of studies evaluating the association between prostate cancer and fat intake have been inconsistent [17]. In our analysis, adjustment for intake for saturated fat did not attenuate the association of processed meat or pork with high-stage disease.

Nitrites in smoked or cured meats (added for preservation or improvement of color and taste) can be transformed into carcinogenic N-nitroso compounds by bacteria in the colon [6]. In addition to the endogenous production of N-nitroso compounds it has been shown that the packaging of processed meats may also be important since N-nitrosamines were detected in processed meats packed in elastic rubber nettings [18]. Our finding of positive associations for the consumption of sausage, bacon, ham/lunch meat, and hot dogs, which contain nitrites as a preservative, with high-stage prostate cancer is consistent with this hypothesis.

Another proposed mechanism is the formation of mutagenic heterocyclic aromatic amines in muscle meat cooked at high temperatures [5]. A recent analysis in the Prostate, Colon, Lung, and Ovary Cancer Trial observed a positive association between the intake of PhIP (2-amino-1-methyl-6-phenylimidazo-[4,5-b]pyridine) and prostate cancer [19], but a study from New Zealand did not find an association between heterocyclic amine intake and prostate cancer risk [20]. We did not observe an association of flame-broiled food consumption at baseline and prostate cancer risk. However, only flame-broiled food consumption in the four weeks prior to the baseline dietary assessment was assessed on the food frequency questionnaire, which may not have adequately reflected the usual intake of heterocyclic amines.

We did not observe a statistically significant association between the total consumption of dairy foods and total prostate cancer when considering tertiles of intake. Milk consumption was not consistently associated with an increased risk of prostate cancer; risk tended to be higher in the middle third. The consumption of cheese tended to be associated with an elevated risk of total and high-stage prostate cancer. Consistent with this result, we noted an increased risk of total prostate cancer in men who consume dairy foods 5+/week compared with ≤1/week. However, this result is based on a small number of cases in the comparison group. The association of dairy food intake with prostate cancer is inconsistent across studies [4]. However, two US studies observed an increased risk of advanced prostate cancer with higher intake of dairy products [1, 21]. Three studies observed an increased, although not statistically significantly positive association between cheese consumption and prostate cancer risk [1, 2, 22]. In our study, calcium intake was not associated with a higher risk of total, low-, or high-stage prostate cancer. The results of prior prospective cohort studies that examined the association of calcium intake with prostate cancer are inconsistent, some showing a strong positive association [21, 23, 24], whereas others do not show an association [2, 25, 26]. One reason for these inconsistent results may be the fact that the average calcium intake differed among these studies. In the present study, median calcium intake from the diet and supplements was 830 mg/day, which is lower than the recommended daily allowance of 1,000 mg/day for men younger than 50 years of age and 1,200 mg for men 50 years and older [27]. Positive associations with prostate cancer, especially with advanced disease were largely limited to men with a high calcium intake of 2,000 mg/day or more in other cohort studies [23, 24] and in CLUE II, only 3.2% of men had a calcium intake of 2,000 mg per day or more. The 15% of men who consumed more than 1,200 mg calcium per day did not have a higher risk of total prostate cancer or advanced disease compared to men with a calcium intake of less than 1,200 mg calcium per day in the CLUE II cohort.

This study has several strengths including its prospective design, examination of broad and narrow food groups in relation to prostate cancer overall and exploration by stage at diagnosis, the examination of calcium from diet separate from supplements, and control for possible confounding by dietary and lifestyle factors. Nevertheless, several other aspects of the study warrant discussion. Half of the men who participated in the CLUE II study returned a food frequency questionnaire that was considered to be valid in a general population. Men who were included in the analysis did not systematically differ from the cohort of eligible men before exclusions for the food frequency questionnaire. However, since the present study was a prospective analysis and dietary habits were assessed at baseline prior to the diagnosis of prostate cancer, the modest proportion of valid food frequency questionnaires is unlikely to have affected the internal validity of our results, although the power to detect statistically significant associations was reduced. For the subgroup analyses in particular, we cannot exclude that the positive associations observed might be due to chance based on the number of tests we conducted or to the small number of advanced cases. However, we do not believe that this is the most likely explanation because they are compatible with previously observed associations, especially on processed meat consumption, and risk of prostate cancer.

In conclusion, in this US cohort, men with a higher intake of processed meat and pork were more likely to be subsequently diagnosed with total and advanced prostate cancer compared to men who consumed lower amounts or did not consume these foods at all. In contrast to prior studies, we did not observe consistent associations between calcium intake and prostate cancer although the level of calcium intake in this cohort was lower than in studies that observed a positive association. Further research is warranted examining the possible higher risk of prostate cancer in men who consume larger amounts of processed meat or pork products like sausages, hot dogs, ham/lunch meat, or bacon.

Acknowledgments

We thank Judy Hoffman-Bolton and Alyce Burke at the George W. Comstock Center for Public Health Research and Prevention in Hagerstown, MD, for their continuing efforts in the ongoing CLUE II study. Supported by National Cancer Institute Grant CA08030, National Institute of Aging Grant AG18033, and Department of Defense Grant DAMD17-94-J-4265. Dr. Rohrmann was supported by the Fund for Research and Progress in Urology, Johns Hopkins Medical Institutions. These data were supplied in part by the Maryland Cancer Registry of the Department of Health and Mental Hygiene, Baltimore, MD, which specifically disclaims responsibility for any analyses, interpretations, or conclusions of this study.

Copyright information

© Springer Science+Business Media B.V 2006