Garlic intake and gastric cancer risk: results from two large prospective US cohort studies

Abstract

Although many case-control studies suggested that garlic intake may reduce gastric cancer risk, evidence from prospective cohort studies has been lacking. We examined the association between garlic intake and subsequent risk of gastric cancer among 77,086 women in the Nurses’ Health Study (1984–2014) and 46,398 men in the Health Professionals Follow-up Study (1986–2014). Relative risks (RRs) and 95% confidence intervals (95% CIs) were estimated using Cox proportional hazards models. We additionally examined garlic intake in relation to Helicobacter pylori (H. pylori) infection among 613 participants using logistic regression. During up to 30 years of follow-up, 292 participants were diagnosed with gastric cancer. The pooled multivariable RR of gastric cancer among participants who ate garlic, as compared with those who did not, were 1.11 (95% CI = 0.81 to 1.51) for the intake of garlic less than once per week, 0.98 (95% CI = 0.71 to 1.36) for one to four times per week, and 1.39 (95% CI = 0.89 to 2.17) for five or more times per week (P for trend = 0.23). Similarly, no statistically significant association was observed cross-sectionally between garlic intake and H. pylori infection (comparing five or more times per week to never, pooled multivariable odds ratio = 1.66, 95% CI = 0.89 to 3.09; P for trend = 0.11). The findings from this large prospective study do not support the hypothesis that high garlic intake reduces risk of gastric cancer.

INTRODUCTION

Over the last several decades, gastric cancer incidence and mortality have declined in most countries around the world¹. However, gastric cancer remains as the fifth most common cancer and the third leading cause of cancer death worldwide².

Garlic has been studied for its antimicrobial, antifungal, antihypertensive, hypoglycemic, and antitumor properties³. In addition to amino acids, vitamins, and micronutrients, garlic contains various organosulfur compounds including diallyl sulfide⁴. Experimental and human studies suggested that these organosulfur compounds in garlic have chemopreventive and other beneficial effects^3–5, affecting each stage of carcinogenesis and influencing physiologic processes that modify cancer risk⁶.

An inverse association between garlic intake and gastric cancer was derived mainly from case-control studies^{5, 7–10}. A recent meta-analysis including 12 case-control studies estimated pooled relative risk (RR) of 0.60 (95% confidence interval, CI, 0.47 to 0.76) for the highest versus lowest category of garlic intake¹⁰. Evidence from cohort studies, however, has been limited and inconclusive. The only cohort study to date to examine garlic and gastric cancer risk focused on garlic supplement intake rather than raw or cooked garlic intake¹¹.

We therefore investigated the association between garlic intake and gastric cancer risk in two large prospective US cohort studies. In addition, because garlic has antibacterial properties and has been suggested to inhibit Helicobacter pylori (H. pylori) growth in experimental studies^{12, 13} and an epidemiological study¹⁴, we examined the cross-sectional association between garlic intake and H. pylori infection, an established risk factor of gastric cancer¹⁵.

MATERIALS AND METHODS

Study Population

The Nurses’ Health Study (NHS) is a prospective cohort study that started in 1976 with 121,700 U.S. female nurses who were 30 to 55 years of age¹⁶. The Health Professionals Follow-up Study (HPFS) is a prospective cohort study that began in 1986 with 51,529 U.S. male health professionals who were 40 to 75 years of age¹⁷. Questionnaires were sent to participants at baseline and every 2 years thereafter to gather information on demographics, lifestyle factors, and medical history. Diet was first assessed in 1980 for NHS and 1986 for HPFS, and generally updated every 4 years.

For the current analysis, we used 1984 for the NHS and 1986 for the HPFS as our baseline when garlic intake was first assessed on the dietary questionnaire. At baseline, we excluded participants who did not provide information on garlic intake, or who had cancer (excluding non-melanoma skin cancer). The final analyses included 77,086 women in the NHS and 46,398 men in the HPFS.

