Journal Pre-proof

Optimization of natural sweeteners for sugar reduction in chocolate flavoured milk and
their impact on sensory attributes

Dipendra Kumar Mahato, Russell Keast, Djin Gie Liem, Catherine Georgina Russell,
Sara Cicerale, Shirani Gamlath

PII: S0958-6946(20)30292-2
DOI: https://doi.org/10.1016/j.idairyj.2020.104922
Reference: INDA 104922

To appear in: International Dairy Journal

Received Date: 27 June 2020

Revised Date: 4 November 2020
Accepted Date: 4 November 2020

Please cite this article as: Mahato, D.K., Keast, R., Liem, D.G., Russell, C.G., Cicerale, S., Gamlath, S.,
Optimization of natural sweeteners for sugar reduction in chocolate flavoured milk and their impact on
sensory attributes, International Dairy Journal, https://doi.org/10.1016/j.idairyj.2020.104922.

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.

© 2020 Elsevier Ltd. All rights reserved.

 1 Optimization of natural sweeteners for sugar reduction in chocolate flavoured
2 milk and their impact on sensory attributes
3
4 Dipendra Kumar Mahato*, Russell Keast, Djin Gie Liem, Catherine Georgina Russell, Sara
5 Cicerale and Shirani Gamlath
6 CASS Food Research Centre, School of Exercise and Nutrition Sciences, Deakin University, Burwood,
7 VIC 3125, Australia.
8
9 *Corresponding author.
10 Email addresses: dmahato@deakin.edu.au; kumar.dipendra2@gmail.com
11

12 ABSTRACT

13 Public Health bodies like the World Health Organization and Public Health England have

of
14 identified addition of sugar to processed foods and beverages as a health concern. Owing to the

ro
15 increasing consumption of sugar sweetened chocolate flavoured milk in Australia among all age

16
-p
groups, this study was conducted to investigate the possibility of achieving a 50% of added sugar
re
17 reduction in chocolate milk using natural non-nutritive sweeteners, namely stevia and monk fruit
lP

18 extract. Optimization of the sweeteners was successfully performed by Response Surface

na

19 Methodology. Variables for 50% reduction of added sugar in chocolate milk were chosen as stevia
ur

20 sweetener (5-100 ppm) and monk fruit extract (50-100 ppm) based on preliminary trials. Sensory

21 attributes (overall liking, appearance, aroma, sweetness, mouthfeel and aftertaste) were taken as
Jo

22 the responses. The process was optimized with a combination of 56.27 ppm of stevia sweetener

23 and 81.90 ppm of monk fruit extract with sensory attributes of 6.78, 6.47, 6.31, 6.47, 6.45 and

24 6.33, respectively for overall liking, appearance, aroma, sweetness, mouthfeel and aftertaste. Use

25 of the two selected sweeteners in combination resulted in sweetness synergy. In addition, the

26 bitter and metallic aftertastes of stevia were masked by the monk fruit extract and helped enhance

27 the overall sensory attributes of the product compared to the control. This study provides

28 insightful information for optimizing sweeteners for sugar reduction in milk-based product.

29

30

1
31 1. Introduction

32 Consumption of added sugars in the diet is associated with an increased risk of several

33 chronic diseases, including overweight, obesity and type 2 diabetes (Malik, Schulze, & Hu, 2006;

34 Te Morenga, Mallard, & Mann, 2013; Patel et al., 2018; Singh et al., 2020). Globally,

35 consumption of added sugar has tripled in the last 50 years (Lustig, Schmidt, & Brindis, 2012).

36 Major contributors to consumption of added sugar are sugar-sweetened beverages which include

37 soft drinks (e.g. soda or pop), fruit and sports drinks, tea, coffee, energy drinks, sweetened milk or

38 milk alternatives, as well as other beverages to which sugar (mainly sucrose, fructose or high

of
39 fructose corn syrup) has been added (Blake, Lancsar, Peeters, & Backholer, 2019; Cleghorn, et al.,

ro
40 2019; Falbe, Thompson, Patel, & Madsen, 2019).

41
-p
Milk is the principal dairy product consumed by children worldwide. Approximately, 60-
re
42 80% of American children’s dairy product consumption is comprised of fluid milk (FPED, 2014).
lP

43 The majority of this milk is flavoured milk (86% of all the milk) in the USA, and further, the

44 majority of flavoured milk available is chocolate (Nicklas, O'neil, & Fulgoni, 2017). In Australia,
na

45 per capita consumption of drinking milk decreased from 105.1 litres in 2014-15 to 98.6 litres in
ur

46 2018-19 due to growth of flavoured milk products and coffee (Dairy Australia, 2019), suggesting
Jo

47 that Australians are increasingly incorporating flavoured milk in their diets. Flavoured milk is

48 considered a core food in the Australian Dietary Guidelines (Brownie, Muggleston, & Oliver,

49 2015) and is used to promote milk intake to ensure that children meet the recommended dietary

50 allowances (RDA) for vitamin D and calcium (Patel et al., 2018). Children’s dairy consumption is

51 typically lower than recommended in the Australian Dietary Guidelines (ADGs) and flavoured

52 milk might help to address this because children tend to consume more flavoured milk than plain

53 milk (Hanks, Just, & Wansink, 2014; Henry et al., 2015).

54

2
55 However, regular consumption of sweetened flavoured milk has been reported to increase

56 energy intake more than 10% compared to non-consumers (Krachler et al., 2006; Ludwig,

57 Peterson, & Gortmaker, 2001; Striegel-Moore et al., 2006) with the increased energy intake linked

58 to overweight, obesity and type 2 diabetes (Malik et al., 2006; Te Morenga et al., 2013; Patel et

59 al., 2018; Singh et al., 2020). The World Health Organization (WHO) recommends less than 10%

60 of total energy intake should come from free sugar and is in line with Public Health England

61 (PHE) recommendation of least a 20% sugar reduction by 2020 in processed foods and beverages

62 (PHE, 2017). Therefore, sugar reduction in chocolate milk is desired to maximise the health

63 benefits of chocolate milk consumption, specifically calcium. The vectors rule of a well-tested

of
64 model of an epidemiological triad (hosts, vectors, and environments) suggests ‘small changes ×

ro
65 large volumes = significant population benefits’ (Egger, Swinburn, & Rossner, 2003; Keast,

66
-p
Sayompark, Sacks, Swinburn, & Riddell, 2011). Even a small reduction can significantly benefit a
re
67 larger population in long terms (Mahato et al., 2020).
lP

68
na

69 Sweet taste is a driver of liking for many foods such as chocolates (de Melo, Bolini, &
ur

70 Efraim, 2009), yogurt and yogurt-like products (Bayarri, Carbonell, Barrios, & Costell, 2011;

71 Thompson, Lopetcharat, & Drake, 2007) along with chocolate milk (Li, Lopetcharat, & Drake,
Jo

72 2015). Sugar reduction greater than 30% affects the liking of dairy products including flavoured

73 yogurt (Chollet, Gille, Schmid, Walther, & Piccinali, 2013) and chocolate milk (Li et al., 2015).

74 Thus, finding a suitable sugar substitute to increase sugar reduction while maintaining sweet taste

75 and liking is important. Reduction of sugar without consumers’ noticing could be the most

76 effective strategy to maintain acceptance, however, this is difficult to achieve as sugar has

77 multiple sensory influences on a product, not just sweetness. This is the challenge for the food

78 industry.

79

80

3
 81 Reduction of sugar and its impact on consumer acceptance can be overcome by the use of

82 high intensity sweeteners. Most previous studies have used artificial non-nutritive sweeteners

83 (NNS) like advantame, neotame, sucralose, or aspartame in dairy-based foods and beverages to

84 replace sucrose (Morais, Morais, Cruz, & Bolini, 2014). However, consumer trends suggest a

85 desire for healthier products with “all-natural” ingredients and therefore, products with natural

86 sweeteners are preferred over artificial sweeteners (Christoph et al., 2011; Li, Lopetcharat, &

87 Drake, 2014). For example, Li et al. (2014) found parents preferred to buy chocolate flavoured

88 milk with added natural NNSs or sucrose over the artificial NNSs for their children. Further, Li et

89 al. (2015) reported parents’ and children's acceptance of naturally sweetened chocolate milk.

of
90 Similarly, Oltman, Lopetcharat, Bastian, & Drake (2015) found general consumers’ (18-70 years)

ro
91 preference of “naturally sweetened” labels for protein beverages. Natural NNSs are derived from

92
-p
plants and comprise of natural compounds (Kim & Kinghorn, 2002), therefore, use of natural
re
93 NNSs like stevia and monk fruit could have better consumer acceptance. Also, an enzymatic
lP

94 treatment (lactase) can increase the sweetness and reduce sugar (up to 50%) which could be an
na

95 alternative to reduce sugar in a natural way in flavoured milk apart from the use of natural
ur

96 sweeteners (arlafoods.co.uk).

