Being Breastfed in Infancy and Risk of Colorectal Cancer and Precursor Lesions

Abstract

BACKGROUND & AIMS:

Emerging evidence implicates the importance of perinatal and early-life exposures in colorectal cancer (CRC) development. However, it remains unclear whether being breastfed in infancy is associated with CRC risk in adult life, particularly early adulthood.

METHODS:

We prospectively investigated the association between history of being breastfed and risk of CRC and its precursor lesions among 66,634 women aged 46–93 years from the Nurses’ Health Study and 92,062 women aged 27–68 years from the Nurses’ Health Study II. Cox regression and logistic regression for clustered data were used to estimate hazard ratios for CRC and odds ratios for CRC precursors, respectively.

RESULTS:

During 3.5 million person-years of follow-up, we identified 1490 incident cases of CRC in 2 cohorts. Having been breastfed was associated with a 23% (95% confidence interval [CI], 10%–38%) increased risk of CRC. The risk of CRC increased with duration of being breastfed (P_trend < .001). These findings were validated using breastfeeding information from the mothers of a subset of participants. Among younger participants from the Nurses’ Health Study II, a significant association was observed between being breastfed and increased risk of high-risk adenomas under age 50 (odds ratio, 1.46; 95% CI, 1.16–1.83). Consistently, having been breastfed was associated with increased risk of CRC among participants aged ≤55 years (hazard ratio, 1.38; 95% CI, 1.06–1.80).

CONCLUSIONS:

Being breastfed in infancy was associated with increased risk of CRC in adulthood, including among younger adults. However, further research is needed to understand the underlying biological mechanisms, as this association does not establish causation.

Colorectal cancer (CRC) is the second most common cause of cancer death in the United States.¹ Emerging evidence implicates the role of perinatal and early-life exposures in CRC development.² After experiencing a significant drop between 1975–1990, the U.S. incidence of CRC has risen significantly among individuals aged <50 years (young-onset) since the mid-1990s and among those aged 50–64 years since 2011.^1,3,4 These shifting patterns reflect increased CRC risk in generations born after 1950–1960 and decreased risk in previous generations,^3,4 implying that population-level changes in early-life exposures may be contributing to the trend.

Breastfeeding has numerous benefits for both mothers and infants. However, a prospective UK cohort, the Million Women Study, reported an unexpected but highly significant association between being breastfed in infancy and increased risk of CRC.⁵ The U.S. breastfeeding rates declined to a low of <30% from the early 20th century to the late 1960s, then rebounded and reached >60% in the 1990s, as public perception of breastfeeding changed.^6,7 Coincidentally, the rise in young-onset CRC occurred in birth cohorts with increased breastfeeding rates. Nonetheless, a case-control study nested in the UK Biobank found no association between being breastfed and risk of young-onset CRC.⁸ Due to limited and inconsistent evidence, a systematic and rigorous study of whether being breastfed in infancy is associated wth CRC risk in adult life, particularly early adulthood, is warranted.

In the current study, we prospectively investigated the association between history of being breastfed and risk of CRC and its precursor lesions across the age range among adult women from 2 large U.S. cohort studies with validated data on early-life exposures and detailed assessments of a broad range of CRC risk factors.

METHODS

Study Population

This study analyzed data from the Nurses’ Health Study (NHS), a study initiated in 1976 in which 121,700 U.S. female nurses aged 30–55 years completed a mailed questionnaire on demographic information, lifestyle choices, and medical history.⁹ Another source of data was the Nurses’ Health Study II (NHSII), which was initiated in 1989 and involved 116,430 U.S. female nurses aged 25–42 years who responded to a similar mailed questionnaire.¹⁰ Since enrollment, participants in both cohorts have updated their information through biennial follow-up questionnaires. This study set the baseline as 1992 for NHS and 1991 for NHSII, which were the years when participants reported on whether they were breastfed as infants (Figure 1A). Supplementary Figure 1 shows the sample size derivation. The analysis of CRC comprised 66,634 women aged 46–93 years from NHS and 92,062 women aged 27–68 years from NHSII, who had no prior history of cancer (Figure 1A). For CRC precursors analysis, we included 47,087 women from NHS and 65,631 women from NHSII who underwent at least 1 lower endoscopy. The study population for the analysis of young-onset precursors was further limited to 21,937 NHSII participants who had undergone lower endoscopies under age 50. The study population for CRC and CRC precursors had similar characteristics (Supplementary Table 1).