Assessment of Garlic Intake

Garlic intake was assessed with a validated semiquantitative food frequency questionnaire (FFQ) in 1984, 1994, 1998, 2002, and 2010 for the NHS and in 1986, 1994, 1998, 2002, and 2010 for the HPFS. Specifying the serving size (1 clove or 4 shakes), the FFQ asked participants to select their average frequency of garlic intake (fresh or powdered) over the past year out of the 9 choices: never or less than once per month, 1–3 per month, 1 per week, 2–4 per week, 5–6 per week, 1 per day, 2–3 per day, 4–5 per day, and ≥ 6 per day. The FFQ also assessed garlic supplement use (yes or no) in 1998 and 2002 for the NHS and biennially from 1996 and 2006 for the HPFS.

Assessment of Covariates

To minimize potential confounding, information on known or suspected risk factors for gastric cancer, such as age, race, smoking status, alcohol consumption, multivitamin use, aspirin use, history of diabetes mellitus, and intakes of total calorie, red/processed meat, fruits, vegetables, and coffee, was collected from the baseline and subsequent questionnaires. Body mass index (BMI) was calculated based on reported height and weight.

For physical activity, participants were asked about the average amount of time spent per week in the previous year in recreational physical activities (e.g. walking, jogging, swimming, etc). Based on this information and intensity of each physical activity as expressed by Metabolic Equivalent of Task (MET)¹⁸, MET-hour/week was calculated to represent total weekly energy expenditure contributed by these activities.

Ascertainment of Gastric Cancer

Gastric cancer cases are identified by self-report or the national death index. Participants reported their diagnosis of gastric cancer through biennial follow-up questionnaires. Follow-up response for each biennial questionnaire was over 90%, and death ascertainment was 98%. We obtain medical records from 90% of cases through permission from the case or next of kin. For cases without records, we link to cancer registries to confirm the diagnosis. Study physicians blinded to the participants’ exposure status reviewed medical records to confirm the self-reported cancer diagnosis and to classify histology and tumor location (proximal or distal). For this analysis, we included incident gastric cancer cases diagnosed between the return of the baseline questionnaire and the end of the analysis (May 31, 2014 for the NHS; January 31, 2014 for the HPFS). Cases with non-adenocarcinoma histology were excluded.

Ascertainment of H. Pylori Infection Status

Between 1989 and 1990 in the NHS and between 1993 and 1995 in the HPFS, a subset of participants in each cohort provided blood samples. Results from blood H. pylori IgG antibody test were available for 350 women and 263 men. All samples were handled identically and together, shipped in the same batch, and assayed in the same analytical run. The H. pylori IgG antibodies were assayed in the laboratory of Dr. Nader Rifai (Children’s Hospital, Boston, MA) using Enzyme-Linked Immunosorbent Assays (ELISA)¹⁹. As validated by the manufacturer (ALPCO Diagnostics, Salem, NH), the sensitivity and specificity for this method are 91% and 95% respectively, which are comparable to the diagnostic performance of other commercially available H. pylori serology kits¹⁹. The mean intra-assay coefficients of variance were <10% for blinded, replicate quality control samples. As per the manufacturer’s instructions, one was defined as H. pylori infected when the level of blood antibody to H. pylori was greater than 7 units/ml.

Statistical Analysis

In examining the association between garlic intake and gastric cancer risk, person-time was accrued from the date of the baseline questionnaire return to the date of gastric cancer diagnosis, death, or end of follow-up, whichever came first. Setting age as the time metric, we used the Cox proportional hazards model to calculate hazard ratios (HRs) and their 95% CIs from each cohort²⁰. These HRs were used to estimate RRs. The results from each cohort were combined using a fixed-effect meta-analysis based on inverse weight²¹.

