Consumption of guava (Psidium guajava L) and noni (Morinda citrifolia L) may protect betel quid-chewing Papua New Guineans against diabetes

Asia Pac J Clin Nutr 2008;17 (4):635-643
635
Original Article

Consumption of guava (Psidium guajava L) and noni
(Morinda citrifolia L) may protect betel quid-chewing
Papua New Guineans against diabetes

Patrick L Owen PhD1, Louis C Martineau PhD2, Dayna Caves BSc1,
Pierre S Haddad PhD2, Teatulohi Matainaho PhD3 and Timothy Johns PhD1

1School of Dietetics and Human Nutrition, Macdonald Campus of McGill University, Quebec, Canada
2Natural Health Products and Metabolic Diseases Laboratory, Department of Pharmacology, Université de
Montréal, Quebec, Canada
3Discipline of Pharmacology, School of Medicine and Health Sciences, University of Papua New Guinea,
Port Moresby, Papua New Guinea


Rapid increase in the incidence of type 2 diabetes (DM2) in Papua New Guinea, coupled with
compelling epidemiological evidence supporting a diabetogenic association with betel quid (BQ)
chewing has lead us to investigate dietary strategies that might offer protection from developing DM2.
We investigated the dietary habits of Kalo residents from coastal Central Province who are avid BQ
chewers yet have a relatively low incidence of DM2 compared to the ethnically similar and adjacent
Wanigelans who abstain from BQ yet have an unusually high incidence of DM2. In Kalo, guava bud
(Psidium guajava L) and noni (Morinda citrifolia L) were consumed much more frequently than in
Wanigela, whereas the inverse was observed for mangrove bean (Bruguiera gymnorrhiza (L) Lam.).
These plants, along with BQ and its component ingredients areca nut (Areca catechu L) and Piper
betle L inflorescence, were assessed for their ability to mediate insulin-dependent and insulin-
independent glucose transport in cultured 3T3-L1 adipocytes. A dose-dependent inhibition of glucose
uptake from methanolic extracts of BQ, areca nut and P. betle inflorescence supports previous reports
of pro-diabetic activity. Conversely, guava bud extract displayed significant insulin-mimetic and
potentiating activity. Noni fruit, noni leaf, commercial noni juice and mangrove bean all displayed
insulin-like activity but had little or no effect on insulin action. Habitual intake of guava and noni is
proposed to offer better protection against DM2 development and/or betel quid diabetogenicity than
cooked mangrove bean. These findings provide empirical support that DM2 risk reduction can be
accomplished using traditional foods and medicines.

Key Words: Psidium, areca, 3T3-L1 cells, plants, medicinal



INTRODUCTION
ing Kalo residents from developing DM2 and from BQ
Populations undergoing nutritional and social transition
diabetogenicity.
due to economic development and globalization are at
Beyond their macronutrient composition and glycemic
greater risk of obesity and type 2 diabetes (DM2).1 Com-
index, several foods and medicinal plants have antidia-
pelling epidemiological and animal studies suggests that
betic properties attributed to various phytochemicals that
betel quid (BQ) chewing, popular in some Pacific nations,
stimulate pancreatic insulin secretion, reduce peripheral
exacerbates this risk.2-4 In Papua New Guinea, chewing
insulin resistance or decrease carbohydrate absorption.8
BQ was reportedly the most significant independent risk
Conversely, some plant compounds such as the nitrosa-
factor for DM2 (odds ratio 3.4; 95% CI. 2.0-5.9).5 How-
mines in areca nut can be diabetogenic.4 Differential con-
ever, DM2 is rare in Kalo, a coastal community where

residents habitually and avidly chew BQ.6 In contrast, the

ethnically similar and geographically adjacent Wanigela
Corresponding Author: Dr Patrick Owen, School of Dietetics
people have been recorded as having an unusually high
and Human Nutrition, McGill University, 21,111 Lakeshore
incidence of DM2 in both urban and rural settings,7 yet
Road, Ste-Anne-de-Bellevue, Quebec, Canada, H9X 3V9
for religious reasons abstain from chewing BQ. A plausi-
Tel: (514) 996-4106; Fax: (514) 398-7739
ble explanation for this apparent paradox is that some
Email: patrick.owen@mcgill.ca
element(s) in the diet, lifestyle or environment is protect-
Manuscript received 29 July 2008. Initial review completed 23
October 2008. Revision accepted 27 November 2008.

---

636
PL Owen, LC Martineau, D Caves, PS Haddad, T Matainaho and T Johns
sumption patterns of pro- and antidiabetic traditional
Participants were weighed without shoes and in light
plants may mediate DM2 risk over and above the risk
clothing to the nearest 100g on a Seca Dial Scale (Vogel
imposed by the nutrition transition.
& Halke, Germany). Standing height was measured using
The aim of this study was to select traditional plants
wooden boards with a measuring tape (0.1 cm precision)
that were prominent and/or distinct in Kalo and Wanigela
that were built locally based on the UNICEF model.
using a pretested ethnomedical questionnaire, and to as-
Waist and hip circumference was measured using a fiber-
sess these plants’ ability to mediate insulin-independent
glass measuring tape. Skin-fold thickness was obtained
and -dependent glucose uptake in cultured 3T3-L1 adipo-
using Lange calipers and body fat % calculated according
cytes, a cell line used extensively to study peripheral tis-
to formulae supplied by the manufacturer for four sites:
sue insulin responsiveness.9 The study was designed as a
tricep, bicep, subscapular and suprailiac. Blood pressure
preliminary screening of commonly used medicinal plants
was taken in duplicate after a participant was asked to sit
for antidiabetic activity, and multivariable analysis with
quietly for five minutes. In cases where hypertension was
community plant use frequency lead to our hypothesis
detected (>120/80), another reading was taken after ten
that guava and noni consumption may counter diabetes
minutes. If still hypertensive, the participant was referred
risk. Focus was directed towards medicinal plants rather
to the Community Health Worker. Participants were en-
than food plants (with the exception of mangrove bean)
couraged to retest their blood pressure at any time during
since their role in non-communicable disease mediation is
the study. Fasting blood glucose (FBG) concentration was
increasingly becoming an area of interest to researchers
determined using a portable glucometer (CardioChekTM
and public health workers who are developing culturally
Analyzer, Polymer Technology Systems Inc, Indiana)
appropriate strategies to promote wellness and health.
after an overnight 12-hour fast. Readings were obtained

