Impact of protein standardization of milk powder with

lactose or permeate on whey protein nitrogen index and

heat classification
Vandna Sikand, Phillip S. Tong, John Walker

To cite this version:

Vandna Sikand, Phillip S. Tong, John Walker. Impact of protein standardization of milk powder
with lactose or permeate on whey protein nitrogen index and heat classification. Dairy Science &
Technology, 2008, 88 (1), pp.105-120. �hal-00895780�

HAL Id: hal-00895780

https://hal.science/hal-00895780
Submitted on 11 May 2020

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
Dairy Sci. Technol. 88 (2008) 105–120 Available online at:

c INRA, EDP Sciences, 2008 www.dairy-journal.org
DOI: 10.1051/dst:2007011
Original article

Impact of protein standardization of milk

powder with lactose or permeate on whey
protein nitrogen index and heat classification
Vandna Sikand1 , Phillip S. Tong2 *, John Walker3

1
Dairy Products Technology Center, California Polytechnic State University, 1 Grand Avenue,
San Luis Obispo, CA, 93407, USA
2
Dairy Science Department, California Polytechnic State University, 1 Grand Avenue,
San Luis Obispo, CA, 93407, USA
3
Statistics Department, California Polytechnic State University, 1 Grand Avenue,
San Luis Obispo, CA, 93407, USA

Abstract – Whey protein nitrogen index (WPNI) is a well-known method of classifying nonfat dry
milk powder (NFDM) based on its heat treatment. This classification scheme provides one criterion
for the selection of NFDM for food applications. However, the effects of variation in NFDM com-
position on WPNI are not well documented. The objective of this study was to determine the effects
of changing protein content in low heat and medium heat NFDM on WPNI value. Edible lactose
powder (ELP) and permeate powder (PP) from skim milk ultrafiltration were used to downward
standardize NFDM from a protein content of 35.5–30% protein on a wet basis. These standardized
powders were analyzed for WPNI by American Dairy Products Institute (ADPI) method. Powders
were also analyzed for protein, ash, fat, moisture and lactose by standard methods to describe the
composition. Using linear regression, WPNI was found to be positively associated with protein con-
tent for both low and medium heat NFDM from several suppliers. For example, a low heat NFDM
(from supplier B) with initial protein content of 34.3% (wet basis) and WPNI of 6.38 mg soluble
whey protein nitrogen (classified as a low heat powder) had its WPNI reduced to less than 6.0 mg
of soluble nitrogen per g of powder when standardized to a protein content of less than or equal
to 31.89% (wet basis) with either ELP or PP. This would reclassify it as medium heat powder. We
conclude that standardization of NFDM with lactose or permeate will change its WPNI value and
may effect its heat classification. We propose a modified approach to calculate WPNI based upon
soluble whey protein nitrogen per g protein. This new WPNI value (protein corrected) would be
independent of powder protein content and hence be more indicative of the actual heat treatment it
is intended to reflect.

whey protein nitrogen index / protein standardization / edible lactose powder / permeate
powder / nonfat dry milk powder

摘要 – 乳 糖 和 超 滤 透 过 液 对 乳 粉 中 乳 蛋 白 标 准 化 后 对 乳 清 蛋 白 态 氮 指 数 和 热 分 类 的 影
响 。 乳清蛋白态氮指数是划分脱脂乳粉热处理程度的良好方法。该分类方法为脱脂乳粉
在食品工业中的应用提供了选择标准。但是脱脂乳粉成分的变化对乳清蛋白态氮指数的
影响还没有予以证实。本文旨在研究低热及中热脱脂乳粉中不同蛋白质含量对乳清蛋白
态氮指数的影响。使用食用乳糖粉或者脱脂乳的超滤透过液, 将脱脂乳粉的湿基蛋白含
量标准化到 35.5%–30.0%, 其乳清蛋白态氮指数测定采用美国乳品协会提供的测定方法,
同时按照标准方法测定其蛋白质、灰分、脂肪、水分和乳糖含量。线性回归分析结果表

* Corresponding author (通讯作者): ptong@calpoly.edu

Article published by EDP Sciences and available at http://www.dairy-journal.org

or http://dx.doi.org/10.1051/dst:2007011
106 V. Sikand et al.

明：乳清蛋白态氮指数与低热及中热脱脂乳粉 , (不同供应商提供) 的蛋白质含量均呈正相

关。例如：低热脱脂乳粉 (B供应商) 在原始蛋白含量为 34.3% (湿基) 、可溶性乳清蛋白态
氮指数为 6.38 mg 时被认为是低热乳粉, 但是当用食用乳糖粉或者超滤透过液将湿基蛋白质
含量降低至 31.89% 时, 则每克乳粉中可溶性乳清蛋白态氮的含量小于 6.0 mg, 这样可能就被
归类为中热脱脂乳粉。这就表明使用乳糖或者超滤透过液进行脱脂乳的标准化可能改变其
乳清蛋白态氮指数, 并影响其热分类结果。因此, 并影响其热分类结果。因此, 建议采用每克
蛋白中可溶性乳清蛋白氮来改进乳清蛋白态氮指数的计算方法。该种方法中乳清蛋白态氮
指数 (蛋白质进行了校正) 与脱脂乳粉中的蛋白质含量无关, 因此更能反应出热处理的真实
的程度。

