ABSTRACT Mushroomsarerichinbioactivecompounds.Thepotentialhealthbenefitsassociatedwithmushroomintakehavegainedrecentresearchattention. We thus conducted a systematic review and meta-analysis to assess the association between mushroom intake and risk of cancer at any site. We searched MEDLINE, Web of Science, and Cochrane Library to identify relevant studies on mushroom intake and cancer published from 1 January, 1966,upto31October,2020.Observationalstudies( n =17)withRRs,HRs,orORsand95%CIsofcancerriskfor ≥2categoriesofmushroomintake were eligible for the present study. Random-effects meta-analyses were conducted. Higher mushroom consumption was associated with lower riskoftotalcancer(pooledRRforthehighestcomparedwiththelowestconsumptiongroups:0.66;95%CI:0.55,0.78; n =17).Highermushroom consumptionwasalsoassociatedwithlowerriskofbreastcancer(pooledRRforthehighestcomparedwiththelowestconsumptiongroups:0.65; 95% CI: 0.52, 0.81;n = 10) and nonbreast cancer (pooled RR for the highest compared with the lowest consumption groups: 0.80; 95% CI: 0.66, 0.97; n = 13). When site-specific cancers were examined, a significant association with mushroom consumption was only observed with breast cancer; this could be due to the small number of studies that were conducted with other cancers. There was evidence of a significant nonlinear dose–responseassociationbetweenmushroomconsumptionandtheriskoftotalcancer( P-nonlinearity =0.001; n =7).Limitationsincludedthe potentialforrecallandselectionbiasincase-controldesigns,whichcomprised11outofthe17studiesincludedinthismeta-analysis,andthelarg e variationintheadjustmentfactorsusedinthefinalmodelsfromeachstudy.Theassociationbetweenhighermushroomconsumptionandlower riskofcancer,particularlybreastcancer,mayindicateapotentialprotectiveroleformushroomsinthediet. Adv Nutr2021;12:1691–1704. Keywords: mushroom,cancerrisk,diet,epidemiology,dose-response,observationalstudies Introduction Cancer constitutes a major threat to public health in both high- and low-income countries. Globally, cancer is consid- ered the second leading cause of death after cardiovascular diseases with an estimated 9.6 million deaths according to GLOBOCAN in 2018 (1). Modifiable risk factors such as a healthy diet are considered to play a significant role in the prevention of cancer (2). Mushrooms have been consumedasafunctionalfoodbymanyculturesforcenturies because of their unique taste, subtle flavor, and role in a Theauthorsreportednofundingreceivedforthisstudy. Authordisclosures:Theauthorsreportnoconflictsofinterest. XGisanEditorialBoardmemberfor Advances in NutritionbutplayednoroleintheJournal’s evaluationofthismanuscript.
SupplementalFigures1–4areavailablefromthe“Supplementarydata”linkintheonline postingofthearticleandfromthesamelinkintheonlinetableofcontentsat AddresscorrespondencetoJPR(e-mail: jrichie@pennstatehealth.psu.edu). healthful diet, being low in calories, carbohydrates, sodium, and fats and cholesterol-free (3–6). Edible mushrooms are alsoric hinb ioacti v eco m po unds,inc l udin gp h yt oc hemicals (alkaloids, phenolic acids, flavonoids, carotenoids) (7, 8), fiber, polysaccharides (9), selenium (10, 11), vitamins (e.g., niacin, thiamin, riboflavin, ascorbic acid, and vitamins B and D) (12–14), and the crucial antioxidants ergothioneine and glutathione which may play a significant role in the prevention of cancer (15–19). Many of the protective effects of mushrooms are thought to be mediated through their antioxidant properties with the unique antioxidant, ergoth- ioneine, thought to be playing an important role (6, 7, 20, 21). Ergothioneine concentrations vary by mushroom type with shiitake, oyster, maitake, and king oyster mushrooms which are widely consumed in Eastern Asian countries havinghigherconcentrationsthanthewhitebutton,crimini, and portabellas mushrooms which are broadly distributed C⃝ TheAuthor(s)2021.PublishedbyOxfordUniversityPressonbehalfoftheAmericanSocietyforNutrition.Allrightsreserved.Forpermissions,plea see-mail: journals.permissions@oup.com. Adv Nutr2021;12:1691–1704; doi:https://doi.org/10.1093/advances/nmab015. 1691 a n de a t e ni nt h eU n i t e dS t a t e s(17, 22). The growing body of evidence from various research groups across the globe regardingthepotentialhealthbenefitsassociatedwithedible mushroom consumption, including reductions in the risk of chronic diseases including cancer, has gained increasing attentionduringthelastfewdecades( 3,5,7). Even though mushrooms are often considered to be a vegetable, they actually belong to the fungal kingdom (22, 23). There are>2000 species of mushrooms in nature,≥25 of which are widely accepted as functional foods for human consumption and commercially cultivated (5, 23). Overall, mushrooms have been reported to have anticancer capabil- ities and protective effects against tumor development (7) and laboratory studies have revealed these anticarcinogenic effects vary according to different types of mushrooms such as shiitake, maitake, andAgaricus bisporus (button mush- room) (8, 24, 25). Several epidemiological observational studies have also reported an inverse association between mushroom consumption and cancer risk (26–32). However, severalotherepidemiologicalstudiesthathaveexaminedthe effectsofmushroomintakeontheriskofcancerhaveyielded nonsignificant associations (33–36). The previous meta- analysis that examined the association between mushroom consumption and cancer risk was limited to only breast cancer and included only a small number of articles (n = 7) (37). Given the inconsistent findings in the literature and lack of comprehensive studies including multiple cancers, we conducted a systematic review and meta-analysis of observational studies to examine the association between mushroom consumption and risk of cancer at any site. We hypothesized that higher mushroom consumption is associatedwithlowerriskofcancer.