Figure 1.

Study overview. (A) Schematic of the prospective cohorts. (B) Proportion of participants who were breastfed by year of birth. (C) Distribution of age at diagnosis among cases of colorectal cancer.

We validated our findings using information obtained from the Nurses’ Mothers’ Study, which was conducted in 2001 when participants in NHS and NHSII were asked for permission to send questionnaires to their mothers.¹¹ The questionnaires aimed to gather information on the mothers’ experiences during their pregnancy with the nurse-daughter and the nurse-daughter’s early-life exposures. The validation analysis included 39,683 nurse participants whose mothers reported on whether they breastfed their nurse-daughter.

The study protocol was approved by the institutional review boards of the Brigham and Women’s Hospital and Harvard T.H. Chan School of Public Health and those of participating state cancer registries as required.

Exposure Assessment

NHS participants were asked in 1992: “Were you breastfed?” Those who answered “yes” (as opposed to “no” or “don’t know”) were asked for how long (≤3, 4–8, or ≥9 months or unknown). NHSII participants were asked the same questions in 1991. Participants in both cohorts were also asked whether they were born prematurely and their birth weight. In 2001, the mothers were asked: “Did you breastfeed this daughter?” Those answering “yes” were asked for how long. When self-reported history of being breastfed was compared with their mothers’ reports, 85.0% of nurses and their mothers gave concordant answers about whether they were breastfed; the kappa coefficient was 0.69 (95% confidence interval [CI], 0.68–0.70), indicating substantial agreement (Supplementary Table 2). When duration of being breastfed was considered, 74.4% of nurses and their mothers gave concordant answers; the Spearman correlation coefficient was 0.77 (P < .001; Supplementary Table 3). Covariate assessment is described in Supplementary Methods.

Outcome Assessment

On each biennial questionnaire, participants reported whether they had been diagnosed with CRC or polyps. Incident CRC could also be identified during follow-up of participant deaths ascertained through reports from the next of kin, the U.S. Postal Service, or the National Death Index. We requested written permission from participants or their next of kin to acquire medical records and pathology reports. Physicians blinded to exposure information reviewed the documents to verify the diagnosis and record information on tumor/polyp characteristics. Two subtypes of colorectal polyps, conventional adenomas (tubular, tubulovillous, and villous adenomas) and serrated polyps (hyperplastic polyps, sessile serrated lesions, and traditional serrated adenomas), were ascertained. Polyp size, number, and anatomic location were recorded separately for adenomas and serrated polyps. Histology for adenomas was also recorded. If multiple polyps were diagnosed within a subsite, size and histology were categorized by the largest and most advanced polyp, respectively. Adenomas were classified as advanced if they had the following features: ≥1 cm in size, tubulovillous or villous histology, or presence of high-grade dysplasia.¹² Consistent with the current recommendations for follow-up after colonoscopy, we considered high-risk patients to be those with any advanced adenoma or ≥3 adenomas and low-risk patients to be all other patients.¹² Serrated polyps were subdivided by size as a predictor of malignant potential.¹³

Statistical Analyses

In the analysis of CRC, person-years for participants were calculated from the return of the baseline questionnaire until the occurrence of CRC, death from any cause, or the end of the follow-up period (June 2014 for NHS and June 2015 for NHSII), whichever came first. Hazard ratios (HRs) were estimated from Cox proportional hazards regression conditioned on age and calendar year of the survey cycle, while updating time-varying covariates at each survey cycle. The base model was adjusted for race/ethnicity, history of CRC screening, and history of lower endoscopy for symptomatic reasons. The multivariable model was adjusted for putative risk factors for CRC, including adult height, family history of CRC, history of inflammatory bowel disease (IBD), history of diabetes, regular aspirin use, menopausal status and hormone therapy use, body mass index (BMI), physical activity, pack-years of smoking, alcohol intake, intakes of total calories, red meat, dietary fiber, total calcium, total folate, and total vitamin D, and Alternate Healthy Eating Index-2010. In a third model, we additionally adjusted for preterm birth and birth weight. In the validation analysis, we additionally adjusted for delivery mode, mother’s age at the nurse’s birth, mother’s usual BMI before pregnancy, mother gaining ≥40-lb weight during pregnancy, mother smoking cigarettes during pregnancy, mother drinking alcohol during pregnancy, mother using vitamins during pregnancy, mother receiving higher education, father receiving higher education, and highest parental occupation. To examine the age-dependent association, an alternative to Cox regression was used to estimate HRs as a smooth function of age using B-spline curves (R, version 4.0.3).