Garlic intake was classified into four categories: never (as reference group), 0–<1 per week, 1–4 per week, and ≥ 5 per week. To account for potential changes in garlic intake over a long-term follow-up, we updated garlic intake whenever new information was collected from the follow-up questionnaires. We calculated the cumulative average using methods for repeated measures described previously²², which better represents a long-term garlic intake and minimizes random within-person variation. Test for trend was examined by modeling the median garlic intake of each category as a continuous variable. To address the concern that occult gastric cancer in the years that preceded diagnosis may have influenced dietary patterns, we repeated the analyses using baseline garlic intake instead of cumulatively updated values; in addition, in a separate sensitivity analysis, we excluded the first 2 years of follow-up data and added a 2-year lag period between garlic intake assessment and each follow-up period. To explore potential heterogeneity of the relationship between garlic intake and gastric cancer risk, we performed subgroup analyses stratified by factors including garlic supplement use, age, BMI, physical activity, smoking, and intakes of alcohol, red meat, fruits, and vegetables. In addition, to address potential differences in the association between risk factors and different subtypes of gastric cancer, we performed stratified analysis of gastric cancer cases by stratifying them into proximal and distal gastric carcinoma.

In examining the association between garlic intake and H. pylori infection, we used logistic regression to calculate odds ratios (ORs) and their 95% CIs. The analysis used the garlic intake and covariates as they were defined for the analysis of garlic intake and gastric cancer risk.

All the statistical tests were two-sided and P values <0.05 were considered statistically significant. Analyses were performed using SAS 9.3 (SAS Institute, Cary, NC).

RESULTS

In the NHS, we observed 138 gastric cancer cases during 30 years of follow-up (2,087,397 person-years), and in the HPFS, we observed 154 cases during 28 years (1,086,656 person-years). Participants with higher garlic intake tended to be non-white, physically active, ever-smokers, drink more alcohol, and have higher intakes of fruits, vegetables, and coffee; they were also more likely to be diabetic (Table 1).

Table 1.

Characteristics of person-time according to frequency of garlic intake^*

Characteristics	Nurses’ Health Study (1984–2014): Garlic intake (1 clove or 4 shakes)				Health Professionals Follow-up Study (1986–2014): Garlic intake (1 clove or 4 shakes)
	Never	0-<1 per week	1–4 per week	≥5 per week	Never	0-<1 per week	1–4 per week	≥5 per week
Person-years	453431	607236	834130	192600	372505	326969	321151	66031
Age, y, mean (SD)	61.5 (10.8)	64.5 (10.8)	63.6 (10.7)	62.2 (10.3)	63.5 (11.6)	65.3 (11.2)	64.4 (11.1)	63.4 (11.0)
Caucasian, %	98.4	98.4	97.5	95.0	95.8	95.0	94.6	92.3
History of diabetes mellitus, %	8.3	8.3	8.3	9.9	7.2	7.9	8.6	8.8
Body-mass index, kg/m2, mean (SD)	25.5 (4.6)	25.4 (4.5)	25.4 (4.6)	26.0 (4.9)	25.7 (3.5)	25.8 (3.3)	25.9 (3.4)	26.3 (3.8)
Physical activity, MET-hours/wk, mean (SD)	14.7 (16.4)	15.4 (15.5)	17.4 (17.2)	19.3 (19.8)	28.0 (28.6)	30.9 (27.9)	32.6 (29.1)	34.2 (30.7)
Smoking, %
Never smokers	52.5	46.6	41.0	38.1	51.4	48.3	43.1	38.7
Past smokers	35.3	41.2	46.2	49.5	41.9	45.3	49.9	53.6
Current smokers	12.2	12.1	12.7	12.4	6.6	6.4	7.0	7.7
Alcohol intake, g/d, mean (SD)	3.9 (7.3)	5.2 (8.2)	7.1 (9.7)	8.4 (11.0)	8.4 (12.5)	10.8 (13.2)	13.5 (14.6)	15.8 (16.6)
Aspirin use, %	35.0	35.9	36.0	35.6	39.7	45.4	44.8	40.9
Multivitamin use, %	48.0	53.9	54.9	54.0	42.5	49.7	49.5	48.8
Red or processed meat intake, servings/d, mean (SD)	1.0 (0.5)	1.0 (0.5)	1.0 (0.5)	1.0 (0.5)	1.1 (0.7)	1.1 (0.7)	1.0 (0.7)	1.0 (0.7)
Fruits intake, servings/d, mean (SD)	2.2 (1.1)	2.2 (1.0)	2.3 (1.1)	2.6 (1.2)	2.3 (1.4)	2.4 (1.4)	2.6 (1.5)	3.0 (1.8)
Vegetables intake, servings/d, mean (SD)	2.5 (1.1)	2.6 (1.0)	3.0 (1.1)	3.6 (1.4)	2.8 (1.4)	3.1 (1.5)	3.6 (1.6)	4.6 (2.0)
Coffee intake, servings/d, mean (SD)	1.6 (1.4)	1.7 (1.4)	1.7 (1.3)	1.8 (1.3)	1.8 (1.7)	1.9 (1.6)	1.9 (1.6)	2.0 (1.6)

