ARTICLE IN PRESS R EVIEW Diet and acne Whitney P. Bowe, MD,a Smita S. Joshi, MD,b and Alan R. Shalita, MDa Brooklyn, New York, and Chicago, Illinois Historically,therelationshipbetweendietandacnehasbeenhighlycontroversial.Beforethe1960s,certain foodswerethoughttoexacerbateacne.However,subsequentstudiesdispelledtheseallegedassociations asmythforalmosthalfacentury.Severalstudiesduringthelastdecadehaveprompteddermatologiststo revisit the potential link between diet and acne. This article critically reviews the literature and discusses howdermatologistsmightaddressdietwhencounselingpatientswithacne.Dermatologistscannolonger dismiss the association between diet and acne. Compelling evidence exists that high glycemic load diets mayexacerbateacne.Dairyingestionappearstobeweaklyassociatedwithacne,andtherolesofomega-3 fattyacids,antioxidants,zinc,vitaminA,anddietaryfiberremaintobeelucidated.Thisstudywaslimited bythelackofrandomizedcontrolledtrialsintheliterature.Wehopethatthisreviewwillencourageothers to explore the effects of diet on acne. (J Am Acad Dermatol 10.1016/j.jaad.2009.07.043.) Keywords:acne;dairy;diet;glycemicindex;glycemicload;nutrition;omega-3fattyacids;vitaminA; vitamins;zinc. H istorically, the relationship between diet Abbreviationsused: and acne has been highly controversial. In the 1930s through 1960s, patients were BMI: bodymassindex DHT: dihydrotestosterone often discouraged from eating a wide selection of GI: glycemicindex foods including chocolate, fats, and sweets. HGL: highglycemicload Following a few critical studies,1,2 these alleged IGF: insulin-likegrowthfactor IGFBP: insulin-likegrowthfactorbindingprotein associationsbetweenfoodsandacneweredispelled LGL: lowglycemicload as myth for almost half a century. Consequently, dietaryrestrictionshavenotbeenrecommendedasa standardpartofacnetherapyfordecades.Recently, paidtotheinfluenceofdairyproductsandcarbohy- however,therelationshipbetweendietandacnehas drateintakeonacneseverity.Althoughthemajority been called back into question. Several carefully ofevidencesupportingalinkbetweendietandacne designed and thought-provokingstudies duringthe falls under one of these two categories, a brief last decade have prompted dermatologists and nu- discussion of other dietary factors implicated in tritioniststorevisittheallegedlinkbetweendietand acne including omega-3 fatty acids, antioxidants, acne. In particular, evidence that the ingestion of zinc, vitamin A, and iodine follows. Very few, if certain dairy products, carbohydrates, or both may any, human data are available to support a role for exacerbate acne has been particularly compelling these factors in acne, but some in vitro and animal andmeritsexamination.Acomprehensivereviewof data exist. Prospective controlled trials, prospective theliteraturetodatefollows,withparticularattention and retrospective cohort studies, case-control stud- ies,andlargecaseseriesexaminingtheroleofdietin From the Department of Dermatology, State University of New acne, published in the English language and avail- YorkDownstateMedicalCenter,aandNorthwesternUniversity able on PubMed, were included in this review FeinbergSchoolofMedicine,Chicago.b (Table I). We have included several studies that we Supported in part by the Skin Cancer Foundation and Orlando believeareofinferiordesigninanefforttoprovide Dermatology and Aesthetic Conference’s Leonard L. Mazur historicalcontextformorerecentdevelopments,and ResidentResearchAward(YoungInvestigatorResearchGrant; DrBowe)andtheAmericanAcneandRosaceaSocietyClinical toaddressseveraldietaryfactorsthat,inouropinion, ResearchGrant(DrBowe). meritfurtherstudy. Conflictsofinterest:Nonedeclared. Reprint requests: Whitney P. Bowe, MD, 450 Clarkson Ave, Brooklyn,NY11203.E-mail:[email protected]. DIET AND ACNE: A HISTORICAL PublishedonlineMarch23,2010. FRAMEWORK 0190-9622/$36.00 Before the 1960s, dietary advice was a standard ª2009bytheAmericanAcademyofDermatology,Inc. doi:10.1016/j.jaad.2009.07.043 part of acne therapy. Data published in 1931 1 ARTICLE IN PRESS 2 Bowe,Joshi, and Shalita JAMACADDERMATOL suggestedthatimpairedglucosetoleranceexistedin In1971,Anderson2alsochallengedtheacneand