In the analysis of CRC precursors, the case-control sets were defined every 2 years and updated through the last questionnaire cycle: all individuals with newly diagnosed colorectal polyps were considered as cases, and individuals who underwent a lower endoscopy in the same cycle but reported no polyp were controls. Once a participant was diagnosed with CRC or polyps, she was not included in all subsequent follow-up cycles. Logistic regression for clustered data was used to account for repeated endoscopies and to estimate odds ratios (ORs). In addition to the aforementioned covariates, models were adjusted for number of reported endoscopies, time since the most recent endoscopy, and reason for the current endoscopy. All analyses were performed with SAS, version 9.4 (SAS Institute, Cary, NC), unless specified otherwise. Two-sided P < .05 was considered statistically significant.

RESULTS

The 2 cohorts span the birth years 1921–1964. Over this period, the proportion of participants who were breastfed declined from 84% to 29% (Figure 1B), aligning with the U.S. historical trend of breastfeeding. Among NHS participants (born in 1921–1946), 42,728 women reported having been breastfed, and 23,906 reported not having been breastfed; among NHSII participants (born in 1947–1964), there were 31,342 breastfed women and 60,720 non-breastfed women. Table 1 and Supplementary Table 4 display participants’ baseline and follow-up characteristics, respectively, while Supplementary Table 5 presents their maternal exposures and parental socioeconomic status, according to their history of being breastfed.

Table 1.

Baseline Characteristics by History of Being Breastfed Among Participants From the Nurses’ Health Study (1992–2014) and the Nurses’ Health Study II (1991–2015)

	Nurses’ Health Study		Nurses’ Health Study II
Characteristic	Not breastfed	Breastfed	Not breastfed	Breastfed
No. of participants	23,906	42,728	60,720	31,342
Age, y	55.7 ± 6.8	59.8 ± 6.9	36.3 ± 4.6	37.1 ± 4.7
Race/ethnicity, %
White	98.7	96.8	97.7	93.6
Black	0.7	1.8	1.2	2.3
Other	0.5	1.4	1.1	4.1
Preterm birth, %	7.3	3.3	8.9	5.7
Birth weight, %
<5.5 lb	14.9	8.5	9.4	5.0
5.5–6.9 lb	32.9	29.5	30.9	29.6
7.0–8.4 lb	41.9	46.8	46.9	50.9
≥8.5 lb	10.3	15.3	12.8	14.4
Early-life body shapea
At age 5 years	2.3 ± 1.4	2.2 ± 1.3	2.5 ± 1.2	2.5 ± 1.2
At age 10 years	2.6 ± 1.5	2.5 ± 1.4	2.8 ± 1.4	2.8 ± 1.3
At age 20 years	2.9 ± 1.2	2.8 ± 1.2	3.1 ± 1.2	3.1 ± 1.1
BMI at age 18 years, kg/m2	21.3 ± 3.0	21.3 ± 2.9	21.3 ± 3.4	21.2 ± 3.2
Adult height, cm	164 ± 6	164 ± 6	165 ± 7	165 ± 7
History of CRC screening, %b	18.9	20.1	1.4	1.2
History of lower endoscopy for symptomatic reasons, %b	17.8	18.9	2.0	2.0
Family history of CRC, %b	13.7	13.8	4.2	4.0
History of inflammatory bowel disease, %b	1.9	1.7	1.1	1.1
History of diabetes, %b	5.8	5.3	1.0	1.0
Regular aspirin use, %b	28.4	28.4	11.2	11.2
Premenopausal, %b	24.7	25.0	96.4	96.6
BMI, kg/m2b	25.9 ± 5.0	26.0 ± 4.9	24.4 ± 5.1	24.3 ± 5.0
Physical activity, MET-h/wkb	17 ± 21	18 ± 19	23 ± 27	22 ± 27
Pack-years of smokingb	14 ± 20	13 ± 19	4 ± 8	4 ± 7
Alcohol intake, g/db	5 ± 10	5 ± 9	3 ± 6	3 ± 6
Red meat intake, servings/db	0.8 ± 0.4	0.8 ± 0.4	0.7 ± 0.5	0.7 ± 0.5
Dietary fiber intake, g/db	18 ± 6	18 ± 5	18 ± 5	19 ± 6
Total calcium intake, mg/db	964 ± 494	997 ± 496	1015 ± 434	1034 ± 445
Total folate intake, μg/db	428 ± 227	434 ± 223	475 ± 293	489 ± 295
Total vitamin D intake, IU/db	336 ± 248	345 ± 250	385 ± 263	397 ± 262
Alternate Healthy Eating Index-2010b	52 ± 11	53 ± 11	48 ± 11	48 ± 11

