A Prospective Analysis of Red and Processed Meat Intake in Relation to Colorectal Cancer in the Black Women's Health Study

ABSTRACT

Background

Black Americans have the highest incidence of colorectal cancer (CRC) of any racial/ethnic group in the United States. High intake of red and processed meats has been associated with an increased CRC risk in predominately White populations. However, 3 prior studies in Black populations, who have been reported to have high intakes of red and processed meats, have reported no associations. Data on a possible association between CRC risk and SFAs and MUFAs, the primary types of fat in red and processed meats, are inconclusive.

Objectives

We prospectively assessed intakes of processed and unprocessed red meat, SFAs, and MUFAs in relation to CRC risk, utilizing data from the Black Women's Health Study (BWHS, 1995–2018).

Methods

Dietary data were derived from validated FFQs completed in 1995 and 2001. Multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were estimated using Cox proportional hazards regression.

Results

Among 52,695 BWHS participants aged 21–69 y at baseline and followed for ≤22 y, 564 women developed incident CRC. Unprocessed red meat intake was associated with a 33% increased CRC risk per 100 g/d (HR: 1.33; 95% CI: 1.03–1.71). Examination of CRC anatomic sites revealed that unprocessed red meat was associated with 2-times increased rectal cancer risk (HR: 2.22; 95% CI: 1.15–4.26). There was no evidence of an interaction with age (p_interaction = 0.4), but unprocessed red meat intake was only associated with a significant increased risk of late-onset CRC (≥50 y of age, HR: 1.41; 95% CI: 1.05–1.88). Processed red meat and total SFA and MUFA intakes were not associated with CRC risk.

Conclusions

Unprocessed red meat intake was associated with an increased CRC risk in the present study, the first positive evidence that red meat plays a role in the etiology of CRC in Black women. The findings suggest prevention opportunities.

See corresponding editorial on page 1181.

Introduction

In 2021, colorectal cancer (CRC) was estimated to be the third most common cancer type and cause of cancer-related death in the United States (1). Black Americans have the highest CRC incidence and mortality rates of any racial/ethnic group, with 5-y relative survival of only 57.4% (2). Further, Black Americans are more likely than other racial/ethnic groups to be diagnosed with or die from CRC at a younger age (2). These disparities persist after adjusting for established risk factors, tumor stage, treatment, and socioeconomic status (3). The etiology of CRC in US Black populations has received relatively little attention, and there are few clues as to why Black Americans suffer disproportionately from CRC—particularly early-onset CRC, defined as arising in persons <50 y of age.

In epidemiologic studies, diets high in red and processed meat have been associated with increased CRC risk (4). One recent case-control study from Italy and Switzerland reported that processed meat, which was ascertained retrospectively, but not red meat, was associated with an increased risk of early-onset CRC (5). A number of mechanisms linking red or processed meat to colorectal carcinogenesis have been postulated, including SFA content (6), mutagens which are by-products of cooking methods (i.e., polycyclic aromatic hydrocarbons, heterocyclic amines, N-nitroso compounds, and advanced glycation end products) (7), heme iron intake (8), and increased insulin resistance (9).

Diet is a key contributor to numerous health disparities (10). It is recognized that diets differ between diverse populations, which could contribute to the disparities in CRC incidence. For example, a recent report suggested that 40% of CRCs in Black Americans are attributable to suboptimal dietary intake, including high intake of red and processed meat, compared with 38% in White Americans (11). Virtually all of the evidence to date on the red or processed meat–CRC association comes from older populations of European ancestry. However, this association may be of particular concern to US Black populations, whose diets encompass a wide range of culinary traditions, from the American South, the Caribbean, South America, and Africa. Until recently, Black Americans have consumed the highest quantities of processed meat of any racial/ethnic group in the United States (12, 13). Thus, it is critical to quantify the red or processed meat–CRC association in diverse populations to understand how diet contributes to disparities in CRC incidence (10). The few studies conducted in US Black populations were restricted to individuals aged ≥40 y and reported null associations between red and processed meat intake and CRC risk (14–16). In the present study, we prospectively assessed the association of total red meat, unprocessed and processed red meat, and dietary SFAs and MUFAs with CRC risk. We also evaluated whether these effects differed for early- compared with late-onset CRC or by anatomic site.

Methods

Study population

The Black Women's Health Study (BWHS) is an ongoing prospective cohort study, which was designed to assess risk factors for disease outcomes in US Black women (17). In 1995, 59,000 women aged 21–69 y were recruited by mailing questionnaires primarily to subscribers of Essence magazine. At study baseline, participants completed a self-administered questionnaire on demographics, medical history, lifestyle factors, and diet. Follow-up has been ongoing for >20 y, with participants completing a questionnaire every 2 y either online or by mail. At the time of this report, follow-up was complete for 85% of potential person-y.