Abbreviation: MET: metabolic equivalent task

^*All variables (except age) are age-adjusted.

In both age-adjusted and multivariable analyses, garlic intake was not significantly associated with risk of gastric cancer (Table 2). The pooled result showed that compared to those who never consumed garlic, participants who consumed garlic ≥ 5 times per week experienced a multivariable-adjusted RR of 1.39 (95% CI = 0.89 to 2.17; P for trend = 0.23). P values for heterogeneity between women and men were greater than 0.05 for all categories of garlic intake. The results remained largely unchanged when we repeated the multivariable analyses using the baseline garlic intake instead of cumulatively updated values or when we excluded the first 2 years of follow-up data and added a 2-year lag period between garlic intake assessment and each follow-up period.

Table 2.

Relative risks and 95% confidence intervals for gastric cancer according to garlic intake^*

	Frequency of garlic intake (1 clove or 4 shakes)				P for trend
	Never	0-<1 per week	1–4 per week	≥5 per week
NHS
Cases	31	44	45	18
Person-years	453431	607236	834130	192600
Age-adjusted	1.00	1.05 (0.66, 1.68)	0.82 (0.51, 1.31)	1.45 (0.81, 2.60)	0.33
Multivariable-adjusted	1.00	1.08 (0.67, 1.73)	0.81 (0.50, 1.31)	1.34 (0.72, 2.47)	0.50
HPFS
Cases	48	47	46	13
Person-years	372505	326969	321151	66031
Age-adjusted	1.00	1.13 (0.75, 1.70)	1.18 (0.78, 1.78)	1.65 (0.89, 3.05)	0.12
Multivariable-adjusted	1.00	1.13 (0.75, 1.72)	1.15 (0.74, 1.77)	1.45 (0.76, 2.78)	0.29
Pooled†
Age-adjusted	1.00	1.09 (0.80, 1.49)	1.01 (0.74, 1.37)	1.54 (1.01, 2.35)	0.07
Multivariable-adjusted	1.00	1.11 (0.81, 1.51)	0.98 (0.71, 1.36)	1.39 (0.89, 2.17)	0.23

^*Multivariable models were adjusted for age (continuous), Caucasian (yes vs. no), body-mass index (<23, 23–24.9, 25–29.9, 30–34.9, ≥35 kg/m²), physical activity (MET-hours/week, quintiles), smoking status (never, past, current), alcohol consumption (women: 0, 0.1–4.9, 5.0–14.9, ≥15 g/d; men: 0, 0.1–4.9, 5.0–29.9, ≥30 g/d), current multivitamin use (yes vs. no), current aspirin use (yes vs. no), personal history of diabetes mellitus (yes vs. no), and intakes of total calorie, red/processed meat, fruits, vegetables, and coffee (quintiles).