within 2 minutes. Using current WHO diagnostic criteria,
SUBJECTS AND METHODS
participants with glucose concentrations above 6.1
Study area
mmol/L were considered to have impaired glucose toler-
A quantitative ethnobotanical survey of medicinal and
ance and those above 7.0 mmol/L to have diabetes.11
food plants was conducted in Kalo (Central Province,
Newly identified hyperglycemic individuals were retested
10.050° S, 148.200° E), Koki (National Capital District,
for confirmation and referred to the Port Moresby Gen-
9.483° S, 147.167° E), and Wanigela (Central Province,
eral Hospital Diabetes Clinic for further testing and
10.050° S, 147.783° E) from January to August, 2004.
treatment. Participants with known diabetes were ex-
Wanigela is a lagoon stilt village located approximately
cluded from the analyses since this may have influenced
400 km east from the capital. Some residents engaged in
dietary and lifestyle patterns.
subsistence fishing and farming, while others relied on

city relatives to provide trade store goods. Koki, a perma-
Plant selection and extract preparation
nent Wanigela settlement located in a relatively affluent
Plants included in the present study were selected from a
suburb of Port Moresby, had direct access to grocery
combination of quantitative and qualitative measurements
stores and open air markets. Both Wanigela and Koki
determined by the frequency of their usage, the EDDC
practice the Seventh-Day Adventist religion which em-
questionnaire and their local sociocultural importance.
phasizes health and diet by advocating vegetarianism and
Plants were extracted in methanol using the Soxtec™
abstinence from stimulants such as betel quid, alcohol,
system as described elsewhere.12 Voucher specimens
caffeine and tobacco. Residents of Kalo, a road-connected
were deposited at the University of Papua New Guinea
village located roughly equidistant between Koki and
and McGill University herbarium. Betel quid (BQ) was
Wanigela, are ethnically similar and are predominantly
prepared by combining approximately 66.6% areca nut
subsistence farmers and fishermen.
extract (AN), 26.6% Piper betle extract (PBI) and 6.6%
Participants were selected by stratified random sam-
calcium hydroxide before extraction. Cooked mangrove
pling according to gender and age and administered a
bean (MBC) was prepared according to traditional meth-
quantitative questionnaire modeled after the Expanded
ods: thin slices were soaked for 1 h and boiled in two
Diabetes Diagnostic Criteria (EDDC) developed by Carl-
changes of sea water. Noni juice (NJ) was purchased
son.10 The questionnaire invited participants to list all
from a local health food store and used in bioassays as
plants indicated for DM2 symptoms. All individuals
purchased (Flora Manufacturing & Distributing Ltd, Bur-
above 16 years were eligible to participate. The survey
naby, BC).
covered a final sample of 365 participants roughly di-

vided between the three villages. Approval from the head
Cell Culture
of each local level government was obtained, followed by
3T3-L1 murine pre-adipocytes were purchased from
a public information session and individual prior in-
American Type Cell Collection (ATCC; Manassas, VA)
formed consent before initiation of the study. Permission
and cultured in a humidified 37 °C 5% CO2: 95% air at-
and ethics approval was obtained from the McGill Uni-
mosphere in Dulbecco’s modified eagle medium contain-
versity Ethics Committee, the Papua New Guinea Medi-
ing 10% fetal bovine serum (FBS) and penicillin-
cal Advisory Board and the Papua New Guinea Depart-
streptomycin antibiotics (Invitrogen Life Technologies,
ment of Environment and Conservation.
Burlington, ON) as described previously.13 Upon 80%

confluence, differentiation was initiated by adding 250
Anthropometric measurements and blood glucose de-
?mol/L 3-isobutylmethylxanthine, 1 ?mol/L dexa-
termination
methasone, and 500 nmol/L insulin for 2 d and further

---

Guava and noni protects against diabetes
637

continued with FBS and insulin until at least 90% of the
3H] glucose (TRK-383, Amersham Biosciences, Baie
adipocytes developed visible lipid droplets.
d’Urfé, QC) for 10 minutes at 37°C without extracts.

After incubation, cells were placed on ice and immedi-
Determination of plant extract concentration for cell
ately rinsed three times with ice-cold KRPB, lysed with
culture
0.1 mol/L NaOH for 30 minutes and scraped. The lysate
Plant extracts were solubilized in dimethyl sulfoxide
and additional rinse were added to 4 mL of liquid scintil-
(DMSO) at a concentration of 200 mg/mL, filter-sterilized,
lation gel (Ready-Gel 586601; Beckman Coulter Inc,
aliquoted, and stored at -20°C. Stock solutions were di-
Fullerton, CA) and incorporated radioactivity was meas-
luted in culture medium to produce a final concentration
ured in a scintillation counter. A well-recognized hypo-
of 200 µg/mL and a final DMSO concentration of 0.1%.
glycemic plant extract, fenugreek seed (Trigonella foe-
Additional concentrations of 100 and 50, as well as 25
num-graecum L) ethanolic extract14, was used as a posi-
and 12.5 µg/mL for PBI, were obtained by serial dilution
tive control at a maximal non-toxic dose of 75 ?g/mL
and addition of DMSO to keep solvent concentration con-
(data not shown). After this period, cells were incubated
stant. A trial to determine maximal non-toxic concentra-
for an additional 3 h in serum-free medium. Results are
tions by observing cellular morphological changes derived from the average of three independent experi-
showed that all plants except PBI were well-tolerated at
ments performed in triplicate and expressed as the change
concentrations up to 200 µg/mL (data not shown).
in glucose uptake activity relative to basal levels obtained