patientswithacne.3Abnormalcarbohydratemetab- diet association by assigning 27 students with re- olism was implicated in acne and patients were ported history of dietary acne triggers to consume recommendedtoavoidconsumingexcessivecarbo- chocolate,milk,roastedpeanuts,orcolafor1week hydrates and high-sugar foods.4 Major textbooks of underdirectsupervision.Theauthordidnotspecify dermatology5-7popular inthe1940sand 1950s per- how many of the 27 subjects were subdivided into petuatedthesebeliefsanddiscouragedfoodssuchas eachgroup.Beforeandafterthetrial,eachsubject’s chocolate, fats, sweets, and acne lesions were mapped carbonatedbeveragesaspart onto a sheet held over the of acne therapy. However, CAPSULE SUMMARY face. Anderson2 found that thecontentionthatarelation- Dermatologists can no longer dismiss the foods produced no d ship existed between diet the association between diet and acne. flares of acne. However, andacnewasabandonedaf- lesion counts were not Thereis reasonablycompelling evidence d terthefollowingtwostudies reported and statistical anal- that high glycemic load diets may showednoassociation. ysis was not performed.15 exacerbate acne. Furthermore, it is unlikely An association between dairy ingestion EARLIER STUDIES d that any group had enough and acne may exist, but evidence is THAT SHOWED NO statistical power to provide weak. ASSOCIATION significant results, given that BETWEEN DIET AND d The role of omega-3 fatty acids, the 27 patients were subdi- ACNE antioxidants, zinc, vitamin A, and dietary vided into 4 treatment cate- In 1969, Fulton et al1 in- fiber in acne is unclear. gories. This study suffered vestigatedtheeffectofchoc- from many limitations in- olate on acne vulgaris in a cluding small sample size crossoversingle-blindedstudy.Atotalof65subjects and lack of controls, blinding, and were assigned to eat either a chocolate bar or a randomization.11,12 control bar with similar appearance and caloric Finally,thestudiesbyFultonetal1andAnderson2 content. Subjects ate the assigned bar daily for 4 havebothbeencritiquedfortheirfailuretotakeinto weeksandthenaftera3-weekrestperiod,consumed considerationsubjects’baselinediets,resultinginan the alternate bar for 4 weeks. Acne was scored as inability to determine if the treatment diet varied worsenedorimprovedifthetotalnumberoflesions from the subjects’ normal diet.12,13,15 This oversight increased or decreased by 30%, respectively. The mayhaveobscuredthestudyresults. authors found that acne severity did not change during the chocolate bar and control bar study MORE MODERN STUDIES REVISIT THE periodsandconcludedthatchocolatedidnotaffect LINK BETWEEN ACNE AND DIET: AN the course ofacne. However,this study was flawed EXAMINATION OF THE INFLUENCE OF forseveralreasons.Thedurationofthestudywasfar DAIRY PRODUCTS ON ACNE SEVERITY tooshortforthechosenoutcomebecausemostacne The studies by Fulton et al1 and Anderson,2 clinicaltrialslast12weeks8-10toallowforthenatural although suffering from major design flaws, were history of comedo formation and evolution. sufficienttodissociatedietfromacneinthemindsof Furthermore, the placebo bar was an inappropriate most dermatologists. Textbooks were revised to controlbecauseithadsimilartotalsugarandtotalfat reflectthisnewacademicconsensus,anddermatol- contentasthechocolatebar.11Moreover,thecontrol ogiststookthestancethatanymumblingsaboutthe barwaslargelycomposedofpartiallyhydrogenated associationbetweendietandacnewereunscientific vegetable oil rather than cocoa butter and cacao and one of the many myths surrounding this ubiq- paste,theformerofwhichcontainshighquantitiesof uitousdisease. trans-fattyacidsthatcontributetoinflammation.12,13 Two decades before the studies of Fulton et al1 Finally, most patients with dietary acne triggers and Anderson,2 Robinson16 took note of what he reportpustularflaresratherthancomedonalacne.14 believed to be an association between dairy intake Fultonetal1groupedallacnelesionstogether.Thus, and acne severity. He documented that, based on ifapatient’sacnelesionsshiftedfromacomedonalto 1925subjects whokept strictfooddiaries, milkwas pustularpredominancewhereasthetotalnumberof the most commonly implicated