^†The results were combined using a fixed-effect model based on inverse variance weight. For all categories of garlic intake, P values for heterogeneity between women (NHS) and men (HPFS) were ≥ 0.25.

No significant interactions were observed between garlic and garlic supplement use, age, BMI, physical activity, smoking, or intakes of red meat, fruits, and vegetables (all P for interaction > 0.05) but there was a marginal significance (P for interaction = 0.05) in the interaction between garlic and intake of alcohol whereby a positive association between garlic intake and gastric cancer was observed in alcohol non-drinkers (Supplementary Table 1).

In addition, no significant associations were observed when we performed separate analyses of proximal and distal gastric cancers, the two different subtypes of gastric cancer (Supplementary Tables 2 and 3). For both proximal and distal gastric cancers, the results remained largely unchanged when we repeated the analyses using the baseline garlic intake or when we excluded the first 2 years of follow-up data and added a 2-year lag period between garlic intake assessment and each follow-up period.

Furthermore, no cross-sectional association was observed between garlic intake and H. pylori infection. The pooled result showed that compared to participants who never ate garlic, those who consumed garlic ≥ 5 times per week experienced a multivariable-adjusted OR of being H. pylori positive of 1.66 (95% CI = 0.89 to 3.09; P for trend = 0.11) (Table 3). The association between garlic intake and H. pylori infection remained unchanged when we excluded non-whites from the analysis.

Table 3.

Odds ratios and 95% confidence intervals for Helicobacter pylori infection according to garlic intake^*

	Frequency of garlic intake (1 clove or 4 shakes)				P for trend
	Never	0-<1 per week	1–4 per week	≥5 per week
NHS
No. of H. pylori infection	23	14	30	10
No. of participants	105	71	135	39
Age-adjusted	1.00	1.00 (0.47, 2.14)	1.11 (0.59, 2.10)	1.35 (0.56, 3.24)	0.47
Multivariable-adjusted	1.00	1.04 (0.47, 2.28)	1.22 (0.63, 2.35)	1.56 (0.62, 3.95)	0.31
HPFS
No. of H. pylori infection	35	12	14	18
No. of participants	119	38	62	44
Age-adjusted	1.00	1.03 (0.45, 2.37)	0.61 (0.29, 1.31)	1.58 (0.73, 3.43)	0.28
Multivariable-adjusted	1.00	1.17 (0.48, 2.86)	0.75 (0.34, 1.69)	1.75 (0.76, 4.01)	0.21
Pooled†
Age-adjusted	1.00	1.01 (0.58, 1.78)	0.87 (0.54, 1.42)	1.48 (0.83, 2.64)	0.20
Multivariable-adjusted	1.00	1.09 (0.61, 1.98)	1.01 (0.60, 1.68)	1.66 (0.89, 3.09)	0.11

^*Multivariable models were adjusted for age (women: <40, 40–44, 45–49, 50–54, 55–59, 60–64, ≥65; men: <50, 50–54, 55–59, 60–64, 65–69, 70–74, ≥75), body-mass index (<23, 23–24.9, 25–29.9, 30–34.9, ≥35 kg/m²), physical activity (MET-hours/week, quintiles), smoking status (never, past, current), alcohol consumption (women: 0, 0.1–4.9, 5.0–14.9, ≥15 g/d; men: 0, 0.1–4.9, 5.0–29.9, ≥30 g/d), and intakes of total calorie, red/processed meat, fruits, vegetables, and coffee (quintiles).

^†The results were combined using a fixed-effect model based on inverse variance weight. For all categories of garlic intake, P values for heterogeneity between women (NHS) and men (HPFS) were ≥ 0.24.

DISCUSSION

In these two large prospective cohorts, garlic intake was not significantly associated with H. pylori infection or gastric cancer risk. Among the hundreds of species in the Allium genus, garlic (Allium sativum) is particularly well-known for its beneficial effects and has been widely consumed, cultivated, and researched worldwide for centuries. To date, the evidence for inverse association between garlic intake and gastric cancer risk was based primarily on the results from case-control studies^{5, 7–10}. Compared to cohort studies, case-control studies are more prone to biases such as recall bias, selection bias, and reverse causation. Therefore, the results from this large prospective cohort study is critical in understanding the association between garlic consumption and gastric cancer risk.