Methods Searchstrategy We followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines and the guidelines established for reporting nonrandomized studies i nt h eC o c h r a n eL i b r a r y(38, 39)t os e l e c tp u b l i c a t i o n st o be included in this meta-analysis and to extract data. The PRISMA approach follows a few key steps after conceptu- alization of the research question including identifying and selecting relevant studies, charting data, presenting summa- rizedresults,discussingtheresultsandtheirlimitations,and revealingfundingsources( 38).Weperformedacomprehen- sivesystematicliteraturesearchofthePubMed(MEDLINE), Web of Science, and Cochrane Library databases to identify relevant observational studies on the association between mushroom consumption and the risk of cancer published from 1 January, 1966, up to 31 October, 2020. The following keywords were used: ((((mushroom OR Mushrooms)) OR “Agaricales”[Mesh])) AND ((((Prevention) OR (risk OR risks))) AND ((“Neoplasms”[Mesh]) OR (cancer OR can- cers))). The reference lists of past systematic reviews (37, 40) and selected publications were manually searched and scrutinized to identify additional pertinent studies. Only articleswrittenintheEnglishlanguagewereincludedinthis meta-analysis. Figure 1depictsthesearchprocess. Eligibilitycriteria Studies were eligible for this systematic review if they met the following criteria: 1)t h e yu s e da no b s e r v a t i o n a l study design (case-control or cohort study design);2)t h e exposureofinterestwasdietaryintakeofediblemushrooms; 3)t h eo u t c o m ew a st h eo c c u r r e n c eo fa n yc a n c e r ;4)t h e authors reported associations in the form of relative risks (RRs), hazard ratios (HRs), or odds ratios (ORs) with 95% confidence intervals (CIs) for≥2c a t e g o r i e so fm u s h r o o m s intake. We identified 19 publications consisting of 8 cohort and 11 case-control studies: 17 publications (20, 26–31, 33– 36,41–46)wereidentifiedbysearchingPubMed(MEDLINE) and 2 publications were identified by the manual search of previous systematic reviews (47, 48). Two publications (44, 45) were excluded from the meta-analysis because mush- roomintakehadbeentreatedasacontinuousvariable,notas discrete categories. The remaining 17 publications (20, 26– 31,33–36,41–43,46–48)wereincludedinthemeta-analysis (Tabl e 1). Cross-sectional studies were excluded from the currentmeta-analysis.
For the present meta-analysis, the primary outcome of interest was the risk of total cancer, and the secondary outcomeofinterestwastheriskofsite-specificcancer. Dataextraction Two independent authors (DMB and PS) initially screened the titles and abstracts of all selected publications eligible for inclusion in the analysis. Disagreements were resolved by discussion to meet a consensus. If necessary, an available third author was consulted in order to reach a consensus. Only relevant observational studies were included in the current meta-analysis. The following data were extracted from each publication: the first author’s name, the year of publication, sex, sample size, dietary assessment, outcome assessment, country in which the study was conducted, study design, cancer type, mean age of study participants, number of cases, categories of mushroom consumption, reported HRs, RRs, or ORs with corresponding 95% CIs, duration of follow-up for cohort studies, and the covariates foradjustmentsinthefinalmultivariableregressionmodels. Meta-poweranalysis The R software dmetar package was used to calculate the statistical power. With an expected average sample size of 1 2i n d i v i d u a l s / g r o u pi nt h es t u d i e si n c l u d e di nt h em e t a - analysis,10astheexpectednumberofstudies,asignificance level of 0.05, and high heterogeneity under a random-effect model, we had a 90% statistical power to detect an overall pooledeffectsizeofRR =0.66.ForanEasterncomparedwith Western region subgroup analysis, we had a 90% statistical power to detect a minimum effect size difference of 0.34. However, in the current study, the effect size difference was 0.47,whichishigherthantheminimumeffectsizedifference. 1692 Ba et al.
Records screened Full-text articles assessed for eligibility (n = 178)
Full-text articles excluded, with reasons (n = 161)
-Animal studies (n = 23)
-Meta-analysis/reviews (n = 78)
-Not relevant exposure/outcome (n = 58)
-Reported mushrooms intake as continuous variable (n = 2)
Articles included in quantitative synthesis (meta-analysis)
(n = 17)
Articles included in qualitative synthesis (n = 17)
Records identified through databases searching and manual search of reference lists: Pubmed, Web of Science, and Cochrane Library (n = 841)
Included Eligibility Screening noitacifitnedI Records excluded after title/abstracts screening (n = 417)
Excluded duplicate studies (n = 246)
(n = 595)
FIGURE1 PRISMAflowchartofthesystematicreviewandmeta-analysisofmushroomconsumptionandriskofcancer.PRISMA,Preferred ReportingItemsforSystematicReviewsandMeta-Analyses. Statisticalanalysis WeusedthereportedORs,RRs,orHRsasthemeasuresofthe associationbetweenmushroomintakeandtheriskofcancer. Accordingtoapreviousstudy,iftheoutcome(e.g.,cancer)is rareinallpopulationsandsubgroups,thedistinctionsamong different measures of RRs (e.g., ORs, rate ratios, and risk ratios)canbeignored( 49);therefore,wecombinedRRsand HRswithORsinthepresentmeta-analysisandreportedthe pooled effect size as RRs as common risk estimates for all studies.Twopublications( 20,34)reportedseparateORsand HRs for a different type of cancer. In these circumstances, we extracted the ORs and HRs for each type of cancer and u s e dr a n d o m - e ff e c t sm o d e l st op o o lt h eR R sw i t h i ne a c h study.Inaddition,1study( 34)reportedassociationsbetween mushroom intake and multiple site-specific cancers such as esophageal and endometrial. In this situation, we only selected cancers that were available in other studies because ≥2 studies are needed for meta-analysis (50). We used the reported RRs for the remaining studies which had only 1typeofcancertoassesstheassociationbetweenmushroom i n t a k ea n dt h er i s ko fc a n c e r .W efi r s tl o gt r a n s f o r m e d all the reported effect sizes of the data to normalize the distributions.SEswerecalculatedbythefollowingequations as described previously ( 51): lower = log (lower 95% CI) and upper = log (upper 95% CI), SE = (upper − lower)/3.92.
To assess the associations between mushroom intake and theriskofcancer,wepooledtheRRdataforthehighestcom- pared with the lowest mushroom intake category from each study, weighted by the inverse of their variances. We used the metagen function from the R package meta to calculate thepooledeffectestimatesusingrandom-effectsmodels( 52). Random-effects models were pooled using DerSimonian and Laird’s method for the association between mushroom Mushroom consumption and risk of cancer 1693 (reference) Sex Sample size, n Dietary assessment Outcomeassessment Country Studydesign Cancersite Meanage, y Total cases, n Mushroom consumption (quantity)
Reportedeffectsizes: HR/RR/OR(95%CI)
Follow-up, y Covariatesinthefully adjustedmodel Leeetal.( 34) M/F Total:112,991; FFQ Self-reportdiagnosisof cancer,verifiedby medicalrecordsto confirmthecancer diagnosis US Cohortstudy Prostate cancer Breast cancer Ovarian cancer Stomach cancer Liver cancer Colorectal cancer 52.9 9561 Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Neveroralmost never <1time/wk 1time/wk 2–4times/wk ≥5times/wk Reference Reference Reference Reference Reference Reference 26 Age,race(whiteornonwhite), height(continuous),BMI (quintiles),familyhistoryof cancer(yesorno),physical examinpast2y(yesorno), historyofcolonoscopyor sigmoidoscopy(yesorno), smokinginpack-years (neversmoker,1–4.9, physicalactivity(quintiles), regularaspirinuse(yesor no),multivitaminuse(yesor no),totalenergyintake (quintiles),alcohol consumption(0,0.1–4.9, ≥30g/d),redandprocessed meatintake(quintiles), prudentdietpattern (quintiles),andWesterndiet pattern(quintiles); prostate-specificantigen testinpast2y(yesorno)for menonly;andmenopausal status(premenopausalor postmenopausal), postmenopausalhormone use(never,past,orcurrent), andmammograminpast 2y(yesorno)forwomen only vanGilsetal.( 35) F Total:285,526 FFQ Cancerregistriesusing ICD-O-2;C50 10European countries Cohortstudy Breastcancer 50.9 3505 ≤0.3g/d >0.3to ≤1.8g/d >1.8to ≤5.2g/d >5.2to ≤11.1g/d >11.1g/d Reference 5.4 Energyintake,alcoholintake, saturatedfat,height, weight,ageatmenarche, parity,oralcontraceptives, HRT,menopausalstatus, smokingstatus,physical activity,education 1694 Ba et al.
(reference) Sex Sample size, n Dietary assessment Outcomeassessment Country Studydesign Cancersite Meanage, y Total cases, n Mushroom consumption (quantity)
Reportedeffectsizes: HR/RR/OR(95%CI)
Follow-up, y Covariatesinthefully adjustedmodel Masalaetal.( 33) F Total:31,510 FFQ Cancerregistriesusing ICD-O-2;C50 Italy Cohortstudy Breastcancer 50.2 1072 <0.4g/d 0.4–0.9g/d 1.0–1.9g/d 2.0–4.0g/d >4.0g/d Reference 11.25 Weight,height,education, numberofchildren,ageat menarche,menopausal status,energyintakeexcept alcohol,alcoholintake, currentuseofhormone therapy,smokingstatus, physicalactivity Leeetal.( 41) F Total:1000 FFQ Medicalrecordsand laboratory pathologyreports, confirmed histopathologically China Case-control Ovariancancer 59.0 500 ≤14g/d >14g/d Reference Not applicable Ageatinterview,ageof menarche,BMI,physical activity,totalenergyintake (kcal/d),parity,oral contraceptiveuse, menopausalstatus, educationlevel (none/primary,secondary, vocational/tertiary), smokingstatus,alcohol drinking,andfamilyhistory ofovarianorbreastcancer Zhangetal.( 42) M/F Total:132,837; FFQ Medicalrecordsand/or histologicalslidesof cancer China Cohortstudy Livercancer 53.7 267 ≤2.82g/d ≤5.88g/d ≤10.83g/d >10.83g/d Reference 10.9 Ageatenrollments,BMI,total energyintake,family income,educationlevel, familyhistoryoflivercancer, historyofchronicviral hepatitis,chronicliver disease,orcirrhosis, diabetes,cholelithiasisor cholecystectomy,vitaminC, E,ormultivitamin supplementation Zhangetal.( 28) F Total:2018 FFQ Histopathologically confirmeddiagnosis China Case-control Breastcancer 48.4 1009 0g/d <2g/d 2to <10g/d ≥10g/d Reference Not applicable Age,residentialarea, education,BMI,ageat menarche,oral contraceptiveuse,HRT, breastcancerinfirst-degree relatives,totalenergyintake, menopausalstatus,alcohol consumption,active smoking,passivesmoking, teadrinking,andphysical activity Zhangetal.( 27) F Total:876 FFQ Breastcancer diagnosedand histologically confirmed China Case-control Breastcancer 47.1 438 <0.8g/d 0.8–2.5g/d 2.5–7.1g/d >7.1g/d Reference Not applicable Ageatmenarche,BMI,history ofbenignbreastdisease, family,physicalactivity, passivesmoking,andtotal energyintake Mushroom consumption and risk of cancer 1695 (reference) Sex Sample size, n Dietary assessment Outcomeassessment Country Studydesign Cancersite Meanage, y Total cases, n Mushroom consumption (quantity)
Reportedeffectsizes: HR/RR/OR(95%CI)
Follow-up, y Covariatesinthefully adjustedmodel Zhangetal.( 46) M Total:36,499 FFQ ICD-O-3,codedasC61 Japan Cohortstudy Prostatecancer 55.7 1204 <1time/wk 1–2times/wk ≥3times/wk Reference 13.2 Familyhistoryofcancer,BMI, educationlevel,smoking status,alcoholdrinking, timespentwalking,5-group meatconsumption, vegetables,fruit,dairy products,coffee,and energyintakes Mizooetal.2013 F Total:936 Self-administered questionnaire JapaneseSingle Nucleotide Polymorphism database Japan Case-control Breastcancer 54.1 472 ≤1time/wk 2–4times/wk ≥5times/wk Reference Not applicable Age Haraetal.( 20) M/F Total:781;M/F numbersnot specified FFQ Histopathological gradingand anatomicalsubsites ofstomachcancers andcolorectal cancers Japan Case-control Stomach cancer Colorectal cancer 58.7 264 Tertile1(3g/1000kcal)
Tertile2(11g/1000kcal)
Tertile3(28g/1000kcal)
Tertile1(4g/1000kcal)
Tertile2(11g/1000kcal)
Tertile3(29g/1000kcal)
Reference Reference Not applicable Forstomachcancerset, adjustedforsmokingstatus (3categories:never,ex, current),familyhistoryof stomachcancer,saltintake (3categories),totalenergy intake(rankvariable,0–2), andJACooperatives membership;forcolorectal cancerset,adjustedfor smokingstatus(3 categories:never,ex, current),alcoholintake(3 categories),familyhistoryof colorectalcancer,total energyintake(rankvariable, 0–2),andJAmembership Koetal.( 36) M/F Total:9724;M: Self-administered FFQ Identifiedthrough recordlinkagewith theCentralCancer Registry Korea Cohortstudy Stomach cancer 57.6 166 Almostnever 1–4times/mo 1–4times/wk ≥1time/d Reference 8.5 Age,sex,cigarettesmoking, BMI,alcoholdrinking,and areaofresidence Shinetal.( 29) F Total:718 FFQ Paraffin-embedded breasttumorsby immunohistochem- istry Korea Case-control Breastcancer 48.1 358 <2.61g/d 2.62to <5.36g/d 5.36to <11.37g/d ≥11.37g/d Reference Not applicable Age,BMI,familyhistoryof breastcancer,current dietarysupplements, education,job,smoking, alcoholintake,physical activity,menopausalstatus, ageatmenarche,parity, totalenergyintake,and vegetableintake Hongetal.( 26) F Total:724 FFQ Mammographyand histologically confirmed Korea Case-control Breastcancer 46.1 362 0g/d 2.45g/d 4.90g/d 9.80g/d 18.3g/d Reference Not applicable Education(y),familyhistoryfor breastcancer,regular exercise( ≥22.5MET-h/wk), BMI,currentsmoker,current drinker,current multivitaminsupplement, 1696 Ba et al.
(reference) Sex Sample size, n Dietary assessment Outcomeassessment Country Studydesign Cancersite Meanage, y Total cases, n Mushroom consumption (quantity)
Reportedeffectsizes: HR/RR/OR(95%CI)
Follow-up, y Covariatesinthefully adjustedmodel numberofchildren, menopausalstatus,energy intake,carbohydrate,soy protein,vitaminE,andfolate Leeetal.( 48) F Total:262 Notspecified Notspecified Korea Case-control Breastcancer Notspecified 103 ≤1time/wk 2–3times/wk 1time/d Reference Not applicable BMI,residence,occupation, familyhistory,delivery miscarriage,breastfeeding, periodsofbreastfeeding, HRT Leeetal.( 30) F Total:378 ShortFFQ method Firstdiagnosisof histopathologically confirmed Korea Case-control Breastcancer 63.0 189 <1time/wk ≥1time/wk Reference Not applicable Age,education,BMI,and familyhistoryofbreast cancer Kimetal.( 43) M/F Total:272; Quantitative foodfrequency method Endoscopic examination confirmedby histologicmethod Korea Case-control Stomach cancer Notspecified 136 Low( <25thpercentile), Medium(25th–75th percentile), High( >75thpercentile)
Reference Not applicable Age,sex,socioeconomic status,familyhistory,and refrigeratoruse Park et al. (47) M/F Total:360; FFQ Confirmedbythe histological diagnosis Korea Case-control Stomach cancer 51.5 126 ≤4–6times/y ≥1time/mo Reference Not applicable Age,sex,education,economic status,andresidence 1HRT,hormonereplacementtherapy;ICD-O,InternationalClassificationofDiseasesforOncology;JA,JapanAgricultural;MET,metabolicequivalen ts. intake and the risk of all cancer. The results derived were graphically presented in forest plots. Potential between- studies heterogeneity was quantified using the Cochran’sQ test and I 2 statistics expressed as a proportion (%) in the estimationoftherandom-effectsmodels( 53).Heterogeneity was determined with a significance ofP < 0.05. Univariate meta-regression was used to examine the association be- tween study designs, location of study (Western compared with Eastern), and breast cancer compared with nonbreast cancer. We selected breast cancer because it appeared to be the most affected site as significant associations with mushroomintakewereonlyobservedforcancersatthissite. Location was examined because mushroom intake is much higher in Eastern countries such as China than in Western countriessuchastheUnitedStates( 17,22). As a secondary analysis, to better understand the shape of the curve relating mushroom intake to cancer risk, we conducted a dose-response meta-analysis using the R software dosresmetapackage.Studiesincludedinthecurrent meta-analysisuseddifferentunitstoreportmushroomintake (e.g.,grams,servings,andfrequencies).Tofacilitatethedose- response meta-analysis, we included studies (n = 7) that reported ≥3 categories of mushroom intake in grams per day. In publications where the median intake values per mushroom intake category were not reported, we estimated the average intake in each category by calculating the midpoint of the upper and lower boundaries. When the upper boundary of the highest category was not provided, w ea s s u m e dt h a ti th a dt h es a m ea m p l i t u d eo fi n t a k ea s the closet adjacent category (54, 55). We then examined the shapeoftherelationbetween mushroomintakeandtherisk ofcancerwitha2-stagerandom-effectsdose-responsemeta- analysis, using restricted cubic splines with 3 knots at fixed percentiles (25%, 50%, and 75%) of mushroom distribution as done in a previous study (37). It is worth noting that the2-stagedose-responsemeta-analysismodelrequiresdata for ≥3exposurecategories,includingthereferencecategory, withineachstudy.
A P value for the nonlinearity of the dose-response meta-analysis was tested using the Wald test. An influence sensitivity analysis was performed by removing 1 individual studyatatime,toexaminetheeffectoftheexcludedstudyon thepooledeffectestima tes.
Publication bias was investigated through the use of funnel plot asymmetry and tested by Egger’s asymmetry test (56)a n dB e g g ’ st e s t(57). All analyses were conducted using R statistical software version 3.4.3 (R Foundation for StatisticalComputing,Vienna,Austria).A Pvalue<0.05was consideredstatisticallysignificant.
Results Literaturesearchandcharacteristicsofstudies The first search from 1 January, 1966, up to 31 October, 2020, produced 450 studies from PubMed, 305 from Web of Science, 84 from the Cochrane Library databases, and 2 from a manual search of reference lists of the previous Mushroom consumption and risk of cancer 1697 FIGURE2 Summaryforestplotofmushroomconsumption(highestcomparedwithlowestcategory)andcancerrisk.Thesquare representsthepointestimateofeachstudyandthesizeisproportionaltoitsweightinthemeta-analysis.Thehorizontallinethroughthe squarerepresentsits95%CI.Thediamondrepresentsthepooledriskratioofthemeta-analysis. meta-analysis. After removing 246 duplicate records, a total of 595 potential related publications were screened for inclusion in the meta-analysis. Four hundred and seventeen records screened were excluded after review of titles and abstracts. One hundred and sixty-one full texts were further excluded because of the following reasons: animal studies (n = 23), meta-analysis/reviews (n = 78), not relevant exposure/outcome (n = 58), and mushroom intake treated as a continuous variable (n = 2). Finally, 17 studies (20, 26–31, 33–36, 41–43, 46–48)( Figure 1)w e r ee l i g i b l ef o r this meta-analysis consisting of 6 cohort studies (33–36, 42, 46)a n d1 1c a s e - c o n t r o ls t u d i e s(20, 26–31, 41, 43, 47, 48). The total number of cancer cases for this meta- analysis was 19,732. The majority of studies included in this meta-analysis were conducted in Asian countries (n = 14, referred to as “Eastern countries” in the article), whereas 2studieswereconductedinEuropeand1intheUnitedStates (referred to as “Western countries” in the article).Table 1 shows characteristics of the studies included in the meta- analysis.
Mushroomconsumptionandriskofcancer Higher mushroom consumption was associated with lower risk of total cancer (pooled RR for the highest compared with the lowest consumption groups: 0.66; 95% CI: 0.55, 0.78; n =17)( Figure 2).Therewassubstantialheterogeneity between studies (I 2 = 77%; P-heterogeneity < 0.01). Higher mushroom consumption was associated with lower risk of total cancer in cohort studies (pooled RR for the highest comparedwiththelowestconsumptiongroups:0.89;95%CI: 0.82,0.97; n =6)andcase-controlstudies(pooledRRforthe highestcomparedwiththelowestconsumptiongroups:0.52; 95%CI:0.41,0.66; n =11)( Supplemental Figure 1).Higher mushroomconsumptionwas alsoassociated withlower risk of breast cancer (pooled RR for the highest compared with the lowest consumption groups: 0.65; 95% CI: 0.52, 0.81; n = 10) and nonbreast cancer (pooled RR for the highest comparedwiththelowestconsumptiongroups:0.80;95%CI: 0.66,0.97; n =13)( Figure 3).
When site-specific cancer was examined, a significant associationwithmushroomconsumptionwasonlyobserved with breast cancer (Figure 4). The lack of association with other cancers could be due to the small number of studies whichexaminedassociationsofmushroomintakewithother site-specific cancers (<6 studies for each site-specific cancer comparedwith10studiesforbreastcancer). Subgroupanalyses,sensitivityanalyses,and publicationbias We performed univariate meta-regression analysis to in- vestigate the potential sources of heterogeneity. Subgroup analysis (Tabl e 2) showed that the significant association b e t w e e nh i g h e rm u s h r o o mi n t a k ea n dt o t a lc a n c e rr i s kw a s observedonlyinstudiesfromEasternregions(pooledRRfor the highest compared with the lowest consumption groups: 0.58; 95% CI: 0.47, 0.71;P = 0.02; n = 14) (Supplemental Figure 2).Whenlimitingtobreastcancerstudiesonly,higher mushroom intake was associated with lower risk of breast canceronlyincase-controlstudies(pooledRRforthehighest compared with the lowest consumption groups: 0.50; 95% 1698 Ba et al.
FIGURE3 Forestplotofmushroomconsumption(highestcomparedwithlowestcategory)andbreastcancerrisk.Thesquare representsthepointestimateofeachstudyandthesizeisproportionaltoitsweightinthemeta-analysis.Thehorizontallinethroughthe squarerepresentsits95%CI.Thediamondrepresentsthepooledriskratioofthemeta-analysis. CI: 0.40, 0.62) but not in cohort studies (Supplemental Figure 3).
In the sensitivity analysis, omitting each study at a time did not indicate any substantial changes of the pooled RRs fromtherandom-effectsmodel.ThepooledRRsrangedfrom 0.63 to 0.70 (P < 0.0001 for all) (Supplemental Figure 4). Inspection of the funnel plot (Figure 5), Egger’s test for asymmetry (P = 0.08), and Begg’s test (P = 0.16) did not indica tethep r esenceo fpublica tio nb ias. Nonlineardose-responsemeta-analysis Seven studies were included in the restricted cubic spline analysis to fit a 2-stage random-effects dose-response meta- analysis.Therewasevidenceofasignificantnonlineardose– response association between mushroom consumption and theriskoftotalcancer( P-nonlinearity =0.001; n =7),with a45%lowerriskathigherintakeof18g/d(RR:0.55;95%CI: 0.36,0.86)thanatintakeof0g/d( Figure 6). Discussion In this systematic review and meta-analysis of observational studies, we found that higher mushroom consumption was associatedwithlowerriskofcancer.Inparticular,breastcan- cerappearedtobethemostaffectedsitebecauseasignificant association with mushroom intake was only observed for cancersatthissite.Importantly,mushroomconsumptionwas associatedwithlowerriskofcancerinbothcohortandcase- controlstudies.Theeffectwasmuchstrongerincase-control studiesthanincohortstudies.Ourmeta-analysisisgenerally consistent with results from a previous meta-analysis of observational studies that indicated an inverse association between mushroom intake and the risk of breast cancer (37). Evidence from epidemiological studies has shown that Mushroom consumption and risk of cancer 1699 FIGURE4 Forestplotofmushroomconsumption(highestcomparedwithlowestcategory)andriskofsite-specificcancers.Thesquare representsthepointestimateofeachstudyandthesizeisproportionaltoitsweightinthemeta-analysis.Thehorizontallinethroughthe squarerepresentsits95%CI.Thediamondrepresentsthepooledriskratioofthemeta-analysis. 1700 Ba et al.
intake and cancer risk Characteristic PooledRR(95%CI) I2 (%)
Pforheterogeneity betweensubgroups Cancertype 0.
Breast 0.65(0.52,0.81) 81 Prostate 0.86(0.72,1.03) 7 Ovarian 0.79(0.62,1.00) 0 Stomach 0.61(0.33,1.11) 73 Colorectal 0.93(0.64,1.34) 19 Studydesign 0.
Cohort 0.89(0.82,0.97) 13 Case-control 0.52(0.41,0.66) 59 Region 0.
Western 0.93(0.85,1.02) 0 Eastern 0.58(0.47,0.71) 68 regular consumption of fruits and vegetables is associated with reduced risk of chronic diseases such as cancer and all-cause mortality (58–61). A dose-response meta-analysis indicated that higher mushroom consumption of 18 g/d was associatedwitha45%lowerriskoftotalcancerthananintake of 0 g/d.
The potential biological mechanisms underlying the associationbetweenmushroomconsumptionandlowerrisk of cancer may stem from their antioxidant properties due to the specific mushroom components ergothioneine and glutathione. Oxidative damage due to an excess of free radicals is an important causal factor in the aging process anddiseasesofagingincludingmanycancers,withamounts being related to poor lifestyle habits such as an unhealthy diet (62). Reactive oxygen and nitrogen species (RONS) are producedbysomeendogenousandexogenousprocesses,and theirnegativeeffectsareneutralizedbyantioxidantdefenses. Oxidative stress occurs from the imbalance between RONS FIGURE5 Funnelplotofmushroomconsumption(highest comparedwithlowestcategory)andcancerriskfromthe 17studiesincludedinthemeta-analysis.Nosignificantpublication biaswasdetected( P =0.08forEgger’stest, P =0.16forBegg’stest). production and these antioxidant defenses and has been associated with several chronic diseases such as cancer that account for a vital portion of death today (62, 63). Given t h es i g n i fi c a n tr o l eo fo x i d a t i v es t r e s si nt h ep a t h o g e n e s i so f many chronic diseases, antioxidants may play a significant role in the control and prevention of chronic diseases (62). Mushrooms are a potent source of key antioxidants that can mitigate oxidative stress and improve human health and promote quality of life (16). In particular, ergothioneine, which is found in very high concentrations in mushrooms, isasulfur-containingaminoacidthathasstrongantioxidant activity and is obtained exclusively through dietary sources (4, 16, 19, 22, 64, 65). In a recent review, it was proposed that ergothioneine could be used as a therapeutic to reduce the severity of and mortality from coronavirus infectious disease2019(COVID-19)( 66).Inaddition,PaulandSnyder (19) indicated that ergothioneine is an important physio- logical cytoprotectant and should be designated as a new FIGURE6 Dose–responseassociationsofmushroom consumptionandcancerrisk.Thesolidlinesrepresentthe best-fittingcubicsplineandthedottedlinesrepresentthe95%CIs. Mushroom consumption and risk of cancer 1701 vitamin. Furthermore, Bruce Ames (67)h a ss u g g e s t e dt h a t ergothioneineisa“longevityvitamin”asdefinedbyhisTriage Theory, with multiple potential functions in the body (e.g., antioxidant, cytoprotective, and antiaging). A very recent study has also suggested that higher plasma ergothioneine was associated with reduced risk of cardiometabolic disease and mortality (68). Researchers from a previous study have also revealed that mushroom species such as oyster mushrooms contain ∼10 times more ergothioneine than other dietary sources such as chicken liver and black beans (64). Mushrooms also contain other bioactive compounds including polysaccharides such asβ-glucans that have been implicated as having antitumor and immunomodulation properties (4, 7, 69–71). Laboratory experiment studies have revealed that anticarcinogenic effects of mushrooms vary with different types of mushrooms such as shiitake, maitake, and A. bisporus (8, 24, 25). Furthermore, in vitro and in vivo studies have suggested that extracts of mushrooms such as A. bisporus (8)a n dAgaricus blazei Murill (72)weremorelikelytoinhibitthegrowthofprostate tumor cell lines in immune-deficient mice and to block prostate tumorigenic progression. Based on their findings, the authors suggested that mushrooms might be effective for the prevention and treatment of human prostate cancer (8,72).
When site-specific cancer was examined, a significant associationwithmushroomconsumptionwasonlyobserved withbreastcancer,notothercancers.Thelackofassociation with other cancers could be due to the small number of studies which examined associations of mushroom intake with other site-specific cancers (<6 studies for each site- s p e c i fi cc a n c e rc o m p a r e dw i t h1 0s t u d i e sf o rb r e a s tc a n c e r ) . I tc o u l db ea l s od u et ot h ef a c tt h a td i e t a r ya n t i o x i d a n t phytochemicalsfrommushroomsareinhibitorsofaromatase activity and possess antiestrogen properties which may suppressbreastcancerproliferation.
An interesting finding in the present meta-analysis is that studies conducted in Eastern regions were inversely associated with the risk of all cancer compared with those in Western regions. The differences in risk between Eastern and Western countries may result from differences in the amounts and types of mushrooms consumed between these regions. Total amounts of mushroom consumption tend to begreaterinEasterncountries( 26,29,41). Our meta-analysis has several strengths. To the best of ourknowledge,thisisthemostcomprehensivemeta-analysis to assess the association between mushroom intake and the risk of cancer at any site. The previous meta-analysis was limitedtoonlybreastcancerandincludedasmallnumberof articles. The majorityof studies in the present meta-analysis (12 out of 17) used validated dietary assessment methods suchasFFQs.Moreover,wehavetestedtherobustnessofour results by conducting several sensitivity analyses and testing for potential publication bias. Lastly, we conducted a dose- response meta-analysis to better understand the shape of the curve relating to mushroom intake and cancer using a restrictedcubicspline.
Notwithstanding, our study has several limitations that should be noted. First, combining studies from distinct p o p u l a t i o n si n c r e a s e st h es a m p l es i z ea n ds t a t i s t i c a lp o w e r ; however, it may also result in heterogeneity because of inequality in the characteristics of the study populations. Second, the majority of the studies (11 out of 17) included in the present analyses used a case-control design, which is subject to recall and selection bias. A third limitation is that publication bias is inevitable in any meta-analysis and our study was limited by the inclusion of only those studies that were published in English, thus, relevant non-English publishedstudiesmayhavebeenmissed.Afourthlimitation is that the adjustment factors used in the final models from eachstudywer enotthesame.
In conclusion, the current meta-analysis showed a sig- nificant inverse association between higher mushroom con- sumptionandlowerriskofcancer.Inparticular,breastcancer appeared to be the most affected site because a significant association with mushroom intake was only observed for c a n c e r sa tt h i ss i t e .Al a c ko fs i g n i fi c a n ta s s o c i a t i o nf o rs i t e - specificcancersinthisstudycouldbeduetothelownumbers of studies specifically for these cancers. Our findings may have important public health implications in the prevention of chronicdiseases andmortality.In addition,theresults are useful for policy makers, contributing to increasing public awareness about the role of the diet on health, and potential protective effects of mushrooms in lowering the risk of cancer. Lastly, findings from this study will also provide usefuldirectionsforfutureepidemiologicalstudiesaboutthe nutritionalbenefitsofmushroomsandhealth.Futureclinical studiesforsite-specificcancersarewarranted. Acknowledgments We thank Esther Dell at the Penn State College of Medicine Libraries for her systematic review consultation about the database and keyword searches. We also thank Dr. Vernon MChinchilliforprovidinghelpfulfeedback.Theauthors’re- sponsibilitieswereasfollows—DMB,XG,andJPR:designed theresearch(projectconception,developmentoftheoverall researchplan,andstudyoversight);DMBandPS:conducted the systematic review and performed the statistical analysis; DMB:analyzedthedata,managedthesystematicreviewand meta-analysis,andwrotethefirstdraftofthemanuscript;and allauthors:review,editing,andapprovedthefinalversionof themanuscript.