Content uploaded by Mahdieh Abbasalizad Farhangi
Author content
All content in this area was uploaded by Mahdieh Abbasalizad Farhangi on Apr 01, 2020
Content may be subject to copyright.
Int J Clin Pract. 2020;74:e13471. wileyonlinelibrary.com/journal/ijcp
|
1 of 16
https://doi.org/10.1111/ijcp.13471
© 2019 John Wiley & Sons Ltd
Received:31Octob er2019
|
Revised:7D ecember2019
|
Accepted:24December2019
DOI:10.1111/ijcp.13471
Dietary acid load, blood pressure, fasting blood sugar and
biomarkers of insulin resistance among adults: Findings from
an updated systematic review and meta-analysis
Parvin Dehghan1| Mahdieh Abbasalizad Farhangi2
1DrugAp pliedRe searchCenter,Tabriz
UniversityofMe dicalS cience s,Tabriz,Ir an
2ResearchCenterforEviden ceBased
Medicine,Heal thManagementan dSafet y
Promoti onResearchInstitute,Tabriz
UniversityofMe dicalS cience s,Tabriz,Ir an
Correspondence
MahdiehAbbas alizadFarhangi ,Attar
Neyshabouri,DaneshgahBlv,Tabriz,Iran.
Email:abbasalizad_m@yahoo.com
Funding information
TabrizUniver sityofM edica lSciences,
Grant/AwardNumber:IR.TBZMED.VCR.
REC.1398.140
Abstract
Objectives: Thereisnoclearsummarisedreportoftheassociationbetweendietary
acidloadcomponentsincludingpotentialrenalacidload(PRAL)andnet-endogenous
acidproduction(NEAP)withcardiometabolicriskfactors.Inthecurrentmeta-analy-
sis,weaimedtosystematicallyreviewandsummarisetheeligibleobservationalstud-
ies evaluatin g the association bet ween PRAL and NE AP with blood pre ssure and
hypertensionandmarkersofglucosehaemostasisamongadults.
Inasystematicsearch from PubMed,SCOPUS,WebofSciences
andCochraneelectronicdatabasesuptoMay2019,relevantstudieswereincludedin
theliteraturereview.ObservationalstudiesevaluatingtheassociationbetweenPRAL
andNEAPwiththesystolicbloodpressure(SBP),diastolicbloodpressure(DBP),fasting
bloodglucose,insulin,homeostaticmodelassessmentofinsulinresistance(HOMA-IR),
haemoglobinA1C(HbA1C),HOMA-βandquantit at iveinsulincheckindex(QU ICKI)and
alsoprevalenceoroddsofhypertension,hyperglycaemiaanddiabeteswereincluded.
Results: Totalnumberofstudiesincludedinthe14separatemeta-analyseswereas
follows: Mean(SD)of SBP (PR AL,n = 12; NEAP,n = 6), mean (SD) of DBP(PRAL,
n=8;NEAP,n=3),mean(SD)ofFBS(PRAL,n=12;NEAP,n=5),mean(SD)ofHbA1C
(PRAL,n=6;NEAP,n=4),mean(SD)ofHOMA-IR(PRAL,n=7),mean(SD)ofinsulin
(PRAL,n=7;NEAP,n=2);ORoftype2diabetesmellitus(T2DM)(PRAL,n=8;NEAP;
n=6),HTNprevalence(PRAL,n=9; NEAP,n=9),T2DM prevalence(PRAL,n= 7;
NEAP,n=6).Accordingtoourresults,beinginthehighestPRALcategorieswasas-
sociatedwithhigherSBP(WMD=0.98;CI:0.51,1.45;P<.001),DBP(WMD=0.61;
CI:0.089,1.135;P=.022),insulin(WMD=−0.235,CI:0.070,0.40 0;P=.005),higher
oddsofdiabetes(OR=1.19;CI:1.092,1.311;P<.001),higherprevalenceofT2DM
(13%and11%inhighestvslowestcategory).While,beinginthehighestcategoryof
NEAP was onlyassociated withhigher odds of diabetes(OR=1.22; CI: 1.14,1.31,
P<.001).Insubgroupanalysis forfindingthepossiblesourceofheterogeneity,the
continent,dietar yassessmenttool,samplesizeandgenderwerethepotentsources
ofheterogeneity.NoassociationbetweenPR AL and NEAPwithHbA1C,HOMA-IR
was reported.
2 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
|
Metabolicriskfactorsincludingraisedbloodpressure,hyperglycae-
mia and insulin resistance are the most important leading causes of
numerousnon-communicablediseases(NCDs)includingcardiovas-
culardisease(CVD),t ype2diabetesmellitus(T2DM)andmetabolic
syndrome,killingmorethan41millionpeopleeachyearequivalent
to 71% of deaths globally.1Thediseasesarearesultofthecombi-
nation ofgenetic,environmentalandbehaviouralrisk factors;diet
is an important changeable risk factor anddietar y modifications
could substantially reduce the disease occurrence and mortality.2
Acid-base balanceistightlyregulatedinhumanand evenitsminor
changeswouldleadtodeleteriouseffectsincludingchronickidney
disease an d its progres sion, impaire d bone homoeos tasis and in-
sulin resistance.3Recently,theroleofdiet-relatedlow-levelmeta-
bolic acidosis in the pathogenesis of metabolic disorders including
metabo lic syndrome, di abetes and CVDs h as been suggest ed by
numerousresearcheshighlightingthetriggeringeffectsofWestern
dietary pattern.4 -7 Several potential underlying mechanisms for the
association between dietary acid load and metabolic disorders have
also been suggested; it has been mentioned that the association of
the higher dietar y acid load with hypertension and insulin resist-
anceisaresultofexcessiveurinar yexecrationofcalciumandmag-
nesium,increasedcortisolandreducedurinarycitrateexcretions.5,6
Reducedinsulinsensitivity,8 insulin secretion9 and reduced insulin
binding to its receptors because of impaired acid-base balance10
are also several otherpossiblesuggested mechanisms. A dietrich
inacidogenicfoodsincludingmeat,fish,cheeseandlowinalkaline
foods including fruits and vegetables are the potential cause of en-
dogenous acid production and elevated dietary acid load.11Infact,
dietisresponsibleformorethan10-folddif ferenceinendogenous
acid production in different individuals.4 The diet-induced acid
loadisestimatedaccordingtopotentialrenalacidload(PR AL)and
net-endogenous acid production (NE AP) according to information
aboutingestedprotein,potassium,calcium,phosphorousandmag-
nesium.12ThePRALcalculationisbasedontheformulafirstsug-
gested by Remer et al13asfol low s:PRAL (m Eq/d )= 0.4 8 88×pr otein
intake (g/d)+ 0.0366 × phosphorus(mg/d) − 0.0205 × potassium
(mg/d) − 0.0125 × cal cium (mg/d) − 0.0263 × mag nesium (mg/d).
While NE AP is calcu lated based o n the Frasset to et al sugge sted
formula14as:EstimatedNEAP(mEq/d)=(54.5×proteinintake
[g/d] ÷ potas sium intake [mEq/d]) − 10.2. The se calculation s are
validated according to the estimated equivalent s in the 24 hours
urine measurement.13,14 Numerous studies are available reporting
the assoc iation betwee n metabolic risk f actors with di etary acid
load as eitherPRAL or NE AP or both of them.7,12,15-20 Theresults
of these studies are inconsistence; several reported the positive as-
sociationbetween metabolicriskfactors5,6, 21 while others not.4,22
Accordingtoourliteraturereview,onlyonemeta-analysiswascar-
ried out evaluating the association bet ween dietary acid load and
riskofT2DMwithliteraturereviewuptoSeptember2017.23While
no study is available summarising the association between dietary
acidload components(eg PR AL orNEAP) withmetabolicrisk fac-
tors. Th erefore, in the cu rrent meta-anal ysis we summarise d the
results of observational studies evaluated the association between
PRAL ofNEAPwithsystolicanddiastolicblood pressure(SBPand
DBP),serumglucose,insulin,HbA1C, markersof insulinresistance
including homoeostatic model assessment of insulin resistance
(HOMA-IR),hypertension(HTN),hyperglyc aemia,prevalenceofdi-
abet es ,hype r te nsionan do ddsofdiab et esinanupd at edsys te mat ic
reviewandmeta-analysis.
|
The Preferred Reporting Items for Systematic Reviews and Meta-
Analyses(PRISMA)was used forwriting this repor t.24Thecompleted
Conclusions: In the current meta-analysis, we found potent negative effects of
highdietary acidload particularly higherPRALscores cardiometabolic risk factors.
Therefore, loweracidogenicfoodingredientsinthedietsaresuggestedforthepre-
ventionofcardiovascularriskfactorsanddiabetes.
Review criteria
The PubMed, SCOPUS , Web of Sciences and Cochrane
electronic databases were systematically searched from
theirinceptionuptoMay2019identifyallstudiesexamin-
ing the associations between dietary acid load and it s com-
ponentswithcardiometabolicriskfactorsincluding blood
pressure,markersofglucosehomoeostasisandriskoftype
2diabetes mellitus (T2DM).Key termsforinsearch strat-
egy are listed in the Material and Methods section.
Message for the clinic
• The current work evaluated the association between
dietar y acid load, blo od pressure, fas ting blood suga r
and biomarkers of insulin resistanceamong adults ina
systematicreviewandmeta-analysis.
• Higher dietary acidloadwasassociatedwith increased
ris kofcardiometab olicriskf actorsincludingb lo od pr es-
sure,bloodglucose,insulinandhigherriskofT2DM.
• Dietar yacidloadcouldbeassumedasaprognosticdiet-
relatedriskfactorforcardiovascularanddiabetesrisk.
|
3 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
PRISMA checklist is provided in the Supporting Information (Table
S1). The 12-item PRISMA ex tension check list was used to writ e the
Abstract.25
|
Wep er f or medas ys temat ic se arc hu si ng Pu bMe d,SCOPUS, Web
of Sciences and Cochrane ele ctronic databases to the s tudies
evaluated the association bet ween dietary acid load and hyper-
tension an d markers of gluco se homoeos tasis includ ing fasting
serumglucose,insulin,HOMA-IR,HOMA-β,HbA1C,quantitative
in s uli n che c k ind e x(Q U I CKI) andh y per g l ycae m i aup toM a y20 19.
Nolanguagerestrictionwasapplied.Moreover,hand-searching
fromreferencelistsof all relevantpapers, previousreviewsand
meta-analyses wasperformed tocoverall relevantpublications.
Strateg yofsearchwascreatedusingacombinationoftheMeSH
(Medical Subject Headings) terms from the PubMed database
and free te xt words were used. Fo r each electro nic database,
search strategy was adopted.The PICO (patient s, intervention,
comparatorand outcome) for studies' selection is presented in
Table1.ThePICOisoneofthemostwidelyusedmodelsoffor-
mulating and structuring clinical questions in connection with
evidence syntheses. The Cochrane Handbook for Systematic
Reviews spe cifies using PICO as a m odel for developin g a re-
view ques tion, thus ensu ring that the relev ant component s of
the question are well defined.26 ,27Theprotocolofthecurrent
study has beenregistered in PROSPERO withtheidentification
numberofCRD42019122272.Moreover,thestudyprotocolhas
alsob eenr eg is te redb yt heet hics comm it te eo fTab rizU ni ve rs it y
of Medical Sciences (Registration number: IR .TBZMED.VCR.
REC.1398 .140).
|
included studies
Oursearchobtained658 potentiallyrelevantarticlesfromPubMed,
SCOPUS, Web of Sciences and Cochrane electronic databases.
Accordingly,156manuscriptswere remained forfulltextscreening
after re moving duplicate s and excluded accord ing to title and ab-
strac t reading. Totally, 124 manus cripts were excl uded becaus e of
theirirrelevantsubject,inappropriatedesign,reviewsincludingmeta-
analysis orsystematic reviews, conferences and seminars, notrele-
vantagegroups,notevaluatingthestudiedparametersorthetarget
association between them or have a design other than observational
designs.Accordingly32manuscriptswereincludedinthesystematic
review.Figure1presentstheflowchar tofthestudywhileandTable
S2representsthedetailsofexcludedstudiesaf terscreening.
|
Inthecurrentsystematicreviewandmet a-analysis,obse rvationalstud-
ieswiththedesignofcross-sectional,case-controlorcohortevaluating
the assoc iation betwe en dietar y acid load and hyp ertension , systolic
anddiastolicbloodpressure,serumorplasma glucose,insulin, HbA1C,
HOMA-I R, HOM A-β,Q UICKI were in cluded. Accor ding to our set of
parameters, we conducted 14 meta-analyses. Meta-analysisincluded
thestudies evaluatedtheoddsratio(OR), relativerisk(RR),prevalence
ormean±SDoftar ge tv ariab le in th eh ighes tv sl ow es tdiet ary acidloa d
categori es. For the sea rch purpos e, we used MESH (M edical Su bject
Heading) and non-MESH keywords including the following: (‘dietary
acid load’ OR ‘dietar y acid-based load’) AND (‘glucose’ OR ‘fasting
serum glucose’OR ‘fasting bloodglucose’OR‘fastingbloodsugar’ OR
‘bloodsugar’OR‘insulin’OR‘insulinresistance’OR‘homeostaticmodel
assessmentofinsulinresistance’OR‘HOMA-IR’OR‘quantit ativeinsulin
checksensitivity’OR‘QUICKI’OR‘insulinsensitivity’OR‘hypertension’
OR ‘systoli c blood pre ssure’ OR ‘diastol ic blood pre ssure’ OR ‘cardi o-
vascularriskfactors’OR‘cardiometabolicriskfactors’OR‘HbA1C’OR
‘glycosylat ed hemoglobin’ OR ‘hyp erglycemia’ OR ‘obesit y’ OR ‘BMI’
OR‘lipidprofile’(TableS3).Thereviewedliteratureswereinsertedinto
the EndNot e software (ve rsion X8, for Wi ndows, Thoms on Reuters).
Consequentlyretrievedcitationsweremerged,duplicationswereelimi-
natedandthereviewprocesshas been facilitated.Titlesandabstracts
ofallar ti cl eshadbeenev al ua te dindepen de nt lybyt wore vi ewers(MAF
and PD). Articles not meeting the eligibility criteria were excluded.
Moreover,thereferencelistsofrelevantreviewar ticlewerealsoevalu-
atedtobe includedasadditionalstudies.Full-tex ts ofrelevant articles
wer er et riev ed ifme etth ee ligibilitycr it er ia,andwe re re -e va lu at ed .A ny
disagreements were discussed and resolved by consensus.
|
The methodological quality assessment of the included papers
wasperformedbya nine-starNewcastle-Ottawascale(NOS)for
ThePICOcriteriausedforthepresentsystematic
review
Description
Participants General adult population
Exposure
(interventions)
Highestcategor yofdiet aryacidload
representedbyhigherscoresofPRALorNEAP
Comparisons Lowestcategoryofdietar yacidload
representedbylowerscoresofPRALorNEAP
Outcome SBP,DBP,FBS,insulin,HOMA-IR,HOMA-β,
HbA1C,QUICKIandprevalenceoroddsof
hypertension,hyperglycemia,diabetes
Study design Obser vationalstudieswitht hedesignofcross-
sectional,casecontrolorcohort
Abbreviations:DBP,diastolicbloodpressure;FBS,fastingbloodsugar;
HbA1C,haemoglobinA1C;HOMA-IR,homeostaticmodelassessmentof
insulinresistance;NEAP,net-endogenousacidproduction;PICO,patients,
intervention,comparatorandoutcome;PR AL,potentialrenalacidload;
QUICKI,quantitativeinsulincheckindex;SBP,systolicbloodpressure.
4 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
quality assessm ent of the cross-sectional, case-control and co-
hortstudies.The9-pointNOS scalehas scoringrangesfrom0 to
9and is categorised intoselection, comparability and ascertain-
ingofoutcome.Studieswithequalormorethan8starswerecat-
egorised as high quality.28Moreover,theAgencyforHealthcare
Researc h and Quality (A HRQ) checklis t was used to assess th e
qualityofcross-sectionalstudies.29Therewerenoqualitycriteria
forinclusio nofthestudiesinthecurrentmeta-analysis.Theitems
werescored ‘1’if the answerwas‘ Yes’,and ‘0’ if the answerwas
‘No’or‘Unclear ’.Thefinalqualit yassessmentsscoreswereasfol-
lows: low quality = 0-3; moderate quality = 4-7;high quality ≥8.
The det ails of quality s coring for all of t he included s tudies are
provided inTablesS4andS5forcohortandcross-sec tionalstud-
ies,respectively.
|
Data were co llected ac cording to a st andard dat a extrac tion form
gatheringtheinformationabouttheauthorsname,publicationyear,
geograp hical area , study desi gn, parti cipants age r ange, mean age
andnumberofcaseandcontrolgroup,dietar yassessmenttool,set-
ting,gender,samplesize,informationabouttheadjustmentforpos-
sibleconfounders,themainfindingsandestimatesofassociations.
Flowdiagramofstudyscreeningandselectionprocess
Records identified through
database searching (n = 656)
Additional records from manual
search of references or other
sources (n = 2)
Relevant papers included in the PRAL meta-analyses:
a. OR of diabetes (n = 8)
b. Diabetes prevalence (n = 7)
c. HTN prevalence (n = 9)
d. CVD prevalence (n = 5)
e. Mean (SD) of SBP and DII (n = 12)
f. Mean (SD) of DBP and DII (n = 8)
g. Mean (SD) of FBS and DII (n = 12)
h. Mean (SD) of HbA1C and DII (n = 6)
i. Mean (SD) of Insulin and DII (n = 7)
j. Mean (SD) of HOMA-IR and DII (n = 7)
Relevant papers included in the NEAP meta-
analyses:
k. OR of diabetes (n = 6)
l. Diabetes prevalence (n = 6)
m. HTN prevalence (n = 9)
n. CVD prevalence (n = 5)
o. Mean (SD) of SBP and DII (n = 3)
p. Mean (SD) of DBP and DII (n = 3)
q. Mean (SD) of FBS and DII (n = 5)
r. Mean (SD) of HbA1C and DII (n = 4)
s. Mean (SD) of Insulin and DII (n = 2)
Records full text screened
(n = 156)
Articles retrieved for detailed
assessment (n = 32)
Records excluded for the following
reasons:
1. Irrelevant (73 studies)
2. Not evaluating the target study
parameters (13studies)
3. Not evaluated the target relationships
(20 studies)
3. Pregnant women or children (five
studies)
4. Other designs (e.g. RCT, animal
studies) (three studies)
5. Review (10 studies)
Duplicate removal (n = 494)
|
5 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
|
Inthecurrent meta-analysis,threemeta-analysis approaches were
used: the association between odds of diabetes and dietary acid
load wasanalysed by estimating the ORs and 95% confidence in-
tervals(CIs)bycalculating the Ln of ORsand its standarderror of
mean (s.e.) as the effect size of the meta-analysis.Pooled OR (and
95%CI)wasestimatedusingaweightedrandomeffect model(the
DerSimonian-Laird approach). The comparison of the continu-
ous varia bles includ ing SBP,DB P, FBS, insuli n, HOMA-IR , HbA1C,
QUICKI,HOMA-β between highest vs lowest category of dietary
acid load as the reference group was performed by measuring the
unsta ndardised mea n differences a s the effect si ze calculated by
pooled estimateofweightedmeandifference( WMD)with95%CI,
andthefixedeffectsandrandomeffectsmodels.Theprevalenceof
diabetes and HTNinhighestvslowestdietar yacidloadcategories
was performed by re-calculating the proportions of interest from
therelevantnumeratoranddenominator.Theoverallproportionsof
interestwerederivedusingmeta-analysistechniquesby metaprop
command i n the STATAa nd presente d along with 95% CIs cal cu-
latedusinganormalapproximation.Cochran'sQtestandI-squared
testwereusedtoidentifybetween-studyheterogeneity;I2 ˂25% ,
noheterogeneity;I2=25%-50%,moderateheterogeneity;I2>50%
large heterogeneity.30Theheterogeneitywasconsideredsignificant
ife it he rtheQstatistichadP<.1 orI2>50% .S ensitivity an al ysiswas
usedtoexplore the extenttowhichinferencesmightdependon a
particular study or a number of publications. Subgroup analysis was
performedtoidentifypossiblesourcesofheterogeneity,ifrequired.
Begg's Funnel plot swereassessedto evaluatethepublicationbias
followed by the Egger's regression asymmetry test andBegg's ad-
justed rank correlationfor formal statistical assessment ofFunnel
plot asym metry. The dat a were analysed u sing STATA version 13
(S TATACor p),and P-valueslessthan.05we reconsideredass t atis ti-
cally significant.
|
|
hypertension associations
Table 2 presents the summary of systematically reviewed stud-
ies evalua ted the associat ion betwee n dietary a cid load (eg PRA L
or NEAP), blood pressure and hypertension prevalence. Totally,
20 studies repor ted the association between HTN, blood pres-
sureand PRALorNEAP among the systematically reviewedlitera-
ture.4,5,7,12,1 5,18,19,22 ,31-42InthestudybyAkteretal18 evaluating the
associationbetweendietaryacidloadandprevalenceofHTNinthe
FurukawaNutritionandHealthStudytheoddsofHTNinsubjectsin
thehighesttertileofPR ALandNE APwas31%and40%morethan
individ uals in lowest ter tile (PR AL; OR: 1.31; CI: 1. 01-1.70 ; NEAP
OR:1.4 0;CI:1.08-1.82)among2028workingJapanesepopulation.
Several other studies also reported similar results of higher preva-
le n c e o f H T N 1 2,3 4, 41o r h i g h erS B P a n d D BPv a l u e s i nhig h e s t vslowe s t
PRALorNEAPgroupings.1 2, 18 ,3 9,4 0 Only one study reported inverse
associationbetween HTN prevalence among NEAP quartiles4 and
several other studies found no difference.7,15,19,22,31-33,35-38,42Inthe
data anal ysis of Rotterd am study by Eng berink et al ,33 SBP in the
highest tertile of PRAL was significantly higher than the lowest.
While,nosignificantdifferencein themeanvalues ofDBPwas ob-
served.InthestudybyKiefte-deJong34higherprevalenceof HTN
inhighestvslowestquintileofNEAPamongNHS,NHS-IIandHPFS
cohorts was reported.
|
markers of glycaemic status and risk of diabetes
associations
The summa ry of the studi es' character istics evaluate d the asso-
ciation between PR AL, NEAP and markers of glucose homoeo-
stasis, insulin resistance and the prevalence of T2DM are also
presente d in Table 2. The associa tion betwee n dietary ac id load
and glyca emic markers, ins ulin resistanc e and the prevalenc e of
diabetes or the odds of diabetes has been reported in 22 stud-
ies.4,6,7,12,15 ,17-22,31-35,37- 42I nt h e s t ud y b yA k t e r et a l P R A La n d N E AP
scores werepositively associated withHOMA-IR values (P-trend:
.045 and . 03, respec tively). NEAP w as also positive ly associated
withHOMA-β values (P-trend:.03).NoassociationbetweenPRAL,
NEAP and FBS orHbA1Cwasreported.6 Similar results indicating
higher HOMA-IRandHbA1Cvalues,18,32 higher insulin concentra-
tions17 and higher odds of insulin resistance39 in top categories of
PRALorNEAPvslowestcategorieshasalsobeenreportedinfour
ot her s tud i es. Int h es t u dyb yA k t er,6no ass o cia t io n bet w eenPR AL ,
NEAP and FBSorHbA1Cwasrepor ted.Similarfindingswerealso
observed in several other studies.7,12,17,22,31,35,37,39,40,42Akteretal6
repor tedthatmeninthehighestquartilesofPRALhad61%higher
odds of developing diabetes compared with the lowest quartile;
while no ass ociation wa s observe d among women . Moreover, no
associationwasreportedamongNEAP scoresandoddsofT2DM.
Similarly,inthestudybyFagherazzi,15hazardratio(HR)forthein-
cidence ofT2DM according to the PR AL and NEAP categories in
the E3N-EPICcohortstudy wasOR: 1.56;CI: 1.29,1.90 and OR:
1.57;CI :1.30,1.89,respectively(P<.001).Inapopulation-based
study by Gæde et al17inDenmark,womeninthefifthquintileof
PRAL were morelikelytodevelop diabetesafter 15 years follow-
up(OR=1.10;CI:0.98,1.25;P=.02).Whilenoassociationamong
men was reported. Similar findings of the higher prevalence of
diabetes or higher odds of diabetes were reported in two other
studies.37, 41InthestudybyKiefte-deJong,34theoddsofT2DMin
highestquintil eofNEAPandPRALwerehighercomparedwitht he
lowest in NHS and NHSIIcohortswhile in the HPFS study these
associations were not significant. Other reports found no associa-
tions bet ween odds or prev alence of T2DM and PR AL or NEAP
6 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
Characteristicsofstudiesincludedinthesystematicreviewowingtoreporttheassociationbetweendiet aryacidload,hypertension,bloodpressure,markersofglucose
homoeostasisandriskofdiabetes
First author/
country
design Age range
size/
population
of cases/
controls
Dietary assessment/
Results Adjusted variables
Akter5/Japan 2015 Cross-
sectional
Both 18-70y 2028/
working
population
676/676 BDHQ/PRAL,NEAP Th eoddsofdevelopin gHTNinhig hestvs
lowestte rtil eofPRA LandNEAPwere
significant(PR ALOR :1.31;CI:1.01-1.70)
(NEAPO R:1.40;CI:1.0 8-1.82)
Age,sex,work-rel atedPA,
leisure -timePA,smoking,alcoh ol
consump tion,nightorrot ating
shiftwo rk,parentalhistor yof
diabete s,BMI
8
Akter6/Japan 2016 Cross-
sectional
Both 19-69y 1732 433/433 DHQ/PRAL ,NEAP PRA LandNE APwerepositive lyassociated
withHOM A-IRscore(P-trend :.045an d
.03resp ecti vely).NE APwasa lsoposi tively
associa tedwithH OMA-β sco re (P-trend:
.03).Noas sociat ionbet weenPR AL ,NEAP
andFBSorHbA1Cwasreported.
Age,sex,PA,smo king,a lcohol,
shiftwo rk,FHD,B MI
8
Akter21/Japan 2016 Cross-
sectional
Men 45-75y 27808 6952/6952 147-itemFFQ/
PRAL,NEAP
Themenint hehighestqua rtilesofPRAL
had61%highero ddsofdeve loping
diabete scomparedwitht helowes t.NE AP
wasnotass ociatedwithoddsofT2DM.
OR:1.61(CI=1.16,2 .24)P=.05
Age,BMI,HTN,s mokin g,alcohol,
PA,FHD,ene rgy,foo dgroups
intake
9
Akter21/Japan 2016 Cross-
sectional
Women 45-75y 36851 9213/9213 147-itemFFQ/NEAP N o signif icant associat ion bet ween di etar y
acidloa dandodd sofT2DMwasobserved.
NEAPwa snotasso ciatedwithoddsof
T2DM.OR:1 .08(CI=0.72,1.62)P=.7
Age,BMI,HTN,s mokin g,alcohol,
PA,FHD,ene rgy,foo dgroups
intake
9
Akter2/Japan 2017 Cross-
sectional
Both 45-75y 92478 23 120/
23118
147-itemFFQ/NEAP Beingatthehighes tPRA Lquar tile
was associated wi th highe r history of
dyslipidemia (P<.001)
Age,sex,publichealt hcenter
area,BMI,smoking,alcoholuse,
PA,histo ryofHTN,diabetes,
dyslipidemia,occupation,and
energ yintake .
9
Amodu4/USA 2013 Cross-
sectional
Both ≥20y 13 2 74 2490/2477 24HR/NEAP T heprevalenceofH TNandT2DMinlowest
quart ileofNE APwassig nific antlyhigher
than the highes t (P<.001) .
Age,sex,eGFR ,race/ethnicit y,
BMI,poverty,educati on,smo king
status,diure ticuse,T2DM,HTN ,
CVD,Alb–Crrati o,albumin,CRP
6
Bahadoran31/Iran 2015 Cross-
sectional
Both 19-70y 5620 140 5/140 5 147-itemFFQ/PRAL Nosignifica ntdifferencei ntheSBP,DBP,
FBSandth epreval enceofHTNand
hyperglycaemia between quartiles was
reported.
Age,sex,anden ergyi ntake; 6
Banerjee32/USA 2018 Cross-
sectional
Both 21-84y 3257 1074/1075 FFQ HOMA-I Rinthehig hestte rtil ewas
significantly higher. No significant
differenceint hepreva lenceofT2DMand
HTNwasreporte d.
Age,sex,T2DM,e ducat ion,
income,religio n,BMI,insulin
resist ance,smoking,HTN,C VD,
energ yintake
6
Engberink33/
Netherland
2012 Cross-
sectional
baseline
data
Both ≥55y 2241/
general
population
747/747 FFQ SBPinthehi ghestt ertileofPRA Lwas
significantly higher. No significant
differenceint hepreva lenceofT2D Mand
themeanvaluesofDBPwasob served.
Age,sex,BMI,s mokin g,educ ation ,
intakesofenerg y,alcohol,fibre,
7
(Continues)
|
7 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
First author/
country
design Age range
size/
population
of cases/
controls
Dietary assessment/
Results Adjusted variables
Fagherazzi15/
France
2014 Cohort-
baseline
data for
HTN
Follow-up
data for
HRof
diabetes
incidence
Women 40-65y 664 85/
general
population
16621/
16622
208-ite mDHQ HRfortheasso ciatio nbetwe enPRALand
NEAPwi thincidentT2DMwere(OR:
1.56;CI:1 .29,1.90)and(OR:1.57;CI :
1.30,1 .89)resp ecti vely(P<.0 01).The
prevalenceofHTNw asnotsign ificantly
different at baseline.
Age,education,smoking,PA ,
HTN,HCL ,T2DM ,alcoho lintake,
intakeofomega3FA,CHO,
energ y,coffee,sugar,ASB ,fruits,
vegetab le,pro cessedmeat,
dietar ypat terns,BMI
8
Gæde17/Denmark 2018 Cohort-
follow-up
data for
HRof
diabetes
incidence
Men 50-6 4y 25808/
general
population
5162 /516 2 192-I te mFF Q/P RA L TheHRforincid enceofdia betesaccordin g
toPRALscorequi ntileswasnot
significantly different between highest vs
lowest.
BMI,fat,ener gy,PACHOintake,
smoking,
8
Gæde17/Denmark 2018 Cohort-
follow-up
data for
HRof
diabetes
incidence
Women 50-64y 288 43/
general
population
5769/5 769 192- Ite mFFQ/P RAL Womeni nthefif thqui ntileofPR ALwer e
morelikelytodevel opdiabetesafter
15yearsfo llow-up(O R=1.10;CI:0.98,
1.25;P=.02)
BMI,fat,ener gy,PA,CHOi ntake,
smoking,
8
Gæde17/Denmark 2018 Cross-
sectional
(Inter99)
Men 30-6 0y 2842/
general
population
569/569 FFQ/PRAL InsulinandHOMA-IRweresignifica ntly
higherintopquint ilevslowe st.Glucose,
HbA1Cnotdiffe rent.
Age,BMI,smok ing,PA,fat,en ergy,
carbo hydrateintake
8
Gæde17/Denmark 2018 Cross-
sectional
(Inter99)
Women 30-60y 27 90/
general
population
558/558 FFQ/PR AL SignificantlyhigherinsulinandlowerHBA1C
in highest quint ile vs lowe st quin tile. No
significantdifferenceinglucose,HOMA-IR
Age,BMI,smok ing,PA,fat,en ergy,
CHOinta ke
8
Haghighatdoost18/
Iran
2015 Cross-
sectional
Both Mena age
66.8
547/general
population
274/273 FFQ/PRAL FBSwassignificantl yloweran dHbA1C,SBP
significantl y higher in highe st vs lowe st
PRALcategories.
Intakeofprotein,f at,cholesterol,
fibre,w holerefinedgrains,fruit ,
meat,potassium,phosphorus,
beans,nuts,vegetab le,BMI
7
Han12/Korea 2016 Cross-
sectional
Both 40-79y 11601/
general
population
4202/3859 24HR/PR AL S BP,DBPandthep revale nceofHTNa nd
in highest tertile was signifi cantl y higher
thanthelowest .FSG,insulin,HOMA-IR
were not different.
Age,sex,PA,familyhis toryof
cardioa ndcerebro-vascular
disease,T2DM,HTN,LDL ,eGFR ,
urinepH
7
Ikizler19/USA 2016 Cross-
sectional
Both Mean age
60.8
63/general
population
21/21 3-daypro spec tive
FD/NEA P
FBS,insulin,SB P,DBPwerenone
significantl y higher in highe st vs lowe st
NEAPtertile.
Age,ra ce,sex,BMI,eG FR,us eof
diuretics
8
Iwase7/Japan 2015 Cross-
sectional
Both Mean aged
65.7±9.3
149/general
population
74/7 5 DHQ/PRAL Nosignifica ntdifferenceinSBP,HbA1Cwas
observed.
Age,sex,seru muricacidand
creatinine,tot alener gyint ake,
CHOandso diumint ake.
8
Continued
(Continues)
8 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
Continued
(Continues)
First author/
country
design Age range
size/
population
of cases/
controls
Dietary assessment/
Results Adjusted variables
Iwase7/Japan 2015 Cross-
sectional
Both Mean aged
65.7±9.3
149/general
population
74/7 5 DHQ/NEA P Nosignifican tdiffe renceinS BP,HbA1Cwas
observed.
Age,sex,seru muricacidand
creatinine,tot alener gy,CHOand
sodiumintake.
8
Jia20/Sweden 2015 Cross-
sectional
Both ≥70y 861/general
population
215/215 7-dayFR/NE AP No sign ifica nt diff erence i n the prevalence
ofdiabetesbetweenNE APquar tile swas
observed.
Sex,BMI,PA,energyintake,
alcoholintake,smokin g,eGFR
education.
7
Kieft e-deJon g34/
USA
2017 Cohort-
NHS-
median
follow-up
data
Both 30-55y 121 70 0/
general
population
14 974 /
11 449
FFQ/NEA P,PRAL Higherpreva lenceofHT Ninhighe stvs
lowestquintileofNEAPwasrepor ted.T he
oddsofT2DMinhighestquint ileofNE AP
andPR ALwerehighercom paredwiththe
lowest.
Age,energyintake,BMI,FHD,
menopa usalstatus,HTNand
HCL,sm oking ,alcoholintake ,
PA,glycaemiclo ad,AHEIindex ,
western dietar y pattern
9
Kieft e-deJon g34/
USA
2017 Cohort-
NHS2-
median
follow-up
data
Women 25-42y 116430/
general
population
13878/
18030
FFQ/NEA P,PRAL Higherpreva lenceofHT Ninhighe stvs
lowestquintileofNEAPwasrepor ted.T he
oddsofT2DMinhighestquint ileofNE AP
andPR ALwerehighercom paredwiththe
lowest.
Age,energyintake,BMI,FHD,
menopa usalstatus,HTNand
HCL,sm oking ,alcoholintake ,
PA,glycaemiclo ad,AHEIindex ,
western dietar y pattern
9
Kieft e-deJon g34/
USA
2017 Cohort-
HPFS-
median
follow-up
data
Men 40-75y 51529/
general
population
7472 /64 28 FFQ/NEAP,PRA L Higherp revale nceofHTNinhighes tvs
lowestquintileofNEAPwasrepor ted.T he
oddsofT2DMinhighestquint ileofNE AP
andPR ALwereno n-signi ficantlyhigher
than the lowest.
Age,energyintake,BMI,FHD,
HTNmeno pausa lstatus,HCL ,
smokin g,alcoholinta ke,PA,
glycae micload,AHEIin dex,
western dietar y pattern
9
Ko35/Korea 2 017 Cross-
sectional
Both ≥65y 136 9/
general
population
342/343 FFQ/eNE AP Nosignif icantd ifferenceinFB S,SBP,DBP,
theprev alenceofd iabete sandHTN
betwe enlowes tandhigh este-NEAP
quartiles was reported.
Age,sex,Diet aryc aloricand
sodium;B MI,smoking,education,
PA;HTN,T2DM,HLP,CVD
7
Krupp36/Ger many 2018 Cross-
sectional
Both 18-79y 7115/
general
population
1358/1356 FFQ/PR AL Nosignifica ntdifferenceb etwee nSBPand
DBPwasob served.Theprevalenceof
HTNwaslowerinfif thquintilecompared
with the first.
Age,sex,BMI,B P,fastingdu ration,
eGFR,FBS,TC,s mokin g,natrium
excretion,alcoholuse,diuretics,
beta-blockers
8
Kucharska37/
Poland
2018 Cross-
sectional
Men ≥20y 276 0/
general
population
920/920 24HR/NEAP No significant di fference in th e preval ence
ofHTNandtheSBPandDBPvalu es
betwe enNEA Ptertileswasrepor ted.
Age,WC 5
Kucharska37/
Poland
2018 Cross-
sectional
Women ≥20y 34 0 9/
general
population
1136/1137 24HR/NEAP Nosignif icantdifferenceinFB S,the
prevalenceofHTN ,diabetesbet ween
PRALtertil eswasobserve d.
Age,WC 5
Kucharska37/
Poland
2018 Cross-
sectional
Men ≥20y 276 0/
general
population
920/920 24HR/ PR AL Nosig nific antdif ferenceinFBS,t he
prevalenceofHTN ,diabetesbet ween
NEAPte rtil eswasobs erve d.
Age,WC 5
Kucharska37/
Poland
2018 Cross-
sectional
Women ≥20y 34 0 9/
general
population
1136/1137 24HR /PR AL Nosignificantdif ferenceinFBS,the
prevalenceofHTN ,betweenPR ALter tiles
wasobserved.Theprev alenceofT2DMin
highes ttert ileofPR ALwashig her.
Age,WC 5
|
9 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
Continued
First author/
country
design Age range
size/
population
of cases/
controls
Dietary assessment/
Results Adjusted variables
Luis38/Sweden 2014 Cross-
sectional
Both 70-71y 673/general
population
224/224 7d-FR/PR AL Nosign ificantdifferencei nSBP,DBP,
theprev alenceofH TN,T2DM ,between
terti lesofPR ALwasob served.
Age,BMI,smok ingstatus,so dium,
alcohol,energyintake,PA,HL P,
education,C VD,T2DM,eGFR,
SBP,thenight time/day timeSBP
ratio
8
Moghadam39/Iran 2016 Cross-
sectional
Both 22-80y 925/general
population
224 FFQ/PRA L SBPandD BPweresignific antlyh igher
inhighestvslowestquar tileofPRAL .
Nosignif icantd ifferenceinfbs,insulin,
HOMA-IR ,HOMA-β,insulinresist anceat
baselinebetw eenquartilesofPRA Lwas
reported.
Age 6
Moghadam39/Iran 2016 Cohort Both 22-80y 925/general
population
224 FFQ/PR AL,NEAP Significantly higher odds of insulin
resist anceint hehighe stvslowe stPR AL
andNEA Pquar tilesafter3yfollow-up.
Nosignif icantd ifferenceinSBP,DBP,
FBG,ins ulin,HOMA-IR ,HOMA-β,i nsulin
resist anceaf ter3yfoll ow-upbet ween
quart ilesofPR ALwasrepor ted.
Age,sex,BMI, PA,dietaryenerg y,
fat,CHO,SFA,fibre
8
Murakami40/Japan 2008 Cross-
sectional
Both 18-22y 1136/
general
population
227/227 DHQ/PRAL Signific antlyhighervaluesofS BP,DBP
betwe endiff erentqu arti lesofPR AL.N o
differenceinFB S,HbA1Cwasobser ved.
Residen tialblock,residentialarea
size,sur veyyear,PA,smoking ,
BMI,WC .
7
Rebholz41/USA 2015 Cross-
sectional
Both 45-64y 15055 3011/3 011 FFQ/N EAP,PRAL Significantlyhighe rpreval enceofT2DM
(13.8%vs10%)an dHTN(37.7%vs31.9%)
inhighestvslowestPRA Lquar tile.
Age,sex,race-center,caloric
intake,d iabete s,HTN,obese,
smokin g,educ ation ,PA,base line
eGFR
8
van-denBerg22/
Denmark
2012 Cross-
sectional
Both ≥18y 707/ re nal
transplant
patients
236/236 FFQ/NEAP Nodiffer enceinth eSBP,DBP,HbA1Cand
theprev alenceofT 2DMorHTNb etwee n
terti lesofNE APwasreporte d.
Age,BSA,sexm edications,e GFR,
timesincetrans plant ation,
smoking
7
Xu42/Sweden 2016 Cross-
sectional
Both 70-71y 911 304/303 7-dFR/PRAL Nosignifica ntdiff erencei ntheFBS ,insulin,
theprev alenceofT 2DM,HTNb etween
PRALtertil eswasobserve d.ThePR ALor
NEAPwe renotassociatedw ithodd sof
T2DMincidence.
Age,BMI,smok ingstatus,PA,
education,gl ucosedisposalrate,
CVD,HTN ,HLP,energ y-adjus ted
fibre,M UFA,PUFA,SFA,CHO
intake,e GFR,UA ER
8
Abbreviations:24HR,24-hourdietar yrecallquestionnaire;AHEI,alternativehealthyeatingindex;ASB,artificiallysweetenedbeverages;BDHQ,briefvalidatedself-administereddiethis tory
questionnaire;BMI,bodymassindex;BSA,bodysurf acearea;CHO,carbohydrate;CRP,C-reactiveprotein;CVD,cardiovasculardisease;DBP,diastolicbloodpressure;eGFR,estimatedglomerular
filtrationrate;FBS,fastingbloodsugar;FD,fooddiary;FHD,familyhistoryofdiabetes;HbA1C,haemoglobinA1C;HCL,hy percholesterolaemia;HLP,hyperlipidaemia;HOMA-IR ,homeostaticmodel
assessmentofinsulinresistance;HR,hazardratio;HTN,hypertension;MUFA ,monounsaturatedfat tyacid;NEAP,net-endogenousacidproduction;PA,physic alactivity;PRAL,potentialrenalacidload;
PUFA,polyunsaturatedfatt yacid;SBP,systolicbloodpressure;SFA,s aturatedfattyacid;T2DM,type2diabetesmellitus;UAER ,urinaryalbuminexcretion;WC ,waistcircumference.
10 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
scores.20-22,32-34,37,42Inverseassociationsofthehigherprevalence
ofT2DMinthelowestcategories ofNE APinthe studybyAmodu
etal,4lowerHbA1Cconcentrationsinthehigher quintile ofPRAL
inacross-s ec tionalan al ys isofI nter 99cohortofG ædee talstudy17
and lower FBS c oncentrations in t he higher categor y of PRAL18
should also be mentioned.
|
DBP across different dietary acid load categories
All of the studies included inthe current meta-analysis hadcross-
sectional designwhile onlyt wo studies were cohort. Inthe cross-
secti onal and even in the coh ort studies , the baseline dat a (data
befo re fo llow-up) we reinc lu dedinth em eta -a na lysis .T her ef or e,the
design of the included studies could not be a source of bias in the
current analysis. Although, formore assurance, wealsoperformed
subgroup analysis according to all of possible confounders includ-
ing design, country, sample size,gender, dietary assessment tool
and stud y quality s core. Totally, in the met a-analysis of th e mean
differenceofSBPin differentPRALcategories 12studieswerein-
cluded;theForestplotispresentedinFigure2.Accordingly,higher
dietar yacidload was associated with 0.97mmHgincrease in SBP
(WMD = 0.98; CI: 0.51, 1.45; P<.001)withthemoderatehetero-
geneity (Heterogeneity chi-squared = 20.73 [df = 11]; P=.036;
I2=49.6%;Tau
2=0.225).Inthemeta-analysisofNEAPandSBP
including six studies (Figure 2), however, no significant associa-
tion was ob served ( WMD = 0.495; CI = −0. 29,1.28; P=.22)and
no evidence of heterogeneity was also present (Heterogeneity
chi-squar ed = 3.13 [df = 5]; P=.68;I
2=0.0%;Tau
2 = 0.00 ). For
the meta-analysis of the association bet ween dietary acid load
identif ied as PRA L and DBP (Figu re 3), totally, eight s tudies were
includedand higherPRALcategorieswere associated withsignifi-
cantincreaseequalto0.61mmHginDBPvalues(WMD=0.61;CI:
0.089,1.135;P=.022)withapartiallyhighlevelofheterogeneity
(Heterogeneity chi-squared = 28.77 [df = 7];P<.001;I2=75.7%;
Tau 2=0.31).Accordingly,inthemeta-analysisofNEAPandDBP
associat ions (Figure 3), no evid ence of association was obser ved
(WMD =0.03; CI =−1.07,1.13;P=.95)andnoheterogeneitywas
reported(Heterogeneitychi-squared=0.1[df=2];P=.95;I2=0 .0% ;
Tau 2=0.00).TheresultsofthesubgroupanalysisofthePRAL-DBP
associations(TableS6)showedthatsubgroupingaccordingtocoun-
try, dieta ry assessm ent tool and gende r significant ly reduced the
amountofheterogeneityandtherefore,theseparameterscouldbe
considered as the possible sources of heterogeneity. Study quality
was not a source of heterogeneity.
|
1C across different dietary acid load categories
Totallythe associationbetweenPR ALandNEAPwithFBS wasre-
porte d in 12 and 5 studie s (Figure 4). No eff ects of diet ary acid
load measured byPRAL and NEAP on the serum FBS were iden-
tified (PR AL: WMD = 0.034, CI: −2.913, 2.981; P=.98andfor
NEAP: W MD = 0.502; CI: −0.164, 1 .168; P=.139).Thehetero-
geneit y was also high for the FB S-PRAL analy sis (Heterogeneit y
chi-squared = 26.24 [df =1];P<.001;I
2=98.6%;Tau2= 26.23);
whileno heterogeneity was observed forFBS-NEAPassociations
(Heterogeneity chi-squared = 2.93 [df = 4]; P=.57;I
2 = 0.0%;
Forestplotillustrating
weighted mean difference in systolic
bloodpressure(SBP)inhighestvslowest
potentialrenalacidload(PRAL)andnet-
endogenousacidproduction(NEAP)
|
11 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
Tau 2 = 26.23). Sensitivity analysis showed no significant altera-
tionsintheobtainedresults. Thesubgroupanalysisfor findingthe
possible source of heterogeneity for the FBS-PRAL associations
ispresented in Table S7 and countryand dietary assessmenttool
foundtobethepossiblesourcesofheterogeneity.TheForestplot
of the asso ciations bet ween PRAL a nd NEAP wit h serum HbA1C
arepresented in Figure5presenting no significant effects of nei-
ther PRAL nor NE AP on the serum glycosylated haemoglobin
(PRAL: WMD = −0.307,CI: −0.954, 0.341; P=.35andforNEAP:
WMD = −0.032; CI: −0.088, 0.024; P=.265)whileagain,the
great heterogeneity was identified in the PRAL-HbA1C analysis
(Heterogeneitychi-squared=2175.30[df=5];P<.001;I2=99.8%;
Tau 2=0. 6 49)b utn otinN E AP- HbA1Cmeta-analysis(Heterogeneity
chi-squared=0.32[df=2];P=.8 5; I2= 0.0 %; Tau 2=0 .00) .S ubg ro up
anal ys isforth ea ss ociat io nb etweenHb A1Can dP R AL, prese nt edin
TableS8re ve aledt ha ts ubg ro uping ac co rdi ng to co untry,con ti nen t,
dietaryassessmenttoolandsamplesizearethepossiblesourcesof
observed heterogeneity.
Forestplotillustrating
weighted mean difference in diastolic
bloodpressure(DBP)inhighestvslowest
potentialrenalacidload(PRAL)andnet-
endogenousacidproduction(NEAP)
Forestplotillustrating
weighted mean difference in fasting
bloodsugar(FBS)inhighestvslowest
potentialrenalacidload(PRAL)andnet-
endogenousacidproduction(NEAP)
12 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
|
load categories
The Fores t plot of the ef fects of PR AL and NE AP on the ser um
insulinconcentrationsispresentedinFigure6.HighdietaryPRAL
values, increases serum insulin concentrations by 0.23 µIU/mL
(WMD = 0.235, CI:0.070,0.400;P=.005),whilethiseffectwas
not obser ved for the NE AP (WMD = −0. 318, CI: −0.039, 0.676;
P=.081).AmodestheterogeneitywasidentifiedinthePRAL-
insulin analysis (Heterogeneity chi-squared = 14.09 [df=6];
P=.029;I2=57.4%;Tau2=0.022)andnotinNEAP-insulinmeta-
analysis (Heterogeneity chi-squared = 0.17 [df = 1]; P=.68;
I2=0.0%;Tau2=0.00).Accordingtosubgroupanalysis(TableS9),
continent,dietary assessmenttool,sample sizeandgender were
thepossiblesourceofheterogeneity.Inevaluatingtheassociation
between HOMA-IR and dietary acid load, eight studies reported
theassoc iatio nbetweenPRALandHOM A-IRwhileonlyon estudy
reportedtheassociationasNEAP6thereforeitwasexcludedfrom
the anal ysis. According to th e meta-analysis r esults summar ised
in Figure 7 as Fo rest plot, no evidence of the ef fects of PRA L
on HOMA-IR w as obtained (W MD = −0.053, CI: − 0.007,0 .113;
P=.085).Thesensitivityanalysisrevealednomeaningfulchange
in the resu lts. Moreove r,be cause of the high h eterogeneit y ob-
tained (Heterogeneity chi-squared =14759.28[df =6];P < .001;
I2=100.0%;Tau
2=0.005)thesubgroupanalysiswasalsoper-
formed andtheresults arepresentedinTableS10andtheresults
introducedno sourceofheterogeneity exceptforthepossibleef-
fect s of dietary a ssessment too l. Moreover, only t wo studies6, 39
reportedtheassociationofHOMA-βwithPRALwhichnosignifi-
cantassociationwasobserved (Datanot shown). Theinformation
ofQUICKIandhyperglycaemia wasabsent inalmost allofthein-
cludedstudies.Thereforenoanalysiswasdone.
|
dietary acid load-hypertension, diabetes and odd's
ratios of diabetes
Totally, nine studi es report ed the preval ence of HTN in the h igh-
estvslowest category ofPRAL. The Forestplotoftheprevalence
ofHTNby subgroups highestvslowest categories ofPRAL is pre-
sented in Figure S1. Accordingly, the prevalence of HTNwas 19%
(CI: 0.19-0.20) in highest a nd lowest categor y of PRAL. No het-
erogenei ty was obse rved in the m eta-analysi s. The Fores t plot of
theprevalenceof HTN in different NEAPcategoriesis reported in
Figure S2indicating19%prevalence ofHTNinlowest and highest
NEAP categories with no evidence of heterogeneity. The Forest
plot ofthe proportions ofdiabetesinlowest vs highest PR AL cat-
egories (Figure S3) presents the 13% (CI: 0.13, 0.14)prevalenceof
T2DMinthehighestvs11%(CI: 0.10-0.12) in the lowest category
ofPRALincludingsevenstudieswithnoevidenceofheterogeneity.
In the Forest plot of the T2DM prevalence in different NEAP cat-
egorie s(F igureS4),9%pre valen cew asr epor tedbothinhighe stand
lowest category of NE AP with no evidence ofheterogeneity. The
Forestplotofthemeta-analysisofoddsofT2DMinhighestvslow-
est PRAL or NEAPcategories is identified inFigure S5. A positive
associationwasobservedbetweendiabetes and PRAL(OR= 1.19;
CI: 1.092 , 1.311;P<.001)andNEAP(OR=1.22;CI:1.14,1.31,
P<.001)inrandomeffectmodel.Inotherword,beinginthehigh-
est cate gory of PRA L and NEAP ma kes individua ls 19% and 22%
morelikelytodevelopdiabetescomparedwiththelowestcategory.
A great between-study heterogeneity was also observed for the
givenresults(forPRAL:Heterogeneitychi-squared=22.55[df=7];
P=.002;I2=69.0%;Tau2=0.0104andforNEAP:Heterogeneity
chi -squared=11.12[df=5];P=. 049;I2=55 .0 %;Tau2=0.00 69).Fo r
findingthepossiblesourceofheterogeneity,thesubgroupanalysis
basedonthediff er encein in cl udedstu di es ispe rform ed (TablesS11
Forestplotillustrating
weighted mean difference in haemoglobin
A1C(HbA1C)inhighestvslowest
potentialrenalacidload(PRAL)andnet-
endogenousacidproduction(NEAP)
|
13 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
andS12).A cco rdi ngly,inthestudieseval uat ingthedietaryacidload
byPRAL and odds of diabetes,countr y and thesamplesize could
beconsidered as the source of heterogeneit y.InNEAP evaluating
studiescountry,design,samplesizeandgenderdifferencecouldbe
a source of heterogeneity.
|
The Funnel plots revealed moderate asymmetr y (Figures S6 and
S7). However, the B egg's and Egge r's tests provi ded no evidence
ofsubstantialpublicationbias forallofthevariables.Exceptionally,
Egger'stest for the FBS was significant as an evidence of possible
public ation bias. Th e provided valu es are as follow s: DBP,Eg ger's
test (P=.087)andBegg'stest(P=0.93);SBP,Egger'stest(P=.72)
andBeg g' stes t(P=0.5 4) ;FBS ,E gger'ste st(P<.00 1)andBeg g' stes t
(P=.09);HOMA-IR,Egger'stest(P=.87)andBegg'stest(P=0.38).
Ins ulin,Eg ge r'ste st (P=.17)a ndBegg'stes t(P=.99) ;HbA1C,Egger's
test (P=.87)andBegg'stest(P =0.09); ORdiabetes, Egger's test
(P=.33)andBegg'stest(P=.11).
|
Inthecurrentmeta-analysis,wesummarisedtheresultsofstudies
reportingtheassociationbetweenPRAL ,NEAPandmetabolicrisk
factors of glucosehomoeostasis, blood pressure, the prevalence
ofdiabetes, HTN and theoddsofdiabetes. Accordingly,beingin
the highe st categor y of PRAL scor es was associate d with higher
SBP,DBP,insulinconcentrationsandhigherprevalenceandriskof
diabetes c ompared with lowe st category. Wher eas, being in the
highest c ategory of NE AP was only ass ociated with high er odds
Forestplotillustrating
weightedmeandifferenceinInsulinin
highest vs lowest potential renal acid
load(PRAL)andnet-endogenousacid
production(NEAP)
Forestplotillustrating
weighted mean difference in homeostatic
model assessment of insulin resistance
(HOMA-IR)inhighestvslowest
potentialrenalacidload(PRAL)andnet-
endogenousacidproduction(NEAP)
14 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
ofdiabetes. No association between markersof glucosehomoeo-
stasis includingfastingblood glucose, HbA1Cand HOMA-IRwith
PRAL o r NEAP was ob served. A nimal foods i ncluding me at, fish,
egg,chicken,cheeseandalsocerealsarerichinsulphur-containing
aminoacids,phosphorousandchloridearepotentially acid form-
ers; whil e vegetable s and fruit s high in malate , citrate an d gluta-
mate are pote ntially base for mers theref ore, animal-bas ed foods
and high contents in western diets are potentially considered as
mostimportantacid-producerdietsandareassociatedwithhigher
ri sko fi n sul inr e sis t anc e ,h i ghb loo d pre s sur ean ddi a bet e sa ses t a b-
lishedinnumerousworks.11 Accordingly,western dietar ypattern
with high di etary a cid load conte nt, is a potent in ducer of meta-
bolic disorders; several studies had revealed significant relation-
ships bet ween weste rn dietar y pattern an d the increa sed risk of
metabolicsyndrome, hypertension and dyslipidemia. Accordingly,
western dietar y pattern with high content of red meat, eggsand
refined grains is associated with increased riskofobesity and in-
crease dlevelsofb loodsuga r,systo licbloodpre ssure,trigl yce rides ,
and reducedlevels of HDL.43-45IthasbeensuggestedthatPRAL
is a more accurate measure of dietary acid load because it consid-
ersdietaryintakeofproteinandnumerousmicronutrients,potas-
sium, calcium phosphorus andmagnesiumand takes intoaccount
theabsorptionrateofthenutrientsintheintestinalborder,unlike
theNEAPscor e, wh ic ho nlyconsid er thediet ar y pr ot ei na ndpot as -
siumintake.15Therefore,thisleadtoPRALbeagoodpredictorof
the effects of acidity on the body.46Insubgroupanalysisofmen
and women separately,theodds of diabetes amongwomen were
strongerthanmeninbothPRALandNEAPassessment.Apossible
explanation isthe difference in sex-hormones affectingacid-base
balance47an da ls op ossib lythehigh ersam pl es iz eofwom en pa r tic-
ipant scomparedwithmenisapossiblesourceofhighereffectsize
amongthem.Asmentionedinthe resultssection, gender,dietary
assessment tool and continent could be a source of heterogeneity
among obs erved assoc iation. In the cu rrent meta-ana lysis, PRAL
andNE APcalculationswerebasedonself-reporteddatagathered
by 24-hour rec all method , 24-hour record met hod and food f re-
quency questionnaire which might be potential sources of bias.
Moreover,differenceintheitemsoftheFFQmightbeasourceof
heterogeneity;asdescribedpreviously,theFFQitemsrangedfrom
63to168itemsandthelocalfoodsintheFFQcouldalsoaffect
the heterogeneity,48although,almostalloftheincludedstudies
usedvalidatedandreliableFFQs.FFQcoversawiderangeofdi-
etaryingredientsandismoreaccuratethan24-hourrecallmethod
reflecting usual dietar y intake in a short period of time; it has
beenconfirmedthatFFQcouldbemorehelpfulinevaluatingthe
diet-diseaserelationships.49Another sourceofheterogeneity,the
continent ,presents the possible role of geographical distribution,
geneticbackgroundandculturalfactorsinfluencingtheassociation
betweendietary acidloadandmet abolicriskfactors.50Inthecur-
rentmeta-analysis,higherPRALscoreswere associatedwithboth
higherSBPandDBPconcentrationsalthoughno difference inthe
oddsofHTN indifferentPRALor NEAPcategories was reported.
Thepossibleunderlying mechanisms are decline in renalfunction
andreducedcitrateexcretion,increasedcalciumandcortisolsecre-
tion.33WedidnotobserveanyassociationbetweenPRAL,NEAP
and marker s of glucose homo eostasis inc luding FBS, HbA1C a nd
HOMA-IR values. Higher FBS concentrations in higher PR AL cat-
ego rieswererepor tedintheHag hi gh at do ostetals tud y18 although
thisassociation did not achievesignificant threshold, while other
studies reported no significant difference.7,12,17,22,31,35,37,39,40,42
Thecurrentmeta-analysis has severallimitations and strengths;
the current meta-analysis included the results of observational
studie s with the cross-se ctional or coho rt design whic h makes
thecausalinferenceimpossible;although,thestudieswerelarge
population-basedstudieswith acceptablequalit y.However,our
study,basedonourknowledge,isthefirstmeta-analysisevaluat-
ingtheassociationbetweendietar yacidload asboth PRALand
NEAP scoreswitha widerangeofmetabolicriskfactorsinclud-
ing systolic and diastolic blood pressure, fasting serum glucose,
HbA1C, i nsu l in,H O MA- I Ra n dthe p rev a len c eof hyp e r t ens i ona n d
diabetes.Inconclusion,inthe currentmet a-analysis,we found a
potent role of high acid content of diet as a possible leading cause
of metabolic abnormalities, high blood pressure, higher insulin
concentrationsandhighprevalenceofhypertension.Wesuggest
interventional studies in this regard for better causal inference.
Theauthorsdeclarethattheyhavenoconflictofinterest.
MAF and PD designedthe research; MAFconducted the research
and per formed stat istical analy sis; NAF and PD wrote t he paper;
both authors read and approved the final manuscript.
TheprotocolofthecurrentstudyhasbeenregisteredinPROSPERO
with the identification number of CRD42019122272. Moreover,
the study protocol has also been registered by the ethics commit-
tee of Tabriz University ofMedicalSciences (Registration number:
IR.TBZMED.VCR.REC.1398.140).
Mahdieh Abbasalizad Farhangi https://orcid.
org/0000-0002-7036-6900
1. WorldHealth Organization. Non Communicable Disease (NCDs).
2018. https://www.who.int/news-room/fact-sheets/detail/nonco
mmunicable-diseasesAccessedJune1,2018
2. Akter S, N anri A, Mizo ue T,e t al. Dietar y acid load a nd mor-
tality among Japanese men and women: the Japan Public
Health Center-based Prospective Study. Am J Clin Nutr.
2017;106(1):146-154.
3. Kraut JA, MadiasNE. Consequencesandtherapyofthemet abolic
acidosisofchronickidneydisease.Pediatr Nephrol.2011;26:19-28.
4. AmoduA ,Abramowit zMK.Dietar yacid,age,andserumbicarbon-
ate levels among adults in the United States. Clin J Am Soc Nephrol.
2013;8(12):2034-2042.
|
15 of 16
DEHGHA N AND ABBA SALIZ AD FARHAN GI
5. AkterS ,EguchiM, KurotaniK, etal.Highdietar yacidload isasso-
ciated wit h increase d prevalence of hy perten sion: The Furuk awa
NutritionandHealthStudy.Nutrition.2015;31(2):298-303.
6. AkterS,EguchiM,KuwaharaK,etal.Highdietar yacidloadisasso-
ciated with insulinresist ance: The Furukawa Nutrition and Health
Stu dy. Clin Nutr.2016;35(2):453-459.
7. Iwase H, Tanaka M, Kob ayashi Y, et al. Lower veget able protein
intake an d higher diet ary acid lo ad associate d with lower ca rbo-
hydrate intakeareriskfactorsformetabolic syndrome in patients
withtype2diab etes:post-hocanalysisofacross-sectionals tudy.J
Diabetes Investig.2015;6(4):465-472.
8. DeFronzo RA, Beckles AD. Glucose intolerance following
chronic metabolic acidosis in man. Am J Physiol Renal Physiol.
1979;236(4):E328.
9. Rebolledo O R, Hern andez RE , Zanet ta AC, Ga gliardin o JJ. Insulin
secretion duringacid-base alterations. Am J Physiol Renal Physiol.
1978;234(4):E426.
10. WhittakerJ,CuthbertC,HammondVA,AlbertiKG.T heef fect sof
metabolic acidosis in vivo on insulin binding to isolated rat adipo-
cytes. Metabolism.1982;31(6):553-557.
11. Adeva MM, Sou to G. Diet-induced met abolic acido sis. Clin Nutr.
2011;30(4):416-421.
12. Han E, Kim G, H ong N, et al. Ass ociation betwe en dietary ac id
load and the risk of cardiovascular disease: nationwide surveys
(KNHANES200 8–2011).Cardiovasc Diabetol.2016;15(1):122.
13. Remer T, Manz F. Estimation of the ren al net acid excretion by
adults consuming diets containing variable amounts of protein. Am
J Clin Nutr.1994;59(6):1356-1361.
14. Frassetto L A, Todd KM, M orris RC Jr, Seb astian A . Estima tion of
net endogenous noncarbonic acid production in humans from
diet potassium and protein contents. Am J Clin Nutr. 1998;68(3):
576-58 3 .
15. Fagherazzi G , Vilier A , Bonne t F,et al. D ietar y acid load a nd risk
of type 2 diabetes: The E3N-EPIC cohort study. Diabetologia.
2014;57(2):313-320.
16. Faure AM, Fisc her K, Dawson -Hughes B, Eg li A, Bisch off-Ferrari
HA. Gender-specific association betwe en dietary acid load and
total lean body mass and its dependency on protein intake in se-
niors. Osteoporos Int.2017;28(12):3 451-34 62.
17. Gæde J,Nielsen T,Madsen ML,etal.Population-basedstudies of
relationships between dietary acidity load, insulin resistance and
incidentdiabetesinDanes .Nutr J.2018;17(1): 91.
18. Haghighatdoost F, Najafabadi MM , Bellissimo N, A zadbakht
L. Association of dietary acid load with cardiovascular disease
risk fac tors in patients with diabetic nephropathy. Nutrition.
2015;31(5):697-702.
19. Iki zlerH O,Z elnickL ,Ruzi ns kiJ,etal .Diet aryaci dloadisass oc iated
withserumbicarbonatebutnotinsulinsensitivityinchronickidney
disease. J Ren Nutr.2016;26(2):93-102.
20. JiaT,Byberg L , Lindholm B,et al. Dietar y acid load, kidney func-
tion,os teoporosis,andriskoffracturesinelderlymenandwomen.
Osteoporosis Int.2014;26(2):563-570.
21. AkterS,Kurot aniK ,KashinoI,etal.Highdietar yacidloadscoreis
associatedwithincreased riskoftype2diabetesinJapanesemen:
The Japa n Public Heal th Center-based Pr ospecti ve Study. J Nutr.
2016;146(5):1076-1083.
2 2. van den Berg E , Engberink M F,Bri nk EJ, et al. Di etary ac id load
and metabolic acidosis in renal transplant recipients. Clin J Am
SocNephrol.2012;7(11):1811-1818.
23. JayediA,Shab-BidarS.Dietar yacidloadandriskoftype2diabetes:
asystematic reviewanddose-responsemeta-analysis of prospec-
tive observational studies. Clin Nutr.2018;23:10-18.
24. Moher DA, Liber ati A, Tetzlaff J,AltmanDG. Preferredrepor ting
ite ms forsystemat ic reviewsan dm et a-analyses:th ePRISM Astate-
ment. Annals Iintern Med.2009;151(4):264-269.
25. Beller EM, G lasziou PP, Altman DG , et al. PRISM A for abstra cts:
reporting systematic reviews in journal and conference abstracts.
PLoS Medicine.2013;10(4):e1001419.
26. Miller SA, Forrest JL. Enhancing your practice through evi-
dence-based decision making: PICO, learning how to ask good
questions. J Evid Based Dent Pract.2001;1(2):136-141.
27. HigginsJPT,GreenS,eds.Cochrane handbook for systematic reviews
of interventions.Cochranebookseries. Hoboken,NJ:WileyOnline
Librar y;20 08.
28. StangA .Criticalevaluationofthe Newcastle-Ottawascaleforthe
assessmentofthequalityofnonran domizedstudiesinmeta-analy-
ses. Eur J Epidemiol.2010;25(9):603-605.
29. Berkey CS, Ro ckett HRH , Field AE , et al. Acti vity, dieta ry inta ke,
and weight changes in a longitudinal study of preadolescent and
adolescent boys and girls. Pediatrics.2000;105(4):e56.
3 0. Higgins JP, Thompson SG . Quantifying heterogeneit y in a me-
ta-analysis.Stat Med.2002;21:1539-1558.
31. Bahador anZ,MirmiranP,KhosraviH,AziziF.Associationsbetween
dietar y acid-bas e load and ca rdiometa bolic risk f actors in a dults:
The Tehran Lipid and Glucose Study. Endocrinol Metab (Seoul).
2015;30(2):201-207.
32. BanerjeeT,TuckerK,GriswoldM,etal.Dietarypotentialrenalacid
load and r isk of albuminu ria and reduced k idney funct ion in the
JacksonHeartStudy.J Ren Nutr.2018;28(4):251-258.
33. Engberink MF, Bakker SJL, Brink EJ, et al. Dietary acid load
and risk of hypertension: The Rotterdam stu dy. Am J Clin Nutr.
2012;95(6) :143 8-14 44.
3 4. Kief te-de JongJC, LiY,Chen MU, et al. Diet-dependentacidload
and type 2 diabetes: pooled results from three prospec tive cohort
studies. Diabetologia.2017;60 (2):270 -279.
35. KoB-J,ChangY,RyuS, etal.Dietar yacidloadandchronickidney
diseaseinelderlyadults: Proteinandpotassiumintake.PLoS ONE.
2017;12(9):e0185069.
36. Krupp D, Esche J, Mensink G, Klenow S, Thamm M, Remer T.
Dietar yacid load and potassium intakeassociate withbloodpres-
sure and hypertension prevalence in a representative sample of the
German adult population. Nutrients.2 018;10 (1):103.
37. KucharskaAM,Szostak-WȩgierekDE,WaskiewiczA,etal.Dietary
acid loa d and cardi ometabo lic risk in th e Polish adu lt populat ion.
Advan Clin E xp Med.2018;27(10):1347-1354 .
38. LuisD,HuangX,RiserusU,etal.EstimateddietaryacidloadIsnot
associated with blood pressure or hypertension incidence in men
whoareapproximately70yearsold.J Nutr.2015;145(2):315-321.
39. Moghadam SK , Bahadoran Z, Mirmiran P, Tohidi M, Azizi F.
Associationbetweendietaryacidloadandinsulinresistance:Tehran
LipidandGlucoseStudy.Prev Nutr Food Sci.2016;21(2) :104-109.
4 0. MurakamiK,SasakiS,TakahashiY,UenishiK.Associationbetween
dietary acid-base load andcardiometab olicrisk factors in young
Japanesewomen.Brit J Nutr.2008;100(3):642-651.
41. Rebholz CM,Coresh J, Grams ME, et al. Dietar yacid loadandin-
cident chronic kidney disease: Results from the ARIC Study.Am J
Nephrol.2015;42(6):427-435.
42. Xu H, Jia T, Huang X, et al. Dietary acid load, insulin sensitiv-
ity and risk of type 2diabetes in community-dwelling older men.
Diabetologia.2014;57(8):1561-1568.
43. FarhangiMA,VajdiM,NajafiM.Diet aryacidloadsignificantlypre-
dicts10-years sur vivalinpatientsunder went coronary ar teryby-
passgraf ting(C ABG)surgery.PLoS ONE.2019;14(10):e022383 0.
4 4. Farhangi MA, Jahangir y L, Asghari-Jafar abadi M, Najafi M .
Association between dietar y patterns andmetabolic syndrome in
asampleofTehranianadults.Ob es Res Clin Pract.2016;10: S6 4-S73.
45. Farhangi MA , Nikniaz L, Ni kniaz Z. Higher d ietary aci d load po-
tentially increases serum triglyceride and obesity prevalence in
adults:anupdatedsystematicreviewandmet a-analysis.PLoS ONE.
2019;14(5):e0216547.
16 of 16
|
DEHGHAN AND ABB ASALI ZAD FARH ANGI
46. MichaudDS ,Troiano RP,Subar AF,etal.Comparisonofestimated
renalnetacidexcretionfromdietaryintakeandbodysizewithurine
pH.J Am Diet Assoc.20 03;103(8):10 01-1007.
47. Orr-WalkerBJ,HorneAM,EvansMC ,etal.Hormone replacement
thera py causes a res piratory al kalosis in nor mal postme nopausal
women. J Clin Endocrinol Metabol.1999;84(6):1997-2001.
48. Mueller M, Hobiger S, Jungbauer A. Anti-inflammatory ac-
tivity of extracts from fruits, herbs and spices. Food Chem.
2010;122(4):987-996.
49. ShimJS,OhK,K imHC .Dietaryassessmentmethodsin epidemio-
logic studies. Epidemiol Health.2014;36:e2014009.
50. SlimaniN,FaheyM,WelchA,etal.Diversityofdietarypatternsob-
servedin theEuropean ProspectiveInvestigation into Cancer and
Nutrition(EPIC)project.Public Health Nutr.2002;5:1311-1328.
Additional supporting information may be found online in the
Suppor tingInformationsection.
DehghanP,AbbasalizadFarhangiM.
Dietar yacidload,bloodpressure,fastingbloodsugarand
biomarkersofinsulinresistanceamongadults:Findingsfrom
anupdatedsystematicreviewandmeta-analysis.Int J Clin
Pract. 2020;74:e13471. https ://doi.o rg /10.1111/ijc p.13471