Outcome

Incident cases for this analysis were women diagnosed with primary colon or rectal cancer [defined based on the International Classification of Diseases for Oncology, Third Edition (ICD-O-3); codes C18.0–C18.9, C19.9, C20.9, and C21.8] from 1 July 1996, 1 y after the baseline questionnaire, through 31 December 2018. CRC was further divided into proximal (ICD-10 codes: C18.0, C18.2–C18.5), distal (C18.6–C18.7), and rectal cancer (C19.9 and C20.9). Cases were ascertained by self-report on follow-up questionnaires, linkage with cancer registries in 24 states (covering 95% of participants), and the National Death Index. Pathology data were obtained from hospitals or cancer registries for 75% of cases, of which 95% were confirmed. All self-reported CRC cases are included unless they were disconfirmed by review of pathology data. Early-onset CRC was defined as arising in women <50 y of age, while late-onset CRC was defined as arising in women ≥50 y of age.

Dietary assessment

Dietary data were collected in 1995 using the National Cancer Institute-Block short-form FFQ (68 line items) and again in 2001 (85 line items). Slight modifications were made to the BWHS FFQ to include food items commonly eaten by Black Americans but not previously included on the Block FFQ (18, 19). The BWHS FFQ has been validated using three 24-h recalls as criterion instruments (20). The energy-adjusted and deattenuated Pearson correlation coefficients between the FFQ and 24-h recalls for protein and saturated fat were 0.78 and 0.63, respectively (20).

For each food, the medium portion size was specified, and the participant was asked to report how often she had consumed the food in the past year and the portion size of the food. The portion sizes used in 1995 were small, medium, and large; in 2001, a super-size portion was also included. A small serving was defined as one-half or less of the medium serving; a large serving was one and one-half times the medium serving; a super-size serving was twice the medium serving. The responses for the frequency of consumption ranged from “never or less than 1 per month” to “2 or more per day.” To calculate consumption in grams of red meat, serving size-adjusted frequency of intake was multiplied by the number of grams in a medium serving, estimated as 85 g or ∼3 ounces (85 g). To calculate fatty acid intakes, the serving size–adjusted grams per day for each specific food was multiplied by its fat content (in grams) per 100 grams fresh weight of the edible portion of food, using DIETCALC software, version 1.4.1 (National Cancer Institute).

Processed red meats included bacon, sausage, hot dogs, and lunchmeats (e.g., ham, bologna, and salami). Total red meat included processed red meat in addition to unprocessed red meat—beef (e.g., hamburgers, steak, roast, and stew), lamb, and pork (e.g., chops, roasts, and dinner ham). Saturated fats included butyric acid (4:0), caproic acid (6:0), caprylic acid (8:0), capric acid (10:0), lauric acid (12:0), myristic acid (14:0), palmitic acid (16:0), and stearic acid (18:0). Monounsaturated fats included palmitoleic acid (16:1), oleic acid (18:1), gadoleic acid (20:1), and erucic acid (22:1).

Meat and fat intake were classified into categories based on the distribution of intake among the complete analytic study population, as well as current dietary recommendations (21, 22). A serving size of red meat was defined as 85 g. Total red meat intake was classified as 0 to <2, 2 to <4, and ≥4 servings/wk, consistent with dietary guidelines of ≤3 servings/wk. Unprocessed red meat was classified as 0 to <1, 1 to <3, and ≥3 servings/wk, and processed red meat as 0 to <0.5, 0.5 to <1 and ≥1 serving/wk. SUFAs and MUFAs were classified into tertiles. Tests of linear trend were performed based on the category-specific medians of meat or fat intake. Because continuous meat and fat intake (expressed as g/d) was right skewed, intakes were analyzed as log₁₀-transformed continuous variables. The HR in relation to a 1-unit change in log₁₀-transformed intakes corresponded to 100 g/d of total and unprocessed red meat intake, 50 g/d of processed red meat intake, 10 g/d of total saturated and monounsaturated fat intake, and 1 g/d of individual fatty acids (4).

Dietary intake was examined using the cumulative average approach; thus, in 2001, dietary data from the 1995 and 2001 FFQ reports were averaged. This method reduces measurement error and provides a statistically more powerful test of diet–disease associations (23). For participants who did not complete the 2001 FFQ, the 1995 dietary values were carried forward.