97 Stevia (Stevia rebaudiana) belongs to the Asteraceae family. The sweetness of stevia is due
Jo

98 to sweet glycosides, stevioside and different types of rebaudiosides (e.g. Reb A, Reb B, Reb D,

99 Reb E, Reb M, etc.) present in stevia leaves (Parker, Lopetcharat, & Drake, 2018). Stevia is

100 commonly used in dairy products such as yogurt and ice cream, baked goods and soft drinks

101 (Guggisberg, Piccinali, & Schreier, 2011; Ozdemir, Arslaner, Ozdemir, & Allahyari, 2015; Pon,

102 Lee, & Chong, 2015). Steviol glycosides are noncarcinogenic but consumption more than 4 mg

103 kg−1 body weight day−1 is unsafe and may change the composition of the gut microbiota (EU

104 regulation 1129/2011) (EFSA, 2010, 2015; Ruiz-Ojeda, Plaza-Díaz, Sáez-Lara, & Gil, 2019).

105 Further, rebaudiosides like Reb A and Reb M can be used in food and beverages for human

4
106 consumption in Australia and New Zealand at the same permitted levels as steviol glycosides

107 (Australia and New Zealand Food Standards Code, 2018).

108 Monk fruit (Siraitia grosvenorii) also known as Luo Han Guo (DuBois & Prakash, 2012;

109 Pawar, Krynitsky, & Rader, 2013), is a fruit native to southern China, that contains sweet

110 glycosides mogroside IV, mogroside V, and mogroside VI (Pawar et al., 2013). Monk fruit has

111 several health beneficial effects such as anti-cancer (Liu et al., 2016), antioxidant (Chen, Wang,

112 Qi, & Xie, 2007), anti-inflammatory (Chun et al., 2014), anti-obesity (Sun et al., 2012) and anti-

113 diabetic properties (Lee, Jeong, Kim, & Nam, 2016; Qi, Chen, Zhang, & Xie, 2008). In a recent

114 study by Ban et al. (2020), rats fed symbiotic yoghurt sweetened with monk fruit extract showed

of
115 greater blood regulation and a significant reduction in insulin resistance and glycosylated

ro
116 haemoglobin compared to rats fed yoghurt sweetened with sucrose. Further, monk fruit is a low

117
-p
cost and high-intensity natural sweetener that can be utilized in food and beverage industries for
re
118 the development of low calories products for diabetics and health-conscious consumers (Pandey
lP

119 & Chauhan, 2019). The ADI has not been established for monk fruit extract because no adverse
na

120 effects have been reported, however, the ADI of monk fruit juice concentrate is 25 mg kg−1 body

weight day−1 (DuBois & Prakash, 2012).

ur

121

122 To achieve sugar reduction in chocolate milk, some studies have tried reformulation. For
Jo

123 example, Rad, Delshadian, Arefhosseini, Alipour, & Jafarabadi (2012) used stevia as a sugar

124 replacer along with inulin as a thickening agent to evaluate the physical properties (sedimentation

125 and viscosity) of chocolate milk. Li et al., (2015) used monk fruit extract and stevia leaf extract

126 separately to reduce sugar in skim chocolate milk and studied the impact on consumer acceptance.

127 Similarly, Bordi Jr, Palchak, Verruma-Bernardi, & Cho (2016) used Reb A stevia to partially

128 substitute for sugar in chocolate milk. Further, Azami, Niakousari, Hashemi, & Torri (2018) used

129 liquorice extract as a sugar substitute in chocolate milk and studied consumer acceptance and

130 physicochemical properties. All these studies used a single sweetener for sugar reduction.

131 However, as no single sweetener has similar functionality to sucrose, the use of two or more

5
132 sweeteners as a blend may provide flavour and taste profiles similar to sucrose. For example, the

133 combined effect of stevia and sucralose improved sensory and physical properties of a sugar-free

134 dairy dessert (Furlán & Campderrós, 2017). Similarly, blending Reb M with Reb B/ Reb D

135 resulted in sweetness synergy (Prakash, Markosyan, & Bunders, 2014). No such experiment

136 involving two sweeteners and the effect of their interactions on the sensory attributes has been

137 reported in chocolate milk for sugar reduction. Therefore, this work aims to optimize two natural

138 NNSs (stevia sweetener containing steviol glycosides, Reb A and stevioside, and monk fruit

139 extract containing mogroside V) for sugar reduction in chocolate flavoured milk for optimum

140 sensory attributes and consumer liking.

of
ro
141 The novelty of this paper is its focuses on two sweeteners (stevia and monk fruit extract) of

142 -p
plant origin and introduces a Response Surface Methodology (RSM)-based approach to optimize
re
143 these sweeteners for sugar reduction in a milk-based product with optimum sensory attributes and
lP

144 consumer liking. The article would be a milestone for the dairy industry to reduce the sugar

145 content of their products in order to meet the demands and pressure from Public Health bodies and
na

146 consumers. Furthermore, as this article provides an insightful approach for optimizing sugar
ur

147 reduction in milk-based drinks, it would also be considered for its optimization methodology.
Jo

148 2. Materials and methods

149 2.1. Raw materials

150 Coles cocoa powder (natural, 100% pure cocoa; 23.3% w/v, protein; 11.0% w/v, fat; and

151 18.7% w/v, carbohydrate), CSR white sugar (refined) and Australian homogenised and

152 pasteurised milk (3.5% w/v, protein; 3.3% w/v, fat; and 4.6% w/v, carbohydrate) were purchased

153 from a supermarket near Deakin University, Burwood, Australia. Starch (NOVATIONTM 3300)

154 was supplied by Ingredion, Westchester, city, IL, USA while the stabilizer (Ticaloid® 750

155 containing carrageenan standardized with maltodextrin) was provided by TIC GUMS,

156 Philadelphia, Ph, USA. Natural non-nutritive sweeteners namely Stevia (TASTEVA® containing

6
157 steviol glycosides, Reb A and stevioside) and Monk fruit extract (PUREFRUITTM containing

158 mogroside V) were made available by TATE & LYLE, London, UK.

159 2.2. Sample preparation

160 The chocolate flavoured milk was prepared as per the process described by Ingredion

161 Singapore Pte Ltd, depicted in Fig. 1. The concentrations of the stabilizer (Ticaloid®750) and the

162 starch (NOVATIONTM 3300) were fixed to 0.02% and 0.5%, respectively based on pre-trials

163 performed to match a commercial reference sample. The stabilizer provides thermostability while

164 the starch improves the rheology or mouthfeel properties of the product (Agoda-Tandjawa, Le

of
165 Garnec, Boulenguer, Gilles, & Langendorff, 2017). The total sugar concentration of commercial

ro
166 chocolate milk in the Australian supermarket varies from 4-12% (including both reduced and full-

167
-p
sugar versions). Our previous consumer study on preference mapping (unpublished data) of ten
re
168 commercial chocolate milk samples with this sugar concentration range using 235 participants
lP

169 (19-63 years, 82.98% female, 65.53% Australian, 24.26% Asian and remaining from other

170 nations), identified OAK CHOCOLATE milk and OAK THICK DEATH BY CHOCOLATE milk
na

171 (Parmalat Food Products PTY LTD, Australia) with sugar concentrations 10.6% and 10.9%,
ur

172 respectively, to have the highest appeal satisfying between 70-80% of consumers assessed.
Jo

173 Therefore, the lower limit (round figure, 10%) was considered as the control point for sugar

174 reduction with OAK chocolate milk as a commercial reference sample in this study.

175 To formulate the control sample, 5% sucrose (white sugar) was added to obtain the final

176 10% total sugar. The remaining 5% sugar was contributed by other ingredients (milk and cocoa

177 powder) in the chocolate milk as calculated by FoodWorks (Xyris Software, Australia Pty Ltd).

178 The software calculates the final nutrient composition of the product based on the yield, nutrient

179 composition and proportion of individual ingredients added to the product. The control sample

180 was prepared to match the commercial reference (OAK CHOCOLATE milk as determined from

181 our previous study) in terms of the sensory perception (overall liking, 6.50; appearance liking,

7
182 6.37; aroma liking 6.34; sweetness liking, 6.17; mouthfeel liking, 6.30; and aftertaste liking, 3.11)

183 and other physicochemical properties (pH, 6.73; total soluble solids, 20.5 °Brix; viscosity, 7.50

184 mPa.s; and water activity, 0.94) as benchmarking trials by 6-10 participants in CASS sensory

185 laboratory. To reduce sugar in chocolate flavoured milk, Stevia sweetener (TASTEVA®) and

186 Monk fruit extract (PUREFRUITTM) were used.

187 2.3. Optimization of natural non-nutritive sweeteners for sugar reduction

188 In the initial trials to decide the concentrations of sweeteners, a single sweetener (stevia

189 sweetener or monk fruit extract) was compared to formulations with combinations of these

of
190 sweeteners. Different concentrations (low, middle and high) of stevia sweetener and monk fruit

ro
191 extract between 5-200 ppm (within the ADI limits for safe consumption) were used for

192
-p
formulations with 20, 50, 80 and 100% reduction of added sugar as preliminary trials, taking
re
193 chocolate milk with 10% total sugar as the control sample. It was found that an individual
lP

194 sweetener needed a higher concentration to achieve a sweetness level compared to the two in

195 combination. Also, other properties like mouthfeel and aftertaste were better when the two
na

196 sweeteners were used together which would account for their synergistic effects. Monk fruit
ur

197 extract showed a better sensory profile than stevia. Therefore, for further refining, the
Jo

198 concentration of stevia was reduced (5-100 ppm) while monk fruit extract (50-100 ppm) was

199 increased. The preliminary trials were repeated using the low, middle and high values of these

200 sweeteners in combination for 20, 50, 80 and 100% reduction of added sugar.