from incubation with the vehicle.
Glucose Uptake Assay
Following the methods of Martineau et al.13 differentiated
Statistical Analysis
and confluent 3T3-L1 adipocytes grown in 12-well plates
Differences in consumption patterns, anthropometric and
were incubated with vehicle (DMSO), plant extract or
clinical measures between sexes and between villages
positive control for 18 hours. Thereafter, cells were rinsed
were assessed using independent Student’s t-tests.

twice with Krebs Ringer phosphate buffer (KRBS) solu-
Spearman’s correlation for non-normally distributed data
tion (20 mM HEPES, 4.05 mM Na2HPO4, 0.95 mM
was used to find associations between consumption pat-
NaH2PO4, 136 mM NaCl, 4.7 mM KCl, 1 mM CaCl2, 1
terns and health parameters. Fasting blood glucose was
mM MgSO4, 5 mM glucose, 0.5% BSA, pH 7.4) at 37°C
first normalized by inverse transformation (100-1/FBG)
and then treated with 0, 1, or 100 nM insulin in this buffer
and multivariate linear regression performed with age,
for 30 minutes in the continued presence or absence of
waist circumference and weight as covariates. The inter-
plant extract. Cells were then washed twice with glucose-
action of plant extract concentration on insulin dose re-
free KRPB and treated with 0.5 ?Ci/mL 2-deoxy-D-[1-
sponse was assessed using 2-factor ANOVA with pair-


Table 1. Characteristics of plants, the area of collection, ethnomedical indications and methanolic extract yield.

Family, species, Voucher
Extract yield
Plant part
English, local name
Ethnomedical indications†
Area collected
Number
(%)
Arecaceae
Stimulant/ sedative, appetite
Areca catechu L (AN),
Seed Betelnut,
Buai suppressor/ stimulant, anti-
Koki, Kalo
24.4
KAL-12
malarial
Antidiabetic, antibacterial,
Myrtaceae
anti-malarial, anti-
Psidium guajava L (GB),
Bud
Guava, Tuava
Kalo 18.4
inflammatory, antidiarrheal,
KAL-01
gastrointestinal tonic
Piperaceae
Inflores-
Piper betle L (PBI),
Pepper, Daka
Antibacterial
Koki, Kalo
12.5
cence
KAL-24
Rhizophoraceae
Cooked
Mangrove
Bruguiera gymnorrhiza (L.)
Antimicrobial, insecticidal Wanigela

2.1
hypocotyls bean, Kavela
Lam. (MBC), WAN-01
Raw
(MBR)

Wanigela
23.5
hypocotyls
Rubiaceae
Morinda citrifolia L (NF)
Fruit Kalo
29.4
Noni, Nono
KAL-03
Analgesic, anti-
inflammatory, hypotensive,
(NJ) Juice
Commercial
antibacterial, tonic
(NL) Leaf

Kalo
23.3
(NR) Root

Koki 17.3
Betel quid (BQ)




13.3

†Ethnomedical information was obtained from interviews with participants and traditional healers.

---

638
PL Owen, LC Martineau, D Caves, PS Haddad, T Matainaho and T Johns
Table 2. Population characteristics of adult men and women of Kalo, and the Wanigelan urban and rural communities
Koki and Wanigela, Papua New Guinea. Data expressed as mean ± SD. The proportion of people who have consumed
the plant at least once during the survey week is also included.


Kalo Wanigela
(semi-rural) Koki
(urban) Wanigela
(rural)
Male Female Male Female Male Female
N
60 61 60 60 60 65
Age (y)
43.2 ± 17.9
42.6 ± 15.3
35.8 ±17.8
42.0 ± 16.5
40.7 ± 16.2
40.7 ± 19.4
Fasting blood glucose (mmol/L)
3.5 ± 1.3
3.6 ± 1.3
4.2 ± 3.5
3.8 ± 1.3
4.8 ± 3.5
3.8 ± 1.3
Diabetes (%)†
3.3*
1.6
10.2a* 5.0 13.3a* 4.6
Weight (kg)
68.3 ± 15.3*
59.1 ± 13.2
67.0 ± 17.9
61.7 ± 17.4 59.1 ± 10.2ab* 48.7 ± 8.3ab
Stature (m)
1.7 ± 0.6*
1.6 ± 0.1
1.6 ± 0.7a*
1.5 ± 0.1a
1.6 ± 0.1a*
1.5 ± 0.1a
Body mass index (kg/m2)
23.6 ± 4.8
23.1 ± 4.4
25.3 ± 5.4
26.0 ± 6.2a
22.3 ± 3.1b*
21.0 ± 3.0b
Overweight (%)‡ 26.6
26.2
48.3a 46.6a 6.7ab 8.5ab
Waist circumference (cm)
85.6 ± 13.6
85.6 ± 14.7
87.4 ± 14.43
85.4 ± 13.2
80.4 ± 8.85
77.14 ± 8.3
Abdominal Obesity (%)§ 15.0*
32.8
13.6a* 35.0a 1.7ab* 10.8ab
Waist: hip ratio (cm)
0.93 ± 0.11
0.90 ± 0.13
0.94 ± 0.09
0.87 ± 0.10a 0.92 ± 0.07b 0.89 ± 0.06ab
% Body fat
21.6 ± 5.5*
32.5 ± 6.7
23.6 ± 7.7*
36.3 ± 6.9a
18.4 ± 4.7ab* 29.4 ± 5.4ab
Tricep skinfold thickness (mm)
19.0 ± 6.7
19.0 ± 6.7
13.9 ± 8.2
22.5 ± 9.3
9.4 ± 4.0
13.5 ± 4.9
Excess body fat (%)¶ 49.2*
49.2
67.2a* 75.0a 33.3ab* 27.7ab
Systolic blood pressure (mm Hg)
131 ± 18.5
128 ± 14.6
129 ± 15.9
130 ± 18.1
128 ± 16.5
133 ± 16.7
Diastolic blood pressure (mm Hg)
75.9 ± 10.2
75.5 ± 7.8
81.4 ± 7a
81.1 ± 12.3a
79.9 ± 11.3
80.6 ± 7.5a
Hypertension (%)†† 15.0*
13.1
22.0*
28.3a 16.7 23.1a
Proportion who consumed the plant (%)