Although no previous cohort studies have examined garlic intake and gastric cancer risk, there have been cohort studies examining allium vegetables as a group or examining garlic supplement use. Those cohort studies, including the European Prospective Investigation into Cancer and Nutrition study²³, Shanghai Women’s and Men’s Health studies²⁴, Swedish Mammography Cohort and Cohort of Swedish Men²⁵ and the Netherlands Cohort Study²⁶, did not support a statistically significant association between total allium intake and gastric cancer risk. Although a different Netherlands Cohort Study by Dorant et al. reported a strong inverse association between intake of onion, another popular allium vegetable, and gastric cancer risk, they found no association between the garlic supplements use and gastric cancer risk (comparing exclusively garlic supplements to no supplements, multivariable RR = 1.27, 95% CI = 0.61 to 2.64)¹¹.

H. pylori infection is a major determinant of gastric cancer^{1, 15}. Garlic has been emphasized historically for its antibacterial effect, and in particular, has been anticipated to inhibit H. pylori infection in experimental settings¹². Therefore, H. pylori infection could act as a mediator between garlic intake and gastric cancer risk. We performed a secondary analysis to examine whether there is an association between garlic intake and H. pylori infection in our cohorts, but found no association. While one epidemiological study in China suggested an inverse association between garlic consumption and H. pylori infection¹⁴, our result was consistent with other studies^27–30 that consistently reported no association between garlic or garlic supplements intake and reduction in H. pylori infection.

The strengths of our study include prospective design, up to 30 years of follow-up with high follow-up rate, and repeated measurement of diet and lifestyle factors. We also had relatively wide variation in garlic intake and detailed information on various risk factors of gastric cancer. Our study also has limitations. Despite the large number of participants in our cohorts, we had limited number of gastric cancer cases. We were able to reduce measurement error by averaging garlic intake cumulatively from multiple time-points, but some measurement error is inevitable. Restriction of the study population to health professionals could reduce the generalizability of the results, but it could also reduce residual confounding by socioeconomic status. In addition, metabolic processes are unlikely to differ between health professionals and the general population.

In our cohorts, we could not examine whether the risk of gastric cancer is varied with the differences in preparation and storage of garlic. At present, we have limited knowledge on the stability of the organosulfur compounds during cooking⁷. An experimental study suggested that cooked garlic no longer has beneficial effects, at least in carcinogenesis³¹. A review article summarized that cooked or processed garlic products have different types of organosulfur compounds, some of which being highly unstable³². The conflicting findings on the anticancer properties of garlic could potentially be explained by differences in its processing and storage, but should further be clarified.

Gastric cancer can be divided into two groups by anatomical sites: cardia (proximal) and noncardia (distal) cancers. Growing evidence suggests that there are differences in risk factors between these subtypes of gastric cancer^{33, 34}. The recent meta-analysis reported that the inverse association of allium vegetables with gastric cancer risk was stronger for distal gastric cancer than proximal cancer¹⁰. In addition, H. pylori infection was found to be a risk factor for distal gastric cancer, but protective against proximal cancer³⁵. However, in our cohorts, we were not able to observe different results between proximal and distal gastric cancer subtypes.

In conclusion, we found no evidence that high intake of garlic reduces gastric cancer risk. Further population studies examining cooked versus uncooked garlic as well as different types of gastric cancer are warranted.

Novelty and impact.

While many previous case-control studies suggested that garlic consumption decreases gastric cancer risk, evidence from prospective cohort studies remains inconsistent. According to the two large prospective US cohort studies, the Nurses’ Health Study and the Health Professionals Follow-up Study, we could not say that high garlic intake reduces gastric cancer risk.