food in acne flares. lesions remained the same, this would have been However, Robinson16 failed to provide summary scoredasunchangedeventhoughaclinicallysignif- statisticsandstatisticalanalyses.Hisreportismerely icantchangewouldhavebeenpresent. anecdotalanddescriptive,basedonhisexperience. Table I.Studies investigating relationship between dietand acne J A M Levelof Conclusionsand A Study Design evidence Subjects Samplesize Intervention limitations C A Robinson,16 Caseseries 2 Subjects 1925 Dietdiary Milkwasmostcommonlyimplicated D D 1949 with acne foodin acneflares. Limitations E R include shortstudy durationof1-2 M A wk and absenceof dataand T O statistical analysis. L Grant and Caseseries 3 Subjects with acne 8 Milkchocolate barfor Chocolatedid notaggravate acne. Anderson,106 5 d;halfof subjects Antacid did notaffectresults. 1965 consumed antacid Limitationsincludesmallsamplesize, before chocolate short studyduration,lack ofcontrol, lack ofrandomization, lackof blinding,andlackofconsiderationof subjects’ baseline diets. Bett et al,107 Cross-sectional 3 Subjects with acne, 16Subjects with Questionnaire Subjects with acne didnotconsume A 1967 survey control subjects acne, 16control more sugarthan control subjects. R T with warts, and subjects withwarts, Limitationsincludesmallsamplesize I sex- andage-matched 16healthy control and lackof validated questionnaire. C healthycontrol subjects L E subjects Fultonet al,11969 Crossover, 1 Adolescents with 30Adolescents, Chocolate baror Severityofacnedidnotchangeduring IN subject-blind acne and adult 35prisoners control barwith chocolate barand controlbarstudy P interventional male prisoners similar caloric periods.Chocolate does notaffect R study with acne composition acne. Limitationsinclude small E sample size,shortstudy duration, S similar amountsofsugar andfat in S both bars,inappropriate grouping, andlackofconsiderationofsubjects’ baseline diets. Anderson,2 Caseseries 3 University students 27 Daily consumption Studyfoods did notproduce any acne 1971 whoreported ofchocolate, milk, flares. Limitationsinclude small B o dietary acne roastedpeanuts, or sample size,lackof control,lack of w triggers cola for1 wk randomization, lack ofblinding, e, J absence oflesioncounts, lack of o s statistical analysis,and failure to h i , consider subjects’ prestudy diets. a n Continued d S h a li ta 3 Table I. Cont’d 4 B Levelof Conclusionsand o Study Design evidence Subjects Samplesize Intervention limitations w e Kligmanet al,98 Case series 2 Part 1:subjects Part 1:136 Part 1: daily Retinol iseffective treatment foracne , J o 1981 with inflammatory Part 2:39 vitaminA300,000 atdosesof300,000Uforwomenand s h acne IU for3-4 mo 400,000-500,000 U formen. Toxicity i, a Part 2:menwith acne Part 2: eitherdaily isminimaland limited mainlyto n vitaminA300,000 xerosis andcheilitis. Limitations d S IU for12wk or includelack ofcontrol,lack of h a daily vitaminA randomization,lack ofblinding, and li 300,000 IU for1 wk lackofdata on higherdosesof ta followed by 400,000 vitaminAinwomen.Women were IU forsecond wk givenoralcontraceptivesthatmay and then500,000 IU haveconfoundedresults. fornext10 wk A Aizawaand Cross-sectional 3 Womenwith acne 30Subjects with OGTT Mildinsulin resistance occurs during R Niimura,45 andage-matched acne,13 control OGTT, however,postprandial T 1996 controlsubjects subjects hyperinsulinemiadoes not I C determinehyperandrogenemia in L patientswith acne. Limitations E includesmallsamplesizeandlackof I generalizability beyond N eumenorrheic women. P Dreno etal,89 RCT, double-blind 1 Subjects with 332 3mo ofEitherelemental Zincandminocyclinearebotheffective R 2001 inflammatory zinc30 mgdailyor intreatment ofinflammatory acne, E acne minocycline 100 mg however,minocycline hassuperior S S daily effect. Cordainet al,37 Cross-sectional 2 Kitavanof Papua 1200Kitavan, Skinexamination No acne wasobserved inany subject. 