乳清蛋白态氮指数 / 乳蛋白标准化 / 食用乳糖粉 / 超滤透过液 / 脱脂乳粉

Résumé – Impact de la standardisation en protéines de poudre de lait avec du lactose ou du

perméat sur l’indice d’azote des protéines solubles et sur sa classification thermique. L’indice
d’azote des protéines solubles (WPNI) est une méthode bien connue pour classifier les poudres de
lait écrémé selon leur traitement thermique. Cette classification procure un critère pour sélectionner
ces poudres pour des applications alimentaires. Cependant, les effets de leurs variations de composi-
tion sur le WPNI n’est pas bien documenté. L’objectif de cette étude était de déterminer les effets de
changements de teneur en protéines dans des poudres de lait écrémé « low heat » et « medium heat »
sur leur WPNI. Des poudres alimentaires de lactose ou de perméat d’ultrafiltration de lait écrémé
ont été utilisées pour standardiser à la baisse la poudre de lait écrémé à partir d’une teneur en pro-
téines de 35,5–30 % sur matière sèche. Le WPNI de ces poudres standardisées a été calculé selon la
méthode de l’ADPI (American Dairy Products Institute). Les poudres ont également été analysées
pour leur teneur en protéines, cendres, matière grasse, eau et lactose par les méthodes standards
pour en connaître la composition. L’utilisation de la régression linéaire a permis de mettre en évi-
dence que le WPNI est positivement associé à la teneur en protéines pour des poudres « low heat »
et « medium heat » provenant de plusieurs fournisseurs. Par exemple, une poudre de lait écrémé
« low heat » (fournisseur B) avec une teneur en protéines initiale de 34,3 % (sur matière sèche) et
un WPNI de 6,38 mg d’azote soluble des protéines de lactosérum (classée comme poudre « low
heat ») avait son WPNI réduit à moins de 6,0 mg d’azote soluble par gramme de poudre quand elle
était standardisée à une teneur en protéines de moins de 31,89 % (sur matière sèche) avec de la
poudre de lactose ou de perméat, ce qui pourrait la reclasser comme poudre « medium heat ». Nous
pouvons conclure que la standardisation de la poudre de lait écrémé par du lactose ou du perméat
change sa valeur de WPNI et peut avoir un effet sur sa classification. Nous proposons une approche
modifiée pour calculer le WPNI basée sur l’azote soluble des protéines de lactosérum par gramme
de protéines. Cette nouvelle valeur de WPNI (corrigée en protéines) serait indépendante de la teneur
en protéines de la poudre et par conséquent indiquerait mieux le traitement thermique réel qu’il est
censé traduire.

indice d’azote des protéines solubles / standardisation en protéines / poudre de lactose alimen-
taire / poudre de perméat / poudre de lait écrémé

1. INTRODUCTION The above relationship was used to es-

tablish the widely practiced heat classi-
fication system based on the Whey Pro-
Prior to drying, skim milk or skim milk tein Nitrogen Index (WPNI) analytical
concentrates can be given an additional method [1]. In this method, the turbidity of
heat treatment to alter the functional prop- a sample is measured and then correlated
erties such as solubility, water holding, to a standard curve to determine the soluble
emulsification, and viscosity of the result- whey protein nitrogen content in 1 g sam-
ing dried powder [16]. It has been shown ple of milk powder [12]. Nonfat dry milk
that this heat treatment is related to the ex- (NFDM) powders that have a WPNI value
tent of whey protein denatured [9]. greater than or equal to 6.0 mg soluble
 Impact of protein standardization on WPNI 107

whey protein nitrogen per g powder are de- document the influence of protein stan-
fined as low heat, between 1.51–5.99 mg dardization on heat classification of milk
soluble whey protein nitrogen per g pow- powders. Thus, the present study was con-
der are defined as medium heat, and less ducted to determine the impact of down-
than or equal to 1.51 mg soluble whey pro- wards protein standardization of nonfat dry
tein nitrogen per g powder are defined as milk on its WPNI value and its subsequent
high heat [1]. heat classification.
The heat classification system for milk
powders has been used commercially over
the years to provide general guidelines for 2. MATERIALS AND METHODS
various food applications. However, the 2.1. Materials
value of the WPNI method has not been
without criticism. It has been reported that Commercial samples of low heat and
the soluble whey protein nitrogen level can medium heat NFDM were obtained from
be influenced by natural variation in the three different manufacturers: A (low
concentration of the proteins in the raw heat); B (low heat); and C (medium heat).
milk and thus, has limited value as an in-
dex of heating history [5, 10, 14, 15, 19].
2.2. Sample preparation
There is an interest in standardizing the
protein content of milks destined for manu- Commercial edible lactose powder
facture into skim milk powder (SMP) [11]. (ELP) and commercial permeate pow-
According to Codex Alimentarius 207 [2], der (PP) from skim milk ultrafiltration
the use of milk derived permeate or edible were obtained and used for downwards
lactose can be used to standardize the pro- protein standardization of low heat and
tein content of milk powder or milk des- medium heat skim milk powder. While
tined for manufacture into SMP to 34% standardization below 34% protein on
protein on a solids-not-fat (SNF) basis. Re- a SNF basis is not allowed by Codex
cently, it has been reported that amend- Alimentarius 207, we studied such values
ment to European Commission (EC) direc- to better confirm the observed trends of
tive will allow for protein standardization the effects of protein standardization of
of SMP with 34% content (SNF) in line SMP. Figure 1 shows protein standard-
with Codex Alimentarius [3]. For exam- ization was achieved by dry blending of
ple, a SMP containing moisture content of predetermined quantity of ELP or PP to
4%, 1.4% fat content and 32.6% protein low heat NFDM to achieve target protein
(wet basis) translates into 34% protein con- content. Protein standardized samples are
tent on a SNF basis. Fluid milk applica- presented according to the level of protein
tion studies on protein standardization of and type of standardization, for example,
milk have focused to improve the yield for NFDM powder standardized with ELP
cheese making [6, 18], functional proper- standardized to 34% protein content was
ties such as heat stability [17], and storage named MPSELP34.
stability of UHT milk [7].
Despite the reported limitations of the
WPNI method, it remains a tool widely uti- 2.3. Methods
lized in commercial practice. In addition, 2.3.1. Whey protein nitrogen index
expectations for consistent composition (WPNI)
and functionality of all dairy ingredients
drive interest in protein standardization. Samples of low heat, medium heat
Further, little information is available to and standardized SMP were analyzed for
 108

NFDM Powder (B)

(34.3% protein content)
(wet basis)

ELP PP
(Amount of ELP / 100 g powder) (Amount of PP / 100 g powder)
V. Sikand et al.