Note: Mean ± SD is presented unless otherwise indicated.

BMI, body mass index; CRC, colorectal cancer; MET, metabolic equivalent; SD, standard deviation.

^aUsing a 9-level pictogram: 1 = most lean body shape, 9 = most overweight body shape.

^bStandardized by age within each study.

Being Breastfed in Infancy and CRC Risk

During 3.5 million person-years of follow-up, we identified 1490 incident cases of CRC, with 1142 in NHS (median [range] age, 71 [50–92] years) and 348 in NHSII (median [range] age, 52 [33–67] years) (Figure 1C). Despite the difference in age range, having been breastfed was associated with increased risk of CRC in both cohorts. The HRs were 1.21 (95% CI, 1.06–1.38) in NHS and 1.31 (95% CI, 1.06–1.63) in NHSII, which did not materially alter after controlling for preterm birth and birth weight or considering only participants without a family history of CRC (Supplementary Table 6). When the 2 cohorts were combined, having been breastfed was associated with a 23% (95% CI, 10%–38%) increased risk of CRC (Table 2). Moreover, the risk of CRC increased with the duration of being breastfed. Compared to non-breastfed women, those who were breastfed for ≤3, 4–8, and ≥9 months had a 14%, 17%, and 36% increased risk, respectively (P_trend < .001). When considering anatomic location, having been breastfed was associated with a 40% (95% CI, 16%–68%) increased risk of proximal colon cancer and a 17% (95% CI, −1%–38%) increased risk of distal CRC (ie, distal colon and rectal cancer) (Supplementary Table 7). When the association was examined across the age range, being breastfed in infancy remained associated with increased risk of CRC even at advanced ages, particularly proximal colon cancer (Supplementary Figure 2).

Table 2.

Being Breastfed in Infancy and its Duration and Colorectal Cancer Risk Among Participants From the Nurses’ Health Study (1992–2014) and the Nurses’ Health Study II (1991–2015) and Among Those With Reports of Their Mothers From the Nurses’ Mothers’ Study (2001)

				Duration of being breastfed
	Not Breastfed	Breastfed	P	≤3 months	4–8 months	≥9 months	P trend a
Nurses’ Health Studies
Person-years	1,914,584	1,572,628		307,384	315,079	242,186
No. of cases	525	965		100	142	193
HR (95% CI)b	1 (reference)	1.21 (1.08–1.36)	.001	1.13 (0.91–1.40)	1.14 (0.94–1.38)	1.35 (1.13–1.61)	.001
HR (95% CI)c	1 (reference)	1.23 (1.10–1.38)	<.001	1.14 (0.92–1.42)	1.17 (0.97–1.42)	1.36 (1.14–1.63)	<.001
HR (95% CI)d	1 (reference)	1.23 (1.09–1.38)	<.001	1.14 (0.92–1.42)	1.16 (0.96–1.41)	1.35 (1.13–1.62)	.001
Nurses’ Mothers’ Study
Person-years	320,088	295,305		150,783	107,676	34,604
No. of cases	55	94		41	35	17
HR (95% CI)b	1 (reference)	1.48 (1.05–2.11)	.03	1.33 (0.87–2.02)	1.52 (0.98–2.35)	2.03 (1.14–3.64)	.009
HR (95% CI)c	1 (reference)	1.58 (1.11–2.24)	.01	1.41 (0.92–2.15)	1.57 (1.01–2.45)	2.31 (1.28–4.19)	.003
HR (95% CI)e	1 (reference)	1.57 (1.10–2.25)	.01	1.39 (0.90–2.13)	1.58 (1.00–2.48)	2.43 (1.33–4.44)	.003

Note: Participants with unknown duration of being breastfed were excluded from the duration analysis.