2002 NewGuinea and 115 Ache´ Difference in prevalenceofacne Ache´ ofParaguay betweenWestern vsKitavan and Ache´ societiesmaybe partly caused byenvironmental factors suchas diet. LGL dietsmay reduce acne. However, these isolatedsubcultures mayhavegenetic and other J environmental factors thataccount A M forabsence ofacne. A Chiu etal,108 Prospective 2 Universitystudents 22 Subject-perceived Worsening perceived dietquality is CA D 2003 cohort with acne dietquality positivelyassociated withacne. D Limitationsincludesmallsamplesize E R anduse oftoolthathas notbeen M A validated formeasuring dietquality. TO L J Adebamowo Retrospective 2 Women 47,355 Food frequency Acneis positivelyassociated with A etal,17 2005 cohort questionnaire of intakeofmilk(particularlyskimmilk), M A distant dietary intake instantbreakfast drink,sherbet, and C A and subject reportof cottagecheese.Associationbetween D D physician-diagnosed acne and dairymay becausedby E R severe acne hormones andbioactivemolecules M presentinmilk.Thesefactorsmaybe AT O more bioavailablein skimmilk. L Limitations ofthis studyinclude its retrospective design, loosely described definitionofacne, subjects’ imprecise recallof diet, failure tocontrol forheredityand socioeconomic status,and low clinical significance offindings. Adebamowo Prospective 2 Girls aged 9-15y 6094 Food frequency Acneis positivelyassociated with A etal,34 2006 cohort atbaseline questionnaire intakeofmilk(particularlyskimmilk). R Limitations includelack ofvalidation T I ofsubjects’ self-report ofacne, low C clinical relevance of findings,and L E failure toaccount forlack oftrend between wholeorlow-fat milkwith I N acne. Smith etal,49 RCT, 1 Male patientsaged 43 LGL diet or PatientsonLGL diet had decreased P R 2007 investigator-blind 15-25 ywith acne conventional HGL totalacnelesions,decreasedweight, E diet (control) decreased freeandrogen index,and S increased IGFBP-1compared with S control subjects. Majorlimitation is thatindependent effects ofweight loss anddietary intervention were notisolated. Furthermore,study did notaccount forpossible differences B indietary fatand fiber intake o w between twogroups. Finally, itis e , difficult togeneralizeresults ofthis Jo s study towomenwho undergo h i cyclical hormonal changesthatmay , a influenceacne. n d Continued Sh a li ta 5 Table I. Cont’d 6 B Levelof Conclusionsand o Study Design evidence Subjects Samplesize Intervention limitations w e Kaymak etal,46 Cross-sectional 3 University students 49Subjects with Foodfrequency Dietary glycemicindex, glycemic load, , J o 2007 with acneand age- acne, 42control questionnaire andinsulin levels were similarin s h and sex-matched subjects patientswith acne compared with i, a controlsubjects controlsubjects. These factors may n notbeinvolvedin pathogenesis of d S acnein youngerpatients. Study h a limitationsinclude lack ofvalidated li questionnaire,failuretoprovidecopy ta ofquestionnaire,questionable timingofinsulinlevelmeasurements, andfailure to accountfor significant differences inIGF-1 andIGFBP-3 A betweenacne andcontrol groups. R Smithet al,48 RCT, 1 Malepatients 43 LGL diet or Patients onLGL diet haddecreased T 2007 investigator-blind aged15-25 y carbohydrate-dense acnelesions, decreased weight, I C with acne diet (control) decreasedBMI,andincreasedinsulin L sensitivitycompared with control E subjects. Majorlimitation isthat I independenteffects ofweight loss N anddietary intervention were not P isolated.Furthermore, studydid not R accountfor possibledifferences in E dietaryfat andfiber intake between S S twogroups. Finally, itis difficultto generalizeresultsofthis studyto womenwhoundergo cyclical hormonalchanges thatmay influenceacne. Adebamowo Prospective 2 Boysaged9-15 y 4273 Foodfrequency Acne ispositively associatedwith et al,21 2008 cohort atbaseline questionnaire intake ofskimmilk.Milkmay influencecomedogenesis through hormonalpathways. Limitations J includelackofvalidationofsubjects’ A M self-report ofacne andfailure to A accountfor lackoftrend between CA D wholeorlow-fat milkwith acne. D E R M A T O L Smithetal,50 Nonrandomized 1 Male patients aged 12 7-dAdmission to IGFBP-1 andIGFBP-3 increased from JA 2008 clinical trial 15-20y with acne researchfacility, baselineinLGLgroup,whereasSHBG M A controlledfeeding decreased frombaseline inHGL C A ofLGL orHGL diet group. Increased glycemicloadmay D D result inincreased activity ofsex E R hormones and IGF-1,thereby