MPSELP33 MPSELP32 MPSELP31 MPSELP30 MPSPP33 MPSPP32 MPSPP31 MPSPP30

(3.79 g) (6.70 g) (9.62 g) (12.53 g) (3.79 g) (6.70 g) (9.62 g) (12.53 g)

Figure 1. Flow diagram for low heat nonfat dry milk (NFDM) powder (B) standardization with edible lactose powder (ELP) or permeate powder (PP)
to different protein contents.
 Impact of protein standardization on WPNI 109

WPNI by American Dairy Products In- heat NFDM, and standardized SMP by dif-
stitute (ADPI) method [1]. ADPI has ference.
adopted the most commonly used WPNI
method, which was originally developed
by Harland and Ashworth [8] and fur- 2.3.3. Statistical analyses
ther modified by Kuramoto et al. [12] and
Leighton [13]. This turbidimetric method Results were evaluated using a multi-
is based on removal of casein and dena- ple linear regression in both Minitab Ver-
tured whey protein (DWPN) from unde- sion 14.2 (Minitab, Inc., State College,
natured whey protein (UWPN) by precip- PA, USA) and Data Desk Version 6.1
itation with saturated NaCl solution. Then, (Data Description, Inc., Ithaca, NY, USA)
2 drops of 10% HCl is added to a test software. Possible predictors of WPNI
tube containing 1 mL sample of filtrate that were included in the model were
and 10 mL of saturated NaCl solution. powder supplier (A, B, C), powder type
Then, the developed turbidity of the sam- (low heat, medium heat), standardization
ple is measured at 420 nm and compared method (ELP, PP), and protein level (35.5–
against a standard curve generated from 30%).
analysis of “standard” milk powders of
WPNI value obtained from ADPI low heat
NFDM (WPNI of 7.32 mg N·g−1 ) and high 3. RESULTS AND DISCUSSION
heat NFDM (WPNI of 0.32 mg N·g−1 )
using the above method. It is assumed 3.1. Descriptive NFDM composition
that developed turbidity corresponds to the
amount of UWPN in the original milk Crude protein analysis of three commer-
powders. Hence, results are reported as cial NFDM low heat (A), low heat (B),
mg soluble whey protein nitrogen per g of and medium heat (C) along with ELP
powder. and PP are presented in Table I. The av-
erage crude protein content observed for
low heat NFDM (A), low heat NFDM (B)
2.3.2. Composition analysis and medium heat NFDM (C) was 35.1%,
34.3%, and 35.5% respectively, 0% for
Samples of low heat and medium heat ELP and 2.40% for the PP. PP does not
NFDM powder were analyzed for protein, have true protein and hence, the 2.4% of
ash, moisture, fat, and lactose content. To- crude protein value really reflects non-
tal nitrogen (TN) was determined in the protein nitrogen content of permeate. A de-
low heat NFDM, medium heat NFDM, tailed analysis of the three samples is dis-
ELP, PP and standardized SMP samples by cussed below.
Kjeldahl method (AOAC, method number
991.20; 33.2.11) [4]. Ash was determined
by ignition at 550 ◦ C in an electric muffle 3.1.1. Low heat NFDM (A)
furnace (AOAC, method number 945.46;
33.2.10) [4]. Fat content was determined Effect of protein standardization on
by Mojonnier method (AOAC, method composition of low heat NFDM (A)
number 989.05; 33.2.26) [4]. Free moisture is summarized in Table II. Low heat
was determined by direct forced air oven NFDM (A) with 35.1% protein content on
drying method (AOAC, method number wet basis (36.3% on dry basis; 36.9% on
990.20; 33.2.44) [4]. Lactose content was SNF basis) was downward standardized to
determined in low heat NFDM, medium 34%, 33%, 32%, 31% and 30% protein
110 V. Sikand et al.

Table I. Crude protein analysis (wet, dry, SNF basis) of the three commercial milk powders (non-
standardized), edible lactose powder and milk permeate powder.

Ingredient Crude protein*

Wet basis 1 Dry basis 2 SNF basis 3
Low heat SMP (A) 35.1 ± 0.0 36.32 ± 0.0 36.9 ± 0.0
Low heat SMP (B) 34.3 ± 0.2 35.66 ± 0.3 35.97 ± 0.3
Medium heat SMP (C) 35.5 ± 0.1 36.71 ± 0.0 37.05 ± 0.1
Edible lactose powder 0.00 ± 0.0 0.00 ± 0.0 0.00 ± 0.0
Milk permeate powder 2.40 ± 0.1 2.50 ± 0.1 2.50 ± 0.1
n = 2.
* Crude protein = Total nitrogen * 6.38.
1 Wet basis = As is.
2 Dry basis = 100 – % free moisture content.
3 SNF basis = 100 – % free moisture content – % fat content.

with ELP. The corresponding protein con- SNF basis) was downward standardized to
tent on dry basis and SNF basis for low 33%, 32%, 31% and 30% protein with ELP
heat NFDM (A) is listed in Table II. WPNI or PP. The corresponding protein content
values for the standardized samples were on dry basis and SNF basis for downward
observed to decrease up to 15% with de- standardized low heat NFDM (B) is listed
crease in protein content from 35.1–30% in Table III.
protein. A decrease in WPNI values of ELP WPNI values for the ELP standard-
standardized samples indicates that ELP ized samples were observed to decrease
does contribute towards WPNI. Similarly, up to 11% as protein was standardized
ash values for the standardized samples downwards from 35.3–30%. A decrease in
were observed to decrease up to 15% with WPNI values of ELP standardized sam-
decrease in protein content from 35.1–30% ples indicates that ELP does contribute to-
protein. The decrease of ash values is at- wards WPNI. Similarly, ash values for the
tributed to the observed low mineral con- ELP standardized samples were observed
tent of the ELP. Moisture content for the to decrease up to 13%, with a decrease in
ELP standardized samples was observed to protein content of the SMP. The decrease
decrease up to 17%, with decrease in pro- of ash values is attributed to the low min-
tein and hence, results in decrease of pro- eral content of the ELP. Moisture content
tein to lactose ratio. Fat content for the ELP for the ELP standardized samples were ob-
standardized samples were observed to de- served to decrease up to 8%, with decrease
crease up to 37% because ELP has negligi- in protein and hence, results in decrease of
ble fat contribution. protein to lactose ratio. Fat content for the
ELP standardized samples were observed
to decrease up to 35% with decrease in
3.1.2. Low heat NFDM (B) protein content from 34.3–30% protein be-
cause ELP has negligible fat contribution.
Effect of protein standardization on WPNI values for the PP standardized
the composition of low heat NFDM (B) samples were observed to decrease up
is summarized in Table III. Low heat to 12% as protein was standardized
NFDM (B) with 34.3% protein content on downwards from 34.3–30%. A decrease in
wet basis (35.7% on dry basis; 36% on WPNI values of PP standardized samples
 Impact of protein standardization on WPNI 111

Table II. Composition analysis (mean ± SD) of low heat NFDM powder (A) standardized with
edible lactose powder.