CI, confidence interval; HR, hazard ratio.

^aCalculated by entering the duration category (not breastfed, ≤3 months, 4–8 months, or ≥9 months) as an ordinal variable.

^bFrom Cox proportional hazards regression conditioned on age in months and calendar year of the survey cycle and adjusted for race/ethnicity (white or non-white), history of colorectal cancer screening (yes or no), and history of lower endoscopy for symptomatic reasons (yes or no).

^cAdditionally adjusted for adult height (in quintiles), family history of colorectal cancer (yes or no), history of inflammatory bowel disease (yes or no), history of diabetes (yes or no), regular aspirin use (yes or no), menopausal status and hormone therapy use (premenopausal, postmenopausal never user, postmenopausal ever user, or unknown), body mass index (in quintiles), physical activity (in quintiles), pack-years of smoking (0, 1–4, 5–19, 20–39, or ≥40), alcohol intake in g/d (0, 0.1–14.9, or ≥15.0), intakes of total calories, red meat, dietary fiber, total calcium, total folate, and total vitamin D (all in quintiles), and Alternate Healthy Eating Index-2010 (in quintiles).

^dAdditionally adjusted for preterm birth (yes or no) and birth weight in pounds (<5.5, 5.5–6.9, 7.0–8.4, or ≥8.5).

^eAdditionally adjusted for preterm birth (yes or no), birth weight in pounds (<5.5, 5.5–6.9, 7.0–8.4, or ≥8.5), delivery mode (cesarean or vaginal birth), mother’s age at the nurse’s birth (continuous), mother’s usual body mass index before pregnancy (continuous), mother gaining ≥40-lb weight during pregnancy (yes or no), mother smoking cigarettes during pregnancy (yes or no), mother drinking alcohol during pregnancy (yes or no), mother using vitamins during pregnancy (yes or no), mother receiving higher education (yes or no), father receiving higher education (yes or no), and highest parental occupation (professional, managerial, clerical, craftsman, service, laborer, farmer, military, or housewife/don’t work).

We validated our findings among a subset of participants whose mothers provided breastfeeding information. Relating this information to incident CRC yielded associations somewhat stronger than those estimated using the information provided by the participants. Compared to non-breastfed women, those who were breastfed, according to their mothers’ reports, had an HR for CRC of 1.58 (95% CI, 1.11–2.24), and those breastfed for ≥9 months had an HR of 2.31 (95% CI, 1.28–4.19) (Table 2). Adjustment for maternal exposure and parental socioeconomic status did not alter the results. We also examined the feeding pattern during the first 3 months of life. Compared to women who were exclusively breastfed, the HRs were 0.79 (95% CI, 0.52–1.22) for those who were non-exclusively breastfed, 0.53 (95% CI, 0.32–0.86) for those who were exclusively fed cow’s milk, and 0.50 (95% CI, 0.29–0.86) for those who were exclusively fed formula (Supplementary Table 8).

Being Breastfed in Infancy and Younger-Onset CRC and Precursor Lesions

Among NHSII participants, who were enrolled at younger ages, we investigated whether being breastfed in infancy was associated with CRC risk during earlier adulthood. From 1991 to 2015, we identified 116 incident cases of CRC aged <50 years (median [range] age, 45 [33–49] years) and 232 cases aged ≥50 years (median [range] age, 55 [50–67] years). Having been breastfed was associated with increased risk of CRC among participants aged ≥50 years (HR, 1.40; 95% CI, 1.08–1.83), but not among those aged <50 years (HR, 1.18; 95% CI, 0.80–1.73), possibly due to a limited number of cases (Figure 2A and Supplementary Table 9).

Figure 2.

Being breastfed in infancy and risk of colorectal cancer and precursor lesions among participants from the Nurses’ Health Study II (1991–2015). (A) Multivariable hazard ratios (HRs) for colorectal cancer and odds ratios (ORs) for precursor lesions, stratified by age (<50 and ≥50 years). (B) Multivariable ORs for adenomas under age 50, according to pathologic characteristics. HRs were estimated from Cox proportional hazards regression and ORs from logistic regression for clustered data. CI, confidence interval.