M contributing toacne development. AT O Limitationsincludesmallsamplesize, L lack ofrandomization, and inability to generalizeto female adolescents. Smithetal,51 RCT 1 Male patients 31 LGL dietor Comparedwithcontrolgroup,LGLgroup 2008 aged15-25y controldiet hadincreasedratioofskinsurfaceSFAsto with acne MUFAs.Thisincreasedratiowas negativelyassociatedwithnumberof acnelesions.Increasedfollicularsebum outflowwasalsocorrelatedwith A increasedproportionofsebumMUFAs. R Limitationsincludefailuretoaccountfor T I confoundingeffectofweightlossinLGL C group,lackofstandardstoensure L compliancewithspecifieddiet,and E inabilitytogeneralizetofemalepatients I N orsubjectsoutsideof15-to25-yage P range. R Rouhaniet al,56 2009 Cross-sectional 2 Subjects adhering 2528 WorldWide Subjectsreportedimprovementofacne E tolow-glycemic Webebased andreductionofacnetreatment S SouthBeach diet questionnaire whileadheringtolow-glycemicSouth S Beachdiet.Limitationsincludelackof validationofsubjects’self-reportof acne,presenceofconfoundingfactors suchasweightlossandexercise, recallbias,andselectionbias. B o w BMI,Bodymassindex;HGL,highglycemicload;IGF,insulin-likegrowthfactor;IGFBP,insulin-likegrowthfactorbindingprotein;LGL,lowglycemicload;MUFA,monounsaturatedfattyacid;OGTT, e , oralglucosetolerancetest;RCT,randomizedcontrolledtrial;SFA,saturatedfattyacid;SHBG,sexhormone-bindingglobulin. J o Keytolevelofevidence:(1)prospectivecontrolledtrial;(2)retrospectivestudy,prospectivecohortstudy,largecross-sectionalstudy,orlargecaseseries;(3)smallcross-sectionalstudy,smallcase sh series,orindividualcasereports. i, a n d S h a li ta 7 ARTICLE IN PRESS 8 Bowe,Joshi, and Shalita JAMACADDERMATOL In 2005, Adebamowo et al17 re-examined this postulated that the fat content of milk itself was alleged connection between dairy and acne. A total unlikely correlated to comedogenicity. In the study of 47,355 adult women were asked to recall their of Adebamowo et al,17 there were also positive high school diet using a validated food frequency associationsfoundbetweenacneandinstantbreak- questionnaire.Subjectswerealsoaskediftheyever fast drink, sherbet, and cottage cheese. These asso- had physician-diagnosed severe acne. The authors ciations were attributed to the milk content of the found that acne was positively associated with foods. reported quantity of milk ingested, particularly This study had several limitations, notably its skimmilk. retrospective design. Adult women were asked to Theauthorsspeculatedthatdairyinfluencesacne recall their high school dietary intakes. Patients’ through hormonal mediators and by increasing recollectionoftheirmilkconsumptioninthedistant plasma insulin-like growth factor (IGF)-1 levels. past may have been imprecise.17 Women were also Withregardtohormonaleffectsofmilk,itisknown askedtorecalliftheyeverhadphysician-diagnosed that milk contains the testosterone precursors severe acne. The word ‘‘severe’’ is subjective and androstenedione and dehydroepiandrosterone- patients may not have known how their physicians sulfate.18 In vivo, endogenous testosterone is graded acne. Another drawback of this study was converted via 5a-reductase to dihydrotestosterone thatitdidnotcontrolforconfoundingvariablessuch (DHT), which stimulates the pilosebaceous unit. as heredity, nationality, and socioeconomic back- However, DHTcan be produced without 5a-reduc- ground.32 Specifically, the study has been criticized tase in the setting of exogenous 5a-reduced molec- for failing to account for the possibility that the ues.19 Milk contains 5a-reduced steroids, notably iodine content of milk might be contributing to the 5a-androstanedione and 5a-pregnanedione, which apparent association between milk and acne.33 are DHT precursors.18 Both testosterone precursors Finally, the clinical significance of the results has and5a-reducedmoleculesarethoughttocontribute been called into question. The percent of patients to the comedogenicity of milk17,18,20 by stimulating whorememberedhavingphysician-diagnosedacne sebum production and hyperkeratinization of the was 6% for those who drank less than one glass of pilosebaceousunit. milk per week and 8% for those who