Sample Target Protein Protein b WPNI b Ash b Moisture b Fat b Lactose* Protein/
type protein (dry matter SNF Lactose
basis)a basis
Control 35.1 36.3 36.9 6.36 ± 0.4 7.66 ± 0.0 3.36 ± 0.1 1.5 ± 0.0 52.4 0.67
MPSELP34 34.0 35.1 35.6 6.04 ± 0.9 7.52 ± 0.1 3.04 ± 0.1 1.45 ± 0.0 54.0 0.63
MPSELP33 33.0 34.0 34.5 5.84 ± 0.2 7.27 ± 0.0 2.93 ± 0.0 1.37 ± 0.1 55.4 0.60
MPSELP32 32.0 33.0 33.4 5.79 ± 2.0 7.00 ± 0.1 2.89 ± 0.1 1.25 ± 0.1 56.9 0.56
MPSELP31 31.0 31.9 32.3 5.54 ± 0.2 6.82 ± 0.0 2.88 ± 0.1 1.11 ± 0.0 58.2 0.53
MPSELP30 30.0 30.9 31.2 5.41 ± 0.2 6.49 ± 0.0 2.79 ± 0.1 0.95 ± 0.1 59.8 0.50

n = 2.
Results expressed in g/100 g of milk powder.
a Dry matter basis = 100 – % free moisture.
b Results based on wet basis.
* Calculated value (by difference).
MPSELP = Milk powder standardized with edible lactose powder.

Table III. Composition analysis (mean ± SD) of low heat NFDM powder (B) standardized with
edible lactose powder or permeate powder.

Sample Target Protein Protein b WPNI b Ash b Moisture b Fat b Lactose* Protein/
type protein (dry matter (SNF Lactose
basis)a basis)
Control 34.3 35.7 36.0 6.38 ± 0.1 7.74 ± 0.0 3.81 ± 0.3 0.83 ± 0.0 53.4 0.64
MPSELP33 33.0 34.1 34.3 6.18 ± 0.0 7.45 ± 0.0 3.18 ± 0.1 0.68 ± 0.0 55.7 0.59
MPSELP32 32.0 33.1 33.3 6.03 ± 0.0 7.23 ± 0.0 3.38 ± 0.1 0.63 ± 0.1 56.8 0.56
MPSELP31 31.0 32.2 32.3 5.89 ± 0.1 7.10 ± 0.1 3.61 ± 0.4 0.57 ± 0.0 57.7 0.54
MPSELP30 30.0 31.1 31.3 5.68 ± 0.1 6.74 ± 0.0 3.50 ± 0.0 0.54 ± 0.0 59.2 0.51

MPSPP33 33.0 34.3 34.5 6.23 ± 0.1 7.73 ± 0.1 3.66 ± 0.3 0.83 ± 0.1 54.8 0.60
MPSPP32 32.0 33.2 33.5 5.98 ± 0.0 7.72 ± 0.1 3.70 ± 0.2 0.72 ± 0.0 55.9 0.57
MPSPP31 31.0 32.2 32.4 5.86 ± 0.0 7.70 ± 0.1 3.68 ± 0.1 0.67 ± 0.1 57.0 0.54
MPSPP30 30.0 31.3 31.5 5.60 ± 0.1 7.70 ± 0.1 4.13 ± 0.1 0.60 ± 0.1 57.6 0.52

n = 2.
Results expressed in g/100 g of milk powder.
a Dry matter basis = 100 – % free moisture.
b Results based on wet basis.
* Calculated value (by difference).
MPSELP = Milk powder standardized with edible lactose powder.
MPSPP = Milk powder standardized with permeate powder.
112 V. Sikand et al.

indicates that PP does contribute towards as protein was standardized downwards

WPNI. Similarly, mean values of ash for from 35.5–30%. A decrease in WPNI val-
PP standardized NFDM decreased up to ues of PP standardized samples indicates
0.3%. The very small changes in ash value that PP does contribute towards WPNI.
are attributed to high mineral content of Similarly, mean values of ash for PP stan-
PP. Fat content for the PP standardized dardized NFDM decreased up to 0.3%.
samples were observed to decrease up The very small changes in ash value are
to 28% with decrease in protein content attributed to high mineral content of PP.
from 34.3–30% protein, as milk powder re- Fat content for the PP standardized sam-
placed with PP has fat content less than ples was observed to decrease up to 25%
0.1%. Moisture content for the PP stan- with a decrease in the protein content from
dardized samples were observed to in- 35.5–30% protein as milk powder replaced
crease up to 8%, with decrease in protein with PP has fat content less than 0.1%.
and hence, results in a decrease of protein Moisture content for the PP standardized
to lactose ratio. samples were observed to increase up to
11%, with decrease in protein and hence,
results in decrease of protein to lactose ra-
3.1.3. Medium heat NFDM (C) tio.