As colorectal polyps are much more common than CRC, we also examined the association between being breastfed and risk of precursor lesions under age 50, including conventional adenomas and serrated polyps, which are believed to arise via distinct pathways.¹⁴ Of the 65,631 women who underwent endoscopy, we identified 1201 women with young-onset adenomas and 1270 with young-onset serrated polyps. Having been breastfed was associated with increased risk of young-onset adenomas (OR, 1.25; 95% CI, 1.11–1.41), but not with young-onset serrated polyps (OR, 0.99; 95% CI, 0.88–1.12) (Figure 2A and Supplementary Table 9).

We next considered the pathologic characteristics of young-onset adenomas (Figure 2B and Supplementary Table 10). Having been breastfed was associated with high-risk adenomas (OR, 1.46; 95% CI, 1.16–1.83), which are associated with increased risk of metachronous advanced neoplasia.¹² This association was driven mainly by advanced adenomas (ie, adenomas ≥1 cm in size [OR, 1.64; 95% CI, 1.20–2.23] or with tubulovillous or villous histology [OR, 1.49; 95% CI, 1.11–1.99]), rather than by presence of ≥3 adenomas (OR, 1.18; 95% CI, 0.55–2.52). When considering anatomic location, having been breastfed was strongly associated with high-risk adenomas in the distal colon and rectum (OR, 1.58; 95% CI, 1.22–2.05), while a non-significant trend was observed for adenomas in the proximal colon (OR, 1.46; 95% CI, 0.94–2.28).

Given the significant findings of the adenoma analysis and the fact that adenomas typically take years to develop into cancer, we sought to examine the association between being breastfed and risk of younger-onset CRC by using a more flexible age cut-off. Through a spline analysis, we found that the association between being breastfed and CRC risk increased with age, peaked around 55 years old, and persisted until the late 50s (Figure 3A). Moreover, we examined the association among participants censored at different ages, and consistently found that having been breastfed was associated with increased risk of CRC among those aged ≤55 years (HR, 1.38; 95% CI, 1.06–1.80), particularly distal CRC (HR, 1.45; 95% CI, 1.03–2.05) (Figure 3B and Supplementary Table 11).

Figure 3.

Age-specific association between being breastfed in infancy and risk of younger-onset colorectal cancer among participants from the Nurses’ Health Study II (1991–2015). (A) Multivariable age-varying hazard ratios from Cox B-spline piecewise regression. Solid curve represents point estimates and dashed curves represent 95% confidence intervals (CIs). (B) Multivariable hazard ratios from Cox proportional hazards regression among participants censored at different ages.

Serrated polyps represent another precursor lesion of CRC. Among NHSII participants aged ≤50 years, we found an association between being breastfed and increased risk of serrated polyps (OR, 1.16; 95% CI, 1.07–1.25) (Figure 2A and Supplementary Table 9), which did not appreciably differ by subsite or polyp size (Supplementary Table 12). This finding was further validated in the older NHS cohort (OR 1.11; 95% CI, 1.03–1.20) (Supplementary Table 13).

DISCUSSION

In 2 prospective cohorts of 158,696 women aged 27–93 years, being breastfed in infancy was associated with a 23% increased risk of CRC. The risk of CRC increased with duration of being breastfed. Among younger NHSII participants, having been breastfed was associated with increased risk of high-risk adenomas under age 50 and CRC among those aged ≥55 years. Additionally, having been breastfed was associated with increased risk of serrated polyps after age 50. Taken together, having been breastfed was associated with increased risk of CRC and precursor lesions across a wide range of ages.

Our findings are generally consistent with 2 previous reports, in which having been breastfed was associated with increased risk of CRC, but not among individuals under age 50.^5,8 Our study further discovered a positive association between duration of being breastfed and CRC risk, with those who were breastfed for ≥9 months having the highest risk. These findings were validated by using breastfeeding information collected from the mothers of a subset of participants, among which the association was even stronger.