201 To determine maximum added sugar reduction that could be achieved without much

202 impact on sensory perception compared to the control chocolate milk (10% total sugar), 6-10

203 participants (25-60 years, 50% female, experts in sensory science and product development) were

204 used to evaluate the reformulations. One participant at a time was provided with samples from

205 low to high concentration and feedback was noted for sensory perception (sweetness, appearance,

8
206 colour, flavour, mouthfeel and aftertaste). Participants were allowed to rinse their mouth between

207 samples as well as re-taste the samples and the control.

208

209 Based on feedback, 50% of added sugar reduction was deemed possibly with lower and

210 upper limits of Stevia sweetener (TASTEVA®, 5 and 100 ppm, respectively) and Monk fruit

211 extract (PUREFRUITTM, 50 and 100 ppm, respectively) and considered for the optimization by

212 RSM experimental design software. Other ingredients like cocoa powder (approx. 20% solution),

213 white sugar (sucrose, 5%), starch (0.5%) and stabilizer (0.02%) were kept fixed for all the 13

of
214 reformulations obtained by RSM design as described in section 2.5. (Experimental design and

ro
215 data analysis). These reformulated samples were taken for sensory evaluation.

216
-p
re
217 2.4. Sensory evaluation
lP

218 2.4.1. Sample size and inclusion/exclusion criteria

na

219 Typically, 100 consumers are required for sensory evaluation using a hedonic scale, which
ur

220 is generally representative of the target market or current users (Kemp, Hollowood, & Hort,
Jo

221 2009). Therefore, 120 participants were recruited allowing a 20% dropout rate. To be included,

222 participants were required to be above 18 years, consume milk and flavoured milk at least once

223 per week, be without diabetes and/or allergies or sensitivities. All the interested candidates

224 provided informed consent before participation. The consumer study was performed as per the

225 approved ethics (HEAG-H 155_2018).

226

227 2.4.2. Recruitment

228 Participants were recruited using two strategies. First, members of the CASS consumer

229 database (n=2000+) were contacted via email through Compusense Cloud software (Compusense

9
230 Inc., Guelph, Ontario, Canada). Members of the CASS consumer database are members of the

231 general community who were recruited via social media, flyers, and posters in the greater

232 Melbourne area. Second, recruitment was conducted in areas surrounding and including Deakin

233 University, Burwood campus utilising social media, flyers, posters and faculty emails. Interested

234 candidates were screened by asking them to complete a short online questionnaire through

235 Compusense Cloud software (Compusense Inc., Guelph, Ontario, Canada) which included

236 inclusion and exclusion criteria mentioned in the earlier section.

237

of
238 2.4.3. Consumer study

ro
239 Among the 120 participants recruited considering a 20% dropout rate, 107 participants (43

240
-p
male and 64 female) aged between 20 and 65 years old evaluated the chocolate milk samples. The
re
241 reformulated chocolate flavoured milk with varying concentrations of sweeteners (Table 1) were
lP

242 evaluated for sensory acceptance. All the reformulations were prepared fresh (as depicted in Fig.
na

243 1) on the day of evaluation and stored at 4°C to ensure the safety of the product.

244 Panelist first evaluated samples for overall liking using 9-Point Hedonic Scale ranging from
ur

245 “1-Dislike Extremely” to “9-Like Extremely”. Second, evaluation focused on the attributes
Jo

246 (namely appearance, aroma, sweetness, mouthfeel and aftertaste) using the same 9-Point Hedonic

247 Scale. Samples (15-20 ml) were presented in cups with three-digit codes in randomised order

248 (Williams' Latin Square design) to avoid carryover effects and bias in the data collection.

249 Evaluation was performed under white light, at room temperature and in sound controlled

250 partitioned sensory booths in the CASS sensory laboratory of Deakin University. All the data

251 collection was done using Compusense software (Compusense Inc., Guelph, Ontario, Canada).

252 The mean values of the liking scores were used for the optimization of sweeteners using RSM

253 (Table 1).

254

10
255 2.5. Experimental design and data analysis

256 Experiments were designed using RSM of Design-Expert 12 statistical software (State Ease,

257 Inc. Minneapolis, Mn, USA). The central composite design (CCD) of RSM was used to optimize

258 concentrations of natural non-nutritive sweeteners for 50% reduction of added sugar. It is based

259 on 2-level factorial designs, having centre and axial points to fit quadratic models. In total, it has 5

260 levels which include +/- axial points, +/- factorial points and the centre point. The number of

261 experiments is calculated based on the formula in the equation 1 (Aslan, 2007):

262

of
263 No. of Experiments = 2k + 2k + centre points………….……(1)

ro
264 Where, k is the number of variables. In the study design with 2 variables (Table 1), the number of

265 experiments is 22 + 2×2 + 5 = 13.

-p
re
lP

266 Therefore, 13 different formulations (5 centre points and 8 non-centre points) of chocolate

267 milk comprising of two independent variables: Stevia sweetener (5-100 ppm) and Monk fruit
na

268 extract (50-100 ppm) were used to fit a quadratic response surface and approximately locate the
ur

269 optimum combination. The centre points help to anchor the design space i.e. to control the design
Jo

270 from external environmental influence, thereby, ensuring that the design space is stable. The

271 effect of these independent variables on the sensory attributes (overall liking, appearance liking,

272 aroma liking, sweetness liking, mouthfeel liking and aftertaste liking) was evaluated by a second-

273 order polynomial as in the equation (2):

274 = +∑ +∑ + …………………………….(2)

275 Where, Y is the response variable; β0 is the intercept; Xi and Xj are independent variables; βi, βij

276 and βii are regression coefficients for linear, interaction and quadratic terms; and k is number of

277 variables. 3D graphs were generated using regression coefficient and analysis of variance

278 (ANOVA) was conducted to find the significant (p < 0.05) effect of the model terms. For model

11
279 fitness, model F-value, lack of fitness and coefficient of determination (R2) were estimated for

280 each response variable.

281

282 3. Results and discussion

283 3.1. Optimization of variables

284 The CCD of the RSM software was used to generate levels of variables namely stevia

285 sweetener (X1) and monk fruit extract (X2) with 13 sets of experiments as shown in Table 1.

286 Based on the suggested model, the significance of the model and lack-of-fit (non-significant) were

of
287 individually checked for all responses. The statistics and adequacy of the models for all responses

ro
288 are shown in Table 2. For the model adequacy, it was found that the values of adjusted R2 were

289 lower than R2 with no more than 0.20.

-p
re
lP

290 After the model fit and adequacy, the numerical and graphical optimizations were carried

291 out for the sugar reduction in chocolate flavoured milk using stevia and monk fruit sweeteners. As
na

292 a part of numerical optimization, the criteria for desired goals, weightage and importance of each
ur

293 variable and response were set as summarized in Table 3. The goal for each sweetener
Jo

294 concentration was kept in range while the output for response variables (i.e. sensory attributes)

295 was desired for maximum. Using the desirability function method, the software predicted the most

296 desirable solution with the desirability value of 0.952 (Table 4). Based on the same set of criteria

297 and outputs for numerical optimization, relevant and statistically significant responses were

298 depicted through the 3D graphs for each response (Fig. 2 a-f). The shaded area represents the

299 independent variable (stevia sweetener and monk fruit extract) conditions and their effects on the

300 response variables (shown by 3D graphs) for optimizing sugar reduction in chocolate flavoured

301 milk.

302 3.2. Effect of concentration of NNSs on the sensory attributes

12
303 3.2.1. Effect on overall liking

304 Consumers are usually asked to provide the hedonic ratings for the overall liking first,

305 before a specific attribute (e.g., sweetness) because they evaluate the product using an integrative

306 frame of mind where all attributes are considered collectively. However, when they evaluate a

307 specific attribute, they switch into an analytical frame of mind and focus only on that specific

308 attribute (Lawless & Heymann, 2010). This prevents the influence of the memory of any

309 particular attribute on the overall liking score.

310 The effect of non-nutritive sweeteners (stevia sweetener and monk fruit extract) on the

of
311 overall liking for 50% reduction of added sugar is presented in Table 1. Both stevia sweetener and

ro
312 monk fruit extract positively affected the overall liking score, however, the effect of stevia

313
-p
sweetener was more prominent than monk fruit extract. This effect is illustrated by the quadratic
re
314 equation obtained by the response surface analysis (RSA) of the data as expressed in eq. (3).
lP

315 Overall liking = + 6.76 + 0.13 × X1 + 0.074 × X2 - 0.21 × X1X2 - 0.17 × X12 - 0.08 × X22 …..(3)
na

316 Where,
ur

317 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)

Jo

318 The range of overall liking scores varied between 6.06 to 6.80 (Table 1 and Table 3) on a

319 9-point hedonic scale for 50% reduction of added sugar. Similar findings were reported by earlier

320 researchers, Li et al. (2015) reported overall liking scores of 6.5 for both monk fruit and stevia

321 leaf extract for 50% reduction as compared to control skim chocolate milk (5.14% sucrose with

322 overall liking 6.9).

323 The overall liking scores followed a quadratic model as depicted in Fig. 2a and Fig. 3.

324 Initially, with an increase of the concentration of stevia sweetener, the overall liking increases up

325 to 6.76 and then tails off while for the monk fruit extract, the tailing off is less prominent (Fig. 3).

326 This can also be seen from the eq. (3) where the quadratic coefficient for stevia sweetener (X12) is

13
327 greater than that of monk fruit extract (X22). Further looking at the interaction between these two

328 sweeteners (X1X2 term in eq. 3), there is a slightly negative impact on the overall liking score.

329 Overall, both sweeteners influenced positively and synergistically increased the overall liking

330 scores with a prediction of 6.78 (Fig. 4 and Table 4).

331

332 3.2.2. Effect on appearance liking

333 The appearance liking is based on the visual appearance of the chocolate milk. The scores

334 for appearance liking ranged in between 6.07 and 6.50 (Table 1 and Table 3). The effect of stevia

of
335 sweetener and monk fruit extract on the appearance is describes as in quadratic eq. (4).

ro
336 Appearance = + 6.46 + 0.06 × X1 + 0.03 × X2 - 0.09 × X1X2 - 0.13 × X12 - 0.09 × X22 …..(4)

337 Where,
-p
re
338 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)
lP

339 Both the sweeteners (stevia sweetener and monk fruit extract) had a positive effect on
na

340 appearance liking of the chocolate milk (Fig. 2b and Fig. 3). Stevia sweetener showed a positive
ur

341 effect double compared to monk fruit extract as evident from their respective coefficient X1 and
Jo

342 X2 (eq. 4). Stevia sweetener and monk fruit extract showed a similar pattern of quadratic

343 behaviour as observed for overall liking (Fig. 3). Similarly, the quadratic coefficient for stevia

344 sweetener (X12) is greater than that of monk fruit extract (X22) as shown in eq. (4) and both the

345 sweeteners contributed positively and synergistically to appearance liking with a prediction of

346 6.47 (Fig. 4 and Table 4). The appearance liking of the current study was similar to Li et al.

347 (2015), where the appearance liking were 6.6 and 6.7, respectively for 50% reduction with monk

348 fruit extract and stevia leaf extract as compared to control skim chocolate milk (5.14% sucrose

349 with appearance liking 6.5).

350

14
351 3.2.3. Effect on aroma liking

352 Aroma is the odour perceived through the nasal cavity while flavour combines both aroma

353 and taste. The quadratic equation describing the relation between stevia sweetener and monk fruit

354 extract and their effect on the aroma is represented in eq. (5).

355

356 Aroma = +6.30 + 0.06 × X1 + 0.05 × X2 - 0.15 × X1X2 - 0.16 × X12 - 0.09 × X22 ……..(5)

357 Where,

of
358 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)

ro
359 The range for aroma liking scores varied between 5.80 to 6.34 (Table 1 and Table 3). Both

360
-p
sweeteners had a positive effect on the aroma liking of chocolate milk (Fig. 2b and Fig. 3). Their
re
361 effects were almost equivalent as evident from the coefficients of X1 and X2 (eq. 5) while monk
lP

362 fruit extract tails off slightly slower than stevia sweetener as observed from their quadratic
na

363 behaviour graph for aroma liking (Fig. 3). Their overall effects on aroma liking were positive and

364 synergistic with a prediction of 6.31 (Fig. 4 and Table 4) which is supported by previous findings
ur

365 where aroma profile was improved in sugar-free dairy dessert with the combined effect of stevia
Jo

366 and sucralose (Furlán & Campderrós, 2017). The aroma liking for chocolate milk in this study

367 was similar to the flavour liking of skim chocolate milk by Li et al. (2015).

368 The aroma of chocolate milk in the current study is mainly due to cocoa powder, sugar and

369 milk. No additional aroma was added during the preparation of the chocolate milk. The addition

370 of aroma could be employed to enhance the sensory perception while reducing sugar. For

371 example, Alcaire, Antunez, Vidal, Gimenez, & Ares (2017) used vanilla flavour to mitigate the

372 effect of a 20% reduction of added sugar in milk desserts. Further, aroma is also associated with

373 perception of sweetness in a particular product (Prescott, Johnstone, & Francis, 2004) and the

374 taste-smell integration in the brain is related to existing experiences with taste-smell

15
375 combinations. Both tastants and odourants are responsible for overlapping activations in a specific

376 part of the brain that leads to enhanced sweetness perception (Small et al., 2004).

377

378

379

380 3.2.4. Effect on sweetness liking

381 Sweetness is a key driver of liking for food products and a heightened liking for sweet tastes

of
382 has been associated with increased intakes of foods with added sucrose (Drewnowski, Mennella,

ro
383 Johnson, & Bellisle, 2012). Therefore, sweetness is given higher importance compared to the

384 -p
other sensory attributes as presented in Table 3. The quadratic equation obtained from RSA data
re
385 shows the effect of stevia sweetener and monk fruit extract on sweetness liking is represented in
lP

386 eq. (6).

Sweetness = + 3.45 + 0.04 × X1 + 0.05 × X2 - 0.0002 × X1X2 - 0.0002 × X12 - 0.0002 × X22….(6)
na

387

Where,
ur

388
Jo

389 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)

390 Sweetness liking scores varied from 5.50 to 6.50 (Table 1 and Table 3) for a 50% reduction

391 of added sugar. Similar findings were reported by Li et al. (2015) where the sweetness liking

392 scores were 6.3 and 6.2, respectively for monk fruit and stevia leaf extract for 50% reduction as

393 compared to control skim chocolate milk (5.14% sucrose with sweetness liking 6.9).

394 Sweetness liking scores followed a quadratic model as shown in Fig. 2d and Fig. 3.

395 Psychophysical curve provides practical formulation information between the perceived sweetness

396 intensity (on Y-axis) and sweetener concentration (on X-axis) (Cardello, Da Silva, & Damasio,

397 1999). Psychophysical behaviour has been studied for dose-response behaviour across 16

16
398 sweeteners (Wee, Tan, & Forde, 2018); for sweet taste perception associated with sweet food

399 liking and intake (Jayasinghe et al., 2017) as well as oral sensitivity to complex carbohydrates and

400 sweet taste in humans (Low, Lacy, McBride, & Keast, 2017). Unlike the psychophysical curve,

401 where sweetness intensity is on the Y-axis, quadratic modelling graph (Fig. 3) uses sweetness

402 liking on Y-axis. Usually, as sweetness levels increase, a pattern of hedonic responses follows an

403 inverted-U as reported across different cultures (Pfaffmann, 1980; Prescott et al., 1992; Prescott et

404 al., 1997; Prescott, 1998). A similar pattern was observed (Fig. 3) where sweetness liking was

405 increased to a point and then gradually tails off. The tailing off is quicker for stevia sweetener

406 compared to monk fruit extract. This is because both stevia (Reb A) and monk fruit have varied

of
407 temporal sweetness and side taste profiles compared to sucrose with monk fruit having prolonged

ro
408 sweet and honey taste in comparison to stevia (Reb A) (Tan, Wee, Tomic, & Forde, 2019). A

409
-p
similar curve referring to “sensory liking relation” was used for the selection of tomato sauce
re
410 (Moskowitz & Marketo, 2001).
lP

411 Further looking at the interaction coefficient (X1X2) and quadratic coefficients for stevia
na

412 sweetener (X12) and monk fruit extract (X22) in eq. 6, the value (0.0002) is almost negligible
ur

413 suggesting a strong positive and synergistic effect of stevia sweetener and monk fruit extract on
Jo

414 sweetness liking with a prediction of 6.47 (Fig. 4 and Table 4) when used in combination. This

415 finding is supported by earlier research where the use of two sweeteners as a blend

416 (Cyclamate/Saccharin blend, 2:1) enhanced sweetness to meet sweetness equivalent to sucrose

417 probably by minimising off-flavour or bitter after-taste in peach nectar (Cardoso & Bolini, 2007).

418 Similarly, blending aspartame (APM) and acesulfame-K (ACE-K) resulted in a sweetness synergy

419 of approximately 30% (DuBois & Prakash, 2012) as well as blending Reb M with Reb B/ Reb D

420 resulted in sweetness synergy with improvement in sweetness intensity (Prakash et al., 2014).

421

422 3.2.5. Effect on mouthfeel liking

17
423 Consumer liking of chocolate milk is largely dependent on the mouthfeel profile next to

424 sweetness (Paixão, Rodrigues, Esmerino, Cruz, & Bolini, 2014). Mouthfeel depends on various

425 ingredients such as sugar, fat, protein, stabilisers like carrageenan and other thickening agents in

426 the chocolate milk (Li et al., 2015). For this study, we fixed the other ingredients (Fig. 1) for all

427 reformulations except for the variables (stevia sweetener and monk fruit extract) to evaluate their

428 effect on sensory attributes. The quadratic equation showing the effect of stevia sweetener and

429 monk fruit extract on mouthfeel liking is represented in eq. (7).

430

of
431 Mouthfeel = + 6.43 + 0.08 × X1 + 0.09 × X2 - 0.19 × X1X2 - 0.21 × X12 - 0.16 × X22……(7)

ro
432 Where,
-p
re
433 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)
lP

434 Values of mouthfeel liking vary from 5.64 to 6.48 (Table 1 and Table 3) for a 50%

435 reduction of sucrose. Li et al. (2015) reported thickness liking scores similar to the mouthfeel
na

436 scores of the present findings i.e. 6.3 and 6.4, respectively for monk fruit and stevia leaf extract
ur

437 for 50% reduction as compared to control skim chocolate milk (5.14% sucrose with thickness
Jo

438 liking 6.5). The mouthfeel liking revealed a quadratic model similar to other sensory attributes as

439 depicted in Fig. 2e and Fig. 3. Mouthfeel liking increased with an increase in concentration up to

440 a certain point and then gradually tails off. Tailing off is slightly quicker for stevia sweetener

441 compared to monk fruit extract (Fig. 3). This could be due to their varying mouth-drying

442 properties (i.e. the sensation of the lack of lubrication or moistness) as studied by Tan et al. (2019)

443 where mouth-drying was higher for stevia (Reb A) compared to monk fruit. Further, the

444 sweeteners (stevia sweetener and monk fruit extract) showed an almost similar positive effect on

445 mouthfeel liking as evident from the coefficients X1 and X2 (eq. 7). The monk fruit extract tails off

446 slightly slower than stevia sweetener for mouthfeel liking, similar to that observed for aroma

18
447 liking (Fig. 3). Overall, the combined effect of both the sweeteners on the mouthfeel liking was

448 positive and synergistic with a prediction of 6.45 (Fig. 4 and Table 4).

449

450 3.2.6. Effect on aftertaste liking

451 Aftertaste is another important sensory attribute for consumer liking and acceptance or

452 rejection of any product. Presence of aftertastes like bitterness and metallic can limit the

453 applications of sweeteners for sugar reduction in foods and beverages (Belščak-Cvitanović et al.,

454 2015). The quadratic equation for the effect of stevia sweetener and monk fruit extract on

of
455 aftertaste liking is represented in eq. (8).

ro
456 Aftertaste = + 6.30 + 0.11× X1 + 0.09 × X2 - 0.22 × X1X2 - 0.28 × X12 - 0.03 × X22…..(8)
-p
re
457 Where,
lP

458 X1 = Stevia sweetener (ppm); X2 = Monk fruit extract (ppm)

na

459 Aftertaste liking scores range from 5.52 to 6.41 (Table 1 and Table 3). Li et al. (2015)
ur

460 reported similar values for aftertaste liking i.e. 5.8 for both monk fruit and stevia leaf extract for
Jo

461 50% reduction as compared to control skim chocolate milk (5.14% sucrose with aftertaste liking

462 6.3). Aftertaste liking scores followed a quadratic model as depicted in Fig. 2f and Fig. 3. Stevia

463 sweetener showed an inverted-U shape while the graph for monk fruit extract is almost linear.

464 This is because stevia sweetener contains Reb A which has metallic and bitter aftertastes (DuBois

465 & Prakash, 2012) that causes the graph to tail off after reaching the optimum point. Tan et al.

466 (2019) further reported bitterness, metallic and chemical taste to be more prominent in the

467 temporal profile of stevia (Reb A) compared to the monk fruit which had a prolonged sweet and

468 honey taste. This accounts for the lower coefficient values of (X2) and (X22) in eq. (8).

469 Therefore, when both the sweeteners were used in combination, the bitter and metallic

470 aftertastes of stevia sweetener were masked by the monk fruit extract. This led to the overall

19
471 positive and synergistic effect of stevia sweetener and monk fruit extract on the aftertaste liking

472 with a prediction of 6.33 (Fig. 4 and Table 4). Similar findings were reported in earlier research,

473 where a blend of two sweeteners (Cyclamate/Saccharin blend, 2:1) minimised the off-flavour and

474 bitter after-taste in peach nectar (Cardoso & Bolini, 2007). Further, blending Reb M with Reb B/

475 Reb D improved the onset and bitterness perception (Prakash et al., 2014). This collectively leads

476 to improved sweetness perception and ultimately the overall liking and acceptance of the product

477 by the consumers.

478 3.3. Optimized chocolate flavoured milk

of
479 Stevia sweetener (5-100 ppm) and Monk fruit extract (50-100 ppm) were used for the

ro
480 optimization for 50% reduction of added sugar (i.e. sucrose, 5% to 2.5%) in chocolate flavoured

481
-p
milk using central composite design (CCD) of the RSM software. Following the numerical (Table
re
482 3) and graphical (Fig. 2) optimization processes, an optimum amount of these sweeteners within
lP

483 the selected range i.e. 56.27 ppm of stevia sweetener and 81.90 ppm of monk fruit extract were

484 predicted for achieving the optimum sensory attributes for 50% sugar reduction in chocolate
na

485 flavoured milk. These concentrations resulted in a positive and synergistic effect on the sensory
ur

486 attributes to achieve optimum liking of 6.78, 6.47, 6.31, 6.47, 6.45 and 6.33, respectively for
Jo

487 overall liking, appearance, aroma, sweetness, mouthfeel and aftertaste as the best outcome in

488 terms of desirability function (0.952) (Table 4). Also, the viscosity (8.38 mPa.s) of the chocolate

489 milk with optimum sweetener concentration was close to the control sample (8.34 mPa.s). Further,

490 the predicted sensory attributes’ liking of the 50% reduced added sugar chocolate milk with stevia

491 and monk fruit extract is comparable to the control with full sugar (Table 1).

492 4. Conclusion

493 Optimization of natural non-nutritive sweeteners namely stevia sweetener (5-100 ppm) and

494 monk fruit extract (50-100 ppm) for 50% sucrose reduction (5% to 2.5%) in chocolate flavoured

495 milk could be predicted based on sensory attributes using a central composite design (CCD) from

20
496 RSM. Multiple response optimizations predicted the concentrations of stevia sweetener and monk

497 fruit extract to be 56.27 ppm and 81.90 ppm, respectively for the optimum sensory attributes for

498 sugar reduction in chocolate flavoured milk without much impact as compared to the control with

499 full sugar. Therefore, the combined use of natural non-nutritive sweeteners could be used to

500 partially reduce sugar in food products to meet WHO and PHE sugar guidelines but still maintain

501 consumer liking and acceptance of the product. From an industrial point of view, the results of

502 this study can assist food companies in overcoming the technical challenges underlying sugar

503 reduction in milk-based beverages and achieve at least a small reduction that could benefit the

504 health of the population significantly. The future recommendation is to consider children as

of
505 consumers as well as their demographic variations for the acceptance of the reduced sugar

ro
506 product.
-p
re
507
lP

508 Acknowledgments
na

509 This research is supported by Deakin University Postgraduate Research (DUPR)

510 Scholarship and the authors would like to thank TATE & LYLE, London, UK for providing the
ur

511 natural non-nutritive sweeteners as well as Simone Lewin, for her support.
Jo

512 Compliance with ethical standards

513 The authors ensure that the consumer study involving humans was carried out in accordance with

514 The Code of Ethics of the World Medical Association (Declaration of Helsinki), as per the

515 approved ethics, Human Ethics Advisory Group, Faculty of Health (HEAG-H 155_2018),

516 Melbourne Burwood Campus, Deakin University.

517 Conflict of interest

518 The authors declare no conflict of interest.

519

21
520 References

521 Agoda-Tandjawa, G., Le Garnec, C., Boulenguer, P., Gilles, M., & Langendorff, V. (2017).

522 Rheological behavior of starch/carrageenan/milk proteins mixed systems: Role of each

523 biopolymer type and chemical characteristics. Food Hydrocolloids, 73, 300-312.

524 Alcaire, F., Antunez, L., Vidal, L., Gimenez, A., & Ares, G. (2017). Aroma-related cross-modal

525 interactions for sugar reduction in milk desserts: Influence on consumer perception. Food

526 Research International, 97, 45-50.

527 Aslan, N. (2007). Application of response surface methodology and central composite rotatable

528 design for modeling the influence of some operating variables of a multi-gravity separator

of
529 for coal cleaning. Fuel, 86, 769-776.

ro
530 Australia and New Zealand Food Standards Code. (2018). Application to Amend the

531
-p
Specifications for Rebaudioside M Under Australia and New Zealand Food Standards
re
532 Code – Standard.
lP

533 Azami, T., Niakousari, M., Hashemi, S. M. B., & Torri, L. (2018). A three-step sensory-based
na

534 approach to maximize consumer acceptability for new low-sugar licorice-chocolate-

ur

535 flavored milk drink. LWT- Food Science and Technology, 91, 375-381.

536 Ban, Q., Cheng, J., Sun, X., Jiang, Y., Zhao, S., Song, X., & Guo, M. (2020). Effects of a
Jo

537 synbiotic yogurt using monk fruit extract as sweetener on glucose regulation and gut

538 microbiota in rats with type 2 diabetes mellitus. Journal of Dairy Science, 103, 2956-2968.

539 Bayarri, S., Carbonell, I., Barrios, E. X., & Costell, E. (2011). Impact of sensory differences on

540 consumer acceptability of yoghurt and yoghurt-like products. International Dairy Journal,

541 21, 111-118.

542 Belščak-Cvitanović, A., Komes, D., Dujmović, M., Karlović, S., Biškić, M., Brnčić, M., & Ježek,

543 D. (2015). Physical, bioactive and sensory quality parameters of reduced sugar chocolates

544 formulated with natural sweeteners as sucrose alternatives. Food Chemistry, 167, 61-70.

22
545 Blake, M. R., Lancsar, E., Peeters, A., & Backholer, K. (2019). Sugar-sweetened beverage price

546 elasticities in a hypothetical convenience store. Social Science & Medicine, 225, 98-107.

547 Bordi Jr, P. L., Palchak, T., Verruma-Bernardi, M. R., & Cho, H. C. (2016). Adult Acceptance of

548 Chocolate Milk Sweetened with Stevia. Journal of Culinary Science & Technology, 14,

549 216-221.

550 Brownie, S., Muggleston, H., & Oliver, C. (2015). The 2013 Australian dietary guidelines and

551 recommendations for older Australians. Australian Family Physician, 44, 311.

552 Cardello, H. M. A. B., Da Silva, M. A. P. A., & Damasio, M. H. (1999). Measurement of the

553 relative sweetness of stevia extract, aspartame and cyclamate/saccharin blend as compared

of
554 to sucrose at different concentrations. Plant Foods for Human Nutrition, 54, 119-129.

ro
555 Cardoso, J. M. P., & Bolini, H. M. A. (2007). Different sweeteners in peach nectar: Ideal and

556
-p
equivalent sweetness. Food Research International, 40, 1249-1253.
re
557 Chen, W., Wang, J., Qi, X., & Xie, B. (2007). The antioxidant activities of natural sweeteners,
lP

558 mogrosides, from fruits of Siraitia grosvenori. International Journal of Food Sciences and
na

559 Nutrition, 58, 548-556.

ur

560 Chollet, M., Gille, D., Schmid, A., Walther, B., & Piccinali, P. (2013). Acceptance of sugar

561 reduction in flavored yogurt. Journal of Dairy Science, 96, 5501-5511.

Jo

562 Christoph, I. B., Peter, G., Rothe, A., Salamon, P., Weber, S. A., & Weible, D. (2011). School

563 milk consumption in Germany–what are important product attributes for children and

564 parents? In EAAE 2011 Congress Change and Uncertainty. ETH Zurich, Zurich,

565 Switzerland.

566 Chun, L., Li-Mei, L., Feng, S., Zhi-Min, W., Hai-Ru, H., Li, D., & Jiang, T.-L. (2014). Chemistry

567 and pharmacology of Siraitia grosvenorii: A review. Chinese Journal of Natural

568 Medicines, 12, 89-102.

23
569 Cleghorn, C., Blakely, T., Mhurchu, C. N., Wilson, N., Neal, B., & Eyles, H. (2019). Estimating

570 the health benefits and cost-savings of a cap on the size of single serve sugar-sweetened

571 beverages. Preventive Medicine, 120, 150-156.

572 Dairy Australia, 2019. Australian dairy industry: IN FOCUS 2019.

573 https://www.dairyaustralia.com.au/industry/production-and-sales/consumption-summary

574 de Melo, L. L. M. M., Bolini, H. M. A., & Efraim, P. (2009). Sensory profile, acceptability, and

575 their relationship for diabetic/reduced calorie chocolates. Food Quality and Preference,

576 20, 138-143.

577 Drewnowski, A., Mennella, J. A., Johnson, S. L., & Bellisle, F. (2012). Sweetness and food

of
578 preference. The Journal of Nutrition, 142, 1142S-1148S.

ro
579 DuBois, G. E., & Prakash, I. (2012). Non-caloric sweeteners, sweetness modulators, and

580
-p
sweetener enhancers. Annual Review of Food Science and Technology, 3, 353-380.
re
581 EFSA. (2010). Scientific opinion on the safety of steviol glycosides for the proposed uses as a
lP

582 food additive. EFSA Panel of Food Additive and Nutrients Sources added to Food (ANS).
na

583 EFSA Journal, 8, 1537.

ur

584 EFSA. (2015). Scientific opinion on the safety of the proposed amendment of the specifications

585 for steviol glycosides (E 960) as a food additive. EFSA Panel of Food Additive and
Jo

586 Nutrient Sources added to Food (ANS). EFSA Journal, 13, 4316.

587 Egger, G., Swinburn, B., & Rossner, S. (2003). Dusting off the epidemiological triad: could it

588 work with obesity? Obesity Reviews, 4, 115-119.

589 Falbe, J., Thompson, H. R., Patel, A., & Madsen, K. A. (2019). Potentially addictive properties of

590 sugar-sweetened beverages among adolescents. Appetite, 133, 130-137.

591 FPED. (2014). What We Eat in America. NHANES 2011-2012, individuals 2 years and over

592 (excluding breast-fed children), day 1 dietary intake data, weighted. Food Patterns

593 Equivalents Database (FPED), 2011-2012.

24
594 Furlán, L. T. R., & Campderrós, M. E. (2017). The combined effects of Stevia and sucralose as

595 sugar substitute and inulin as fat mimetic on the physicochemical properties of sugar-free

596 reduced-fat dairy dessert. International Journal of Gastronomy and Food Science, 10, 16-

597 23.

598 Guggisberg, D., Piccinali, P., & Schreier, K. (2011). Effects of sugar substitution with Stevia,

599 Actilight™ and Stevia combinations or Palatinose™ on rheological and sensory

600 characteristics of low-fat and whole milk set yogurt. International Dairy Journal, 21, 636-

601 644.

602 Hanks, A. S., Just, D. R., & Wansink, B. (2014). Chocolate milk consequences: a pilot study

of
603 evaluating the consequences of banning chocolate milk in school cafeterias. PLOS One, 9,

ro
604 e91022.

605
-p
Henry, C., Whiting, S. J., Phillips, T., Finch, S. L., Zello, G. A., & Vatanparast, H. (2015). Impact
re
606 of the removal of chocolate milk from school milk programs for children in Saskatoon,
lP

607 Canada. Applied Physiology, Nutrition, and Metabolism, 40, 245-250.

na

608 Jayasinghe, S. N., Kruger, R., Walsh, D. C., Cao, G., Rivers, S., Richter, M., & Breier, B. H.
ur

609 (2017). Is sweet taste perception associated with sweet food liking and intake? Nutrients,

610 9, 750.
Jo

611 Keast, R., Sayompark, D., Sacks, G., Swinburn, B., & Riddell, L. (2011). The influence of

612 caffeine on energy content of sugar-sweetened beverages:‘the caffeine–calorie effect’.

613 European Journal of Clinical Nutrition, 65, 1338.

614 Kemp, D. S. E., Hollowood, D. T., & Hort, D. J. (2009). Sensory test methods. Sensory

615 Evaluation: A practical handbook, 66-137.

616 Kim, N.-C., & Kinghorn, A. D. (2002). Highly sweet compounds of plant origin. Archives of

617 Pharmacal Research, 25, 725-746.

25
618 Krachler, B., Eliasson, M., Stenlund, H., Johansson, I., Hallmans, G., & Lindahl, B. (2006).

619 Reported food intake and distribution of body fat: a repeated cross-sectional study.

620 Nutrition Journal, 5, 34.

621 Lawless, H. T., & Heymann, H. (2010). Sensory evaluation of food: principles and practices:

622 Springer Science & Business Media.

623 Lee, Y. J., Jeong, J., Kim, M. O., & Nam, J.-O. (2016). The positive effect of luohanguo as sugar

624 substitute on blood glucose and metabolism in streptozotocin-induced diabetic mice.

625 Applied Microscopy, 46, 140-149.

626 Li, X. E., Lopetcharat, K., & Drake, M. A. (2014). Extrinsic attributes that influence parents’

of
627 purchase of chocolate milk for their children. Journal of Food Science, 79, S1407-S1415.

ro
628 Li, X. E., Lopetcharat, K., & Drake, M. A. (2015). Parents’ and children's acceptance of skim

629
-p
chocolate milks sweetened by monk fruit and stevia leaf extracts. Journal of Food Science,
re
630 80, S1083-S1092.
lP

631 Liu, C., Dai, L., Liu, Y., Rong, L., Dou, D., Sun, Y., & Ma, L. (2016). Antiproliferative activity of
na

632 triterpene glycoside nutrient from monk fruit in colorectal cancer and throat cancer.
ur

633 Nutrients, 8, 360.

634 Low, J. Y. Q., Lacy, K. E., McBride, R. L., & Keast, R. S. J. (2017). Evidence supporting oral
Jo

635 sensitivity to complex carbohydrates independent of sweet taste sensitivity in humans.

636 PLOS One, 12, e0188784.

637 Ludwig, D. S., Peterson, K. E., & Gortmaker, S. L. (2001). Relation between consumption of

638 sugar-sweetened drinks and childhood obesity: a prospective, observational analysis. The

639 Lancet, 357, 505-508.

640 Lustig, R. H., Schmidt, L. A., & Brindis, C. D. (2012). Public health: the toxic truth about sugar.

641 Nature, 482, 27.

26
642 Mahato, D. K., Keast, R., Liem, D. G., Russell, C. G., Cicerale, S., & Gamlath, S. (2020). Sugar

643 Reduction in Dairy Food: An Overview with Flavoured Milk as an Example. Foods, 9(10),

644 1400.

645 Malik, V. S., Schulze, M. B., & Hu, F. B. (2006). Intake of sugar-sweetened beverages and weight

646 gain: a systematic review. The American Journal of Clinical Nutrition, 84, 274-288.

647 Morais, E., Morais, A., Cruz, A., & Bolini, H. (2014). Development of chocolate dairy dessert

648 with addition of prebiotics and replacement of sucrose with different high-intensity

649 sweeteners. Journal of Dairy Science, 97, 2600-2609.

650 Moskowitz, H. R., & Marketo, C. (2001). Selecting products for Category Appraisal Studies—

of
651 fewer products do almost as well as many products. Journal of Sensory Studies, 16, 537-

ro
652 549.

653
-p
Nicklas, T. A., O'neil, C., & Fulgoni, V. (2017). Flavored milk consumers drank more milk and
re
654 had a higher prevalence of meeting calcium recommendation than nonconsumers. Journal
lP

655 of School Health, 87, 650-657.

na

656 Oltman, A. E., Lopetcharat, K., Bastian, E., & Drake, M. A. (2015). Identifying key attributes for
ur

657 protein beverages. Journal of Food Science, 80, S1383-S1390.

658 Ozdemir, C., Arslaner, A., Ozdemir, S., & Allahyari, M. (2015). The production of ice cream
Jo

659 using stevia as a sweetener. Journal of Food Science and Technology, 52, 7545-7548.

660 Paixão, J., Rodrigues, J., Esmerino, E., Cruz, A., & Bolini, H. (2014). Influence of temperature

661 and fat content on ideal sucrose concentration, sweetening power, and sweetness

662 equivalence of different sweeteners in chocolate milk beverage. Journal of Dairy Science,

663 97, 7344-7353.

664 Pandey, A. K., & Chauhan, O. P. (2019). Monk fruit (Siraitia grosvenorii)-health aspects and

665 food applications. Pantnagar Journal of Research, 17, 191-198.

666 Parker, M. N., Lopetcharat, K., & Drake, M. A. (2018). Consumer acceptance of natural

667 sweeteners in protein beverages. Journal of Dairy Science, 101, 8875-8889.

27
668 Patel, A. I., Moghadam, S. D., Freedman, M., Hazari, A., Fang, M. L., & Allen, I. E. (2018). The

669 association of flavored milk consumption with milk and energy intake, and obesity: A

670 systematic review. Preventive Medicine, 111, 151-162.

671 Pawar, R. S., Krynitsky, A. J., & Rader, J. I. (2013). Sweeteners from plants—with emphasis on

672 Stevia rebaudiana (Bertoni) and Siraitia grosvenorii (Swingle). Analytical and

673 Bioanalytical Chemistry, 405, 4397-4407.

674 Pfaffmann, C. (1980). Wundt's schema of sensory affect in the light of research on gustatory

675 preferences. Psychological Research, 42, 165-174.

676 PHE. (2017). Sugar reduction: achieving the 20%. A technical report outlining progress to date,

of
677 guidelines for industry, 2015 baseline levels in key foods and next steps. London, UK:

ro
678 Public Health England. In.

679
-p
Pon, S., Lee, W., & Chong, G. (2015). Textural and rheological properties of stevia ice cream.
re
680 International Food Research Journal, 22, 1544.
lP

681 Prakash, I., Markosyan, A., & Bunders, C. (2014). Development of next generation stevia
na

682 sweetener: rebaudioside M. Foods, 3, 162-175.

ur

683 Prescott, J., Laing, D., Bell, G., Yoshida, M., Gillmore, R., Allen, S., Yamazaki, K., & Ishii, R.

684 (1992). Hedonic responses to taste solutions: a cross-cultural study of Japanese and
Jo

685 Australians. Chemical Senses, 17, 801-809.

686 Prescott, J., Bell, G. A., Gillmore, R., Yoshida, M., O'sullivan, M., Korac, S., Allen, S., &

687 Yamazaki, K. (1997). Cross-cultural comparisons of Japanese and Australian responses to

688 manipulations of sweetness in foods. Food Quality and Preference, 8, 45-55.

689 Prescott, J. (1998). Comparisons of taste perceptions and preferences of Japanese and Australian

690 consumers: Overview and implications for cross-cultural sensory research. Food Quality

691 and Preference, 9, 393-402.

692 Prescott, J., Johnstone, V., & Francis, J. (2004). Odor–taste interactions: effects of attentional

693 strategies during exposure. Chemical Senses, 29, 331-340.

28
694 Qi, X.-Y., Chen, W.-J., Zhang, L.-Q., & Xie, B.-J. (2008). Mogrosides extract from Siraitia

695 grosvenori scavenges free radicals in vitro and lowers oxidative stress, serum glucose, and

696 lipid levels in alloxan-induced diabetic mice. Nutrition Research, 28, 278-284.

697 Rad, A. H., Delshadian, Z., Arefhosseini, S. R., Alipour, B., & Jafarabadi, M. A. (2012). Effect of

698 inulin and stevia on some physical properties of chocolate milk. Health Promotion

699 Perspectives, 2, 42.

700 Ruiz-Ojeda, F. J., Plaza-Díaz, J., Sáez-Lara, M. J., & Gil, A. (2019). Effects of sweeteners on the

701 gut microbiota: a review of experimental studies and clinical trials. Advances in Nutrition,

702 10, S31-S48.

of
703 Singh, J., Rasane, P., Kaur, S., Kumar, V., Dhawan, K., Mahato, D. K., Malhotra, S., Sarma, C.,

ro
704 Kaur, D. & Bhattacharya, J. (2020). Nutritional interventions and considerations for the

705
-p
development of low calorie or sugar free foods. Current Diabetes Reviews, 16, 301-312.
re
706 Small, D. M., Voss, J., Mak, Y. E., Simmons, K. B., Parrish, T., & Gitelman, D. (2004).
lP

707 Experience-dependent neural integration of taste and smell in the human brain. Journal of
na

708 Neurophysiology, 92, 1892-1903.

ur

709 Striegel-Moore, R. H., Thompson, D., Affenito, S. G., Franko, D. L., Obarzanek, E., Barton, B.

710 A., Schreiber, G. B., Daniels, S. R., Schmidt, M., & Crawford, P. B. (2006). Correlates of
Jo

711 beverage intake in adolescent girls: the National Heart, Lung, and Blood Institute Growth

712 and Health Study. The Journal of Pediatrics, 148, 183-187.

713 Sun, B.-S., Chen, Y.-P., Wang, Y.-B., Tang, S.-W., Pan, F.-Y., Li, Z., & Sung, C.-K. (2012).

714 Anti-obesity effects of mogrosides extracted from the fruits of Siraitia grosvenorii

715 (Cucurbitaceae). African Journal of Pharmacy Pharmacology, 6, 1492-1501.

716 Tan, V. W. K., Wee, M. S. M., Tomic, O., & Forde, C. G. (2019). Temporal sweetness and side

717 tastes profiles of 16 sweeteners using temporal check-all-that-apply (TCATA). Food

718 Research International, 121, 39-47.

29
719 Te Morenga, L., Mallard, S., & Mann, J. (2013). Dietary sugars and body weight: systematic

720 review and meta-analyses of randomised controlled trials and cohort studies. British

721 Medical Journal, 346, e7492.

722 Thompson, J. L., Lopetcharat, K., & Drake, M. A. (2007). Preferences for commercial strawberry

723 drinkable yogurts among African American, Caucasian, and Hispanic consumers in the

724 United States. Journal of Dairy Science, 90, 4974-4987.

725 Wee, M., Tan, V., & Forde, C. (2018). A comparison of psychophysical dose-response behaviour

726 across 16 sweeteners. Nutrients, 10, 1632.

of
727

ro
728
-p
re
lP
na
ur
Jo

30
Fig. 1. Flow diagram for the preparation of chocolate flavoured milk.

Fig. 2. Effect of stevia sweetener and monk fruit extract concentration on different sensory
attributes: (a) Overall liking; (b) Appearance liking; (c) Aroma liking; (d) Sweetness liking;
(e) Mouthfeel liking and (f) Aftertaste liking, for 50% reduction of added sugar in chocolate
flavoured milk.

Fig. 3. Quadratic modelling graph for concentration of natural sweeteners (stevia sweetener
and monk fruit extract) and the prediction for individual sensory attribute: overall liking;
appearance; aroma; sweetness; mouthfeel and aftertaste. The cross point on the middle
prediction line denotes the optimum concentration of each sweetener for individual sensory

of
attribute for maximum desirability function.

ro
Fig. 4. Interaction effects of stevia sweetener (denoted by black prediction lines) and monk
fruit extract (denoted by red prediction lines) for the sugar reduction in chocolate flavoured
-p
milk. The dotted lines denote 95% confidence interval (CI) while the cross point denotes the
re
optimum prediction for individual sensory attribute after the synergistic effect of the two
sweeteners.
lP

Table 1 The experimental design and sensory scores* for optimization with different
na

concentrations of natural sweeteners (Stevia sweetener and Monk fruit extract) for 50%
reduction of added sugar in chocolate flavoured milk by Response Surface Methodology
ur

(RSM).
Jo

Table 2 Model statistics (Analysis of Variance, ANOVA) and adequacy of the models for
sensory attributes of chocolate milk.

Table 3 Criteria and outputs of the numerical optimization for added sugar reduction in
chocolate flavoured milk using stevia sweetener and monk fruit extract.

Table 4 Most desirable solution of the experiment.

Table 1
Experiment Variables: Natural sweeteners Response: Sensory attributes

X1: Stevia X2: Monk fruit Overall Appearance Aroma Sweetness Mouthfeel Aftertaste
sweetener (ppm) extract (ppm) liking
1 5.00 50.00 6.06 6.07 5.80 5.50 5.64 5.52

f
oo
2 100.00 50.00 6.70 6.35 6.20 6.17 6.25 6.15
3 52.50 75.00 6.76 6.42 6.26 6.48 6.40 6.29

pr
4 100.00 100.00 6.46 6.20 5.95 5.75 6.00 5.92

e-
5 52.50 75.00 6.72 6.45 6.30 6.42 6.38 6.35

Pr
6 119.68 75.00 6.67 6.32 6.12 5.89 6.15 5.95

al
7 52.50 75.00 6.74 6.50 6.28 6.45 6.44 6.33

rn
8 52.50 39.64 6.55 6.25 6.03 6.08 6.02 6.18
9 52.50 75.00 u 6.80 6.46 6.32 6.46 6.46 6.25
Jo
10 0.00 75.00 6.24 6.13 5.90 5.63 5.98 5.62
11 5.00 100.00 6.65 6.29 6.15 6.21 6.18 6.18
12 52.50 110.36 6.72 6.37 6.24 6.35 6.30 6.41
13 52.50 75.00 6.78 6.48 6.34 6.50 6.48 6.30
Control - - 6.88 6.64 6.39 6.28 6.55 6.50
*Scored on a 9-point hedonic scale; ppm- parts per million; Control- Full sugar version (10%)
Table 2
Response Model Lack-of-fit R2 Adjusted R2 Std. dev. C.V.% F value

(P value) (P value)

*Overall liking Quadratic 0.07 0.97 0.94 0.05 0.82 39.77

(< 0.0001)

f
oo
*Appearance Quadratic 0.27 0.96 0.93 0.04 0.56 34.07

pr
(< 0.0001)

e-
*Aroma Quadratic 0.12 0.96 0.93 0.05 0.75 33.08

Pr
(< 0.0001)

al
rn
*Sweetness Quadratic 0.06 0.99 0.98 0.05 0.88 97.59

(< 0.0001) u
Jo
*Mouthfeel Quadratic 0.09 0.96 0.93 0.07 1.07 31.54

(0.0001)

*Aftertaste Quadratic 0.07 0.97 0.95 0.06 1.05 44.36

(< 0.0001)

*Significant at P value < 0.05. Model is significant while Lack-of-fit is non-significant.

Table 3

Name Goal Lower Upper Lower Upper Importance**

Limit Limit Weight* Weight*
X1: Stevia sweetener is in range 5 100 1 1 3
X2: Monk fruit is in range 50 100 1 1 3
extract
Overall liking maximize 6.06 6.80 1 1 5

f
oo
Appearance maximize 6.07 6.50 1 1

pr
Aroma maximize 5.80 6.34 1 1 3

e-
Sweetness maximize 5.50 6.50 1 1 5

Pr
Mouthfeel maximize 5.64 6.48 1 1 3
Aftertaste maximize 5.52 6.41 1 1 3

al
rn
*Weight range (0.10 to 10)-fine tunes the optimization process for the best solution. A low weight (near 0.10) allows more solutions that don’t quite meet the optimal goal while a

u
high weight (close to 10) causes the optimization to seek a solution close to or beyond the stated goal. So, from a practical standpoint, the weight is kept at 1. **Importance: 1 is
Jo
assigned for the response with the lowest importance (+); 3 for medium importance (+++) and 5 for critical importance (+++++).
Table 4
Solutions Stevia Monk Fruit Overall Appearance Aroma Sweetness Mouthfeel Aftertaste Desirability

sweetener extract liking

(TASTEVA®) (PUREFRUITTM)

(ppm) (ppm)

1 56.268 81.903 6.779 6.465 6.307 6.471 6.445 6.330 0.952 *Selected

f
oo
2 56.271 81.619 6.779 6.466 6.307 6.472 6.455 6.329 0.952

pr
* Response Surface Methodology (RSM) selected solution based on the criteria set in Table 3

e-
Pr
al
u rn
Jo
 Cocoa powder

Heat in water (approx. 20% solution) at around 90°C for 30 min

Premix stabilizer (Ticaloid®750, 0.02%), sucrose (5%), and NOVATIONTM 3300 starch

(0.5%)

Add Stevia sweetener (TASTEVA®, 5-100 ppm) and Monk fruit extract (PUREFRUITTM,

50-100 ppm)

Dissolve in water and heat at 70°C for 10 min

of
ro
Add cocoa solution to the premix

-p
Cool the solution to 20°C
re
Add pasteurized milk (3.5% w/v, protein; 3.3% w/v, fat; and 4.6% w/v, carbohydrate)
lP

Homogenize at 5000rpm for 10 min

na

Heat the mixture at 75°C for 15 min

ur

Cool the mixture to 30°C

Jo

Packaging

Storage at 4°C

Fig. 1.
 (a) (b) (c)

f
oo
pr
e-
Pr
(d) (e)
(f)

al
rn
u
Jo

Fig. 2.
Jo
ur
na
lP
re
-p
ro
Fig. 3.

of
Fig. 4.
Jo
urn
al
Pr
e-
pr
oo
f
 Highlights

1. It is possible to achieve a 50% of added sugar reduction in chocolate milk using stevia
and monk fruit compounds
2. The use of two sweeteners provide synergistic effects on sensory attributes
3. 56.27 ppm of stevia sweetener and 81.90 ppm of monk fruit extract showed optimum
sensory response with overall liking score of 6.78
4. The optimum sensory response is comparable to the control sample with full sugar
having overall liking score of 6.88

of
ro
-p
re
lP
na
ur
Jo
 Credit Author Statement

Russell Keast, Shirani Gamlath: Conceptualization of the manuscript; Dipendra

Kumar Mahato: Data collection, Writing- Original draft preparation. Djin Gie Liem:

Visualization, Investigation. Catherine Georgina Russell: Supervision.: Shirani Gamlath:

Software, Validation.: Sara Cicerale: Writing- Reviewing and Editing

of
ro
-p
re
lP
na
ur
Jo
Declaration of interests

☒ The authors declare that they have no known competing financial interests or personal relationships
that could have appeared to influence the work reported in this paper.

☐The authors declare the following financial interests/personal relationships which may be considered
as potential competing interests:

Signature:

of
ro
-p
re
lP
na
ur
Jo