Effect of protein standardization on the

composition of medium heat NFDM (C) 3.2. Linear regression analysis
is summarized in Table IV. Medium heat
NFDM (C) with 35.5% protein content Based on the results obtained from Ta-
on wet basis (36.7% on dry basis; 37.0% bles II–IV, the relationship between pro-
on SNF basis) was downward standard- tein level and WPNI was analyzed us-
ized to 34%, 32% and 30% protein with ing two multiple linear regression models.
ELP or PP. WPNI values for ELP standard- Two models are necessary because low and
ized samples were observed to decrease medium heat samples were not available
up to 18% as protein was standardized from all three suppliers. Each model was
downwards from 35.5–30%. A decrease in tested for different effects within the ex-
WPNI values of ELP standardized sam- periment. Model 1 tests for differences be-
ples indicates that ELP does contribute to- tween low and medium heat powders from
wards WPNI. Similarly, ash values for the supplier C only. Model 2 uses all three sup-
ELP standardized samples were observed pliers for only low heat powder.
to decrease up to 15%, with a decrease in
protein content of the SMP. The decrease
of ash values is attributed to the low min- 3.2.1. Model 1
eral content of the ELP. Moisture content
for the ELP standardized samples was ob- Model 1 compares WPNI and protein
served to decrease up to 9%, with decrease level for both medium and low heat NFDM
in protein and hence, results in decrease of powders. Only samples from Supplier C
protein to lactose ratio. Fat content for the were used because only Supplier C had
ELP standardized samples was observed to both low heat and medium heat NFDM.
decrease up to 56% with decrease in pro- The powders were standardized with either
tein content from 35.5–30% because ELP ELP or PP to adjust each sample to a spec-
has negligible fat contribution. ified target protein level between 30% and
WPNI values for PP standardized sam- 34%. WPNI values for ELP standardized
ples were observed to decrease up to 15% low heat milk powder were found to be
 Impact of protein standardization on WPNI 113

Table IV. Composition analysis (mean ± SD) of medium heat NFDM (C) standardized with edible
lactose powder or permeate powder.

Sample a Ta rget Protein Protein b WPNI b Ash b Moisture b Fat b Lactose* Protein/
type protein (dry matter (SNF Lactose
basis)a basis)

Control 35.5 36.7 37.0 3.43 ± 0.0 7.75 ± 0.0 3.30 ± 0.3 0.88 ± 0.1 52.6 0.68
MPSELP34 34.0 35.1 35.4 3.11 ± 0.1 7.47 ± 0.0 3.25 ± 0.3 0.67 ± 0.1 54.6 0.62
MPSELP32 32.0 33.0 33.2 3.00 ± 0.1 7.05 ± 0.0 3.09 ± 0.2 0.44 ± 0.1 57.4 0.56
MPSELP30 30.0 30.9 31.0 2.80 ± 0.2 6.61 ± 0.0 2.99 ± 0.3 0.39 ± 0.0 60.0 0.50

MPSPP34 34.0 35.2 35.5 3.32 ± 0.1 7.80 ± 0.0 3.40 ± 0.1 0.84 ± 0.1 53.9 0.63
MPSPP32 32.0 33.2 33.4 3.16 ± 0.3 7.76 ± 0.0 3.57 ± 0.1 0.74 ± 0.1 55.9 0.57
MPSPP30 30.0 31.1 31.4 2.92 ± 0.0 7.78 ± 0.0 3.65 ± 0.2 0.66 ± 0.1 57.9 0.52

n = 2.
Results expressed in g/100 g of milk powder.
a Dry matter basis = 100 – % free moisture.
b Results based on wet basis.
* Calculated value (by difference).
MPSELP = Milk powder standardized with edible lactose powder.
MPSPP = Milk powder standardized with permeate powder.

6.38 ± 0.2, 5.96 ± 0.0, 5.61 ± 0.1 mg WPNI. The equation for Model 1A is
nitrogen per g powder for 34%, 32% and
E (WPNI) = β0 + βPro (xPro − 32)
30% protein respectively. WPNI values
+ βMed xMed + βPP xPP + βPro∗Med (xPro − 32)
for PP standardized low heat milk powder
× xMed + βPro∗PP × (xPro − 32) xPP
were found to be 6.39 ± 0.2, 6.09 ± 0.1,
+ βMed∗PP xMed xPP + βPro∗Med∗PP (xPro − 32)
and 5.74 ± 0.1 mg nitrogen per g pow-
× xMed xPP
der for 34%, 32% and 30% protein respec-
tively. WPNI value for control (unadjusted) where E(WPNI) is mean WPNI, (xPro –
low heat milk powder sample was found 32) is the target protein level centered to a
to be 6.90 ± 0.1 mg nitrogen per g pow- value of 32%, and xMed and xPP are indi-
der for 35.5% protein. Initially, the control cator variables representing medium heat
samples were omitted from the analysis so NFDM and standardization with perme-
that the interaction of standardization and ate. The β terms represent the effects of
protein level could be tested. Because the these variables and their interactions on the
control samples did not share any target WPNI level.
protein levels with the ELP and PP sam- Model 1A explains 99.6% of the vari-
ples, this interaction cannot be tested with ation in the WPNI values of the samples.
control samples in the data. After omit- Initial tests showed no significant differ-
ting the control samples, data from 24 stan- ence in mean WPNI between the lactose
dardized powder samples were analyzed and permeate samples (p = 0.38), so
in Model 1A. The purpose of this model all terms involving standardization effects
was to estimate whether the different stan- were removed to simplify the model, and
dardizations and powder types have differ- the control samples were restored. We re-
ing relationships between protein level and fer to this model as Model 1B, which has
114 V. Sikand et al.

Scatterplot of WPNI vs target protein

7
Low heat
classification
6

5 low heat
WPNI

medium heat
4 Medium heat
classification

2
30 31 32 33 34 35 36
Target protein

Figure 2. Effect of protein standardization on WPNI (mg of WPN·g−1 of powder) of low heat or
medium heat SMP.

equation heat powder,

E (WPNI) = β0 + βPro (xPro − 32) E(WPNI) = 6.053 + 0.206(xPro − 32)

+ βMed xMed + βPro∗Med (xPro − 32) xMed .
where (xPro – 32) is the deviation in per-
centage points from 32% protein. There-
This model explains 99.5% of the variation
fore, at 32% protein, the low heat pow-
in WPNI. After adjusting for protein level,
der is estimated to have a mean WPNI of
mean WPNI of medium heat powder was
6.053 mg nitrogen per g powder. Each in-
significantly lower than that of low heat
crease of one percentage point in protein
powder (p < 0.001). At the median pro-
is associated with an increase of 0.206 mg
tein level of 32%, medium heat powder had
nitrogen per g powder in mean WPNI for
an average WPNI that was 3.10 mg nitro-
low heat powder. With 95% confidence the
gen per g of powder, lower than low heat
change in WPNI per percent of protein is
powder. After adjusting for NFDM pow-
between 0.170 and 0.242 mg nitrogen per g
der type, protein level was positively asso-
powder.
ciated with WPNI (p < 0.001). In addition,
For medium heat powder,
protein level had a greater effect on WPNI
in low heat powder than in medium heat E(WPNI) = 2.952 + 0.119(xPro − 32).
powder (p = 0.002). These effects are de-
scribed in greater detail below. Therefore, at 32% protein, medium heat
Model 1B produced two equations to powder is estimated to have a mean WPNI
describe the relationship between protein of 2.952 mg nitrogen per g powder. Each
and WPNI. These equations are displayed increase of 1 percentage point in protein
along with the data in Figure 2. For low is associated with an increase of 0.119 mg
 Impact of protein standardization on WPNI 115

nitrogen per g powder in mean WPNI for Model 2B explains 87.9% of the varia-
medium heat powder. With 95% confi- tion in WPNI values. After adjusting for
dence the change in WPNI per percent of supplier, protein level was positively as-
protein is between 0.083 and 0.156 mg ni- sociated with WPNI (p < 0.001). The
trogen per g powder, which is substantially effect of protein on WPNI did not differ
lower than the effect of changing protein significantly from one supplier to another
on low heat powders. (p = 0.24). After adjusting for protein
level, powder from supplier A had a mean
WPNI significantly lower than either sup-
3.2.2. Model 2 plier B or C (p < 0.001). This difference
is not surprising given that each suppli-
Model 2 compares WPNI and protein ers’ thermal processing of milk evapora-
for the low heat powders from suppliers A, tion, spray drying conditions (all of which
B, and C. The purpose of Model 2 was can effect WPNI) are likely different. No
to determine if the effect of protein on significant difference was detected in the
WPNI differed based on standardization or mean WPNI of powder from suppliers B
supplier. As in the earlier model, samples and C (p = 0.26).
were standardized with either ELP or PP After simplifying the model by drop-
to target protein levels between 30% and ping insignificant terms, the equations de-
34%. As before, the control samples were scribing the relationship between protein
initially omitted from the analysis to test and WPNI in low heat powders from our
for interaction effects. Data from 41 sam- three suppliers are
ples were used in this analysis, Model 2A,
which has equation Supplier A: E (WPNI) = 5.741 + 0.187
× (xPro − 32)
E (WPNI) = β0 + βPro (xPro − 32) Supplier B: E (WPNI) = 6.021 + 0.187
+ βA xA + βB xB + βPer xPP + βPro∗A (xPro − 32) × (xPro − 32)
×xA + βPro∗B (xPro − 32) xB + βPro∗PP Supplier C: E (WPNI) = 6.063 + 0.187
× (xPro − 32) xPP + βB∗PP xB xPP + βPro∗B∗PP × (xPro − 32).
× (xProt − 32) xB xPP
Regardless of supplier, each increase of
where E(WPNI), (xPro – 32), and xPP are one percentage point in protein is associ-
defined as before while xA and xB are indi- ated with an increase of 0.187 mg nitro-
cator variables for suppliers A and B. gen per g powder in the mean WPNI. With
Model 2A explains 89.4% of the vari- 95% confidence the true change in WPNI
ation in the WPNI values of the samples. per percent of protein is between 0.168 and
The poorer fit is due to larger variation 0.207 mg nitrogen per g of powder.
in the WPNI values from supplier A. Ini- Figure 3 shows that some low heat
tial tests again showed no significant dif- powder samples had WPNI of less than
ference in mean WPNI between the lac- 6.0 mg nitrogen per g powder, which
tose and permeate samples (p = 0.38), would change the heat classification of the
so all terms involving standardization ef- powder from low to medium heat. For
fects were removed from the model, and supplier A, the change from low heat to
the control samples were restored. We refer medium heat classification is estimated to
to this model as Model 2B with equation occur when protein is 33.39% on wet ba-
sis (34.6 on dry basis; 35.1 on SNF basis).
E (WPNI) = β0 + βPro (xPro − 32) + βA xA These changes in classification are signif-
+ βB xB + βPro∗A (xProt − 32) xB + βPro∗B icant because they occur at protein values
× (xProt − 32) xC . before the legal minimum of 34% protein
116 V. Sikand et al.

on SNF basis. For suppliers B and C, the protein content of 35.5, we define low heat
change from low heat to medium heat clas- NFDM as having a WPNIpcorr value greater
sification is estimated to occur when pro- than or equal to 16.9 mg nitrogen per g pro-
tein is 31.89% and 31.66% on wet basis tein, and we define medium heat NFDM as
(33.1 on dry basis, 33.3 on SNF basis and having a WPNIpcorr value between 4.22 and
32.7 on dry basis, 33 on SNF basis) respec- 16.9 mg nitrogen per g protein. These cut-
tively. off values and additional values for other
The results of the present study indicate protein levels are shown in Tables V–VII.
that WPNI value of SMP is influenced by To validate these proposed WPNIpcorr
the protein level; however, protein has a levels for low and medium heat NFDM,
larger effect on WPNI for low heat pow- a linear regression model was fit to
ders compared to medium heat powders. In the WPNIpcorr data for supplier C using
some cases, this change in WPNI changes the powder type, standardization, target
the actual heat classification (Fig. 2), sug- percent protein level, and their interac-
gesting that selection of milk powder based tions as predictors. As before the model
on traditional WPNI value (standard un- (Model 3A) was initially fit without us-
corrected) could be misleading under some ing the control samples in order to test the
situations of protein standardization. effect of standardization type and its in-
teractions. Model 3A has the same equa-
tion as Model 1A, but with WPNIpcorr as
3.3. Proposed modification of heat the response variable. This model explains
classification measure 99.6% of the variation in WPNIpcorr . The
analysis shows no significant effect of stan-
The turbidimetric method is the stan- dardization on WPNIpcorr (p = 0.33).
dard method for heat classification of milk Removing all terms containing stan-
powder. Since there is natural variation in dardization type and restoring the control
the protein concentration of raw milk, any samples, we create Model 3B (Fig. 4).
indicator of functionality of SMP turbid- This model explains 99.5% of the varia-
ity should correct for the protein level. Fur- tion in WPNIpcorr . This analysis shows that
ther, in light of allowable downwards pro- medium heat NFDM powder has a lower
tein standardization and its effects reported mean WPNIpcorr than low heat NFDM
here on traditional WPNI value, use of un- powder (p < 0.01). However, there is no
corrected WPNI may not be desirable. We significant difference in mean WPNIpcorr
propose a modified classification measure for different protein levels (p = 0.15),
using mg of whey protein nitrogen per g of and the effect of protein on WPNIpcorr does
protein (WPNIpcorr ). not differ significantly between low and
Low heat NFDM is defined as having medium heat powders (p = 0.66).
a WPNI of 6.0 mg soluble nitrogen per g After removing the insignificant terms
powder or more. Dividing this value by a containing protein level, Model 3C was
percent target protein level of 35.5 yields used to estimate the WPNIpcorr for both
a WPNIpcorr of 16.9 mg soluble nitrogen types of NFDM with 35.5% (supplier C)
per g protein. Medium heat NFDM is de- protein content. Mean WPNIpcorr of all low
fined as having a WPNI between 1.5 and heat powder from supplier C is between
6.0 mg soluble nitrogen per g powder. 18.72 and 19.16 mg soluble nitrogen per g
Dividing these values by the target pro- protein. Mean WPNIpcorr for all medium
tein percentage of 35.5 yields a WPNIpcorr heat powder from supplier C is between
between 4.2 and 16.9 mg soluble nitro- 9.04 and 9.48 mg soluble nitrogen per g
gen per g protein. Therefore for a percent protein. Individual samples of low heat
 Impact of protein standardization on WPNI 117

Scatterplot of WPNI vs target protein

7.00

6.75

6.50 low heat

supplier
A
WPNI

6.25
B
C
6.00

5.75
medium heat
5.50 Transition from
low to medium heat

30 31 32 33 34 35 36
Target protein

Figure 3. Effect of protein standardization on WPNI (mg of WPN·g−1 of powder) of low heat SMP.

Table V. Effect of absolute protein content on heat classification of milk powder based on soluble
whey protein nitrogen·g−1 of protein.

Traditional Modified WPNIpcorr

WPNI

Heat NFDM NFDM (A) NFDM (B) NFDM (C )

classification mg of soluble mg of soluble mg of soluble mg of soluble
WPN·g–1 powder WPN·g–1 protein WPN·g–1 protein WPN·g–1 protein
(Protein = 34.3 g·100 g–1) (Protein = 35.1 g·100 g–1) (Protein = 35.5 g·100 g–1)

Low heat > 6.0 > 17.49 > 17.09 > 16.90
Medium heat 1.5–6.0 4.37–17.49 4.27–17.09 4.22–16.90
High heat < 1.5 < 4.37 < 4.27 < 4.22

Protein is expressed g·100 g–1 of powder as 34.3 or 0.343; 35.1 or 0.351; and 35.5 or 0.355.

powder may vary from 18.08 to 19.79 mg Any indicator of functionality of SMP
soluble nitrogen per g protein, while in- needs to account for the protein level. For
dividual samples of medium heat powder example, PP standardized low heat (B)
may vary from 8.40 to 10.11 mg soluble SMP sample with a target of 32.0% pro-
nitrogen per g protein. All estimates are tein content and a WPNI value of 5.98 mg
made with 95% confidence. The model es- soluble nitrogen per g powder would be
timates of WPNIpcorr for individual sam- a medium heat milk powder (Tab. VII) as
ples of low and medium heat powders fit per current nonfat dry milk heat classi-
comfortably within the ranges proposed for fication (WPNI uncorrected). However, if
35.5% protein in Table V. WPNI values are expressed as mg soluble
118 V. Sikand et al.

Table VI. Changes on heat classification of low heat SMP (A) due to protein standardization and
method of expression of WPNI.

Sample Target WPNI Powder classification WPNI Modified Powder classification

type protein (traditional) based on WPN·g–1 protein based on
WPNI·g–1 powder (WPNIpcorr) WPNI·g–1 protein

Control 35.1 6.36 Low 18.11 Low

MPSELP34 34.0 6.04 Low 17.76 Low
MPSELP33 33.0 5.84 Medium 17.69 Low
MPSELP32 32.0 5.79 Medium 18.09 Low
MPSELP31 31.0 5.54 Medium 17.87 Low
MPSELP30 30.0 5.41 Medium 18.03 Low

MPSELP = Milk powder standardized with edible lactose powder.

WPNIpcorr = Protein corrected WPNI.

Table VII. Changes on heat classification of low heat SMP (B) due to protein standardization and
method of expression of WPNI.

Sample Target Traditional Powder classification WPN.g−1 protein Powder classification

type protein WPNI based on (WPNIpcorr) based on
WPNI.g−1 powder WPNI.g−1 protein
Control 34.3 6.38 Low (control) 18.60 Low
MPSELP33 33.0 6.18 Low 18.72 Low
MPSELP32 32.0 6.03 Low 18.84 Low
MPSELP31 31.0 5.89 Medium 19.00 Low
MPSELP30 30.0 5.68 Medium 18.93 Low
MPSPP33 33.0 6.23 Low 18.87 Low
MPSPP32 32.0 5.98 Medium 18.68 Low
MPSPP31 31.0 5.86 Medium 18.90 Low
MPSPP30 30.0 5.60 Medium 18.66 Low

MPSELP = Milk powder standardized with edible lactose powder.

MPSPP = Milk powder standardized with permeate powder.
WPNIpcorr = Protein corrected WPNI.

nitrogen per g protein (WPNIpcorr ), then misclassification of low heat powder to a

the WPNI value of 5.98 mg soluble ni- medium heat powder due to protein con-
trogen per g powder at 32% protein has a tent.
WPNIpcorr value of 18.68 mg soluble nitro- Further, medium heat or high heat
gen per g protein and would still be clas- powder classification will be promptly
sified as low heat powder. In other words, achieved more often when standardizing
this modification in WPNI use is to prevent downwards protein content. For example,
 Impact of protein standardization on WPNI 119

S catterplot of WPN . g -1 protein vs. target protein

20

18
low heat
Tran sitio n fro m lo w h eat to med iu m h eat c lassific atio n
16
WPN.g -1 protein

m edium heat low heat

14 m edum heat

12

10

30 31 32 33 34 35 36
Targ et protein

Figure 4. Effect of protein standardization on WPNI (mg of WPN·g−1 of protein) of low heat and
medium heat SMP.

low heat milk powder with a WPNI value provide the assurance in consistent perfor-
greater than or equal to 6.0 mg nitrogen mance. Re-expression of WPNI per g pro-
per g powder, but close to medium heat tein prevents erroneous heat classification
classification value less than or equal to due to protein content variation. This pro-
5.9 mg nitrogen per g powder can achieve posed modified WPNI (protein corrected)
medium heat classification upon its protein would enhance the value of this method
standardization as WPNI is positively re- as a means to select milk powders for
lated to protein level. Similarly, a medium their use in food applications. In light of
heat milk powder with a WPNI value readily available protein testing today, re-
closer to 1.5 mg of nitrogen per g powder expression of WPNI to account for the
can achieve high heat classification upon potential protein content impact in heat
its protein standardization. classification may be justified. The present
study suggests milk with higher protein
content will have higher WPNI when ev-
4. CONCLUSIONS erything else is same. However, we also
We have demonstrated that a powder note that WPNI was different between sup-
with higher protein content will inherently pliers’ at equal protein content in milk
have a greater chance of having a higher powders. Further study will be needed to
traditional WPNI value (uncorrected for explain the effects of plant processing con-
protein) at equal heat treatment than a pow- ditions on SMP heat classification.
der with lower protein content. Conversely,
achieving a medium-heat classification or Acknowledgements: Authors acknowledge
higher heat classification for a high protein Dairy Management Inc., US Dairy Export
powder can be more difficult. Specifying Council and California Dairy Research Foun-
dation for their financial support and the
a WPNI value for milk powder purchases commercial dairy ingredient manufacturers for
or broad heat classification without a con- providing milk powder, lactose and permeate
sistent protein level may not necessarily samples for this study.
120 V. Sikand et al.

REFERENCES limitation of their use in evaluating the heat

treatment of milk, J. Dairy Sci. 38 (1955)
[1] Anonymous, Standards for grades of dry 858-869.
milks including methods of analysis, [11] Jensen K., Milk powders: Specifications in
American Dairy Products Institute, in: relation to the products to be manufac-
Bulletin 916 (1971). tured, in: Recombination of Milk and Milk
[2] Anonymous, Codex standard for milk Products, International Dairy Federation,
powders and cream powder, in: Codex Brussels, Belgium (1990), Special Issue
Alimentarius 207, 1999. 9001,104–125.
[3] Anonymous, EU-27 Dairy and Products [12] Kuramoto S., Jenness R., Coulter S.T.,
European Commission Simplifies Milk Choi R.P., Standardization of the Harland-
Regime 2007, in: GAIN Report, USDA Ashworth test for whey protein nitrogen, J.
Foreign Agricultural Service, 1–3. Dairy Sci. 42 (1959) 28–38.
[4] Association of Official Analytical Chemists,
[13] Leighton F.R., Determination of whey pro-
Official Methods of Analysis, Vol. I, 16th
tein index of skim milk powder, Aust. J.
edn., AOAC, Gaithersburg, MD, 1995.
Dairy Technol. 17 (1962) 186–188.
[5] Augustin M.A., Ingredients for recombina-
tion of dairy products, in: Proceedings of 3rd [14] O’Conner F., McKenna B.M., O’Sullivan
International Symposium on Recombined A.C., Bacteriological and whey protein de-
Milk and Milk Products, Special Issue 9902, naturation aspects of heating processes used
International Dairy Federation, Brussels, in the manufacture of low-heat skim-milk
Belgium (1999) 66–70. powder, Ir. J. Agric. Res. 8 (1969) 417–430.
[6] Guinee L.J., O’Kennedy B.T., Kelly P.M., [15] O’Sullivan A.C., O’Conner F., Aspects of
Effect of milk protein standardization using heat treatment classified skim milk powder
different methods on the composition and manufacture, Dairy Ind. 36 (1971) 340–344.
yields of Cheddar cheese, J. Dairy Sci. 89
(2006) 468–482. [16] Pearce R.J., Food functionality: success or
failure for dairy based ingredients, Aust. J.
[7] Hardham J.F., Effect of protein standardiza- Dairy Technol. 50 (1995) 15–23.
tion of milk by addition of UF milk permeate
on the composition and storage stability of [17] Rattray W., Jelen P., Thermal stability of
UHT processed milk, Aust. J. Dairy Technol. skim milk with protein content standardized
53 (1998) 22-27. by the addition of ultrafiltration permeates,
[8] Harland H.A., Ashworth U.S., A rapid Int. Dairy J. 6 (1996) 157–170.
method for estimation of whey proteins as [18] Rehman S.U., Farkye N.Y., Considine T.,
an indication of baking quality of nonfat dry Schaffner A., Effects of standardization of
milk solids, Food Res. 12 (1947) 247–251. whole milk with dry milk protein concen-
[9] Harland H.A., Coulter S.T., Jenness R., The trate on the yield and ripening of reduced-
effect of the various steps in the manufacture fat Cheddar cheese, J. Dairy Sci. 86 (2003)
on the extent of serum protein denaturation 1608–1615.
in nonfat dry milk solids, J. Dairy Sci. 35 [19] Sanderson W.B., Seasonal variations affect-
(1952) 363–368. ing the determination of the whey protein ni-
[10] Harland H.A., Coulter S.T., Jenness R., trogen index of skim milk powder, N. Z. J.
Natural variation of milk serum proteins as a Dairy Sci. Technol. 5 (1970) 48–52.