Covariates

Covariates and effect modifiers were obtained at baseline (1995) and updated in 2001. Covariates were selected a priori based on known associations between dietary exposures, potential covariates, and CRC: total energy intake (kcal/d), fiber (g/d), discretionary fat (g/d), BMI (in kg/m²) <25, 25–29.9, or ≥30 kg/m², history of type 2 diabetes, alcohol intake (g/d), cigarette smoking (never, current, former; pack-years), aspirin use (nonregular or regular use, defined as ≥3 d/wk), vigorous physical activity (none, <5 h/wk, ≥5 h/wk), and years of education (≤12, 13–15, 16, ≥17 y).

Exclusions

Of the 59,000 women enrolled at baseline, participants were excluded due to missing dietary data (i.e., >10 blank items) or implausible energy intake (i.e., ≤500 kcal/d and ≥3800 kcal/d) in 1995 (n = 6110), prevalent CRC (n = 134) in 1995, or death within the first year (n = 61). Thus, the current study includes an analytic sample of 52,695 women. The Institutional Review Board of Boston University approved the BWHS protocol and reviews the study annually.

Statistical analysis

Cox proportional hazards regression models, with follow-up time as the underlying time metric, were used to estimate HRs and 95% CIs for association between meat (i.e., total red meat, unprocessed red meat, processed red meat) and fatty acids (i.e., monounsaturated and saturated) with incidence of CRC. Follow-up time was calculated from 1 July 1996 to the first of the following events: occurrence of incident CRC, loss to follow-up, date of death, or end of study follow-up in 2018. Proportional hazards assumptions were tested utilizing an interaction with log(time) in models with confounders, and no violations of the assumption are reported (P > 0.05). All models were stratified by continuous age and time period (2-y questionnaire cycle) and adjusted for total energy intake (using the standard multivariate approach), fiber, discretionary fat, BMI, history of type 2 diabetes, alcohol intake, cigarette smoking, aspirin use, vigorous physical activity, and years of education.

To examine early- compared with late-onset CRC, effect measure modification by age was assessed using likelihood ratio tests comparing regression models with and without a multiplicative term. We also examined joint associations of total red meat intake (<4 or ≥4 servings/wk) with other dietary and lifestyle factors: BMI (<30 or ≥30), waist-to-hip ratio (≤0.75 or >0.75), vigorous activity (ever or never), intake of fruits and vegetables (≥5 or <5 servings/d), whole grains (>0.5 or ≤0.5 servings/d), dairy (>1 or ≤1 servings/d), and discretionary fat (≤45 or >45 g/d). All P values are 2-sided. Statistical analyses were conducted using SAS version 9.4 (SAS Institute).

Results

After exclusions, the analysis included 564 incident CRC cases (420 late onset, 144 early onset) over a median follow-up of 22 y. Median total red meat intake in the BWHS was 34.9 g/d. Compared with women consuming the lowest amount, women consuming ≥4 servings per wk of total red meat were more likely at baseline to be younger, have a higher BMI, have lower educational attainment, smoke cigarettes, drink alcohol, be physically inactive, have a history of type 2 diabetes, and have higher overall energy intake (Table 1 and Supplemental Table S1).

TABLE 1.

Baseline characteristics of study participants by total red meat consumption, Black Women's Health Study¹

	Total red meat servings/wk
	0–<2 (0–<170 g/wk) (n = 18,590)	2–<4 (170–<340 g/wk) (n = 16,749)	≥4 (≥340 g/wk) (n = 17,356)
Age, y (mean ± SD)	40.2 ± 11.0	39.1 ± 10.6	37.1 ± 10.2
BMI (mean ± SD)	26.7 ± 5.7	28.0 ± 6.5	29.3 ± 7.4
Waist-to-hip ratio (mean ± SD)	0.8 ± 0.1	0.8 ± 0.1	0.8 ± 0.1
Education, y (%)
≤12	14	18	22
13–15	33	36	39
16	26	25	22
≥17	26	21	17
Smoking status (%)
Current	11	16	20
Past	21	20	19
Never	69	65	61
Smoking,2 pack-y (mean ± SD)	3.5 ± 8.5	4.3 ± 9.3	5.1 ± 10.3
Alcohol consumption (%)
Never	61	55	52
Past	14	14	15
Current	25	30	33
Missing	1	1	1
Vigorous exercise, h/wk (%)
None	26	32	38
<5	53	53	48
≥5	18	12	10
Missing	4	3	3
Diabetes (%)	3	4	6

¹Values are means ± SDs or percentages of participants and are standardized to the age distributions of the study population.

²Among current and past smokers.

Unprocessed red meat was associated with a 33% increased risk of CRC per 100g/d (HR: 1.33; 95% CI: 1.03–1.71; P = 0.03), but there was no association with processed red meat (per 50 g/d HR: 1.02; 95% CI: 0.83–1.26; P = 0.8, Table 2). As ∼70% of total red meat consumption is unprocessed red meat (Supplemental Table S1), the results for total red meat were comparable to the results for unprocessed red meat: total red meat intake was associated with a nonsignificant 26% increased risk of CRC per 100 g/d (HR: 1.26; 95% CI: 0.94–1.69; P = 0.1, Table 2). Similar results were seen in the categorical analysis: compared with low intake of total red meat (<2 servings/wk), consuming ≥4 servings/wk of total red meat was associated with nonsignificant 25% increased risk of CRC (HR: 1.25; 95% CI: 0.96–1.63; p_trend = 0.1). There was no evidence of an interaction with age (p_interaction = 0.4). However, the association with unprocessed red meat was significant for late-onset CRC (HR_100g/d: 1.41; 95% CI: 1.05–1.88; P = 0.02) but was nonsignificant for early-onset CRC (HR_100g/d: 1.15; 95% CI: 0.70–1.90; P = 0.6). There was no association between processed red meat intake and CRC for either age group.

TABLE 2.

Adjusted HRs and 95% CIs for associations of red meat intake with risk of CRC by age in the BWHS¹

	Participant age, y
	All			≥50			<50
	Cases, n	Person-y	HR (95% CI)2	Cases, n	Person-y	HR (95% CI)2	Cases, n	Person-y	HR (95% CI)2
Total red meat, servings/wk
0–<2 (0–<170 g/wk)	187	375,098	1.00	144	182,795	1.00	43	192,303	1.00
2–<4 (170–<340 g/wk)	192	371,319	1.14 (0.92, 1.41)	150	166,094	1.25 (0.98, 1.60)	42	205,225	0.84 (0.53, 1.32)
≥4 (≥340 g/wk)	185	369,760	1.25 (0.96, 1.63)	126	138,260	1.31 (0.96, 1.79)	59	231,500	1.08 (0.65, 1.79)
Ptrend3			0.1			0.1			0.6
Continuous, per 100 g/d			1.26 (0.94, 1.69)			1.38 (0.98, 1.94)			0.99 (0.56, 1.75)
Processed meat, servings/wk
0–<0.5 (0–<42.5 g/wk)	190	372,769	1.00	147	170,967	1.00	43	201,802	1.00
0.5–<1 (42.5–<85 g/wk)	163	292,516	1.16 (0.93, 1.45)	124	133,648	1.19 (0.92, 1.53)	39	158,868	1.08 (0.69, 1.70)
≥1 (≥85 g/wk)	211	450,891	1.02 (0.81, 1.29)	149	182,533	1.04 (0.79, 1.36)	62	268,358	0.96 (0.61, 1.52)
Ptrend3			0.9			0.9			0.7
Continuous (per 50 g/d)			1.02 (0.83, 1.26)			1.06 (0.83, 1.36)			0.93 (0.62, 1.38)
Unprocessed meat, servings/wk
0–<1 (0–<85 g/wk)	139	287,700	1.00	112	142,804	1.00	27	144,896	1.00
1–<3 (85–<255 g/wk)	267	510,430	1.16 (0.93, 1.44)	198	227,476	1.15 (0.90, 1.47)	69	282,953	1.20 (0.75, 1.90)
≥3 (≥255 g/wk)	158	318,047	1.24 (0.94, 1.65)	110	116,868	1.29 (0.93, 1.79)	48	201,179	1.17 (0.67, 2.06)
Ptrend3			0.2			0.1			0.7
Continuous, per 100 g/d			1.33 (1.03, 1.71)			1.41 (1.05, 1.88)			1.15 (0.70, 1.90)

¹BWHS, Black Women's Health Study; CRC, colorectal cancer.

²The model was adjusted for age (continuous), total energy intake (kcal/d), fiber (g/d), discretionary fat (g/d), BMI (<25, 25–29.9, or ≥30), history of type 2 diabetes, alcohol intake (g/d), cigarette smoking (never, current, former; pack-y), aspirin use (nonregular or regular use, defined as ≥3 d/wk), vigorous physical activity (none, <5 h/wk, ≥5 h/wk), and years of education (≤12, 13–15, 16, ≥17 y).

³Tests of linear trend were calculated by assigning the median of each category as scores.

Results stratified by anatomic site are shown in Table 3. Total and unprocessed red meat intake was associated with a 2-times increased risk of rectal cancer per 100 g/d (HR: 2.13; 95% CI: 1.00–4.53; P = 0.05; and HR: 2.22; 95% CI: 1.15–4.26; P = 0.02, respectively), although the CIs are wide. There was little to no association between total or unprocessed red meat and proximal (HR_100g/d^: 0.99; 95% CI: 0.63–1.54, P = 0.9; and HR: 0.91; 95% CI: 0.63–1.32, P = 0.6, respectively) or distal cancer (HR_100g/d: 0.92; 95% CI: 0.47–1.82, P = 0.8, and HR: 1.28; 95% CI: 0.69–2.36, P = 0.4, respectively). No associations were found for any of the CRC subtypes and processed red meat.

TABLE 3.

Adjusted¹ HRs and 95% CIs for associations of red meat intake with risk of colorectal cancer, stratified by anatomic site, Black Women's Health Study

	Colon cancer
	Proximal		Distal		Rectal
Total red meat servings/wk	Cases, n	HR (95% CI)1	Cases, n	HR (95% CI)1	Cases, n	HR (95% CI)1
0–<2 (0–<170 g/wk)	85	1.00	34	1.00	27	1.00
2–<4 (170–<340 g/wk)	73	0.92 (0.66, 1.29)	28	0.97 (0.57, 1.67)	38	1.58 (0.93, 2.68)
≥4 (≥340 g/wk)	64	0.88 (0.57, 1.33)	30	1.43 (0.75, 2.71)	39	1.72 (0.91, 3.24)
Ptrend2		0.5		0.3		0.1
Continuous, per 100 g/d		0.99 (0.63, 1.54)		0.92 (0.47, 1.82)		2.13 (1.00, 4.53)
Processed meat servings/wk
0–<0.5 (0–<42.5 g/wk)	74	1.00	42	1.00	29	1.00
0.5–<1 (42.5–<85 g/wk)	67	1.23 (0.87, 1.74)	21	0.61 (0.35, 1.07)	31	1.56 (0.91, 2.70)
≥1 (≥85 g/wk)	81	1.04 (0.72, 1.51)	29	0.62 (0.35, 1.09)	44	1.43 (0.81, 2.51)
Ptrend2		0.9		0.1		0.4
Continuous, per 50 g/d		1.08 (0.77, 1.50)		0.72 (0.45, 1.16)		1.29 (0.77, 2.14)
Unprocessed meat, servings/wk
0–<1 (0–<85 g/wk)	66	1.00	25	1.00	18	1.00
1–<3 (85–<255 g/wk)	105	0.94 (0.68, 1.31)	45	1.12 (0.67, 1.89)	54	1.91 (1.07, 3.40)
≥3 (≥255 g/wk)	51	0.79 (0.50, 1.23)	22	1.13 (0.56, 2.28)	32	1.87 (0.92, 3.79)
Ptrend2		0.3		0.8		0.2
Continuous, per 100 g/d		0.91 (0.63, 1.32)		1.28 (0.69, 2.36)		2.22 (1.15, 4.26)

¹Adjusted for age (continuous), total energy intake (kcal/d), fiber (g/d), discretionary fat (g/d), BMI (<25, 25–29.9, ≥30), history of type 2 diabetes, alcohol intake (g/d), cigarette smoking (never, current, former; pack-y), aspirin use (nonregular or regular use, defined as ≥3 d/wk), vigorous physical activity (none, <5 h/wk, ≥5 h/wk), and years of education (≤12, 13–15, 16, ≥17 y).

²Tests of linear trend were calculated by assigning the median of each category as scores.

We examined the joint effects of total red meat intake with other dietary and lifestyle factors (Supplemental Figure S1). Among women with low red meat intake, lower fruit and vegetable intake (<5 servings/d) was associated with 37% higher CRC risk (95% CI: 1.00–1.88, P = 0.05) compared with higher fruit and vegetable intake. Among women with higher fruit and vegetable intake, higher total red meat intake (≥4 servings/d) was associated with 39% higher CRC risk (95% CI: 1.04–1.86, P = 0.03) compared with lower total red meat intake. Women consuming higher intakes of total red meat in combination with lower intake of fruit and vegetables had the highest risk of developing CRC (HR: 1.41; 95% CI: 1.05–1.89, P = 0.02). Similarly, lower intake of whole grains (≤0.5 serving/d) and higher intake of total red meat was associated with a 41% increased risk of CRC (95% CI: 1.05–1.90, P = 0.02), compared with women with high whole grain and low total red meat intake.

Overall, there was no association between saturated or monounsaturated fats with increased risk of CRC (Tables 4 and 5). For intake of 20.0 g/d or more of saturated fat, the OR was 0.89 (95% CI: 0.58–1.39, p_trend = 0.5). For intake of 21.5 g/d or more of monounsaturated fat, the OR was 1.00 (95% CI: 0.70–1.43, p_trend = 0.9).

TABLE 4.

Adjusted HRs and 95% CIs for associations of SFA intake with risk of colorectal cancer, Black Women's Health Study

	Cases, n	Person-y	HR (95% CI)1
Total saturated fat intake, g/d
0.7–<10.0	115	211,840	1.00
10.0–<20.0	272	532,039	0.98 (0.75, 1.27)
≥20.0	173	370,777	0.89 (0.58, 1.39)
Ptrend2			0.5
Continuous, per 10 g/d			0.66 (0.23, 1.91)
Individual SFA intake
Fatty acid 4:0 (butyric acid)
Tertile 1	186	317,912	1.00
Tertile 2	209	455,927	0.89 (0.72, 1.10)
Tertile 3	161	332,950	0.94 (0.71, 1.24)
Ptrend2			0.9
Continuous (per 1 g/d)			0.78 (0.53, 1.17)
Fatty acid 6:0 (caproic acid)
Tertile 1	95	169,697	1.00
Tertile 2	241	499,349	0.99 (0.77, 1.27)
Tertile 3	220	435,876	1.08 (0.81, 1.46)
Ptrend2			0.5
Continuous, per 1 g/d			1.06 (0.63, 1.78d)
Fatty acid 8:0 (caprylic acid)
Tertile 1	154	250,335	1.00
Tertile 2	269	582,975	0.87 (0.70, 1.08)
Tertile 3	132	275,760	0.88 (0.64, 1.21)
Ptrend2			0.4
Continuous, per 1 g/d			0.98 (0.48, 1.98)
Fatty acid 10:0 (capric acid)
Tertile 1	218	392,511	1.00
Tertile 2	142	296,423	0.94 (0.75, 1.17)
Tertile 3	195	419,296	0.90 (0.69, 1.17)
Ptrend2			0.5
Continuous, per 1 g/d			0.76 (0.48, 1.19)
Fatty acid 12:0 (lauric acid)
Tertile 1	199	358,279	1.00
Tertile 2	180	357,789	1.01 (0.81, 1.26)
Tertile 3	173	393,005	0.87 (0.65, 1.17)
Ptrend2			0.2
Continuous, per 1 g/d			0.68 (0.44, 1.06)
Fatty acid 14:0 (myristic acid)
Tertile 1	183	331,578	1.00
Tertile 2	195	394,287	1.04 (0.83, 1.31)
Tertile 3	177	384,455	0.99 (0.72, 1.35)
Ptrend2			0.7
Continuous, per 1 g/d			0.79 (0.47, 1.33)
Fatty acid 16:0 (palmitic acid)
Tertile 1	174	362,342	1.00
Tertile 2	207	378,266	1.25 (0.98, 1.60)
Tertile 3	181	374,644	1.29 (0.86, 1.96)
Ptrend2			0.3
Continuous, per 1 g/d			0.84 (0.26, 2.72)
Fatty acid 18:0 (stearic acid)
Tertile 1	177	354,444	1.00
Tertile 2	200	385,835	1.17 (0.92, 1.49)
Tertile 3	184	374,622	1.32 (0.88, 1.97)
Ptrend2			0.2
Continuous, per 1 g/d			0.68 (0.23, 1.96)

¹Adjusted for age (continuous), total energy intake (kcal/d), fiber (g/d), BMI (<25, 25–29.9, ≥30), history of type 2 diabetes, alcohol intake (g/d), cigarette smoking (never, current, former; pack-y), aspirin use (nonregular or regular use, defined as ≥3 d/wk), vigorous physical activity (none, <5 h/wk, ≥5 h/wk), and years of education (≤12, 13–15, 16, ≥17 y).

²Tests for linear trend were calculated by assigning the median of each category as scores.

TABLE 5.

Colorectal cancer risk across categories of total and individual MUFAs

Total MUFAs, g/d	Cases, n	Person-y	HR (95% CI)1
0–<15.0	172	343,039	1.00
15.0–<21.5	159	306,157	1.05 (0.83, 1.35)
≥21.5	233	466,980	1.00 (0.70, 1.43)
Ptrend2			0.9
Continuous, per 10 g/d			0.83 (0.30, 2.29)
Individual MUFAs
Fatty acid 16:1 (palmitoleic acid)
Tertile 1	176	362,159	1.00
Tertile 2	199	374,057	1.19 (0.95, 1.50)
Tertile 3	188	377,579	1.26 (0.91, 1.75)
Ptrend2			0.2
Continuous, per 1 g/d			1.23 (0.62, 2.46)
Fatty Acid 18:1 (oleic acid)
Tertile 1	178	361,155	1.00
Tertile 2	194	379,334	1.12 (0.87, 1.43)
Tertile 3	191	375,261	1.23 (0.82, 1.85)
Ptrend2			0.4
Continuous, per 1 g/d			1.04 (0.30, 3.57)
Fatty Acid 20:1 (gadoleic acid)
Tertile 1	159	334,821	1.00
Tertile 2	195	396,244	1.01 (0.80, 1.26)
Tertile 3	210	383,173	1.01 (0.76, 1.33)
Ptrend2			0.9
Continuous, per 1 g/d			1.01 (0.61, 1.68)
Fatty acid 22:1 (erucic acid)
Tertile 1	200	412,020	1.00
Tertile 2	152	320,232	0.86 (0.69, 1.07)
Tertile 3	206	374,347	0.89 (0.71, 1.11)
Ptrend2			0.4
Continuous, per 1 g/d			0.99 (0.26, 3.88)

¹Adjusted for age (continuous), total energy intake (kcal/d), fiber (g/d), BMI (<25, 25–29.9, ≥30), history of type 2 diabetes, alcohol intake (g/d), cigarette smoking (never, current, former; pack-years), aspirin use (nonregular or regular use, defined as ≥3 d/wk), vigorous physical activity (none, <5 h/wk, ≥5 h/wk), and years of education (≤12, 13–15, 16, ≥17 y).

²Tests for linear trend were calculated by assigning the median of each category as scores.

Discussion

In this prospective cohort study of Black women, higher intake of unprocessed red meats was associated with a 33% increased risk of incident CRC. There was no evidence of an interaction by age. In particular, unprocessed red meat intake was associated with 2-times increased risk of rectal cancer. Processed red meat intake was not associated with CRC risk, and there were little to no associations with saturated or monounsaturated fats.

Examining diet–cancer associations among different populations is critical, as dietary habits are known to vary between populations—particularly by race/ethnicity (25, 26). Eight cohort studies from predominately White populations have reported that red meat (per 100 g/d) is associated with a 12% increased risk of CRC (95% CI: 1.00, 1.25) (4). Thus, the World Cancer Research Fund/American Institute for Cancer Research (WCRF/AICR) concluded that red meat is a probable cause of CRC (4). However, red meat was not associated with CRC risk based on 4 cohort studies of predominately White women (RR: 1.02; 95% CI: 078, 1.33). Only 3 studies have examined the red meat–CRC association in a US Black population, including 2 case-control studies from North Carolina [n = 255 distal colon cases (14) and n = 276 colon cancer cases (15)] and the Multiethnic Cohort Study (n = 648 CRC cases) (16). None of these studies reported an association between red meat intake, processed or unprocessed, and CRC risk (14–16). Differences between the current study and prior literature could be due to socioeconomic status, age, and/or sex, which are factors known to influence dietary habits. Compared with the participants of prior studies, the BWHS study participants are younger, all female, and more highly educated.

The WCRF/AICR concluded that processed meat is a convincing cause of CRC (4). Ten cohort studies from predominately White populations have reported that processed meat (per 50 g/d) is associated with a 16% increased risk of CRC (95% CI: 1.08, 1.26), for both women and men (4). The WCRF/AICR also reported processed meat was associated with increased risks of both colon and rectal cancer. However, recent studies, including the Nurses’ Health Study, Health Professionals Follow-up Study, and UK Biobank, have reported differences by CRC anatomic site (27, 28). Notably, all 3 of these cohorts have found no association between processed meat and proximal colon cancer (27, 28), the most common type of CRC in Black Americans (29). Thus, the lack of an association between processed meat and CRC in the BWHS could be in part due to the majority of CRCs being proximal colon cancer.

Our current study was able to utilize a prospective dietary assessment to examine etiology in early- compared with late-onset CRC. We found that unprocessed, but not processed, red meat was associated with an increased CRC risk for both early- and late-onset CRC. One prior case-control study from Italy and Switzerland reported that processed meat, but not total red meat, was associated with a 56% increased risk of early-onset CRC (5). However, this prior study of early-onset CRC obtained dietary data at the time of cancer diagnosis for the 2 preceding y (5), and recall may have been biased (30).

The observed association of red meat consumption with CRC risk could have to do with foods that are consumed in conjunction with red meat as part of the overall dietary pattern. Understanding the association between red meat and CRC development requires separating it from its associated food groups. Prior studies have reported that high intakes of fruits/vegetables and fiber are associated with reduced risk of CRC (4, 31). In the current study, we report that women following healthy dietary habits (i.e., high intake of fruits/vegetables or whole grains combined with low red meat intake) had the lowest CRC risk; conversely, low intakes of fruits/vegetables or whole grains and/or high red meat intake was associated with increased CRC risk. This is of particular importance for Black Americans, as prior studies have reported that red and processed meat intake is above the guideline recommended level even among Black Americans who consume a more plant-based dietary pattern (32, 33). Our findings are consistent with what has been reported in the cardiovascular literature, whereby the increased risk of cardiovascular disease associated with high red meat intake was not modified by fruit and vegetable intake (34).

There are several possible mechanisms that could potentially explain an adverse association of red meat with colorectal carcinogenesis, including mutagens from cooking methods (7), heme iron intake (8), or increased insulin resistance (9). Cooking methods and doneness of red meat can affect levels of carcinogens in meat (i.e., polycyclic aromatic hydrocarbons, heterocyclic amines, N-nitroso compounds, and advanced glycation end products) (7). However, the Multiethnic Cohort Study reported no associations between red meat doneness or heterocyclic amines and CRC risk (35). Heme iron is present at high levels in red meat and can also promote endogenous formation of N-nitroso compounds (8). Red meat (9) and saturated fat (36–38) are associated with increasing insulin resistance, which can lead to hyperinsulinemia and subsequent type 2 diabetes. Type 2 diabetes has consistently been associated with an increased risk of CRC (39).

High amounts of red meats are considered to be unhealthy due to their relatively high content of SFAs, which are often assumed to be less healthy compared with other dietary fats such as MUFAs or PUFAs, especially in conjunction with the Western diet. Results from animal studies have reported that diets high in SFA increase CRC risk (40, 41). Postulated mechanisms underlying the SFA–CRC association include inflammation, oxidative stress, increased bile acid production by gut microbes, and stimulation of oncogenic pathways through the epithelial–mesenchymal transition (41, 42). However, there is limited epidemiological evidence to suggest that saturated fat is associated with increased CRC risk. In a recent meta-analysis of 9 prospective cohort studies, the associations between SFA or MUFA and CRC risk were null (43). Similarly, 1 prior case-control study in a Black American population reported no association between higher intakes of SFA or MUFA and distal CRC risk (14). We found no indication that higher intakes of total SFA or MUFA increased CRC risk. This suggests that our findings of an association between red meat intake and CRC risk are not attributable to the SFA and MUFA content of the diet.

Strengths of our study include the prospective design; thus, error in reporting of dietary data is more likely to be nondifferential. Further, dietary assessment was determined at 2 time points, and cumulative averages were used to better reflect long-term dietary intakes. There was a relatively large number of CRC cases, including 144 early-onset CRC. Cases were identified through self-report and through repeated linkage with cancer registries and the National Death Index, with 85% confirmed via hospital, registry, and death records.

Limitations of the current study include generalizability, potential measurement error, and lack of information on CRC screening. The BWHS population is more highly educated than the general US population of Black women. However, they represent 83% of Black women of the same ages nationally who have completed high school or a higher level of education (44), and the estimated fat intake from the FFQs are consistent with estimates from nationally representative US Black adult populations (45). Thus, these results are likely generalizable to most Black American women. FFQs have known measurement errors, but they are nevertheless useful for ranking individuals’ dietary intake, which was our primary objective (23). The BWHS FFQ was validated with criterion instruments (i.e., 3-d food records and three 24-h recalls) and showed high correlations (20), which are within the range of acceptable validity and comparable to those of other dietary validation studies (46–49). Despite the fact that we adjusted for a wide range of potential confounders, unmeasured and residual confounding may still exist. Finally, we do not have information on CRC screening or prior polyps for participants at study baseline.

In summary, we found that higher consumption of red meat—specifically, unprocessed red meat—was associated with a 33% increased risk of incident CRC; this was especially notable for rectal cancer. Our results update and support the recommendation by the World Cancer Research Fund and American Institutes for Cancer Research to limit the consumption of red meat for cancer prevention.

Notes

This research was funded by National Institutes of Health grants U01 CA164974 and R01 CA058420, the Karin Grunebaum Cancer Research Foundation, and the Boston University Peter Paul Career Development Professorship. The funding sources had no role in the design or conduct of the study.

Author disclosures: The authors report no conflicts of interest.

Supplemental Table 1 and Supplemental Figure 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

Abbreviations used: BWHS, Black Women's Health Study; CRC, colorectal cancer; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research

Data Availability

Data described in this article, code book, and analytic code will not be made publically available. Information on the procedure to obtain and access data from the Black Women’s Health Study is described at http://www.bu.edu/bwhs under the information for Researchers.