drank more It has been proposed that milk also increases than 3 glasses of milk per day.32 In a study with comedogenicitythroughinteractionswiththeIGF-1 enoughpower,suchasthis47,355membercohort,it pathway.21Milk,particularlyskimmilk,ispositively is relatively easy to show a statistically significant correlated with higher plasma IGF-1 levels.22-24 association between two variables. However, the IGF-1 stimulates synthesis of androgens in both incrementallyincreasedprevalenceofacneinthose ovarian25andtesticular26tissuesandinhibitshepatic who consumed higher amounts of milk may not synthesis of sex hormone-binding globulin27 result- havebeenclinicallysignificant.Thisstudymightalso inginincreasedbioavailabilityofandrogens.15Both have been subject to information bias. The cohort IGF-128and androgens29,30increasesebumproduc- studiedattendedhighschoolbetweenthe1960sand tion,whichisimplicatedinacne. 1990s. During these years, although the medical Whenbroken downintocategoriesofmilktype, community began to view the link between diet including whole, low fat, and skim, Adebamowo and acne as myth, this information didnot immedi- etal17foundthatonlyskimmilkshowedastatistical ately disseminate to the public. Therefore, sweet correlation with acne. This may be surprising given foods and beverages such as soda were often that one might expect bovine hormones to be avoided, even during the 1960s through 1980s, by concentrated in the lipid fraction of milk.31 The thosewhohadacneinanattempttoamelioratetheir authors hypothesized that the bioavailability of the disease. If these subjects were actively avoiding factors responsible for comedogenicity of milk may soda,theymighthavebeenmorelikelytoconsume be increased by skim milk processing.17 They also milk. postulatedthatskimmilkismoreacnegenicbecause, In 2006, Adebamowo et al34 set out to uphold in comparison with whole milk, skim milk contains their previous findings, this time using a prospec- lessestrogen,ahormoneknowntoreduceacne.17It tive study design. The researchers examined 6094 is unknown how processing affects hormone levels girls aged 9 to 15 years at baseline who were but it has been documented that fermentation that prospectively followed up for up to 3 years. occurs with cheese production results in additional Subjects completed validated food frequency ques- testosterone production from androgen precur- tionnaires and self-reported the frequency and sors.20 Because the association between milk and amount of pimples they experienced. Positive as- acne was more marked for skim milk, it was sociations were found between prevalence of acne ARTICLE IN PRESS JAMACADDERMATOL Bowe,Joshi, andShalita 9 and intake of total milk, whole milk, low-fat milk, reports rather than objective measures. In addition, and skim milk. These trends were significant even they were all observational as opposed to random- when girls using oral contraceptives and girls aged ized, controlled trials. Lastly, the associations that 11 years or older were each excluded. Like the weredetectedwereallweakbyepidemiologicstan- previous study, there was also no association dards in that the odds ratios of associations were between dairy fat and acne. The advantage of this consistently close to 1. However, the authors did study was its prospective design. However, this demonstrateanassociationbetweenmilkandacnein study was limited by the lack of validation of the 3separatepopulations,andthus,furtherinvestigation subjects’ self-report of acne. Severity of acne as throughrandomizedcontrolledtrialsiswarranted. assessed by a physician does not always correlate with self-report of acne. A recent study measuring MORE MODERN STUDIES REVISIT THE the validity of self-report for acne showed that LINK BETWEEN ACNE AND DIET: AN college students cannot accurately report that they EXAMINATION OF THE INFLUENCE OF have acne.35 CARBOHYDRATE INTAKE ON ACNE Again, although statistically significant, the prev- SEVERITY alence ratios demonstrating an association between Although those mentioned above were working milk and acne were very low (maximum of 1.24). out the link between dairy and acne, others were Thus, the clinical