Within the NHSII cohort, we examined the association between being breastfed and risk of younger-onset CRC and its precursor lesions. Notably, this cohort comprised women born between 1947–1964, who were among the first U.S. generation to experience a rise in young-onset CRC.^3,4 The participants have been followed since the early 1990s, coinciding with the time when the incidence of young-onset CRC began to increase.^3,4 Therefore, this cohort offers a unique opportunity to study young-onset CRC. In this study, having been breastfed was associated with increased risk of conventional adenomas under age 50, particularly high-risk adenomas in the distal colon and rectum. Adenomas are considered the main precursor of young-onset CRC, given that the typical location (the distal colon and rectum) and molecular characteristics (microsatellite stable, CpG island methylator phenotype-low, and BRAF wild-type) of young-onset CRC align with CRC arising from the adenoma-carcinoma sequence.^15,16 We also observed a significant association between being breastfed and increased risk of CRC, particularly distal CRC, among individuals aged ≤55 years. Notably, the rising incidence of younger-onset CRC in the United States, driven by distal CRC, coincides with increased breastfeeding rates, necessitating further studies to establish a definitive link between the two.

Conventional adenomas and serrated polyps, despite their distinct characteristics, share many common risk factors.¹⁷ We found an association between being breastfed and increased risk of serrated polyps after age 50. Previous research has demonstrated that individuals with serrated polyps have a substantially increased risk of proximal colon cancer,¹⁸ which could explain the observed association between being breastfed and increased risk of proximal colon cancer that persisted into late adulthood. It is worth noting that we were unable to distinguish hyperplastic polyps that are generally considered benign from sessile serrated lesions and traditional serrated adenomas. This is because the criteria for diagnosing serrated polyps have evolved over time,¹⁹ and the majority of endoscopies in the study were performed before the routine application of standardized diagnostic criteria.

Breastfeeding is widely recognized for its benefits. While being breastfed is a protective factor for IBD,²⁰ which increases CRC risk,²¹ our findings suggest a robust association between being breastfed and increased risk of CRC, independent of IBD history. Further investigation is needed to understand the underlying mechanisms. One long-standing but controversial hypothesis is that human milk may contain a virus similar to the mouse mammary tumor virus, which could be transmitted to offspring and increase their risk of cancer.²² Breastfeeding has been shown to shape the infant’s gut microbiota differently from non-breastfed infants,²³ and there is growing evidence that the gut microbiome plays a role in CRC development.²⁴ Additionally, the lower risk of CRC in non-breastfed individuals may be linked to the potential benefits of cow’s milk and formula fortified with vitamins and minerals. While consuming specific vitamins and minerals, such as vitamin D, in adulthood has been associated with decreased risk of colorectal neoplasia, including young-onset CRC,²⁵ the effects of consuming these nutrients during early life are poorly understood.

This study has important strengths. In the 2 cohorts, participants had a wide range of ages and birth years and were prospectively followed for >20 years. Detailed information on colorectal polyps by subtype, anatomic location, and malignant potential was extracted by investigators blinded to exposure information. We validated our findings in a subset of participants with reports of their mothers. The current study also has limitations. While we considered and adjusted for a broad range of CRC risk factors and potential confounders, residual confounding is still possible. The primary analysis relied on self-reported history of being breastfed, which was subject to error. However, the misclassification is expected to be non-differential and bias our findings toward the null. Indeed, stronger associations were observed when using breastfeeding information from the mothers. Finally, our study population primarily consisted of white women, which may limit the generalizability to men or other racial and ethnic groups.

In conclusion, being breastfed in infancy was associated with increased risk of CRC in adulthood, including among younger adults. However, it is important to note that this association does not establish causation. Furthermore, our study does not serve as a discouragement of breastfeeding, but rather highlights the pressing need to investigate the underlying mechanisms and explore potential interventions. Breastfeeding has well-established and essential benefits that likely outweigh any potential increased risk of CRC in adulthood.

What You Need to Know.

Background

Whether being breastfed in infancy is associated with CRC risk in adult life, particularly during early adulthood, remains unclear.

Findings

Among women participating in 2 large prospective U.S. cohorts, a history of being breastfed was found to be associated with a 23% increase in colorectal cancer risk, as well as an elevated risk of younger-onset colorectal cancer and adenomas.

Implications for patient care

It is important to note that the observed association does not establish causation. Further research is needed to understand the underlying biological mechanisms.

Abbreviations used in this paper:

BMI

body mass index

CI

confidence interval

CRC

colorectal cancer

HR

hazard ratio

IBD

inflammatory bowel disease

MET

metabolic equivalent

NHS

Nurses’ Health Study

NHSII

Nurses’ Health Study II

OR

odds ratio

SD

standard deviation