Betel quid
93.3
93.4
33.3a 1.7a 6.7ab* 0.00a
Guava bud
15.0
14.8
1.7a 2.0a 3.3a* 0.00a
Noni 21.7
18.0
2.0a* 8.2a 11.7a* 0.00a
Mangrove bean
0.0
0.0
48.3a* 86.7a 93.3ab* 100.a

†Newly diagnosed diabetes was defined as fasting blood glucose >7.0 mmol/L; ‡BMI>25.0; §Waist circumference >102 cm for men, >80 cm
for women; ¶% body fat >20% for men; >32% for women according to guidelines from the American Council on Exercise;
††Systolic/diastolic blood pressure >140/90 mm Hg
ap<0.05 between same sex vs from Kalo
bp<0.05 between same sex vs from Koki
*p<0.05 between sexes of the same village

wise comparisons. All data are presented as mean ± SEM
Consumption patterns of these plants along with the
from at least 3 independent experiments performed in
health characteristics of males and females of each com-
triplicate with significance set at p?0.05. Analysis was
munity are presented in Table 2. Both sexes living in rural
performed with SPSS 15.0 for Windows (SPSS, Inc.,
Wanigela were significantly less heavy, had a lower BMI,
Chicago, IL, USA).
a lower WHR indicating less prevalence of abdominal

obesity and a lower total body fat percentage compared to
RESULTS
age and sex-matched residents of urban Koki and semi-
Plant and human population characteristics
rural Kalo. Despite this, Wanigelans had a prevalence of
Collection details of the plants used in this study, their
DM2 (males: 13.3%; females: 2.8%) and hypertension
ethnopharmacological indications and extract yield are
(males: 18.3%; females: 23.1%) equal to that of their ur-
presented in Table 1. Although Papua New Guineans
ban counterparts. One dietary factor that is shared be-
consider the classification of foods and medicines as a
tween the two communities is the consumption of man-
continuum, most recognize mangrove bean as principally
grove bean. Almost all of the Wanigelans interviewed
a food, noni and guava bud as principally medicines, and
relied on mangrove bean as their primary source of en-
betel quid as a habit akin to chewing gum, but with rec-
ergy, and roughly half of the males and 87% of the fe-
ognized medical properties. Mangrove bean, although
males in urban Koki had consumed it at least once during
considered a food plant, was included in this study be-
the week. Considered more of a famine food in Kalo,
cause of its distinctiveness as being one of the world’s
none of the villagers had consumed mangrove bean dur-
few remaining human dietary staples that are still har-
ing the survey period. Virtually all Kalo villagers regu-
vested from wild populations. It was also included since
larly chewed betel quid, roughly 15% of the population
little of its phytochemistry and pharmacology is known.
had consumed an infusion of guava buds and young

---

Guava and noni protects against diabetes
639

50 ?g/mL and inhibited uptake at 100 and 200 ?g/mL


(Figure 2d). Raw mangrove bean (MBR) exhibited strong
l)

ro
nt
activity at all tested concentrations, however cooking of
MBR reduced its activity by 65% suggesting that removal
n
ly co

baseline
m
o
c
l
e-o
of tannins was responsible for the loss in activity (data
e
fr
g
vehi
not shown). Of the BQ ingredients, AN elicited a 40-60%
n
e

to
increase in glucose uptake (Figure 3a) while PBI had no
% Cha
(Relativ
effect at 12.5 ?g/mL and at higher concentrations inhib-
ited glucose transport (Figure 3b). As expected, BQ ex-
hibited a combination of these effects corresponding to
Insulin nmol/L

the proportion of its PBI and AN content (Figure 3c).
Figure 1. Percent change in glucose uptake from vehicle

(DMSO)-treated cells in 3T3-L1 adipocytes treated with guava
Insulin potentiating activity of plant extracts in 3T3-L1
bud (GB) MeOH extract in the absence (0 nmol/L) and presence (1
adipocytes
and 100 nmol/L) of insulin. Bars represent the mean ± SEM of
n=9 samples from 3 independent experiments. *p<0.05; **p<0.01;
The ability of plant extracts to potentiate insulin-
***p<0.0001 vs. insulin (control).
stimulated glucose uptake was assessed at insulin concen-
trations of 1 and 100 nmol/L (shaded middle and right-
leaves at least once during the week, and about a fifth had
most group of bars, respectively in Figures 1-3). Guava
used noni (part not specified). Urban and rural sex differ-
bud extract displayed significant insulin potentiating ac-
ences in noni use differed in that slightly more women in
tivity at 200 ?g/mL, increasing glucose uptake by a factor
Koki had used noni (8.2 %), while in rural Wanigela,
of 2.5 compared to insulin alone at 1 nmol/L and 1.6
males were more frequent users (11.7 %).
times greater than at 100 nmol/L (Figure 1). No insulin

potentiating effect was observed for any of the noni ex-
tracts (Figure 2a-d), and in the case of NL and NR 200
Insulin-like activity of plant extracts in 3T3-L1 adipo-
?g/mL inhibited glucose uptake. A potent insulin-
cytes
The ability of plant extracts to stimulate glucose uptake in
potentiating effect was observed for uncooked MBR,
3T3-L1 adipocytes was first assessed in the absence of
where 50 ?g/mL produced a 140% increase in glucose
insulin (unshaded left-most group of bars in Figures 1-3).
transport at an insulin concentration of 1 nmol/L and a
Of the extracts tested, GB had the highest insulin-like
45% increase at 100 nmol/L (data not shown). The
activity, increasing glucose uptake in a dose-dependent
cooked form of mangrove bean however had no effect on
manner up to more than 150% at a concentration of 200
insulin-mediated glucose uptake (data not shown), further
?g/mL (Figure 1). All noni part extracts, except for the
indicating a participation of tannins in MBR bioactivity.
root, had significant insulin-like activity at all concentra-
Areca nut (AN) extract reduced insulin’s action on glu-
tions; the commercially prepared fruit juice having had
cose uptake in 3T3-L1 adipocytes in a dose-dependent
the largest effect (92% increase from baseline at 100
manner (Figure 3a) while all tested concentrations of PBI
?g/mL) (Figure 2). In contrast, the root had no effect at
inhibited insulin-mediated glucose uptake to below base-

A
B
e
n
t
r
o
l
)
ne
n
t
r
o
l
)
n
t
r
o
l
)
li
co
co
co
se
s
e
l
i
n
a

b
o
n
ly

ba
o
n
ly
m
m
i
cle-
r
o
i
cle-
i
cle-
f

f
r
o
ge
t
o
veh
an
t
o
veh
C
e
v
e
D
Change
v
e
Change
Ch
ativ
%
%
ativ
%
elati
(Rel
(R
(Rel
Insul
u in
i n
n mo
m l
o /L
Insul
u in n
i
mo
m l
o /L

Figure 2. Percent change in glucose uptake from vehicle (DMSO)-treated cells in 3T3-L1 adipocytes treated with A: noni fruit (NF), B:
commercial noni fruit juice (NJ), C: noni leaf (NL), and D: noni root (NR) MeOH extracts in the absence (0 nmol/L) and presence (1 and
100 nmol/L) of insulin. Bars represent the mean ± SEM of n=9 samples from 3 independent experiments. *p<0.05; **p<0.01; ***p<0.0001
vs. insulin (control).

---

640
PL Owen, LC Martineau, D Caves, PS Haddad, T Matainaho and T Johns
Table 3. Spearman’s correlation coefficients for the relationship between intake of betel quid, guava and mangrove
bean, and fasting blood glucose, diabetes status and selected anthropometric and vascular antecedents.


Betel quid (g/wk)
Guava fruit (g/wk)† Mangrove
bean
(g/wk)
Fasting blood glucose (mmol/L)
-0.154**
-0.213***
0.137*
Type 2 diabetes‡ -0.111*
-0.116*
0.047
ns
Standardized ?§
-0.01 ns -0.120*
0.140*
Weight (kg)
0.273***
0.092ns -0.315***
Stature (m)
0.409***
0.233***
-0.397***
Body Mass Index (kg/m2)
0.101*
-0.040 ns -0.149**
Waist circumference (cm)
0.109*
-0.032 ns -0.190***
% Body fat
-0.08 ns -0.109*
-0.023
ns
Tricep skinfold thickness (cm)
0.176**
-0.008 ns -0.154**
Systolic blood pressure (mm Hg)
-0.081 ns -0.091
ns 0.071
ns
Diastolic blood pressure (mm Hg)
-0.184***
-0.132*
-0.190***

†Guava bud tea intake data was supplemented with guava fruit because of infrequent intake.
‡Nondiabetic=0; diabetic=1.
§Continuous dependent variable fasting blood glucose was normalized by inverse transformation (100-1/p) and food intake was entered in
multivariable analysis with age, weight and waist circumference as covariables.
*p<0.05; **p<0.005; ***p<0.0001; ns =non-significant
line levels (Figure 3b). When considered together in the
BQ admixture, glucose uptake gradually declined with

increased dose (Figure 3c). With the exception of MBR, a
A
significant interaction was observed in all cases between
extract concentration and insulin dose response (data not
shown).

Association between plant intake and population health
parameters
Associations between the frequency and amount of plant
ingested and selected health parameters were assessed
using Spearman’s correlation for non-normally distrib-
B
uted data (Table 3). Since medicinal plants were not in-
e

k
gested as frequently as food plants, but rather used to re-
t
a
n
t
r
o
l
)

p
co
lieve symptoms as they arose, correlations could not be
e
u
o
n
ly
c
os
reliably calculated for GB and noni. As an alternative,
u
e-

gl
guava bud tea intake data was replaced with frequency
e

i
n

v
e
h
i
cl
g
data of guava fruit, and this was found to have an inverse
an
e to
h
iv
at
correlation with FBG (r=-0.21, p<0.0001) and prevalence
C
el
%
(
R
of DM2 (r=-0.12, p<0.05). This relationship remained
significant when FBG was inverse-transformed and con-
trolled for age, central adiposity and weight (?=-0.117,
C
p=0.024). A simple inverse correlation between BQ and
FBG and DM2 was also observed (r=-0.15, p<0.005; r=-
0.11, p<0.05, respectively), but this disappeared when
covariates were included (?=-0.-014, p=0.787). In con-
trast, a significant relationship between MBC and FBG
was observed (r=0.14, p<0.05) and remained significant
when controlled for age, weight and central adiposity
(?=0.139, p=0.010). There was no correlation between
MBC consumption and DM2 prevalence.
Insulin nmol/L
Mangrove bean consumption was associated with

lighter body weight, shorter stature, smaller BMI, a thin-
Figure 3. Percent change in glucose uptake from vehicle (DMSO)-
ner waistline and less subcutaneous fat as assessed by
treated cells in differentiated 3T3-L1 adipocytes treated with com-
TSF thickness. In contrast, all these parameters increased
ponents of betel quid. A: areca nut (AN), B: P betle inflorescence
significantly with BQ consumption. Those who consumed
(PBI) and C: betel quid (BQ) MeOH extracts in the absence (0
nmol/L) and presence (1 and 100 nmol/L) of insulin. Bars repre-
more guava fruit tended to have less % body fat (r=-0.11,
sent the mean ± SEM of n=9 samples from 3 independent experi-
p<0.05) and lower DBP (r=-0.13, p<0.05), although the
ments. *p<0.05; **p<0.01; ***p<0.0001 vs. insulin (control).
latter was more strongly correlated with BQ (r=-0.18,
Bars higher than the control indicate insulin-like or insulin-
p<0.0001) and MBC (r=-0.19, p<0.0001) intake.
potentiating activity; lower bars indicate inhibition of glucose
uptake.

---

Guava and noni protects against diabetes
641

DISCUSSION
guava may be gallic acid since this common phenolic
Despite the social perturbations affecting all residents of
molecule has been reported to stimulate glucose uptake
Papua New Guinea, relatively high rates of adiposity, as
and enhance insulin sensitivity by activating peroxisome
well as habitual BQ utilization, the low incidence of DM2
proliferator-activated receptor (PPAR)-?.23 In addition,
in Kalo residents suggests the mediation of protective
guava fruit may exert indirect benefits through its rich
environmental factors. Our findings indicate that con-
antioxidant and soluble fibre content.24 A variety of other
sumption of certain plants, in particular guava and noni,
bioactive compounds found in guava leaves including
may confer anti-diabetic effects via enhanced insulin-
essential oils, saponins, and flavonoids, could contribute
independent and insulin-dependent glucose uptake despite
to an antidiabetic effect, requiring further research towards
representing only a minor component of the diet. While
identification of responsible agents.
only insulin-mimetic and insulin-sensitizing activities
Noni, perhaps the most popular herbal medicine in the
were tested, plant products may exert additional antidia-
Pacific Islands, displayed decent insulin-mimetic activity,
betic properties through other biological mechanisms
especially in the form of the commercial fruit juice (92.0
such as increased insulin secretion, reduced glucose ± 22.3 % increase at 100?g/mL compared to basal glu-
production, and reduced glucose absorption, as well as cose intake levels). None of the noni extracts however,
cytoprotective, antioxidant, anti-glycation, and anti hy-
were able to stimulate insulin-mediated glucose uptake,
pertensive effects.8
and at 200 ?g/mL, root extract inhibited glucose uptake.
Betel quid chewing is an integral component of Papua
This bell-shaped pattern could help explain a report for a
New Guinean culture, particularly in strengthening social
related species, Morinda officinale, where the ethanolic
bonds between individuals and clans. Although the use of
root extract displayed both hypo- and hyperglycemic ef-
BQ as an herbal therapy should not be surprising given its
fects in STZ-diabetic mice depending on the dose.25 Noni
socio-cultural importance, a growing trend of prescribing
is known to be rich in anthraquinones, especially the roots,
it as a treatment for DM2 by some traditional medical
and studies suggest that these may be anti-diabetic bioac-
practitioners is alarming considering its association with
tive compounds.26
increased DM2 risk.5 Areca nut, the main component of
Cooked mangrove bean extract lost roughly 80% of to-
BQ, is believed to promote DM2 by damaging islet ?-
tal phenol content and 65% of insulin-like activity com-
cells via the unstable free-radical generating nitrosated
pared to the raw plant extract. Nevertheless, a modest yet
derivates of its alkaloids.15 This is supported epidemi-
significant insulin-mimetic effect suggests that cooked
ologically by a 1995-1999 survey in Papua New Guinea
mangrove bean could be of some benefit in preventing
which reported BQ chewing as an independent predictor
DM2. This, however, was not supported in the present
of high fasting capillary blood glucose (?=1.032, p=0.005),
study since our survey suggests a positive correlation
surpassing the effect of age (?=0.554, p=0.028), BMI
between MBC frequency intake and FBG despite an oth-
(?=0.507, p=0.061) and region of origin (?=-1.582,
erwise favourable association with weight, waist circum-
p=0.056).5 Considering that BQ chewing has been prac-
ference, TSF, and DBP. That DM2 had only been de-
ticed for centuries without any association with DM2, a
tected in the village from a 1986 survey,27 concurrent
relationship may be justified if the myriad of dietary,
with the introduction of non-staple items such as polished
economic and social confounders related to moderniza-
rice, refined flour, sugar and edible oils and not in earlier
tion is considered. A previous study suggests that BQ
diabetes surveys dating from 1977,28 suggests the associa-
chewing may indirectly increase susceptibility to DM2 by
tion between MBC and FBG is due to other factors that
promoting weight gain and increased waist size.2 Such a
have statistical collinearity with residence in Wanigela,
relationship has been observed in the present study with
thereby limiting our ability to distinguish the effect of a
weight (r=0.27, p<0.0001), BMI (r=0.10, p<0.05) and
single food item on chronic disease development.
waist circumference (r=0.11, p<0.05), despite an inverse
Medicinal plants constitute only a minor component of
correlation with DM2 (r=-0.11, p<0.05). However, the
total dietary intake and habitual food habits clearly have a
latter was rendered non-significant when age, weight and
greater influence on non-communicable disease risk.
central obesity were accounted for in the analysis (?=-
Dietary analysis showed that Wanigela residents con-
0.014, p=0.787). Our cell culture experiments suggest a
sumed significantly less rice than Kalo and as vegetarians,
novel diabetogenic mechanism for BQ whereby its indi-
abstained from meat. There were no significant differ-
vidual components inhibit the effects of insulin (Figure
ences in the amount of fruits and vegetables consumed
3c). Inhibition of insulin-mediated glucose uptake in 3T3-
between the communities except for guava which was
L1 adipocytes may be caused by the cytotoxic nature of
consumed in greater quantities in Kalo.29 The impact of
PBI observed here at concentrations as low as 25 ?g/mL
consuming the plants included in the present study on
and also observed in other cell lines.12,16-18
metabolic health is unknown when considered holistically
Guava bud extract displayed particularly strong insu-
with diet and physical activity, although it is plausible
lin-like and insulin-potentiating activity in our study. that they may have an influence on nutrient metabolism
Assuming good bioavailability, this plant product may be
and long-term health. Our next objective is to assess this
useful for the prevention and treatment of DM2. Indeed,
interaction in a study designed specifically to test whether
antidiabetic effects of guava leaf have been reported in
guava and noni consumption could effectively counter
genetically19 and chemically-induced20,21 animal models
DM2 and betel quid diabetogenicity.
of DM2. Guava juice was also reported to be hypoglyce-

mic in healthy and alloxan diabetic mice as well as in
CONCLUSION
healthy and diabetic humans.22 A bioactive constituent of

---

642
PL Owen, LC Martineau, D Caves, PS Haddad, T Matainaho and T Johns
The present study is a first step in determining whether
9. Thomson MJ, Williams MG, Frost SC. Development of
consumption of particular plants can protect against DM2
insulin resistance in 3T3-L1 adipocytes. J Biol Chem. 1997;
risk or its comorbidities in light of the risks associated
272:7759-64.
with nutritional transition and those purported to be in-
10. Carlson TJ. Ethnobotanical Field Research Manual for Non-
curred by chewing BQ. Of the plants that form Kalo’s
Insulin Dependent Diabetes Mellitus. South San Francisco,
traditional food and medical systems, guava bud and noni
CA: Shaman Pharmaceuticals, 1995.
11. WHO. Definition and diagnosis of diabetes mellitus and
stood out as being distinctive from those of rural Wani-
intermediate hyperglycemia: report of a WHO/IDF consul-
gela. Using cultured 3T3-L1 adipocytes as a model of
tation. Geneva: WHO, 2006:50.
insulin resistance, we demonstrated that guava bud and
12. Owen PL, Matainaho T, Sirois MG, Johns T. Endothelial
noni extract possessed potent insulin-mimetic activity, the
cytoprotection from oxidized LDL by some crude Melane-
former also having effective insulin-potentiating activity.
sian plant extracts is not related to their antioxidant capacity.
Habitual ingestion of these may protect against BQ diabe-
J Biochem Mol Toxicol. 2007;21:231-42.
togenicity or DM2 associated with socioeconomic transi-
13. Martineau LC, Couture A, Spoor D, Benhaddou-Andaloussi
tion. A large prospective study that controls for con-
A, Harris C, Meddah B, et al. Antidiabetic properties of the
founders is required to reliably discern any dietary asso-
Canadian lowbush blueberry Vaccinium angustifolium Ait.
ciation with lifestyle-related diseases, especially for in-
Phytomedicine. 2006;13:612-23.
frequently consumed plants such as guava and noni. 14. Vats V, Grover JK, Rathi SS. Evaluation of antihypergly-
cemic and hypoglycemic effect of Trigonella foenum-
Likewise, different experimental models, particularly
graecum Linn, Ocimum sanctum Linn and Pterocarpus
animal studies and clinical studies are needed to more
marsupium Linn in normal and alloxanized diabetic rats. J
accurately explore antidiabetic potential and mechanisms
Ethnopharmacol. 2002;79:95-100.
of action for these plants.
15. Boucher BJ, Mannan N. Metabolic effects of the consump-

tion of Areca catechu. Addiction Biol. 2002;7:103-10.
ACKNOWLEDGEMENT
16. Jeng JH, Kuo ML, Hahn LJ, Kuo MYP. Genotoxic and non-
Support for this study was received from the Natural Science
genotoxic effects of betel ingredients on oral mucosal fibro-
and Engineering Research Council of Canada (PGS D for PLO
blasts in vitro. J Dent Res. 1994;73:1043-9.
and the Discovery Grant for TJ); and the International Devel-
17. Jeng JH, Hahn LJ, Lin B-R, Hsieh CC, Chan CP, Chang
opment Research Council of Canada: Ecosystems Approaches
MC. Effects of areca nut, inflorescence Piper betle extracts
to Human Health Training Award. The authors are indebted to
and arecoline on cytotoxicity, total and unscheduled DNA
Ricky Guria, Henry Urai, Perry Otio and Bremiche Valina for
synthesis in cultured gingival keratinocytes. J Oral Pathol
assisting with the collection of FFQ and plant samples. Osia
Med. 1999;28:64-71.
Gideon from the UPNG herbarium aided in plant identification
18. IARC. Betel-quid and Areca-nut Chewing and Some Areca-
and facilitated exportation of plant material.
nut-derived Nitrosamines. Lyon: IARC Press, 2004.

19. Oh WK, Lee CH, Lee MS, Bae EY, Sohn CB, Oh H, Kim
AUTHOR DISCLOSURES
BY, Ahn JS. Antidiabetic effects of extracts from Psidium
The authors declare that there are no conflicts related to the
guajava. J Ethnopharmacol. 2005;96:411-5.
publication of this study.
20. Ojewole JAO. Hypoglycemic and hypotensive effects of

Psidium guajava Linn. (Myrtaceae) leaf aqueous extract.
REFERENCES
Methods Finds Exp Clin Pharmacol. 2005;27:689-95.
1. Popkin BM, Gordon-Larsen P. The nutrition transition:
21. Mukhtar HM, Ansari SH, Ali M, Naved T, Bhat ZA. Effect
worldwide obesity dynamics and their determinants. Int J
of water extract of Psidium guajava leaves on alloxan-
Obes. 2004;28:S2-S9.
induced diabetic rats. Pharmazie. 2004;59:734-5.
2. Mannan N, Boucher BJ, Evans SJW. Increased waist size
22. Cheng JT, Yang RS. Hypoglycemic effect of guava juice in
and weight in relation to consumption of Areca catechu (be-
mice and human subjects. Am J Chin Med. 1983;11:74-6.
tel-nut); a risk factor for increased glycaemia in Asians in
23. Huang THW, Peng G, Kota BP, Li GQ, Yamahara J, Rou-
East London. Br J Nutr. 2000;83:267-77.
fogalis BD, Li Y. Anti-diabetic action of Punica granatum
3. Boucher BJ, Ewen SWB, Stowers JM. Betel nut (Areca
flower extract: activation of PPAR-? and identification of
catechu) consumption and the induction of glucose intoler-
an active compound. Toxicol Appl Pharmacol. 2005;207:
ance in adult CD1 mice and in their F1 and F2 offspring.
160-9.
Diabetologia. 1994;37:49-55.
24. Singh RB, Rastogi SS, Singh NK, Ghosh S, Gupta S, Niaz
4. Tung T-H, Chiu Y-H, Chen L-S, Wu H-M, Boucher BJ,
MA. Can guava fruit intake decrease blood pressure and
Chen TH-H. A population-based study of the association
blood lipids? J Hum Hypertens. 1993;7:33-8.
between areca nut chewing and Type 2 diabetes mellitus in
25. Soon YY, Tan BKH. Evaluation of the hypoglycemic and
men (Keelung Community-based Integrated Screening pro-
antioxidant activities of Morinda officinalis in streptozoto-
gramme No. 2). Diabetologia. 2004;47:1776-81.
cin-induced diabetic rats. Singapore Med J. 2002;43:77-85.
5. Benjamin AL. Community screening for diabetes in the
26. Abe D, Saito T, Sekiya K. Sennidin stimulates glucose in-
National Capital District, Papua New Guinea: is betelnut
corporation in rat adipocytes. Life Sci. 2006;79:1027-33.
chewing a risk factor for diabetes? P N G Med J. 2001;44:101-7.
27. Patel M, Jamrozik K, Allen O, Martin FIR, Eng J, Dean B.
6. Dowse GK. Betel-nut chewing and diabetes in Papua New
A high prevalence of diabetes in a rural village in Papua
Guinea and elsewhere. Diabetologia 1994;37:1062-1063.
New Guinea. Diabetes Res Clin Pract. 1986;2:97-103.
7. Dowse GK, Spark RA, Mavo B, Hodge AM, Erasmus RT,
28. Martin FIR, Wyatt GB, Griew AR, Mathews JD, Campbell
Gwalimu M, Knight LT, Koki G, Zimmet PZ. Extraordi-
DG. Diabetic surveys in Papua New Guinea - results and
nary prevalence of non-insulin-dependent diabetes mellitus
implications. P N G Med J 1981;24:188-194.
and bimodal plasma glucose distribution in the Wanigela
29. Owen PL. Functional diversity of Indigenous diets in
people of Papua New Guinea. Med J Aust. 1994;160:767-74.
coastal Papua New Guinea: Role in the nutrition transition
8. Marles R, Farnsworth N. Antidiabetic plants and their ac-
and noncommunicable disease risk. School of Dietetics and
tive constituents. Phytomedicine. 1995;2:137-65.
Human Nutrition. Montreal: McGill University, 2008:244.

---

Guava and noni protects against diabetes
643

Original Article

Consumption of guava (Psidium guajava L) and noni
(Morinda citrifolia L) may protect betel quid-chewing
Papua New Guineans against diabetes

Patrick L Owen
1
2
1
PhD , Louis C Martineau PhD , Dayna Caves BSc ,
Pierre S Haddad
2
3
1
PhD , Teatulohi Matainaho PhD and Timothy Johns PhD

1School of Dietetics and Human Nutrition, Macdonald Campus of McGill University, Quebec, Canada
2Natural Health Products and Metabolic Diseases Laboratory, Department of Pharmacology, Université
de Montréal, Quebec, Canada
3Discipline of Pharmacology, School of Medicine and Health Sciences, University of Papua New Guinea,
Port Moresby, Papua New Guinea

?????(Psidium guajava L.)????(Morinda ci-
trifolia L.)???????????????????
???

??????(DM2)????????????????????????
????????????????????(BQ)???????????
??????????? DM2 ????????????????????
Kalo ???????????????????????? DM2 ?????
??????????? Wanigelans???????????? DM2 ??
??? Kalo ???????(Psidium guajava L.)????(????)(Morinda
citrifolia L.)????? Wanigela ??????????(Bruguiera gymnor-
rhiza (L.) Lam.)?????????????????????????-?
???(Areca catechu L.)????(Piper betle L.)??????? 3T3-L1 ??
????????????????????????????????
????????????????????????????????
?????????????????????????????????
?????????????????????????????????
?????????????????????????????????
??????????????????????? DM2 ????????
?????????????????????????????????
???????????? DM2 ????

?????????????3T3-L1 ????????

---