relevance of these results remains focusingonthepotentialassociationbetweencarbo- in question. In 2008, Adebamowo et al21 demon- hydrateingestionandacneseverity.In2002,Cordain strated similar results in a cohort of 4237 boys et al37 performed a cross-sectional study in which followedupprospectively.Liketheadolescentgirls, 1300subjectsfromtwonon-Westernizedsocieties,the these boys all completed food frequency question- KitavanIslandersofPapuaNewGuineaandtheAche´ naires and were asked about their pimples. There hunter-gatherersofParaguay,underwentskinexam- was a positive association between intake of skim inationbyageneralpractitionertrainedinthedetec- milkandprevalenceofacne.Themultivariateprev- tionanddiagnosisofacne.Amongthesesubjects,no alence ratio of acne adjusted for age, height, and casesofacnewerereported.Cordainetal37suggested energy intake was 1.19 (95% confidence interval thattheabsenceofacneinthesesocietiesmayhave 1.01-1.40, P value\.02) for those who consumed beenadirectconsequenceoftheirdiets.TheKitavans skim milk more than twice a day compared with andAche´ssubsistonlowglycemicload(LGL)diets, thosewhoconsumedlessthanoneservingperweek devoid of Western refined foods such as cereals, ofskimmilk.Asidefromthissingleweakassociation, chips,cookies,andbread. this study did not find a significant association Theglycemicindex(GI),developedin1981,isa between total milk consumption and acne. There relativecomparisonofthepotentialofvariousfoods wasalsonoassociationbetweenacnewithtotalfat, toincreasebloodglucosebasedonequalamountsof dairy fat, nor vitamin A and thus these compounds carbohydratesinthefood.38Theconceptofglycemic were not thought to contribute to the acnegenic load was developed in 1997, and is calculated by propertiesofmilk. multiplying the GI times the carbohydrate content/ Thistrialdidhavetheadvantageofaprospective servingsize.39Ineffect,theglycemicloadtakesinto design. However, once again, the authors did not account both the quality and quantity of carbohy- account for the fact that there was no significant drates ingested. Dietary glycemic load is a measure trend between intake of whole orlow-fatmilk with of the blood glucose- and insulin-increasing poten- acne.31 As stated previously, bovine hormones tial, as it represents both the rate of carbohydrate would likely be concentrated in the lipid fractions absorption(theGI)andthequantityofcarbohydrate of milk, making this repeated lack of association consumed.40 Examples of low- and high-glycemic between fattier milks and acne perplexing.31 foodscanbefoundinTableII. However,itistheoreticallypossiblethatthefindings In a subsequent review article, Cordain et al41 maybearesultofdifferencesintheglycemicloadof postulatedthatdiet-inducedhyperinsulinemialeads full versus skim milk. For a 250-g serving size, the toa cascade of endocrine responses that may influ- glycemicloadoffull-fatmilkhasbeenreportedtobe ence the development of acne through androgens, 3whereastheglycemicloadofskimmilkis4.36The IGF-1, IGF binding protein (IGFBP)-3, and retinoid nearly equivalent glycemic loads of skim milk and signaling pathways. The authors purport that a full-fat milk can hardly explain the latter’s stronger hyperinsulinemic diet acts as a risk factor in the associationwithacne. developmentof acnethrough influences on follicu- DrawbacksofallthedairystudiesbyAdebamowo lar epithelial growth, keratinization, and androgen- etal17,21,34werethattheywerebasedonsubjects’self- mediatedsebumsecretion. ARTICLE IN PRESS 10 Bowe,Joshi, and Shalita JAMACADDERMATOL Table II. Examplesof low-and high-glycemic with acne and 13 control subjects. Serum-free foods36 testosterone, DHT, and dehydroepiandrosterone- sulfate were significantly higher in the acne group. Glycemic Glycemicloady Basal insulin levels were similar between the two Food index* (per100-gserving) groups, but upon administration of a 75-g, 2-hour Carrots,raw 16 1.6 Peanuts 14 1.7 oral glucose tolerance test, summed insulin levels Oranges, raw 42 3.9 were significantly higher in the acne group. Plums,raw 39 3.9 However, there was no significant difference in the Apples, raw 38 4.8 change in serum testosterone or free testosterone Porridgemadefrom 58 5.1 during the oral glucose tolerance test between the rolled oats two groups. The authors concluded that there was Chickpeas,boiled 28 5.6 mildinsulinresistanceduringtheoralglucosetoler- Spaghetti,white,boiled 42 11.0 ance test in patients with acne, but that the post- Whiteflour bread 70 32.7 prandial hyperinsulinemia did not determine hyperandrogenemiainpatientswithacne.Thisstudy *Listedglycemicindexiswhenglucosesugarisusedasreference food. waslimitedbysmall samplesizeandlackofgener- yAdapted from Foster-Powell et al.36 Glycemic load is estimated alizabilitybeyondeumenorrheicwomen. bymultiplyingglycemicindexbygramsofcarbohydrateper100-g A recent cross-sectional study by Kaymak et al46 servinganddividingby100. measured fasting glucose, insulin, IGF-1, IGFBP-3, and leptin levels in 49 Turkish university students The proposition of Cordain et al,37 linking the with acne and 42 healthy control subjects. All sub- absenceofacneintheKitavanandAche´ toasingle jectsalsocompletedafoodfrequencyquestionnaire variablesuchasdietaryGI,althoughintriguing,has from which investigators calculated overall GI and been heavily criticized. Not only do these isolated dietary glycemic load. No significant differences in subcultureshavesimilardiets,theyalsosharesimilar serum glucose, insulin levels, leptin levels, overall genetic compositionand environmental factors.42,43 GI,anddietaryglycemicloadbetweenpatientswith The findings of Cordain et al37 may have been acne and control subjects were found. The authors bolstered if the acne-free subjects were given diets concludedthatthesevariablesarenotinvolvedinthe rich in high-glycemic foods with subsequent devel- pathogenesisofacne. opmentofacne.42 This study has been criticized for using a diet- Recent noteworthy advances have been made in assessmenttoolthatwasnevervalidated,andfornot ourunderstandingofthecomplexinterrelationships providing it in the article, making one unable to among hyperandrogenism, hyperinsulinemia, and assesstherobustnessoftheinstrument.47Inaddition, acne.Wesummarizethehighlightsoftheseadvances thetimingofphlebotomyhasbeencriticized.Fasting below. However, despite our evolving knowledge, levels of serum markers were used, and do not the link between hyperinsulinemic, high-GI diets represent insulin level excursions and total insulin and acne pathogenesis remains tenuous. Recent exposureoverthecourseoftheday.47Todetectmild studiespresentconflictingresults. insulinresistance,whichisthekindexpectedinthis Ithasbeenpostulatedthatgrowthhormonemay otherwisehealthypopulation,repeatedmeasuresof be involved in the pathogenesis of acne. IGF-1, a serum markers must be drawn at multiple intervals surrogate marker of growth hormone, is often used postprandially. Another drawback of the study was asanindicatorofgrowthhormonesecretionbecause thefailuretoassignGIandglycemicloadvaluesfor it has little diurnal variation. In 1995, Aizawa and meat, poultry, fish, vegetables, and cheese. The Niimura44 sought to investigate the relationship be- authors stated that these foods are not likely to tweenacneand IGF-1.They studied 82postadoles- induce a significant increase in blood glucose. cent eumenorrheic women with acne and 31 age- However, dietary fish rich in omega-3 fatty acids andsex-matchedcontrolsubjectsandfoundthatthe can improve glycemic control, whereas processed IGF-1 levels were significantly greater in patients meats can impair glycemic control. In addition, with acne. However, there was no correlation be- vegetable cooking technique can affect glycemic tween IGF-1 and testosterone, free testosterone, load.47Byfocusingonlyoncarbohydrates,thestudy dehydroepiandrosterone-sulfate, or acne severity. failed to account for the impact of other foods. Inasubsequentstudy,AizawaandNiimura45sought Interestingly, the levels of IGF-1 were significantly to correlate basal insulin and glucose-stimulated higher among patients with acne as compared with insulin levels with androgen levels in women with control subjects.Furthermore, the levels of IGFBP-3 acne. They investigated 30 eumenorrheic women weresignificantlyloweramongpatientswithacneas
Description: